diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,66533 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.3838120209924975,
+  "eval_steps": 500,
+  "global_step": 95000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 4.040126536763131e-05,
+      "grad_norm": 720.1376953125,
+      "learning_rate": 8.080808080808081e-09,
+      "loss": 97.7333,
+      "step": 10
+    },
+    {
+      "epoch": 8.080253073526263e-05,
+      "grad_norm": 765.2322387695312,
+      "learning_rate": 1.6161616161616162e-08,
+      "loss": 64.9714,
+      "step": 20
+    },
+    {
+      "epoch": 0.00012120379610289395,
+      "grad_norm": 1526.16015625,
+      "learning_rate": 2.4242424242424243e-08,
+      "loss": 141.7881,
+      "step": 30
+    },
+    {
+      "epoch": 0.00016160506147052525,
+      "grad_norm": 1654.4871826171875,
+      "learning_rate": 3.2323232323232324e-08,
+      "loss": 94.7421,
+      "step": 40
+    },
+    {
+      "epoch": 0.00020200632683815657,
+      "grad_norm": 706.5513305664062,
+      "learning_rate": 4.040404040404041e-08,
+      "loss": 79.4392,
+      "step": 50
+    },
+    {
+      "epoch": 0.0002424075922057879,
+      "grad_norm": 1566.697998046875,
+      "learning_rate": 4.8484848484848486e-08,
+      "loss": 167.6519,
+      "step": 60
+    },
+    {
+      "epoch": 0.0002828088575734192,
+      "grad_norm": 1094.059326171875,
+      "learning_rate": 5.656565656565657e-08,
+      "loss": 74.5029,
+      "step": 70
+    },
+    {
+      "epoch": 0.0003232101229410505,
+      "grad_norm": 2690.986328125,
+      "learning_rate": 6.464646464646465e-08,
+      "loss": 101.7366,
+      "step": 80
+    },
+    {
+      "epoch": 0.0003636113883086818,
+      "grad_norm": 845.0662231445312,
+      "learning_rate": 7.272727272727274e-08,
+      "loss": 109.2449,
+      "step": 90
+    },
+    {
+      "epoch": 0.00040401265367631315,
+      "grad_norm": 1515.5186767578125,
+      "learning_rate": 8.080808080808082e-08,
+      "loss": 68.3294,
+      "step": 100
+    },
+    {
+      "epoch": 0.00044441391904394446,
+      "grad_norm": 875.564453125,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 96.3639,
+      "step": 110
+    },
+    {
+      "epoch": 0.0004848151844115758,
+      "grad_norm": 342.3182067871094,
+      "learning_rate": 9.696969696969697e-08,
+      "loss": 112.5309,
+      "step": 120
+    },
+    {
+      "epoch": 0.000525216449779207,
+      "grad_norm": 1264.6597900390625,
+      "learning_rate": 1.0505050505050506e-07,
+      "loss": 118.3208,
+      "step": 130
+    },
+    {
+      "epoch": 0.0005656177151468384,
+      "grad_norm": 1327.964111328125,
+      "learning_rate": 1.1313131313131314e-07,
+      "loss": 75.7366,
+      "step": 140
+    },
+    {
+      "epoch": 0.0006060189805144697,
+      "grad_norm": 1309.533447265625,
+      "learning_rate": 1.2121212121212122e-07,
+      "loss": 109.1942,
+      "step": 150
+    },
+    {
+      "epoch": 0.000646420245882101,
+      "grad_norm": 683.5692749023438,
+      "learning_rate": 1.292929292929293e-07,
+      "loss": 129.5559,
+      "step": 160
+    },
+    {
+      "epoch": 0.0006868215112497323,
+      "grad_norm": 965.5596923828125,
+      "learning_rate": 1.3737373737373738e-07,
+      "loss": 67.0329,
+      "step": 170
+    },
+    {
+      "epoch": 0.0007272227766173637,
+      "grad_norm": 1409.6192626953125,
+      "learning_rate": 1.4545454545454548e-07,
+      "loss": 100.9175,
+      "step": 180
+    },
+    {
+      "epoch": 0.000767624041984995,
+      "grad_norm": 936.8455810546875,
+      "learning_rate": 1.5353535353535356e-07,
+      "loss": 114.6267,
+      "step": 190
+    },
+    {
+      "epoch": 0.0008080253073526263,
+      "grad_norm": 1041.8524169921875,
+      "learning_rate": 1.6161616161616163e-07,
+      "loss": 100.7278,
+      "step": 200
+    },
+    {
+      "epoch": 0.0008484265727202576,
+      "grad_norm": 1001.06982421875,
+      "learning_rate": 1.6969696969696974e-07,
+      "loss": 83.2612,
+      "step": 210
+    },
+    {
+      "epoch": 0.0008888278380878889,
+      "grad_norm": 542.4169921875,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 65.0905,
+      "step": 220
+    },
+    {
+      "epoch": 0.0009292291034555202,
+      "grad_norm": 1276.65185546875,
+      "learning_rate": 1.858585858585859e-07,
+      "loss": 95.6987,
+      "step": 230
+    },
+    {
+      "epoch": 0.0009696303688231516,
+      "grad_norm": 550.302734375,
+      "learning_rate": 1.9393939393939395e-07,
+      "loss": 118.7955,
+      "step": 240
+    },
+    {
+      "epoch": 0.0010100316341907828,
+      "grad_norm": 1489.1029052734375,
+      "learning_rate": 2.0202020202020205e-07,
+      "loss": 124.1141,
+      "step": 250
+    },
+    {
+      "epoch": 0.001050432899558414,
+      "grad_norm": 1103.9412841796875,
+      "learning_rate": 2.1010101010101013e-07,
+      "loss": 101.3522,
+      "step": 260
+    },
+    {
+      "epoch": 0.0010908341649260454,
+      "grad_norm": 1566.489013671875,
+      "learning_rate": 2.181818181818182e-07,
+      "loss": 97.4709,
+      "step": 270
+    },
+    {
+      "epoch": 0.0011312354302936767,
+      "grad_norm": 1549.5975341796875,
+      "learning_rate": 2.2626262626262628e-07,
+      "loss": 74.9938,
+      "step": 280
+    },
+    {
+      "epoch": 0.001171636695661308,
+      "grad_norm": 1458.78564453125,
+      "learning_rate": 2.343434343434344e-07,
+      "loss": 86.4949,
+      "step": 290
+    },
+    {
+      "epoch": 0.0012120379610289394,
+      "grad_norm": 767.0146484375,
+      "learning_rate": 2.4242424242424244e-07,
+      "loss": 64.3565,
+      "step": 300
+    },
+    {
+      "epoch": 0.0012524392263965707,
+      "grad_norm": 1175.2060546875,
+      "learning_rate": 2.505050505050505e-07,
+      "loss": 58.2902,
+      "step": 310
+    },
+    {
+      "epoch": 0.001292840491764202,
+      "grad_norm": 666.94580078125,
+      "learning_rate": 2.585858585858586e-07,
+      "loss": 65.0699,
+      "step": 320
+    },
+    {
+      "epoch": 0.0013332417571318333,
+      "grad_norm": 411.40283203125,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 79.5953,
+      "step": 330
+    },
+    {
+      "epoch": 0.0013736430224994647,
+      "grad_norm": 800.8609619140625,
+      "learning_rate": 2.7474747474747475e-07,
+      "loss": 113.2127,
+      "step": 340
+    },
+    {
+      "epoch": 0.001414044287867096,
+      "grad_norm": 438.9530029296875,
+      "learning_rate": 2.828282828282829e-07,
+      "loss": 77.6877,
+      "step": 350
+    },
+    {
+      "epoch": 0.0014544455532347273,
+      "grad_norm": 708.1030883789062,
+      "learning_rate": 2.9090909090909096e-07,
+      "loss": 59.7696,
+      "step": 360
+    },
+    {
+      "epoch": 0.0014948468186023586,
+      "grad_norm": 472.1529235839844,
+      "learning_rate": 2.9898989898989904e-07,
+      "loss": 94.6443,
+      "step": 370
+    },
+    {
+      "epoch": 0.00153524808396999,
+      "grad_norm": 1017.7798461914062,
+      "learning_rate": 3.070707070707071e-07,
+      "loss": 80.9425,
+      "step": 380
+    },
+    {
+      "epoch": 0.0015756493493376213,
+      "grad_norm": 874.1619262695312,
+      "learning_rate": 3.151515151515152e-07,
+      "loss": 78.2897,
+      "step": 390
+    },
+    {
+      "epoch": 0.0016160506147052526,
+      "grad_norm": 738.783203125,
+      "learning_rate": 3.2323232323232327e-07,
+      "loss": 124.2281,
+      "step": 400
+    },
+    {
+      "epoch": 0.001656451880072884,
+      "grad_norm": 1742.195556640625,
+      "learning_rate": 3.3131313131313135e-07,
+      "loss": 90.681,
+      "step": 410
+    },
+    {
+      "epoch": 0.0016968531454405152,
+      "grad_norm": 495.2870178222656,
+      "learning_rate": 3.393939393939395e-07,
+      "loss": 53.7599,
+      "step": 420
+    },
+    {
+      "epoch": 0.0017372544108081465,
+      "grad_norm": 599.09033203125,
+      "learning_rate": 3.474747474747475e-07,
+      "loss": 73.8856,
+      "step": 430
+    },
+    {
+      "epoch": 0.0017776556761757779,
+      "grad_norm": 701.4446411132812,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 104.5027,
+      "step": 440
+    },
+    {
+      "epoch": 0.0018180569415434092,
+      "grad_norm": 1392.83642578125,
+      "learning_rate": 3.6363636363636366e-07,
+      "loss": 82.1324,
+      "step": 450
+    },
+    {
+      "epoch": 0.0018584582069110405,
+      "grad_norm": 368.55596923828125,
+      "learning_rate": 3.717171717171718e-07,
+      "loss": 90.4273,
+      "step": 460
+    },
+    {
+      "epoch": 0.0018988594722786718,
+      "grad_norm": 966.1232299804688,
+      "learning_rate": 3.7979797979797987e-07,
+      "loss": 80.2726,
+      "step": 470
+    },
+    {
+      "epoch": 0.0019392607376463031,
+      "grad_norm": 727.8269653320312,
+      "learning_rate": 3.878787878787879e-07,
+      "loss": 72.9839,
+      "step": 480
+    },
+    {
+      "epoch": 0.0019796620030139342,
+      "grad_norm": 1065.9033203125,
+      "learning_rate": 3.9595959595959597e-07,
+      "loss": 76.9423,
+      "step": 490
+    },
+    {
+      "epoch": 0.0020200632683815656,
+      "grad_norm": 487.7821960449219,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 85.3323,
+      "step": 500
+    },
+    {
+      "epoch": 0.002060464533749197,
+      "grad_norm": 943.0349731445312,
+      "learning_rate": 4.121212121212122e-07,
+      "loss": 103.3431,
+      "step": 510
+    },
+    {
+      "epoch": 0.002100865799116828,
+      "grad_norm": 1148.1529541015625,
+      "learning_rate": 4.2020202020202026e-07,
+      "loss": 94.2576,
+      "step": 520
+    },
+    {
+      "epoch": 0.0021412670644844595,
+      "grad_norm": 751.3715209960938,
+      "learning_rate": 4.282828282828283e-07,
+      "loss": 74.1272,
+      "step": 530
+    },
+    {
+      "epoch": 0.002181668329852091,
+      "grad_norm": 940.9969482421875,
+      "learning_rate": 4.363636363636364e-07,
+      "loss": 57.4998,
+      "step": 540
+    },
+    {
+      "epoch": 0.002222069595219722,
+      "grad_norm": 1486.8199462890625,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 86.582,
+      "step": 550
+    },
+    {
+      "epoch": 0.0022624708605873535,
+      "grad_norm": 794.8299560546875,
+      "learning_rate": 4.5252525252525257e-07,
+      "loss": 100.8996,
+      "step": 560
+    },
+    {
+      "epoch": 0.002302872125954985,
+      "grad_norm": 475.3540344238281,
+      "learning_rate": 4.6060606060606064e-07,
+      "loss": 52.334,
+      "step": 570
+    },
+    {
+      "epoch": 0.002343273391322616,
+      "grad_norm": 1649.77392578125,
+      "learning_rate": 4.686868686868688e-07,
+      "loss": 120.0299,
+      "step": 580
+    },
+    {
+      "epoch": 0.0023836746566902474,
+      "grad_norm": 1001.852294921875,
+      "learning_rate": 4.767676767676768e-07,
+      "loss": 99.2672,
+      "step": 590
+    },
+    {
+      "epoch": 0.0024240759220578788,
+      "grad_norm": 755.6719970703125,
+      "learning_rate": 4.848484848484849e-07,
+      "loss": 60.3841,
+      "step": 600
+    },
+    {
+      "epoch": 0.00246447718742551,
+      "grad_norm": 1413.067626953125,
+      "learning_rate": 4.929292929292929e-07,
+      "loss": 99.4986,
+      "step": 610
+    },
+    {
+      "epoch": 0.0025048784527931414,
+      "grad_norm": 514.0801391601562,
+      "learning_rate": 5.01010101010101e-07,
+      "loss": 82.1478,
+      "step": 620
+    },
+    {
+      "epoch": 0.0025452797181607727,
+      "grad_norm": 446.5071105957031,
+      "learning_rate": 5.090909090909092e-07,
+      "loss": 90.5055,
+      "step": 630
+    },
+    {
+      "epoch": 0.002585680983528404,
+      "grad_norm": 1363.0565185546875,
+      "learning_rate": 5.171717171717172e-07,
+      "loss": 75.8123,
+      "step": 640
+    },
+    {
+      "epoch": 0.0026260822488960354,
+      "grad_norm": 1498.7562255859375,
+      "learning_rate": 5.252525252525253e-07,
+      "loss": 97.2051,
+      "step": 650
+    },
+    {
+      "epoch": 0.0026664835142636667,
+      "grad_norm": 819.4781494140625,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 92.2455,
+      "step": 660
+    },
+    {
+      "epoch": 0.002706884779631298,
+      "grad_norm": 739.6301879882812,
+      "learning_rate": 5.414141414141415e-07,
+      "loss": 60.5293,
+      "step": 670
+    },
+    {
+      "epoch": 0.0027472860449989293,
+      "grad_norm": 610.8301391601562,
+      "learning_rate": 5.494949494949495e-07,
+      "loss": 76.4783,
+      "step": 680
+    },
+    {
+      "epoch": 0.0027876873103665606,
+      "grad_norm": 796.4779663085938,
+      "learning_rate": 5.575757575757576e-07,
+      "loss": 76.7366,
+      "step": 690
+    },
+    {
+      "epoch": 0.002828088575734192,
+      "grad_norm": 372.6942443847656,
+      "learning_rate": 5.656565656565658e-07,
+      "loss": 55.3669,
+      "step": 700
+    },
+    {
+      "epoch": 0.0028684898411018233,
+      "grad_norm": 887.7366333007812,
+      "learning_rate": 5.737373737373738e-07,
+      "loss": 91.2889,
+      "step": 710
+    },
+    {
+      "epoch": 0.0029088911064694546,
+      "grad_norm": 869.73486328125,
+      "learning_rate": 5.818181818181819e-07,
+      "loss": 90.853,
+      "step": 720
+    },
+    {
+      "epoch": 0.002949292371837086,
+      "grad_norm": 900.1129760742188,
+      "learning_rate": 5.898989898989899e-07,
+      "loss": 99.8146,
+      "step": 730
+    },
+    {
+      "epoch": 0.0029896936372047172,
+      "grad_norm": 882.5169067382812,
+      "learning_rate": 5.979797979797981e-07,
+      "loss": 94.4538,
+      "step": 740
+    },
+    {
+      "epoch": 0.0030300949025723486,
+      "grad_norm": 599.550537109375,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 66.0833,
+      "step": 750
+    },
+    {
+      "epoch": 0.00307049616793998,
+      "grad_norm": 513.547607421875,
+      "learning_rate": 6.141414141414142e-07,
+      "loss": 51.7566,
+      "step": 760
+    },
+    {
+      "epoch": 0.003110897433307611,
+      "grad_norm": 645.5187377929688,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 68.8667,
+      "step": 770
+    },
+    {
+      "epoch": 0.0031512986986752425,
+      "grad_norm": 473.2290344238281,
+      "learning_rate": 6.303030303030304e-07,
+      "loss": 58.1628,
+      "step": 780
+    },
+    {
+      "epoch": 0.003191699964042874,
+      "grad_norm": 912.64404296875,
+      "learning_rate": 6.383838383838384e-07,
+      "loss": 90.716,
+      "step": 790
+    },
+    {
+      "epoch": 0.003232101229410505,
+      "grad_norm": 647.8580322265625,
+      "learning_rate": 6.464646464646465e-07,
+      "loss": 55.9714,
+      "step": 800
+    },
+    {
+      "epoch": 0.0032725024947781365,
+      "grad_norm": 768.3733520507812,
+      "learning_rate": 6.545454545454547e-07,
+      "loss": 83.1954,
+      "step": 810
+    },
+    {
+      "epoch": 0.003312903760145768,
+      "grad_norm": 482.00775146484375,
+      "learning_rate": 6.626262626262627e-07,
+      "loss": 63.0688,
+      "step": 820
+    },
+    {
+      "epoch": 0.003353305025513399,
+      "grad_norm": 411.4869079589844,
+      "learning_rate": 6.707070707070708e-07,
+      "loss": 58.3106,
+      "step": 830
+    },
+    {
+      "epoch": 0.0033937062908810304,
+      "grad_norm": 590.60546875,
+      "learning_rate": 6.78787878787879e-07,
+      "loss": 81.1757,
+      "step": 840
+    },
+    {
+      "epoch": 0.0034341075562486618,
+      "grad_norm": 267.18963623046875,
+      "learning_rate": 6.868686868686869e-07,
+      "loss": 74.0499,
+      "step": 850
+    },
+    {
+      "epoch": 0.003474508821616293,
+      "grad_norm": 553.4140625,
+      "learning_rate": 6.94949494949495e-07,
+      "loss": 93.1003,
+      "step": 860
+    },
+    {
+      "epoch": 0.0035149100869839244,
+      "grad_norm": 624.134033203125,
+      "learning_rate": 7.03030303030303e-07,
+      "loss": 53.7593,
+      "step": 870
+    },
+    {
+      "epoch": 0.0035553113523515557,
+      "grad_norm": 529.945556640625,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 48.8026,
+      "step": 880
+    },
+    {
+      "epoch": 0.003595712617719187,
+      "grad_norm": 564.1666870117188,
+      "learning_rate": 7.191919191919193e-07,
+      "loss": 76.9729,
+      "step": 890
+    },
+    {
+      "epoch": 0.0036361138830868184,
+      "grad_norm": 433.373779296875,
+      "learning_rate": 7.272727272727273e-07,
+      "loss": 64.9687,
+      "step": 900
+    },
+    {
+      "epoch": 0.0036765151484544497,
+      "grad_norm": 671.010986328125,
+      "learning_rate": 7.353535353535354e-07,
+      "loss": 86.2554,
+      "step": 910
+    },
+    {
+      "epoch": 0.003716916413822081,
+      "grad_norm": 481.1733093261719,
+      "learning_rate": 7.434343434343436e-07,
+      "loss": 54.4597,
+      "step": 920
+    },
+    {
+      "epoch": 0.0037573176791897123,
+      "grad_norm": 643.103271484375,
+      "learning_rate": 7.515151515151516e-07,
+      "loss": 64.1825,
+      "step": 930
+    },
+    {
+      "epoch": 0.0037977189445573436,
+      "grad_norm": 732.6908569335938,
+      "learning_rate": 7.595959595959597e-07,
+      "loss": 67.7345,
+      "step": 940
+    },
+    {
+      "epoch": 0.003838120209924975,
+      "grad_norm": 401.0261535644531,
+      "learning_rate": 7.676767676767677e-07,
+      "loss": 61.4696,
+      "step": 950
+    },
+    {
+      "epoch": 0.0038785214752926063,
+      "grad_norm": 491.2078552246094,
+      "learning_rate": 7.757575757575758e-07,
+      "loss": 54.7679,
+      "step": 960
+    },
+    {
+      "epoch": 0.003918922740660237,
+      "grad_norm": 496.06292724609375,
+      "learning_rate": 7.838383838383839e-07,
+      "loss": 74.1232,
+      "step": 970
+    },
+    {
+      "epoch": 0.0039593240060278685,
+      "grad_norm": 612.330810546875,
+      "learning_rate": 7.919191919191919e-07,
+      "loss": 48.3253,
+      "step": 980
+    },
+    {
+      "epoch": 0.0039997252713955,
+      "grad_norm": 961.5505981445312,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 97.4158,
+      "step": 990
+    },
+    {
+      "epoch": 0.004040126536763131,
+      "grad_norm": 458.24176025390625,
+      "learning_rate": 8.080808080808082e-07,
+      "loss": 73.8477,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0040805278021307624,
+      "grad_norm": 902.3980712890625,
+      "learning_rate": 8.161616161616162e-07,
+      "loss": 84.1415,
+      "step": 1010
+    },
+    {
+      "epoch": 0.004120929067498394,
+      "grad_norm": 1041.8564453125,
+      "learning_rate": 8.242424242424244e-07,
+      "loss": 91.096,
+      "step": 1020
+    },
+    {
+      "epoch": 0.004161330332866025,
+      "grad_norm": 483.5507507324219,
+      "learning_rate": 8.323232323232324e-07,
+      "loss": 71.8416,
+      "step": 1030
+    },
+    {
+      "epoch": 0.004201731598233656,
+      "grad_norm": 461.9169006347656,
+      "learning_rate": 8.404040404040405e-07,
+      "loss": 109.2555,
+      "step": 1040
+    },
+    {
+      "epoch": 0.004242132863601288,
+      "grad_norm": 716.3146362304688,
+      "learning_rate": 8.484848484848486e-07,
+      "loss": 67.7536,
+      "step": 1050
+    },
+    {
+      "epoch": 0.004282534128968919,
+      "grad_norm": 583.708251953125,
+      "learning_rate": 8.565656565656566e-07,
+      "loss": 79.6363,
+      "step": 1060
+    },
+    {
+      "epoch": 0.00432293539433655,
+      "grad_norm": 644.1818237304688,
+      "learning_rate": 8.646464646464647e-07,
+      "loss": 86.4302,
+      "step": 1070
+    },
+    {
+      "epoch": 0.004363336659704182,
+      "grad_norm": 549.10400390625,
+      "learning_rate": 8.727272727272728e-07,
+      "loss": 48.8068,
+      "step": 1080
+    },
+    {
+      "epoch": 0.004403737925071813,
+      "grad_norm": 651.5703735351562,
+      "learning_rate": 8.808080808080808e-07,
+      "loss": 62.0918,
+      "step": 1090
+    },
+    {
+      "epoch": 0.004444139190439444,
+      "grad_norm": 553.933837890625,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 51.3118,
+      "step": 1100
+    },
+    {
+      "epoch": 0.004484540455807076,
+      "grad_norm": 559.3411254882812,
+      "learning_rate": 8.96969696969697e-07,
+      "loss": 78.899,
+      "step": 1110
+    },
+    {
+      "epoch": 0.004524941721174707,
+      "grad_norm": 1178.9420166015625,
+      "learning_rate": 9.050505050505051e-07,
+      "loss": 59.7121,
+      "step": 1120
+    },
+    {
+      "epoch": 0.004565342986542338,
+      "grad_norm": 470.557373046875,
+      "learning_rate": 9.131313131313133e-07,
+      "loss": 77.5851,
+      "step": 1130
+    },
+    {
+      "epoch": 0.00460574425190997,
+      "grad_norm": 823.9404907226562,
+      "learning_rate": 9.212121212121213e-07,
+      "loss": 88.5502,
+      "step": 1140
+    },
+    {
+      "epoch": 0.004646145517277601,
+      "grad_norm": 858.840087890625,
+      "learning_rate": 9.292929292929294e-07,
+      "loss": 103.8964,
+      "step": 1150
+    },
+    {
+      "epoch": 0.004686546782645232,
+      "grad_norm": 914.1184692382812,
+      "learning_rate": 9.373737373737376e-07,
+      "loss": 77.9602,
+      "step": 1160
+    },
+    {
+      "epoch": 0.0047269480480128636,
+      "grad_norm": 686.4456787109375,
+      "learning_rate": 9.454545454545455e-07,
+      "loss": 76.4946,
+      "step": 1170
+    },
+    {
+      "epoch": 0.004767349313380495,
+      "grad_norm": 487.4496765136719,
+      "learning_rate": 9.535353535353536e-07,
+      "loss": 99.8757,
+      "step": 1180
+    },
+    {
+      "epoch": 0.004807750578748126,
+      "grad_norm": 520.9862670898438,
+      "learning_rate": 9.616161616161617e-07,
+      "loss": 93.658,
+      "step": 1190
+    },
+    {
+      "epoch": 0.0048481518441157575,
+      "grad_norm": 390.9288635253906,
+      "learning_rate": 9.696969696969698e-07,
+      "loss": 87.906,
+      "step": 1200
+    },
+    {
+      "epoch": 0.004888553109483389,
+      "grad_norm": 436.38543701171875,
+      "learning_rate": 9.77777777777778e-07,
+      "loss": 47.0411,
+      "step": 1210
+    },
+    {
+      "epoch": 0.00492895437485102,
+      "grad_norm": 671.6222534179688,
+      "learning_rate": 9.858585858585858e-07,
+      "loss": 51.2458,
+      "step": 1220
+    },
+    {
+      "epoch": 0.0049693556402186515,
+      "grad_norm": 691.9678344726562,
+      "learning_rate": 9.93939393939394e-07,
+      "loss": 54.2135,
+      "step": 1230
+    },
+    {
+      "epoch": 0.005009756905586283,
+      "grad_norm": 849.5625610351562,
+      "learning_rate": 1.002020202020202e-06,
+      "loss": 85.6906,
+      "step": 1240
+    },
+    {
+      "epoch": 0.005050158170953914,
+      "grad_norm": 790.7261352539062,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 118.3505,
+      "step": 1250
+    },
+    {
+      "epoch": 0.0050905594363215454,
+      "grad_norm": 838.802734375,
+      "learning_rate": 1.0181818181818183e-06,
+      "loss": 66.9146,
+      "step": 1260
+    },
+    {
+      "epoch": 0.005130960701689177,
+      "grad_norm": 871.91162109375,
+      "learning_rate": 1.0262626262626264e-06,
+      "loss": 94.0071,
+      "step": 1270
+    },
+    {
+      "epoch": 0.005171361967056808,
+      "grad_norm": 321.37939453125,
+      "learning_rate": 1.0343434343434344e-06,
+      "loss": 74.266,
+      "step": 1280
+    },
+    {
+      "epoch": 0.005211763232424439,
+      "grad_norm": 1083.537353515625,
+      "learning_rate": 1.0424242424242426e-06,
+      "loss": 84.3127,
+      "step": 1290
+    },
+    {
+      "epoch": 0.005252164497792071,
+      "grad_norm": 755.6991577148438,
+      "learning_rate": 1.0505050505050506e-06,
+      "loss": 67.5171,
+      "step": 1300
+    },
+    {
+      "epoch": 0.005292565763159702,
+      "grad_norm": 496.10772705078125,
+      "learning_rate": 1.0585858585858587e-06,
+      "loss": 59.7423,
+      "step": 1310
+    },
+    {
+      "epoch": 0.005332967028527333,
+      "grad_norm": 840.3679809570312,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 69.084,
+      "step": 1320
+    },
+    {
+      "epoch": 0.005373368293894965,
+      "grad_norm": 913.7398681640625,
+      "learning_rate": 1.0747474747474747e-06,
+      "loss": 54.4863,
+      "step": 1330
+    },
+    {
+      "epoch": 0.005413769559262596,
+      "grad_norm": 1077.27392578125,
+      "learning_rate": 1.082828282828283e-06,
+      "loss": 90.3723,
+      "step": 1340
+    },
+    {
+      "epoch": 0.005454170824630227,
+      "grad_norm": 587.8721313476562,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 61.4766,
+      "step": 1350
+    },
+    {
+      "epoch": 0.005494572089997859,
+      "grad_norm": 651.0157470703125,
+      "learning_rate": 1.098989898989899e-06,
+      "loss": 58.8504,
+      "step": 1360
+    },
+    {
+      "epoch": 0.00553497335536549,
+      "grad_norm": 1200.4676513671875,
+      "learning_rate": 1.1070707070707072e-06,
+      "loss": 78.9724,
+      "step": 1370
+    },
+    {
+      "epoch": 0.005575374620733121,
+      "grad_norm": 1426.4935302734375,
+      "learning_rate": 1.1151515151515153e-06,
+      "loss": 97.0092,
+      "step": 1380
+    },
+    {
+      "epoch": 0.005615775886100753,
+      "grad_norm": 668.6116333007812,
+      "learning_rate": 1.1232323232323233e-06,
+      "loss": 78.3742,
+      "step": 1390
+    },
+    {
+      "epoch": 0.005656177151468384,
+      "grad_norm": 213.35760498046875,
+      "learning_rate": 1.1313131313131315e-06,
+      "loss": 101.9843,
+      "step": 1400
+    },
+    {
+      "epoch": 0.005696578416836015,
+      "grad_norm": 407.4592590332031,
+      "learning_rate": 1.1393939393939395e-06,
+      "loss": 67.6436,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0057369796822036466,
+      "grad_norm": 1257.8907470703125,
+      "learning_rate": 1.1474747474747476e-06,
+      "loss": 85.1267,
+      "step": 1420
+    },
+    {
+      "epoch": 0.005777380947571278,
+      "grad_norm": 671.1593017578125,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 71.1813,
+      "step": 1430
+    },
+    {
+      "epoch": 0.005817782212938909,
+      "grad_norm": 592.0625610351562,
+      "learning_rate": 1.1636363636363638e-06,
+      "loss": 70.4293,
+      "step": 1440
+    },
+    {
+      "epoch": 0.0058581834783065405,
+      "grad_norm": 425.52978515625,
+      "learning_rate": 1.1717171717171719e-06,
+      "loss": 60.6906,
+      "step": 1450
+    },
+    {
+      "epoch": 0.005898584743674172,
+      "grad_norm": 1379.921630859375,
+      "learning_rate": 1.1797979797979799e-06,
+      "loss": 69.5714,
+      "step": 1460
+    },
+    {
+      "epoch": 0.005938986009041803,
+      "grad_norm": 246.26141357421875,
+      "learning_rate": 1.187878787878788e-06,
+      "loss": 64.0959,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0059793872744094345,
+      "grad_norm": 399.0498046875,
+      "learning_rate": 1.1959595959595961e-06,
+      "loss": 58.6223,
+      "step": 1480
+    },
+    {
+      "epoch": 0.006019788539777066,
+      "grad_norm": 723.107421875,
+      "learning_rate": 1.2040404040404042e-06,
+      "loss": 99.6874,
+      "step": 1490
+    },
+    {
+      "epoch": 0.006060189805144697,
+      "grad_norm": 556.9105224609375,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 48.2795,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0061005910705123284,
+      "grad_norm": 1261.1217041015625,
+      "learning_rate": 1.2202020202020202e-06,
+      "loss": 63.467,
+      "step": 1510
+    },
+    {
+      "epoch": 0.00614099233587996,
+      "grad_norm": 832.5606689453125,
+      "learning_rate": 1.2282828282828285e-06,
+      "loss": 56.5807,
+      "step": 1520
+    },
+    {
+      "epoch": 0.006181393601247591,
+      "grad_norm": 561.32275390625,
+      "learning_rate": 1.2363636363636365e-06,
+      "loss": 50.7661,
+      "step": 1530
+    },
+    {
+      "epoch": 0.006221794866615222,
+      "grad_norm": 607.7224731445312,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 68.9423,
+      "step": 1540
+    },
+    {
+      "epoch": 0.006262196131982854,
+      "grad_norm": 728.4628295898438,
+      "learning_rate": 1.2525252525252527e-06,
+      "loss": 70.7681,
+      "step": 1550
+    },
+    {
+      "epoch": 0.006302597397350485,
+      "grad_norm": 666.349853515625,
+      "learning_rate": 1.2606060606060608e-06,
+      "loss": 56.895,
+      "step": 1560
+    },
+    {
+      "epoch": 0.006342998662718116,
+      "grad_norm": 1740.4351806640625,
+      "learning_rate": 1.268686868686869e-06,
+      "loss": 60.3912,
+      "step": 1570
+    },
+    {
+      "epoch": 0.006383399928085748,
+      "grad_norm": 1566.6124267578125,
+      "learning_rate": 1.2767676767676768e-06,
+      "loss": 93.7447,
+      "step": 1580
+    },
+    {
+      "epoch": 0.006423801193453379,
+      "grad_norm": 462.4256591796875,
+      "learning_rate": 1.2848484848484848e-06,
+      "loss": 73.4026,
+      "step": 1590
+    },
+    {
+      "epoch": 0.00646420245882101,
+      "grad_norm": 1016.7862548828125,
+      "learning_rate": 1.292929292929293e-06,
+      "loss": 76.6085,
+      "step": 1600
+    },
+    {
+      "epoch": 0.006504603724188642,
+      "grad_norm": 657.1649780273438,
+      "learning_rate": 1.301010101010101e-06,
+      "loss": 75.5045,
+      "step": 1610
+    },
+    {
+      "epoch": 0.006545004989556273,
+      "grad_norm": 958.7669067382812,
+      "learning_rate": 1.3090909090909093e-06,
+      "loss": 65.2289,
+      "step": 1620
+    },
+    {
+      "epoch": 0.006585406254923904,
+      "grad_norm": 542.748779296875,
+      "learning_rate": 1.3171717171717172e-06,
+      "loss": 87.3519,
+      "step": 1630
+    },
+    {
+      "epoch": 0.006625807520291536,
+      "grad_norm": 719.8927612304688,
+      "learning_rate": 1.3252525252525254e-06,
+      "loss": 75.2894,
+      "step": 1640
+    },
+    {
+      "epoch": 0.006666208785659167,
+      "grad_norm": 909.3445434570312,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 82.4471,
+      "step": 1650
+    },
+    {
+      "epoch": 0.006706610051026798,
+      "grad_norm": 589.6493530273438,
+      "learning_rate": 1.3414141414141417e-06,
+      "loss": 40.8706,
+      "step": 1660
+    },
+    {
+      "epoch": 0.0067470113163944296,
+      "grad_norm": 409.3217468261719,
+      "learning_rate": 1.3494949494949497e-06,
+      "loss": 60.2111,
+      "step": 1670
+    },
+    {
+      "epoch": 0.006787412581762061,
+      "grad_norm": 750.4218139648438,
+      "learning_rate": 1.357575757575758e-06,
+      "loss": 53.4837,
+      "step": 1680
+    },
+    {
+      "epoch": 0.006827813847129692,
+      "grad_norm": 807.2741088867188,
+      "learning_rate": 1.3656565656565657e-06,
+      "loss": 65.8104,
+      "step": 1690
+    },
+    {
+      "epoch": 0.0068682151124973235,
+      "grad_norm": 909.712890625,
+      "learning_rate": 1.3737373737373738e-06,
+      "loss": 54.6645,
+      "step": 1700
+    },
+    {
+      "epoch": 0.006908616377864955,
+      "grad_norm": 736.7755126953125,
+      "learning_rate": 1.381818181818182e-06,
+      "loss": 74.1443,
+      "step": 1710
+    },
+    {
+      "epoch": 0.006949017643232586,
+      "grad_norm": 1265.934814453125,
+      "learning_rate": 1.38989898989899e-06,
+      "loss": 64.3544,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0069894189086002175,
+      "grad_norm": 587.2948608398438,
+      "learning_rate": 1.3979797979797982e-06,
+      "loss": 51.3367,
+      "step": 1730
+    },
+    {
+      "epoch": 0.007029820173967849,
+      "grad_norm": 580.3670043945312,
+      "learning_rate": 1.406060606060606e-06,
+      "loss": 79.2176,
+      "step": 1740
+    },
+    {
+      "epoch": 0.00707022143933548,
+      "grad_norm": 727.06298828125,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 85.4666,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0071106227047031114,
+      "grad_norm": 370.42242431640625,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 76.675,
+      "step": 1760
+    },
+    {
+      "epoch": 0.007151023970070743,
+      "grad_norm": 316.96478271484375,
+      "learning_rate": 1.4303030303030306e-06,
+      "loss": 64.5112,
+      "step": 1770
+    },
+    {
+      "epoch": 0.007191425235438374,
+      "grad_norm": 630.2029418945312,
+      "learning_rate": 1.4383838383838386e-06,
+      "loss": 86.1591,
+      "step": 1780
+    },
+    {
+      "epoch": 0.007231826500806005,
+      "grad_norm": 313.6539611816406,
+      "learning_rate": 1.4464646464646464e-06,
+      "loss": 43.7239,
+      "step": 1790
+    },
+    {
+      "epoch": 0.007272227766173637,
+      "grad_norm": 443.1033630371094,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 45.1063,
+      "step": 1800
+    },
+    {
+      "epoch": 0.007312629031541268,
+      "grad_norm": 435.49420166015625,
+      "learning_rate": 1.4626262626262627e-06,
+      "loss": 83.8737,
+      "step": 1810
+    },
+    {
+      "epoch": 0.007353030296908899,
+      "grad_norm": 600.2357177734375,
+      "learning_rate": 1.470707070707071e-06,
+      "loss": 92.0006,
+      "step": 1820
+    },
+    {
+      "epoch": 0.007393431562276531,
+      "grad_norm": 720.475341796875,
+      "learning_rate": 1.478787878787879e-06,
+      "loss": 83.276,
+      "step": 1830
+    },
+    {
+      "epoch": 0.007433832827644162,
+      "grad_norm": 837.3829956054688,
+      "learning_rate": 1.4868686868686872e-06,
+      "loss": 63.679,
+      "step": 1840
+    },
+    {
+      "epoch": 0.007474234093011793,
+      "grad_norm": 749.28564453125,
+      "learning_rate": 1.494949494949495e-06,
+      "loss": 66.9486,
+      "step": 1850
+    },
+    {
+      "epoch": 0.007514635358379425,
+      "grad_norm": 447.3569030761719,
+      "learning_rate": 1.5030303030303032e-06,
+      "loss": 75.0599,
+      "step": 1860
+    },
+    {
+      "epoch": 0.007555036623747056,
+      "grad_norm": 687.4884033203125,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 93.3957,
+      "step": 1870
+    },
+    {
+      "epoch": 0.007595437889114687,
+      "grad_norm": 926.2616577148438,
+      "learning_rate": 1.5191919191919195e-06,
+      "loss": 54.0039,
+      "step": 1880
+    },
+    {
+      "epoch": 0.007635839154482319,
+      "grad_norm": 1626.1761474609375,
+      "learning_rate": 1.5272727272727275e-06,
+      "loss": 78.25,
+      "step": 1890
+    },
+    {
+      "epoch": 0.00767624041984995,
+      "grad_norm": 894.1566772460938,
+      "learning_rate": 1.5353535353535353e-06,
+      "loss": 72.2371,
+      "step": 1900
+    },
+    {
+      "epoch": 0.007716641685217581,
+      "grad_norm": 1152.4473876953125,
+      "learning_rate": 1.5434343434343435e-06,
+      "loss": 63.9651,
+      "step": 1910
+    },
+    {
+      "epoch": 0.0077570429505852126,
+      "grad_norm": 378.2879333496094,
+      "learning_rate": 1.5515151515151516e-06,
+      "loss": 69.7281,
+      "step": 1920
+    },
+    {
+      "epoch": 0.007797444215952844,
+      "grad_norm": 668.77294921875,
+      "learning_rate": 1.5595959595959598e-06,
+      "loss": 75.3944,
+      "step": 1930
+    },
+    {
+      "epoch": 0.007837845481320474,
+      "grad_norm": 1068.0228271484375,
+      "learning_rate": 1.5676767676767678e-06,
+      "loss": 70.1825,
+      "step": 1940
+    },
+    {
+      "epoch": 0.007878246746688106,
+      "grad_norm": 1168.907958984375,
+      "learning_rate": 1.5757575757575759e-06,
+      "loss": 76.5451,
+      "step": 1950
+    },
+    {
+      "epoch": 0.007918648012055737,
+      "grad_norm": 556.515625,
+      "learning_rate": 1.5838383838383839e-06,
+      "loss": 75.1146,
+      "step": 1960
+    },
+    {
+      "epoch": 0.007959049277423368,
+      "grad_norm": 4213.0146484375,
+      "learning_rate": 1.5919191919191921e-06,
+      "loss": 86.7434,
+      "step": 1970
+    },
+    {
+      "epoch": 0.007999450542791,
+      "grad_norm": 190.214111328125,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 87.13,
+      "step": 1980
+    },
+    {
+      "epoch": 0.008039851808158631,
+      "grad_norm": 732.1063232421875,
+      "learning_rate": 1.6080808080808084e-06,
+      "loss": 62.2306,
+      "step": 1990
+    },
+    {
+      "epoch": 0.008080253073526262,
+      "grad_norm": 601.2501220703125,
+      "learning_rate": 1.6161616161616164e-06,
+      "loss": 50.099,
+      "step": 2000
+    },
+    {
+      "epoch": 0.008120654338893894,
+      "grad_norm": 982.2610473632812,
+      "learning_rate": 1.6242424242424242e-06,
+      "loss": 83.7663,
+      "step": 2010
+    },
+    {
+      "epoch": 0.008161055604261525,
+      "grad_norm": 738.6604614257812,
+      "learning_rate": 1.6323232323232325e-06,
+      "loss": 79.6428,
+      "step": 2020
+    },
+    {
+      "epoch": 0.008201456869629156,
+      "grad_norm": 517.3074951171875,
+      "learning_rate": 1.6404040404040405e-06,
+      "loss": 65.7427,
+      "step": 2030
+    },
+    {
+      "epoch": 0.008241858134996788,
+      "grad_norm": 908.445556640625,
+      "learning_rate": 1.6484848484848487e-06,
+      "loss": 78.0444,
+      "step": 2040
+    },
+    {
+      "epoch": 0.008282259400364419,
+      "grad_norm": 1011.9039306640625,
+      "learning_rate": 1.6565656565656567e-06,
+      "loss": 51.7269,
+      "step": 2050
+    },
+    {
+      "epoch": 0.00832266066573205,
+      "grad_norm": 917.6417236328125,
+      "learning_rate": 1.6646464646464648e-06,
+      "loss": 86.5933,
+      "step": 2060
+    },
+    {
+      "epoch": 0.008363061931099681,
+      "grad_norm": 795.33544921875,
+      "learning_rate": 1.6727272727272728e-06,
+      "loss": 108.7873,
+      "step": 2070
+    },
+    {
+      "epoch": 0.008403463196467313,
+      "grad_norm": 554.3195190429688,
+      "learning_rate": 1.680808080808081e-06,
+      "loss": 60.6646,
+      "step": 2080
+    },
+    {
+      "epoch": 0.008443864461834944,
+      "grad_norm": 674.0402221679688,
+      "learning_rate": 1.688888888888889e-06,
+      "loss": 61.3736,
+      "step": 2090
+    },
+    {
+      "epoch": 0.008484265727202575,
+      "grad_norm": 620.1227416992188,
+      "learning_rate": 1.6969696969696973e-06,
+      "loss": 46.0749,
+      "step": 2100
+    },
+    {
+      "epoch": 0.008524666992570207,
+      "grad_norm": 664.8109741210938,
+      "learning_rate": 1.705050505050505e-06,
+      "loss": 47.4902,
+      "step": 2110
+    },
+    {
+      "epoch": 0.008565068257937838,
+      "grad_norm": 698.6871337890625,
+      "learning_rate": 1.7131313131313131e-06,
+      "loss": 73.7654,
+      "step": 2120
+    },
+    {
+      "epoch": 0.00860546952330547,
+      "grad_norm": 557.4563598632812,
+      "learning_rate": 1.7212121212121214e-06,
+      "loss": 74.8754,
+      "step": 2130
+    },
+    {
+      "epoch": 0.0086458707886731,
+      "grad_norm": 951.68701171875,
+      "learning_rate": 1.7292929292929294e-06,
+      "loss": 75.8206,
+      "step": 2140
+    },
+    {
+      "epoch": 0.008686272054040732,
+      "grad_norm": 607.4713134765625,
+      "learning_rate": 1.7373737373737376e-06,
+      "loss": 65.7856,
+      "step": 2150
+    },
+    {
+      "epoch": 0.008726673319408363,
+      "grad_norm": 561.9832763671875,
+      "learning_rate": 1.7454545454545456e-06,
+      "loss": 71.6938,
+      "step": 2160
+    },
+    {
+      "epoch": 0.008767074584775995,
+      "grad_norm": 576.9796142578125,
+      "learning_rate": 1.7535353535353537e-06,
+      "loss": 69.7876,
+      "step": 2170
+    },
+    {
+      "epoch": 0.008807475850143626,
+      "grad_norm": 774.2403564453125,
+      "learning_rate": 1.7616161616161617e-06,
+      "loss": 70.9791,
+      "step": 2180
+    },
+    {
+      "epoch": 0.008847877115511257,
+      "grad_norm": 826.15625,
+      "learning_rate": 1.76969696969697e-06,
+      "loss": 46.3716,
+      "step": 2190
+    },
+    {
+      "epoch": 0.008888278380878889,
+      "grad_norm": 614.9705200195312,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 55.0711,
+      "step": 2200
+    },
+    {
+      "epoch": 0.00892867964624652,
+      "grad_norm": 385.35986328125,
+      "learning_rate": 1.7858585858585862e-06,
+      "loss": 55.8608,
+      "step": 2210
+    },
+    {
+      "epoch": 0.008969080911614151,
+      "grad_norm": 900.0477905273438,
+      "learning_rate": 1.793939393939394e-06,
+      "loss": 63.3841,
+      "step": 2220
+    },
+    {
+      "epoch": 0.009009482176981783,
+      "grad_norm": 484.96966552734375,
+      "learning_rate": 1.802020202020202e-06,
+      "loss": 80.3578,
+      "step": 2230
+    },
+    {
+      "epoch": 0.009049883442349414,
+      "grad_norm": 598.5676879882812,
+      "learning_rate": 1.8101010101010103e-06,
+      "loss": 72.6313,
+      "step": 2240
+    },
+    {
+      "epoch": 0.009090284707717045,
+      "grad_norm": 540.750244140625,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 57.7766,
+      "step": 2250
+    },
+    {
+      "epoch": 0.009130685973084677,
+      "grad_norm": 1184.3426513671875,
+      "learning_rate": 1.8262626262626265e-06,
+      "loss": 48.5757,
+      "step": 2260
+    },
+    {
+      "epoch": 0.009171087238452308,
+      "grad_norm": 752.9002075195312,
+      "learning_rate": 1.8343434343434343e-06,
+      "loss": 97.9035,
+      "step": 2270
+    },
+    {
+      "epoch": 0.00921148850381994,
+      "grad_norm": 420.1020202636719,
+      "learning_rate": 1.8424242424242426e-06,
+      "loss": 75.47,
+      "step": 2280
+    },
+    {
+      "epoch": 0.00925188976918757,
+      "grad_norm": 605.6198120117188,
+      "learning_rate": 1.8505050505050506e-06,
+      "loss": 54.201,
+      "step": 2290
+    },
+    {
+      "epoch": 0.009292291034555202,
+      "grad_norm": 908.86376953125,
+      "learning_rate": 1.8585858585858588e-06,
+      "loss": 42.1971,
+      "step": 2300
+    },
+    {
+      "epoch": 0.009332692299922833,
+      "grad_norm": 719.4215087890625,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 75.1888,
+      "step": 2310
+    },
+    {
+      "epoch": 0.009373093565290464,
+      "grad_norm": 834.7787475585938,
+      "learning_rate": 1.874747474747475e-06,
+      "loss": 69.2066,
+      "step": 2320
+    },
+    {
+      "epoch": 0.009413494830658096,
+      "grad_norm": 1182.236572265625,
+      "learning_rate": 1.882828282828283e-06,
+      "loss": 58.0279,
+      "step": 2330
+    },
+    {
+      "epoch": 0.009453896096025727,
+      "grad_norm": 804.4097900390625,
+      "learning_rate": 1.890909090909091e-06,
+      "loss": 65.1326,
+      "step": 2340
+    },
+    {
+      "epoch": 0.009494297361393358,
+      "grad_norm": 503.9187316894531,
+      "learning_rate": 1.8989898989898992e-06,
+      "loss": 90.3938,
+      "step": 2350
+    },
+    {
+      "epoch": 0.00953469862676099,
+      "grad_norm": 667.1215209960938,
+      "learning_rate": 1.9070707070707072e-06,
+      "loss": 92.9535,
+      "step": 2360
+    },
+    {
+      "epoch": 0.009575099892128621,
+      "grad_norm": 962.7706909179688,
+      "learning_rate": 1.9151515151515154e-06,
+      "loss": 62.9176,
+      "step": 2370
+    },
+    {
+      "epoch": 0.009615501157496252,
+      "grad_norm": 644.2401733398438,
+      "learning_rate": 1.9232323232323235e-06,
+      "loss": 52.7077,
+      "step": 2380
+    },
+    {
+      "epoch": 0.009655902422863884,
+      "grad_norm": 651.3950805664062,
+      "learning_rate": 1.9313131313131315e-06,
+      "loss": 65.4711,
+      "step": 2390
+    },
+    {
+      "epoch": 0.009696303688231515,
+      "grad_norm": 240.46392822265625,
+      "learning_rate": 1.9393939393939395e-06,
+      "loss": 63.148,
+      "step": 2400
+    },
+    {
+      "epoch": 0.009736704953599146,
+      "grad_norm": 590.9519653320312,
+      "learning_rate": 1.9474747474747475e-06,
+      "loss": 46.7516,
+      "step": 2410
+    },
+    {
+      "epoch": 0.009777106218966778,
+      "grad_norm": 698.1563110351562,
+      "learning_rate": 1.955555555555556e-06,
+      "loss": 52.9271,
+      "step": 2420
+    },
+    {
+      "epoch": 0.009817507484334409,
+      "grad_norm": 657.922119140625,
+      "learning_rate": 1.9636363636363636e-06,
+      "loss": 70.142,
+      "step": 2430
+    },
+    {
+      "epoch": 0.00985790874970204,
+      "grad_norm": 598.1620483398438,
+      "learning_rate": 1.9717171717171716e-06,
+      "loss": 47.5373,
+      "step": 2440
+    },
+    {
+      "epoch": 0.009898310015069672,
+      "grad_norm": 277.993408203125,
+      "learning_rate": 1.97979797979798e-06,
+      "loss": 51.4112,
+      "step": 2450
+    },
+    {
+      "epoch": 0.009938711280437303,
+      "grad_norm": 552.108642578125,
+      "learning_rate": 1.987878787878788e-06,
+      "loss": 34.7859,
+      "step": 2460
+    },
+    {
+      "epoch": 0.009979112545804934,
+      "grad_norm": 859.3973999023438,
+      "learning_rate": 1.995959595959596e-06,
+      "loss": 72.8732,
+      "step": 2470
+    },
+    {
+      "epoch": 0.010019513811172566,
+      "grad_norm": 838.69140625,
+      "learning_rate": 2.004040404040404e-06,
+      "loss": 57.4993,
+      "step": 2480
+    },
+    {
+      "epoch": 0.010059915076540197,
+      "grad_norm": 505.9194030761719,
+      "learning_rate": 2.012121212121212e-06,
+      "loss": 85.4388,
+      "step": 2490
+    },
+    {
+      "epoch": 0.010100316341907828,
+      "grad_norm": 369.4288330078125,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 32.7626,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01014071760727546,
+      "grad_norm": 877.2677612304688,
+      "learning_rate": 2.0282828282828286e-06,
+      "loss": 60.406,
+      "step": 2510
+    },
+    {
+      "epoch": 0.010181118872643091,
+      "grad_norm": 718.6683959960938,
+      "learning_rate": 2.0363636363636367e-06,
+      "loss": 58.8499,
+      "step": 2520
+    },
+    {
+      "epoch": 0.010221520138010722,
+      "grad_norm": 620.8602905273438,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 40.9196,
+      "step": 2530
+    },
+    {
+      "epoch": 0.010261921403378354,
+      "grad_norm": 448.3660583496094,
+      "learning_rate": 2.0525252525252527e-06,
+      "loss": 62.7339,
+      "step": 2540
+    },
+    {
+      "epoch": 0.010302322668745985,
+      "grad_norm": 698.1956787109375,
+      "learning_rate": 2.0606060606060607e-06,
+      "loss": 94.1049,
+      "step": 2550
+    },
+    {
+      "epoch": 0.010342723934113616,
+      "grad_norm": 422.0584716796875,
+      "learning_rate": 2.0686868686868688e-06,
+      "loss": 52.2436,
+      "step": 2560
+    },
+    {
+      "epoch": 0.010383125199481247,
+      "grad_norm": 1026.939697265625,
+      "learning_rate": 2.0767676767676768e-06,
+      "loss": 81.9264,
+      "step": 2570
+    },
+    {
+      "epoch": 0.010423526464848879,
+      "grad_norm": 672.6967163085938,
+      "learning_rate": 2.0848484848484852e-06,
+      "loss": 49.8656,
+      "step": 2580
+    },
+    {
+      "epoch": 0.01046392773021651,
+      "grad_norm": 1137.779052734375,
+      "learning_rate": 2.092929292929293e-06,
+      "loss": 49.7138,
+      "step": 2590
+    },
+    {
+      "epoch": 0.010504328995584141,
+      "grad_norm": 709.0454711914062,
+      "learning_rate": 2.1010101010101013e-06,
+      "loss": 50.5514,
+      "step": 2600
+    },
+    {
+      "epoch": 0.010544730260951773,
+      "grad_norm": 1094.946044921875,
+      "learning_rate": 2.1090909090909093e-06,
+      "loss": 54.4812,
+      "step": 2610
+    },
+    {
+      "epoch": 0.010585131526319404,
+      "grad_norm": 695.2977294921875,
+      "learning_rate": 2.1171717171717173e-06,
+      "loss": 58.17,
+      "step": 2620
+    },
+    {
+      "epoch": 0.010625532791687035,
+      "grad_norm": 639.0679931640625,
+      "learning_rate": 2.1252525252525254e-06,
+      "loss": 66.6043,
+      "step": 2630
+    },
+    {
+      "epoch": 0.010665934057054667,
+      "grad_norm": 443.9601135253906,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 88.869,
+      "step": 2640
+    },
+    {
+      "epoch": 0.010706335322422298,
+      "grad_norm": 601.3858032226562,
+      "learning_rate": 2.1414141414141414e-06,
+      "loss": 63.8113,
+      "step": 2650
+    },
+    {
+      "epoch": 0.01074673658778993,
+      "grad_norm": 216.58883666992188,
+      "learning_rate": 2.1494949494949494e-06,
+      "loss": 53.4207,
+      "step": 2660
+    },
+    {
+      "epoch": 0.01078713785315756,
+      "grad_norm": 796.6296997070312,
+      "learning_rate": 2.157575757575758e-06,
+      "loss": 73.9355,
+      "step": 2670
+    },
+    {
+      "epoch": 0.010827539118525192,
+      "grad_norm": 429.7906188964844,
+      "learning_rate": 2.165656565656566e-06,
+      "loss": 52.2577,
+      "step": 2680
+    },
+    {
+      "epoch": 0.010867940383892823,
+      "grad_norm": 961.33740234375,
+      "learning_rate": 2.173737373737374e-06,
+      "loss": 63.543,
+      "step": 2690
+    },
+    {
+      "epoch": 0.010908341649260455,
+      "grad_norm": 1032.55810546875,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 75.2953,
+      "step": 2700
+    },
+    {
+      "epoch": 0.010948742914628086,
+      "grad_norm": 707.4940185546875,
+      "learning_rate": 2.18989898989899e-06,
+      "loss": 155.0811,
+      "step": 2710
+    },
+    {
+      "epoch": 0.010989144179995717,
+      "grad_norm": 632.8768310546875,
+      "learning_rate": 2.197979797979798e-06,
+      "loss": 111.3545,
+      "step": 2720
+    },
+    {
+      "epoch": 0.011029545445363349,
+      "grad_norm": 422.77349853515625,
+      "learning_rate": 2.2060606060606064e-06,
+      "loss": 48.8855,
+      "step": 2730
+    },
+    {
+      "epoch": 0.01106994671073098,
+      "grad_norm": 1005.6026611328125,
+      "learning_rate": 2.2141414141414145e-06,
+      "loss": 71.3869,
+      "step": 2740
+    },
+    {
+      "epoch": 0.011110347976098611,
+      "grad_norm": 618.6168212890625,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 50.2819,
+      "step": 2750
+    },
+    {
+      "epoch": 0.011150749241466243,
+      "grad_norm": 661.1751708984375,
+      "learning_rate": 2.2303030303030305e-06,
+      "loss": 65.9274,
+      "step": 2760
+    },
+    {
+      "epoch": 0.011191150506833874,
+      "grad_norm": 718.5821533203125,
+      "learning_rate": 2.2383838383838385e-06,
+      "loss": 45.7298,
+      "step": 2770
+    },
+    {
+      "epoch": 0.011231551772201505,
+      "grad_norm": 375.16595458984375,
+      "learning_rate": 2.2464646464646466e-06,
+      "loss": 64.9801,
+      "step": 2780
+    },
+    {
+      "epoch": 0.011271953037569137,
+      "grad_norm": 290.95556640625,
+      "learning_rate": 2.254545454545455e-06,
+      "loss": 53.4231,
+      "step": 2790
+    },
+    {
+      "epoch": 0.011312354302936768,
+      "grad_norm": 753.8040771484375,
+      "learning_rate": 2.262626262626263e-06,
+      "loss": 54.899,
+      "step": 2800
+    },
+    {
+      "epoch": 0.0113527555683044,
+      "grad_norm": 948.993896484375,
+      "learning_rate": 2.2707070707070706e-06,
+      "loss": 139.1114,
+      "step": 2810
+    },
+    {
+      "epoch": 0.01139315683367203,
+      "grad_norm": 564.3215942382812,
+      "learning_rate": 2.278787878787879e-06,
+      "loss": 55.8284,
+      "step": 2820
+    },
+    {
+      "epoch": 0.011433558099039662,
+      "grad_norm": 884.8452758789062,
+      "learning_rate": 2.286868686868687e-06,
+      "loss": 74.0817,
+      "step": 2830
+    },
+    {
+      "epoch": 0.011473959364407293,
+      "grad_norm": 691.4619750976562,
+      "learning_rate": 2.294949494949495e-06,
+      "loss": 65.5481,
+      "step": 2840
+    },
+    {
+      "epoch": 0.011514360629774924,
+      "grad_norm": 513.1780395507812,
+      "learning_rate": 2.303030303030303e-06,
+      "loss": 62.6853,
+      "step": 2850
+    },
+    {
+      "epoch": 0.011554761895142556,
+      "grad_norm": 446.5616760253906,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 64.6615,
+      "step": 2860
+    },
+    {
+      "epoch": 0.011595163160510187,
+      "grad_norm": 1361.8853759765625,
+      "learning_rate": 2.3191919191919192e-06,
+      "loss": 74.9464,
+      "step": 2870
+    },
+    {
+      "epoch": 0.011635564425877818,
+      "grad_norm": 511.1175537109375,
+      "learning_rate": 2.3272727272727277e-06,
+      "loss": 40.8995,
+      "step": 2880
+    },
+    {
+      "epoch": 0.01167596569124545,
+      "grad_norm": 935.4153442382812,
+      "learning_rate": 2.3353535353535357e-06,
+      "loss": 72.9418,
+      "step": 2890
+    },
+    {
+      "epoch": 0.011716366956613081,
+      "grad_norm": 504.27166748046875,
+      "learning_rate": 2.3434343434343437e-06,
+      "loss": 77.0979,
+      "step": 2900
+    },
+    {
+      "epoch": 0.011756768221980712,
+      "grad_norm": 407.9814147949219,
+      "learning_rate": 2.3515151515151517e-06,
+      "loss": 65.9649,
+      "step": 2910
+    },
+    {
+      "epoch": 0.011797169487348344,
+      "grad_norm": 463.7010192871094,
+      "learning_rate": 2.3595959595959598e-06,
+      "loss": 51.9431,
+      "step": 2920
+    },
+    {
+      "epoch": 0.011837570752715975,
+      "grad_norm": 493.0938720703125,
+      "learning_rate": 2.367676767676768e-06,
+      "loss": 57.3446,
+      "step": 2930
+    },
+    {
+      "epoch": 0.011877972018083606,
+      "grad_norm": 465.1329345703125,
+      "learning_rate": 2.375757575757576e-06,
+      "loss": 56.2858,
+      "step": 2940
+    },
+    {
+      "epoch": 0.011918373283451238,
+      "grad_norm": 853.31396484375,
+      "learning_rate": 2.3838383838383843e-06,
+      "loss": 66.7674,
+      "step": 2950
+    },
+    {
+      "epoch": 0.011958774548818869,
+      "grad_norm": 1173.5980224609375,
+      "learning_rate": 2.3919191919191923e-06,
+      "loss": 85.9143,
+      "step": 2960
+    },
+    {
+      "epoch": 0.0119991758141865,
+      "grad_norm": 1093.7786865234375,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 55.0169,
+      "step": 2970
+    },
+    {
+      "epoch": 0.012039577079554132,
+      "grad_norm": 525.7014770507812,
+      "learning_rate": 2.4080808080808083e-06,
+      "loss": 46.8544,
+      "step": 2980
+    },
+    {
+      "epoch": 0.012079978344921763,
+      "grad_norm": 1659.3507080078125,
+      "learning_rate": 2.4161616161616164e-06,
+      "loss": 95.7587,
+      "step": 2990
+    },
+    {
+      "epoch": 0.012120379610289394,
+      "grad_norm": 520.4998779296875,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 51.6112,
+      "step": 3000
+    },
+    {
+      "epoch": 0.012160780875657026,
+      "grad_norm": 1072.08447265625,
+      "learning_rate": 2.432323232323233e-06,
+      "loss": 80.0246,
+      "step": 3010
+    },
+    {
+      "epoch": 0.012201182141024657,
+      "grad_norm": 371.0710144042969,
+      "learning_rate": 2.4404040404040404e-06,
+      "loss": 59.9696,
+      "step": 3020
+    },
+    {
+      "epoch": 0.012241583406392288,
+      "grad_norm": 1110.0040283203125,
+      "learning_rate": 2.4484848484848485e-06,
+      "loss": 51.8267,
+      "step": 3030
+    },
+    {
+      "epoch": 0.01228198467175992,
+      "grad_norm": 664.203369140625,
+      "learning_rate": 2.456565656565657e-06,
+      "loss": 72.4027,
+      "step": 3040
+    },
+    {
+      "epoch": 0.01232238593712755,
+      "grad_norm": 804.1693725585938,
+      "learning_rate": 2.464646464646465e-06,
+      "loss": 82.5,
+      "step": 3050
+    },
+    {
+      "epoch": 0.012362787202495182,
+      "grad_norm": 426.992919921875,
+      "learning_rate": 2.472727272727273e-06,
+      "loss": 51.2883,
+      "step": 3060
+    },
+    {
+      "epoch": 0.012403188467862813,
+      "grad_norm": 946.348876953125,
+      "learning_rate": 2.480808080808081e-06,
+      "loss": 53.2797,
+      "step": 3070
+    },
+    {
+      "epoch": 0.012443589733230445,
+      "grad_norm": 925.2742309570312,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 64.6453,
+      "step": 3080
+    },
+    {
+      "epoch": 0.012483990998598076,
+      "grad_norm": 773.533203125,
+      "learning_rate": 2.496969696969697e-06,
+      "loss": 64.05,
+      "step": 3090
+    },
+    {
+      "epoch": 0.012524392263965707,
+      "grad_norm": 1721.887939453125,
+      "learning_rate": 2.5050505050505055e-06,
+      "loss": 89.4751,
+      "step": 3100
+    },
+    {
+      "epoch": 0.012564793529333339,
+      "grad_norm": 502.3291320800781,
+      "learning_rate": 2.5131313131313135e-06,
+      "loss": 73.6394,
+      "step": 3110
+    },
+    {
+      "epoch": 0.01260519479470097,
+      "grad_norm": 951.5177612304688,
+      "learning_rate": 2.5212121212121215e-06,
+      "loss": 69.3922,
+      "step": 3120
+    },
+    {
+      "epoch": 0.012645596060068601,
+      "grad_norm": 528.2192993164062,
+      "learning_rate": 2.5292929292929296e-06,
+      "loss": 55.9364,
+      "step": 3130
+    },
+    {
+      "epoch": 0.012685997325436233,
+      "grad_norm": 760.628662109375,
+      "learning_rate": 2.537373737373738e-06,
+      "loss": 51.1709,
+      "step": 3140
+    },
+    {
+      "epoch": 0.012726398590803864,
+      "grad_norm": 718.6551513671875,
+      "learning_rate": 2.5454545454545456e-06,
+      "loss": 88.1984,
+      "step": 3150
+    },
+    {
+      "epoch": 0.012766799856171495,
+      "grad_norm": 1814.95654296875,
+      "learning_rate": 2.5535353535353536e-06,
+      "loss": 82.0715,
+      "step": 3160
+    },
+    {
+      "epoch": 0.012807201121539127,
+      "grad_norm": 724.8761596679688,
+      "learning_rate": 2.5616161616161617e-06,
+      "loss": 50.4141,
+      "step": 3170
+    },
+    {
+      "epoch": 0.012847602386906758,
+      "grad_norm": 720.0292358398438,
+      "learning_rate": 2.5696969696969697e-06,
+      "loss": 79.8761,
+      "step": 3180
+    },
+    {
+      "epoch": 0.01288800365227439,
+      "grad_norm": 543.6854858398438,
+      "learning_rate": 2.577777777777778e-06,
+      "loss": 57.894,
+      "step": 3190
+    },
+    {
+      "epoch": 0.01292840491764202,
+      "grad_norm": 1304.784912109375,
+      "learning_rate": 2.585858585858586e-06,
+      "loss": 61.9256,
+      "step": 3200
+    },
+    {
+      "epoch": 0.012968806183009652,
+      "grad_norm": 749.076904296875,
+      "learning_rate": 2.593939393939394e-06,
+      "loss": 86.3883,
+      "step": 3210
+    },
+    {
+      "epoch": 0.013009207448377283,
+      "grad_norm": 744.2839965820312,
+      "learning_rate": 2.602020202020202e-06,
+      "loss": 58.4392,
+      "step": 3220
+    },
+    {
+      "epoch": 0.013049608713744915,
+      "grad_norm": 642.6310424804688,
+      "learning_rate": 2.6101010101010107e-06,
+      "loss": 74.5423,
+      "step": 3230
+    },
+    {
+      "epoch": 0.013090009979112546,
+      "grad_norm": 1457.65380859375,
+      "learning_rate": 2.6181818181818187e-06,
+      "loss": 68.783,
+      "step": 3240
+    },
+    {
+      "epoch": 0.013130411244480177,
+      "grad_norm": 900.1226196289062,
+      "learning_rate": 2.6262626262626267e-06,
+      "loss": 63.59,
+      "step": 3250
+    },
+    {
+      "epoch": 0.013170812509847809,
+      "grad_norm": 727.9959716796875,
+      "learning_rate": 2.6343434343434343e-06,
+      "loss": 74.8924,
+      "step": 3260
+    },
+    {
+      "epoch": 0.01321121377521544,
+      "grad_norm": 569.1546630859375,
+      "learning_rate": 2.6424242424242423e-06,
+      "loss": 75.5204,
+      "step": 3270
+    },
+    {
+      "epoch": 0.013251615040583071,
+      "grad_norm": 579.5902099609375,
+      "learning_rate": 2.6505050505050508e-06,
+      "loss": 52.7214,
+      "step": 3280
+    },
+    {
+      "epoch": 0.013292016305950703,
+      "grad_norm": 725.4396362304688,
+      "learning_rate": 2.658585858585859e-06,
+      "loss": 80.6275,
+      "step": 3290
+    },
+    {
+      "epoch": 0.013332417571318334,
+      "grad_norm": 752.127685546875,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 62.3925,
+      "step": 3300
+    },
+    {
+      "epoch": 0.013372818836685965,
+      "grad_norm": 730.8565063476562,
+      "learning_rate": 2.674747474747475e-06,
+      "loss": 46.3678,
+      "step": 3310
+    },
+    {
+      "epoch": 0.013413220102053596,
+      "grad_norm": 706.3470458984375,
+      "learning_rate": 2.6828282828282833e-06,
+      "loss": 55.6273,
+      "step": 3320
+    },
+    {
+      "epoch": 0.013453621367421228,
+      "grad_norm": 850.8480834960938,
+      "learning_rate": 2.6909090909090913e-06,
+      "loss": 76.168,
+      "step": 3330
+    },
+    {
+      "epoch": 0.013494022632788859,
+      "grad_norm": 547.6309814453125,
+      "learning_rate": 2.6989898989898994e-06,
+      "loss": 75.9509,
+      "step": 3340
+    },
+    {
+      "epoch": 0.01353442389815649,
+      "grad_norm": 2081.936279296875,
+      "learning_rate": 2.7070707070707074e-06,
+      "loss": 74.0594,
+      "step": 3350
+    },
+    {
+      "epoch": 0.013574825163524122,
+      "grad_norm": 795.8720703125,
+      "learning_rate": 2.715151515151516e-06,
+      "loss": 51.1241,
+      "step": 3360
+    },
+    {
+      "epoch": 0.013615226428891753,
+      "grad_norm": 741.885498046875,
+      "learning_rate": 2.7232323232323234e-06,
+      "loss": 93.6878,
+      "step": 3370
+    },
+    {
+      "epoch": 0.013655627694259384,
+      "grad_norm": 540.7803955078125,
+      "learning_rate": 2.7313131313131315e-06,
+      "loss": 41.2008,
+      "step": 3380
+    },
+    {
+      "epoch": 0.013696028959627016,
+      "grad_norm": 510.04937744140625,
+      "learning_rate": 2.7393939393939395e-06,
+      "loss": 64.7001,
+      "step": 3390
+    },
+    {
+      "epoch": 0.013736430224994647,
+      "grad_norm": 743.4458618164062,
+      "learning_rate": 2.7474747474747475e-06,
+      "loss": 59.3923,
+      "step": 3400
+    },
+    {
+      "epoch": 0.013776831490362278,
+      "grad_norm": 824.0260009765625,
+      "learning_rate": 2.755555555555556e-06,
+      "loss": 103.2661,
+      "step": 3410
+    },
+    {
+      "epoch": 0.01381723275572991,
+      "grad_norm": 809.0377807617188,
+      "learning_rate": 2.763636363636364e-06,
+      "loss": 65.098,
+      "step": 3420
+    },
+    {
+      "epoch": 0.013857634021097541,
+      "grad_norm": 1693.0699462890625,
+      "learning_rate": 2.771717171717172e-06,
+      "loss": 50.6526,
+      "step": 3430
+    },
+    {
+      "epoch": 0.013898035286465172,
+      "grad_norm": 519.3567504882812,
+      "learning_rate": 2.77979797979798e-06,
+      "loss": 32.3857,
+      "step": 3440
+    },
+    {
+      "epoch": 0.013938436551832804,
+      "grad_norm": 293.765380859375,
+      "learning_rate": 2.7878787878787885e-06,
+      "loss": 44.1302,
+      "step": 3450
+    },
+    {
+      "epoch": 0.013978837817200435,
+      "grad_norm": 879.6158447265625,
+      "learning_rate": 2.7959595959595965e-06,
+      "loss": 49.3538,
+      "step": 3460
+    },
+    {
+      "epoch": 0.014019239082568066,
+      "grad_norm": 616.3663330078125,
+      "learning_rate": 2.804040404040404e-06,
+      "loss": 53.0307,
+      "step": 3470
+    },
+    {
+      "epoch": 0.014059640347935698,
+      "grad_norm": 987.908447265625,
+      "learning_rate": 2.812121212121212e-06,
+      "loss": 63.739,
+      "step": 3480
+    },
+    {
+      "epoch": 0.014100041613303329,
+      "grad_norm": 767.0664672851562,
+      "learning_rate": 2.82020202020202e-06,
+      "loss": 59.8414,
+      "step": 3490
+    },
+    {
+      "epoch": 0.01414044287867096,
+      "grad_norm": 1181.593994140625,
+      "learning_rate": 2.8282828282828286e-06,
+      "loss": 55.1145,
+      "step": 3500
+    },
+    {
+      "epoch": 0.014180844144038592,
+      "grad_norm": 512.4702758789062,
+      "learning_rate": 2.8363636363636366e-06,
+      "loss": 49.1521,
+      "step": 3510
+    },
+    {
+      "epoch": 0.014221245409406223,
+      "grad_norm": 889.2345581054688,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 46.9908,
+      "step": 3520
+    },
+    {
+      "epoch": 0.014261646674773854,
+      "grad_norm": 616.6620483398438,
+      "learning_rate": 2.8525252525252527e-06,
+      "loss": 55.8178,
+      "step": 3530
+    },
+    {
+      "epoch": 0.014302047940141486,
+      "grad_norm": 950.4310302734375,
+      "learning_rate": 2.860606060606061e-06,
+      "loss": 44.9625,
+      "step": 3540
+    },
+    {
+      "epoch": 0.014342449205509117,
+      "grad_norm": 678.1963500976562,
+      "learning_rate": 2.868686868686869e-06,
+      "loss": 47.2252,
+      "step": 3550
+    },
+    {
+      "epoch": 0.014382850470876748,
+      "grad_norm": 274.08941650390625,
+      "learning_rate": 2.876767676767677e-06,
+      "loss": 54.4971,
+      "step": 3560
+    },
+    {
+      "epoch": 0.01442325173624438,
+      "grad_norm": 702.7638549804688,
+      "learning_rate": 2.884848484848485e-06,
+      "loss": 51.9829,
+      "step": 3570
+    },
+    {
+      "epoch": 0.01446365300161201,
+      "grad_norm": 377.993408203125,
+      "learning_rate": 2.892929292929293e-06,
+      "loss": 38.773,
+      "step": 3580
+    },
+    {
+      "epoch": 0.014504054266979642,
+      "grad_norm": 606.4713745117188,
+      "learning_rate": 2.9010101010101012e-06,
+      "loss": 48.4606,
+      "step": 3590
+    },
+    {
+      "epoch": 0.014544455532347273,
+      "grad_norm": 941.6731567382812,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 49.9611,
+      "step": 3600
+    },
+    {
+      "epoch": 0.014584856797714905,
+      "grad_norm": 576.7796630859375,
+      "learning_rate": 2.9171717171717173e-06,
+      "loss": 77.4996,
+      "step": 3610
+    },
+    {
+      "epoch": 0.014625258063082536,
+      "grad_norm": 582.7783203125,
+      "learning_rate": 2.9252525252525253e-06,
+      "loss": 65.8574,
+      "step": 3620
+    },
+    {
+      "epoch": 0.014665659328450167,
+      "grad_norm": 340.5007019042969,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 56.075,
+      "step": 3630
+    },
+    {
+      "epoch": 0.014706060593817799,
+      "grad_norm": 1056.101806640625,
+      "learning_rate": 2.941414141414142e-06,
+      "loss": 54.8923,
+      "step": 3640
+    },
+    {
+      "epoch": 0.01474646185918543,
+      "grad_norm": 697.7027587890625,
+      "learning_rate": 2.94949494949495e-06,
+      "loss": 68.0702,
+      "step": 3650
+    },
+    {
+      "epoch": 0.014786863124553061,
+      "grad_norm": 711.9849243164062,
+      "learning_rate": 2.957575757575758e-06,
+      "loss": 57.5173,
+      "step": 3660
+    },
+    {
+      "epoch": 0.014827264389920693,
+      "grad_norm": 645.8955078125,
+      "learning_rate": 2.9656565656565663e-06,
+      "loss": 77.3891,
+      "step": 3670
+    },
+    {
+      "epoch": 0.014867665655288324,
+      "grad_norm": 836.5162353515625,
+      "learning_rate": 2.9737373737373743e-06,
+      "loss": 54.3535,
+      "step": 3680
+    },
+    {
+      "epoch": 0.014908066920655955,
+      "grad_norm": 756.8634643554688,
+      "learning_rate": 2.981818181818182e-06,
+      "loss": 73.0932,
+      "step": 3690
+    },
+    {
+      "epoch": 0.014948468186023587,
+      "grad_norm": 840.5918579101562,
+      "learning_rate": 2.98989898989899e-06,
+      "loss": 119.6765,
+      "step": 3700
+    },
+    {
+      "epoch": 0.014988869451391218,
+      "grad_norm": 485.3866271972656,
+      "learning_rate": 2.997979797979798e-06,
+      "loss": 57.1866,
+      "step": 3710
+    },
+    {
+      "epoch": 0.01502927071675885,
+      "grad_norm": 1064.0423583984375,
+      "learning_rate": 3.0060606060606064e-06,
+      "loss": 65.6966,
+      "step": 3720
+    },
+    {
+      "epoch": 0.01506967198212648,
+      "grad_norm": 899.6082763671875,
+      "learning_rate": 3.0141414141414144e-06,
+      "loss": 77.0435,
+      "step": 3730
+    },
+    {
+      "epoch": 0.015110073247494112,
+      "grad_norm": 542.0571899414062,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 61.2996,
+      "step": 3740
+    },
+    {
+      "epoch": 0.015150474512861743,
+      "grad_norm": 1062.953369140625,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 81.9733,
+      "step": 3750
+    },
+    {
+      "epoch": 0.015190875778229375,
+      "grad_norm": 423.0398864746094,
+      "learning_rate": 3.038383838383839e-06,
+      "loss": 49.3342,
+      "step": 3760
+    },
+    {
+      "epoch": 0.015231277043597006,
+      "grad_norm": 587.5816650390625,
+      "learning_rate": 3.046464646464647e-06,
+      "loss": 71.2666,
+      "step": 3770
+    },
+    {
+      "epoch": 0.015271678308964637,
+      "grad_norm": 827.7647705078125,
+      "learning_rate": 3.054545454545455e-06,
+      "loss": 58.1244,
+      "step": 3780
+    },
+    {
+      "epoch": 0.015312079574332269,
+      "grad_norm": 819.0059814453125,
+      "learning_rate": 3.0626262626262626e-06,
+      "loss": 69.2099,
+      "step": 3790
+    },
+    {
+      "epoch": 0.0153524808396999,
+      "grad_norm": 538.2634887695312,
+      "learning_rate": 3.0707070707070706e-06,
+      "loss": 41.0661,
+      "step": 3800
+    },
+    {
+      "epoch": 0.015392882105067531,
+      "grad_norm": 755.6787109375,
+      "learning_rate": 3.078787878787879e-06,
+      "loss": 63.0487,
+      "step": 3810
+    },
+    {
+      "epoch": 0.015433283370435162,
+      "grad_norm": 507.8043518066406,
+      "learning_rate": 3.086868686868687e-06,
+      "loss": 49.7004,
+      "step": 3820
+    },
+    {
+      "epoch": 0.015473684635802794,
+      "grad_norm": 783.618896484375,
+      "learning_rate": 3.094949494949495e-06,
+      "loss": 63.6786,
+      "step": 3830
+    },
+    {
+      "epoch": 0.015514085901170425,
+      "grad_norm": 573.7366943359375,
+      "learning_rate": 3.103030303030303e-06,
+      "loss": 51.0182,
+      "step": 3840
+    },
+    {
+      "epoch": 0.015554487166538056,
+      "grad_norm": 435.3841247558594,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 58.01,
+      "step": 3850
+    },
+    {
+      "epoch": 0.015594888431905688,
+      "grad_norm": 979.0087890625,
+      "learning_rate": 3.1191919191919196e-06,
+      "loss": 49.2491,
+      "step": 3860
+    },
+    {
+      "epoch": 0.01563528969727332,
+      "grad_norm": 361.8955078125,
+      "learning_rate": 3.1272727272727276e-06,
+      "loss": 56.3938,
+      "step": 3870
+    },
+    {
+      "epoch": 0.01567569096264095,
+      "grad_norm": 323.97381591796875,
+      "learning_rate": 3.1353535353535357e-06,
+      "loss": 49.1725,
+      "step": 3880
+    },
+    {
+      "epoch": 0.01571609222800858,
+      "grad_norm": 501.0435485839844,
+      "learning_rate": 3.143434343434344e-06,
+      "loss": 52.2216,
+      "step": 3890
+    },
+    {
+      "epoch": 0.01575649349337621,
+      "grad_norm": 1032.385009765625,
+      "learning_rate": 3.1515151515151517e-06,
+      "loss": 60.1772,
+      "step": 3900
+    },
+    {
+      "epoch": 0.015796894758743844,
+      "grad_norm": 400.3407287597656,
+      "learning_rate": 3.1595959595959597e-06,
+      "loss": 53.0336,
+      "step": 3910
+    },
+    {
+      "epoch": 0.015837296024111474,
+      "grad_norm": 961.6312866210938,
+      "learning_rate": 3.1676767676767678e-06,
+      "loss": 46.5452,
+      "step": 3920
+    },
+    {
+      "epoch": 0.015877697289479107,
+      "grad_norm": 697.0581665039062,
+      "learning_rate": 3.1757575757575758e-06,
+      "loss": 69.2422,
+      "step": 3930
+    },
+    {
+      "epoch": 0.015918098554846737,
+      "grad_norm": 744.8246459960938,
+      "learning_rate": 3.1838383838383842e-06,
+      "loss": 46.6851,
+      "step": 3940
+    },
+    {
+      "epoch": 0.01595849982021437,
+      "grad_norm": 454.99267578125,
+      "learning_rate": 3.1919191919191923e-06,
+      "loss": 69.5364,
+      "step": 3950
+    },
+    {
+      "epoch": 0.015998901085582,
+      "grad_norm": 890.9744262695312,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 70.4928,
+      "step": 3960
+    },
+    {
+      "epoch": 0.016039302350949632,
+      "grad_norm": 381.5486755371094,
+      "learning_rate": 3.2080808080808083e-06,
+      "loss": 52.5701,
+      "step": 3970
+    },
+    {
+      "epoch": 0.016079703616317262,
+      "grad_norm": 722.4985961914062,
+      "learning_rate": 3.2161616161616168e-06,
+      "loss": 39.8047,
+      "step": 3980
+    },
+    {
+      "epoch": 0.016120104881684895,
+      "grad_norm": 457.2010498046875,
+      "learning_rate": 3.2242424242424248e-06,
+      "loss": 73.1557,
+      "step": 3990
+    },
+    {
+      "epoch": 0.016160506147052525,
+      "grad_norm": 412.5587463378906,
+      "learning_rate": 3.232323232323233e-06,
+      "loss": 79.8474,
+      "step": 4000
+    },
+    {
+      "epoch": 0.016200907412420158,
+      "grad_norm": 432.9417419433594,
+      "learning_rate": 3.2404040404040404e-06,
+      "loss": 58.3798,
+      "step": 4010
+    },
+    {
+      "epoch": 0.016241308677787787,
+      "grad_norm": 883.053466796875,
+      "learning_rate": 3.2484848484848484e-06,
+      "loss": 56.1412,
+      "step": 4020
+    },
+    {
+      "epoch": 0.01628170994315542,
+      "grad_norm": 433.01910400390625,
+      "learning_rate": 3.256565656565657e-06,
+      "loss": 78.8629,
+      "step": 4030
+    },
+    {
+      "epoch": 0.01632211120852305,
+      "grad_norm": 590.1842651367188,
+      "learning_rate": 3.264646464646465e-06,
+      "loss": 74.5015,
+      "step": 4040
+    },
+    {
+      "epoch": 0.016362512473890683,
+      "grad_norm": 1087.239501953125,
+      "learning_rate": 3.272727272727273e-06,
+      "loss": 47.2005,
+      "step": 4050
+    },
+    {
+      "epoch": 0.016402913739258312,
+      "grad_norm": 547.9854736328125,
+      "learning_rate": 3.280808080808081e-06,
+      "loss": 88.357,
+      "step": 4060
+    },
+    {
+      "epoch": 0.016443315004625945,
+      "grad_norm": 1154.8641357421875,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 55.9507,
+      "step": 4070
+    },
+    {
+      "epoch": 0.016483716269993575,
+      "grad_norm": 696.9087524414062,
+      "learning_rate": 3.2969696969696974e-06,
+      "loss": 56.5099,
+      "step": 4080
+    },
+    {
+      "epoch": 0.016524117535361208,
+      "grad_norm": 612.0421142578125,
+      "learning_rate": 3.3050505050505054e-06,
+      "loss": 73.3457,
+      "step": 4090
+    },
+    {
+      "epoch": 0.016564518800728838,
+      "grad_norm": 935.5057373046875,
+      "learning_rate": 3.3131313131313135e-06,
+      "loss": 50.4606,
+      "step": 4100
+    },
+    {
+      "epoch": 0.01660492006609647,
+      "grad_norm": 888.5339965820312,
+      "learning_rate": 3.321212121212121e-06,
+      "loss": 56.5149,
+      "step": 4110
+    },
+    {
+      "epoch": 0.0166453213314641,
+      "grad_norm": 392.083740234375,
+      "learning_rate": 3.3292929292929295e-06,
+      "loss": 66.5389,
+      "step": 4120
+    },
+    {
+      "epoch": 0.016685722596831733,
+      "grad_norm": 340.5673522949219,
+      "learning_rate": 3.3373737373737375e-06,
+      "loss": 75.7454,
+      "step": 4130
+    },
+    {
+      "epoch": 0.016726123862199363,
+      "grad_norm": 668.51025390625,
+      "learning_rate": 3.3454545454545456e-06,
+      "loss": 52.3896,
+      "step": 4140
+    },
+    {
+      "epoch": 0.016766525127566996,
+      "grad_norm": 728.2340698242188,
+      "learning_rate": 3.3535353535353536e-06,
+      "loss": 80.3662,
+      "step": 4150
+    },
+    {
+      "epoch": 0.016806926392934626,
+      "grad_norm": 497.7481384277344,
+      "learning_rate": 3.361616161616162e-06,
+      "loss": 68.5763,
+      "step": 4160
+    },
+    {
+      "epoch": 0.01684732765830226,
+      "grad_norm": 589.8648071289062,
+      "learning_rate": 3.36969696969697e-06,
+      "loss": 87.7933,
+      "step": 4170
+    },
+    {
+      "epoch": 0.016887728923669888,
+      "grad_norm": 1049.1983642578125,
+      "learning_rate": 3.377777777777778e-06,
+      "loss": 58.2885,
+      "step": 4180
+    },
+    {
+      "epoch": 0.01692813018903752,
+      "grad_norm": 472.6341247558594,
+      "learning_rate": 3.385858585858586e-06,
+      "loss": 48.8271,
+      "step": 4190
+    },
+    {
+      "epoch": 0.01696853145440515,
+      "grad_norm": 648.27734375,
+      "learning_rate": 3.3939393939393946e-06,
+      "loss": 64.2691,
+      "step": 4200
+    },
+    {
+      "epoch": 0.017008932719772784,
+      "grad_norm": 446.6534729003906,
+      "learning_rate": 3.4020202020202026e-06,
+      "loss": 65.1291,
+      "step": 4210
+    },
+    {
+      "epoch": 0.017049333985140414,
+      "grad_norm": 212.90321350097656,
+      "learning_rate": 3.41010101010101e-06,
+      "loss": 41.2398,
+      "step": 4220
+    },
+    {
+      "epoch": 0.017089735250508047,
+      "grad_norm": 1060.3450927734375,
+      "learning_rate": 3.4181818181818182e-06,
+      "loss": 74.0737,
+      "step": 4230
+    },
+    {
+      "epoch": 0.017130136515875676,
+      "grad_norm": 622.3221435546875,
+      "learning_rate": 3.4262626262626262e-06,
+      "loss": 63.4254,
+      "step": 4240
+    },
+    {
+      "epoch": 0.01717053778124331,
+      "grad_norm": 1164.900634765625,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 76.6035,
+      "step": 4250
+    },
+    {
+      "epoch": 0.01721093904661094,
+      "grad_norm": 513.646728515625,
+      "learning_rate": 3.4424242424242427e-06,
+      "loss": 66.3594,
+      "step": 4260
+    },
+    {
+      "epoch": 0.017251340311978572,
+      "grad_norm": 457.9349060058594,
+      "learning_rate": 3.4505050505050507e-06,
+      "loss": 55.117,
+      "step": 4270
+    },
+    {
+      "epoch": 0.0172917415773462,
+      "grad_norm": 786.7095336914062,
+      "learning_rate": 3.4585858585858588e-06,
+      "loss": 68.2138,
+      "step": 4280
+    },
+    {
+      "epoch": 0.017332142842713835,
+      "grad_norm": 131.28732299804688,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 33.2112,
+      "step": 4290
+    },
+    {
+      "epoch": 0.017372544108081464,
+      "grad_norm": 1183.358642578125,
+      "learning_rate": 3.4747474747474752e-06,
+      "loss": 57.2951,
+      "step": 4300
+    },
+    {
+      "epoch": 0.017412945373449097,
+      "grad_norm": 770.9044799804688,
+      "learning_rate": 3.4828282828282833e-06,
+      "loss": 73.1749,
+      "step": 4310
+    },
+    {
+      "epoch": 0.017453346638816727,
+      "grad_norm": 920.8295288085938,
+      "learning_rate": 3.4909090909090913e-06,
+      "loss": 43.6484,
+      "step": 4320
+    },
+    {
+      "epoch": 0.01749374790418436,
+      "grad_norm": 613.7767333984375,
+      "learning_rate": 3.498989898989899e-06,
+      "loss": 50.3909,
+      "step": 4330
+    },
+    {
+      "epoch": 0.01753414916955199,
+      "grad_norm": 936.6824951171875,
+      "learning_rate": 3.5070707070707073e-06,
+      "loss": 54.8212,
+      "step": 4340
+    },
+    {
+      "epoch": 0.017574550434919622,
+      "grad_norm": 512.0675048828125,
+      "learning_rate": 3.5151515151515154e-06,
+      "loss": 67.7541,
+      "step": 4350
+    },
+    {
+      "epoch": 0.017614951700287252,
+      "grad_norm": 522.534912109375,
+      "learning_rate": 3.5232323232323234e-06,
+      "loss": 59.2163,
+      "step": 4360
+    },
+    {
+      "epoch": 0.017655352965654885,
+      "grad_norm": 696.8623046875,
+      "learning_rate": 3.5313131313131314e-06,
+      "loss": 60.4836,
+      "step": 4370
+    },
+    {
+      "epoch": 0.017695754231022515,
+      "grad_norm": 370.84930419921875,
+      "learning_rate": 3.53939393939394e-06,
+      "loss": 54.7472,
+      "step": 4380
+    },
+    {
+      "epoch": 0.017736155496390148,
+      "grad_norm": 624.1700439453125,
+      "learning_rate": 3.547474747474748e-06,
+      "loss": 41.7864,
+      "step": 4390
+    },
+    {
+      "epoch": 0.017776556761757777,
+      "grad_norm": 512.7784423828125,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 78.3068,
+      "step": 4400
+    },
+    {
+      "epoch": 0.01781695802712541,
+      "grad_norm": 596.0513916015625,
+      "learning_rate": 3.563636363636364e-06,
+      "loss": 64.5097,
+      "step": 4410
+    },
+    {
+      "epoch": 0.01785735929249304,
+      "grad_norm": 603.3121948242188,
+      "learning_rate": 3.5717171717171724e-06,
+      "loss": 44.8857,
+      "step": 4420
+    },
+    {
+      "epoch": 0.017897760557860673,
+      "grad_norm": 702.3116455078125,
+      "learning_rate": 3.57979797979798e-06,
+      "loss": 66.0556,
+      "step": 4430
+    },
+    {
+      "epoch": 0.017938161823228303,
+      "grad_norm": 762.5350952148438,
+      "learning_rate": 3.587878787878788e-06,
+      "loss": 54.1978,
+      "step": 4440
+    },
+    {
+      "epoch": 0.017978563088595936,
+      "grad_norm": 578.8521728515625,
+      "learning_rate": 3.595959595959596e-06,
+      "loss": 63.9503,
+      "step": 4450
+    },
+    {
+      "epoch": 0.018018964353963565,
+      "grad_norm": 514.289306640625,
+      "learning_rate": 3.604040404040404e-06,
+      "loss": 63.0047,
+      "step": 4460
+    },
+    {
+      "epoch": 0.0180593656193312,
+      "grad_norm": 760.0272216796875,
+      "learning_rate": 3.6121212121212125e-06,
+      "loss": 77.3559,
+      "step": 4470
+    },
+    {
+      "epoch": 0.018099766884698828,
+      "grad_norm": 1489.2890625,
+      "learning_rate": 3.6202020202020205e-06,
+      "loss": 67.4747,
+      "step": 4480
+    },
+    {
+      "epoch": 0.01814016815006646,
+      "grad_norm": 639.5055541992188,
+      "learning_rate": 3.6282828282828286e-06,
+      "loss": 62.7511,
+      "step": 4490
+    },
+    {
+      "epoch": 0.01818056941543409,
+      "grad_norm": 459.6601257324219,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 60.6712,
+      "step": 4500
+    },
+    {
+      "epoch": 0.018220970680801724,
+      "grad_norm": 614.9654541015625,
+      "learning_rate": 3.644444444444445e-06,
+      "loss": 54.301,
+      "step": 4510
+    },
+    {
+      "epoch": 0.018261371946169353,
+      "grad_norm": 439.7833557128906,
+      "learning_rate": 3.652525252525253e-06,
+      "loss": 43.6239,
+      "step": 4520
+    },
+    {
+      "epoch": 0.018301773211536986,
+      "grad_norm": 796.3854370117188,
+      "learning_rate": 3.660606060606061e-06,
+      "loss": 61.3992,
+      "step": 4530
+    },
+    {
+      "epoch": 0.018342174476904616,
+      "grad_norm": 343.8157653808594,
+      "learning_rate": 3.6686868686868687e-06,
+      "loss": 65.5689,
+      "step": 4540
+    },
+    {
+      "epoch": 0.01838257574227225,
+      "grad_norm": 1090.6224365234375,
+      "learning_rate": 3.6767676767676767e-06,
+      "loss": 76.7682,
+      "step": 4550
+    },
+    {
+      "epoch": 0.01842297700763988,
+      "grad_norm": 599.2227783203125,
+      "learning_rate": 3.684848484848485e-06,
+      "loss": 68.6533,
+      "step": 4560
+    },
+    {
+      "epoch": 0.01846337827300751,
+      "grad_norm": 679.0711669921875,
+      "learning_rate": 3.692929292929293e-06,
+      "loss": 65.9553,
+      "step": 4570
+    },
+    {
+      "epoch": 0.01850377953837514,
+      "grad_norm": 979.4942626953125,
+      "learning_rate": 3.701010101010101e-06,
+      "loss": 67.687,
+      "step": 4580
+    },
+    {
+      "epoch": 0.018544180803742774,
+      "grad_norm": 210.69126892089844,
+      "learning_rate": 3.7090909090909092e-06,
+      "loss": 59.4653,
+      "step": 4590
+    },
+    {
+      "epoch": 0.018584582069110404,
+      "grad_norm": 434.259033203125,
+      "learning_rate": 3.7171717171717177e-06,
+      "loss": 65.9339,
+      "step": 4600
+    },
+    {
+      "epoch": 0.018624983334478037,
+      "grad_norm": 854.1275024414062,
+      "learning_rate": 3.7252525252525257e-06,
+      "loss": 45.3619,
+      "step": 4610
+    },
+    {
+      "epoch": 0.018665384599845666,
+      "grad_norm": 311.56829833984375,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 38.0432,
+      "step": 4620
+    },
+    {
+      "epoch": 0.0187057858652133,
+      "grad_norm": 498.4246520996094,
+      "learning_rate": 3.7414141414141418e-06,
+      "loss": 48.5221,
+      "step": 4630
+    },
+    {
+      "epoch": 0.01874618713058093,
+      "grad_norm": 801.4674072265625,
+      "learning_rate": 3.74949494949495e-06,
+      "loss": 58.1732,
+      "step": 4640
+    },
+    {
+      "epoch": 0.018786588395948562,
+      "grad_norm": 1160.4736328125,
+      "learning_rate": 3.757575757575758e-06,
+      "loss": 53.0421,
+      "step": 4650
+    },
+    {
+      "epoch": 0.01882698966131619,
+      "grad_norm": 897.707763671875,
+      "learning_rate": 3.765656565656566e-06,
+      "loss": 79.2011,
+      "step": 4660
+    },
+    {
+      "epoch": 0.018867390926683825,
+      "grad_norm": 400.8449401855469,
+      "learning_rate": 3.773737373737374e-06,
+      "loss": 60.0867,
+      "step": 4670
+    },
+    {
+      "epoch": 0.018907792192051454,
+      "grad_norm": 739.8035888671875,
+      "learning_rate": 3.781818181818182e-06,
+      "loss": 44.1463,
+      "step": 4680
+    },
+    {
+      "epoch": 0.018948193457419087,
+      "grad_norm": 1055.0322265625,
+      "learning_rate": 3.7898989898989903e-06,
+      "loss": 57.8223,
+      "step": 4690
+    },
+    {
+      "epoch": 0.018988594722786717,
+      "grad_norm": 1063.527587890625,
+      "learning_rate": 3.7979797979797984e-06,
+      "loss": 77.483,
+      "step": 4700
+    },
+    {
+      "epoch": 0.01902899598815435,
+      "grad_norm": 560.8001708984375,
+      "learning_rate": 3.8060606060606064e-06,
+      "loss": 53.0922,
+      "step": 4710
+    },
+    {
+      "epoch": 0.01906939725352198,
+      "grad_norm": 692.53466796875,
+      "learning_rate": 3.8141414141414144e-06,
+      "loss": 51.8766,
+      "step": 4720
+    },
+    {
+      "epoch": 0.019109798518889613,
+      "grad_norm": 1038.6710205078125,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 40.9016,
+      "step": 4730
+    },
+    {
+      "epoch": 0.019150199784257242,
+      "grad_norm": 739.777099609375,
+      "learning_rate": 3.830303030303031e-06,
+      "loss": 71.4101,
+      "step": 4740
+    },
+    {
+      "epoch": 0.019190601049624875,
+      "grad_norm": 540.2933959960938,
+      "learning_rate": 3.8383838383838385e-06,
+      "loss": 47.5912,
+      "step": 4750
+    },
+    {
+      "epoch": 0.019231002314992505,
+      "grad_norm": 454.8396911621094,
+      "learning_rate": 3.846464646464647e-06,
+      "loss": 49.2231,
+      "step": 4760
+    },
+    {
+      "epoch": 0.019271403580360138,
+      "grad_norm": 538.3920288085938,
+      "learning_rate": 3.8545454545454545e-06,
+      "loss": 56.6818,
+      "step": 4770
+    },
+    {
+      "epoch": 0.019311804845727767,
+      "grad_norm": 567.5830078125,
+      "learning_rate": 3.862626262626263e-06,
+      "loss": 69.7806,
+      "step": 4780
+    },
+    {
+      "epoch": 0.0193522061110954,
+      "grad_norm": 234.95065307617188,
+      "learning_rate": 3.8707070707070706e-06,
+      "loss": 40.516,
+      "step": 4790
+    },
+    {
+      "epoch": 0.01939260737646303,
+      "grad_norm": 947.2410888671875,
+      "learning_rate": 3.878787878787879e-06,
+      "loss": 52.9704,
+      "step": 4800
+    },
+    {
+      "epoch": 0.019433008641830663,
+      "grad_norm": 435.2585754394531,
+      "learning_rate": 3.8868686868686875e-06,
+      "loss": 48.7174,
+      "step": 4810
+    },
+    {
+      "epoch": 0.019473409907198293,
+      "grad_norm": 815.7978515625,
+      "learning_rate": 3.894949494949495e-06,
+      "loss": 55.1616,
+      "step": 4820
+    },
+    {
+      "epoch": 0.019513811172565926,
+      "grad_norm": 1187.990478515625,
+      "learning_rate": 3.9030303030303035e-06,
+      "loss": 67.6024,
+      "step": 4830
+    },
+    {
+      "epoch": 0.019554212437933555,
+      "grad_norm": 1282.373046875,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 50.7462,
+      "step": 4840
+    },
+    {
+      "epoch": 0.01959461370330119,
+      "grad_norm": 662.55908203125,
+      "learning_rate": 3.9191919191919196e-06,
+      "loss": 64.9835,
+      "step": 4850
+    },
+    {
+      "epoch": 0.019635014968668818,
+      "grad_norm": 443.5242614746094,
+      "learning_rate": 3.927272727272727e-06,
+      "loss": 37.5332,
+      "step": 4860
+    },
+    {
+      "epoch": 0.01967541623403645,
+      "grad_norm": 1131.5301513671875,
+      "learning_rate": 3.935353535353536e-06,
+      "loss": 82.1557,
+      "step": 4870
+    },
+    {
+      "epoch": 0.01971581749940408,
+      "grad_norm": 799.3508911132812,
+      "learning_rate": 3.943434343434343e-06,
+      "loss": 51.2066,
+      "step": 4880
+    },
+    {
+      "epoch": 0.019756218764771714,
+      "grad_norm": 510.3349914550781,
+      "learning_rate": 3.951515151515152e-06,
+      "loss": 43.9406,
+      "step": 4890
+    },
+    {
+      "epoch": 0.019796620030139343,
+      "grad_norm": 549.172119140625,
+      "learning_rate": 3.95959595959596e-06,
+      "loss": 65.3976,
+      "step": 4900
+    },
+    {
+      "epoch": 0.019837021295506976,
+      "grad_norm": 763.05224609375,
+      "learning_rate": 3.967676767676768e-06,
+      "loss": 55.32,
+      "step": 4910
+    },
+    {
+      "epoch": 0.019877422560874606,
+      "grad_norm": 427.2909851074219,
+      "learning_rate": 3.975757575757576e-06,
+      "loss": 39.9464,
+      "step": 4920
+    },
+    {
+      "epoch": 0.01991782382624224,
+      "grad_norm": 838.1200561523438,
+      "learning_rate": 3.983838383838385e-06,
+      "loss": 50.7108,
+      "step": 4930
+    },
+    {
+      "epoch": 0.01995822509160987,
+      "grad_norm": 784.1903686523438,
+      "learning_rate": 3.991919191919192e-06,
+      "loss": 75.4968,
+      "step": 4940
+    },
+    {
+      "epoch": 0.0199986263569775,
+      "grad_norm": 336.4251708984375,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 60.2289,
+      "step": 4950
+    },
+    {
+      "epoch": 0.02003902762234513,
+      "grad_norm": 976.969970703125,
+      "learning_rate": 4.008080808080808e-06,
+      "loss": 50.0926,
+      "step": 4960
+    },
+    {
+      "epoch": 0.020079428887712764,
+      "grad_norm": 1279.132568359375,
+      "learning_rate": 4.016161616161616e-06,
+      "loss": 60.1504,
+      "step": 4970
+    },
+    {
+      "epoch": 0.020119830153080394,
+      "grad_norm": 711.7320556640625,
+      "learning_rate": 4.024242424242424e-06,
+      "loss": 49.4476,
+      "step": 4980
+    },
+    {
+      "epoch": 0.020160231418448027,
+      "grad_norm": 423.693603515625,
+      "learning_rate": 4.032323232323233e-06,
+      "loss": 62.753,
+      "step": 4990
+    },
+    {
+      "epoch": 0.020200632683815656,
+      "grad_norm": 613.124267578125,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 49.9839,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02024103394918329,
+      "grad_norm": 984.8370971679688,
+      "learning_rate": 4.048484848484849e-06,
+      "loss": 60.4485,
+      "step": 5010
+    },
+    {
+      "epoch": 0.02028143521455092,
+      "grad_norm": 614.8654174804688,
+      "learning_rate": 4.056565656565657e-06,
+      "loss": 50.103,
+      "step": 5020
+    },
+    {
+      "epoch": 0.020321836479918552,
+      "grad_norm": 766.6077270507812,
+      "learning_rate": 4.064646464646465e-06,
+      "loss": 71.7173,
+      "step": 5030
+    },
+    {
+      "epoch": 0.020362237745286182,
+      "grad_norm": 1153.7392578125,
+      "learning_rate": 4.072727272727273e-06,
+      "loss": 70.789,
+      "step": 5040
+    },
+    {
+      "epoch": 0.020402639010653815,
+      "grad_norm": 788.94189453125,
+      "learning_rate": 4.080808080808081e-06,
+      "loss": 54.5462,
+      "step": 5050
+    },
+    {
+      "epoch": 0.020443040276021444,
+      "grad_norm": 1231.907958984375,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 62.8956,
+      "step": 5060
+    },
+    {
+      "epoch": 0.020483441541389077,
+      "grad_norm": 335.8381652832031,
+      "learning_rate": 4.096969696969697e-06,
+      "loss": 48.22,
+      "step": 5070
+    },
+    {
+      "epoch": 0.020523842806756707,
+      "grad_norm": 2388.536376953125,
+      "learning_rate": 4.105050505050505e-06,
+      "loss": 68.5397,
+      "step": 5080
+    },
+    {
+      "epoch": 0.02056424407212434,
+      "grad_norm": 880.5589599609375,
+      "learning_rate": 4.113131313131313e-06,
+      "loss": 51.2133,
+      "step": 5090
+    },
+    {
+      "epoch": 0.02060464533749197,
+      "grad_norm": 444.1842956542969,
+      "learning_rate": 4.1212121212121215e-06,
+      "loss": 45.3639,
+      "step": 5100
+    },
+    {
+      "epoch": 0.020645046602859603,
+      "grad_norm": 497.71258544921875,
+      "learning_rate": 4.12929292929293e-06,
+      "loss": 65.4038,
+      "step": 5110
+    },
+    {
+      "epoch": 0.020685447868227232,
+      "grad_norm": 620.34912109375,
+      "learning_rate": 4.1373737373737375e-06,
+      "loss": 49.9131,
+      "step": 5120
+    },
+    {
+      "epoch": 0.020725849133594865,
+      "grad_norm": 902.3446655273438,
+      "learning_rate": 4.145454545454546e-06,
+      "loss": 45.9957,
+      "step": 5130
+    },
+    {
+      "epoch": 0.020766250398962495,
+      "grad_norm": 505.1466064453125,
+      "learning_rate": 4.1535353535353536e-06,
+      "loss": 43.1903,
+      "step": 5140
+    },
+    {
+      "epoch": 0.020806651664330128,
+      "grad_norm": 401.0746765136719,
+      "learning_rate": 4.161616161616162e-06,
+      "loss": 59.4582,
+      "step": 5150
+    },
+    {
+      "epoch": 0.020847052929697758,
+      "grad_norm": 622.1741943359375,
+      "learning_rate": 4.1696969696969705e-06,
+      "loss": 42.9341,
+      "step": 5160
+    },
+    {
+      "epoch": 0.02088745419506539,
+      "grad_norm": 633.169921875,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 64.0073,
+      "step": 5170
+    },
+    {
+      "epoch": 0.02092785546043302,
+      "grad_norm": 709.371337890625,
+      "learning_rate": 4.185858585858586e-06,
+      "loss": 55.1911,
+      "step": 5180
+    },
+    {
+      "epoch": 0.020968256725800653,
+      "grad_norm": 536.787841796875,
+      "learning_rate": 4.193939393939394e-06,
+      "loss": 59.7044,
+      "step": 5190
+    },
+    {
+      "epoch": 0.021008657991168283,
+      "grad_norm": 710.5322875976562,
+      "learning_rate": 4.2020202020202026e-06,
+      "loss": 55.8415,
+      "step": 5200
+    },
+    {
+      "epoch": 0.021049059256535916,
+      "grad_norm": 747.7606811523438,
+      "learning_rate": 4.21010101010101e-06,
+      "loss": 46.1033,
+      "step": 5210
+    },
+    {
+      "epoch": 0.021089460521903546,
+      "grad_norm": 470.2503356933594,
+      "learning_rate": 4.218181818181819e-06,
+      "loss": 51.2424,
+      "step": 5220
+    },
+    {
+      "epoch": 0.02112986178727118,
+      "grad_norm": 723.833984375,
+      "learning_rate": 4.226262626262626e-06,
+      "loss": 51.1689,
+      "step": 5230
+    },
+    {
+      "epoch": 0.021170263052638808,
+      "grad_norm": 692.9503173828125,
+      "learning_rate": 4.234343434343435e-06,
+      "loss": 66.355,
+      "step": 5240
+    },
+    {
+      "epoch": 0.02121066431800644,
+      "grad_norm": 3034.588623046875,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 85.8846,
+      "step": 5250
+    },
+    {
+      "epoch": 0.02125106558337407,
+      "grad_norm": 912.686767578125,
+      "learning_rate": 4.250505050505051e-06,
+      "loss": 69.7366,
+      "step": 5260
+    },
+    {
+      "epoch": 0.021291466848741704,
+      "grad_norm": 941.4260864257812,
+      "learning_rate": 4.258585858585859e-06,
+      "loss": 67.8435,
+      "step": 5270
+    },
+    {
+      "epoch": 0.021331868114109333,
+      "grad_norm": 484.5862731933594,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 56.8078,
+      "step": 5280
+    },
+    {
+      "epoch": 0.021372269379476967,
+      "grad_norm": 703.8028564453125,
+      "learning_rate": 4.274747474747475e-06,
+      "loss": 56.7962,
+      "step": 5290
+    },
+    {
+      "epoch": 0.021412670644844596,
+      "grad_norm": 406.5282897949219,
+      "learning_rate": 4.282828282828283e-06,
+      "loss": 50.0921,
+      "step": 5300
+    },
+    {
+      "epoch": 0.02145307191021223,
+      "grad_norm": 909.0770263671875,
+      "learning_rate": 4.290909090909091e-06,
+      "loss": 68.9491,
+      "step": 5310
+    },
+    {
+      "epoch": 0.02149347317557986,
+      "grad_norm": 605.3233642578125,
+      "learning_rate": 4.298989898989899e-06,
+      "loss": 48.6208,
+      "step": 5320
+    },
+    {
+      "epoch": 0.021533874440947492,
+      "grad_norm": 360.8162536621094,
+      "learning_rate": 4.307070707070707e-06,
+      "loss": 67.9922,
+      "step": 5330
+    },
+    {
+      "epoch": 0.02157427570631512,
+      "grad_norm": 547.1395263671875,
+      "learning_rate": 4.315151515151516e-06,
+      "loss": 49.9828,
+      "step": 5340
+    },
+    {
+      "epoch": 0.021614676971682754,
+      "grad_norm": 298.53076171875,
+      "learning_rate": 4.323232323232323e-06,
+      "loss": 51.9223,
+      "step": 5350
+    },
+    {
+      "epoch": 0.021655078237050384,
+      "grad_norm": 239.08287048339844,
+      "learning_rate": 4.331313131313132e-06,
+      "loss": 93.9693,
+      "step": 5360
+    },
+    {
+      "epoch": 0.021695479502418017,
+      "grad_norm": 400.78326416015625,
+      "learning_rate": 4.33939393939394e-06,
+      "loss": 61.8255,
+      "step": 5370
+    },
+    {
+      "epoch": 0.021735880767785647,
+      "grad_norm": 844.6863403320312,
+      "learning_rate": 4.347474747474748e-06,
+      "loss": 80.0776,
+      "step": 5380
+    },
+    {
+      "epoch": 0.02177628203315328,
+      "grad_norm": 1113.2698974609375,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 54.753,
+      "step": 5390
+    },
+    {
+      "epoch": 0.02181668329852091,
+      "grad_norm": 776.0969848632812,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 63.0497,
+      "step": 5400
+    },
+    {
+      "epoch": 0.021857084563888542,
+      "grad_norm": 448.30889892578125,
+      "learning_rate": 4.3717171717171715e-06,
+      "loss": 67.8514,
+      "step": 5410
+    },
+    {
+      "epoch": 0.021897485829256172,
+      "grad_norm": 1007.500732421875,
+      "learning_rate": 4.37979797979798e-06,
+      "loss": 63.2784,
+      "step": 5420
+    },
+    {
+      "epoch": 0.021937887094623805,
+      "grad_norm": 755.4855346679688,
+      "learning_rate": 4.387878787878788e-06,
+      "loss": 64.7918,
+      "step": 5430
+    },
+    {
+      "epoch": 0.021978288359991435,
+      "grad_norm": 468.41009521484375,
+      "learning_rate": 4.395959595959596e-06,
+      "loss": 64.8188,
+      "step": 5440
+    },
+    {
+      "epoch": 0.022018689625359068,
+      "grad_norm": 952.0288696289062,
+      "learning_rate": 4.4040404040404044e-06,
+      "loss": 66.8628,
+      "step": 5450
+    },
+    {
+      "epoch": 0.022059090890726697,
+      "grad_norm": 921.4072875976562,
+      "learning_rate": 4.412121212121213e-06,
+      "loss": 69.8722,
+      "step": 5460
+    },
+    {
+      "epoch": 0.02209949215609433,
+      "grad_norm": 971.798583984375,
+      "learning_rate": 4.4202020202020205e-06,
+      "loss": 45.01,
+      "step": 5470
+    },
+    {
+      "epoch": 0.02213989342146196,
+      "grad_norm": 2802.30419921875,
+      "learning_rate": 4.428282828282829e-06,
+      "loss": 70.787,
+      "step": 5480
+    },
+    {
+      "epoch": 0.022180294686829593,
+      "grad_norm": 528.7456665039062,
+      "learning_rate": 4.436363636363637e-06,
+      "loss": 52.1586,
+      "step": 5490
+    },
+    {
+      "epoch": 0.022220695952197222,
+      "grad_norm": 775.1383056640625,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 62.1388,
+      "step": 5500
+    },
+    {
+      "epoch": 0.022261097217564856,
+      "grad_norm": 653.5880737304688,
+      "learning_rate": 4.452525252525253e-06,
+      "loss": 39.7751,
+      "step": 5510
+    },
+    {
+      "epoch": 0.022301498482932485,
+      "grad_norm": 493.3662109375,
+      "learning_rate": 4.460606060606061e-06,
+      "loss": 49.2763,
+      "step": 5520
+    },
+    {
+      "epoch": 0.022341899748300118,
+      "grad_norm": 717.1765747070312,
+      "learning_rate": 4.468686868686869e-06,
+      "loss": 56.1161,
+      "step": 5530
+    },
+    {
+      "epoch": 0.022382301013667748,
+      "grad_norm": 907.2646484375,
+      "learning_rate": 4.476767676767677e-06,
+      "loss": 52.5763,
+      "step": 5540
+    },
+    {
+      "epoch": 0.02242270227903538,
+      "grad_norm": 680.5430908203125,
+      "learning_rate": 4.4848484848484855e-06,
+      "loss": 64.6257,
+      "step": 5550
+    },
+    {
+      "epoch": 0.02246310354440301,
+      "grad_norm": 587.7493896484375,
+      "learning_rate": 4.492929292929293e-06,
+      "loss": 66.1097,
+      "step": 5560
+    },
+    {
+      "epoch": 0.022503504809770643,
+      "grad_norm": 351.2096252441406,
+      "learning_rate": 4.501010101010102e-06,
+      "loss": 35.0575,
+      "step": 5570
+    },
+    {
+      "epoch": 0.022543906075138273,
+      "grad_norm": 868.7583618164062,
+      "learning_rate": 4.50909090909091e-06,
+      "loss": 52.3459,
+      "step": 5580
+    },
+    {
+      "epoch": 0.022584307340505906,
+      "grad_norm": 413.28155517578125,
+      "learning_rate": 4.517171717171718e-06,
+      "loss": 56.9695,
+      "step": 5590
+    },
+    {
+      "epoch": 0.022624708605873536,
+      "grad_norm": 649.5818481445312,
+      "learning_rate": 4.525252525252526e-06,
+      "loss": 49.7981,
+      "step": 5600
+    },
+    {
+      "epoch": 0.02266510987124117,
+      "grad_norm": 1128.8115234375,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 55.7752,
+      "step": 5610
+    },
+    {
+      "epoch": 0.0227055111366088,
+      "grad_norm": 1171.7386474609375,
+      "learning_rate": 4.541414141414141e-06,
+      "loss": 78.1653,
+      "step": 5620
+    },
+    {
+      "epoch": 0.02274591240197643,
+      "grad_norm": 988.4644775390625,
+      "learning_rate": 4.54949494949495e-06,
+      "loss": 52.3605,
+      "step": 5630
+    },
+    {
+      "epoch": 0.02278631366734406,
+      "grad_norm": 667.2089233398438,
+      "learning_rate": 4.557575757575758e-06,
+      "loss": 58.1055,
+      "step": 5640
+    },
+    {
+      "epoch": 0.022826714932711694,
+      "grad_norm": 1140.114990234375,
+      "learning_rate": 4.565656565656566e-06,
+      "loss": 68.9946,
+      "step": 5650
+    },
+    {
+      "epoch": 0.022867116198079324,
+      "grad_norm": 975.0390014648438,
+      "learning_rate": 4.573737373737374e-06,
+      "loss": 65.694,
+      "step": 5660
+    },
+    {
+      "epoch": 0.022907517463446957,
+      "grad_norm": 974.479736328125,
+      "learning_rate": 4.581818181818183e-06,
+      "loss": 48.7883,
+      "step": 5670
+    },
+    {
+      "epoch": 0.022947918728814586,
+      "grad_norm": 474.9225158691406,
+      "learning_rate": 4.58989898989899e-06,
+      "loss": 47.1188,
+      "step": 5680
+    },
+    {
+      "epoch": 0.02298831999418222,
+      "grad_norm": 1234.3985595703125,
+      "learning_rate": 4.597979797979799e-06,
+      "loss": 76.4564,
+      "step": 5690
+    },
+    {
+      "epoch": 0.02302872125954985,
+      "grad_norm": 937.4863891601562,
+      "learning_rate": 4.606060606060606e-06,
+      "loss": 53.0915,
+      "step": 5700
+    },
+    {
+      "epoch": 0.023069122524917482,
+      "grad_norm": 687.7938842773438,
+      "learning_rate": 4.614141414141414e-06,
+      "loss": 58.9133,
+      "step": 5710
+    },
+    {
+      "epoch": 0.02310952379028511,
+      "grad_norm": 445.2677001953125,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 47.943,
+      "step": 5720
+    },
+    {
+      "epoch": 0.023149925055652745,
+      "grad_norm": 380.9918212890625,
+      "learning_rate": 4.630303030303031e-06,
+      "loss": 54.5093,
+      "step": 5730
+    },
+    {
+      "epoch": 0.023190326321020374,
+      "grad_norm": 399.1734924316406,
+      "learning_rate": 4.6383838383838384e-06,
+      "loss": 60.505,
+      "step": 5740
+    },
+    {
+      "epoch": 0.023230727586388007,
+      "grad_norm": 1005.51953125,
+      "learning_rate": 4.646464646464647e-06,
+      "loss": 53.1637,
+      "step": 5750
+    },
+    {
+      "epoch": 0.023271128851755637,
+      "grad_norm": 1141.5557861328125,
+      "learning_rate": 4.654545454545455e-06,
+      "loss": 85.439,
+      "step": 5760
+    },
+    {
+      "epoch": 0.02331153011712327,
+      "grad_norm": 719.6810913085938,
+      "learning_rate": 4.662626262626263e-06,
+      "loss": 53.3294,
+      "step": 5770
+    },
+    {
+      "epoch": 0.0233519313824909,
+      "grad_norm": 804.4564208984375,
+      "learning_rate": 4.670707070707071e-06,
+      "loss": 58.7554,
+      "step": 5780
+    },
+    {
+      "epoch": 0.023392332647858533,
+      "grad_norm": 738.3641967773438,
+      "learning_rate": 4.678787878787879e-06,
+      "loss": 58.9601,
+      "step": 5790
+    },
+    {
+      "epoch": 0.023432733913226162,
+      "grad_norm": 442.52630615234375,
+      "learning_rate": 4.6868686868686874e-06,
+      "loss": 58.0687,
+      "step": 5800
+    },
+    {
+      "epoch": 0.023473135178593795,
+      "grad_norm": 684.3228759765625,
+      "learning_rate": 4.694949494949496e-06,
+      "loss": 57.9766,
+      "step": 5810
+    },
+    {
+      "epoch": 0.023513536443961425,
+      "grad_norm": 1048.6868896484375,
+      "learning_rate": 4.7030303030303035e-06,
+      "loss": 65.6569,
+      "step": 5820
+    },
+    {
+      "epoch": 0.023553937709329058,
+      "grad_norm": 417.07696533203125,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 49.7707,
+      "step": 5830
+    },
+    {
+      "epoch": 0.023594338974696687,
+      "grad_norm": 1276.1221923828125,
+      "learning_rate": 4.7191919191919195e-06,
+      "loss": 51.5833,
+      "step": 5840
+    },
+    {
+      "epoch": 0.02363474024006432,
+      "grad_norm": 1004.6600341796875,
+      "learning_rate": 4.727272727272728e-06,
+      "loss": 84.6636,
+      "step": 5850
+    },
+    {
+      "epoch": 0.02367514150543195,
+      "grad_norm": 709.0186157226562,
+      "learning_rate": 4.735353535353536e-06,
+      "loss": 62.0867,
+      "step": 5860
+    },
+    {
+      "epoch": 0.023715542770799583,
+      "grad_norm": 817.0234375,
+      "learning_rate": 4.743434343434344e-06,
+      "loss": 52.1775,
+      "step": 5870
+    },
+    {
+      "epoch": 0.023755944036167213,
+      "grad_norm": 246.22628784179688,
+      "learning_rate": 4.751515151515152e-06,
+      "loss": 62.8754,
+      "step": 5880
+    },
+    {
+      "epoch": 0.023796345301534846,
+      "grad_norm": 927.6898193359375,
+      "learning_rate": 4.75959595959596e-06,
+      "loss": 54.2966,
+      "step": 5890
+    },
+    {
+      "epoch": 0.023836746566902475,
+      "grad_norm": 1191.2314453125,
+      "learning_rate": 4.7676767676767685e-06,
+      "loss": 102.11,
+      "step": 5900
+    },
+    {
+      "epoch": 0.02387714783227011,
+      "grad_norm": 573.662109375,
+      "learning_rate": 4.775757575757576e-06,
+      "loss": 63.7849,
+      "step": 5910
+    },
+    {
+      "epoch": 0.023917549097637738,
+      "grad_norm": 677.00732421875,
+      "learning_rate": 4.783838383838385e-06,
+      "loss": 32.899,
+      "step": 5920
+    },
+    {
+      "epoch": 0.02395795036300537,
+      "grad_norm": 739.2052001953125,
+      "learning_rate": 4.791919191919192e-06,
+      "loss": 42.3564,
+      "step": 5930
+    },
+    {
+      "epoch": 0.023998351628373,
+      "grad_norm": 591.5802001953125,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 54.1494,
+      "step": 5940
+    },
+    {
+      "epoch": 0.024038752893740634,
+      "grad_norm": 676.9740600585938,
+      "learning_rate": 4.808080808080808e-06,
+      "loss": 51.5359,
+      "step": 5950
+    },
+    {
+      "epoch": 0.024079154159108263,
+      "grad_norm": 688.4370727539062,
+      "learning_rate": 4.816161616161617e-06,
+      "loss": 65.8977,
+      "step": 5960
+    },
+    {
+      "epoch": 0.024119555424475896,
+      "grad_norm": 881.8679809570312,
+      "learning_rate": 4.824242424242424e-06,
+      "loss": 55.0991,
+      "step": 5970
+    },
+    {
+      "epoch": 0.024159956689843526,
+      "grad_norm": 994.0008544921875,
+      "learning_rate": 4.832323232323233e-06,
+      "loss": 59.3008,
+      "step": 5980
+    },
+    {
+      "epoch": 0.02420035795521116,
+      "grad_norm": 653.2454223632812,
+      "learning_rate": 4.840404040404041e-06,
+      "loss": 27.4822,
+      "step": 5990
+    },
+    {
+      "epoch": 0.02424075922057879,
+      "grad_norm": 288.7889404296875,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 50.8106,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02428116048594642,
+      "grad_norm": 615.9229125976562,
+      "learning_rate": 4.856565656565657e-06,
+      "loss": 53.6111,
+      "step": 6010
+    },
+    {
+      "epoch": 0.02432156175131405,
+      "grad_norm": 957.2933349609375,
+      "learning_rate": 4.864646464646466e-06,
+      "loss": 58.5743,
+      "step": 6020
+    },
+    {
+      "epoch": 0.024361963016681684,
+      "grad_norm": 1033.9827880859375,
+      "learning_rate": 4.872727272727273e-06,
+      "loss": 69.179,
+      "step": 6030
+    },
+    {
+      "epoch": 0.024402364282049314,
+      "grad_norm": 730.6578979492188,
+      "learning_rate": 4.880808080808081e-06,
+      "loss": 46.347,
+      "step": 6040
+    },
+    {
+      "epoch": 0.024442765547416947,
+      "grad_norm": 446.7236022949219,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 46.9909,
+      "step": 6050
+    },
+    {
+      "epoch": 0.024483166812784576,
+      "grad_norm": 411.31451416015625,
+      "learning_rate": 4.896969696969697e-06,
+      "loss": 50.5173,
+      "step": 6060
+    },
+    {
+      "epoch": 0.02452356807815221,
+      "grad_norm": 962.7041625976562,
+      "learning_rate": 4.905050505050505e-06,
+      "loss": 66.097,
+      "step": 6070
+    },
+    {
+      "epoch": 0.02456396934351984,
+      "grad_norm": 950.1348876953125,
+      "learning_rate": 4.913131313131314e-06,
+      "loss": 66.6417,
+      "step": 6080
+    },
+    {
+      "epoch": 0.02460437060888747,
+      "grad_norm": 651.8181762695312,
+      "learning_rate": 4.9212121212121214e-06,
+      "loss": 46.0087,
+      "step": 6090
+    },
+    {
+      "epoch": 0.0246447718742551,
+      "grad_norm": 925.5591430664062,
+      "learning_rate": 4.92929292929293e-06,
+      "loss": 60.3818,
+      "step": 6100
+    },
+    {
+      "epoch": 0.02468517313962273,
+      "grad_norm": 683.6544799804688,
+      "learning_rate": 4.937373737373738e-06,
+      "loss": 61.0053,
+      "step": 6110
+    },
+    {
+      "epoch": 0.024725574404990364,
+      "grad_norm": 627.9014892578125,
+      "learning_rate": 4.945454545454546e-06,
+      "loss": 66.3312,
+      "step": 6120
+    },
+    {
+      "epoch": 0.024765975670357994,
+      "grad_norm": 835.2926025390625,
+      "learning_rate": 4.953535353535354e-06,
+      "loss": 64.1998,
+      "step": 6130
+    },
+    {
+      "epoch": 0.024806376935725627,
+      "grad_norm": 979.1827392578125,
+      "learning_rate": 4.961616161616162e-06,
+      "loss": 62.5498,
+      "step": 6140
+    },
+    {
+      "epoch": 0.024846778201093257,
+      "grad_norm": 491.91552734375,
+      "learning_rate": 4.9696969696969696e-06,
+      "loss": 59.011,
+      "step": 6150
+    },
+    {
+      "epoch": 0.02488717946646089,
+      "grad_norm": 494.9542541503906,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 76.8512,
+      "step": 6160
+    },
+    {
+      "epoch": 0.02492758073182852,
+      "grad_norm": 627.1669311523438,
+      "learning_rate": 4.9858585858585865e-06,
+      "loss": 64.9759,
+      "step": 6170
+    },
+    {
+      "epoch": 0.024967981997196152,
+      "grad_norm": 668.6937866210938,
+      "learning_rate": 4.993939393939394e-06,
+      "loss": 64.4498,
+      "step": 6180
+    },
+    {
+      "epoch": 0.025008383262563782,
+      "grad_norm": 561.9140014648438,
+      "learning_rate": 5.0020202020202025e-06,
+      "loss": 49.3964,
+      "step": 6190
+    },
+    {
+      "epoch": 0.025048784527931415,
+      "grad_norm": 549.7935180664062,
+      "learning_rate": 5.010101010101011e-06,
+      "loss": 64.5368,
+      "step": 6200
+    },
+    {
+      "epoch": 0.025089185793299044,
+      "grad_norm": 2008.2236328125,
+      "learning_rate": 5.0181818181818186e-06,
+      "loss": 53.3066,
+      "step": 6210
+    },
+    {
+      "epoch": 0.025129587058666678,
+      "grad_norm": 431.6338806152344,
+      "learning_rate": 5.026262626262627e-06,
+      "loss": 78.1808,
+      "step": 6220
+    },
+    {
+      "epoch": 0.025169988324034307,
+      "grad_norm": 277.0874328613281,
+      "learning_rate": 5.034343434343435e-06,
+      "loss": 51.0862,
+      "step": 6230
+    },
+    {
+      "epoch": 0.02521038958940194,
+      "grad_norm": 601.8732299804688,
+      "learning_rate": 5.042424242424243e-06,
+      "loss": 43.3031,
+      "step": 6240
+    },
+    {
+      "epoch": 0.02525079085476957,
+      "grad_norm": 357.4873352050781,
+      "learning_rate": 5.0505050505050515e-06,
+      "loss": 71.1351,
+      "step": 6250
+    },
+    {
+      "epoch": 0.025291192120137203,
+      "grad_norm": 666.0474243164062,
+      "learning_rate": 5.058585858585859e-06,
+      "loss": 48.8687,
+      "step": 6260
+    },
+    {
+      "epoch": 0.025331593385504832,
+      "grad_norm": 2769.57568359375,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 85.213,
+      "step": 6270
+    },
+    {
+      "epoch": 0.025371994650872465,
+      "grad_norm": 1185.662841796875,
+      "learning_rate": 5.074747474747476e-06,
+      "loss": 64.6072,
+      "step": 6280
+    },
+    {
+      "epoch": 0.025412395916240095,
+      "grad_norm": 593.606689453125,
+      "learning_rate": 5.082828282828284e-06,
+      "loss": 45.3778,
+      "step": 6290
+    },
+    {
+      "epoch": 0.025452797181607728,
+      "grad_norm": 685.597900390625,
+      "learning_rate": 5.090909090909091e-06,
+      "loss": 63.415,
+      "step": 6300
+    },
+    {
+      "epoch": 0.025493198446975358,
+      "grad_norm": 779.54150390625,
+      "learning_rate": 5.098989898989899e-06,
+      "loss": 42.4922,
+      "step": 6310
+    },
+    {
+      "epoch": 0.02553359971234299,
+      "grad_norm": 748.0287475585938,
+      "learning_rate": 5.107070707070707e-06,
+      "loss": 54.5869,
+      "step": 6320
+    },
+    {
+      "epoch": 0.02557400097771062,
+      "grad_norm": 1047.96630859375,
+      "learning_rate": 5.115151515151515e-06,
+      "loss": 71.8537,
+      "step": 6330
+    },
+    {
+      "epoch": 0.025614402243078253,
+      "grad_norm": 481.7173156738281,
+      "learning_rate": 5.123232323232323e-06,
+      "loss": 53.4357,
+      "step": 6340
+    },
+    {
+      "epoch": 0.025654803508445883,
+      "grad_norm": 469.4927673339844,
+      "learning_rate": 5.131313131313132e-06,
+      "loss": 55.997,
+      "step": 6350
+    },
+    {
+      "epoch": 0.025695204773813516,
+      "grad_norm": 882.4183349609375,
+      "learning_rate": 5.139393939393939e-06,
+      "loss": 61.6789,
+      "step": 6360
+    },
+    {
+      "epoch": 0.025735606039181146,
+      "grad_norm": 710.2070922851562,
+      "learning_rate": 5.147474747474748e-06,
+      "loss": 50.3809,
+      "step": 6370
+    },
+    {
+      "epoch": 0.02577600730454878,
+      "grad_norm": 979.2042846679688,
+      "learning_rate": 5.155555555555556e-06,
+      "loss": 57.65,
+      "step": 6380
+    },
+    {
+      "epoch": 0.025816408569916408,
+      "grad_norm": 3295.435302734375,
+      "learning_rate": 5.163636363636364e-06,
+      "loss": 95.8132,
+      "step": 6390
+    },
+    {
+      "epoch": 0.02585680983528404,
+      "grad_norm": 612.2908935546875,
+      "learning_rate": 5.171717171717172e-06,
+      "loss": 44.9435,
+      "step": 6400
+    },
+    {
+      "epoch": 0.02589721110065167,
+      "grad_norm": 163.25262451171875,
+      "learning_rate": 5.17979797979798e-06,
+      "loss": 43.0925,
+      "step": 6410
+    },
+    {
+      "epoch": 0.025937612366019304,
+      "grad_norm": 727.7279663085938,
+      "learning_rate": 5.187878787878788e-06,
+      "loss": 50.2254,
+      "step": 6420
+    },
+    {
+      "epoch": 0.025978013631386934,
+      "grad_norm": 561.295654296875,
+      "learning_rate": 5.195959595959597e-06,
+      "loss": 55.9661,
+      "step": 6430
+    },
+    {
+      "epoch": 0.026018414896754567,
+      "grad_norm": 1451.63916015625,
+      "learning_rate": 5.204040404040404e-06,
+      "loss": 69.4304,
+      "step": 6440
+    },
+    {
+      "epoch": 0.026058816162122196,
+      "grad_norm": 1265.6790771484375,
+      "learning_rate": 5.212121212121213e-06,
+      "loss": 66.7605,
+      "step": 6450
+    },
+    {
+      "epoch": 0.02609921742748983,
+      "grad_norm": 646.0961303710938,
+      "learning_rate": 5.220202020202021e-06,
+      "loss": 45.969,
+      "step": 6460
+    },
+    {
+      "epoch": 0.02613961869285746,
+      "grad_norm": 676.3788452148438,
+      "learning_rate": 5.228282828282829e-06,
+      "loss": 56.7987,
+      "step": 6470
+    },
+    {
+      "epoch": 0.026180019958225092,
+      "grad_norm": 510.5483093261719,
+      "learning_rate": 5.236363636363637e-06,
+      "loss": 62.0078,
+      "step": 6480
+    },
+    {
+      "epoch": 0.02622042122359272,
+      "grad_norm": 1271.3424072265625,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 53.2245,
+      "step": 6490
+    },
+    {
+      "epoch": 0.026260822488960354,
+      "grad_norm": 408.92352294921875,
+      "learning_rate": 5.252525252525253e-06,
+      "loss": 61.5336,
+      "step": 6500
+    },
+    {
+      "epoch": 0.026301223754327984,
+      "grad_norm": 644.672119140625,
+      "learning_rate": 5.26060606060606e-06,
+      "loss": 54.1448,
+      "step": 6510
+    },
+    {
+      "epoch": 0.026341625019695617,
+      "grad_norm": 477.9904479980469,
+      "learning_rate": 5.268686868686869e-06,
+      "loss": 71.9779,
+      "step": 6520
+    },
+    {
+      "epoch": 0.026382026285063247,
+      "grad_norm": 963.8379516601562,
+      "learning_rate": 5.276767676767677e-06,
+      "loss": 77.2526,
+      "step": 6530
+    },
+    {
+      "epoch": 0.02642242755043088,
+      "grad_norm": 1079.2596435546875,
+      "learning_rate": 5.284848484848485e-06,
+      "loss": 55.9919,
+      "step": 6540
+    },
+    {
+      "epoch": 0.02646282881579851,
+      "grad_norm": 949.0852661132812,
+      "learning_rate": 5.292929292929293e-06,
+      "loss": 48.6409,
+      "step": 6550
+    },
+    {
+      "epoch": 0.026503230081166142,
+      "grad_norm": 684.5231323242188,
+      "learning_rate": 5.3010101010101016e-06,
+      "loss": 75.0948,
+      "step": 6560
+    },
+    {
+      "epoch": 0.026543631346533772,
+      "grad_norm": 534.61474609375,
+      "learning_rate": 5.309090909090909e-06,
+      "loss": 43.8272,
+      "step": 6570
+    },
+    {
+      "epoch": 0.026584032611901405,
+      "grad_norm": 593.8630981445312,
+      "learning_rate": 5.317171717171718e-06,
+      "loss": 61.2505,
+      "step": 6580
+    },
+    {
+      "epoch": 0.026624433877269035,
+      "grad_norm": 500.44586181640625,
+      "learning_rate": 5.325252525252525e-06,
+      "loss": 36.6023,
+      "step": 6590
+    },
+    {
+      "epoch": 0.026664835142636668,
+      "grad_norm": 534.3717651367188,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 60.64,
+      "step": 6600
+    },
+    {
+      "epoch": 0.026705236408004297,
+      "grad_norm": 948.6107788085938,
+      "learning_rate": 5.341414141414142e-06,
+      "loss": 57.0894,
+      "step": 6610
+    },
+    {
+      "epoch": 0.02674563767337193,
+      "grad_norm": 684.484619140625,
+      "learning_rate": 5.34949494949495e-06,
+      "loss": 53.3891,
+      "step": 6620
+    },
+    {
+      "epoch": 0.02678603893873956,
+      "grad_norm": 613.2217407226562,
+      "learning_rate": 5.357575757575758e-06,
+      "loss": 71.4286,
+      "step": 6630
+    },
+    {
+      "epoch": 0.026826440204107193,
+      "grad_norm": 568.6145629882812,
+      "learning_rate": 5.365656565656567e-06,
+      "loss": 50.2748,
+      "step": 6640
+    },
+    {
+      "epoch": 0.026866841469474823,
+      "grad_norm": 398.4335632324219,
+      "learning_rate": 5.373737373737374e-06,
+      "loss": 42.2858,
+      "step": 6650
+    },
+    {
+      "epoch": 0.026907242734842456,
+      "grad_norm": 712.74755859375,
+      "learning_rate": 5.381818181818183e-06,
+      "loss": 44.9092,
+      "step": 6660
+    },
+    {
+      "epoch": 0.026947644000210085,
+      "grad_norm": 739.2601318359375,
+      "learning_rate": 5.38989898989899e-06,
+      "loss": 51.0345,
+      "step": 6670
+    },
+    {
+      "epoch": 0.026988045265577718,
+      "grad_norm": 663.2088623046875,
+      "learning_rate": 5.397979797979799e-06,
+      "loss": 42.3987,
+      "step": 6680
+    },
+    {
+      "epoch": 0.027028446530945348,
+      "grad_norm": 567.5929565429688,
+      "learning_rate": 5.406060606060607e-06,
+      "loss": 43.0136,
+      "step": 6690
+    },
+    {
+      "epoch": 0.02706884779631298,
+      "grad_norm": 1150.5491943359375,
+      "learning_rate": 5.414141414141415e-06,
+      "loss": 76.9021,
+      "step": 6700
+    },
+    {
+      "epoch": 0.02710924906168061,
+      "grad_norm": 568.4838256835938,
+      "learning_rate": 5.422222222222223e-06,
+      "loss": 69.4932,
+      "step": 6710
+    },
+    {
+      "epoch": 0.027149650327048244,
+      "grad_norm": 589.513671875,
+      "learning_rate": 5.430303030303032e-06,
+      "loss": 60.2286,
+      "step": 6720
+    },
+    {
+      "epoch": 0.027190051592415873,
+      "grad_norm": 948.9300537109375,
+      "learning_rate": 5.438383838383838e-06,
+      "loss": 81.8324,
+      "step": 6730
+    },
+    {
+      "epoch": 0.027230452857783506,
+      "grad_norm": 470.6671447753906,
+      "learning_rate": 5.446464646464647e-06,
+      "loss": 56.1316,
+      "step": 6740
+    },
+    {
+      "epoch": 0.027270854123151136,
+      "grad_norm": 1675.1475830078125,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 73.5525,
+      "step": 6750
+    },
+    {
+      "epoch": 0.02731125538851877,
+      "grad_norm": 754.5110473632812,
+      "learning_rate": 5.462626262626263e-06,
+      "loss": 61.0356,
+      "step": 6760
+    },
+    {
+      "epoch": 0.0273516566538864,
+      "grad_norm": 430.7125244140625,
+      "learning_rate": 5.4707070707070705e-06,
+      "loss": 53.5754,
+      "step": 6770
+    },
+    {
+      "epoch": 0.02739205791925403,
+      "grad_norm": 847.1322631835938,
+      "learning_rate": 5.478787878787879e-06,
+      "loss": 65.6341,
+      "step": 6780
+    },
+    {
+      "epoch": 0.02743245918462166,
+      "grad_norm": 977.4957275390625,
+      "learning_rate": 5.486868686868687e-06,
+      "loss": 70.9679,
+      "step": 6790
+    },
+    {
+      "epoch": 0.027472860449989294,
+      "grad_norm": 2050.4521484375,
+      "learning_rate": 5.494949494949495e-06,
+      "loss": 99.4739,
+      "step": 6800
+    },
+    {
+      "epoch": 0.027513261715356924,
+      "grad_norm": 1008.5648193359375,
+      "learning_rate": 5.5030303030303034e-06,
+      "loss": 58.5655,
+      "step": 6810
+    },
+    {
+      "epoch": 0.027553662980724557,
+      "grad_norm": 401.62506103515625,
+      "learning_rate": 5.511111111111112e-06,
+      "loss": 61.3661,
+      "step": 6820
+    },
+    {
+      "epoch": 0.027594064246092186,
+      "grad_norm": 997.3637084960938,
+      "learning_rate": 5.5191919191919195e-06,
+      "loss": 49.5676,
+      "step": 6830
+    },
+    {
+      "epoch": 0.02763446551145982,
+      "grad_norm": 576.8031616210938,
+      "learning_rate": 5.527272727272728e-06,
+      "loss": 41.5905,
+      "step": 6840
+    },
+    {
+      "epoch": 0.02767486677682745,
+      "grad_norm": 550.9244384765625,
+      "learning_rate": 5.5353535353535355e-06,
+      "loss": 61.1382,
+      "step": 6850
+    },
+    {
+      "epoch": 0.027715268042195082,
+      "grad_norm": 626.9200439453125,
+      "learning_rate": 5.543434343434344e-06,
+      "loss": 53.2902,
+      "step": 6860
+    },
+    {
+      "epoch": 0.02775566930756271,
+      "grad_norm": 752.208984375,
+      "learning_rate": 5.5515151515151524e-06,
+      "loss": 57.3701,
+      "step": 6870
+    },
+    {
+      "epoch": 0.027796070572930345,
+      "grad_norm": 772.6936645507812,
+      "learning_rate": 5.55959595959596e-06,
+      "loss": 60.7988,
+      "step": 6880
+    },
+    {
+      "epoch": 0.027836471838297974,
+      "grad_norm": 533.5892944335938,
+      "learning_rate": 5.5676767676767685e-06,
+      "loss": 60.7892,
+      "step": 6890
+    },
+    {
+      "epoch": 0.027876873103665607,
+      "grad_norm": 756.2522583007812,
+      "learning_rate": 5.575757575757577e-06,
+      "loss": 48.367,
+      "step": 6900
+    },
+    {
+      "epoch": 0.027917274369033237,
+      "grad_norm": 600.169921875,
+      "learning_rate": 5.5838383838383845e-06,
+      "loss": 57.2744,
+      "step": 6910
+    },
+    {
+      "epoch": 0.02795767563440087,
+      "grad_norm": 800.9529418945312,
+      "learning_rate": 5.591919191919193e-06,
+      "loss": 56.2487,
+      "step": 6920
+    },
+    {
+      "epoch": 0.0279980768997685,
+      "grad_norm": 691.3702392578125,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 42.3235,
+      "step": 6930
+    },
+    {
+      "epoch": 0.028038478165136133,
+      "grad_norm": 625.58642578125,
+      "learning_rate": 5.608080808080808e-06,
+      "loss": 51.2433,
+      "step": 6940
+    },
+    {
+      "epoch": 0.028078879430503762,
+      "grad_norm": 591.58837890625,
+      "learning_rate": 5.616161616161616e-06,
+      "loss": 56.8397,
+      "step": 6950
+    },
+    {
+      "epoch": 0.028119280695871395,
+      "grad_norm": 828.33935546875,
+      "learning_rate": 5.624242424242424e-06,
+      "loss": 40.0593,
+      "step": 6960
+    },
+    {
+      "epoch": 0.028159681961239025,
+      "grad_norm": 0.0,
+      "learning_rate": 5.632323232323233e-06,
+      "loss": 32.3671,
+      "step": 6970
+    },
+    {
+      "epoch": 0.028200083226606658,
+      "grad_norm": 1996.341796875,
+      "learning_rate": 5.64040404040404e-06,
+      "loss": 85.5505,
+      "step": 6980
+    },
+    {
+      "epoch": 0.028240484491974287,
+      "grad_norm": 596.9549560546875,
+      "learning_rate": 5.648484848484849e-06,
+      "loss": 52.3962,
+      "step": 6990
+    },
+    {
+      "epoch": 0.02828088575734192,
+      "grad_norm": 404.708740234375,
+      "learning_rate": 5.656565656565657e-06,
+      "loss": 47.1833,
+      "step": 7000
+    },
+    {
+      "epoch": 0.02832128702270955,
+      "grad_norm": 912.1199340820312,
+      "learning_rate": 5.664646464646465e-06,
+      "loss": 55.7482,
+      "step": 7010
+    },
+    {
+      "epoch": 0.028361688288077183,
+      "grad_norm": 833.735107421875,
+      "learning_rate": 5.672727272727273e-06,
+      "loss": 81.9062,
+      "step": 7020
+    },
+    {
+      "epoch": 0.028402089553444813,
+      "grad_norm": 486.2171630859375,
+      "learning_rate": 5.680808080808081e-06,
+      "loss": 44.0852,
+      "step": 7030
+    },
+    {
+      "epoch": 0.028442490818812446,
+      "grad_norm": 755.2271728515625,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 52.617,
+      "step": 7040
+    },
+    {
+      "epoch": 0.028482892084180075,
+      "grad_norm": 734.7620849609375,
+      "learning_rate": 5.696969696969698e-06,
+      "loss": 63.221,
+      "step": 7050
+    },
+    {
+      "epoch": 0.02852329334954771,
+      "grad_norm": 699.6790771484375,
+      "learning_rate": 5.705050505050505e-06,
+      "loss": 48.4481,
+      "step": 7060
+    },
+    {
+      "epoch": 0.028563694614915338,
+      "grad_norm": 819.5504760742188,
+      "learning_rate": 5.713131313131314e-06,
+      "loss": 51.6106,
+      "step": 7070
+    },
+    {
+      "epoch": 0.02860409588028297,
+      "grad_norm": 597.2007446289062,
+      "learning_rate": 5.721212121212122e-06,
+      "loss": 61.2867,
+      "step": 7080
+    },
+    {
+      "epoch": 0.0286444971456506,
+      "grad_norm": 970.8818969726562,
+      "learning_rate": 5.72929292929293e-06,
+      "loss": 47.3044,
+      "step": 7090
+    },
+    {
+      "epoch": 0.028684898411018234,
+      "grad_norm": 459.5702819824219,
+      "learning_rate": 5.737373737373738e-06,
+      "loss": 57.9656,
+      "step": 7100
+    },
+    {
+      "epoch": 0.028725299676385863,
+      "grad_norm": 581.9468383789062,
+      "learning_rate": 5.745454545454546e-06,
+      "loss": 47.3413,
+      "step": 7110
+    },
+    {
+      "epoch": 0.028765700941753496,
+      "grad_norm": 467.80316162109375,
+      "learning_rate": 5.753535353535354e-06,
+      "loss": 67.7175,
+      "step": 7120
+    },
+    {
+      "epoch": 0.028806102207121126,
+      "grad_norm": 797.3532104492188,
+      "learning_rate": 5.761616161616163e-06,
+      "loss": 47.0771,
+      "step": 7130
+    },
+    {
+      "epoch": 0.02884650347248876,
+      "grad_norm": 725.4122314453125,
+      "learning_rate": 5.76969696969697e-06,
+      "loss": 76.4885,
+      "step": 7140
+    },
+    {
+      "epoch": 0.02888690473785639,
+      "grad_norm": 1118.2633056640625,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 63.7839,
+      "step": 7150
+    },
+    {
+      "epoch": 0.02892730600322402,
+      "grad_norm": 626.3515014648438,
+      "learning_rate": 5.785858585858586e-06,
+      "loss": 74.4487,
+      "step": 7160
+    },
+    {
+      "epoch": 0.02896770726859165,
+      "grad_norm": 886.15234375,
+      "learning_rate": 5.793939393939394e-06,
+      "loss": 45.6227,
+      "step": 7170
+    },
+    {
+      "epoch": 0.029008108533959284,
+      "grad_norm": 1027.4027099609375,
+      "learning_rate": 5.8020202020202025e-06,
+      "loss": 68.2278,
+      "step": 7180
+    },
+    {
+      "epoch": 0.029048509799326914,
+      "grad_norm": 725.6253051757812,
+      "learning_rate": 5.81010101010101e-06,
+      "loss": 60.2867,
+      "step": 7190
+    },
+    {
+      "epoch": 0.029088911064694547,
+      "grad_norm": 689.9947509765625,
+      "learning_rate": 5.8181818181818185e-06,
+      "loss": 37.6077,
+      "step": 7200
+    },
+    {
+      "epoch": 0.029129312330062176,
+      "grad_norm": 1027.0823974609375,
+      "learning_rate": 5.826262626262626e-06,
+      "loss": 50.7096,
+      "step": 7210
+    },
+    {
+      "epoch": 0.02916971359542981,
+      "grad_norm": 226.7436981201172,
+      "learning_rate": 5.834343434343435e-06,
+      "loss": 64.8211,
+      "step": 7220
+    },
+    {
+      "epoch": 0.02921011486079744,
+      "grad_norm": 675.5316162109375,
+      "learning_rate": 5.842424242424243e-06,
+      "loss": 63.2848,
+      "step": 7230
+    },
+    {
+      "epoch": 0.029250516126165072,
+      "grad_norm": 564.1721801757812,
+      "learning_rate": 5.850505050505051e-06,
+      "loss": 49.1681,
+      "step": 7240
+    },
+    {
+      "epoch": 0.029290917391532702,
+      "grad_norm": 891.8140869140625,
+      "learning_rate": 5.858585858585859e-06,
+      "loss": 58.443,
+      "step": 7250
+    },
+    {
+      "epoch": 0.029331318656900335,
+      "grad_norm": 782.2761840820312,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 48.1456,
+      "step": 7260
+    },
+    {
+      "epoch": 0.029371719922267964,
+      "grad_norm": 1056.0394287109375,
+      "learning_rate": 5.874747474747475e-06,
+      "loss": 48.6662,
+      "step": 7270
+    },
+    {
+      "epoch": 0.029412121187635597,
+      "grad_norm": 440.3533630371094,
+      "learning_rate": 5.882828282828284e-06,
+      "loss": 66.0266,
+      "step": 7280
+    },
+    {
+      "epoch": 0.029452522453003227,
+      "grad_norm": 819.3785400390625,
+      "learning_rate": 5.890909090909091e-06,
+      "loss": 70.9092,
+      "step": 7290
+    },
+    {
+      "epoch": 0.02949292371837086,
+      "grad_norm": 1346.183349609375,
+      "learning_rate": 5.8989898989899e-06,
+      "loss": 54.89,
+      "step": 7300
+    },
+    {
+      "epoch": 0.02953332498373849,
+      "grad_norm": 1040.2781982421875,
+      "learning_rate": 5.907070707070708e-06,
+      "loss": 52.2293,
+      "step": 7310
+    },
+    {
+      "epoch": 0.029573726249106123,
+      "grad_norm": 609.7448120117188,
+      "learning_rate": 5.915151515151516e-06,
+      "loss": 46.7856,
+      "step": 7320
+    },
+    {
+      "epoch": 0.029614127514473752,
+      "grad_norm": 2038.486328125,
+      "learning_rate": 5.923232323232324e-06,
+      "loss": 60.1004,
+      "step": 7330
+    },
+    {
+      "epoch": 0.029654528779841385,
+      "grad_norm": 626.9315185546875,
+      "learning_rate": 5.9313131313131326e-06,
+      "loss": 62.3969,
+      "step": 7340
+    },
+    {
+      "epoch": 0.029694930045209015,
+      "grad_norm": 464.0664978027344,
+      "learning_rate": 5.93939393939394e-06,
+      "loss": 46.8368,
+      "step": 7350
+    },
+    {
+      "epoch": 0.029735331310576648,
+      "grad_norm": 830.5698852539062,
+      "learning_rate": 5.947474747474749e-06,
+      "loss": 47.1381,
+      "step": 7360
+    },
+    {
+      "epoch": 0.029775732575944278,
+      "grad_norm": 924.8589477539062,
+      "learning_rate": 5.955555555555555e-06,
+      "loss": 42.5227,
+      "step": 7370
+    },
+    {
+      "epoch": 0.02981613384131191,
+      "grad_norm": 627.990478515625,
+      "learning_rate": 5.963636363636364e-06,
+      "loss": 47.545,
+      "step": 7380
+    },
+    {
+      "epoch": 0.02985653510667954,
+      "grad_norm": 1128.0264892578125,
+      "learning_rate": 5.9717171717171714e-06,
+      "loss": 65.1939,
+      "step": 7390
+    },
+    {
+      "epoch": 0.029896936372047173,
+      "grad_norm": 680.4015502929688,
+      "learning_rate": 5.97979797979798e-06,
+      "loss": 77.6617,
+      "step": 7400
+    },
+    {
+      "epoch": 0.029937337637414803,
+      "grad_norm": 1081.4503173828125,
+      "learning_rate": 5.987878787878788e-06,
+      "loss": 55.418,
+      "step": 7410
+    },
+    {
+      "epoch": 0.029977738902782436,
+      "grad_norm": 789.0983276367188,
+      "learning_rate": 5.995959595959596e-06,
+      "loss": 71.1661,
+      "step": 7420
+    },
+    {
+      "epoch": 0.030018140168150065,
+      "grad_norm": 1137.5018310546875,
+      "learning_rate": 6.004040404040404e-06,
+      "loss": 54.8057,
+      "step": 7430
+    },
+    {
+      "epoch": 0.0300585414335177,
+      "grad_norm": 728.9630737304688,
+      "learning_rate": 6.012121212121213e-06,
+      "loss": 45.3661,
+      "step": 7440
+    },
+    {
+      "epoch": 0.030098942698885328,
+      "grad_norm": 769.5979614257812,
+      "learning_rate": 6.0202020202020204e-06,
+      "loss": 58.8569,
+      "step": 7450
+    },
+    {
+      "epoch": 0.03013934396425296,
+      "grad_norm": 620.3338012695312,
+      "learning_rate": 6.028282828282829e-06,
+      "loss": 48.3413,
+      "step": 7460
+    },
+    {
+      "epoch": 0.03017974522962059,
+      "grad_norm": 1105.1793212890625,
+      "learning_rate": 6.0363636363636365e-06,
+      "loss": 49.1405,
+      "step": 7470
+    },
+    {
+      "epoch": 0.030220146494988224,
+      "grad_norm": 873.118896484375,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 74.7576,
+      "step": 7480
+    },
+    {
+      "epoch": 0.030260547760355853,
+      "grad_norm": 657.729248046875,
+      "learning_rate": 6.052525252525253e-06,
+      "loss": 59.2427,
+      "step": 7490
+    },
+    {
+      "epoch": 0.030300949025723486,
+      "grad_norm": 1126.5247802734375,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 62.4204,
+      "step": 7500
+    },
+    {
+      "epoch": 0.030341350291091116,
+      "grad_norm": 751.9967041015625,
+      "learning_rate": 6.068686868686869e-06,
+      "loss": 46.9996,
+      "step": 7510
+    },
+    {
+      "epoch": 0.03038175155645875,
+      "grad_norm": 620.5112915039062,
+      "learning_rate": 6.076767676767678e-06,
+      "loss": 50.917,
+      "step": 7520
+    },
+    {
+      "epoch": 0.03042215282182638,
+      "grad_norm": 566.5572509765625,
+      "learning_rate": 6.0848484848484855e-06,
+      "loss": 63.4834,
+      "step": 7530
+    },
+    {
+      "epoch": 0.030462554087194012,
+      "grad_norm": 622.100830078125,
+      "learning_rate": 6.092929292929294e-06,
+      "loss": 63.3863,
+      "step": 7540
+    },
+    {
+      "epoch": 0.03050295535256164,
+      "grad_norm": 990.3203125,
+      "learning_rate": 6.1010101010101015e-06,
+      "loss": 41.6357,
+      "step": 7550
+    },
+    {
+      "epoch": 0.030543356617929274,
+      "grad_norm": 746.3700561523438,
+      "learning_rate": 6.10909090909091e-06,
+      "loss": 50.8004,
+      "step": 7560
+    },
+    {
+      "epoch": 0.030583757883296904,
+      "grad_norm": 684.2229614257812,
+      "learning_rate": 6.117171717171718e-06,
+      "loss": 41.9685,
+      "step": 7570
+    },
+    {
+      "epoch": 0.030624159148664537,
+      "grad_norm": 642.2493896484375,
+      "learning_rate": 6.125252525252525e-06,
+      "loss": 61.6354,
+      "step": 7580
+    },
+    {
+      "epoch": 0.030664560414032167,
+      "grad_norm": 554.5538330078125,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 87.8135,
+      "step": 7590
+    },
+    {
+      "epoch": 0.0307049616793998,
+      "grad_norm": 1042.8443603515625,
+      "learning_rate": 6.141414141414141e-06,
+      "loss": 62.6359,
+      "step": 7600
+    },
+    {
+      "epoch": 0.03074536294476743,
+      "grad_norm": 473.3791198730469,
+      "learning_rate": 6.14949494949495e-06,
+      "loss": 57.5903,
+      "step": 7610
+    },
+    {
+      "epoch": 0.030785764210135062,
+      "grad_norm": 466.8446960449219,
+      "learning_rate": 6.157575757575758e-06,
+      "loss": 60.731,
+      "step": 7620
+    },
+    {
+      "epoch": 0.030826165475502692,
+      "grad_norm": 891.0825805664062,
+      "learning_rate": 6.165656565656566e-06,
+      "loss": 60.5271,
+      "step": 7630
+    },
+    {
+      "epoch": 0.030866566740870325,
+      "grad_norm": 262.8304138183594,
+      "learning_rate": 6.173737373737374e-06,
+      "loss": 55.9255,
+      "step": 7640
+    },
+    {
+      "epoch": 0.030906968006237955,
+      "grad_norm": 946.0525512695312,
+      "learning_rate": 6.181818181818182e-06,
+      "loss": 53.6557,
+      "step": 7650
+    },
+    {
+      "epoch": 0.030947369271605588,
+      "grad_norm": 556.9738159179688,
+      "learning_rate": 6.18989898989899e-06,
+      "loss": 50.9775,
+      "step": 7660
+    },
+    {
+      "epoch": 0.030987770536973217,
+      "grad_norm": 555.402099609375,
+      "learning_rate": 6.197979797979799e-06,
+      "loss": 60.4395,
+      "step": 7670
+    },
+    {
+      "epoch": 0.03102817180234085,
+      "grad_norm": 961.5715942382812,
+      "learning_rate": 6.206060606060606e-06,
+      "loss": 74.3546,
+      "step": 7680
+    },
+    {
+      "epoch": 0.03106857306770848,
+      "grad_norm": 2129.817626953125,
+      "learning_rate": 6.214141414141415e-06,
+      "loss": 83.1414,
+      "step": 7690
+    },
+    {
+      "epoch": 0.031108974333076113,
+      "grad_norm": 704.78564453125,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 49.4473,
+      "step": 7700
+    },
+    {
+      "epoch": 0.031149375598443742,
+      "grad_norm": 881.543701171875,
+      "learning_rate": 6.230303030303031e-06,
+      "loss": 67.96,
+      "step": 7710
+    },
+    {
+      "epoch": 0.031189776863811376,
+      "grad_norm": 648.9303588867188,
+      "learning_rate": 6.238383838383839e-06,
+      "loss": 52.5546,
+      "step": 7720
+    },
+    {
+      "epoch": 0.031230178129179005,
+      "grad_norm": 561.0586547851562,
+      "learning_rate": 6.246464646464647e-06,
+      "loss": 59.6244,
+      "step": 7730
+    },
+    {
+      "epoch": 0.03127057939454664,
+      "grad_norm": 509.1589050292969,
+      "learning_rate": 6.254545454545455e-06,
+      "loss": 45.3432,
+      "step": 7740
+    },
+    {
+      "epoch": 0.03131098065991427,
+      "grad_norm": 648.4699096679688,
+      "learning_rate": 6.262626262626264e-06,
+      "loss": 67.8853,
+      "step": 7750
+    },
+    {
+      "epoch": 0.0313513819252819,
+      "grad_norm": 1544.9857177734375,
+      "learning_rate": 6.270707070707071e-06,
+      "loss": 55.752,
+      "step": 7760
+    },
+    {
+      "epoch": 0.03139178319064953,
+      "grad_norm": 654.1318969726562,
+      "learning_rate": 6.27878787878788e-06,
+      "loss": 51.3248,
+      "step": 7770
+    },
+    {
+      "epoch": 0.03143218445601716,
+      "grad_norm": 464.28411865234375,
+      "learning_rate": 6.286868686868688e-06,
+      "loss": 58.6119,
+      "step": 7780
+    },
+    {
+      "epoch": 0.031472585721384796,
+      "grad_norm": 543.3377075195312,
+      "learning_rate": 6.294949494949495e-06,
+      "loss": 63.7197,
+      "step": 7790
+    },
+    {
+      "epoch": 0.03151298698675242,
+      "grad_norm": 1178.969970703125,
+      "learning_rate": 6.303030303030303e-06,
+      "loss": 56.3949,
+      "step": 7800
+    },
+    {
+      "epoch": 0.031553388252120056,
+      "grad_norm": 534.0205688476562,
+      "learning_rate": 6.311111111111111e-06,
+      "loss": 41.2432,
+      "step": 7810
+    },
+    {
+      "epoch": 0.03159378951748769,
+      "grad_norm": 965.5466918945312,
+      "learning_rate": 6.3191919191919195e-06,
+      "loss": 56.5476,
+      "step": 7820
+    },
+    {
+      "epoch": 0.03163419078285532,
+      "grad_norm": 429.6799011230469,
+      "learning_rate": 6.327272727272727e-06,
+      "loss": 49.7253,
+      "step": 7830
+    },
+    {
+      "epoch": 0.03167459204822295,
+      "grad_norm": 1150.05126953125,
+      "learning_rate": 6.3353535353535355e-06,
+      "loss": 55.4475,
+      "step": 7840
+    },
+    {
+      "epoch": 0.03171499331359058,
+      "grad_norm": 996.5704956054688,
+      "learning_rate": 6.343434343434344e-06,
+      "loss": 53.3563,
+      "step": 7850
+    },
+    {
+      "epoch": 0.031755394578958214,
+      "grad_norm": 896.9970092773438,
+      "learning_rate": 6.3515151515151516e-06,
+      "loss": 51.2789,
+      "step": 7860
+    },
+    {
+      "epoch": 0.03179579584432585,
+      "grad_norm": 833.8233642578125,
+      "learning_rate": 6.35959595959596e-06,
+      "loss": 71.7647,
+      "step": 7870
+    },
+    {
+      "epoch": 0.03183619710969347,
+      "grad_norm": 2155.960205078125,
+      "learning_rate": 6.3676767676767685e-06,
+      "loss": 67.7663,
+      "step": 7880
+    },
+    {
+      "epoch": 0.031876598375061106,
+      "grad_norm": 685.2928466796875,
+      "learning_rate": 6.375757575757576e-06,
+      "loss": 42.8484,
+      "step": 7890
+    },
+    {
+      "epoch": 0.03191699964042874,
+      "grad_norm": 633.02978515625,
+      "learning_rate": 6.3838383838383845e-06,
+      "loss": 46.8481,
+      "step": 7900
+    },
+    {
+      "epoch": 0.03195740090579637,
+      "grad_norm": 413.897216796875,
+      "learning_rate": 6.391919191919192e-06,
+      "loss": 69.5535,
+      "step": 7910
+    },
+    {
+      "epoch": 0.031997802171164,
+      "grad_norm": 367.6637878417969,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 62.3668,
+      "step": 7920
+    },
+    {
+      "epoch": 0.03203820343653163,
+      "grad_norm": 580.1402587890625,
+      "learning_rate": 6.408080808080809e-06,
+      "loss": 58.8996,
+      "step": 7930
+    },
+    {
+      "epoch": 0.032078604701899265,
+      "grad_norm": 779.8268432617188,
+      "learning_rate": 6.416161616161617e-06,
+      "loss": 49.8146,
+      "step": 7940
+    },
+    {
+      "epoch": 0.0321190059672669,
+      "grad_norm": 712.888427734375,
+      "learning_rate": 6.424242424242425e-06,
+      "loss": 61.1355,
+      "step": 7950
+    },
+    {
+      "epoch": 0.032159407232634524,
+      "grad_norm": 760.9315185546875,
+      "learning_rate": 6.4323232323232335e-06,
+      "loss": 96.3001,
+      "step": 7960
+    },
+    {
+      "epoch": 0.03219980849800216,
+      "grad_norm": 949.9454345703125,
+      "learning_rate": 6.440404040404041e-06,
+      "loss": 54.5059,
+      "step": 7970
+    },
+    {
+      "epoch": 0.03224020976336979,
+      "grad_norm": 1105.3018798828125,
+      "learning_rate": 6.4484848484848496e-06,
+      "loss": 86.9141,
+      "step": 7980
+    },
+    {
+      "epoch": 0.03228061102873742,
+      "grad_norm": 622.3108520507812,
+      "learning_rate": 6.456565656565658e-06,
+      "loss": 52.3702,
+      "step": 7990
+    },
+    {
+      "epoch": 0.03232101229410505,
+      "grad_norm": 511.1920166015625,
+      "learning_rate": 6.464646464646466e-06,
+      "loss": 65.486,
+      "step": 8000
+    },
+    {
+      "epoch": 0.03236141355947268,
+      "grad_norm": 787.6452026367188,
+      "learning_rate": 6.472727272727272e-06,
+      "loss": 63.4064,
+      "step": 8010
+    },
+    {
+      "epoch": 0.032401814824840315,
+      "grad_norm": 641.1221923828125,
+      "learning_rate": 6.480808080808081e-06,
+      "loss": 67.2346,
+      "step": 8020
+    },
+    {
+      "epoch": 0.03244221609020795,
+      "grad_norm": 849.4371948242188,
+      "learning_rate": 6.488888888888889e-06,
+      "loss": 74.775,
+      "step": 8030
+    },
+    {
+      "epoch": 0.032482617355575574,
+      "grad_norm": 281.5003662109375,
+      "learning_rate": 6.496969696969697e-06,
+      "loss": 48.3455,
+      "step": 8040
+    },
+    {
+      "epoch": 0.03252301862094321,
+      "grad_norm": 635.1185302734375,
+      "learning_rate": 6.505050505050505e-06,
+      "loss": 58.5666,
+      "step": 8050
+    },
+    {
+      "epoch": 0.03256341988631084,
+      "grad_norm": 930.7824096679688,
+      "learning_rate": 6.513131313131314e-06,
+      "loss": 73.5296,
+      "step": 8060
+    },
+    {
+      "epoch": 0.03260382115167847,
+      "grad_norm": 582.126708984375,
+      "learning_rate": 6.521212121212121e-06,
+      "loss": 34.6027,
+      "step": 8070
+    },
+    {
+      "epoch": 0.0326442224170461,
+      "grad_norm": 562.6720581054688,
+      "learning_rate": 6.52929292929293e-06,
+      "loss": 60.8401,
+      "step": 8080
+    },
+    {
+      "epoch": 0.03268462368241373,
+      "grad_norm": 306.123046875,
+      "learning_rate": 6.537373737373737e-06,
+      "loss": 44.4158,
+      "step": 8090
+    },
+    {
+      "epoch": 0.032725024947781366,
+      "grad_norm": 991.8218994140625,
+      "learning_rate": 6.545454545454546e-06,
+      "loss": 45.6267,
+      "step": 8100
+    },
+    {
+      "epoch": 0.032765426213149,
+      "grad_norm": 835.2438354492188,
+      "learning_rate": 6.553535353535354e-06,
+      "loss": 53.1318,
+      "step": 8110
+    },
+    {
+      "epoch": 0.032805827478516625,
+      "grad_norm": 1006.3707885742188,
+      "learning_rate": 6.561616161616162e-06,
+      "loss": 53.602,
+      "step": 8120
+    },
+    {
+      "epoch": 0.03284622874388426,
+      "grad_norm": 350.9178466796875,
+      "learning_rate": 6.56969696969697e-06,
+      "loss": 49.0478,
+      "step": 8130
+    },
+    {
+      "epoch": 0.03288663000925189,
+      "grad_norm": 542.7490844726562,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 42.3374,
+      "step": 8140
+    },
+    {
+      "epoch": 0.032927031274619524,
+      "grad_norm": 453.57965087890625,
+      "learning_rate": 6.585858585858586e-06,
+      "loss": 50.3855,
+      "step": 8150
+    },
+    {
+      "epoch": 0.03296743253998715,
+      "grad_norm": 887.0755615234375,
+      "learning_rate": 6.593939393939395e-06,
+      "loss": 43.7275,
+      "step": 8160
+    },
+    {
+      "epoch": 0.03300783380535478,
+      "grad_norm": 388.6866760253906,
+      "learning_rate": 6.602020202020203e-06,
+      "loss": 44.9985,
+      "step": 8170
+    },
+    {
+      "epoch": 0.033048235070722416,
+      "grad_norm": 672.9983520507812,
+      "learning_rate": 6.610101010101011e-06,
+      "loss": 53.5537,
+      "step": 8180
+    },
+    {
+      "epoch": 0.03308863633609005,
+      "grad_norm": 634.128173828125,
+      "learning_rate": 6.618181818181819e-06,
+      "loss": 52.5657,
+      "step": 8190
+    },
+    {
+      "epoch": 0.033129037601457675,
+      "grad_norm": 735.7168579101562,
+      "learning_rate": 6.626262626262627e-06,
+      "loss": 61.5097,
+      "step": 8200
+    },
+    {
+      "epoch": 0.03316943886682531,
+      "grad_norm": 506.9273376464844,
+      "learning_rate": 6.634343434343435e-06,
+      "loss": 59.6884,
+      "step": 8210
+    },
+    {
+      "epoch": 0.03320984013219294,
+      "grad_norm": 568.7894287109375,
+      "learning_rate": 6.642424242424242e-06,
+      "loss": 54.3135,
+      "step": 8220
+    },
+    {
+      "epoch": 0.033250241397560575,
+      "grad_norm": 916.6472778320312,
+      "learning_rate": 6.650505050505051e-06,
+      "loss": 46.6363,
+      "step": 8230
+    },
+    {
+      "epoch": 0.0332906426629282,
+      "grad_norm": 511.55718994140625,
+      "learning_rate": 6.658585858585859e-06,
+      "loss": 45.2044,
+      "step": 8240
+    },
+    {
+      "epoch": 0.033331043928295834,
+      "grad_norm": 680.4920654296875,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 51.8027,
+      "step": 8250
+    },
+    {
+      "epoch": 0.03337144519366347,
+      "grad_norm": 524.2860107421875,
+      "learning_rate": 6.674747474747475e-06,
+      "loss": 46.2791,
+      "step": 8260
+    },
+    {
+      "epoch": 0.0334118464590311,
+      "grad_norm": 1008.0855712890625,
+      "learning_rate": 6.682828282828283e-06,
+      "loss": 70.7169,
+      "step": 8270
+    },
+    {
+      "epoch": 0.033452247724398726,
+      "grad_norm": 586.5045776367188,
+      "learning_rate": 6.690909090909091e-06,
+      "loss": 77.9191,
+      "step": 8280
+    },
+    {
+      "epoch": 0.03349264898976636,
+      "grad_norm": 613.2836303710938,
+      "learning_rate": 6.6989898989899e-06,
+      "loss": 63.8899,
+      "step": 8290
+    },
+    {
+      "epoch": 0.03353305025513399,
+      "grad_norm": 339.6265869140625,
+      "learning_rate": 6.707070707070707e-06,
+      "loss": 71.7176,
+      "step": 8300
+    },
+    {
+      "epoch": 0.033573451520501625,
+      "grad_norm": 341.3080749511719,
+      "learning_rate": 6.715151515151516e-06,
+      "loss": 49.7896,
+      "step": 8310
+    },
+    {
+      "epoch": 0.03361385278586925,
+      "grad_norm": 1058.944580078125,
+      "learning_rate": 6.723232323232324e-06,
+      "loss": 61.7188,
+      "step": 8320
+    },
+    {
+      "epoch": 0.033654254051236884,
+      "grad_norm": 392.35272216796875,
+      "learning_rate": 6.731313131313132e-06,
+      "loss": 54.2229,
+      "step": 8330
+    },
+    {
+      "epoch": 0.03369465531660452,
+      "grad_norm": 725.1533813476562,
+      "learning_rate": 6.73939393939394e-06,
+      "loss": 48.29,
+      "step": 8340
+    },
+    {
+      "epoch": 0.03373505658197215,
+      "grad_norm": 741.072265625,
+      "learning_rate": 6.747474747474749e-06,
+      "loss": 64.0867,
+      "step": 8350
+    },
+    {
+      "epoch": 0.033775457847339777,
+      "grad_norm": 450.0721130371094,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 49.962,
+      "step": 8360
+    },
+    {
+      "epoch": 0.03381585911270741,
+      "grad_norm": 967.4086303710938,
+      "learning_rate": 6.763636363636365e-06,
+      "loss": 46.306,
+      "step": 8370
+    },
+    {
+      "epoch": 0.03385626037807504,
+      "grad_norm": 537.5621337890625,
+      "learning_rate": 6.771717171717172e-06,
+      "loss": 29.0172,
+      "step": 8380
+    },
+    {
+      "epoch": 0.033896661643442676,
+      "grad_norm": 1187.185546875,
+      "learning_rate": 6.779797979797981e-06,
+      "loss": 67.4143,
+      "step": 8390
+    },
+    {
+      "epoch": 0.0339370629088103,
+      "grad_norm": 893.6007690429688,
+      "learning_rate": 6.787878787878789e-06,
+      "loss": 39.529,
+      "step": 8400
+    },
+    {
+      "epoch": 0.033977464174177935,
+      "grad_norm": 713.060546875,
+      "learning_rate": 6.795959595959597e-06,
+      "loss": 55.9818,
+      "step": 8410
+    },
+    {
+      "epoch": 0.03401786543954557,
+      "grad_norm": 406.208251953125,
+      "learning_rate": 6.804040404040405e-06,
+      "loss": 49.2516,
+      "step": 8420
+    },
+    {
+      "epoch": 0.0340582667049132,
+      "grad_norm": 644.3673095703125,
+      "learning_rate": 6.812121212121212e-06,
+      "loss": 41.1035,
+      "step": 8430
+    },
+    {
+      "epoch": 0.03409866797028083,
+      "grad_norm": 759.6295166015625,
+      "learning_rate": 6.82020202020202e-06,
+      "loss": 58.1698,
+      "step": 8440
+    },
+    {
+      "epoch": 0.03413906923564846,
+      "grad_norm": 564.6557006835938,
+      "learning_rate": 6.828282828282828e-06,
+      "loss": 31.1043,
+      "step": 8450
+    },
+    {
+      "epoch": 0.03417947050101609,
+      "grad_norm": 787.0310668945312,
+      "learning_rate": 6.8363636363636364e-06,
+      "loss": 50.7432,
+      "step": 8460
+    },
+    {
+      "epoch": 0.034219871766383726,
+      "grad_norm": 373.0039367675781,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 37.4919,
+      "step": 8470
+    },
+    {
+      "epoch": 0.03426027303175135,
+      "grad_norm": 747.8372192382812,
+      "learning_rate": 6.8525252525252525e-06,
+      "loss": 63.8979,
+      "step": 8480
+    },
+    {
+      "epoch": 0.034300674297118985,
+      "grad_norm": 407.1216125488281,
+      "learning_rate": 6.860606060606061e-06,
+      "loss": 69.6129,
+      "step": 8490
+    },
+    {
+      "epoch": 0.03434107556248662,
+      "grad_norm": 70.61812591552734,
+      "learning_rate": 6.868686868686869e-06,
+      "loss": 78.9777,
+      "step": 8500
+    },
+    {
+      "epoch": 0.03438147682785425,
+      "grad_norm": 599.4298095703125,
+      "learning_rate": 6.876767676767677e-06,
+      "loss": 68.5256,
+      "step": 8510
+    },
+    {
+      "epoch": 0.03442187809322188,
+      "grad_norm": 726.5167236328125,
+      "learning_rate": 6.8848484848484854e-06,
+      "loss": 89.8193,
+      "step": 8520
+    },
+    {
+      "epoch": 0.03446227935858951,
+      "grad_norm": 612.718505859375,
+      "learning_rate": 6.892929292929294e-06,
+      "loss": 50.8533,
+      "step": 8530
+    },
+    {
+      "epoch": 0.034502680623957144,
+      "grad_norm": 1743.5341796875,
+      "learning_rate": 6.9010101010101015e-06,
+      "loss": 50.107,
+      "step": 8540
+    },
+    {
+      "epoch": 0.03454308188932478,
+      "grad_norm": 702.13671875,
+      "learning_rate": 6.90909090909091e-06,
+      "loss": 74.444,
+      "step": 8550
+    },
+    {
+      "epoch": 0.0345834831546924,
+      "grad_norm": 875.4811401367188,
+      "learning_rate": 6.9171717171717175e-06,
+      "loss": 44.2529,
+      "step": 8560
+    },
+    {
+      "epoch": 0.034623884420060036,
+      "grad_norm": 697.1666870117188,
+      "learning_rate": 6.925252525252526e-06,
+      "loss": 87.5896,
+      "step": 8570
+    },
+    {
+      "epoch": 0.03466428568542767,
+      "grad_norm": 751.409423828125,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 46.6455,
+      "step": 8580
+    },
+    {
+      "epoch": 0.0347046869507953,
+      "grad_norm": 2989.8251953125,
+      "learning_rate": 6.941414141414142e-06,
+      "loss": 58.6519,
+      "step": 8590
+    },
+    {
+      "epoch": 0.03474508821616293,
+      "grad_norm": 701.8047485351562,
+      "learning_rate": 6.9494949494949505e-06,
+      "loss": 89.5045,
+      "step": 8600
+    },
+    {
+      "epoch": 0.03478548948153056,
+      "grad_norm": 460.89923095703125,
+      "learning_rate": 6.957575757575759e-06,
+      "loss": 85.3149,
+      "step": 8610
+    },
+    {
+      "epoch": 0.034825890746898194,
+      "grad_norm": 860.1593627929688,
+      "learning_rate": 6.9656565656565665e-06,
+      "loss": 45.1932,
+      "step": 8620
+    },
+    {
+      "epoch": 0.03486629201226583,
+      "grad_norm": 987.288330078125,
+      "learning_rate": 6.973737373737375e-06,
+      "loss": 64.3151,
+      "step": 8630
+    },
+    {
+      "epoch": 0.034906693277633453,
+      "grad_norm": 1131.6240234375,
+      "learning_rate": 6.981818181818183e-06,
+      "loss": 61.3024,
+      "step": 8640
+    },
+    {
+      "epoch": 0.03494709454300109,
+      "grad_norm": 621.4078979492188,
+      "learning_rate": 6.98989898989899e-06,
+      "loss": 43.6811,
+      "step": 8650
+    },
+    {
+      "epoch": 0.03498749580836872,
+      "grad_norm": 588.7035522460938,
+      "learning_rate": 6.997979797979798e-06,
+      "loss": 52.8447,
+      "step": 8660
+    },
+    {
+      "epoch": 0.03502789707373635,
+      "grad_norm": 549.5409545898438,
+      "learning_rate": 7.006060606060606e-06,
+      "loss": 54.9615,
+      "step": 8670
+    },
+    {
+      "epoch": 0.03506829833910398,
+      "grad_norm": 900.7230834960938,
+      "learning_rate": 7.014141414141415e-06,
+      "loss": 36.6513,
+      "step": 8680
+    },
+    {
+      "epoch": 0.03510869960447161,
+      "grad_norm": 306.8768310546875,
+      "learning_rate": 7.022222222222222e-06,
+      "loss": 39.3492,
+      "step": 8690
+    },
+    {
+      "epoch": 0.035149100869839245,
+      "grad_norm": 581.7448120117188,
+      "learning_rate": 7.030303030303031e-06,
+      "loss": 62.2904,
+      "step": 8700
+    },
+    {
+      "epoch": 0.03518950213520688,
+      "grad_norm": 781.9658813476562,
+      "learning_rate": 7.038383838383839e-06,
+      "loss": 53.9546,
+      "step": 8710
+    },
+    {
+      "epoch": 0.035229903400574504,
+      "grad_norm": 751.5908813476562,
+      "learning_rate": 7.046464646464647e-06,
+      "loss": 54.8535,
+      "step": 8720
+    },
+    {
+      "epoch": 0.03527030466594214,
+      "grad_norm": 1163.4237060546875,
+      "learning_rate": 7.054545454545455e-06,
+      "loss": 55.7728,
+      "step": 8730
+    },
+    {
+      "epoch": 0.03531070593130977,
+      "grad_norm": 1224.25634765625,
+      "learning_rate": 7.062626262626263e-06,
+      "loss": 70.6655,
+      "step": 8740
+    },
+    {
+      "epoch": 0.0353511071966774,
+      "grad_norm": 679.2452392578125,
+      "learning_rate": 7.070707070707071e-06,
+      "loss": 53.1984,
+      "step": 8750
+    },
+    {
+      "epoch": 0.03539150846204503,
+      "grad_norm": 1009.7100830078125,
+      "learning_rate": 7.07878787878788e-06,
+      "loss": 53.5107,
+      "step": 8760
+    },
+    {
+      "epoch": 0.03543190972741266,
+      "grad_norm": 429.02496337890625,
+      "learning_rate": 7.086868686868687e-06,
+      "loss": 50.9108,
+      "step": 8770
+    },
+    {
+      "epoch": 0.035472310992780295,
+      "grad_norm": 705.5595092773438,
+      "learning_rate": 7.094949494949496e-06,
+      "loss": 39.2461,
+      "step": 8780
+    },
+    {
+      "epoch": 0.03551271225814793,
+      "grad_norm": 774.8412475585938,
+      "learning_rate": 7.103030303030304e-06,
+      "loss": 58.4899,
+      "step": 8790
+    },
+    {
+      "epoch": 0.035553113523515555,
+      "grad_norm": 746.4476318359375,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 49.5306,
+      "step": 8800
+    },
+    {
+      "epoch": 0.03559351478888319,
+      "grad_norm": 1140.700927734375,
+      "learning_rate": 7.11919191919192e-06,
+      "loss": 98.9858,
+      "step": 8810
+    },
+    {
+      "epoch": 0.03563391605425082,
+      "grad_norm": 822.6146240234375,
+      "learning_rate": 7.127272727272728e-06,
+      "loss": 40.658,
+      "step": 8820
+    },
+    {
+      "epoch": 0.035674317319618454,
+      "grad_norm": 305.0389709472656,
+      "learning_rate": 7.135353535353536e-06,
+      "loss": 65.953,
+      "step": 8830
+    },
+    {
+      "epoch": 0.03571471858498608,
+      "grad_norm": 983.6098022460938,
+      "learning_rate": 7.143434343434345e-06,
+      "loss": 55.4964,
+      "step": 8840
+    },
+    {
+      "epoch": 0.03575511985035371,
+      "grad_norm": 801.6506958007812,
+      "learning_rate": 7.151515151515152e-06,
+      "loss": 52.7301,
+      "step": 8850
+    },
+    {
+      "epoch": 0.035795521115721346,
+      "grad_norm": 346.94573974609375,
+      "learning_rate": 7.15959595959596e-06,
+      "loss": 55.3774,
+      "step": 8860
+    },
+    {
+      "epoch": 0.03583592238108898,
+      "grad_norm": 660.070556640625,
+      "learning_rate": 7.1676767676767676e-06,
+      "loss": 46.3488,
+      "step": 8870
+    },
+    {
+      "epoch": 0.035876323646456605,
+      "grad_norm": 420.3671569824219,
+      "learning_rate": 7.175757575757576e-06,
+      "loss": 57.9034,
+      "step": 8880
+    },
+    {
+      "epoch": 0.03591672491182424,
+      "grad_norm": 734.4971313476562,
+      "learning_rate": 7.1838383838383845e-06,
+      "loss": 51.1084,
+      "step": 8890
+    },
+    {
+      "epoch": 0.03595712617719187,
+      "grad_norm": 622.8480224609375,
+      "learning_rate": 7.191919191919192e-06,
+      "loss": 63.6594,
+      "step": 8900
+    },
+    {
+      "epoch": 0.035997527442559504,
+      "grad_norm": 1083.3779296875,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 48.5284,
+      "step": 8910
+    },
+    {
+      "epoch": 0.03603792870792713,
+      "grad_norm": 795.3521118164062,
+      "learning_rate": 7.208080808080808e-06,
+      "loss": 60.9741,
+      "step": 8920
+    },
+    {
+      "epoch": 0.036078329973294763,
+      "grad_norm": 763.7685546875,
+      "learning_rate": 7.2161616161616166e-06,
+      "loss": 67.9574,
+      "step": 8930
+    },
+    {
+      "epoch": 0.0361187312386624,
+      "grad_norm": 299.5180969238281,
+      "learning_rate": 7.224242424242425e-06,
+      "loss": 44.8309,
+      "step": 8940
+    },
+    {
+      "epoch": 0.03615913250403003,
+      "grad_norm": 793.3358154296875,
+      "learning_rate": 7.232323232323233e-06,
+      "loss": 57.0138,
+      "step": 8950
+    },
+    {
+      "epoch": 0.036199533769397656,
+      "grad_norm": 698.6807861328125,
+      "learning_rate": 7.240404040404041e-06,
+      "loss": 50.4727,
+      "step": 8960
+    },
+    {
+      "epoch": 0.03623993503476529,
+      "grad_norm": 1897.521484375,
+      "learning_rate": 7.2484848484848495e-06,
+      "loss": 66.5708,
+      "step": 8970
+    },
+    {
+      "epoch": 0.03628033630013292,
+      "grad_norm": 390.52337646484375,
+      "learning_rate": 7.256565656565657e-06,
+      "loss": 32.5167,
+      "step": 8980
+    },
+    {
+      "epoch": 0.036320737565500555,
+      "grad_norm": 1146.5927734375,
+      "learning_rate": 7.2646464646464656e-06,
+      "loss": 74.7776,
+      "step": 8990
+    },
+    {
+      "epoch": 0.03636113883086818,
+      "grad_norm": 698.8363037109375,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 52.9729,
+      "step": 9000
+    },
+    {
+      "epoch": 0.036401540096235814,
+      "grad_norm": 787.6583862304688,
+      "learning_rate": 7.280808080808082e-06,
+      "loss": 58.5929,
+      "step": 9010
+    },
+    {
+      "epoch": 0.03644194136160345,
+      "grad_norm": 1728.040771484375,
+      "learning_rate": 7.28888888888889e-06,
+      "loss": 58.7803,
+      "step": 9020
+    },
+    {
+      "epoch": 0.03648234262697108,
+      "grad_norm": 504.9664001464844,
+      "learning_rate": 7.296969696969698e-06,
+      "loss": 28.0724,
+      "step": 9030
+    },
+    {
+      "epoch": 0.036522743892338706,
+      "grad_norm": 1262.524169921875,
+      "learning_rate": 7.305050505050506e-06,
+      "loss": 67.8205,
+      "step": 9040
+    },
+    {
+      "epoch": 0.03656314515770634,
+      "grad_norm": 425.6719970703125,
+      "learning_rate": 7.3131313131313146e-06,
+      "loss": 39.1797,
+      "step": 9050
+    },
+    {
+      "epoch": 0.03660354642307397,
+      "grad_norm": 1008.1500244140625,
+      "learning_rate": 7.321212121212122e-06,
+      "loss": 70.5822,
+      "step": 9060
+    },
+    {
+      "epoch": 0.036643947688441605,
+      "grad_norm": 1135.142333984375,
+      "learning_rate": 7.32929292929293e-06,
+      "loss": 63.7957,
+      "step": 9070
+    },
+    {
+      "epoch": 0.03668434895380923,
+      "grad_norm": 610.756103515625,
+      "learning_rate": 7.337373737373737e-06,
+      "loss": 48.8291,
+      "step": 9080
+    },
+    {
+      "epoch": 0.036724750219176865,
+      "grad_norm": 295.875,
+      "learning_rate": 7.345454545454546e-06,
+      "loss": 42.3916,
+      "step": 9090
+    },
+    {
+      "epoch": 0.0367651514845445,
+      "grad_norm": 814.1820068359375,
+      "learning_rate": 7.353535353535353e-06,
+      "loss": 81.7657,
+      "step": 9100
+    },
+    {
+      "epoch": 0.03680555274991213,
+      "grad_norm": 800.7321166992188,
+      "learning_rate": 7.361616161616162e-06,
+      "loss": 38.7879,
+      "step": 9110
+    },
+    {
+      "epoch": 0.03684595401527976,
+      "grad_norm": 1052.885009765625,
+      "learning_rate": 7.36969696969697e-06,
+      "loss": 51.3777,
+      "step": 9120
+    },
+    {
+      "epoch": 0.03688635528064739,
+      "grad_norm": 571.583984375,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 43.6872,
+      "step": 9130
+    },
+    {
+      "epoch": 0.03692675654601502,
+      "grad_norm": 452.5201110839844,
+      "learning_rate": 7.385858585858586e-06,
+      "loss": 45.5216,
+      "step": 9140
+    },
+    {
+      "epoch": 0.03696715781138265,
+      "grad_norm": 636.5348510742188,
+      "learning_rate": 7.393939393939395e-06,
+      "loss": 55.7327,
+      "step": 9150
+    },
+    {
+      "epoch": 0.03700755907675028,
+      "grad_norm": 746.4474487304688,
+      "learning_rate": 7.402020202020202e-06,
+      "loss": 88.9879,
+      "step": 9160
+    },
+    {
+      "epoch": 0.037047960342117915,
+      "grad_norm": 1030.46044921875,
+      "learning_rate": 7.410101010101011e-06,
+      "loss": 67.4612,
+      "step": 9170
+    },
+    {
+      "epoch": 0.03708836160748555,
+      "grad_norm": 632.0867919921875,
+      "learning_rate": 7.4181818181818185e-06,
+      "loss": 48.0889,
+      "step": 9180
+    },
+    {
+      "epoch": 0.037128762872853174,
+      "grad_norm": 709.0557250976562,
+      "learning_rate": 7.426262626262627e-06,
+      "loss": 56.7526,
+      "step": 9190
+    },
+    {
+      "epoch": 0.03716916413822081,
+      "grad_norm": 910.6807250976562,
+      "learning_rate": 7.434343434343435e-06,
+      "loss": 61.0249,
+      "step": 9200
+    },
+    {
+      "epoch": 0.03720956540358844,
+      "grad_norm": 1631.1051025390625,
+      "learning_rate": 7.442424242424243e-06,
+      "loss": 53.2277,
+      "step": 9210
+    },
+    {
+      "epoch": 0.037249966668956074,
+      "grad_norm": 637.2019653320312,
+      "learning_rate": 7.450505050505051e-06,
+      "loss": 48.1968,
+      "step": 9220
+    },
+    {
+      "epoch": 0.0372903679343237,
+      "grad_norm": 881.9324951171875,
+      "learning_rate": 7.45858585858586e-06,
+      "loss": 56.8194,
+      "step": 9230
+    },
+    {
+      "epoch": 0.03733076919969133,
+      "grad_norm": 986.95263671875,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 53.6152,
+      "step": 9240
+    },
+    {
+      "epoch": 0.037371170465058966,
+      "grad_norm": 767.5629272460938,
+      "learning_rate": 7.474747474747476e-06,
+      "loss": 49.1764,
+      "step": 9250
+    },
+    {
+      "epoch": 0.0374115717304266,
+      "grad_norm": 2124.801513671875,
+      "learning_rate": 7.4828282828282835e-06,
+      "loss": 42.4815,
+      "step": 9260
+    },
+    {
+      "epoch": 0.037451972995794225,
+      "grad_norm": 409.00555419921875,
+      "learning_rate": 7.490909090909092e-06,
+      "loss": 60.798,
+      "step": 9270
+    },
+    {
+      "epoch": 0.03749237426116186,
+      "grad_norm": 271.09124755859375,
+      "learning_rate": 7.4989898989899e-06,
+      "loss": 50.9011,
+      "step": 9280
+    },
+    {
+      "epoch": 0.03753277552652949,
+      "grad_norm": 974.7747192382812,
+      "learning_rate": 7.507070707070707e-06,
+      "loss": 62.7403,
+      "step": 9290
+    },
+    {
+      "epoch": 0.037573176791897124,
+      "grad_norm": 697.45166015625,
+      "learning_rate": 7.515151515151516e-06,
+      "loss": 71.2546,
+      "step": 9300
+    },
+    {
+      "epoch": 0.03761357805726475,
+      "grad_norm": 536.0864868164062,
+      "learning_rate": 7.523232323232323e-06,
+      "loss": 46.562,
+      "step": 9310
+    },
+    {
+      "epoch": 0.03765397932263238,
+      "grad_norm": 801.9717407226562,
+      "learning_rate": 7.531313131313132e-06,
+      "loss": 68.1762,
+      "step": 9320
+    },
+    {
+      "epoch": 0.037694380588000016,
+      "grad_norm": 780.1590576171875,
+      "learning_rate": 7.53939393939394e-06,
+      "loss": 72.8615,
+      "step": 9330
+    },
+    {
+      "epoch": 0.03773478185336765,
+      "grad_norm": 597.0780029296875,
+      "learning_rate": 7.547474747474748e-06,
+      "loss": 51.8024,
+      "step": 9340
+    },
+    {
+      "epoch": 0.037775183118735275,
+      "grad_norm": 392.8757019042969,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 25.0151,
+      "step": 9350
+    },
+    {
+      "epoch": 0.03781558438410291,
+      "grad_norm": 908.3223266601562,
+      "learning_rate": 7.563636363636364e-06,
+      "loss": 85.6096,
+      "step": 9360
+    },
+    {
+      "epoch": 0.03785598564947054,
+      "grad_norm": 596.0296020507812,
+      "learning_rate": 7.571717171717172e-06,
+      "loss": 35.1797,
+      "step": 9370
+    },
+    {
+      "epoch": 0.037896386914838175,
+      "grad_norm": 587.8231201171875,
+      "learning_rate": 7.579797979797981e-06,
+      "loss": 60.454,
+      "step": 9380
+    },
+    {
+      "epoch": 0.0379367881802058,
+      "grad_norm": 616.9268798828125,
+      "learning_rate": 7.587878787878788e-06,
+      "loss": 50.2102,
+      "step": 9390
+    },
+    {
+      "epoch": 0.037977189445573434,
+      "grad_norm": 486.89788818359375,
+      "learning_rate": 7.595959595959597e-06,
+      "loss": 49.0462,
+      "step": 9400
+    },
+    {
+      "epoch": 0.03801759071094107,
+      "grad_norm": 655.9788208007812,
+      "learning_rate": 7.604040404040405e-06,
+      "loss": 51.1414,
+      "step": 9410
+    },
+    {
+      "epoch": 0.0380579919763087,
+      "grad_norm": 652.9426879882812,
+      "learning_rate": 7.612121212121213e-06,
+      "loss": 44.0585,
+      "step": 9420
+    },
+    {
+      "epoch": 0.038098393241676326,
+      "grad_norm": 315.17327880859375,
+      "learning_rate": 7.620202020202021e-06,
+      "loss": 46.1256,
+      "step": 9430
+    },
+    {
+      "epoch": 0.03813879450704396,
+      "grad_norm": 1290.96533203125,
+      "learning_rate": 7.628282828282829e-06,
+      "loss": 51.2887,
+      "step": 9440
+    },
+    {
+      "epoch": 0.03817919577241159,
+      "grad_norm": 945.685302734375,
+      "learning_rate": 7.636363636363638e-06,
+      "loss": 51.7154,
+      "step": 9450
+    },
+    {
+      "epoch": 0.038219597037779225,
+      "grad_norm": 572.252685546875,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 55.3707,
+      "step": 9460
+    },
+    {
+      "epoch": 0.03825999830314685,
+      "grad_norm": 1046.2398681640625,
+      "learning_rate": 7.652525252525253e-06,
+      "loss": 54.6761,
+      "step": 9470
+    },
+    {
+      "epoch": 0.038300399568514484,
+      "grad_norm": 964.0022583007812,
+      "learning_rate": 7.660606060606062e-06,
+      "loss": 65.9791,
+      "step": 9480
+    },
+    {
+      "epoch": 0.03834080083388212,
+      "grad_norm": 619.1194458007812,
+      "learning_rate": 7.66868686868687e-06,
+      "loss": 43.4894,
+      "step": 9490
+    },
+    {
+      "epoch": 0.03838120209924975,
+      "grad_norm": 337.8772277832031,
+      "learning_rate": 7.676767676767677e-06,
+      "loss": 58.457,
+      "step": 9500
+    },
+    {
+      "epoch": 0.03842160336461738,
+      "grad_norm": 542.2553100585938,
+      "learning_rate": 7.684848484848485e-06,
+      "loss": 43.3362,
+      "step": 9510
+    },
+    {
+      "epoch": 0.03846200462998501,
+      "grad_norm": 740.5171508789062,
+      "learning_rate": 7.692929292929294e-06,
+      "loss": 74.0348,
+      "step": 9520
+    },
+    {
+      "epoch": 0.03850240589535264,
+      "grad_norm": 974.9364624023438,
+      "learning_rate": 7.7010101010101e-06,
+      "loss": 44.931,
+      "step": 9530
+    },
+    {
+      "epoch": 0.038542807160720276,
+      "grad_norm": 519.744140625,
+      "learning_rate": 7.709090909090909e-06,
+      "loss": 33.287,
+      "step": 9540
+    },
+    {
+      "epoch": 0.0385832084260879,
+      "grad_norm": 451.38677978515625,
+      "learning_rate": 7.717171717171717e-06,
+      "loss": 40.3233,
+      "step": 9550
+    },
+    {
+      "epoch": 0.038623609691455535,
+      "grad_norm": 675.4090576171875,
+      "learning_rate": 7.725252525252526e-06,
+      "loss": 41.2818,
+      "step": 9560
+    },
+    {
+      "epoch": 0.03866401095682317,
+      "grad_norm": 4203.88623046875,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 79.4917,
+      "step": 9570
+    },
+    {
+      "epoch": 0.0387044122221908,
+      "grad_norm": 688.3527221679688,
+      "learning_rate": 7.741414141414141e-06,
+      "loss": 46.4714,
+      "step": 9580
+    },
+    {
+      "epoch": 0.03874481348755843,
+      "grad_norm": 985.7579345703125,
+      "learning_rate": 7.74949494949495e-06,
+      "loss": 49.4049,
+      "step": 9590
+    },
+    {
+      "epoch": 0.03878521475292606,
+      "grad_norm": 498.4870300292969,
+      "learning_rate": 7.757575757575758e-06,
+      "loss": 54.2156,
+      "step": 9600
+    },
+    {
+      "epoch": 0.03882561601829369,
+      "grad_norm": 474.2466125488281,
+      "learning_rate": 7.765656565656566e-06,
+      "loss": 43.1809,
+      "step": 9610
+    },
+    {
+      "epoch": 0.038866017283661326,
+      "grad_norm": 777.8629150390625,
+      "learning_rate": 7.773737373737375e-06,
+      "loss": 50.0144,
+      "step": 9620
+    },
+    {
+      "epoch": 0.03890641854902895,
+      "grad_norm": 911.9942016601562,
+      "learning_rate": 7.781818181818183e-06,
+      "loss": 71.0392,
+      "step": 9630
+    },
+    {
+      "epoch": 0.038946819814396585,
+      "grad_norm": 745.5965576171875,
+      "learning_rate": 7.78989898989899e-06,
+      "loss": 52.3377,
+      "step": 9640
+    },
+    {
+      "epoch": 0.03898722107976422,
+      "grad_norm": 655.3526611328125,
+      "learning_rate": 7.797979797979799e-06,
+      "loss": 56.1896,
+      "step": 9650
+    },
+    {
+      "epoch": 0.03902762234513185,
+      "grad_norm": 986.2277221679688,
+      "learning_rate": 7.806060606060607e-06,
+      "loss": 66.5987,
+      "step": 9660
+    },
+    {
+      "epoch": 0.03906802361049948,
+      "grad_norm": 2253.599365234375,
+      "learning_rate": 7.814141414141415e-06,
+      "loss": 54.8736,
+      "step": 9670
+    },
+    {
+      "epoch": 0.03910842487586711,
+      "grad_norm": 809.8319091796875,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 60.677,
+      "step": 9680
+    },
+    {
+      "epoch": 0.039148826141234744,
+      "grad_norm": 745.8842163085938,
+      "learning_rate": 7.83030303030303e-06,
+      "loss": 54.0499,
+      "step": 9690
+    },
+    {
+      "epoch": 0.03918922740660238,
+      "grad_norm": 392.3405456542969,
+      "learning_rate": 7.838383838383839e-06,
+      "loss": 57.1388,
+      "step": 9700
+    },
+    {
+      "epoch": 0.03922962867197,
+      "grad_norm": 391.06524658203125,
+      "learning_rate": 7.846464646464646e-06,
+      "loss": 67.0702,
+      "step": 9710
+    },
+    {
+      "epoch": 0.039270029937337636,
+      "grad_norm": 964.3352661132812,
+      "learning_rate": 7.854545454545454e-06,
+      "loss": 49.3004,
+      "step": 9720
+    },
+    {
+      "epoch": 0.03931043120270527,
+      "grad_norm": 1257.4691162109375,
+      "learning_rate": 7.862626262626263e-06,
+      "loss": 34.1305,
+      "step": 9730
+    },
+    {
+      "epoch": 0.0393508324680729,
+      "grad_norm": 408.74609375,
+      "learning_rate": 7.870707070707071e-06,
+      "loss": 52.1865,
+      "step": 9740
+    },
+    {
+      "epoch": 0.03939123373344053,
+      "grad_norm": 1199.1552734375,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 54.2215,
+      "step": 9750
+    },
+    {
+      "epoch": 0.03943163499880816,
+      "grad_norm": 1096.5802001953125,
+      "learning_rate": 7.886868686868686e-06,
+      "loss": 61.8289,
+      "step": 9760
+    },
+    {
+      "epoch": 0.039472036264175794,
+      "grad_norm": 1617.402587890625,
+      "learning_rate": 7.894949494949495e-06,
+      "loss": 55.392,
+      "step": 9770
+    },
+    {
+      "epoch": 0.03951243752954343,
+      "grad_norm": 603.028076171875,
+      "learning_rate": 7.903030303030303e-06,
+      "loss": 38.076,
+      "step": 9780
+    },
+    {
+      "epoch": 0.039552838794911054,
+      "grad_norm": 712.2247314453125,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 59.4955,
+      "step": 9790
+    },
+    {
+      "epoch": 0.03959324006027869,
+      "grad_norm": 474.9757995605469,
+      "learning_rate": 7.91919191919192e-06,
+      "loss": 41.3934,
+      "step": 9800
+    },
+    {
+      "epoch": 0.03963364132564632,
+      "grad_norm": 895.0010375976562,
+      "learning_rate": 7.927272727272729e-06,
+      "loss": 71.613,
+      "step": 9810
+    },
+    {
+      "epoch": 0.03967404259101395,
+      "grad_norm": 343.2088928222656,
+      "learning_rate": 7.935353535353535e-06,
+      "loss": 56.6161,
+      "step": 9820
+    },
+    {
+      "epoch": 0.03971444385638158,
+      "grad_norm": 317.080078125,
+      "learning_rate": 7.943434343434344e-06,
+      "loss": 58.7523,
+      "step": 9830
+    },
+    {
+      "epoch": 0.03975484512174921,
+      "grad_norm": 1206.803955078125,
+      "learning_rate": 7.951515151515152e-06,
+      "loss": 60.3558,
+      "step": 9840
+    },
+    {
+      "epoch": 0.039795246387116845,
+      "grad_norm": 568.8405151367188,
+      "learning_rate": 7.95959595959596e-06,
+      "loss": 50.1119,
+      "step": 9850
+    },
+    {
+      "epoch": 0.03983564765248448,
+      "grad_norm": 1495.424560546875,
+      "learning_rate": 7.96767676767677e-06,
+      "loss": 63.5327,
+      "step": 9860
+    },
+    {
+      "epoch": 0.039876048917852104,
+      "grad_norm": 889.3040771484375,
+      "learning_rate": 7.975757575757576e-06,
+      "loss": 59.4066,
+      "step": 9870
+    },
+    {
+      "epoch": 0.03991645018321974,
+      "grad_norm": 515.6483154296875,
+      "learning_rate": 7.983838383838384e-06,
+      "loss": 44.5552,
+      "step": 9880
+    },
+    {
+      "epoch": 0.03995685144858737,
+      "grad_norm": 422.6609802246094,
+      "learning_rate": 7.991919191919193e-06,
+      "loss": 34.7969,
+      "step": 9890
+    },
+    {
+      "epoch": 0.039997252713955,
+      "grad_norm": 504.5275573730469,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 42.1655,
+      "step": 9900
+    },
+    {
+      "epoch": 0.04003765397932263,
+      "grad_norm": 388.7497253417969,
+      "learning_rate": 8.00808080808081e-06,
+      "loss": 64.0802,
+      "step": 9910
+    },
+    {
+      "epoch": 0.04007805524469026,
+      "grad_norm": 692.829345703125,
+      "learning_rate": 8.016161616161617e-06,
+      "loss": 56.8257,
+      "step": 9920
+    },
+    {
+      "epoch": 0.040118456510057895,
+      "grad_norm": 646.9595336914062,
+      "learning_rate": 8.024242424242425e-06,
+      "loss": 68.7429,
+      "step": 9930
+    },
+    {
+      "epoch": 0.04015885777542553,
+      "grad_norm": 870.9974365234375,
+      "learning_rate": 8.032323232323232e-06,
+      "loss": 28.9334,
+      "step": 9940
+    },
+    {
+      "epoch": 0.040199259040793155,
+      "grad_norm": 696.292236328125,
+      "learning_rate": 8.04040404040404e-06,
+      "loss": 38.459,
+      "step": 9950
+    },
+    {
+      "epoch": 0.04023966030616079,
+      "grad_norm": 849.521728515625,
+      "learning_rate": 8.048484848484849e-06,
+      "loss": 53.8687,
+      "step": 9960
+    },
+    {
+      "epoch": 0.04028006157152842,
+      "grad_norm": 874.6473388671875,
+      "learning_rate": 8.056565656565657e-06,
+      "loss": 77.5796,
+      "step": 9970
+    },
+    {
+      "epoch": 0.040320462836896054,
+      "grad_norm": 419.1984558105469,
+      "learning_rate": 8.064646464646466e-06,
+      "loss": 62.392,
+      "step": 9980
+    },
+    {
+      "epoch": 0.04036086410226368,
+      "grad_norm": 518.1126098632812,
+      "learning_rate": 8.072727272727274e-06,
+      "loss": 50.2856,
+      "step": 9990
+    },
+    {
+      "epoch": 0.04040126536763131,
+      "grad_norm": 429.58343505859375,
+      "learning_rate": 8.08080808080808e-06,
+      "loss": 44.5308,
+      "step": 10000
+    },
+    {
+      "epoch": 0.040441666632998946,
+      "grad_norm": 321.3651123046875,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 43.4247,
+      "step": 10010
+    },
+    {
+      "epoch": 0.04048206789836658,
+      "grad_norm": 651.4296264648438,
+      "learning_rate": 8.096969696969698e-06,
+      "loss": 46.7586,
+      "step": 10020
+    },
+    {
+      "epoch": 0.040522469163734205,
+      "grad_norm": 424.0990905761719,
+      "learning_rate": 8.105050505050506e-06,
+      "loss": 66.6896,
+      "step": 10030
+    },
+    {
+      "epoch": 0.04056287042910184,
+      "grad_norm": 675.9149780273438,
+      "learning_rate": 8.113131313131315e-06,
+      "loss": 43.7758,
+      "step": 10040
+    },
+    {
+      "epoch": 0.04060327169446947,
+      "grad_norm": 139.72412109375,
+      "learning_rate": 8.121212121212121e-06,
+      "loss": 55.9011,
+      "step": 10050
+    },
+    {
+      "epoch": 0.040643672959837104,
+      "grad_norm": 748.906005859375,
+      "learning_rate": 8.12929292929293e-06,
+      "loss": 43.4391,
+      "step": 10060
+    },
+    {
+      "epoch": 0.04068407422520473,
+      "grad_norm": 669.5646362304688,
+      "learning_rate": 8.137373737373738e-06,
+      "loss": 58.3727,
+      "step": 10070
+    },
+    {
+      "epoch": 0.040724475490572364,
+      "grad_norm": 742.7377319335938,
+      "learning_rate": 8.145454545454547e-06,
+      "loss": 65.0417,
+      "step": 10080
+    },
+    {
+      "epoch": 0.04076487675594,
+      "grad_norm": 404.1474304199219,
+      "learning_rate": 8.153535353535355e-06,
+      "loss": 68.0475,
+      "step": 10090
+    },
+    {
+      "epoch": 0.04080527802130763,
+      "grad_norm": 676.8277587890625,
+      "learning_rate": 8.161616161616162e-06,
+      "loss": 50.2154,
+      "step": 10100
+    },
+    {
+      "epoch": 0.040845679286675256,
+      "grad_norm": 904.6346435546875,
+      "learning_rate": 8.16969696969697e-06,
+      "loss": 61.044,
+      "step": 10110
+    },
+    {
+      "epoch": 0.04088608055204289,
+      "grad_norm": 542.9082641601562,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 40.1249,
+      "step": 10120
+    },
+    {
+      "epoch": 0.04092648181741052,
+      "grad_norm": 664.3563842773438,
+      "learning_rate": 8.185858585858587e-06,
+      "loss": 80.5697,
+      "step": 10130
+    },
+    {
+      "epoch": 0.040966883082778155,
+      "grad_norm": 893.8529663085938,
+      "learning_rate": 8.193939393939394e-06,
+      "loss": 48.2468,
+      "step": 10140
+    },
+    {
+      "epoch": 0.04100728434814578,
+      "grad_norm": 582.698486328125,
+      "learning_rate": 8.202020202020202e-06,
+      "loss": 53.0039,
+      "step": 10150
+    },
+    {
+      "epoch": 0.041047685613513414,
+      "grad_norm": 1301.6468505859375,
+      "learning_rate": 8.21010101010101e-06,
+      "loss": 62.4941,
+      "step": 10160
+    },
+    {
+      "epoch": 0.04108808687888105,
+      "grad_norm": 643.629150390625,
+      "learning_rate": 8.21818181818182e-06,
+      "loss": 56.7416,
+      "step": 10170
+    },
+    {
+      "epoch": 0.04112848814424868,
+      "grad_norm": 599.5493774414062,
+      "learning_rate": 8.226262626262626e-06,
+      "loss": 71.8773,
+      "step": 10180
+    },
+    {
+      "epoch": 0.041168889409616306,
+      "grad_norm": 1008.1700439453125,
+      "learning_rate": 8.234343434343434e-06,
+      "loss": 59.6597,
+      "step": 10190
+    },
+    {
+      "epoch": 0.04120929067498394,
+      "grad_norm": 476.3126525878906,
+      "learning_rate": 8.242424242424243e-06,
+      "loss": 37.4634,
+      "step": 10200
+    },
+    {
+      "epoch": 0.04124969194035157,
+      "grad_norm": 644.0550537109375,
+      "learning_rate": 8.250505050505051e-06,
+      "loss": 55.9344,
+      "step": 10210
+    },
+    {
+      "epoch": 0.041290093205719205,
+      "grad_norm": 624.4482421875,
+      "learning_rate": 8.25858585858586e-06,
+      "loss": 65.2902,
+      "step": 10220
+    },
+    {
+      "epoch": 0.04133049447108683,
+      "grad_norm": 636.2765502929688,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 37.1124,
+      "step": 10230
+    },
+    {
+      "epoch": 0.041370895736454465,
+      "grad_norm": 1039.8856201171875,
+      "learning_rate": 8.274747474747475e-06,
+      "loss": 55.8901,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0414112970018221,
+      "grad_norm": 1239.593505859375,
+      "learning_rate": 8.282828282828283e-06,
+      "loss": 79.6421,
+      "step": 10250
+    },
+    {
+      "epoch": 0.04145169826718973,
+      "grad_norm": 909.7822875976562,
+      "learning_rate": 8.290909090909092e-06,
+      "loss": 63.3099,
+      "step": 10260
+    },
+    {
+      "epoch": 0.04149209953255736,
+      "grad_norm": 340.58721923828125,
+      "learning_rate": 8.2989898989899e-06,
+      "loss": 55.8349,
+      "step": 10270
+    },
+    {
+      "epoch": 0.04153250079792499,
+      "grad_norm": 674.6702270507812,
+      "learning_rate": 8.307070707070707e-06,
+      "loss": 87.4816,
+      "step": 10280
+    },
+    {
+      "epoch": 0.04157290206329262,
+      "grad_norm": 802.2882690429688,
+      "learning_rate": 8.315151515151516e-06,
+      "loss": 40.7146,
+      "step": 10290
+    },
+    {
+      "epoch": 0.041613303328660256,
+      "grad_norm": 707.7451171875,
+      "learning_rate": 8.323232323232324e-06,
+      "loss": 63.5479,
+      "step": 10300
+    },
+    {
+      "epoch": 0.04165370459402788,
+      "grad_norm": 777.0239868164062,
+      "learning_rate": 8.331313131313132e-06,
+      "loss": 43.5335,
+      "step": 10310
+    },
+    {
+      "epoch": 0.041694105859395515,
+      "grad_norm": 514.3717041015625,
+      "learning_rate": 8.339393939393941e-06,
+      "loss": 83.1662,
+      "step": 10320
+    },
+    {
+      "epoch": 0.04173450712476315,
+      "grad_norm": 292.78375244140625,
+      "learning_rate": 8.34747474747475e-06,
+      "loss": 37.0555,
+      "step": 10330
+    },
+    {
+      "epoch": 0.04177490839013078,
+      "grad_norm": 664.9932861328125,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 38.5216,
+      "step": 10340
+    },
+    {
+      "epoch": 0.04181530965549841,
+      "grad_norm": 735.2276611328125,
+      "learning_rate": 8.363636363636365e-06,
+      "loss": 54.7254,
+      "step": 10350
+    },
+    {
+      "epoch": 0.04185571092086604,
+      "grad_norm": 1044.8675537109375,
+      "learning_rate": 8.371717171717171e-06,
+      "loss": 52.2735,
+      "step": 10360
+    },
+    {
+      "epoch": 0.041896112186233674,
+      "grad_norm": 1418.629150390625,
+      "learning_rate": 8.37979797979798e-06,
+      "loss": 85.6306,
+      "step": 10370
+    },
+    {
+      "epoch": 0.04193651345160131,
+      "grad_norm": 560.5042114257812,
+      "learning_rate": 8.387878787878788e-06,
+      "loss": 71.9104,
+      "step": 10380
+    },
+    {
+      "epoch": 0.04197691471696893,
+      "grad_norm": 305.0915222167969,
+      "learning_rate": 8.395959595959597e-06,
+      "loss": 54.2431,
+      "step": 10390
+    },
+    {
+      "epoch": 0.042017315982336566,
+      "grad_norm": 502.9292297363281,
+      "learning_rate": 8.404040404040405e-06,
+      "loss": 55.7189,
+      "step": 10400
+    },
+    {
+      "epoch": 0.0420577172477042,
+      "grad_norm": 408.15557861328125,
+      "learning_rate": 8.412121212121212e-06,
+      "loss": 55.843,
+      "step": 10410
+    },
+    {
+      "epoch": 0.04209811851307183,
+      "grad_norm": 480.6004638671875,
+      "learning_rate": 8.42020202020202e-06,
+      "loss": 48.1143,
+      "step": 10420
+    },
+    {
+      "epoch": 0.04213851977843946,
+      "grad_norm": 990.3383178710938,
+      "learning_rate": 8.428282828282829e-06,
+      "loss": 55.1379,
+      "step": 10430
+    },
+    {
+      "epoch": 0.04217892104380709,
+      "grad_norm": 629.5912475585938,
+      "learning_rate": 8.436363636363637e-06,
+      "loss": 61.4927,
+      "step": 10440
+    },
+    {
+      "epoch": 0.042219322309174724,
+      "grad_norm": 1151.8690185546875,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 73.6629,
+      "step": 10450
+    },
+    {
+      "epoch": 0.04225972357454236,
+      "grad_norm": 1047.275634765625,
+      "learning_rate": 8.452525252525252e-06,
+      "loss": 50.8134,
+      "step": 10460
+    },
+    {
+      "epoch": 0.04230012483990998,
+      "grad_norm": 480.5417175292969,
+      "learning_rate": 8.460606060606061e-06,
+      "loss": 42.938,
+      "step": 10470
+    },
+    {
+      "epoch": 0.042340526105277616,
+      "grad_norm": 547.515869140625,
+      "learning_rate": 8.46868686868687e-06,
+      "loss": 39.9703,
+      "step": 10480
+    },
+    {
+      "epoch": 0.04238092737064525,
+      "grad_norm": 678.3375244140625,
+      "learning_rate": 8.476767676767678e-06,
+      "loss": 62.9904,
+      "step": 10490
+    },
+    {
+      "epoch": 0.04242132863601288,
+      "grad_norm": 849.853759765625,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 53.3405,
+      "step": 10500
+    },
+    {
+      "epoch": 0.04246172990138051,
+      "grad_norm": 493.50836181640625,
+      "learning_rate": 8.492929292929295e-06,
+      "loss": 52.3254,
+      "step": 10510
+    },
+    {
+      "epoch": 0.04250213116674814,
+      "grad_norm": 891.0341796875,
+      "learning_rate": 8.501010101010101e-06,
+      "loss": 53.4351,
+      "step": 10520
+    },
+    {
+      "epoch": 0.042542532432115775,
+      "grad_norm": 343.7546691894531,
+      "learning_rate": 8.50909090909091e-06,
+      "loss": 51.0014,
+      "step": 10530
+    },
+    {
+      "epoch": 0.04258293369748341,
+      "grad_norm": 500.97979736328125,
+      "learning_rate": 8.517171717171718e-06,
+      "loss": 50.3782,
+      "step": 10540
+    },
+    {
+      "epoch": 0.042623334962851034,
+      "grad_norm": 294.78076171875,
+      "learning_rate": 8.525252525252527e-06,
+      "loss": 33.7601,
+      "step": 10550
+    },
+    {
+      "epoch": 0.04266373622821867,
+      "grad_norm": 993.7897338867188,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 44.46,
+      "step": 10560
+    },
+    {
+      "epoch": 0.0427041374935863,
+      "grad_norm": 837.2728881835938,
+      "learning_rate": 8.541414141414142e-06,
+      "loss": 52.4295,
+      "step": 10570
+    },
+    {
+      "epoch": 0.04274453875895393,
+      "grad_norm": 1006.4978637695312,
+      "learning_rate": 8.54949494949495e-06,
+      "loss": 46.5704,
+      "step": 10580
+    },
+    {
+      "epoch": 0.04278494002432156,
+      "grad_norm": 1683.9364013671875,
+      "learning_rate": 8.557575757575757e-06,
+      "loss": 61.555,
+      "step": 10590
+    },
+    {
+      "epoch": 0.04282534128968919,
+      "grad_norm": 664.7928466796875,
+      "learning_rate": 8.565656565656566e-06,
+      "loss": 45.5131,
+      "step": 10600
+    },
+    {
+      "epoch": 0.042865742555056825,
+      "grad_norm": 706.4365234375,
+      "learning_rate": 8.573737373737374e-06,
+      "loss": 71.1823,
+      "step": 10610
+    },
+    {
+      "epoch": 0.04290614382042446,
+      "grad_norm": 1057.157470703125,
+      "learning_rate": 8.581818181818183e-06,
+      "loss": 66.9543,
+      "step": 10620
+    },
+    {
+      "epoch": 0.042946545085792084,
+      "grad_norm": 763.0455322265625,
+      "learning_rate": 8.589898989898991e-06,
+      "loss": 49.7286,
+      "step": 10630
+    },
+    {
+      "epoch": 0.04298694635115972,
+      "grad_norm": 917.1838989257812,
+      "learning_rate": 8.597979797979798e-06,
+      "loss": 61.1907,
+      "step": 10640
+    },
+    {
+      "epoch": 0.04302734761652735,
+      "grad_norm": 821.6635131835938,
+      "learning_rate": 8.606060606060606e-06,
+      "loss": 40.5262,
+      "step": 10650
+    },
+    {
+      "epoch": 0.043067748881894984,
+      "grad_norm": 557.9569091796875,
+      "learning_rate": 8.614141414141415e-06,
+      "loss": 62.2163,
+      "step": 10660
+    },
+    {
+      "epoch": 0.04310815014726261,
+      "grad_norm": 374.8048095703125,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 52.1193,
+      "step": 10670
+    },
+    {
+      "epoch": 0.04314855141263024,
+      "grad_norm": 622.052978515625,
+      "learning_rate": 8.630303030303032e-06,
+      "loss": 39.8787,
+      "step": 10680
+    },
+    {
+      "epoch": 0.043188952677997876,
+      "grad_norm": 496.65679931640625,
+      "learning_rate": 8.63838383838384e-06,
+      "loss": 41.2797,
+      "step": 10690
+    },
+    {
+      "epoch": 0.04322935394336551,
+      "grad_norm": 773.8624877929688,
+      "learning_rate": 8.646464646464647e-06,
+      "loss": 47.26,
+      "step": 10700
+    },
+    {
+      "epoch": 0.043269755208733135,
+      "grad_norm": 397.5426025390625,
+      "learning_rate": 8.654545454545455e-06,
+      "loss": 55.0543,
+      "step": 10710
+    },
+    {
+      "epoch": 0.04331015647410077,
+      "grad_norm": 890.8135375976562,
+      "learning_rate": 8.662626262626264e-06,
+      "loss": 59.0316,
+      "step": 10720
+    },
+    {
+      "epoch": 0.0433505577394684,
+      "grad_norm": 1218.11376953125,
+      "learning_rate": 8.670707070707072e-06,
+      "loss": 59.9817,
+      "step": 10730
+    },
+    {
+      "epoch": 0.043390959004836034,
+      "grad_norm": 981.9630737304688,
+      "learning_rate": 8.67878787878788e-06,
+      "loss": 50.446,
+      "step": 10740
+    },
+    {
+      "epoch": 0.04343136027020366,
+      "grad_norm": 421.39764404296875,
+      "learning_rate": 8.686868686868687e-06,
+      "loss": 59.1779,
+      "step": 10750
+    },
+    {
+      "epoch": 0.04347176153557129,
+      "grad_norm": 994.2972412109375,
+      "learning_rate": 8.694949494949496e-06,
+      "loss": 71.2787,
+      "step": 10760
+    },
+    {
+      "epoch": 0.043512162800938926,
+      "grad_norm": 926.8892211914062,
+      "learning_rate": 8.703030303030304e-06,
+      "loss": 67.1323,
+      "step": 10770
+    },
+    {
+      "epoch": 0.04355256406630656,
+      "grad_norm": 652.7195434570312,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 45.3532,
+      "step": 10780
+    },
+    {
+      "epoch": 0.043592965331674186,
+      "grad_norm": 632.1166381835938,
+      "learning_rate": 8.71919191919192e-06,
+      "loss": 57.5382,
+      "step": 10790
+    },
+    {
+      "epoch": 0.04363336659704182,
+      "grad_norm": 785.2573852539062,
+      "learning_rate": 8.727272727272728e-06,
+      "loss": 64.4019,
+      "step": 10800
+    },
+    {
+      "epoch": 0.04367376786240945,
+      "grad_norm": 859.1141357421875,
+      "learning_rate": 8.735353535353536e-06,
+      "loss": 77.0335,
+      "step": 10810
+    },
+    {
+      "epoch": 0.043714169127777085,
+      "grad_norm": 894.4269409179688,
+      "learning_rate": 8.743434343434343e-06,
+      "loss": 49.1961,
+      "step": 10820
+    },
+    {
+      "epoch": 0.04375457039314471,
+      "grad_norm": 564.3938598632812,
+      "learning_rate": 8.751515151515151e-06,
+      "loss": 51.9728,
+      "step": 10830
+    },
+    {
+      "epoch": 0.043794971658512344,
+      "grad_norm": 1009.34521484375,
+      "learning_rate": 8.75959595959596e-06,
+      "loss": 65.0657,
+      "step": 10840
+    },
+    {
+      "epoch": 0.04383537292387998,
+      "grad_norm": 631.89111328125,
+      "learning_rate": 8.767676767676768e-06,
+      "loss": 61.9598,
+      "step": 10850
+    },
+    {
+      "epoch": 0.04387577418924761,
+      "grad_norm": 716.18994140625,
+      "learning_rate": 8.775757575757577e-06,
+      "loss": 49.4842,
+      "step": 10860
+    },
+    {
+      "epoch": 0.043916175454615236,
+      "grad_norm": 1062.8853759765625,
+      "learning_rate": 8.783838383838385e-06,
+      "loss": 71.7064,
+      "step": 10870
+    },
+    {
+      "epoch": 0.04395657671998287,
+      "grad_norm": 729.2897338867188,
+      "learning_rate": 8.791919191919192e-06,
+      "loss": 63.324,
+      "step": 10880
+    },
+    {
+      "epoch": 0.0439969779853505,
+      "grad_norm": 528.548828125,
+      "learning_rate": 8.8e-06,
+      "loss": 66.7493,
+      "step": 10890
+    },
+    {
+      "epoch": 0.044037379250718135,
+      "grad_norm": 838.0880737304688,
+      "learning_rate": 8.808080808080809e-06,
+      "loss": 54.1123,
+      "step": 10900
+    },
+    {
+      "epoch": 0.04407778051608576,
+      "grad_norm": 623.3431396484375,
+      "learning_rate": 8.816161616161617e-06,
+      "loss": 43.8357,
+      "step": 10910
+    },
+    {
+      "epoch": 0.044118181781453394,
+      "grad_norm": 415.62939453125,
+      "learning_rate": 8.824242424242426e-06,
+      "loss": 53.8524,
+      "step": 10920
+    },
+    {
+      "epoch": 0.04415858304682103,
+      "grad_norm": 618.20654296875,
+      "learning_rate": 8.832323232323233e-06,
+      "loss": 47.9192,
+      "step": 10930
+    },
+    {
+      "epoch": 0.04419898431218866,
+      "grad_norm": 708.5919799804688,
+      "learning_rate": 8.840404040404041e-06,
+      "loss": 50.4365,
+      "step": 10940
+    },
+    {
+      "epoch": 0.04423938557755629,
+      "grad_norm": 791.4614868164062,
+      "learning_rate": 8.84848484848485e-06,
+      "loss": 35.4549,
+      "step": 10950
+    },
+    {
+      "epoch": 0.04427978684292392,
+      "grad_norm": 508.3524169921875,
+      "learning_rate": 8.856565656565658e-06,
+      "loss": 36.9709,
+      "step": 10960
+    },
+    {
+      "epoch": 0.04432018810829155,
+      "grad_norm": 152.70664978027344,
+      "learning_rate": 8.864646464646466e-06,
+      "loss": 41.8174,
+      "step": 10970
+    },
+    {
+      "epoch": 0.044360589373659186,
+      "grad_norm": 322.36065673828125,
+      "learning_rate": 8.872727272727275e-06,
+      "loss": 51.6953,
+      "step": 10980
+    },
+    {
+      "epoch": 0.04440099063902681,
+      "grad_norm": 553.2557983398438,
+      "learning_rate": 8.880808080808082e-06,
+      "loss": 75.7669,
+      "step": 10990
+    },
+    {
+      "epoch": 0.044441391904394445,
+      "grad_norm": 490.9820251464844,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 50.3666,
+      "step": 11000
+    },
+    {
+      "epoch": 0.04448179316976208,
+      "grad_norm": 680.9080810546875,
+      "learning_rate": 8.896969696969697e-06,
+      "loss": 50.7332,
+      "step": 11010
+    },
+    {
+      "epoch": 0.04452219443512971,
+      "grad_norm": 994.4353637695312,
+      "learning_rate": 8.905050505050505e-06,
+      "loss": 54.6636,
+      "step": 11020
+    },
+    {
+      "epoch": 0.04456259570049734,
+      "grad_norm": 879.1863403320312,
+      "learning_rate": 8.913131313131314e-06,
+      "loss": 55.7514,
+      "step": 11030
+    },
+    {
+      "epoch": 0.04460299696586497,
+      "grad_norm": 453.1380310058594,
+      "learning_rate": 8.921212121212122e-06,
+      "loss": 76.496,
+      "step": 11040
+    },
+    {
+      "epoch": 0.0446433982312326,
+      "grad_norm": 364.94403076171875,
+      "learning_rate": 8.92929292929293e-06,
+      "loss": 49.3788,
+      "step": 11050
+    },
+    {
+      "epoch": 0.044683799496600236,
+      "grad_norm": 522.0982055664062,
+      "learning_rate": 8.937373737373737e-06,
+      "loss": 35.623,
+      "step": 11060
+    },
+    {
+      "epoch": 0.04472420076196786,
+      "grad_norm": 362.5142822265625,
+      "learning_rate": 8.945454545454546e-06,
+      "loss": 43.0607,
+      "step": 11070
+    },
+    {
+      "epoch": 0.044764602027335496,
+      "grad_norm": 925.5531616210938,
+      "learning_rate": 8.953535353535354e-06,
+      "loss": 57.3,
+      "step": 11080
+    },
+    {
+      "epoch": 0.04480500329270313,
+      "grad_norm": 797.6778564453125,
+      "learning_rate": 8.961616161616163e-06,
+      "loss": 44.5526,
+      "step": 11090
+    },
+    {
+      "epoch": 0.04484540455807076,
+      "grad_norm": 772.5079345703125,
+      "learning_rate": 8.969696969696971e-06,
+      "loss": 58.54,
+      "step": 11100
+    },
+    {
+      "epoch": 0.04488580582343839,
+      "grad_norm": 803.4046630859375,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 40.9027,
+      "step": 11110
+    },
+    {
+      "epoch": 0.04492620708880602,
+      "grad_norm": 814.4190063476562,
+      "learning_rate": 8.985858585858586e-06,
+      "loss": 54.8803,
+      "step": 11120
+    },
+    {
+      "epoch": 0.044966608354173654,
+      "grad_norm": 554.0245971679688,
+      "learning_rate": 8.993939393939395e-06,
+      "loss": 30.8268,
+      "step": 11130
+    },
+    {
+      "epoch": 0.04500700961954129,
+      "grad_norm": 861.0400390625,
+      "learning_rate": 9.002020202020203e-06,
+      "loss": 95.88,
+      "step": 11140
+    },
+    {
+      "epoch": 0.04504741088490891,
+      "grad_norm": 626.3646850585938,
+      "learning_rate": 9.010101010101012e-06,
+      "loss": 51.77,
+      "step": 11150
+    },
+    {
+      "epoch": 0.045087812150276546,
+      "grad_norm": 1681.829345703125,
+      "learning_rate": 9.01818181818182e-06,
+      "loss": 88.4558,
+      "step": 11160
+    },
+    {
+      "epoch": 0.04512821341564418,
+      "grad_norm": 1307.4827880859375,
+      "learning_rate": 9.026262626262627e-06,
+      "loss": 49.2158,
+      "step": 11170
+    },
+    {
+      "epoch": 0.04516861468101181,
+      "grad_norm": 690.13623046875,
+      "learning_rate": 9.034343434343435e-06,
+      "loss": 46.9865,
+      "step": 11180
+    },
+    {
+      "epoch": 0.04520901594637944,
+      "grad_norm": 1026.7313232421875,
+      "learning_rate": 9.042424242424244e-06,
+      "loss": 41.4207,
+      "step": 11190
+    },
+    {
+      "epoch": 0.04524941721174707,
+      "grad_norm": 1039.1217041015625,
+      "learning_rate": 9.050505050505052e-06,
+      "loss": 65.3547,
+      "step": 11200
+    },
+    {
+      "epoch": 0.045289818477114704,
+      "grad_norm": 605.8031616210938,
+      "learning_rate": 9.058585858585859e-06,
+      "loss": 51.7349,
+      "step": 11210
+    },
+    {
+      "epoch": 0.04533021974248234,
+      "grad_norm": 567.4804077148438,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 68.7462,
+      "step": 11220
+    },
+    {
+      "epoch": 0.045370621007849964,
+      "grad_norm": 801.2859497070312,
+      "learning_rate": 9.074747474747476e-06,
+      "loss": 38.204,
+      "step": 11230
+    },
+    {
+      "epoch": 0.0454110222732176,
+      "grad_norm": 715.8931274414062,
+      "learning_rate": 9.082828282828283e-06,
+      "loss": 78.7292,
+      "step": 11240
+    },
+    {
+      "epoch": 0.04545142353858523,
+      "grad_norm": 383.57635498046875,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 52.5061,
+      "step": 11250
+    },
+    {
+      "epoch": 0.04549182480395286,
+      "grad_norm": 658.3374633789062,
+      "learning_rate": 9.0989898989899e-06,
+      "loss": 60.3061,
+      "step": 11260
+    },
+    {
+      "epoch": 0.04553222606932049,
+      "grad_norm": 825.5767822265625,
+      "learning_rate": 9.107070707070708e-06,
+      "loss": 61.6082,
+      "step": 11270
+    },
+    {
+      "epoch": 0.04557262733468812,
+      "grad_norm": 673.5419921875,
+      "learning_rate": 9.115151515151516e-06,
+      "loss": 52.7299,
+      "step": 11280
+    },
+    {
+      "epoch": 0.045613028600055755,
+      "grad_norm": 2270.1708984375,
+      "learning_rate": 9.123232323232323e-06,
+      "loss": 87.2664,
+      "step": 11290
+    },
+    {
+      "epoch": 0.04565342986542339,
+      "grad_norm": 682.08544921875,
+      "learning_rate": 9.131313131313132e-06,
+      "loss": 49.0,
+      "step": 11300
+    },
+    {
+      "epoch": 0.045693831130791014,
+      "grad_norm": 435.1982727050781,
+      "learning_rate": 9.13939393939394e-06,
+      "loss": 46.8469,
+      "step": 11310
+    },
+    {
+      "epoch": 0.04573423239615865,
+      "grad_norm": 513.4351806640625,
+      "learning_rate": 9.147474747474748e-06,
+      "loss": 59.5729,
+      "step": 11320
+    },
+    {
+      "epoch": 0.04577463366152628,
+      "grad_norm": 701.8012084960938,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 50.215,
+      "step": 11330
+    },
+    {
+      "epoch": 0.04581503492689391,
+      "grad_norm": 1011.1109619140625,
+      "learning_rate": 9.163636363636365e-06,
+      "loss": 61.7965,
+      "step": 11340
+    },
+    {
+      "epoch": 0.04585543619226154,
+      "grad_norm": 1000.3032836914062,
+      "learning_rate": 9.171717171717172e-06,
+      "loss": 66.5579,
+      "step": 11350
+    },
+    {
+      "epoch": 0.04589583745762917,
+      "grad_norm": 892.2225341796875,
+      "learning_rate": 9.17979797979798e-06,
+      "loss": 77.5938,
+      "step": 11360
+    },
+    {
+      "epoch": 0.045936238722996806,
+      "grad_norm": 473.8362121582031,
+      "learning_rate": 9.187878787878789e-06,
+      "loss": 43.2337,
+      "step": 11370
+    },
+    {
+      "epoch": 0.04597663998836444,
+      "grad_norm": 602.120361328125,
+      "learning_rate": 9.195959595959597e-06,
+      "loss": 47.1484,
+      "step": 11380
+    },
+    {
+      "epoch": 0.046017041253732065,
+      "grad_norm": 481.7225341796875,
+      "learning_rate": 9.204040404040406e-06,
+      "loss": 44.948,
+      "step": 11390
+    },
+    {
+      "epoch": 0.0460574425190997,
+      "grad_norm": 617.6996459960938,
+      "learning_rate": 9.212121212121213e-06,
+      "loss": 45.0998,
+      "step": 11400
+    },
+    {
+      "epoch": 0.04609784378446733,
+      "grad_norm": 921.5098876953125,
+      "learning_rate": 9.220202020202021e-06,
+      "loss": 49.7482,
+      "step": 11410
+    },
+    {
+      "epoch": 0.046138245049834964,
+      "grad_norm": 1062.4364013671875,
+      "learning_rate": 9.228282828282828e-06,
+      "loss": 60.8331,
+      "step": 11420
+    },
+    {
+      "epoch": 0.04617864631520259,
+      "grad_norm": 991.2988891601562,
+      "learning_rate": 9.236363636363636e-06,
+      "loss": 62.9908,
+      "step": 11430
+    },
+    {
+      "epoch": 0.04621904758057022,
+      "grad_norm": 694.0717163085938,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 55.5939,
+      "step": 11440
+    },
+    {
+      "epoch": 0.046259448845937856,
+      "grad_norm": 468.04827880859375,
+      "learning_rate": 9.252525252525253e-06,
+      "loss": 42.1948,
+      "step": 11450
+    },
+    {
+      "epoch": 0.04629985011130549,
+      "grad_norm": 618.5086669921875,
+      "learning_rate": 9.260606060606062e-06,
+      "loss": 57.7817,
+      "step": 11460
+    },
+    {
+      "epoch": 0.046340251376673115,
+      "grad_norm": 668.7138061523438,
+      "learning_rate": 9.268686868686868e-06,
+      "loss": 47.0043,
+      "step": 11470
+    },
+    {
+      "epoch": 0.04638065264204075,
+      "grad_norm": 611.4515380859375,
+      "learning_rate": 9.276767676767677e-06,
+      "loss": 65.1178,
+      "step": 11480
+    },
+    {
+      "epoch": 0.04642105390740838,
+      "grad_norm": 465.3556823730469,
+      "learning_rate": 9.284848484848485e-06,
+      "loss": 49.5862,
+      "step": 11490
+    },
+    {
+      "epoch": 0.046461455172776014,
+      "grad_norm": 646.4593505859375,
+      "learning_rate": 9.292929292929294e-06,
+      "loss": 47.2889,
+      "step": 11500
+    },
+    {
+      "epoch": 0.04650185643814364,
+      "grad_norm": 465.8161926269531,
+      "learning_rate": 9.301010101010102e-06,
+      "loss": 49.2758,
+      "step": 11510
+    },
+    {
+      "epoch": 0.046542257703511274,
+      "grad_norm": 124.87579345703125,
+      "learning_rate": 9.30909090909091e-06,
+      "loss": 51.9798,
+      "step": 11520
+    },
+    {
+      "epoch": 0.04658265896887891,
+      "grad_norm": 455.3066711425781,
+      "learning_rate": 9.317171717171717e-06,
+      "loss": 48.4502,
+      "step": 11530
+    },
+    {
+      "epoch": 0.04662306023424654,
+      "grad_norm": 561.8794555664062,
+      "learning_rate": 9.325252525252526e-06,
+      "loss": 60.7198,
+      "step": 11540
+    },
+    {
+      "epoch": 0.046663461499614166,
+      "grad_norm": 1191.8280029296875,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 51.1963,
+      "step": 11550
+    },
+    {
+      "epoch": 0.0467038627649818,
+      "grad_norm": 757.8594360351562,
+      "learning_rate": 9.341414141414143e-06,
+      "loss": 67.2604,
+      "step": 11560
+    },
+    {
+      "epoch": 0.04674426403034943,
+      "grad_norm": 581.9176025390625,
+      "learning_rate": 9.349494949494951e-06,
+      "loss": 48.4778,
+      "step": 11570
+    },
+    {
+      "epoch": 0.046784665295717065,
+      "grad_norm": 326.5763244628906,
+      "learning_rate": 9.357575757575758e-06,
+      "loss": 35.4996,
+      "step": 11580
+    },
+    {
+      "epoch": 0.04682506656108469,
+      "grad_norm": 985.1808471679688,
+      "learning_rate": 9.365656565656566e-06,
+      "loss": 55.5719,
+      "step": 11590
+    },
+    {
+      "epoch": 0.046865467826452324,
+      "grad_norm": 626.5885009765625,
+      "learning_rate": 9.373737373737375e-06,
+      "loss": 37.6512,
+      "step": 11600
+    },
+    {
+      "epoch": 0.04690586909181996,
+      "grad_norm": 634.4691162109375,
+      "learning_rate": 9.381818181818183e-06,
+      "loss": 34.5778,
+      "step": 11610
+    },
+    {
+      "epoch": 0.04694627035718759,
+      "grad_norm": 934.1278686523438,
+      "learning_rate": 9.389898989898992e-06,
+      "loss": 46.384,
+      "step": 11620
+    },
+    {
+      "epoch": 0.046986671622555216,
+      "grad_norm": 784.7304077148438,
+      "learning_rate": 9.397979797979799e-06,
+      "loss": 68.3392,
+      "step": 11630
+    },
+    {
+      "epoch": 0.04702707288792285,
+      "grad_norm": 336.8243408203125,
+      "learning_rate": 9.406060606060607e-06,
+      "loss": 52.2806,
+      "step": 11640
+    },
+    {
+      "epoch": 0.04706747415329048,
+      "grad_norm": 854.4815063476562,
+      "learning_rate": 9.414141414141414e-06,
+      "loss": 56.4582,
+      "step": 11650
+    },
+    {
+      "epoch": 0.047107875418658116,
+      "grad_norm": 478.9867858886719,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 38.2133,
+      "step": 11660
+    },
+    {
+      "epoch": 0.04714827668402574,
+      "grad_norm": 349.7056884765625,
+      "learning_rate": 9.43030303030303e-06,
+      "loss": 42.0987,
+      "step": 11670
+    },
+    {
+      "epoch": 0.047188677949393375,
+      "grad_norm": 771.971923828125,
+      "learning_rate": 9.438383838383839e-06,
+      "loss": 92.2668,
+      "step": 11680
+    },
+    {
+      "epoch": 0.04722907921476101,
+      "grad_norm": 1389.6817626953125,
+      "learning_rate": 9.446464646464648e-06,
+      "loss": 93.1496,
+      "step": 11690
+    },
+    {
+      "epoch": 0.04726948048012864,
+      "grad_norm": 484.66351318359375,
+      "learning_rate": 9.454545454545456e-06,
+      "loss": 44.8321,
+      "step": 11700
+    },
+    {
+      "epoch": 0.04730988174549627,
+      "grad_norm": 1661.9954833984375,
+      "learning_rate": 9.462626262626263e-06,
+      "loss": 38.4946,
+      "step": 11710
+    },
+    {
+      "epoch": 0.0473502830108639,
+      "grad_norm": 616.4440307617188,
+      "learning_rate": 9.470707070707071e-06,
+      "loss": 51.6212,
+      "step": 11720
+    },
+    {
+      "epoch": 0.04739068427623153,
+      "grad_norm": 665.550048828125,
+      "learning_rate": 9.47878787878788e-06,
+      "loss": 63.4329,
+      "step": 11730
+    },
+    {
+      "epoch": 0.047431085541599166,
+      "grad_norm": 837.4865112304688,
+      "learning_rate": 9.486868686868688e-06,
+      "loss": 52.9131,
+      "step": 11740
+    },
+    {
+      "epoch": 0.04747148680696679,
+      "grad_norm": 695.9431762695312,
+      "learning_rate": 9.494949494949497e-06,
+      "loss": 49.8683,
+      "step": 11750
+    },
+    {
+      "epoch": 0.047511888072334425,
+      "grad_norm": 584.3406982421875,
+      "learning_rate": 9.503030303030303e-06,
+      "loss": 32.603,
+      "step": 11760
+    },
+    {
+      "epoch": 0.04755228933770206,
+      "grad_norm": 1272.3975830078125,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 56.4731,
+      "step": 11770
+    },
+    {
+      "epoch": 0.04759269060306969,
+      "grad_norm": 849.8159790039062,
+      "learning_rate": 9.51919191919192e-06,
+      "loss": 44.4619,
+      "step": 11780
+    },
+    {
+      "epoch": 0.04763309186843732,
+      "grad_norm": 967.6990356445312,
+      "learning_rate": 9.527272727272729e-06,
+      "loss": 52.3391,
+      "step": 11790
+    },
+    {
+      "epoch": 0.04767349313380495,
+      "grad_norm": 563.64990234375,
+      "learning_rate": 9.535353535353537e-06,
+      "loss": 58.6493,
+      "step": 11800
+    },
+    {
+      "epoch": 0.047713894399172584,
+      "grad_norm": 410.8050537109375,
+      "learning_rate": 9.543434343434344e-06,
+      "loss": 70.926,
+      "step": 11810
+    },
+    {
+      "epoch": 0.04775429566454022,
+      "grad_norm": 932.15185546875,
+      "learning_rate": 9.551515151515152e-06,
+      "loss": 54.6105,
+      "step": 11820
+    },
+    {
+      "epoch": 0.04779469692990784,
+      "grad_norm": 741.9338989257812,
+      "learning_rate": 9.55959595959596e-06,
+      "loss": 62.3345,
+      "step": 11830
+    },
+    {
+      "epoch": 0.047835098195275476,
+      "grad_norm": 920.0084228515625,
+      "learning_rate": 9.56767676767677e-06,
+      "loss": 62.9895,
+      "step": 11840
+    },
+    {
+      "epoch": 0.04787549946064311,
+      "grad_norm": 612.9961547851562,
+      "learning_rate": 9.575757575757576e-06,
+      "loss": 48.8344,
+      "step": 11850
+    },
+    {
+      "epoch": 0.04791590072601074,
+      "grad_norm": 679.0940551757812,
+      "learning_rate": 9.583838383838384e-06,
+      "loss": 50.3198,
+      "step": 11860
+    },
+    {
+      "epoch": 0.04795630199137837,
+      "grad_norm": 680.0853881835938,
+      "learning_rate": 9.591919191919193e-06,
+      "loss": 71.3714,
+      "step": 11870
+    },
+    {
+      "epoch": 0.047996703256746,
+      "grad_norm": 651.5006103515625,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 83.7444,
+      "step": 11880
+    },
+    {
+      "epoch": 0.048037104522113634,
+      "grad_norm": 803.4596557617188,
+      "learning_rate": 9.608080808080808e-06,
+      "loss": 69.2039,
+      "step": 11890
+    },
+    {
+      "epoch": 0.04807750578748127,
+      "grad_norm": 5523.23193359375,
+      "learning_rate": 9.616161616161616e-06,
+      "loss": 72.4104,
+      "step": 11900
+    },
+    {
+      "epoch": 0.04811790705284889,
+      "grad_norm": 638.7211303710938,
+      "learning_rate": 9.624242424242425e-06,
+      "loss": 50.5673,
+      "step": 11910
+    },
+    {
+      "epoch": 0.048158308318216526,
+      "grad_norm": 909.378173828125,
+      "learning_rate": 9.632323232323233e-06,
+      "loss": 39.5691,
+      "step": 11920
+    },
+    {
+      "epoch": 0.04819870958358416,
+      "grad_norm": 615.8162841796875,
+      "learning_rate": 9.640404040404042e-06,
+      "loss": 37.2438,
+      "step": 11930
+    },
+    {
+      "epoch": 0.04823911084895179,
+      "grad_norm": 655.8267822265625,
+      "learning_rate": 9.648484848484849e-06,
+      "loss": 52.3094,
+      "step": 11940
+    },
+    {
+      "epoch": 0.04827951211431942,
+      "grad_norm": 766.4174194335938,
+      "learning_rate": 9.656565656565657e-06,
+      "loss": 68.0634,
+      "step": 11950
+    },
+    {
+      "epoch": 0.04831991337968705,
+      "grad_norm": 637.963623046875,
+      "learning_rate": 9.664646464646465e-06,
+      "loss": 48.027,
+      "step": 11960
+    },
+    {
+      "epoch": 0.048360314645054685,
+      "grad_norm": 570.1321411132812,
+      "learning_rate": 9.672727272727274e-06,
+      "loss": 43.3436,
+      "step": 11970
+    },
+    {
+      "epoch": 0.04840071591042232,
+      "grad_norm": 970.7443237304688,
+      "learning_rate": 9.680808080808082e-06,
+      "loss": 55.089,
+      "step": 11980
+    },
+    {
+      "epoch": 0.048441117175789944,
+      "grad_norm": 1095.2872314453125,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 95.5435,
+      "step": 11990
+    },
+    {
+      "epoch": 0.04848151844115758,
+      "grad_norm": 219.54226684570312,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 60.4978,
+      "step": 12000
+    },
+    {
+      "epoch": 0.04852191970652521,
+      "grad_norm": 605.0892944335938,
+      "learning_rate": 9.705050505050506e-06,
+      "loss": 50.5544,
+      "step": 12010
+    },
+    {
+      "epoch": 0.04856232097189284,
+      "grad_norm": 574.8591918945312,
+      "learning_rate": 9.713131313131314e-06,
+      "loss": 48.281,
+      "step": 12020
+    },
+    {
+      "epoch": 0.04860272223726047,
+      "grad_norm": 529.4436645507812,
+      "learning_rate": 9.721212121212123e-06,
+      "loss": 65.4045,
+      "step": 12030
+    },
+    {
+      "epoch": 0.0486431235026281,
+      "grad_norm": 1770.2958984375,
+      "learning_rate": 9.729292929292931e-06,
+      "loss": 59.0963,
+      "step": 12040
+    },
+    {
+      "epoch": 0.048683524767995735,
+      "grad_norm": 462.4886169433594,
+      "learning_rate": 9.737373737373738e-06,
+      "loss": 66.9072,
+      "step": 12050
+    },
+    {
+      "epoch": 0.04872392603336337,
+      "grad_norm": 541.0017700195312,
+      "learning_rate": 9.745454545454547e-06,
+      "loss": 53.1198,
+      "step": 12060
+    },
+    {
+      "epoch": 0.048764327298730994,
+      "grad_norm": 901.8868408203125,
+      "learning_rate": 9.753535353535353e-06,
+      "loss": 47.6384,
+      "step": 12070
+    },
+    {
+      "epoch": 0.04880472856409863,
+      "grad_norm": 1244.428955078125,
+      "learning_rate": 9.761616161616162e-06,
+      "loss": 48.4008,
+      "step": 12080
+    },
+    {
+      "epoch": 0.04884512982946626,
+      "grad_norm": 1056.5343017578125,
+      "learning_rate": 9.76969696969697e-06,
+      "loss": 89.3727,
+      "step": 12090
+    },
+    {
+      "epoch": 0.048885531094833894,
+      "grad_norm": 539.5419921875,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 60.6288,
+      "step": 12100
+    },
+    {
+      "epoch": 0.04892593236020152,
+      "grad_norm": 662.9339599609375,
+      "learning_rate": 9.785858585858587e-06,
+      "loss": 43.1786,
+      "step": 12110
+    },
+    {
+      "epoch": 0.04896633362556915,
+      "grad_norm": 789.1748657226562,
+      "learning_rate": 9.793939393939394e-06,
+      "loss": 55.6451,
+      "step": 12120
+    },
+    {
+      "epoch": 0.049006734890936786,
+      "grad_norm": 712.3914184570312,
+      "learning_rate": 9.802020202020202e-06,
+      "loss": 55.9028,
+      "step": 12130
+    },
+    {
+      "epoch": 0.04904713615630442,
+      "grad_norm": 552.1458740234375,
+      "learning_rate": 9.81010101010101e-06,
+      "loss": 55.985,
+      "step": 12140
+    },
+    {
+      "epoch": 0.049087537421672045,
+      "grad_norm": 785.1761474609375,
+      "learning_rate": 9.81818181818182e-06,
+      "loss": 67.8702,
+      "step": 12150
+    },
+    {
+      "epoch": 0.04912793868703968,
+      "grad_norm": 2131.36669921875,
+      "learning_rate": 9.826262626262628e-06,
+      "loss": 70.2431,
+      "step": 12160
+    },
+    {
+      "epoch": 0.04916833995240731,
+      "grad_norm": 712.6942138671875,
+      "learning_rate": 9.834343434343434e-06,
+      "loss": 76.463,
+      "step": 12170
+    },
+    {
+      "epoch": 0.04920874121777494,
+      "grad_norm": 720.9066162109375,
+      "learning_rate": 9.842424242424243e-06,
+      "loss": 51.4022,
+      "step": 12180
+    },
+    {
+      "epoch": 0.04924914248314257,
+      "grad_norm": 563.6546020507812,
+      "learning_rate": 9.850505050505051e-06,
+      "loss": 57.7309,
+      "step": 12190
+    },
+    {
+      "epoch": 0.0492895437485102,
+      "grad_norm": 1423.6729736328125,
+      "learning_rate": 9.85858585858586e-06,
+      "loss": 45.7969,
+      "step": 12200
+    },
+    {
+      "epoch": 0.049329945013877836,
+      "grad_norm": 649.6131591796875,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 58.7549,
+      "step": 12210
+    },
+    {
+      "epoch": 0.04937034627924546,
+      "grad_norm": 614.8030395507812,
+      "learning_rate": 9.874747474747477e-06,
+      "loss": 44.4937,
+      "step": 12220
+    },
+    {
+      "epoch": 0.049410747544613096,
+      "grad_norm": 1071.0455322265625,
+      "learning_rate": 9.882828282828283e-06,
+      "loss": 64.4793,
+      "step": 12230
+    },
+    {
+      "epoch": 0.04945114880998073,
+      "grad_norm": 502.5149230957031,
+      "learning_rate": 9.890909090909092e-06,
+      "loss": 51.8935,
+      "step": 12240
+    },
+    {
+      "epoch": 0.04949155007534836,
+      "grad_norm": 635.7072143554688,
+      "learning_rate": 9.8989898989899e-06,
+      "loss": 77.3302,
+      "step": 12250
+    },
+    {
+      "epoch": 0.04953195134071599,
+      "grad_norm": 243.91246032714844,
+      "learning_rate": 9.907070707070709e-06,
+      "loss": 59.5737,
+      "step": 12260
+    },
+    {
+      "epoch": 0.04957235260608362,
+      "grad_norm": 769.7122802734375,
+      "learning_rate": 9.915151515151515e-06,
+      "loss": 48.6593,
+      "step": 12270
+    },
+    {
+      "epoch": 0.049612753871451254,
+      "grad_norm": 923.63330078125,
+      "learning_rate": 9.923232323232324e-06,
+      "loss": 51.4512,
+      "step": 12280
+    },
+    {
+      "epoch": 0.04965315513681889,
+      "grad_norm": 374.1686096191406,
+      "learning_rate": 9.931313131313132e-06,
+      "loss": 37.1888,
+      "step": 12290
+    },
+    {
+      "epoch": 0.04969355640218651,
+      "grad_norm": 1896.8494873046875,
+      "learning_rate": 9.939393939393939e-06,
+      "loss": 62.5259,
+      "step": 12300
+    },
+    {
+      "epoch": 0.049733957667554146,
+      "grad_norm": 845.9722290039062,
+      "learning_rate": 9.947474747474748e-06,
+      "loss": 45.9669,
+      "step": 12310
+    },
+    {
+      "epoch": 0.04977435893292178,
+      "grad_norm": 628.6317138671875,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 48.4875,
+      "step": 12320
+    },
+    {
+      "epoch": 0.04981476019828941,
+      "grad_norm": 661.7048950195312,
+      "learning_rate": 9.963636363636364e-06,
+      "loss": 50.1969,
+      "step": 12330
+    },
+    {
+      "epoch": 0.04985516146365704,
+      "grad_norm": 302.84808349609375,
+      "learning_rate": 9.971717171717173e-06,
+      "loss": 78.5323,
+      "step": 12340
+    },
+    {
+      "epoch": 0.04989556272902467,
+      "grad_norm": 778.6891479492188,
+      "learning_rate": 9.97979797979798e-06,
+      "loss": 58.8246,
+      "step": 12350
+    },
+    {
+      "epoch": 0.049935963994392304,
+      "grad_norm": 918.199951171875,
+      "learning_rate": 9.987878787878788e-06,
+      "loss": 49.6424,
+      "step": 12360
+    },
+    {
+      "epoch": 0.04997636525975994,
+      "grad_norm": 1064.10302734375,
+      "learning_rate": 9.995959595959597e-06,
+      "loss": 61.3562,
+      "step": 12370
+    },
+    {
+      "epoch": 0.050016766525127564,
+      "grad_norm": 1215.5438232421875,
+      "learning_rate": 9.99999995027162e-06,
+      "loss": 59.4952,
+      "step": 12380
+    },
+    {
+      "epoch": 0.0500571677904952,
+      "grad_norm": 1015.760986328125,
+      "learning_rate": 9.99999955244457e-06,
+      "loss": 58.8701,
+      "step": 12390
+    },
+    {
+      "epoch": 0.05009756905586283,
+      "grad_norm": 547.9041748046875,
+      "learning_rate": 9.999998756790503e-06,
+      "loss": 63.0136,
+      "step": 12400
+    },
+    {
+      "epoch": 0.05013797032123046,
+      "grad_norm": 871.2403564453125,
+      "learning_rate": 9.999997563309483e-06,
+      "loss": 69.3615,
+      "step": 12410
+    },
+    {
+      "epoch": 0.05017837158659809,
+      "grad_norm": 801.6481323242188,
+      "learning_rate": 9.999995972001602e-06,
+      "loss": 46.739,
+      "step": 12420
+    },
+    {
+      "epoch": 0.05021877285196572,
+      "grad_norm": 704.8037109375,
+      "learning_rate": 9.99999398286699e-06,
+      "loss": 74.7781,
+      "step": 12430
+    },
+    {
+      "epoch": 0.050259174117333355,
+      "grad_norm": 741.2653198242188,
+      "learning_rate": 9.999991595905803e-06,
+      "loss": 68.1409,
+      "step": 12440
+    },
+    {
+      "epoch": 0.05029957538270099,
+      "grad_norm": 575.7168579101562,
+      "learning_rate": 9.999988811118232e-06,
+      "loss": 48.7718,
+      "step": 12450
+    },
+    {
+      "epoch": 0.050339976648068614,
+      "grad_norm": 867.4459838867188,
+      "learning_rate": 9.999985628504498e-06,
+      "loss": 61.8401,
+      "step": 12460
+    },
+    {
+      "epoch": 0.05038037791343625,
+      "grad_norm": 525.7904663085938,
+      "learning_rate": 9.999982048064854e-06,
+      "loss": 36.3457,
+      "step": 12470
+    },
+    {
+      "epoch": 0.05042077917880388,
+      "grad_norm": 480.1571044921875,
+      "learning_rate": 9.999978069799585e-06,
+      "loss": 75.6131,
+      "step": 12480
+    },
+    {
+      "epoch": 0.05046118044417151,
+      "grad_norm": 545.6644897460938,
+      "learning_rate": 9.999973693709008e-06,
+      "loss": 60.0911,
+      "step": 12490
+    },
+    {
+      "epoch": 0.05050158170953914,
+      "grad_norm": 547.3186645507812,
+      "learning_rate": 9.99996891979347e-06,
+      "loss": 51.344,
+      "step": 12500
+    },
+    {
+      "epoch": 0.05054198297490677,
+      "grad_norm": 934.025146484375,
+      "learning_rate": 9.999963748053354e-06,
+      "loss": 54.9807,
+      "step": 12510
+    },
+    {
+      "epoch": 0.050582384240274406,
+      "grad_norm": 963.864501953125,
+      "learning_rate": 9.999958178489069e-06,
+      "loss": 56.3869,
+      "step": 12520
+    },
+    {
+      "epoch": 0.05062278550564204,
+      "grad_norm": 762.842041015625,
+      "learning_rate": 9.999952211101056e-06,
+      "loss": 44.2664,
+      "step": 12530
+    },
+    {
+      "epoch": 0.050663186771009665,
+      "grad_norm": 1445.8870849609375,
+      "learning_rate": 9.999945845889795e-06,
+      "loss": 73.7543,
+      "step": 12540
+    },
+    {
+      "epoch": 0.0507035880363773,
+      "grad_norm": 1076.540283203125,
+      "learning_rate": 9.999939082855788e-06,
+      "loss": 53.5912,
+      "step": 12550
+    },
+    {
+      "epoch": 0.05074398930174493,
+      "grad_norm": 438.9979248046875,
+      "learning_rate": 9.999931921999575e-06,
+      "loss": 58.8434,
+      "step": 12560
+    },
+    {
+      "epoch": 0.050784390567112564,
+      "grad_norm": 381.0602111816406,
+      "learning_rate": 9.999924363321726e-06,
+      "loss": 76.014,
+      "step": 12570
+    },
+    {
+      "epoch": 0.05082479183248019,
+      "grad_norm": 394.7668762207031,
+      "learning_rate": 9.999916406822843e-06,
+      "loss": 73.4046,
+      "step": 12580
+    },
+    {
+      "epoch": 0.05086519309784782,
+      "grad_norm": 225.88870239257812,
+      "learning_rate": 9.999908052503557e-06,
+      "loss": 69.8275,
+      "step": 12590
+    },
+    {
+      "epoch": 0.050905594363215456,
+      "grad_norm": 512.4158325195312,
+      "learning_rate": 9.999899300364534e-06,
+      "loss": 40.6297,
+      "step": 12600
+    },
+    {
+      "epoch": 0.05094599562858309,
+      "grad_norm": 404.0837707519531,
+      "learning_rate": 9.99989015040647e-06,
+      "loss": 48.8008,
+      "step": 12610
+    },
+    {
+      "epoch": 0.050986396893950715,
+      "grad_norm": 693.5153198242188,
+      "learning_rate": 9.999880602630092e-06,
+      "loss": 61.1755,
+      "step": 12620
+    },
+    {
+      "epoch": 0.05102679815931835,
+      "grad_norm": 750.3567504882812,
+      "learning_rate": 9.999870657036161e-06,
+      "loss": 59.9361,
+      "step": 12630
+    },
+    {
+      "epoch": 0.05106719942468598,
+      "grad_norm": 229.11830139160156,
+      "learning_rate": 9.99986031362547e-06,
+      "loss": 49.4575,
+      "step": 12640
+    },
+    {
+      "epoch": 0.051107600690053615,
+      "grad_norm": 1311.7059326171875,
+      "learning_rate": 9.99984957239884e-06,
+      "loss": 71.9959,
+      "step": 12650
+    },
+    {
+      "epoch": 0.05114800195542124,
+      "grad_norm": 416.9798889160156,
+      "learning_rate": 9.999838433357124e-06,
+      "loss": 51.3517,
+      "step": 12660
+    },
+    {
+      "epoch": 0.051188403220788874,
+      "grad_norm": 955.900634765625,
+      "learning_rate": 9.99982689650121e-06,
+      "loss": 43.4182,
+      "step": 12670
+    },
+    {
+      "epoch": 0.05122880448615651,
+      "grad_norm": 705.177978515625,
+      "learning_rate": 9.999814961832018e-06,
+      "loss": 49.622,
+      "step": 12680
+    },
+    {
+      "epoch": 0.05126920575152414,
+      "grad_norm": 713.4158935546875,
+      "learning_rate": 9.999802629350492e-06,
+      "loss": 54.9794,
+      "step": 12690
+    },
+    {
+      "epoch": 0.051309607016891766,
+      "grad_norm": 408.42449951171875,
+      "learning_rate": 9.99978989905762e-06,
+      "loss": 59.9696,
+      "step": 12700
+    },
+    {
+      "epoch": 0.0513500082822594,
+      "grad_norm": 513.5841064453125,
+      "learning_rate": 9.999776770954411e-06,
+      "loss": 51.2767,
+      "step": 12710
+    },
+    {
+      "epoch": 0.05139040954762703,
+      "grad_norm": 621.69140625,
+      "learning_rate": 9.99976324504191e-06,
+      "loss": 43.1716,
+      "step": 12720
+    },
+    {
+      "epoch": 0.051430810812994665,
+      "grad_norm": 666.1728515625,
+      "learning_rate": 9.999749321321192e-06,
+      "loss": 54.0102,
+      "step": 12730
+    },
+    {
+      "epoch": 0.05147121207836229,
+      "grad_norm": 600.5795288085938,
+      "learning_rate": 9.999734999793369e-06,
+      "loss": 47.6661,
+      "step": 12740
+    },
+    {
+      "epoch": 0.051511613343729924,
+      "grad_norm": 628.9915161132812,
+      "learning_rate": 9.999720280459576e-06,
+      "loss": 43.5194,
+      "step": 12750
+    },
+    {
+      "epoch": 0.05155201460909756,
+      "grad_norm": 417.6671447753906,
+      "learning_rate": 9.999705163320987e-06,
+      "loss": 54.9336,
+      "step": 12760
+    },
+    {
+      "epoch": 0.05159241587446519,
+      "grad_norm": 1952.5086669921875,
+      "learning_rate": 9.999689648378801e-06,
+      "loss": 79.7189,
+      "step": 12770
+    },
+    {
+      "epoch": 0.051632817139832816,
+      "grad_norm": 537.900634765625,
+      "learning_rate": 9.999673735634259e-06,
+      "loss": 72.6476,
+      "step": 12780
+    },
+    {
+      "epoch": 0.05167321840520045,
+      "grad_norm": 757.7594604492188,
+      "learning_rate": 9.99965742508862e-06,
+      "loss": 71.4085,
+      "step": 12790
+    },
+    {
+      "epoch": 0.05171361967056808,
+      "grad_norm": 515.7596435546875,
+      "learning_rate": 9.999640716743186e-06,
+      "loss": 89.0427,
+      "step": 12800
+    },
+    {
+      "epoch": 0.051754020935935716,
+      "grad_norm": 644.1132202148438,
+      "learning_rate": 9.999623610599287e-06,
+      "loss": 62.6025,
+      "step": 12810
+    },
+    {
+      "epoch": 0.05179442220130334,
+      "grad_norm": 652.4735107421875,
+      "learning_rate": 9.999606106658282e-06,
+      "loss": 50.8262,
+      "step": 12820
+    },
+    {
+      "epoch": 0.051834823466670975,
+      "grad_norm": 646.9786987304688,
+      "learning_rate": 9.999588204921562e-06,
+      "loss": 70.1996,
+      "step": 12830
+    },
+    {
+      "epoch": 0.05187522473203861,
+      "grad_norm": 792.312744140625,
+      "learning_rate": 9.999569905390556e-06,
+      "loss": 55.6008,
+      "step": 12840
+    },
+    {
+      "epoch": 0.05191562599740624,
+      "grad_norm": 635.0096435546875,
+      "learning_rate": 9.999551208066716e-06,
+      "loss": 47.4141,
+      "step": 12850
+    },
+    {
+      "epoch": 0.05195602726277387,
+      "grad_norm": 364.86016845703125,
+      "learning_rate": 9.99953211295153e-06,
+      "loss": 43.5858,
+      "step": 12860
+    },
+    {
+      "epoch": 0.0519964285281415,
+      "grad_norm": 928.7925415039062,
+      "learning_rate": 9.999512620046523e-06,
+      "loss": 60.4772,
+      "step": 12870
+    },
+    {
+      "epoch": 0.05203682979350913,
+      "grad_norm": 1354.784912109375,
+      "learning_rate": 9.999492729353238e-06,
+      "loss": 58.8818,
+      "step": 12880
+    },
+    {
+      "epoch": 0.052077231058876766,
+      "grad_norm": 715.9904174804688,
+      "learning_rate": 9.999472440873261e-06,
+      "loss": 46.8977,
+      "step": 12890
+    },
+    {
+      "epoch": 0.05211763232424439,
+      "grad_norm": 543.8614501953125,
+      "learning_rate": 9.999451754608208e-06,
+      "loss": 71.3133,
+      "step": 12900
+    },
+    {
+      "epoch": 0.052158033589612025,
+      "grad_norm": 497.0710754394531,
+      "learning_rate": 9.999430670559723e-06,
+      "loss": 59.7093,
+      "step": 12910
+    },
+    {
+      "epoch": 0.05219843485497966,
+      "grad_norm": 863.6559448242188,
+      "learning_rate": 9.999409188729484e-06,
+      "loss": 60.5322,
+      "step": 12920
+    },
+    {
+      "epoch": 0.05223883612034729,
+      "grad_norm": 2148.5078125,
+      "learning_rate": 9.999387309119198e-06,
+      "loss": 56.2072,
+      "step": 12930
+    },
+    {
+      "epoch": 0.05227923738571492,
+      "grad_norm": 1144.13037109375,
+      "learning_rate": 9.999365031730609e-06,
+      "loss": 58.272,
+      "step": 12940
+    },
+    {
+      "epoch": 0.05231963865108255,
+      "grad_norm": 2413.07763671875,
+      "learning_rate": 9.99934235656549e-06,
+      "loss": 70.858,
+      "step": 12950
+    },
+    {
+      "epoch": 0.052360039916450184,
+      "grad_norm": 906.7877197265625,
+      "learning_rate": 9.999319283625641e-06,
+      "loss": 62.3502,
+      "step": 12960
+    },
+    {
+      "epoch": 0.05240044118181782,
+      "grad_norm": 534.9559326171875,
+      "learning_rate": 9.999295812912902e-06,
+      "loss": 51.2167,
+      "step": 12970
+    },
+    {
+      "epoch": 0.05244084244718544,
+      "grad_norm": 652.7603759765625,
+      "learning_rate": 9.999271944429139e-06,
+      "loss": 48.0079,
+      "step": 12980
+    },
+    {
+      "epoch": 0.052481243712553076,
+      "grad_norm": 381.3694763183594,
+      "learning_rate": 9.99924767817625e-06,
+      "loss": 35.3519,
+      "step": 12990
+    },
+    {
+      "epoch": 0.05252164497792071,
+      "grad_norm": 1747.3424072265625,
+      "learning_rate": 9.999223014156167e-06,
+      "loss": 49.7469,
+      "step": 13000
+    },
+    {
+      "epoch": 0.05256204624328834,
+      "grad_norm": 862.7362670898438,
+      "learning_rate": 9.999197952370851e-06,
+      "loss": 55.1828,
+      "step": 13010
+    },
+    {
+      "epoch": 0.05260244750865597,
+      "grad_norm": 786.5454711914062,
+      "learning_rate": 9.9991724928223e-06,
+      "loss": 37.0467,
+      "step": 13020
+    },
+    {
+      "epoch": 0.0526428487740236,
+      "grad_norm": 314.32000732421875,
+      "learning_rate": 9.999146635512535e-06,
+      "loss": 45.9797,
+      "step": 13030
+    },
+    {
+      "epoch": 0.052683250039391234,
+      "grad_norm": 694.8716430664062,
+      "learning_rate": 9.999120380443614e-06,
+      "loss": 46.7638,
+      "step": 13040
+    },
+    {
+      "epoch": 0.05272365130475887,
+      "grad_norm": 303.11627197265625,
+      "learning_rate": 9.99909372761763e-06,
+      "loss": 36.8872,
+      "step": 13050
+    },
+    {
+      "epoch": 0.05276405257012649,
+      "grad_norm": 517.9519653320312,
+      "learning_rate": 9.9990666770367e-06,
+      "loss": 45.6657,
+      "step": 13060
+    },
+    {
+      "epoch": 0.052804453835494126,
+      "grad_norm": 1371.8948974609375,
+      "learning_rate": 9.999039228702975e-06,
+      "loss": 89.3382,
+      "step": 13070
+    },
+    {
+      "epoch": 0.05284485510086176,
+      "grad_norm": 864.37890625,
+      "learning_rate": 9.999011382618644e-06,
+      "loss": 58.1169,
+      "step": 13080
+    },
+    {
+      "epoch": 0.05288525636622939,
+      "grad_norm": 522.5619506835938,
+      "learning_rate": 9.998983138785919e-06,
+      "loss": 42.7408,
+      "step": 13090
+    },
+    {
+      "epoch": 0.05292565763159702,
+      "grad_norm": 912.5321044921875,
+      "learning_rate": 9.998954497207045e-06,
+      "loss": 50.1915,
+      "step": 13100
+    },
+    {
+      "epoch": 0.05296605889696465,
+      "grad_norm": 584.2015380859375,
+      "learning_rate": 9.998925457884307e-06,
+      "loss": 46.6389,
+      "step": 13110
+    },
+    {
+      "epoch": 0.053006460162332285,
+      "grad_norm": 835.9763793945312,
+      "learning_rate": 9.99889602082001e-06,
+      "loss": 38.9002,
+      "step": 13120
+    },
+    {
+      "epoch": 0.05304686142769992,
+      "grad_norm": 533.8273315429688,
+      "learning_rate": 9.998866186016501e-06,
+      "loss": 56.9394,
+      "step": 13130
+    },
+    {
+      "epoch": 0.053087262693067544,
+      "grad_norm": 463.4489440917969,
+      "learning_rate": 9.99883595347615e-06,
+      "loss": 56.6824,
+      "step": 13140
+    },
+    {
+      "epoch": 0.05312766395843518,
+      "grad_norm": 407.6231994628906,
+      "learning_rate": 9.998805323201364e-06,
+      "loss": 47.9757,
+      "step": 13150
+    },
+    {
+      "epoch": 0.05316806522380281,
+      "grad_norm": 490.29644775390625,
+      "learning_rate": 9.998774295194579e-06,
+      "loss": 58.1626,
+      "step": 13160
+    },
+    {
+      "epoch": 0.05320846648917044,
+      "grad_norm": 1248.9224853515625,
+      "learning_rate": 9.998742869458264e-06,
+      "loss": 55.3795,
+      "step": 13170
+    },
+    {
+      "epoch": 0.05324886775453807,
+      "grad_norm": 1146.1629638671875,
+      "learning_rate": 9.998711045994922e-06,
+      "loss": 79.3144,
+      "step": 13180
+    },
+    {
+      "epoch": 0.0532892690199057,
+      "grad_norm": 511.3272705078125,
+      "learning_rate": 9.998678824807082e-06,
+      "loss": 40.7286,
+      "step": 13190
+    },
+    {
+      "epoch": 0.053329670285273335,
+      "grad_norm": 711.6715087890625,
+      "learning_rate": 9.99864620589731e-06,
+      "loss": 44.7338,
+      "step": 13200
+    },
+    {
+      "epoch": 0.05337007155064097,
+      "grad_norm": 602.4719848632812,
+      "learning_rate": 9.998613189268197e-06,
+      "loss": 59.2179,
+      "step": 13210
+    },
+    {
+      "epoch": 0.053410472816008595,
+      "grad_norm": 600.1394653320312,
+      "learning_rate": 9.998579774922377e-06,
+      "loss": 39.6892,
+      "step": 13220
+    },
+    {
+      "epoch": 0.05345087408137623,
+      "grad_norm": 232.3237762451172,
+      "learning_rate": 9.998545962862503e-06,
+      "loss": 49.2781,
+      "step": 13230
+    },
+    {
+      "epoch": 0.05349127534674386,
+      "grad_norm": 589.8071899414062,
+      "learning_rate": 9.998511753091267e-06,
+      "loss": 52.6975,
+      "step": 13240
+    },
+    {
+      "epoch": 0.053531676612111494,
+      "grad_norm": 620.8717651367188,
+      "learning_rate": 9.998477145611389e-06,
+      "loss": 93.4042,
+      "step": 13250
+    },
+    {
+      "epoch": 0.05357207787747912,
+      "grad_norm": 435.4456787109375,
+      "learning_rate": 9.998442140425625e-06,
+      "loss": 46.4562,
+      "step": 13260
+    },
+    {
+      "epoch": 0.05361247914284675,
+      "grad_norm": 427.4245910644531,
+      "learning_rate": 9.998406737536761e-06,
+      "loss": 48.8901,
+      "step": 13270
+    },
+    {
+      "epoch": 0.053652880408214386,
+      "grad_norm": 708.8045654296875,
+      "learning_rate": 9.998370936947614e-06,
+      "loss": 57.0132,
+      "step": 13280
+    },
+    {
+      "epoch": 0.05369328167358202,
+      "grad_norm": 746.7619018554688,
+      "learning_rate": 9.998334738661028e-06,
+      "loss": 51.5482,
+      "step": 13290
+    },
+    {
+      "epoch": 0.053733682938949645,
+      "grad_norm": 613.72265625,
+      "learning_rate": 9.998298142679888e-06,
+      "loss": 62.2609,
+      "step": 13300
+    },
+    {
+      "epoch": 0.05377408420431728,
+      "grad_norm": 507.6068420410156,
+      "learning_rate": 9.998261149007104e-06,
+      "loss": 36.7143,
+      "step": 13310
+    },
+    {
+      "epoch": 0.05381448546968491,
+      "grad_norm": 515.9215698242188,
+      "learning_rate": 9.998223757645618e-06,
+      "loss": 57.8858,
+      "step": 13320
+    },
+    {
+      "epoch": 0.053854886735052544,
+      "grad_norm": 480.4765625,
+      "learning_rate": 9.998185968598407e-06,
+      "loss": 73.2252,
+      "step": 13330
+    },
+    {
+      "epoch": 0.05389528800042017,
+      "grad_norm": 892.7890014648438,
+      "learning_rate": 9.998147781868477e-06,
+      "loss": 58.0664,
+      "step": 13340
+    },
+    {
+      "epoch": 0.0539356892657878,
+      "grad_norm": 459.78326416015625,
+      "learning_rate": 9.998109197458865e-06,
+      "loss": 42.0378,
+      "step": 13350
+    },
+    {
+      "epoch": 0.053976090531155436,
+      "grad_norm": 864.340576171875,
+      "learning_rate": 9.998070215372645e-06,
+      "loss": 63.3425,
+      "step": 13360
+    },
+    {
+      "epoch": 0.05401649179652307,
+      "grad_norm": 818.0133666992188,
+      "learning_rate": 9.998030835612914e-06,
+      "loss": 54.6549,
+      "step": 13370
+    },
+    {
+      "epoch": 0.054056893061890696,
+      "grad_norm": 750.6216430664062,
+      "learning_rate": 9.997991058182807e-06,
+      "loss": 42.7049,
+      "step": 13380
+    },
+    {
+      "epoch": 0.05409729432725833,
+      "grad_norm": 749.8779907226562,
+      "learning_rate": 9.997950883085492e-06,
+      "loss": 74.4298,
+      "step": 13390
+    },
+    {
+      "epoch": 0.05413769559262596,
+      "grad_norm": 454.1107482910156,
+      "learning_rate": 9.99791031032416e-06,
+      "loss": 45.0433,
+      "step": 13400
+    },
+    {
+      "epoch": 0.054178096857993595,
+      "grad_norm": 867.2178955078125,
+      "learning_rate": 9.997869339902043e-06,
+      "loss": 47.866,
+      "step": 13410
+    },
+    {
+      "epoch": 0.05421849812336122,
+      "grad_norm": 679.61962890625,
+      "learning_rate": 9.9978279718224e-06,
+      "loss": 67.6122,
+      "step": 13420
+    },
+    {
+      "epoch": 0.054258899388728854,
+      "grad_norm": 528.4546508789062,
+      "learning_rate": 9.99778620608852e-06,
+      "loss": 40.4731,
+      "step": 13430
+    },
+    {
+      "epoch": 0.05429930065409649,
+      "grad_norm": 557.4871826171875,
+      "learning_rate": 9.997744042703731e-06,
+      "loss": 62.1209,
+      "step": 13440
+    },
+    {
+      "epoch": 0.05433970191946412,
+      "grad_norm": 1284.357177734375,
+      "learning_rate": 9.997701481671384e-06,
+      "loss": 65.1331,
+      "step": 13450
+    },
+    {
+      "epoch": 0.054380103184831746,
+      "grad_norm": 566.5718383789062,
+      "learning_rate": 9.997658522994867e-06,
+      "loss": 56.0474,
+      "step": 13460
+    },
+    {
+      "epoch": 0.05442050445019938,
+      "grad_norm": 1788.369873046875,
+      "learning_rate": 9.997615166677597e-06,
+      "loss": 67.9537,
+      "step": 13470
+    },
+    {
+      "epoch": 0.05446090571556701,
+      "grad_norm": 524.1817016601562,
+      "learning_rate": 9.997571412723024e-06,
+      "loss": 58.4599,
+      "step": 13480
+    },
+    {
+      "epoch": 0.054501306980934645,
+      "grad_norm": 730.0363159179688,
+      "learning_rate": 9.99752726113463e-06,
+      "loss": 56.8917,
+      "step": 13490
+    },
+    {
+      "epoch": 0.05454170824630227,
+      "grad_norm": 287.1725158691406,
+      "learning_rate": 9.997482711915926e-06,
+      "loss": 54.6312,
+      "step": 13500
+    },
+    {
+      "epoch": 0.054582109511669905,
+      "grad_norm": 1575.38916015625,
+      "learning_rate": 9.99743776507046e-06,
+      "loss": 54.69,
+      "step": 13510
+    },
+    {
+      "epoch": 0.05462251077703754,
+      "grad_norm": 533.496337890625,
+      "learning_rate": 9.997392420601804e-06,
+      "loss": 42.1878,
+      "step": 13520
+    },
+    {
+      "epoch": 0.05466291204240517,
+      "grad_norm": 590.1005249023438,
+      "learning_rate": 9.99734667851357e-06,
+      "loss": 41.1869,
+      "step": 13530
+    },
+    {
+      "epoch": 0.0547033133077728,
+      "grad_norm": 997.8026733398438,
+      "learning_rate": 9.997300538809394e-06,
+      "loss": 65.3091,
+      "step": 13540
+    },
+    {
+      "epoch": 0.05474371457314043,
+      "grad_norm": 539.83935546875,
+      "learning_rate": 9.99725400149295e-06,
+      "loss": 60.0879,
+      "step": 13550
+    },
+    {
+      "epoch": 0.05478411583850806,
+      "grad_norm": 683.99267578125,
+      "learning_rate": 9.997207066567939e-06,
+      "loss": 54.2406,
+      "step": 13560
+    },
+    {
+      "epoch": 0.054824517103875696,
+      "grad_norm": 767.1815185546875,
+      "learning_rate": 9.997159734038096e-06,
+      "loss": 58.345,
+      "step": 13570
+    },
+    {
+      "epoch": 0.05486491836924332,
+      "grad_norm": 0.0,
+      "learning_rate": 9.997112003907186e-06,
+      "loss": 34.0798,
+      "step": 13580
+    },
+    {
+      "epoch": 0.054905319634610955,
+      "grad_norm": 378.8706970214844,
+      "learning_rate": 9.997063876179007e-06,
+      "loss": 49.5251,
+      "step": 13590
+    },
+    {
+      "epoch": 0.05494572089997859,
+      "grad_norm": 901.79736328125,
+      "learning_rate": 9.997015350857391e-06,
+      "loss": 41.9663,
+      "step": 13600
+    },
+    {
+      "epoch": 0.05498612216534622,
+      "grad_norm": 800.0772705078125,
+      "learning_rate": 9.996966427946195e-06,
+      "loss": 66.8479,
+      "step": 13610
+    },
+    {
+      "epoch": 0.05502652343071385,
+      "grad_norm": 1176.25537109375,
+      "learning_rate": 9.996917107449313e-06,
+      "loss": 39.7329,
+      "step": 13620
+    },
+    {
+      "epoch": 0.05506692469608148,
+      "grad_norm": 1161.61767578125,
+      "learning_rate": 9.99686738937067e-06,
+      "loss": 68.4346,
+      "step": 13630
+    },
+    {
+      "epoch": 0.05510732596144911,
+      "grad_norm": 690.8731689453125,
+      "learning_rate": 9.996817273714222e-06,
+      "loss": 53.2709,
+      "step": 13640
+    },
+    {
+      "epoch": 0.055147727226816746,
+      "grad_norm": 691.3014526367188,
+      "learning_rate": 9.996766760483955e-06,
+      "loss": 75.7811,
+      "step": 13650
+    },
+    {
+      "epoch": 0.05518812849218437,
+      "grad_norm": 556.6007690429688,
+      "learning_rate": 9.996715849683889e-06,
+      "loss": 75.581,
+      "step": 13660
+    },
+    {
+      "epoch": 0.055228529757552006,
+      "grad_norm": 480.1729431152344,
+      "learning_rate": 9.996664541318076e-06,
+      "loss": 51.7071,
+      "step": 13670
+    },
+    {
+      "epoch": 0.05526893102291964,
+      "grad_norm": 533.1707153320312,
+      "learning_rate": 9.996612835390596e-06,
+      "loss": 35.2482,
+      "step": 13680
+    },
+    {
+      "epoch": 0.05530933228828727,
+      "grad_norm": 604.6953125,
+      "learning_rate": 9.996560731905565e-06,
+      "loss": 59.7485,
+      "step": 13690
+    },
+    {
+      "epoch": 0.0553497335536549,
+      "grad_norm": 634.8004760742188,
+      "learning_rate": 9.996508230867126e-06,
+      "loss": 59.5475,
+      "step": 13700
+    },
+    {
+      "epoch": 0.05539013481902253,
+      "grad_norm": 742.4111938476562,
+      "learning_rate": 9.996455332279458e-06,
+      "loss": 49.5485,
+      "step": 13710
+    },
+    {
+      "epoch": 0.055430536084390164,
+      "grad_norm": 513.7229614257812,
+      "learning_rate": 9.99640203614677e-06,
+      "loss": 53.4414,
+      "step": 13720
+    },
+    {
+      "epoch": 0.0554709373497578,
+      "grad_norm": 717.775390625,
+      "learning_rate": 9.996348342473304e-06,
+      "loss": 45.8425,
+      "step": 13730
+    },
+    {
+      "epoch": 0.05551133861512542,
+      "grad_norm": 696.4607543945312,
+      "learning_rate": 9.99629425126333e-06,
+      "loss": 78.0228,
+      "step": 13740
+    },
+    {
+      "epoch": 0.055551739880493056,
+      "grad_norm": 477.80926513671875,
+      "learning_rate": 9.996239762521152e-06,
+      "loss": 61.2231,
+      "step": 13750
+    },
+    {
+      "epoch": 0.05559214114586069,
+      "grad_norm": 970.312744140625,
+      "learning_rate": 9.996184876251105e-06,
+      "loss": 44.8878,
+      "step": 13760
+    },
+    {
+      "epoch": 0.05563254241122832,
+      "grad_norm": 692.9142456054688,
+      "learning_rate": 9.996129592457558e-06,
+      "loss": 42.4326,
+      "step": 13770
+    },
+    {
+      "epoch": 0.05567294367659595,
+      "grad_norm": 437.6714782714844,
+      "learning_rate": 9.996073911144907e-06,
+      "loss": 45.8802,
+      "step": 13780
+    },
+    {
+      "epoch": 0.05571334494196358,
+      "grad_norm": 466.60186767578125,
+      "learning_rate": 9.996017832317583e-06,
+      "loss": 49.2506,
+      "step": 13790
+    },
+    {
+      "epoch": 0.055753746207331215,
+      "grad_norm": 712.9346313476562,
+      "learning_rate": 9.995961355980052e-06,
+      "loss": 77.4249,
+      "step": 13800
+    },
+    {
+      "epoch": 0.05579414747269885,
+      "grad_norm": 711.3295288085938,
+      "learning_rate": 9.995904482136803e-06,
+      "loss": 38.404,
+      "step": 13810
+    },
+    {
+      "epoch": 0.055834548738066474,
+      "grad_norm": 1094.4012451171875,
+      "learning_rate": 9.99584721079236e-06,
+      "loss": 60.306,
+      "step": 13820
+    },
+    {
+      "epoch": 0.05587495000343411,
+      "grad_norm": 259.01220703125,
+      "learning_rate": 9.995789541951287e-06,
+      "loss": 57.5724,
+      "step": 13830
+    },
+    {
+      "epoch": 0.05591535126880174,
+      "grad_norm": 930.34521484375,
+      "learning_rate": 9.995731475618163e-06,
+      "loss": 44.2708,
+      "step": 13840
+    },
+    {
+      "epoch": 0.05595575253416937,
+      "grad_norm": 635.6669311523438,
+      "learning_rate": 9.995673011797615e-06,
+      "loss": 55.8103,
+      "step": 13850
+    },
+    {
+      "epoch": 0.055996153799537,
+      "grad_norm": 1038.2628173828125,
+      "learning_rate": 9.995614150494293e-06,
+      "loss": 56.7732,
+      "step": 13860
+    },
+    {
+      "epoch": 0.05603655506490463,
+      "grad_norm": 646.6795654296875,
+      "learning_rate": 9.995554891712879e-06,
+      "loss": 67.5958,
+      "step": 13870
+    },
+    {
+      "epoch": 0.056076956330272265,
+      "grad_norm": 635.2930908203125,
+      "learning_rate": 9.995495235458087e-06,
+      "loss": 56.4254,
+      "step": 13880
+    },
+    {
+      "epoch": 0.0561173575956399,
+      "grad_norm": 477.6964111328125,
+      "learning_rate": 9.99543518173467e-06,
+      "loss": 33.5029,
+      "step": 13890
+    },
+    {
+      "epoch": 0.056157758861007524,
+      "grad_norm": 750.3272094726562,
+      "learning_rate": 9.995374730547397e-06,
+      "loss": 43.8864,
+      "step": 13900
+    },
+    {
+      "epoch": 0.05619816012637516,
+      "grad_norm": 421.46722412109375,
+      "learning_rate": 9.995313881901085e-06,
+      "loss": 44.306,
+      "step": 13910
+    },
+    {
+      "epoch": 0.05623856139174279,
+      "grad_norm": 1255.6578369140625,
+      "learning_rate": 9.995252635800572e-06,
+      "loss": 69.9456,
+      "step": 13920
+    },
+    {
+      "epoch": 0.05627896265711042,
+      "grad_norm": 935.6283569335938,
+      "learning_rate": 9.995190992250732e-06,
+      "loss": 50.333,
+      "step": 13930
+    },
+    {
+      "epoch": 0.05631936392247805,
+      "grad_norm": 394.6123046875,
+      "learning_rate": 9.995128951256469e-06,
+      "loss": 49.9401,
+      "step": 13940
+    },
+    {
+      "epoch": 0.05635976518784568,
+      "grad_norm": 439.6722412109375,
+      "learning_rate": 9.99506651282272e-06,
+      "loss": 70.1002,
+      "step": 13950
+    },
+    {
+      "epoch": 0.056400166453213316,
+      "grad_norm": 432.78179931640625,
+      "learning_rate": 9.995003676954454e-06,
+      "loss": 50.2052,
+      "step": 13960
+    },
+    {
+      "epoch": 0.05644056771858095,
+      "grad_norm": 581.228271484375,
+      "learning_rate": 9.994940443656668e-06,
+      "loss": 64.6687,
+      "step": 13970
+    },
+    {
+      "epoch": 0.056480968983948575,
+      "grad_norm": 927.3905639648438,
+      "learning_rate": 9.994876812934395e-06,
+      "loss": 47.3174,
+      "step": 13980
+    },
+    {
+      "epoch": 0.05652137024931621,
+      "grad_norm": 336.7020568847656,
+      "learning_rate": 9.994812784792698e-06,
+      "loss": 65.6403,
+      "step": 13990
+    },
+    {
+      "epoch": 0.05656177151468384,
+      "grad_norm": 407.6746520996094,
+      "learning_rate": 9.99474835923667e-06,
+      "loss": 47.5244,
+      "step": 14000
+    },
+    {
+      "epoch": 0.056602172780051474,
+      "grad_norm": 1338.405517578125,
+      "learning_rate": 9.994683536271437e-06,
+      "loss": 67.1354,
+      "step": 14010
+    },
+    {
+      "epoch": 0.0566425740454191,
+      "grad_norm": 967.8305053710938,
+      "learning_rate": 9.994618315902161e-06,
+      "loss": 50.1595,
+      "step": 14020
+    },
+    {
+      "epoch": 0.05668297531078673,
+      "grad_norm": 447.31695556640625,
+      "learning_rate": 9.994552698134023e-06,
+      "loss": 44.7543,
+      "step": 14030
+    },
+    {
+      "epoch": 0.056723376576154366,
+      "grad_norm": 630.2139892578125,
+      "learning_rate": 9.994486682972253e-06,
+      "loss": 60.6165,
+      "step": 14040
+    },
+    {
+      "epoch": 0.056763777841522,
+      "grad_norm": 676.4772338867188,
+      "learning_rate": 9.994420270422096e-06,
+      "loss": 63.8667,
+      "step": 14050
+    },
+    {
+      "epoch": 0.056804179106889625,
+      "grad_norm": 724.5189819335938,
+      "learning_rate": 9.994353460488842e-06,
+      "loss": 61.9814,
+      "step": 14060
+    },
+    {
+      "epoch": 0.05684458037225726,
+      "grad_norm": 885.9684448242188,
+      "learning_rate": 9.994286253177803e-06,
+      "loss": 54.9514,
+      "step": 14070
+    },
+    {
+      "epoch": 0.05688498163762489,
+      "grad_norm": 431.4734191894531,
+      "learning_rate": 9.994218648494327e-06,
+      "loss": 42.4939,
+      "step": 14080
+    },
+    {
+      "epoch": 0.056925382902992525,
+      "grad_norm": 875.316650390625,
+      "learning_rate": 9.994150646443793e-06,
+      "loss": 38.2235,
+      "step": 14090
+    },
+    {
+      "epoch": 0.05696578416836015,
+      "grad_norm": 1274.8563232421875,
+      "learning_rate": 9.994082247031613e-06,
+      "loss": 68.8717,
+      "step": 14100
+    },
+    {
+      "epoch": 0.057006185433727784,
+      "grad_norm": 1349.3104248046875,
+      "learning_rate": 9.99401345026323e-06,
+      "loss": 39.0644,
+      "step": 14110
+    },
+    {
+      "epoch": 0.05704658669909542,
+      "grad_norm": 648.0847778320312,
+      "learning_rate": 9.993944256144115e-06,
+      "loss": 49.0192,
+      "step": 14120
+    },
+    {
+      "epoch": 0.05708698796446305,
+      "grad_norm": 655.9918823242188,
+      "learning_rate": 9.993874664679774e-06,
+      "loss": 53.7475,
+      "step": 14130
+    },
+    {
+      "epoch": 0.057127389229830676,
+      "grad_norm": 857.4654541015625,
+      "learning_rate": 9.993804675875744e-06,
+      "loss": 76.6148,
+      "step": 14140
+    },
+    {
+      "epoch": 0.05716779049519831,
+      "grad_norm": 238.42205810546875,
+      "learning_rate": 9.993734289737596e-06,
+      "loss": 47.7305,
+      "step": 14150
+    },
+    {
+      "epoch": 0.05720819176056594,
+      "grad_norm": 334.1250915527344,
+      "learning_rate": 9.993663506270928e-06,
+      "loss": 51.6321,
+      "step": 14160
+    },
+    {
+      "epoch": 0.057248593025933575,
+      "grad_norm": 1000.7257080078125,
+      "learning_rate": 9.993592325481373e-06,
+      "loss": 49.4137,
+      "step": 14170
+    },
+    {
+      "epoch": 0.0572889942913012,
+      "grad_norm": 605.1682739257812,
+      "learning_rate": 9.993520747374594e-06,
+      "loss": 50.634,
+      "step": 14180
+    },
+    {
+      "epoch": 0.057329395556668834,
+      "grad_norm": 473.75982666015625,
+      "learning_rate": 9.993448771956285e-06,
+      "loss": 49.1249,
+      "step": 14190
+    },
+    {
+      "epoch": 0.05736979682203647,
+      "grad_norm": 443.055419921875,
+      "learning_rate": 9.993376399232175e-06,
+      "loss": 34.1855,
+      "step": 14200
+    },
+    {
+      "epoch": 0.0574101980874041,
+      "grad_norm": 857.4747924804688,
+      "learning_rate": 9.993303629208023e-06,
+      "loss": 50.4225,
+      "step": 14210
+    },
+    {
+      "epoch": 0.057450599352771727,
+      "grad_norm": 372.2861328125,
+      "learning_rate": 9.993230461889616e-06,
+      "loss": 41.4766,
+      "step": 14220
+    },
+    {
+      "epoch": 0.05749100061813936,
+      "grad_norm": 622.736328125,
+      "learning_rate": 9.993156897282776e-06,
+      "loss": 58.5451,
+      "step": 14230
+    },
+    {
+      "epoch": 0.05753140188350699,
+      "grad_norm": 574.2984619140625,
+      "learning_rate": 9.99308293539336e-06,
+      "loss": 53.2113,
+      "step": 14240
+    },
+    {
+      "epoch": 0.057571803148874626,
+      "grad_norm": 706.0330810546875,
+      "learning_rate": 9.993008576227248e-06,
+      "loss": 54.6104,
+      "step": 14250
+    },
+    {
+      "epoch": 0.05761220441424225,
+      "grad_norm": 802.1617431640625,
+      "learning_rate": 9.992933819790358e-06,
+      "loss": 71.0195,
+      "step": 14260
+    },
+    {
+      "epoch": 0.057652605679609885,
+      "grad_norm": 197.7603302001953,
+      "learning_rate": 9.992858666088638e-06,
+      "loss": 45.4237,
+      "step": 14270
+    },
+    {
+      "epoch": 0.05769300694497752,
+      "grad_norm": 407.2517395019531,
+      "learning_rate": 9.992783115128072e-06,
+      "loss": 58.2364,
+      "step": 14280
+    },
+    {
+      "epoch": 0.05773340821034515,
+      "grad_norm": 300.0452575683594,
+      "learning_rate": 9.992707166914662e-06,
+      "loss": 65.9405,
+      "step": 14290
+    },
+    {
+      "epoch": 0.05777380947571278,
+      "grad_norm": 937.140625,
+      "learning_rate": 9.992630821454458e-06,
+      "loss": 48.6805,
+      "step": 14300
+    },
+    {
+      "epoch": 0.05781421074108041,
+      "grad_norm": 1485.9613037109375,
+      "learning_rate": 9.992554078753534e-06,
+      "loss": 58.1787,
+      "step": 14310
+    },
+    {
+      "epoch": 0.05785461200644804,
+      "grad_norm": 659.9443359375,
+      "learning_rate": 9.992476938817994e-06,
+      "loss": 43.7871,
+      "step": 14320
+    },
+    {
+      "epoch": 0.057895013271815676,
+      "grad_norm": 609.6917114257812,
+      "learning_rate": 9.992399401653976e-06,
+      "loss": 37.274,
+      "step": 14330
+    },
+    {
+      "epoch": 0.0579354145371833,
+      "grad_norm": 126.08211517333984,
+      "learning_rate": 9.99232146726765e-06,
+      "loss": 57.0369,
+      "step": 14340
+    },
+    {
+      "epoch": 0.057975815802550935,
+      "grad_norm": 256.6801452636719,
+      "learning_rate": 9.992243135665217e-06,
+      "loss": 59.119,
+      "step": 14350
+    },
+    {
+      "epoch": 0.05801621706791857,
+      "grad_norm": 1137.1373291015625,
+      "learning_rate": 9.992164406852908e-06,
+      "loss": 80.0415,
+      "step": 14360
+    },
+    {
+      "epoch": 0.0580566183332862,
+      "grad_norm": 1009.8486938476562,
+      "learning_rate": 9.992085280836988e-06,
+      "loss": 50.1561,
+      "step": 14370
+    },
+    {
+      "epoch": 0.05809701959865383,
+      "grad_norm": 829.55712890625,
+      "learning_rate": 9.992005757623753e-06,
+      "loss": 54.28,
+      "step": 14380
+    },
+    {
+      "epoch": 0.05813742086402146,
+      "grad_norm": 812.1973876953125,
+      "learning_rate": 9.991925837219532e-06,
+      "loss": 58.4078,
+      "step": 14390
+    },
+    {
+      "epoch": 0.058177822129389094,
+      "grad_norm": 364.135498046875,
+      "learning_rate": 9.991845519630679e-06,
+      "loss": 40.6019,
+      "step": 14400
+    },
+    {
+      "epoch": 0.05821822339475673,
+      "grad_norm": 855.4879150390625,
+      "learning_rate": 9.991764804863588e-06,
+      "loss": 76.5647,
+      "step": 14410
+    },
+    {
+      "epoch": 0.05825862466012435,
+      "grad_norm": 405.6043395996094,
+      "learning_rate": 9.991683692924682e-06,
+      "loss": 47.4223,
+      "step": 14420
+    },
+    {
+      "epoch": 0.058299025925491986,
+      "grad_norm": 449.4226989746094,
+      "learning_rate": 9.991602183820412e-06,
+      "loss": 40.5298,
+      "step": 14430
+    },
+    {
+      "epoch": 0.05833942719085962,
+      "grad_norm": 823.6030883789062,
+      "learning_rate": 9.991520277557266e-06,
+      "loss": 42.4834,
+      "step": 14440
+    },
+    {
+      "epoch": 0.05837982845622725,
+      "grad_norm": 757.8836669921875,
+      "learning_rate": 9.991437974141759e-06,
+      "loss": 47.1578,
+      "step": 14450
+    },
+    {
+      "epoch": 0.05842022972159488,
+      "grad_norm": 873.563720703125,
+      "learning_rate": 9.99135527358044e-06,
+      "loss": 52.538,
+      "step": 14460
+    },
+    {
+      "epoch": 0.05846063098696251,
+      "grad_norm": 670.3134765625,
+      "learning_rate": 9.991272175879888e-06,
+      "loss": 39.9001,
+      "step": 14470
+    },
+    {
+      "epoch": 0.058501032252330144,
+      "grad_norm": 329.8004455566406,
+      "learning_rate": 9.991188681046718e-06,
+      "loss": 82.909,
+      "step": 14480
+    },
+    {
+      "epoch": 0.05854143351769778,
+      "grad_norm": 725.716064453125,
+      "learning_rate": 9.991104789087568e-06,
+      "loss": 66.8185,
+      "step": 14490
+    },
+    {
+      "epoch": 0.058581834783065403,
+      "grad_norm": 359.1539611816406,
+      "learning_rate": 9.991020500009118e-06,
+      "loss": 54.1102,
+      "step": 14500
+    },
+    {
+      "epoch": 0.05862223604843304,
+      "grad_norm": 618.91650390625,
+      "learning_rate": 9.990935813818073e-06,
+      "loss": 39.747,
+      "step": 14510
+    },
+    {
+      "epoch": 0.05866263731380067,
+      "grad_norm": 668.1556396484375,
+      "learning_rate": 9.99085073052117e-06,
+      "loss": 48.4924,
+      "step": 14520
+    },
+    {
+      "epoch": 0.0587030385791683,
+      "grad_norm": 480.24945068359375,
+      "learning_rate": 9.990765250125179e-06,
+      "loss": 50.9185,
+      "step": 14530
+    },
+    {
+      "epoch": 0.05874343984453593,
+      "grad_norm": 811.7920532226562,
+      "learning_rate": 9.990679372636902e-06,
+      "loss": 53.3772,
+      "step": 14540
+    },
+    {
+      "epoch": 0.05878384110990356,
+      "grad_norm": 577.9950561523438,
+      "learning_rate": 9.99059309806317e-06,
+      "loss": 51.9823,
+      "step": 14550
+    },
+    {
+      "epoch": 0.058824242375271195,
+      "grad_norm": 699.646728515625,
+      "learning_rate": 9.990506426410851e-06,
+      "loss": 56.4946,
+      "step": 14560
+    },
+    {
+      "epoch": 0.05886464364063883,
+      "grad_norm": 371.0194091796875,
+      "learning_rate": 9.990419357686839e-06,
+      "loss": 41.3928,
+      "step": 14570
+    },
+    {
+      "epoch": 0.058905044906006454,
+      "grad_norm": 644.16796875,
+      "learning_rate": 9.99033189189806e-06,
+      "loss": 44.1936,
+      "step": 14580
+    },
+    {
+      "epoch": 0.05894544617137409,
+      "grad_norm": 817.6756591796875,
+      "learning_rate": 9.990244029051475e-06,
+      "loss": 52.6435,
+      "step": 14590
+    },
+    {
+      "epoch": 0.05898584743674172,
+      "grad_norm": 966.6154174804688,
+      "learning_rate": 9.990155769154077e-06,
+      "loss": 65.1172,
+      "step": 14600
+    },
+    {
+      "epoch": 0.05902624870210935,
+      "grad_norm": 668.8129272460938,
+      "learning_rate": 9.990067112212884e-06,
+      "loss": 47.6556,
+      "step": 14610
+    },
+    {
+      "epoch": 0.05906664996747698,
+      "grad_norm": 334.03900146484375,
+      "learning_rate": 9.989978058234952e-06,
+      "loss": 63.2436,
+      "step": 14620
+    },
+    {
+      "epoch": 0.05910705123284461,
+      "grad_norm": 658.3740844726562,
+      "learning_rate": 9.989888607227369e-06,
+      "loss": 40.0384,
+      "step": 14630
+    },
+    {
+      "epoch": 0.059147452498212245,
+      "grad_norm": 637.34716796875,
+      "learning_rate": 9.989798759197247e-06,
+      "loss": 51.4579,
+      "step": 14640
+    },
+    {
+      "epoch": 0.05918785376357988,
+      "grad_norm": 511.2720642089844,
+      "learning_rate": 9.989708514151739e-06,
+      "loss": 52.719,
+      "step": 14650
+    },
+    {
+      "epoch": 0.059228255028947505,
+      "grad_norm": 412.48040771484375,
+      "learning_rate": 9.989617872098026e-06,
+      "loss": 38.988,
+      "step": 14660
+    },
+    {
+      "epoch": 0.05926865629431514,
+      "grad_norm": 1462.632080078125,
+      "learning_rate": 9.989526833043316e-06,
+      "loss": 56.2337,
+      "step": 14670
+    },
+    {
+      "epoch": 0.05930905755968277,
+      "grad_norm": 746.0077514648438,
+      "learning_rate": 9.989435396994856e-06,
+      "loss": 50.376,
+      "step": 14680
+    },
+    {
+      "epoch": 0.059349458825050404,
+      "grad_norm": 894.7488403320312,
+      "learning_rate": 9.989343563959919e-06,
+      "loss": 51.9415,
+      "step": 14690
+    },
+    {
+      "epoch": 0.05938986009041803,
+      "grad_norm": 378.21697998046875,
+      "learning_rate": 9.989251333945813e-06,
+      "loss": 63.1732,
+      "step": 14700
+    },
+    {
+      "epoch": 0.05943026135578566,
+      "grad_norm": 433.5731201171875,
+      "learning_rate": 9.989158706959875e-06,
+      "loss": 57.7156,
+      "step": 14710
+    },
+    {
+      "epoch": 0.059470662621153296,
+      "grad_norm": 931.3536987304688,
+      "learning_rate": 9.989065683009477e-06,
+      "loss": 70.65,
+      "step": 14720
+    },
+    {
+      "epoch": 0.05951106388652093,
+      "grad_norm": 1730.477783203125,
+      "learning_rate": 9.988972262102018e-06,
+      "loss": 58.9854,
+      "step": 14730
+    },
+    {
+      "epoch": 0.059551465151888555,
+      "grad_norm": 1209.090576171875,
+      "learning_rate": 9.988878444244937e-06,
+      "loss": 47.5035,
+      "step": 14740
+    },
+    {
+      "epoch": 0.05959186641725619,
+      "grad_norm": 221.4032440185547,
+      "learning_rate": 9.988784229445689e-06,
+      "loss": 45.2581,
+      "step": 14750
+    },
+    {
+      "epoch": 0.05963226768262382,
+      "grad_norm": 404.3892822265625,
+      "learning_rate": 9.988689617711777e-06,
+      "loss": 55.4668,
+      "step": 14760
+    },
+    {
+      "epoch": 0.059672668947991454,
+      "grad_norm": 582.4869384765625,
+      "learning_rate": 9.988594609050726e-06,
+      "loss": 56.3967,
+      "step": 14770
+    },
+    {
+      "epoch": 0.05971307021335908,
+      "grad_norm": 248.2957305908203,
+      "learning_rate": 9.988499203470097e-06,
+      "loss": 44.0815,
+      "step": 14780
+    },
+    {
+      "epoch": 0.059753471478726713,
+      "grad_norm": 945.0496826171875,
+      "learning_rate": 9.988403400977482e-06,
+      "loss": 54.5601,
+      "step": 14790
+    },
+    {
+      "epoch": 0.05979387274409435,
+      "grad_norm": 446.46405029296875,
+      "learning_rate": 9.9883072015805e-06,
+      "loss": 49.9845,
+      "step": 14800
+    },
+    {
+      "epoch": 0.05983427400946198,
+      "grad_norm": 347.37139892578125,
+      "learning_rate": 9.98821060528681e-06,
+      "loss": 65.609,
+      "step": 14810
+    },
+    {
+      "epoch": 0.059874675274829606,
+      "grad_norm": 461.4508972167969,
+      "learning_rate": 9.988113612104093e-06,
+      "loss": 52.9646,
+      "step": 14820
+    },
+    {
+      "epoch": 0.05991507654019724,
+      "grad_norm": 736.7156982421875,
+      "learning_rate": 9.988016222040067e-06,
+      "loss": 42.4082,
+      "step": 14830
+    },
+    {
+      "epoch": 0.05995547780556487,
+      "grad_norm": 729.3582763671875,
+      "learning_rate": 9.987918435102484e-06,
+      "loss": 44.0376,
+      "step": 14840
+    },
+    {
+      "epoch": 0.059995879070932505,
+      "grad_norm": 404.2114562988281,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 77.3696,
+      "step": 14850
+    },
+    {
+      "epoch": 0.06003628033630013,
+      "grad_norm": 954.4871215820312,
+      "learning_rate": 9.987721670637794e-06,
+      "loss": 48.0571,
+      "step": 14860
+    },
+    {
+      "epoch": 0.060076681601667764,
+      "grad_norm": 511.55133056640625,
+      "learning_rate": 9.987622693126342e-06,
+      "loss": 55.6837,
+      "step": 14870
+    },
+    {
+      "epoch": 0.0601170828670354,
+      "grad_norm": 572.0296020507812,
+      "learning_rate": 9.987523318772644e-06,
+      "loss": 51.224,
+      "step": 14880
+    },
+    {
+      "epoch": 0.06015748413240303,
+      "grad_norm": 771.312255859375,
+      "learning_rate": 9.987423547584605e-06,
+      "loss": 35.857,
+      "step": 14890
+    },
+    {
+      "epoch": 0.060197885397770656,
+      "grad_norm": 634.609375,
+      "learning_rate": 9.987323379570161e-06,
+      "loss": 63.9532,
+      "step": 14900
+    },
+    {
+      "epoch": 0.06023828666313829,
+      "grad_norm": 719.4342651367188,
+      "learning_rate": 9.987222814737287e-06,
+      "loss": 51.5199,
+      "step": 14910
+    },
+    {
+      "epoch": 0.06027868792850592,
+      "grad_norm": 544.3135375976562,
+      "learning_rate": 9.987121853093982e-06,
+      "loss": 58.8402,
+      "step": 14920
+    },
+    {
+      "epoch": 0.060319089193873555,
+      "grad_norm": 1171.758056640625,
+      "learning_rate": 9.987020494648279e-06,
+      "loss": 54.2668,
+      "step": 14930
+    },
+    {
+      "epoch": 0.06035949045924118,
+      "grad_norm": 688.6409912109375,
+      "learning_rate": 9.986918739408241e-06,
+      "loss": 52.9917,
+      "step": 14940
+    },
+    {
+      "epoch": 0.060399891724608815,
+      "grad_norm": 450.6348876953125,
+      "learning_rate": 9.986816587381966e-06,
+      "loss": 53.4728,
+      "step": 14950
+    },
+    {
+      "epoch": 0.06044029298997645,
+      "grad_norm": 403.31341552734375,
+      "learning_rate": 9.986714038577582e-06,
+      "loss": 36.6673,
+      "step": 14960
+    },
+    {
+      "epoch": 0.06048069425534408,
+      "grad_norm": 844.4390258789062,
+      "learning_rate": 9.986611093003249e-06,
+      "loss": 52.0443,
+      "step": 14970
+    },
+    {
+      "epoch": 0.06052109552071171,
+      "grad_norm": 254.19454956054688,
+      "learning_rate": 9.986507750667157e-06,
+      "loss": 56.2579,
+      "step": 14980
+    },
+    {
+      "epoch": 0.06056149678607934,
+      "grad_norm": 566.4449462890625,
+      "learning_rate": 9.986404011577525e-06,
+      "loss": 39.8158,
+      "step": 14990
+    },
+    {
+      "epoch": 0.06060189805144697,
+      "grad_norm": 648.5628662109375,
+      "learning_rate": 9.986299875742612e-06,
+      "loss": 53.3486,
+      "step": 15000
+    },
+    {
+      "epoch": 0.060642299316814606,
+      "grad_norm": 856.6651000976562,
+      "learning_rate": 9.986195343170703e-06,
+      "loss": 53.0422,
+      "step": 15010
+    },
+    {
+      "epoch": 0.06068270058218223,
+      "grad_norm": 770.17431640625,
+      "learning_rate": 9.986090413870114e-06,
+      "loss": 47.2132,
+      "step": 15020
+    },
+    {
+      "epoch": 0.060723101847549865,
+      "grad_norm": 400.9591979980469,
+      "learning_rate": 9.985985087849193e-06,
+      "loss": 67.7665,
+      "step": 15030
+    },
+    {
+      "epoch": 0.0607635031129175,
+      "grad_norm": 694.5767211914062,
+      "learning_rate": 9.98587936511632e-06,
+      "loss": 43.7988,
+      "step": 15040
+    },
+    {
+      "epoch": 0.06080390437828513,
+      "grad_norm": 466.62896728515625,
+      "learning_rate": 9.98577324567991e-06,
+      "loss": 50.6092,
+      "step": 15050
+    },
+    {
+      "epoch": 0.06084430564365276,
+      "grad_norm": 737.1053466796875,
+      "learning_rate": 9.985666729548404e-06,
+      "loss": 47.2564,
+      "step": 15060
+    },
+    {
+      "epoch": 0.06088470690902039,
+      "grad_norm": 696.4453125,
+      "learning_rate": 9.985559816730277e-06,
+      "loss": 54.4192,
+      "step": 15070
+    },
+    {
+      "epoch": 0.060925108174388024,
+      "grad_norm": 1044.3509521484375,
+      "learning_rate": 9.985452507234037e-06,
+      "loss": 60.5334,
+      "step": 15080
+    },
+    {
+      "epoch": 0.06096550943975566,
+      "grad_norm": 1355.4449462890625,
+      "learning_rate": 9.98534480106822e-06,
+      "loss": 57.1139,
+      "step": 15090
+    },
+    {
+      "epoch": 0.06100591070512328,
+      "grad_norm": 586.2362060546875,
+      "learning_rate": 9.985236698241396e-06,
+      "loss": 57.273,
+      "step": 15100
+    },
+    {
+      "epoch": 0.061046311970490916,
+      "grad_norm": 781.879150390625,
+      "learning_rate": 9.985128198762168e-06,
+      "loss": 40.2266,
+      "step": 15110
+    },
+    {
+      "epoch": 0.06108671323585855,
+      "grad_norm": 831.5631713867188,
+      "learning_rate": 9.98501930263917e-06,
+      "loss": 61.6051,
+      "step": 15120
+    },
+    {
+      "epoch": 0.06112711450122618,
+      "grad_norm": 538.1463012695312,
+      "learning_rate": 9.984910009881062e-06,
+      "loss": 34.2463,
+      "step": 15130
+    },
+    {
+      "epoch": 0.06116751576659381,
+      "grad_norm": 1016.048828125,
+      "learning_rate": 9.984800320496542e-06,
+      "loss": 56.4194,
+      "step": 15140
+    },
+    {
+      "epoch": 0.06120791703196144,
+      "grad_norm": 537.7840576171875,
+      "learning_rate": 9.984690234494338e-06,
+      "loss": 70.5479,
+      "step": 15150
+    },
+    {
+      "epoch": 0.061248318297329074,
+      "grad_norm": 625.3380737304688,
+      "learning_rate": 9.98457975188321e-06,
+      "loss": 77.5726,
+      "step": 15160
+    },
+    {
+      "epoch": 0.06128871956269671,
+      "grad_norm": 1212.03857421875,
+      "learning_rate": 9.984468872671945e-06,
+      "loss": 49.5992,
+      "step": 15170
+    },
+    {
+      "epoch": 0.06132912082806433,
+      "grad_norm": 1049.4886474609375,
+      "learning_rate": 9.984357596869369e-06,
+      "loss": 62.8757,
+      "step": 15180
+    },
+    {
+      "epoch": 0.061369522093431966,
+      "grad_norm": 1094.6876220703125,
+      "learning_rate": 9.984245924484334e-06,
+      "loss": 56.43,
+      "step": 15190
+    },
+    {
+      "epoch": 0.0614099233587996,
+      "grad_norm": 503.16534423828125,
+      "learning_rate": 9.984133855525723e-06,
+      "loss": 55.6275,
+      "step": 15200
+    },
+    {
+      "epoch": 0.061450324624167225,
+      "grad_norm": 431.4841613769531,
+      "learning_rate": 9.984021390002458e-06,
+      "loss": 64.4734,
+      "step": 15210
+    },
+    {
+      "epoch": 0.06149072588953486,
+      "grad_norm": 968.3142700195312,
+      "learning_rate": 9.983908527923486e-06,
+      "loss": 62.6523,
+      "step": 15220
+    },
+    {
+      "epoch": 0.06153112715490249,
+      "grad_norm": 652.2515869140625,
+      "learning_rate": 9.983795269297782e-06,
+      "loss": 30.2004,
+      "step": 15230
+    },
+    {
+      "epoch": 0.061571528420270125,
+      "grad_norm": 660.2927856445312,
+      "learning_rate": 9.983681614134363e-06,
+      "loss": 43.4263,
+      "step": 15240
+    },
+    {
+      "epoch": 0.06161192968563775,
+      "grad_norm": 536.5686645507812,
+      "learning_rate": 9.98356756244227e-06,
+      "loss": 55.396,
+      "step": 15250
+    },
+    {
+      "epoch": 0.061652330951005384,
+      "grad_norm": 542.5938720703125,
+      "learning_rate": 9.983453114230575e-06,
+      "loss": 41.3936,
+      "step": 15260
+    },
+    {
+      "epoch": 0.06169273221637302,
+      "grad_norm": 971.2131958007812,
+      "learning_rate": 9.98333826950839e-06,
+      "loss": 66.6024,
+      "step": 15270
+    },
+    {
+      "epoch": 0.06173313348174065,
+      "grad_norm": 422.6085510253906,
+      "learning_rate": 9.983223028284847e-06,
+      "loss": 37.8949,
+      "step": 15280
+    },
+    {
+      "epoch": 0.061773534747108276,
+      "grad_norm": 339.5025634765625,
+      "learning_rate": 9.983107390569118e-06,
+      "loss": 38.7237,
+      "step": 15290
+    },
+    {
+      "epoch": 0.06181393601247591,
+      "grad_norm": 916.392333984375,
+      "learning_rate": 9.982991356370404e-06,
+      "loss": 39.1873,
+      "step": 15300
+    },
+    {
+      "epoch": 0.06185433727784354,
+      "grad_norm": 678.418212890625,
+      "learning_rate": 9.982874925697937e-06,
+      "loss": 64.8346,
+      "step": 15310
+    },
+    {
+      "epoch": 0.061894738543211175,
+      "grad_norm": 854.1286010742188,
+      "learning_rate": 9.982758098560978e-06,
+      "loss": 52.7229,
+      "step": 15320
+    },
+    {
+      "epoch": 0.0619351398085788,
+      "grad_norm": 747.940673828125,
+      "learning_rate": 9.982640874968827e-06,
+      "loss": 40.8252,
+      "step": 15330
+    },
+    {
+      "epoch": 0.061975541073946434,
+      "grad_norm": 982.86181640625,
+      "learning_rate": 9.98252325493081e-06,
+      "loss": 65.4864,
+      "step": 15340
+    },
+    {
+      "epoch": 0.06201594233931407,
+      "grad_norm": 356.97271728515625,
+      "learning_rate": 9.982405238456281e-06,
+      "loss": 52.2651,
+      "step": 15350
+    },
+    {
+      "epoch": 0.0620563436046817,
+      "grad_norm": 685.7958374023438,
+      "learning_rate": 9.982286825554636e-06,
+      "loss": 44.4038,
+      "step": 15360
+    },
+    {
+      "epoch": 0.06209674487004933,
+      "grad_norm": 834.1498413085938,
+      "learning_rate": 9.982168016235292e-06,
+      "loss": 44.5143,
+      "step": 15370
+    },
+    {
+      "epoch": 0.06213714613541696,
+      "grad_norm": 433.9917907714844,
+      "learning_rate": 9.982048810507706e-06,
+      "loss": 44.4605,
+      "step": 15380
+    },
+    {
+      "epoch": 0.06217754740078459,
+      "grad_norm": 908.0674438476562,
+      "learning_rate": 9.98192920838136e-06,
+      "loss": 43.4592,
+      "step": 15390
+    },
+    {
+      "epoch": 0.062217948666152226,
+      "grad_norm": 1588.20703125,
+      "learning_rate": 9.98180920986577e-06,
+      "loss": 56.2675,
+      "step": 15400
+    },
+    {
+      "epoch": 0.06225834993151985,
+      "grad_norm": 886.8815307617188,
+      "learning_rate": 9.981688814970485e-06,
+      "loss": 55.398,
+      "step": 15410
+    },
+    {
+      "epoch": 0.062298751196887485,
+      "grad_norm": 1576.02685546875,
+      "learning_rate": 9.981568023705085e-06,
+      "loss": 50.5813,
+      "step": 15420
+    },
+    {
+      "epoch": 0.06233915246225512,
+      "grad_norm": 594.6544799804688,
+      "learning_rate": 9.981446836079178e-06,
+      "loss": 49.5407,
+      "step": 15430
+    },
+    {
+      "epoch": 0.06237955372762275,
+      "grad_norm": 942.7835083007812,
+      "learning_rate": 9.981325252102408e-06,
+      "loss": 75.0559,
+      "step": 15440
+    },
+    {
+      "epoch": 0.06241995499299038,
+      "grad_norm": 762.1746826171875,
+      "learning_rate": 9.98120327178445e-06,
+      "loss": 42.7199,
+      "step": 15450
+    },
+    {
+      "epoch": 0.06246035625835801,
+      "grad_norm": 575.0107421875,
+      "learning_rate": 9.981080895135007e-06,
+      "loss": 40.931,
+      "step": 15460
+    },
+    {
+      "epoch": 0.06250075752372564,
+      "grad_norm": 765.7384033203125,
+      "learning_rate": 9.980958122163818e-06,
+      "loss": 47.4123,
+      "step": 15470
+    },
+    {
+      "epoch": 0.06254115878909328,
+      "grad_norm": 1211.600341796875,
+      "learning_rate": 9.980834952880652e-06,
+      "loss": 51.5481,
+      "step": 15480
+    },
+    {
+      "epoch": 0.0625815600544609,
+      "grad_norm": 678.6597290039062,
+      "learning_rate": 9.980711387295306e-06,
+      "loss": 79.6206,
+      "step": 15490
+    },
+    {
+      "epoch": 0.06262196131982854,
+      "grad_norm": 808.155517578125,
+      "learning_rate": 9.980587425417612e-06,
+      "loss": 75.9096,
+      "step": 15500
+    },
+    {
+      "epoch": 0.06266236258519617,
+      "grad_norm": 1321.018310546875,
+      "learning_rate": 9.980463067257437e-06,
+      "loss": 47.1829,
+      "step": 15510
+    },
+    {
+      "epoch": 0.0627027638505638,
+      "grad_norm": 705.9486083984375,
+      "learning_rate": 9.980338312824672e-06,
+      "loss": 38.1095,
+      "step": 15520
+    },
+    {
+      "epoch": 0.06274316511593143,
+      "grad_norm": 801.2616577148438,
+      "learning_rate": 9.980213162129244e-06,
+      "loss": 53.8346,
+      "step": 15530
+    },
+    {
+      "epoch": 0.06278356638129906,
+      "grad_norm": 499.783935546875,
+      "learning_rate": 9.980087615181111e-06,
+      "loss": 50.7611,
+      "step": 15540
+    },
+    {
+      "epoch": 0.06282396764666669,
+      "grad_norm": 978.2482299804688,
+      "learning_rate": 9.979961671990263e-06,
+      "loss": 54.354,
+      "step": 15550
+    },
+    {
+      "epoch": 0.06286436891203433,
+      "grad_norm": 611.9962768554688,
+      "learning_rate": 9.979835332566719e-06,
+      "loss": 49.5433,
+      "step": 15560
+    },
+    {
+      "epoch": 0.06290477017740195,
+      "grad_norm": 2557.07958984375,
+      "learning_rate": 9.97970859692053e-06,
+      "loss": 69.103,
+      "step": 15570
+    },
+    {
+      "epoch": 0.06294517144276959,
+      "grad_norm": 484.30908203125,
+      "learning_rate": 9.979581465061784e-06,
+      "loss": 60.9328,
+      "step": 15580
+    },
+    {
+      "epoch": 0.06298557270813722,
+      "grad_norm": 1626.028564453125,
+      "learning_rate": 9.979453937000594e-06,
+      "loss": 77.7794,
+      "step": 15590
+    },
+    {
+      "epoch": 0.06302597397350485,
+      "grad_norm": 600.4717407226562,
+      "learning_rate": 9.979326012747106e-06,
+      "loss": 62.6274,
+      "step": 15600
+    },
+    {
+      "epoch": 0.06306637523887249,
+      "grad_norm": 863.0751953125,
+      "learning_rate": 9.9791976923115e-06,
+      "loss": 44.0777,
+      "step": 15610
+    },
+    {
+      "epoch": 0.06310677650424011,
+      "grad_norm": 1112.2489013671875,
+      "learning_rate": 9.979068975703984e-06,
+      "loss": 54.6989,
+      "step": 15620
+    },
+    {
+      "epoch": 0.06314717776960774,
+      "grad_norm": 579.8980712890625,
+      "learning_rate": 9.978939862934802e-06,
+      "loss": 51.3433,
+      "step": 15630
+    },
+    {
+      "epoch": 0.06318757903497538,
+      "grad_norm": 577.5842895507812,
+      "learning_rate": 9.978810354014223e-06,
+      "loss": 52.9068,
+      "step": 15640
+    },
+    {
+      "epoch": 0.063227980300343,
+      "grad_norm": 1737.6309814453125,
+      "learning_rate": 9.978680448952556e-06,
+      "loss": 60.6608,
+      "step": 15650
+    },
+    {
+      "epoch": 0.06326838156571064,
+      "grad_norm": 931.930908203125,
+      "learning_rate": 9.978550147760133e-06,
+      "loss": 65.9854,
+      "step": 15660
+    },
+    {
+      "epoch": 0.06330878283107827,
+      "grad_norm": 847.0628662109375,
+      "learning_rate": 9.978419450447325e-06,
+      "loss": 47.0742,
+      "step": 15670
+    },
+    {
+      "epoch": 0.0633491840964459,
+      "grad_norm": 538.1990356445312,
+      "learning_rate": 9.978288357024527e-06,
+      "loss": 69.5365,
+      "step": 15680
+    },
+    {
+      "epoch": 0.06338958536181354,
+      "grad_norm": 920.2020263671875,
+      "learning_rate": 9.978156867502173e-06,
+      "loss": 49.8007,
+      "step": 15690
+    },
+    {
+      "epoch": 0.06342998662718116,
+      "grad_norm": 922.1722412109375,
+      "learning_rate": 9.978024981890724e-06,
+      "loss": 45.6619,
+      "step": 15700
+    },
+    {
+      "epoch": 0.06347038789254879,
+      "grad_norm": 262.210205078125,
+      "learning_rate": 9.977892700200673e-06,
+      "loss": 47.8528,
+      "step": 15710
+    },
+    {
+      "epoch": 0.06351078915791643,
+      "grad_norm": 781.1572875976562,
+      "learning_rate": 9.977760022442545e-06,
+      "loss": 66.2854,
+      "step": 15720
+    },
+    {
+      "epoch": 0.06355119042328405,
+      "grad_norm": 808.654052734375,
+      "learning_rate": 9.977626948626897e-06,
+      "loss": 46.0415,
+      "step": 15730
+    },
+    {
+      "epoch": 0.0635915916886517,
+      "grad_norm": 924.575927734375,
+      "learning_rate": 9.977493478764316e-06,
+      "loss": 63.1324,
+      "step": 15740
+    },
+    {
+      "epoch": 0.06363199295401932,
+      "grad_norm": 1043.2413330078125,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 60.2796,
+      "step": 15750
+    },
+    {
+      "epoch": 0.06367239421938695,
+      "grad_norm": 331.38427734375,
+      "learning_rate": 9.97722535094087e-06,
+      "loss": 50.0805,
+      "step": 15760
+    },
+    {
+      "epoch": 0.06371279548475459,
+      "grad_norm": 742.5581665039062,
+      "learning_rate": 9.977090693001336e-06,
+      "loss": 50.485,
+      "step": 15770
+    },
+    {
+      "epoch": 0.06375319675012221,
+      "grad_norm": 692.9002075195312,
+      "learning_rate": 9.976955639057539e-06,
+      "loss": 48.5586,
+      "step": 15780
+    },
+    {
+      "epoch": 0.06379359801548984,
+      "grad_norm": 700.7842407226562,
+      "learning_rate": 9.976820189120223e-06,
+      "loss": 44.3135,
+      "step": 15790
+    },
+    {
+      "epoch": 0.06383399928085748,
+      "grad_norm": 693.6561889648438,
+      "learning_rate": 9.976684343200164e-06,
+      "loss": 65.8412,
+      "step": 15800
+    },
+    {
+      "epoch": 0.0638744005462251,
+      "grad_norm": 785.6405029296875,
+      "learning_rate": 9.976548101308173e-06,
+      "loss": 48.4629,
+      "step": 15810
+    },
+    {
+      "epoch": 0.06391480181159274,
+      "grad_norm": 640.048828125,
+      "learning_rate": 9.976411463455088e-06,
+      "loss": 37.2934,
+      "step": 15820
+    },
+    {
+      "epoch": 0.06395520307696037,
+      "grad_norm": 874.2314453125,
+      "learning_rate": 9.976274429651783e-06,
+      "loss": 41.366,
+      "step": 15830
+    },
+    {
+      "epoch": 0.063995604342328,
+      "grad_norm": 1200.9598388671875,
+      "learning_rate": 9.976136999909156e-06,
+      "loss": 34.9769,
+      "step": 15840
+    },
+    {
+      "epoch": 0.06403600560769564,
+      "grad_norm": 639.4553833007812,
+      "learning_rate": 9.97599917423815e-06,
+      "loss": 47.3031,
+      "step": 15850
+    },
+    {
+      "epoch": 0.06407640687306326,
+      "grad_norm": 745.8543090820312,
+      "learning_rate": 9.975860952649724e-06,
+      "loss": 65.3151,
+      "step": 15860
+    },
+    {
+      "epoch": 0.06411680813843089,
+      "grad_norm": 644.9210815429688,
+      "learning_rate": 9.975722335154876e-06,
+      "loss": 54.8319,
+      "step": 15870
+    },
+    {
+      "epoch": 0.06415720940379853,
+      "grad_norm": 836.5652465820312,
+      "learning_rate": 9.975583321764638e-06,
+      "loss": 39.6489,
+      "step": 15880
+    },
+    {
+      "epoch": 0.06419761066916616,
+      "grad_norm": 776.029052734375,
+      "learning_rate": 9.975443912490073e-06,
+      "loss": 83.7785,
+      "step": 15890
+    },
+    {
+      "epoch": 0.0642380119345338,
+      "grad_norm": 620.9443359375,
+      "learning_rate": 9.975304107342268e-06,
+      "loss": 49.9498,
+      "step": 15900
+    },
+    {
+      "epoch": 0.06427841319990142,
+      "grad_norm": 780.52099609375,
+      "learning_rate": 9.97516390633235e-06,
+      "loss": 45.0322,
+      "step": 15910
+    },
+    {
+      "epoch": 0.06431881446526905,
+      "grad_norm": 570.9188232421875,
+      "learning_rate": 9.975023309471473e-06,
+      "loss": 35.3322,
+      "step": 15920
+    },
+    {
+      "epoch": 0.06435921573063669,
+      "grad_norm": 650.2386474609375,
+      "learning_rate": 9.974882316770823e-06,
+      "loss": 60.9373,
+      "step": 15930
+    },
+    {
+      "epoch": 0.06439961699600431,
+      "grad_norm": 369.92730712890625,
+      "learning_rate": 9.974740928241617e-06,
+      "loss": 41.9204,
+      "step": 15940
+    },
+    {
+      "epoch": 0.06444001826137194,
+      "grad_norm": 427.5985412597656,
+      "learning_rate": 9.974599143895107e-06,
+      "loss": 47.1646,
+      "step": 15950
+    },
+    {
+      "epoch": 0.06448041952673958,
+      "grad_norm": 552.6906127929688,
+      "learning_rate": 9.974456963742573e-06,
+      "loss": 50.5909,
+      "step": 15960
+    },
+    {
+      "epoch": 0.0645208207921072,
+      "grad_norm": 895.8667602539062,
+      "learning_rate": 9.97431438779533e-06,
+      "loss": 59.4875,
+      "step": 15970
+    },
+    {
+      "epoch": 0.06456122205747485,
+      "grad_norm": 756.911376953125,
+      "learning_rate": 9.974171416064719e-06,
+      "loss": 45.0479,
+      "step": 15980
+    },
+    {
+      "epoch": 0.06460162332284247,
+      "grad_norm": 492.8658142089844,
+      "learning_rate": 9.974028048562118e-06,
+      "loss": 59.8826,
+      "step": 15990
+    },
+    {
+      "epoch": 0.0646420245882101,
+      "grad_norm": 1029.951904296875,
+      "learning_rate": 9.973884285298932e-06,
+      "loss": 62.1363,
+      "step": 16000
+    },
+    {
+      "epoch": 0.06468242585357774,
+      "grad_norm": 1486.369873046875,
+      "learning_rate": 9.9737401262866e-06,
+      "loss": 58.4914,
+      "step": 16010
+    },
+    {
+      "epoch": 0.06472282711894536,
+      "grad_norm": 532.9174194335938,
+      "learning_rate": 9.973595571536593e-06,
+      "loss": 58.1295,
+      "step": 16020
+    },
+    {
+      "epoch": 0.06476322838431299,
+      "grad_norm": 768.2650756835938,
+      "learning_rate": 9.973450621060412e-06,
+      "loss": 58.3537,
+      "step": 16030
+    },
+    {
+      "epoch": 0.06480362964968063,
+      "grad_norm": 626.2802734375,
+      "learning_rate": 9.97330527486959e-06,
+      "loss": 74.6434,
+      "step": 16040
+    },
+    {
+      "epoch": 0.06484403091504826,
+      "grad_norm": 805.8840942382812,
+      "learning_rate": 9.973159532975691e-06,
+      "loss": 50.8522,
+      "step": 16050
+    },
+    {
+      "epoch": 0.0648844321804159,
+      "grad_norm": 612.8935546875,
+      "learning_rate": 9.973013395390314e-06,
+      "loss": 43.9846,
+      "step": 16060
+    },
+    {
+      "epoch": 0.06492483344578352,
+      "grad_norm": 711.977294921875,
+      "learning_rate": 9.972866862125083e-06,
+      "loss": 54.3085,
+      "step": 16070
+    },
+    {
+      "epoch": 0.06496523471115115,
+      "grad_norm": 1257.6363525390625,
+      "learning_rate": 9.972719933191657e-06,
+      "loss": 45.2739,
+      "step": 16080
+    },
+    {
+      "epoch": 0.06500563597651879,
+      "grad_norm": 828.5887451171875,
+      "learning_rate": 9.97257260860173e-06,
+      "loss": 51.8254,
+      "step": 16090
+    },
+    {
+      "epoch": 0.06504603724188641,
+      "grad_norm": 885.0379638671875,
+      "learning_rate": 9.972424888367019e-06,
+      "loss": 44.1245,
+      "step": 16100
+    },
+    {
+      "epoch": 0.06508643850725404,
+      "grad_norm": 699.1367797851562,
+      "learning_rate": 9.972276772499281e-06,
+      "loss": 49.8983,
+      "step": 16110
+    },
+    {
+      "epoch": 0.06512683977262168,
+      "grad_norm": 619.27099609375,
+      "learning_rate": 9.9721282610103e-06,
+      "loss": 43.9222,
+      "step": 16120
+    },
+    {
+      "epoch": 0.0651672410379893,
+      "grad_norm": 1254.580322265625,
+      "learning_rate": 9.971979353911891e-06,
+      "loss": 65.6907,
+      "step": 16130
+    },
+    {
+      "epoch": 0.06520764230335695,
+      "grad_norm": 856.28759765625,
+      "learning_rate": 9.971830051215905e-06,
+      "loss": 73.6939,
+      "step": 16140
+    },
+    {
+      "epoch": 0.06524804356872457,
+      "grad_norm": 703.4512329101562,
+      "learning_rate": 9.97168035293422e-06,
+      "loss": 53.0731,
+      "step": 16150
+    },
+    {
+      "epoch": 0.0652884448340922,
+      "grad_norm": 1020.3666381835938,
+      "learning_rate": 9.971530259078743e-06,
+      "loss": 42.901,
+      "step": 16160
+    },
+    {
+      "epoch": 0.06532884609945984,
+      "grad_norm": 570.0337524414062,
+      "learning_rate": 9.971379769661422e-06,
+      "loss": 68.6924,
+      "step": 16170
+    },
+    {
+      "epoch": 0.06536924736482747,
+      "grad_norm": 586.2837524414062,
+      "learning_rate": 9.971228884694228e-06,
+      "loss": 43.7809,
+      "step": 16180
+    },
+    {
+      "epoch": 0.06540964863019509,
+      "grad_norm": 636.9314575195312,
+      "learning_rate": 9.971077604189166e-06,
+      "loss": 68.6276,
+      "step": 16190
+    },
+    {
+      "epoch": 0.06545004989556273,
+      "grad_norm": 369.6883850097656,
+      "learning_rate": 9.970925928158275e-06,
+      "loss": 38.0615,
+      "step": 16200
+    },
+    {
+      "epoch": 0.06549045116093036,
+      "grad_norm": 545.9331665039062,
+      "learning_rate": 9.970773856613617e-06,
+      "loss": 60.5622,
+      "step": 16210
+    },
+    {
+      "epoch": 0.065530852426298,
+      "grad_norm": 390.1427001953125,
+      "learning_rate": 9.970621389567301e-06,
+      "loss": 51.0867,
+      "step": 16220
+    },
+    {
+      "epoch": 0.06557125369166562,
+      "grad_norm": 473.8525085449219,
+      "learning_rate": 9.97046852703145e-06,
+      "loss": 72.4376,
+      "step": 16230
+    },
+    {
+      "epoch": 0.06561165495703325,
+      "grad_norm": 807.6546630859375,
+      "learning_rate": 9.970315269018231e-06,
+      "loss": 48.2503,
+      "step": 16240
+    },
+    {
+      "epoch": 0.06565205622240089,
+      "grad_norm": 255.02505493164062,
+      "learning_rate": 9.970161615539837e-06,
+      "loss": 40.2138,
+      "step": 16250
+    },
+    {
+      "epoch": 0.06569245748776852,
+      "grad_norm": 471.5314636230469,
+      "learning_rate": 9.970007566608492e-06,
+      "loss": 55.3855,
+      "step": 16260
+    },
+    {
+      "epoch": 0.06573285875313614,
+      "grad_norm": 512.8792724609375,
+      "learning_rate": 9.969853122236455e-06,
+      "loss": 62.5609,
+      "step": 16270
+    },
+    {
+      "epoch": 0.06577326001850378,
+      "grad_norm": 733.0510864257812,
+      "learning_rate": 9.969698282436013e-06,
+      "loss": 87.3586,
+      "step": 16280
+    },
+    {
+      "epoch": 0.06581366128387141,
+      "grad_norm": 451.9583740234375,
+      "learning_rate": 9.969543047219487e-06,
+      "loss": 50.8946,
+      "step": 16290
+    },
+    {
+      "epoch": 0.06585406254923905,
+      "grad_norm": 638.0576171875,
+      "learning_rate": 9.969387416599227e-06,
+      "loss": 77.3127,
+      "step": 16300
+    },
+    {
+      "epoch": 0.06589446381460667,
+      "grad_norm": 542.7348022460938,
+      "learning_rate": 9.969231390587618e-06,
+      "loss": 53.3758,
+      "step": 16310
+    },
+    {
+      "epoch": 0.0659348650799743,
+      "grad_norm": 553.0634155273438,
+      "learning_rate": 9.969074969197072e-06,
+      "loss": 57.5647,
+      "step": 16320
+    },
+    {
+      "epoch": 0.06597526634534194,
+      "grad_norm": 298.8785400390625,
+      "learning_rate": 9.968918152440036e-06,
+      "loss": 52.9646,
+      "step": 16330
+    },
+    {
+      "epoch": 0.06601566761070957,
+      "grad_norm": 943.7760009765625,
+      "learning_rate": 9.968760940328987e-06,
+      "loss": 52.4363,
+      "step": 16340
+    },
+    {
+      "epoch": 0.06605606887607719,
+      "grad_norm": 841.0722045898438,
+      "learning_rate": 9.968603332876435e-06,
+      "loss": 65.1616,
+      "step": 16350
+    },
+    {
+      "epoch": 0.06609647014144483,
+      "grad_norm": 310.6723327636719,
+      "learning_rate": 9.968445330094915e-06,
+      "loss": 49.5367,
+      "step": 16360
+    },
+    {
+      "epoch": 0.06613687140681246,
+      "grad_norm": 364.29095458984375,
+      "learning_rate": 9.968286931997004e-06,
+      "loss": 42.1819,
+      "step": 16370
+    },
+    {
+      "epoch": 0.0661772726721801,
+      "grad_norm": 806.6517333984375,
+      "learning_rate": 9.968128138595304e-06,
+      "loss": 59.3689,
+      "step": 16380
+    },
+    {
+      "epoch": 0.06621767393754772,
+      "grad_norm": 540.7401733398438,
+      "learning_rate": 9.967968949902448e-06,
+      "loss": 50.0984,
+      "step": 16390
+    },
+    {
+      "epoch": 0.06625807520291535,
+      "grad_norm": 734.2467041015625,
+      "learning_rate": 9.967809365931102e-06,
+      "loss": 39.6099,
+      "step": 16400
+    },
+    {
+      "epoch": 0.06629847646828299,
+      "grad_norm": 934.7858276367188,
+      "learning_rate": 9.967649386693964e-06,
+      "loss": 46.0471,
+      "step": 16410
+    },
+    {
+      "epoch": 0.06633887773365062,
+      "grad_norm": 823.3958129882812,
+      "learning_rate": 9.967489012203765e-06,
+      "loss": 40.7156,
+      "step": 16420
+    },
+    {
+      "epoch": 0.06637927899901824,
+      "grad_norm": 1129.1982421875,
+      "learning_rate": 9.967328242473261e-06,
+      "loss": 42.2821,
+      "step": 16430
+    },
+    {
+      "epoch": 0.06641968026438588,
+      "grad_norm": 509.66717529296875,
+      "learning_rate": 9.967167077515246e-06,
+      "loss": 54.3899,
+      "step": 16440
+    },
+    {
+      "epoch": 0.06646008152975351,
+      "grad_norm": 269.69049072265625,
+      "learning_rate": 9.967005517342544e-06,
+      "loss": 41.5783,
+      "step": 16450
+    },
+    {
+      "epoch": 0.06650048279512115,
+      "grad_norm": 178.01014709472656,
+      "learning_rate": 9.966843561968005e-06,
+      "loss": 43.4332,
+      "step": 16460
+    },
+    {
+      "epoch": 0.06654088406048878,
+      "grad_norm": 854.1151733398438,
+      "learning_rate": 9.966681211404521e-06,
+      "loss": 57.2772,
+      "step": 16470
+    },
+    {
+      "epoch": 0.0665812853258564,
+      "grad_norm": 602.7614135742188,
+      "learning_rate": 9.966518465665007e-06,
+      "loss": 35.6911,
+      "step": 16480
+    },
+    {
+      "epoch": 0.06662168659122404,
+      "grad_norm": 882.269775390625,
+      "learning_rate": 9.966355324762412e-06,
+      "loss": 44.7084,
+      "step": 16490
+    },
+    {
+      "epoch": 0.06666208785659167,
+      "grad_norm": 649.4912719726562,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 63.0644,
+      "step": 16500
+    },
+    {
+      "epoch": 0.0667024891219593,
+      "grad_norm": 466.97381591796875,
+      "learning_rate": 9.966027857519931e-06,
+      "loss": 29.1349,
+      "step": 16510
+    },
+    {
+      "epoch": 0.06674289038732693,
+      "grad_norm": 271.84954833984375,
+      "learning_rate": 9.9658635312061e-06,
+      "loss": 58.9536,
+      "step": 16520
+    },
+    {
+      "epoch": 0.06678329165269456,
+      "grad_norm": 553.5524291992188,
+      "learning_rate": 9.965698809781298e-06,
+      "loss": 46.6408,
+      "step": 16530
+    },
+    {
+      "epoch": 0.0668236929180622,
+      "grad_norm": 872.4886474609375,
+      "learning_rate": 9.965533693258632e-06,
+      "loss": 45.775,
+      "step": 16540
+    },
+    {
+      "epoch": 0.06686409418342983,
+      "grad_norm": 670.1477661132812,
+      "learning_rate": 9.965368181651239e-06,
+      "loss": 75.9548,
+      "step": 16550
+    },
+    {
+      "epoch": 0.06690449544879745,
+      "grad_norm": 708.0873413085938,
+      "learning_rate": 9.965202274972288e-06,
+      "loss": 53.3991,
+      "step": 16560
+    },
+    {
+      "epoch": 0.06694489671416509,
+      "grad_norm": 487.0908203125,
+      "learning_rate": 9.965035973234977e-06,
+      "loss": 50.9436,
+      "step": 16570
+    },
+    {
+      "epoch": 0.06698529797953272,
+      "grad_norm": 1012.3402099609375,
+      "learning_rate": 9.964869276452542e-06,
+      "loss": 61.3239,
+      "step": 16580
+    },
+    {
+      "epoch": 0.06702569924490034,
+      "grad_norm": 987.9320068359375,
+      "learning_rate": 9.964702184638244e-06,
+      "loss": 65.6153,
+      "step": 16590
+    },
+    {
+      "epoch": 0.06706610051026798,
+      "grad_norm": 512.2110595703125,
+      "learning_rate": 9.964534697805377e-06,
+      "loss": 48.5166,
+      "step": 16600
+    },
+    {
+      "epoch": 0.06710650177563561,
+      "grad_norm": 300.725341796875,
+      "learning_rate": 9.96436681596727e-06,
+      "loss": 53.5148,
+      "step": 16610
+    },
+    {
+      "epoch": 0.06714690304100325,
+      "grad_norm": 512.7298583984375,
+      "learning_rate": 9.964198539137277e-06,
+      "loss": 59.9783,
+      "step": 16620
+    },
+    {
+      "epoch": 0.06718730430637088,
+      "grad_norm": 1161.68994140625,
+      "learning_rate": 9.964029867328791e-06,
+      "loss": 74.497,
+      "step": 16630
+    },
+    {
+      "epoch": 0.0672277055717385,
+      "grad_norm": 1012.3746337890625,
+      "learning_rate": 9.963860800555228e-06,
+      "loss": 78.1185,
+      "step": 16640
+    },
+    {
+      "epoch": 0.06726810683710614,
+      "grad_norm": 596.1810302734375,
+      "learning_rate": 9.963691338830045e-06,
+      "loss": 46.9206,
+      "step": 16650
+    },
+    {
+      "epoch": 0.06730850810247377,
+      "grad_norm": 375.6921081542969,
+      "learning_rate": 9.963521482166718e-06,
+      "loss": 42.1555,
+      "step": 16660
+    },
+    {
+      "epoch": 0.0673489093678414,
+      "grad_norm": 482.91064453125,
+      "learning_rate": 9.96335123057877e-06,
+      "loss": 54.3135,
+      "step": 16670
+    },
+    {
+      "epoch": 0.06738931063320903,
+      "grad_norm": 398.011962890625,
+      "learning_rate": 9.963180584079741e-06,
+      "loss": 41.0378,
+      "step": 16680
+    },
+    {
+      "epoch": 0.06742971189857666,
+      "grad_norm": 317.195068359375,
+      "learning_rate": 9.963009542683214e-06,
+      "loss": 53.7055,
+      "step": 16690
+    },
+    {
+      "epoch": 0.0674701131639443,
+      "grad_norm": 198.84494018554688,
+      "learning_rate": 9.962838106402791e-06,
+      "loss": 43.8947,
+      "step": 16700
+    },
+    {
+      "epoch": 0.06751051442931193,
+      "grad_norm": 284.9634704589844,
+      "learning_rate": 9.962666275252117e-06,
+      "loss": 58.5298,
+      "step": 16710
+    },
+    {
+      "epoch": 0.06755091569467955,
+      "grad_norm": 675.8781127929688,
+      "learning_rate": 9.962494049244866e-06,
+      "loss": 54.8316,
+      "step": 16720
+    },
+    {
+      "epoch": 0.06759131696004719,
+      "grad_norm": 852.4727783203125,
+      "learning_rate": 9.962321428394735e-06,
+      "loss": 52.3011,
+      "step": 16730
+    },
+    {
+      "epoch": 0.06763171822541482,
+      "grad_norm": 344.2173767089844,
+      "learning_rate": 9.962148412715464e-06,
+      "loss": 43.6536,
+      "step": 16740
+    },
+    {
+      "epoch": 0.06767211949078245,
+      "grad_norm": 601.1047973632812,
+      "learning_rate": 9.961975002220816e-06,
+      "loss": 44.1749,
+      "step": 16750
+    },
+    {
+      "epoch": 0.06771252075615009,
+      "grad_norm": 740.0291137695312,
+      "learning_rate": 9.96180119692459e-06,
+      "loss": 42.4861,
+      "step": 16760
+    },
+    {
+      "epoch": 0.06775292202151771,
+      "grad_norm": 573.695556640625,
+      "learning_rate": 9.961626996840613e-06,
+      "loss": 39.4819,
+      "step": 16770
+    },
+    {
+      "epoch": 0.06779332328688535,
+      "grad_norm": 616.494384765625,
+      "learning_rate": 9.961452401982748e-06,
+      "loss": 54.9687,
+      "step": 16780
+    },
+    {
+      "epoch": 0.06783372455225298,
+      "grad_norm": 780.5487060546875,
+      "learning_rate": 9.961277412364884e-06,
+      "loss": 54.1724,
+      "step": 16790
+    },
+    {
+      "epoch": 0.0678741258176206,
+      "grad_norm": 1156.9290771484375,
+      "learning_rate": 9.961102028000948e-06,
+      "loss": 68.4075,
+      "step": 16800
+    },
+    {
+      "epoch": 0.06791452708298824,
+      "grad_norm": 1188.955810546875,
+      "learning_rate": 9.96092624890489e-06,
+      "loss": 67.3586,
+      "step": 16810
+    },
+    {
+      "epoch": 0.06795492834835587,
+      "grad_norm": 912.7185668945312,
+      "learning_rate": 9.960750075090698e-06,
+      "loss": 51.0991,
+      "step": 16820
+    },
+    {
+      "epoch": 0.0679953296137235,
+      "grad_norm": 918.0380859375,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 46.7497,
+      "step": 16830
+    },
+    {
+      "epoch": 0.06803573087909114,
+      "grad_norm": 1461.614501953125,
+      "learning_rate": 9.960396543364013e-06,
+      "loss": 52.6995,
+      "step": 16840
+    },
+    {
+      "epoch": 0.06807613214445876,
+      "grad_norm": 1134.6422119140625,
+      "learning_rate": 9.96021918547965e-06,
+      "loss": 52.5664,
+      "step": 16850
+    },
+    {
+      "epoch": 0.0681165334098264,
+      "grad_norm": 1380.5745849609375,
+      "learning_rate": 9.96004143293341e-06,
+      "loss": 50.9108,
+      "step": 16860
+    },
+    {
+      "epoch": 0.06815693467519403,
+      "grad_norm": 706.7164306640625,
+      "learning_rate": 9.959863285739436e-06,
+      "loss": 75.2603,
+      "step": 16870
+    },
+    {
+      "epoch": 0.06819733594056165,
+      "grad_norm": 847.0613403320312,
+      "learning_rate": 9.959684743911904e-06,
+      "loss": 56.6586,
+      "step": 16880
+    },
+    {
+      "epoch": 0.0682377372059293,
+      "grad_norm": 560.4253540039062,
+      "learning_rate": 9.959505807465018e-06,
+      "loss": 47.3038,
+      "step": 16890
+    },
+    {
+      "epoch": 0.06827813847129692,
+      "grad_norm": 759.576171875,
+      "learning_rate": 9.959326476413016e-06,
+      "loss": 60.7899,
+      "step": 16900
+    },
+    {
+      "epoch": 0.06831853973666455,
+      "grad_norm": 742.4689331054688,
+      "learning_rate": 9.959146750770167e-06,
+      "loss": 56.8813,
+      "step": 16910
+    },
+    {
+      "epoch": 0.06835894100203219,
+      "grad_norm": 1035.506103515625,
+      "learning_rate": 9.95896663055077e-06,
+      "loss": 47.8974,
+      "step": 16920
+    },
+    {
+      "epoch": 0.06839934226739981,
+      "grad_norm": 734.8804931640625,
+      "learning_rate": 9.958786115769157e-06,
+      "loss": 59.3255,
+      "step": 16930
+    },
+    {
+      "epoch": 0.06843974353276745,
+      "grad_norm": 514.1586303710938,
+      "learning_rate": 9.958605206439692e-06,
+      "loss": 56.9457,
+      "step": 16940
+    },
+    {
+      "epoch": 0.06848014479813508,
+      "grad_norm": 523.6434326171875,
+      "learning_rate": 9.958423902576764e-06,
+      "loss": 37.5847,
+      "step": 16950
+    },
+    {
+      "epoch": 0.0685205460635027,
+      "grad_norm": 855.633056640625,
+      "learning_rate": 9.958242204194804e-06,
+      "loss": 62.8069,
+      "step": 16960
+    },
+    {
+      "epoch": 0.06856094732887034,
+      "grad_norm": 1315.413818359375,
+      "learning_rate": 9.958060111308267e-06,
+      "loss": 63.7991,
+      "step": 16970
+    },
+    {
+      "epoch": 0.06860134859423797,
+      "grad_norm": 460.4294738769531,
+      "learning_rate": 9.957877623931642e-06,
+      "loss": 55.9995,
+      "step": 16980
+    },
+    {
+      "epoch": 0.0686417498596056,
+      "grad_norm": 2325.47412109375,
+      "learning_rate": 9.95769474207945e-06,
+      "loss": 81.635,
+      "step": 16990
+    },
+    {
+      "epoch": 0.06868215112497324,
+      "grad_norm": 521.9066162109375,
+      "learning_rate": 9.957511465766236e-06,
+      "loss": 120.9248,
+      "step": 17000
+    },
+    {
+      "epoch": 0.06872255239034086,
+      "grad_norm": 650.1915893554688,
+      "learning_rate": 9.957327795006589e-06,
+      "loss": 61.7534,
+      "step": 17010
+    },
+    {
+      "epoch": 0.0687629536557085,
+      "grad_norm": 781.476806640625,
+      "learning_rate": 9.95714372981512e-06,
+      "loss": 52.5771,
+      "step": 17020
+    },
+    {
+      "epoch": 0.06880335492107613,
+      "grad_norm": 1087.2088623046875,
+      "learning_rate": 9.956959270206474e-06,
+      "loss": 55.3713,
+      "step": 17030
+    },
+    {
+      "epoch": 0.06884375618644376,
+      "grad_norm": 891.9490356445312,
+      "learning_rate": 9.956774416195329e-06,
+      "loss": 50.7214,
+      "step": 17040
+    },
+    {
+      "epoch": 0.0688841574518114,
+      "grad_norm": 289.277587890625,
+      "learning_rate": 9.956589167796392e-06,
+      "loss": 58.5371,
+      "step": 17050
+    },
+    {
+      "epoch": 0.06892455871717902,
+      "grad_norm": 641.5277709960938,
+      "learning_rate": 9.956403525024402e-06,
+      "loss": 46.9823,
+      "step": 17060
+    },
+    {
+      "epoch": 0.06896495998254665,
+      "grad_norm": 1088.70458984375,
+      "learning_rate": 9.956217487894131e-06,
+      "loss": 49.6527,
+      "step": 17070
+    },
+    {
+      "epoch": 0.06900536124791429,
+      "grad_norm": 795.5383911132812,
+      "learning_rate": 9.95603105642038e-06,
+      "loss": 41.3176,
+      "step": 17080
+    },
+    {
+      "epoch": 0.06904576251328191,
+      "grad_norm": 531.6299438476562,
+      "learning_rate": 9.955844230617985e-06,
+      "loss": 48.9703,
+      "step": 17090
+    },
+    {
+      "epoch": 0.06908616377864955,
+      "grad_norm": 1016.78271484375,
+      "learning_rate": 9.955657010501807e-06,
+      "loss": 54.1437,
+      "step": 17100
+    },
+    {
+      "epoch": 0.06912656504401718,
+      "grad_norm": 447.7895202636719,
+      "learning_rate": 9.955469396086743e-06,
+      "loss": 62.9068,
+      "step": 17110
+    },
+    {
+      "epoch": 0.0691669663093848,
+      "grad_norm": 729.5142822265625,
+      "learning_rate": 9.955281387387724e-06,
+      "loss": 39.2298,
+      "step": 17120
+    },
+    {
+      "epoch": 0.06920736757475245,
+      "grad_norm": 582.5546875,
+      "learning_rate": 9.955092984419705e-06,
+      "loss": 52.08,
+      "step": 17130
+    },
+    {
+      "epoch": 0.06924776884012007,
+      "grad_norm": 921.3667602539062,
+      "learning_rate": 9.954904187197679e-06,
+      "loss": 57.2562,
+      "step": 17140
+    },
+    {
+      "epoch": 0.0692881701054877,
+      "grad_norm": 556.9216918945312,
+      "learning_rate": 9.954714995736667e-06,
+      "loss": 47.0871,
+      "step": 17150
+    },
+    {
+      "epoch": 0.06932857137085534,
+      "grad_norm": 1060.7427978515625,
+      "learning_rate": 9.95452541005172e-06,
+      "loss": 54.6456,
+      "step": 17160
+    },
+    {
+      "epoch": 0.06936897263622296,
+      "grad_norm": 492.2557678222656,
+      "learning_rate": 9.954335430157926e-06,
+      "loss": 57.3385,
+      "step": 17170
+    },
+    {
+      "epoch": 0.0694093739015906,
+      "grad_norm": 812.3903198242188,
+      "learning_rate": 9.9541450560704e-06,
+      "loss": 46.2137,
+      "step": 17180
+    },
+    {
+      "epoch": 0.06944977516695823,
+      "grad_norm": 1028.221923828125,
+      "learning_rate": 9.953954287804286e-06,
+      "loss": 77.8132,
+      "step": 17190
+    },
+    {
+      "epoch": 0.06949017643232586,
+      "grad_norm": 511.1866149902344,
+      "learning_rate": 9.953763125374767e-06,
+      "loss": 53.9452,
+      "step": 17200
+    },
+    {
+      "epoch": 0.0695305776976935,
+      "grad_norm": 617.6327514648438,
+      "learning_rate": 9.953571568797049e-06,
+      "loss": 49.3006,
+      "step": 17210
+    },
+    {
+      "epoch": 0.06957097896306112,
+      "grad_norm": 725.1810913085938,
+      "learning_rate": 9.953379618086377e-06,
+      "loss": 53.0108,
+      "step": 17220
+    },
+    {
+      "epoch": 0.06961138022842875,
+      "grad_norm": 1026.266357421875,
+      "learning_rate": 9.95318727325802e-06,
+      "loss": 60.5115,
+      "step": 17230
+    },
+    {
+      "epoch": 0.06965178149379639,
+      "grad_norm": 457.9506530761719,
+      "learning_rate": 9.952994534327283e-06,
+      "loss": 40.233,
+      "step": 17240
+    },
+    {
+      "epoch": 0.06969218275916401,
+      "grad_norm": 95.95256042480469,
+      "learning_rate": 9.952801401309504e-06,
+      "loss": 54.0698,
+      "step": 17250
+    },
+    {
+      "epoch": 0.06973258402453165,
+      "grad_norm": 369.8098449707031,
+      "learning_rate": 9.952607874220048e-06,
+      "loss": 29.6986,
+      "step": 17260
+    },
+    {
+      "epoch": 0.06977298528989928,
+      "grad_norm": 605.2461547851562,
+      "learning_rate": 9.952413953074312e-06,
+      "loss": 48.9784,
+      "step": 17270
+    },
+    {
+      "epoch": 0.06981338655526691,
+      "grad_norm": 1227.542724609375,
+      "learning_rate": 9.952219637887725e-06,
+      "loss": 63.727,
+      "step": 17280
+    },
+    {
+      "epoch": 0.06985378782063455,
+      "grad_norm": 554.7730102539062,
+      "learning_rate": 9.952024928675752e-06,
+      "loss": 55.0353,
+      "step": 17290
+    },
+    {
+      "epoch": 0.06989418908600217,
+      "grad_norm": 1237.1422119140625,
+      "learning_rate": 9.951829825453881e-06,
+      "loss": 41.5354,
+      "step": 17300
+    },
+    {
+      "epoch": 0.0699345903513698,
+      "grad_norm": 508.1934814453125,
+      "learning_rate": 9.951634328237635e-06,
+      "loss": 54.589,
+      "step": 17310
+    },
+    {
+      "epoch": 0.06997499161673744,
+      "grad_norm": 568.03076171875,
+      "learning_rate": 9.951438437042572e-06,
+      "loss": 50.1381,
+      "step": 17320
+    },
+    {
+      "epoch": 0.07001539288210507,
+      "grad_norm": 330.10528564453125,
+      "learning_rate": 9.951242151884275e-06,
+      "loss": 33.1387,
+      "step": 17330
+    },
+    {
+      "epoch": 0.0700557941474727,
+      "grad_norm": 426.9512634277344,
+      "learning_rate": 9.951045472778365e-06,
+      "loss": 48.9173,
+      "step": 17340
+    },
+    {
+      "epoch": 0.07009619541284033,
+      "grad_norm": 860.360595703125,
+      "learning_rate": 9.950848399740488e-06,
+      "loss": 52.8169,
+      "step": 17350
+    },
+    {
+      "epoch": 0.07013659667820796,
+      "grad_norm": 694.7147827148438,
+      "learning_rate": 9.950650932786325e-06,
+      "loss": 53.9027,
+      "step": 17360
+    },
+    {
+      "epoch": 0.0701769979435756,
+      "grad_norm": 2214.466552734375,
+      "learning_rate": 9.95045307193159e-06,
+      "loss": 72.7385,
+      "step": 17370
+    },
+    {
+      "epoch": 0.07021739920894322,
+      "grad_norm": 922.2384033203125,
+      "learning_rate": 9.95025481719202e-06,
+      "loss": 69.9777,
+      "step": 17380
+    },
+    {
+      "epoch": 0.07025780047431085,
+      "grad_norm": 1792.91162109375,
+      "learning_rate": 9.950056168583395e-06,
+      "loss": 48.3632,
+      "step": 17390
+    },
+    {
+      "epoch": 0.07029820173967849,
+      "grad_norm": 718.42431640625,
+      "learning_rate": 9.949857126121519e-06,
+      "loss": 30.7003,
+      "step": 17400
+    },
+    {
+      "epoch": 0.07033860300504612,
+      "grad_norm": 378.0965881347656,
+      "learning_rate": 9.949657689822226e-06,
+      "loss": 66.112,
+      "step": 17410
+    },
+    {
+      "epoch": 0.07037900427041376,
+      "grad_norm": 893.2484130859375,
+      "learning_rate": 9.949457859701388e-06,
+      "loss": 71.9019,
+      "step": 17420
+    },
+    {
+      "epoch": 0.07041940553578138,
+      "grad_norm": 536.0899658203125,
+      "learning_rate": 9.949257635774903e-06,
+      "loss": 57.2,
+      "step": 17430
+    },
+    {
+      "epoch": 0.07045980680114901,
+      "grad_norm": 724.5510864257812,
+      "learning_rate": 9.9490570180587e-06,
+      "loss": 41.6391,
+      "step": 17440
+    },
+    {
+      "epoch": 0.07050020806651665,
+      "grad_norm": 1140.1925048828125,
+      "learning_rate": 9.948856006568746e-06,
+      "loss": 34.9596,
+      "step": 17450
+    },
+    {
+      "epoch": 0.07054060933188427,
+      "grad_norm": 1027.14697265625,
+      "learning_rate": 9.94865460132103e-06,
+      "loss": 47.5769,
+      "step": 17460
+    },
+    {
+      "epoch": 0.0705810105972519,
+      "grad_norm": 1001.2021484375,
+      "learning_rate": 9.948452802331578e-06,
+      "loss": 63.4763,
+      "step": 17470
+    },
+    {
+      "epoch": 0.07062141186261954,
+      "grad_norm": 751.27978515625,
+      "learning_rate": 9.948250609616449e-06,
+      "loss": 55.1303,
+      "step": 17480
+    },
+    {
+      "epoch": 0.07066181312798717,
+      "grad_norm": 355.3070068359375,
+      "learning_rate": 9.948048023191728e-06,
+      "loss": 43.9042,
+      "step": 17490
+    },
+    {
+      "epoch": 0.0707022143933548,
+      "grad_norm": 516.75244140625,
+      "learning_rate": 9.947845043073533e-06,
+      "loss": 44.1306,
+      "step": 17500
+    },
+    {
+      "epoch": 0.07074261565872243,
+      "grad_norm": 1007.9006958007812,
+      "learning_rate": 9.947641669278016e-06,
+      "loss": 61.7492,
+      "step": 17510
+    },
+    {
+      "epoch": 0.07078301692409006,
+      "grad_norm": 462.08758544921875,
+      "learning_rate": 9.947437901821358e-06,
+      "loss": 38.7414,
+      "step": 17520
+    },
+    {
+      "epoch": 0.0708234181894577,
+      "grad_norm": 893.343505859375,
+      "learning_rate": 9.947233740719772e-06,
+      "loss": 50.9898,
+      "step": 17530
+    },
+    {
+      "epoch": 0.07086381945482532,
+      "grad_norm": 501.1763916015625,
+      "learning_rate": 9.947029185989501e-06,
+      "loss": 49.4258,
+      "step": 17540
+    },
+    {
+      "epoch": 0.07090422072019295,
+      "grad_norm": 495.0986022949219,
+      "learning_rate": 9.946824237646823e-06,
+      "loss": 68.988,
+      "step": 17550
+    },
+    {
+      "epoch": 0.07094462198556059,
+      "grad_norm": 327.663818359375,
+      "learning_rate": 9.946618895708043e-06,
+      "loss": 73.2408,
+      "step": 17560
+    },
+    {
+      "epoch": 0.07098502325092822,
+      "grad_norm": 346.1497497558594,
+      "learning_rate": 9.946413160189498e-06,
+      "loss": 39.7363,
+      "step": 17570
+    },
+    {
+      "epoch": 0.07102542451629586,
+      "grad_norm": 592.0716552734375,
+      "learning_rate": 9.946207031107562e-06,
+      "loss": 49.571,
+      "step": 17580
+    },
+    {
+      "epoch": 0.07106582578166348,
+      "grad_norm": 463.5035400390625,
+      "learning_rate": 9.94600050847863e-06,
+      "loss": 58.9509,
+      "step": 17590
+    },
+    {
+      "epoch": 0.07110622704703111,
+      "grad_norm": 518.01904296875,
+      "learning_rate": 9.945793592319137e-06,
+      "loss": 57.2311,
+      "step": 17600
+    },
+    {
+      "epoch": 0.07114662831239875,
+      "grad_norm": 1007.0841064453125,
+      "learning_rate": 9.945586282645545e-06,
+      "loss": 50.4529,
+      "step": 17610
+    },
+    {
+      "epoch": 0.07118702957776638,
+      "grad_norm": 822.48583984375,
+      "learning_rate": 9.945378579474351e-06,
+      "loss": 62.1675,
+      "step": 17620
+    },
+    {
+      "epoch": 0.071227430843134,
+      "grad_norm": 550.595458984375,
+      "learning_rate": 9.945170482822079e-06,
+      "loss": 44.7671,
+      "step": 17630
+    },
+    {
+      "epoch": 0.07126783210850164,
+      "grad_norm": 550.9453125,
+      "learning_rate": 9.944961992705288e-06,
+      "loss": 40.3771,
+      "step": 17640
+    },
+    {
+      "epoch": 0.07130823337386927,
+      "grad_norm": 926.0831298828125,
+      "learning_rate": 9.944753109140564e-06,
+      "loss": 83.9147,
+      "step": 17650
+    },
+    {
+      "epoch": 0.07134863463923691,
+      "grad_norm": 303.1770324707031,
+      "learning_rate": 9.94454383214453e-06,
+      "loss": 47.9595,
+      "step": 17660
+    },
+    {
+      "epoch": 0.07138903590460453,
+      "grad_norm": 736.1362915039062,
+      "learning_rate": 9.944334161733835e-06,
+      "loss": 58.8802,
+      "step": 17670
+    },
+    {
+      "epoch": 0.07142943716997216,
+      "grad_norm": 927.33935546875,
+      "learning_rate": 9.944124097925161e-06,
+      "loss": 65.2757,
+      "step": 17680
+    },
+    {
+      "epoch": 0.0714698384353398,
+      "grad_norm": 337.0379638671875,
+      "learning_rate": 9.943913640735224e-06,
+      "loss": 52.1521,
+      "step": 17690
+    },
+    {
+      "epoch": 0.07151023970070743,
+      "grad_norm": 814.1279907226562,
+      "learning_rate": 9.94370279018077e-06,
+      "loss": 54.99,
+      "step": 17700
+    },
+    {
+      "epoch": 0.07155064096607505,
+      "grad_norm": 695.0791625976562,
+      "learning_rate": 9.94349154627857e-06,
+      "loss": 48.8546,
+      "step": 17710
+    },
+    {
+      "epoch": 0.07159104223144269,
+      "grad_norm": 629.86279296875,
+      "learning_rate": 9.943279909045438e-06,
+      "loss": 45.39,
+      "step": 17720
+    },
+    {
+      "epoch": 0.07163144349681032,
+      "grad_norm": 572.5151977539062,
+      "learning_rate": 9.94306787849821e-06,
+      "loss": 72.4478,
+      "step": 17730
+    },
+    {
+      "epoch": 0.07167184476217796,
+      "grad_norm": 473.1834411621094,
+      "learning_rate": 9.942855454653755e-06,
+      "loss": 44.4805,
+      "step": 17740
+    },
+    {
+      "epoch": 0.07171224602754558,
+      "grad_norm": 653.8013305664062,
+      "learning_rate": 9.942642637528977e-06,
+      "loss": 49.7141,
+      "step": 17750
+    },
+    {
+      "epoch": 0.07175264729291321,
+      "grad_norm": 724.1669921875,
+      "learning_rate": 9.942429427140807e-06,
+      "loss": 42.9303,
+      "step": 17760
+    },
+    {
+      "epoch": 0.07179304855828085,
+      "grad_norm": 848.2758178710938,
+      "learning_rate": 9.942215823506211e-06,
+      "loss": 71.6385,
+      "step": 17770
+    },
+    {
+      "epoch": 0.07183344982364848,
+      "grad_norm": 537.5114135742188,
+      "learning_rate": 9.942001826642184e-06,
+      "loss": 55.1928,
+      "step": 17780
+    },
+    {
+      "epoch": 0.0718738510890161,
+      "grad_norm": 517.9396362304688,
+      "learning_rate": 9.941787436565751e-06,
+      "loss": 42.2545,
+      "step": 17790
+    },
+    {
+      "epoch": 0.07191425235438374,
+      "grad_norm": 1042.4097900390625,
+      "learning_rate": 9.941572653293974e-06,
+      "loss": 41.3418,
+      "step": 17800
+    },
+    {
+      "epoch": 0.07195465361975137,
+      "grad_norm": 686.4190673828125,
+      "learning_rate": 9.941357476843938e-06,
+      "loss": 32.3788,
+      "step": 17810
+    },
+    {
+      "epoch": 0.07199505488511901,
+      "grad_norm": 411.7096862792969,
+      "learning_rate": 9.941141907232766e-06,
+      "loss": 59.7465,
+      "step": 17820
+    },
+    {
+      "epoch": 0.07203545615048663,
+      "grad_norm": 838.7464599609375,
+      "learning_rate": 9.940925944477608e-06,
+      "loss": 51.814,
+      "step": 17830
+    },
+    {
+      "epoch": 0.07207585741585426,
+      "grad_norm": 1071.185546875,
+      "learning_rate": 9.940709588595649e-06,
+      "loss": 53.1841,
+      "step": 17840
+    },
+    {
+      "epoch": 0.0721162586812219,
+      "grad_norm": 970.0137939453125,
+      "learning_rate": 9.940492839604103e-06,
+      "loss": 59.9117,
+      "step": 17850
+    },
+    {
+      "epoch": 0.07215665994658953,
+      "grad_norm": 441.033935546875,
+      "learning_rate": 9.940275697520216e-06,
+      "loss": 38.621,
+      "step": 17860
+    },
+    {
+      "epoch": 0.07219706121195715,
+      "grad_norm": 347.8889465332031,
+      "learning_rate": 9.940058162361264e-06,
+      "loss": 41.1021,
+      "step": 17870
+    },
+    {
+      "epoch": 0.0722374624773248,
+      "grad_norm": 579.1629028320312,
+      "learning_rate": 9.939840234144556e-06,
+      "loss": 54.0829,
+      "step": 17880
+    },
+    {
+      "epoch": 0.07227786374269242,
+      "grad_norm": 941.6212158203125,
+      "learning_rate": 9.939621912887431e-06,
+      "loss": 50.8714,
+      "step": 17890
+    },
+    {
+      "epoch": 0.07231826500806006,
+      "grad_norm": 860.2476196289062,
+      "learning_rate": 9.93940319860726e-06,
+      "loss": 56.4514,
+      "step": 17900
+    },
+    {
+      "epoch": 0.07235866627342769,
+      "grad_norm": 678.0944213867188,
+      "learning_rate": 9.939184091321445e-06,
+      "loss": 51.4201,
+      "step": 17910
+    },
+    {
+      "epoch": 0.07239906753879531,
+      "grad_norm": 1071.39013671875,
+      "learning_rate": 9.938964591047421e-06,
+      "loss": 35.9462,
+      "step": 17920
+    },
+    {
+      "epoch": 0.07243946880416295,
+      "grad_norm": 311.69110107421875,
+      "learning_rate": 9.938744697802651e-06,
+      "loss": 53.9317,
+      "step": 17930
+    },
+    {
+      "epoch": 0.07247987006953058,
+      "grad_norm": 1138.3892822265625,
+      "learning_rate": 9.938524411604631e-06,
+      "loss": 71.4564,
+      "step": 17940
+    },
+    {
+      "epoch": 0.0725202713348982,
+      "grad_norm": 670.2659301757812,
+      "learning_rate": 9.938303732470888e-06,
+      "loss": 54.5628,
+      "step": 17950
+    },
+    {
+      "epoch": 0.07256067260026584,
+      "grad_norm": 587.255615234375,
+      "learning_rate": 9.938082660418981e-06,
+      "loss": 70.5084,
+      "step": 17960
+    },
+    {
+      "epoch": 0.07260107386563347,
+      "grad_norm": 466.6063537597656,
+      "learning_rate": 9.937861195466498e-06,
+      "loss": 53.5537,
+      "step": 17970
+    },
+    {
+      "epoch": 0.07264147513100111,
+      "grad_norm": 783.6781616210938,
+      "learning_rate": 9.937639337631064e-06,
+      "loss": 36.8183,
+      "step": 17980
+    },
+    {
+      "epoch": 0.07268187639636874,
+      "grad_norm": 849.5465698242188,
+      "learning_rate": 9.937417086930328e-06,
+      "loss": 42.0539,
+      "step": 17990
+    },
+    {
+      "epoch": 0.07272227766173636,
+      "grad_norm": 1289.6552734375,
+      "learning_rate": 9.937194443381972e-06,
+      "loss": 46.7966,
+      "step": 18000
+    },
+    {
+      "epoch": 0.072762678927104,
+      "grad_norm": 865.15283203125,
+      "learning_rate": 9.936971407003714e-06,
+      "loss": 60.911,
+      "step": 18010
+    },
+    {
+      "epoch": 0.07280308019247163,
+      "grad_norm": 406.6366882324219,
+      "learning_rate": 9.936747977813299e-06,
+      "loss": 46.7317,
+      "step": 18020
+    },
+    {
+      "epoch": 0.07284348145783925,
+      "grad_norm": 571.2393188476562,
+      "learning_rate": 9.936524155828503e-06,
+      "loss": 40.7895,
+      "step": 18030
+    },
+    {
+      "epoch": 0.0728838827232069,
+      "grad_norm": 671.0640869140625,
+      "learning_rate": 9.936299941067137e-06,
+      "loss": 37.075,
+      "step": 18040
+    },
+    {
+      "epoch": 0.07292428398857452,
+      "grad_norm": 735.3406372070312,
+      "learning_rate": 9.93607533354704e-06,
+      "loss": 64.778,
+      "step": 18050
+    },
+    {
+      "epoch": 0.07296468525394216,
+      "grad_norm": 1079.908447265625,
+      "learning_rate": 9.935850333286081e-06,
+      "loss": 64.6785,
+      "step": 18060
+    },
+    {
+      "epoch": 0.07300508651930979,
+      "grad_norm": 553.8833618164062,
+      "learning_rate": 9.935624940302165e-06,
+      "loss": 43.7391,
+      "step": 18070
+    },
+    {
+      "epoch": 0.07304548778467741,
+      "grad_norm": 821.9501342773438,
+      "learning_rate": 9.93539915461322e-06,
+      "loss": 42.4855,
+      "step": 18080
+    },
+    {
+      "epoch": 0.07308588905004505,
+      "grad_norm": 380.239013671875,
+      "learning_rate": 9.935172976237218e-06,
+      "loss": 51.201,
+      "step": 18090
+    },
+    {
+      "epoch": 0.07312629031541268,
+      "grad_norm": 550.2764282226562,
+      "learning_rate": 9.934946405192152e-06,
+      "loss": 35.0583,
+      "step": 18100
+    },
+    {
+      "epoch": 0.0731666915807803,
+      "grad_norm": 510.43701171875,
+      "learning_rate": 9.934719441496048e-06,
+      "loss": 45.7056,
+      "step": 18110
+    },
+    {
+      "epoch": 0.07320709284614794,
+      "grad_norm": 601.5700073242188,
+      "learning_rate": 9.934492085166965e-06,
+      "loss": 42.423,
+      "step": 18120
+    },
+    {
+      "epoch": 0.07324749411151557,
+      "grad_norm": 621.2982177734375,
+      "learning_rate": 9.934264336222992e-06,
+      "loss": 55.7587,
+      "step": 18130
+    },
+    {
+      "epoch": 0.07328789537688321,
+      "grad_norm": 913.1992797851562,
+      "learning_rate": 9.934036194682253e-06,
+      "loss": 48.2963,
+      "step": 18140
+    },
+    {
+      "epoch": 0.07332829664225084,
+      "grad_norm": 993.3046264648438,
+      "learning_rate": 9.933807660562898e-06,
+      "loss": 52.526,
+      "step": 18150
+    },
+    {
+      "epoch": 0.07336869790761846,
+      "grad_norm": 387.90020751953125,
+      "learning_rate": 9.933578733883109e-06,
+      "loss": 51.2602,
+      "step": 18160
+    },
+    {
+      "epoch": 0.0734090991729861,
+      "grad_norm": 622.7506103515625,
+      "learning_rate": 9.933349414661103e-06,
+      "loss": 53.1697,
+      "step": 18170
+    },
+    {
+      "epoch": 0.07344950043835373,
+      "grad_norm": 767.485595703125,
+      "learning_rate": 9.933119702915125e-06,
+      "loss": 47.9134,
+      "step": 18180
+    },
+    {
+      "epoch": 0.07348990170372136,
+      "grad_norm": 542.6588134765625,
+      "learning_rate": 9.932889598663452e-06,
+      "loss": 53.2502,
+      "step": 18190
+    },
+    {
+      "epoch": 0.073530302969089,
+      "grad_norm": 543.2275390625,
+      "learning_rate": 9.932659101924393e-06,
+      "loss": 50.1185,
+      "step": 18200
+    },
+    {
+      "epoch": 0.07357070423445662,
+      "grad_norm": 629.7283935546875,
+      "learning_rate": 9.932428212716287e-06,
+      "loss": 58.6988,
+      "step": 18210
+    },
+    {
+      "epoch": 0.07361110549982426,
+      "grad_norm": 523.290283203125,
+      "learning_rate": 9.932196931057505e-06,
+      "loss": 51.7237,
+      "step": 18220
+    },
+    {
+      "epoch": 0.07365150676519189,
+      "grad_norm": 1247.166259765625,
+      "learning_rate": 9.931965256966449e-06,
+      "loss": 49.4384,
+      "step": 18230
+    },
+    {
+      "epoch": 0.07369190803055951,
+      "grad_norm": 862.195068359375,
+      "learning_rate": 9.931733190461552e-06,
+      "loss": 59.3807,
+      "step": 18240
+    },
+    {
+      "epoch": 0.07373230929592715,
+      "grad_norm": 806.7147827148438,
+      "learning_rate": 9.931500731561279e-06,
+      "loss": 57.1338,
+      "step": 18250
+    },
+    {
+      "epoch": 0.07377271056129478,
+      "grad_norm": 417.43731689453125,
+      "learning_rate": 9.931267880284124e-06,
+      "loss": 43.645,
+      "step": 18260
+    },
+    {
+      "epoch": 0.0738131118266624,
+      "grad_norm": 608.3236694335938,
+      "learning_rate": 9.931034636648616e-06,
+      "loss": 57.7599,
+      "step": 18270
+    },
+    {
+      "epoch": 0.07385351309203005,
+      "grad_norm": 921.3759765625,
+      "learning_rate": 9.930801000673314e-06,
+      "loss": 53.7194,
+      "step": 18280
+    },
+    {
+      "epoch": 0.07389391435739767,
+      "grad_norm": 536.0777587890625,
+      "learning_rate": 9.930566972376803e-06,
+      "loss": 59.2549,
+      "step": 18290
+    },
+    {
+      "epoch": 0.0739343156227653,
+      "grad_norm": 792.5118408203125,
+      "learning_rate": 9.930332551777709e-06,
+      "loss": 60.8778,
+      "step": 18300
+    },
+    {
+      "epoch": 0.07397471688813294,
+      "grad_norm": 699.9194946289062,
+      "learning_rate": 9.930097738894679e-06,
+      "loss": 60.5394,
+      "step": 18310
+    },
+    {
+      "epoch": 0.07401511815350056,
+      "grad_norm": 360.6986389160156,
+      "learning_rate": 9.929862533746398e-06,
+      "loss": 53.1862,
+      "step": 18320
+    },
+    {
+      "epoch": 0.0740555194188682,
+      "grad_norm": 572.6876220703125,
+      "learning_rate": 9.92962693635158e-06,
+      "loss": 81.4816,
+      "step": 18330
+    },
+    {
+      "epoch": 0.07409592068423583,
+      "grad_norm": 703.1517944335938,
+      "learning_rate": 9.929390946728972e-06,
+      "loss": 56.8753,
+      "step": 18340
+    },
+    {
+      "epoch": 0.07413632194960346,
+      "grad_norm": 594.4182739257812,
+      "learning_rate": 9.929154564897347e-06,
+      "loss": 37.7829,
+      "step": 18350
+    },
+    {
+      "epoch": 0.0741767232149711,
+      "grad_norm": 536.6315307617188,
+      "learning_rate": 9.928917790875519e-06,
+      "loss": 41.1827,
+      "step": 18360
+    },
+    {
+      "epoch": 0.07421712448033872,
+      "grad_norm": 790.8310546875,
+      "learning_rate": 9.92868062468232e-06,
+      "loss": 67.6239,
+      "step": 18370
+    },
+    {
+      "epoch": 0.07425752574570635,
+      "grad_norm": 640.1522216796875,
+      "learning_rate": 9.928443066336624e-06,
+      "loss": 51.2941,
+      "step": 18380
+    },
+    {
+      "epoch": 0.07429792701107399,
+      "grad_norm": 588.3639526367188,
+      "learning_rate": 9.92820511585733e-06,
+      "loss": 63.7813,
+      "step": 18390
+    },
+    {
+      "epoch": 0.07433832827644161,
+      "grad_norm": 385.4635925292969,
+      "learning_rate": 9.927966773263375e-06,
+      "loss": 33.1393,
+      "step": 18400
+    },
+    {
+      "epoch": 0.07437872954180925,
+      "grad_norm": 1008.00439453125,
+      "learning_rate": 9.92772803857372e-06,
+      "loss": 64.6085,
+      "step": 18410
+    },
+    {
+      "epoch": 0.07441913080717688,
+      "grad_norm": 670.56787109375,
+      "learning_rate": 9.927488911807359e-06,
+      "loss": 38.1293,
+      "step": 18420
+    },
+    {
+      "epoch": 0.07445953207254451,
+      "grad_norm": 943.517333984375,
+      "learning_rate": 9.927249392983319e-06,
+      "loss": 53.9865,
+      "step": 18430
+    },
+    {
+      "epoch": 0.07449993333791215,
+      "grad_norm": 616.2203979492188,
+      "learning_rate": 9.927009482120658e-06,
+      "loss": 30.1008,
+      "step": 18440
+    },
+    {
+      "epoch": 0.07454033460327977,
+      "grad_norm": 559.6461181640625,
+      "learning_rate": 9.926769179238467e-06,
+      "loss": 47.3283,
+      "step": 18450
+    },
+    {
+      "epoch": 0.0745807358686474,
+      "grad_norm": 646.236572265625,
+      "learning_rate": 9.926528484355859e-06,
+      "loss": 77.804,
+      "step": 18460
+    },
+    {
+      "epoch": 0.07462113713401504,
+      "grad_norm": 509.07177734375,
+      "learning_rate": 9.926287397491992e-06,
+      "loss": 50.2232,
+      "step": 18470
+    },
+    {
+      "epoch": 0.07466153839938267,
+      "grad_norm": 522.69580078125,
+      "learning_rate": 9.926045918666045e-06,
+      "loss": 44.2353,
+      "step": 18480
+    },
+    {
+      "epoch": 0.0747019396647503,
+      "grad_norm": 729.4260864257812,
+      "learning_rate": 9.925804047897231e-06,
+      "loss": 69.1156,
+      "step": 18490
+    },
+    {
+      "epoch": 0.07474234093011793,
+      "grad_norm": 730.46142578125,
+      "learning_rate": 9.925561785204797e-06,
+      "loss": 46.6243,
+      "step": 18500
+    },
+    {
+      "epoch": 0.07478274219548556,
+      "grad_norm": 406.8343505859375,
+      "learning_rate": 9.925319130608015e-06,
+      "loss": 47.7084,
+      "step": 18510
+    },
+    {
+      "epoch": 0.0748231434608532,
+      "grad_norm": 547.1348876953125,
+      "learning_rate": 9.925076084126194e-06,
+      "loss": 40.9297,
+      "step": 18520
+    },
+    {
+      "epoch": 0.07486354472622082,
+      "grad_norm": 466.9611511230469,
+      "learning_rate": 9.924832645778674e-06,
+      "loss": 46.5333,
+      "step": 18530
+    },
+    {
+      "epoch": 0.07490394599158845,
+      "grad_norm": 1125.0511474609375,
+      "learning_rate": 9.924588815584822e-06,
+      "loss": 49.7393,
+      "step": 18540
+    },
+    {
+      "epoch": 0.07494434725695609,
+      "grad_norm": 287.8588562011719,
+      "learning_rate": 9.924344593564038e-06,
+      "loss": 49.0368,
+      "step": 18550
+    },
+    {
+      "epoch": 0.07498474852232372,
+      "grad_norm": 753.5078125,
+      "learning_rate": 9.924099979735754e-06,
+      "loss": 33.4634,
+      "step": 18560
+    },
+    {
+      "epoch": 0.07502514978769136,
+      "grad_norm": 688.8632202148438,
+      "learning_rate": 9.923854974119434e-06,
+      "loss": 69.1773,
+      "step": 18570
+    },
+    {
+      "epoch": 0.07506555105305898,
+      "grad_norm": 571.031005859375,
+      "learning_rate": 9.92360957673457e-06,
+      "loss": 54.5468,
+      "step": 18580
+    },
+    {
+      "epoch": 0.07510595231842661,
+      "grad_norm": 305.73028564453125,
+      "learning_rate": 9.923363787600688e-06,
+      "loss": 34.359,
+      "step": 18590
+    },
+    {
+      "epoch": 0.07514635358379425,
+      "grad_norm": 647.7716674804688,
+      "learning_rate": 9.923117606737347e-06,
+      "loss": 84.9632,
+      "step": 18600
+    },
+    {
+      "epoch": 0.07518675484916187,
+      "grad_norm": 385.5935363769531,
+      "learning_rate": 9.92287103416413e-06,
+      "loss": 53.0548,
+      "step": 18610
+    },
+    {
+      "epoch": 0.0752271561145295,
+      "grad_norm": 567.6202392578125,
+      "learning_rate": 9.922624069900658e-06,
+      "loss": 43.9075,
+      "step": 18620
+    },
+    {
+      "epoch": 0.07526755737989714,
+      "grad_norm": 758.9103393554688,
+      "learning_rate": 9.922376713966581e-06,
+      "loss": 47.5629,
+      "step": 18630
+    },
+    {
+      "epoch": 0.07530795864526477,
+      "grad_norm": 764.8362426757812,
+      "learning_rate": 9.92212896638158e-06,
+      "loss": 42.1043,
+      "step": 18640
+    },
+    {
+      "epoch": 0.0753483599106324,
+      "grad_norm": 551.5079345703125,
+      "learning_rate": 9.921880827165367e-06,
+      "loss": 38.0223,
+      "step": 18650
+    },
+    {
+      "epoch": 0.07538876117600003,
+      "grad_norm": 1067.0855712890625,
+      "learning_rate": 9.921632296337683e-06,
+      "loss": 73.4578,
+      "step": 18660
+    },
+    {
+      "epoch": 0.07542916244136766,
+      "grad_norm": 715.7173461914062,
+      "learning_rate": 9.921383373918305e-06,
+      "loss": 60.8801,
+      "step": 18670
+    },
+    {
+      "epoch": 0.0754695637067353,
+      "grad_norm": 515.8720092773438,
+      "learning_rate": 9.92113405992704e-06,
+      "loss": 43.01,
+      "step": 18680
+    },
+    {
+      "epoch": 0.07550996497210292,
+      "grad_norm": 863.5546264648438,
+      "learning_rate": 9.92088435438372e-06,
+      "loss": 42.6416,
+      "step": 18690
+    },
+    {
+      "epoch": 0.07555036623747055,
+      "grad_norm": 549.4111938476562,
+      "learning_rate": 9.920634257308217e-06,
+      "loss": 59.0651,
+      "step": 18700
+    },
+    {
+      "epoch": 0.07559076750283819,
+      "grad_norm": 493.9446105957031,
+      "learning_rate": 9.920383768720429e-06,
+      "loss": 58.4141,
+      "step": 18710
+    },
+    {
+      "epoch": 0.07563116876820582,
+      "grad_norm": 807.5221557617188,
+      "learning_rate": 9.920132888640286e-06,
+      "loss": 72.1741,
+      "step": 18720
+    },
+    {
+      "epoch": 0.07567157003357346,
+      "grad_norm": 319.9321594238281,
+      "learning_rate": 9.91988161708775e-06,
+      "loss": 58.0552,
+      "step": 18730
+    },
+    {
+      "epoch": 0.07571197129894108,
+      "grad_norm": 585.4686279296875,
+      "learning_rate": 9.919629954082813e-06,
+      "loss": 47.1483,
+      "step": 18740
+    },
+    {
+      "epoch": 0.07575237256430871,
+      "grad_norm": 393.1877136230469,
+      "learning_rate": 9.919377899645497e-06,
+      "loss": 38.655,
+      "step": 18750
+    },
+    {
+      "epoch": 0.07579277382967635,
+      "grad_norm": 889.9515380859375,
+      "learning_rate": 9.91912545379586e-06,
+      "loss": 42.3758,
+      "step": 18760
+    },
+    {
+      "epoch": 0.07583317509504398,
+      "grad_norm": 468.6612548828125,
+      "learning_rate": 9.918872616553986e-06,
+      "loss": 45.4651,
+      "step": 18770
+    },
+    {
+      "epoch": 0.0758735763604116,
+      "grad_norm": 976.6599731445312,
+      "learning_rate": 9.918619387939991e-06,
+      "loss": 40.7768,
+      "step": 18780
+    },
+    {
+      "epoch": 0.07591397762577924,
+      "grad_norm": 862.0733032226562,
+      "learning_rate": 9.918365767974025e-06,
+      "loss": 34.8569,
+      "step": 18790
+    },
+    {
+      "epoch": 0.07595437889114687,
+      "grad_norm": 2882.824951171875,
+      "learning_rate": 9.91811175667627e-06,
+      "loss": 62.0223,
+      "step": 18800
+    },
+    {
+      "epoch": 0.07599478015651451,
+      "grad_norm": 709.4342651367188,
+      "learning_rate": 9.91785735406693e-06,
+      "loss": 58.1651,
+      "step": 18810
+    },
+    {
+      "epoch": 0.07603518142188213,
+      "grad_norm": 626.4570922851562,
+      "learning_rate": 9.917602560166253e-06,
+      "loss": 55.8888,
+      "step": 18820
+    },
+    {
+      "epoch": 0.07607558268724976,
+      "grad_norm": 685.3681030273438,
+      "learning_rate": 9.917347374994507e-06,
+      "loss": 54.2592,
+      "step": 18830
+    },
+    {
+      "epoch": 0.0761159839526174,
+      "grad_norm": 559.7869873046875,
+      "learning_rate": 9.917091798571998e-06,
+      "loss": 40.9733,
+      "step": 18840
+    },
+    {
+      "epoch": 0.07615638521798503,
+      "grad_norm": 249.24891662597656,
+      "learning_rate": 9.916835830919062e-06,
+      "loss": 51.8981,
+      "step": 18850
+    },
+    {
+      "epoch": 0.07619678648335265,
+      "grad_norm": 913.8907470703125,
+      "learning_rate": 9.916579472056064e-06,
+      "loss": 61.8862,
+      "step": 18860
+    },
+    {
+      "epoch": 0.07623718774872029,
+      "grad_norm": 861.2070922851562,
+      "learning_rate": 9.916322722003402e-06,
+      "loss": 52.7893,
+      "step": 18870
+    },
+    {
+      "epoch": 0.07627758901408792,
+      "grad_norm": 541.4371337890625,
+      "learning_rate": 9.916065580781504e-06,
+      "loss": 44.1705,
+      "step": 18880
+    },
+    {
+      "epoch": 0.07631799027945556,
+      "grad_norm": 567.7680053710938,
+      "learning_rate": 9.91580804841083e-06,
+      "loss": 49.3585,
+      "step": 18890
+    },
+    {
+      "epoch": 0.07635839154482318,
+      "grad_norm": 402.49945068359375,
+      "learning_rate": 9.915550124911866e-06,
+      "loss": 56.762,
+      "step": 18900
+    },
+    {
+      "epoch": 0.07639879281019081,
+      "grad_norm": 649.5604248046875,
+      "learning_rate": 9.915291810305141e-06,
+      "loss": 40.8407,
+      "step": 18910
+    },
+    {
+      "epoch": 0.07643919407555845,
+      "grad_norm": 680.6880493164062,
+      "learning_rate": 9.915033104611204e-06,
+      "loss": 80.9502,
+      "step": 18920
+    },
+    {
+      "epoch": 0.07647959534092608,
+      "grad_norm": 558.8545532226562,
+      "learning_rate": 9.914774007850641e-06,
+      "loss": 44.2588,
+      "step": 18930
+    },
+    {
+      "epoch": 0.0765199966062937,
+      "grad_norm": 451.441162109375,
+      "learning_rate": 9.914514520044065e-06,
+      "loss": 36.8717,
+      "step": 18940
+    },
+    {
+      "epoch": 0.07656039787166134,
+      "grad_norm": 356.0220947265625,
+      "learning_rate": 9.914254641212124e-06,
+      "loss": 52.1456,
+      "step": 18950
+    },
+    {
+      "epoch": 0.07660079913702897,
+      "grad_norm": 519.8460693359375,
+      "learning_rate": 9.913994371375494e-06,
+      "loss": 66.9399,
+      "step": 18960
+    },
+    {
+      "epoch": 0.07664120040239661,
+      "grad_norm": 247.09814453125,
+      "learning_rate": 9.913733710554886e-06,
+      "loss": 42.8186,
+      "step": 18970
+    },
+    {
+      "epoch": 0.07668160166776423,
+      "grad_norm": 1293.5992431640625,
+      "learning_rate": 9.913472658771034e-06,
+      "loss": 55.2946,
+      "step": 18980
+    },
+    {
+      "epoch": 0.07672200293313186,
+      "grad_norm": 826.9439086914062,
+      "learning_rate": 9.913211216044715e-06,
+      "loss": 64.7492,
+      "step": 18990
+    },
+    {
+      "epoch": 0.0767624041984995,
+      "grad_norm": 1064.3778076171875,
+      "learning_rate": 9.912949382396728e-06,
+      "loss": 80.7947,
+      "step": 19000
+    },
+    {
+      "epoch": 0.07680280546386713,
+      "grad_norm": 521.4830322265625,
+      "learning_rate": 9.912687157847905e-06,
+      "loss": 39.9748,
+      "step": 19010
+    },
+    {
+      "epoch": 0.07684320672923475,
+      "grad_norm": 837.2169799804688,
+      "learning_rate": 9.91242454241911e-06,
+      "loss": 52.3863,
+      "step": 19020
+    },
+    {
+      "epoch": 0.0768836079946024,
+      "grad_norm": 650.2366333007812,
+      "learning_rate": 9.912161536131242e-06,
+      "loss": 47.731,
+      "step": 19030
+    },
+    {
+      "epoch": 0.07692400925997002,
+      "grad_norm": 887.7250366210938,
+      "learning_rate": 9.911898139005222e-06,
+      "loss": 62.2921,
+      "step": 19040
+    },
+    {
+      "epoch": 0.07696441052533766,
+      "grad_norm": 911.5108032226562,
+      "learning_rate": 9.91163435106201e-06,
+      "loss": 71.3875,
+      "step": 19050
+    },
+    {
+      "epoch": 0.07700481179070529,
+      "grad_norm": 821.865478515625,
+      "learning_rate": 9.911370172322595e-06,
+      "loss": 70.9358,
+      "step": 19060
+    },
+    {
+      "epoch": 0.07704521305607291,
+      "grad_norm": 741.1298217773438,
+      "learning_rate": 9.911105602807996e-06,
+      "loss": 48.0397,
+      "step": 19070
+    },
+    {
+      "epoch": 0.07708561432144055,
+      "grad_norm": 423.426025390625,
+      "learning_rate": 9.910840642539261e-06,
+      "loss": 49.4986,
+      "step": 19080
+    },
+    {
+      "epoch": 0.07712601558680818,
+      "grad_norm": 509.99359130859375,
+      "learning_rate": 9.910575291537476e-06,
+      "loss": 60.3422,
+      "step": 19090
+    },
+    {
+      "epoch": 0.0771664168521758,
+      "grad_norm": 507.8599548339844,
+      "learning_rate": 9.91030954982375e-06,
+      "loss": 47.7275,
+      "step": 19100
+    },
+    {
+      "epoch": 0.07720681811754344,
+      "grad_norm": 711.5784912109375,
+      "learning_rate": 9.910043417419228e-06,
+      "loss": 52.5352,
+      "step": 19110
+    },
+    {
+      "epoch": 0.07724721938291107,
+      "grad_norm": 597.524169921875,
+      "learning_rate": 9.909776894345086e-06,
+      "loss": 43.659,
+      "step": 19120
+    },
+    {
+      "epoch": 0.07728762064827871,
+      "grad_norm": 603.5820922851562,
+      "learning_rate": 9.909509980622532e-06,
+      "loss": 60.1314,
+      "step": 19130
+    },
+    {
+      "epoch": 0.07732802191364634,
+      "grad_norm": 570.4426879882812,
+      "learning_rate": 9.909242676272797e-06,
+      "loss": 50.9336,
+      "step": 19140
+    },
+    {
+      "epoch": 0.07736842317901396,
+      "grad_norm": 450.529296875,
+      "learning_rate": 9.908974981317155e-06,
+      "loss": 32.5167,
+      "step": 19150
+    },
+    {
+      "epoch": 0.0774088244443816,
+      "grad_norm": 423.9222106933594,
+      "learning_rate": 9.9087068957769e-06,
+      "loss": 33.1907,
+      "step": 19160
+    },
+    {
+      "epoch": 0.07744922570974923,
+      "grad_norm": 509.5530700683594,
+      "learning_rate": 9.908438419673367e-06,
+      "loss": 25.4662,
+      "step": 19170
+    },
+    {
+      "epoch": 0.07748962697511685,
+      "grad_norm": 1166.0137939453125,
+      "learning_rate": 9.908169553027916e-06,
+      "loss": 57.6205,
+      "step": 19180
+    },
+    {
+      "epoch": 0.0775300282404845,
+      "grad_norm": 660.1497802734375,
+      "learning_rate": 9.90790029586194e-06,
+      "loss": 85.3185,
+      "step": 19190
+    },
+    {
+      "epoch": 0.07757042950585212,
+      "grad_norm": 768.8164672851562,
+      "learning_rate": 9.907630648196857e-06,
+      "loss": 43.815,
+      "step": 19200
+    },
+    {
+      "epoch": 0.07761083077121976,
+      "grad_norm": 623.1567993164062,
+      "learning_rate": 9.907360610054132e-06,
+      "loss": 65.0855,
+      "step": 19210
+    },
+    {
+      "epoch": 0.07765123203658739,
+      "grad_norm": 430.18414306640625,
+      "learning_rate": 9.907090181455241e-06,
+      "loss": 65.7018,
+      "step": 19220
+    },
+    {
+      "epoch": 0.07769163330195501,
+      "grad_norm": 758.6072387695312,
+      "learning_rate": 9.906819362421707e-06,
+      "loss": 44.9825,
+      "step": 19230
+    },
+    {
+      "epoch": 0.07773203456732265,
+      "grad_norm": 684.4268188476562,
+      "learning_rate": 9.906548152975076e-06,
+      "loss": 65.7489,
+      "step": 19240
+    },
+    {
+      "epoch": 0.07777243583269028,
+      "grad_norm": 976.61767578125,
+      "learning_rate": 9.906276553136924e-06,
+      "loss": 55.8617,
+      "step": 19250
+    },
+    {
+      "epoch": 0.0778128370980579,
+      "grad_norm": 886.5833129882812,
+      "learning_rate": 9.906004562928865e-06,
+      "loss": 70.1672,
+      "step": 19260
+    },
+    {
+      "epoch": 0.07785323836342554,
+      "grad_norm": 1017.2815551757812,
+      "learning_rate": 9.905732182372538e-06,
+      "loss": 48.9207,
+      "step": 19270
+    },
+    {
+      "epoch": 0.07789363962879317,
+      "grad_norm": 805.864013671875,
+      "learning_rate": 9.905459411489617e-06,
+      "loss": 43.7626,
+      "step": 19280
+    },
+    {
+      "epoch": 0.07793404089416081,
+      "grad_norm": 319.4452209472656,
+      "learning_rate": 9.905186250301802e-06,
+      "loss": 27.7071,
+      "step": 19290
+    },
+    {
+      "epoch": 0.07797444215952844,
+      "grad_norm": 660.695068359375,
+      "learning_rate": 9.904912698830828e-06,
+      "loss": 55.5006,
+      "step": 19300
+    },
+    {
+      "epoch": 0.07801484342489606,
+      "grad_norm": 516.4996948242188,
+      "learning_rate": 9.904638757098464e-06,
+      "loss": 43.2097,
+      "step": 19310
+    },
+    {
+      "epoch": 0.0780552446902637,
+      "grad_norm": 2025.7215576171875,
+      "learning_rate": 9.9043644251265e-06,
+      "loss": 77.0012,
+      "step": 19320
+    },
+    {
+      "epoch": 0.07809564595563133,
+      "grad_norm": 510.74237060546875,
+      "learning_rate": 9.90408970293677e-06,
+      "loss": 50.4969,
+      "step": 19330
+    },
+    {
+      "epoch": 0.07813604722099896,
+      "grad_norm": 1480.844970703125,
+      "learning_rate": 9.903814590551127e-06,
+      "loss": 72.9755,
+      "step": 19340
+    },
+    {
+      "epoch": 0.0781764484863666,
+      "grad_norm": 656.4638671875,
+      "learning_rate": 9.903539087991462e-06,
+      "loss": 48.8749,
+      "step": 19350
+    },
+    {
+      "epoch": 0.07821684975173422,
+      "grad_norm": 558.57177734375,
+      "learning_rate": 9.903263195279698e-06,
+      "loss": 64.6075,
+      "step": 19360
+    },
+    {
+      "epoch": 0.07825725101710186,
+      "grad_norm": 626.6182861328125,
+      "learning_rate": 9.902986912437784e-06,
+      "loss": 44.9975,
+      "step": 19370
+    },
+    {
+      "epoch": 0.07829765228246949,
+      "grad_norm": 2065.308349609375,
+      "learning_rate": 9.902710239487702e-06,
+      "loss": 66.3444,
+      "step": 19380
+    },
+    {
+      "epoch": 0.07833805354783711,
+      "grad_norm": 1237.13134765625,
+      "learning_rate": 9.902433176451466e-06,
+      "loss": 68.9546,
+      "step": 19390
+    },
+    {
+      "epoch": 0.07837845481320475,
+      "grad_norm": 989.8242797851562,
+      "learning_rate": 9.902155723351124e-06,
+      "loss": 66.7895,
+      "step": 19400
+    },
+    {
+      "epoch": 0.07841885607857238,
+      "grad_norm": 771.996826171875,
+      "learning_rate": 9.901877880208747e-06,
+      "loss": 60.2761,
+      "step": 19410
+    },
+    {
+      "epoch": 0.07845925734394,
+      "grad_norm": 981.9373168945312,
+      "learning_rate": 9.901599647046443e-06,
+      "loss": 69.1419,
+      "step": 19420
+    },
+    {
+      "epoch": 0.07849965860930765,
+      "grad_norm": 709.3343505859375,
+      "learning_rate": 9.901321023886351e-06,
+      "loss": 65.5082,
+      "step": 19430
+    },
+    {
+      "epoch": 0.07854005987467527,
+      "grad_norm": 411.25970458984375,
+      "learning_rate": 9.901042010750641e-06,
+      "loss": 49.9508,
+      "step": 19440
+    },
+    {
+      "epoch": 0.07858046114004291,
+      "grad_norm": 709.1156616210938,
+      "learning_rate": 9.900762607661509e-06,
+      "loss": 50.5245,
+      "step": 19450
+    },
+    {
+      "epoch": 0.07862086240541054,
+      "grad_norm": 137.40362548828125,
+      "learning_rate": 9.900482814641188e-06,
+      "loss": 47.2816,
+      "step": 19460
+    },
+    {
+      "epoch": 0.07866126367077816,
+      "grad_norm": 792.5342407226562,
+      "learning_rate": 9.90020263171194e-06,
+      "loss": 65.7329,
+      "step": 19470
+    },
+    {
+      "epoch": 0.0787016649361458,
+      "grad_norm": 1262.7125244140625,
+      "learning_rate": 9.899922058896058e-06,
+      "loss": 34.923,
+      "step": 19480
+    },
+    {
+      "epoch": 0.07874206620151343,
+      "grad_norm": 1053.7841796875,
+      "learning_rate": 9.899641096215865e-06,
+      "loss": 67.22,
+      "step": 19490
+    },
+    {
+      "epoch": 0.07878246746688106,
+      "grad_norm": 619.6085205078125,
+      "learning_rate": 9.899359743693715e-06,
+      "loss": 43.8119,
+      "step": 19500
+    },
+    {
+      "epoch": 0.0788228687322487,
+      "grad_norm": 406.4720458984375,
+      "learning_rate": 9.899078001351996e-06,
+      "loss": 38.7725,
+      "step": 19510
+    },
+    {
+      "epoch": 0.07886326999761632,
+      "grad_norm": 821.3504638671875,
+      "learning_rate": 9.898795869213125e-06,
+      "loss": 50.2778,
+      "step": 19520
+    },
+    {
+      "epoch": 0.07890367126298396,
+      "grad_norm": 948.050048828125,
+      "learning_rate": 9.898513347299549e-06,
+      "loss": 55.1774,
+      "step": 19530
+    },
+    {
+      "epoch": 0.07894407252835159,
+      "grad_norm": 1106.9036865234375,
+      "learning_rate": 9.898230435633747e-06,
+      "loss": 51.0458,
+      "step": 19540
+    },
+    {
+      "epoch": 0.07898447379371921,
+      "grad_norm": 676.792236328125,
+      "learning_rate": 9.897947134238228e-06,
+      "loss": 53.7979,
+      "step": 19550
+    },
+    {
+      "epoch": 0.07902487505908685,
+      "grad_norm": 544.3801879882812,
+      "learning_rate": 9.897663443135534e-06,
+      "loss": 55.7435,
+      "step": 19560
+    },
+    {
+      "epoch": 0.07906527632445448,
+      "grad_norm": 374.0148010253906,
+      "learning_rate": 9.897379362348239e-06,
+      "loss": 49.8657,
+      "step": 19570
+    },
+    {
+      "epoch": 0.07910567758982211,
+      "grad_norm": 647.5410766601562,
+      "learning_rate": 9.897094891898942e-06,
+      "loss": 39.7925,
+      "step": 19580
+    },
+    {
+      "epoch": 0.07914607885518975,
+      "grad_norm": 770.840576171875,
+      "learning_rate": 9.89681003181028e-06,
+      "loss": 33.4735,
+      "step": 19590
+    },
+    {
+      "epoch": 0.07918648012055737,
+      "grad_norm": 732.4638671875,
+      "learning_rate": 9.896524782104917e-06,
+      "loss": 64.5618,
+      "step": 19600
+    },
+    {
+      "epoch": 0.07922688138592501,
+      "grad_norm": 487.14825439453125,
+      "learning_rate": 9.89623914280555e-06,
+      "loss": 45.1367,
+      "step": 19610
+    },
+    {
+      "epoch": 0.07926728265129264,
+      "grad_norm": 540.7517700195312,
+      "learning_rate": 9.895953113934904e-06,
+      "loss": 48.6443,
+      "step": 19620
+    },
+    {
+      "epoch": 0.07930768391666027,
+      "grad_norm": 639.5513916015625,
+      "learning_rate": 9.895666695515739e-06,
+      "loss": 55.7388,
+      "step": 19630
+    },
+    {
+      "epoch": 0.0793480851820279,
+      "grad_norm": 936.28857421875,
+      "learning_rate": 9.895379887570842e-06,
+      "loss": 55.8839,
+      "step": 19640
+    },
+    {
+      "epoch": 0.07938848644739553,
+      "grad_norm": 513.8567504882812,
+      "learning_rate": 9.895092690123036e-06,
+      "loss": 45.4252,
+      "step": 19650
+    },
+    {
+      "epoch": 0.07942888771276316,
+      "grad_norm": 274.3640441894531,
+      "learning_rate": 9.894805103195168e-06,
+      "loss": 47.4719,
+      "step": 19660
+    },
+    {
+      "epoch": 0.0794692889781308,
+      "grad_norm": 879.132568359375,
+      "learning_rate": 9.894517126810122e-06,
+      "loss": 57.8774,
+      "step": 19670
+    },
+    {
+      "epoch": 0.07950969024349842,
+      "grad_norm": 460.71124267578125,
+      "learning_rate": 9.894228760990811e-06,
+      "loss": 44.0254,
+      "step": 19680
+    },
+    {
+      "epoch": 0.07955009150886606,
+      "grad_norm": 655.6797485351562,
+      "learning_rate": 9.893940005760181e-06,
+      "loss": 67.7663,
+      "step": 19690
+    },
+    {
+      "epoch": 0.07959049277423369,
+      "grad_norm": 320.0049133300781,
+      "learning_rate": 9.893650861141204e-06,
+      "loss": 39.7949,
+      "step": 19700
+    },
+    {
+      "epoch": 0.07963089403960132,
+      "grad_norm": 940.2349853515625,
+      "learning_rate": 9.893361327156887e-06,
+      "loss": 50.6715,
+      "step": 19710
+    },
+    {
+      "epoch": 0.07967129530496896,
+      "grad_norm": 772.8973388671875,
+      "learning_rate": 9.893071403830265e-06,
+      "loss": 45.953,
+      "step": 19720
+    },
+    {
+      "epoch": 0.07971169657033658,
+      "grad_norm": 950.0260009765625,
+      "learning_rate": 9.892781091184409e-06,
+      "loss": 56.717,
+      "step": 19730
+    },
+    {
+      "epoch": 0.07975209783570421,
+      "grad_norm": 1299.931396484375,
+      "learning_rate": 9.892490389242417e-06,
+      "loss": 50.0108,
+      "step": 19740
+    },
+    {
+      "epoch": 0.07979249910107185,
+      "grad_norm": 764.660888671875,
+      "learning_rate": 9.892199298027416e-06,
+      "loss": 41.7661,
+      "step": 19750
+    },
+    {
+      "epoch": 0.07983290036643947,
+      "grad_norm": 2445.427490234375,
+      "learning_rate": 9.891907817562572e-06,
+      "loss": 59.8738,
+      "step": 19760
+    },
+    {
+      "epoch": 0.07987330163180711,
+      "grad_norm": 1094.4820556640625,
+      "learning_rate": 9.891615947871072e-06,
+      "loss": 80.5574,
+      "step": 19770
+    },
+    {
+      "epoch": 0.07991370289717474,
+      "grad_norm": 910.0176391601562,
+      "learning_rate": 9.89132368897614e-06,
+      "loss": 42.2061,
+      "step": 19780
+    },
+    {
+      "epoch": 0.07995410416254237,
+      "grad_norm": 740.8135986328125,
+      "learning_rate": 9.891031040901031e-06,
+      "loss": 53.9417,
+      "step": 19790
+    },
+    {
+      "epoch": 0.07999450542791,
+      "grad_norm": 852.3710327148438,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 45.9227,
+      "step": 19800
+    },
+    {
+      "epoch": 0.08003490669327763,
+      "grad_norm": 653.4926147460938,
+      "learning_rate": 9.890444577303448e-06,
+      "loss": 50.5123,
+      "step": 19810
+    },
+    {
+      "epoch": 0.08007530795864526,
+      "grad_norm": 474.41845703125,
+      "learning_rate": 9.890150761827639e-06,
+      "loss": 51.1544,
+      "step": 19820
+    },
+    {
+      "epoch": 0.0801157092240129,
+      "grad_norm": 1015.6337890625,
+      "learning_rate": 9.889856557264975e-06,
+      "loss": 79.7073,
+      "step": 19830
+    },
+    {
+      "epoch": 0.08015611048938052,
+      "grad_norm": 645.1337890625,
+      "learning_rate": 9.889561963638866e-06,
+      "loss": 69.1597,
+      "step": 19840
+    },
+    {
+      "epoch": 0.08019651175474816,
+      "grad_norm": 784.7030029296875,
+      "learning_rate": 9.889266980972752e-06,
+      "loss": 73.7766,
+      "step": 19850
+    },
+    {
+      "epoch": 0.08023691302011579,
+      "grad_norm": 779.0071411132812,
+      "learning_rate": 9.888971609290103e-06,
+      "loss": 34.1661,
+      "step": 19860
+    },
+    {
+      "epoch": 0.08027731428548342,
+      "grad_norm": 639.2666015625,
+      "learning_rate": 9.88867584861442e-06,
+      "loss": 40.5408,
+      "step": 19870
+    },
+    {
+      "epoch": 0.08031771555085106,
+      "grad_norm": 564.05224609375,
+      "learning_rate": 9.888379698969236e-06,
+      "loss": 70.8067,
+      "step": 19880
+    },
+    {
+      "epoch": 0.08035811681621868,
+      "grad_norm": 371.7476806640625,
+      "learning_rate": 9.888083160378114e-06,
+      "loss": 59.5962,
+      "step": 19890
+    },
+    {
+      "epoch": 0.08039851808158631,
+      "grad_norm": 625.6640014648438,
+      "learning_rate": 9.887786232864648e-06,
+      "loss": 40.9848,
+      "step": 19900
+    },
+    {
+      "epoch": 0.08043891934695395,
+      "grad_norm": 1052.5758056640625,
+      "learning_rate": 9.887488916452463e-06,
+      "loss": 77.8972,
+      "step": 19910
+    },
+    {
+      "epoch": 0.08047932061232158,
+      "grad_norm": 565.8422241210938,
+      "learning_rate": 9.887191211165217e-06,
+      "loss": 44.9603,
+      "step": 19920
+    },
+    {
+      "epoch": 0.08051972187768922,
+      "grad_norm": 747.18603515625,
+      "learning_rate": 9.886893117026593e-06,
+      "loss": 38.7456,
+      "step": 19930
+    },
+    {
+      "epoch": 0.08056012314305684,
+      "grad_norm": 567.5298461914062,
+      "learning_rate": 9.886594634060314e-06,
+      "loss": 44.1234,
+      "step": 19940
+    },
+    {
+      "epoch": 0.08060052440842447,
+      "grad_norm": 441.0209655761719,
+      "learning_rate": 9.886295762290125e-06,
+      "loss": 39.1584,
+      "step": 19950
+    },
+    {
+      "epoch": 0.08064092567379211,
+      "grad_norm": 574.0647583007812,
+      "learning_rate": 9.885996501739808e-06,
+      "loss": 75.4105,
+      "step": 19960
+    },
+    {
+      "epoch": 0.08068132693915973,
+      "grad_norm": 1946.658447265625,
+      "learning_rate": 9.885696852433174e-06,
+      "loss": 96.2204,
+      "step": 19970
+    },
+    {
+      "epoch": 0.08072172820452736,
+      "grad_norm": 719.7885131835938,
+      "learning_rate": 9.885396814394062e-06,
+      "loss": 50.1303,
+      "step": 19980
+    },
+    {
+      "epoch": 0.080762129469895,
+      "grad_norm": 526.1619262695312,
+      "learning_rate": 9.885096387646346e-06,
+      "loss": 42.3297,
+      "step": 19990
+    },
+    {
+      "epoch": 0.08080253073526263,
+      "grad_norm": 761.8712768554688,
+      "learning_rate": 9.88479557221393e-06,
+      "loss": 60.0437,
+      "step": 20000
+    },
+    {
+      "epoch": 0.08084293200063027,
+      "grad_norm": 653.3501586914062,
+      "learning_rate": 9.88449436812075e-06,
+      "loss": 55.2075,
+      "step": 20010
+    },
+    {
+      "epoch": 0.08088333326599789,
+      "grad_norm": 293.372314453125,
+      "learning_rate": 9.88419277539077e-06,
+      "loss": 46.4932,
+      "step": 20020
+    },
+    {
+      "epoch": 0.08092373453136552,
+      "grad_norm": 561.39208984375,
+      "learning_rate": 9.883890794047985e-06,
+      "loss": 64.5671,
+      "step": 20030
+    },
+    {
+      "epoch": 0.08096413579673316,
+      "grad_norm": 581.0809936523438,
+      "learning_rate": 9.883588424116424e-06,
+      "loss": 66.5158,
+      "step": 20040
+    },
+    {
+      "epoch": 0.08100453706210078,
+      "grad_norm": 554.6759033203125,
+      "learning_rate": 9.883285665620145e-06,
+      "loss": 59.3087,
+      "step": 20050
+    },
+    {
+      "epoch": 0.08104493832746841,
+      "grad_norm": 387.6012268066406,
+      "learning_rate": 9.882982518583238e-06,
+      "loss": 56.412,
+      "step": 20060
+    },
+    {
+      "epoch": 0.08108533959283605,
+      "grad_norm": 1148.7908935546875,
+      "learning_rate": 9.882678983029819e-06,
+      "loss": 82.5292,
+      "step": 20070
+    },
+    {
+      "epoch": 0.08112574085820368,
+      "grad_norm": 892.7615356445312,
+      "learning_rate": 9.882375058984044e-06,
+      "loss": 65.469,
+      "step": 20080
+    },
+    {
+      "epoch": 0.08116614212357132,
+      "grad_norm": 790.3026123046875,
+      "learning_rate": 9.882070746470092e-06,
+      "loss": 51.3404,
+      "step": 20090
+    },
+    {
+      "epoch": 0.08120654338893894,
+      "grad_norm": 743.3460083007812,
+      "learning_rate": 9.881766045512176e-06,
+      "loss": 50.7929,
+      "step": 20100
+    },
+    {
+      "epoch": 0.08124694465430657,
+      "grad_norm": 916.11474609375,
+      "learning_rate": 9.88146095613454e-06,
+      "loss": 52.6558,
+      "step": 20110
+    },
+    {
+      "epoch": 0.08128734591967421,
+      "grad_norm": 452.7943420410156,
+      "learning_rate": 9.881155478361459e-06,
+      "loss": 37.1106,
+      "step": 20120
+    },
+    {
+      "epoch": 0.08132774718504183,
+      "grad_norm": 318.2351379394531,
+      "learning_rate": 9.880849612217238e-06,
+      "loss": 65.855,
+      "step": 20130
+    },
+    {
+      "epoch": 0.08136814845040946,
+      "grad_norm": 1544.2833251953125,
+      "learning_rate": 9.880543357726214e-06,
+      "loss": 66.3424,
+      "step": 20140
+    },
+    {
+      "epoch": 0.0814085497157771,
+      "grad_norm": 201.214111328125,
+      "learning_rate": 9.880236714912754e-06,
+      "loss": 64.8169,
+      "step": 20150
+    },
+    {
+      "epoch": 0.08144895098114473,
+      "grad_norm": 290.1680908203125,
+      "learning_rate": 9.879929683801254e-06,
+      "loss": 46.0704,
+      "step": 20160
+    },
+    {
+      "epoch": 0.08148935224651237,
+      "grad_norm": 661.8184814453125,
+      "learning_rate": 9.879622264416147e-06,
+      "loss": 46.8915,
+      "step": 20170
+    },
+    {
+      "epoch": 0.08152975351188,
+      "grad_norm": 481.2290954589844,
+      "learning_rate": 9.87931445678189e-06,
+      "loss": 50.5384,
+      "step": 20180
+    },
+    {
+      "epoch": 0.08157015477724762,
+      "grad_norm": 854.9109497070312,
+      "learning_rate": 9.879006260922975e-06,
+      "loss": 47.7714,
+      "step": 20190
+    },
+    {
+      "epoch": 0.08161055604261526,
+      "grad_norm": 550.2037353515625,
+      "learning_rate": 9.878697676863922e-06,
+      "loss": 56.8208,
+      "step": 20200
+    },
+    {
+      "epoch": 0.08165095730798289,
+      "grad_norm": 560.0890502929688,
+      "learning_rate": 9.878388704629286e-06,
+      "loss": 43.5966,
+      "step": 20210
+    },
+    {
+      "epoch": 0.08169135857335051,
+      "grad_norm": 470.9778137207031,
+      "learning_rate": 9.87807934424365e-06,
+      "loss": 72.7084,
+      "step": 20220
+    },
+    {
+      "epoch": 0.08173175983871815,
+      "grad_norm": 1365.6583251953125,
+      "learning_rate": 9.877769595731629e-06,
+      "loss": 73.7771,
+      "step": 20230
+    },
+    {
+      "epoch": 0.08177216110408578,
+      "grad_norm": 0.0,
+      "learning_rate": 9.877459459117864e-06,
+      "loss": 60.936,
+      "step": 20240
+    },
+    {
+      "epoch": 0.08181256236945342,
+      "grad_norm": 881.4310302734375,
+      "learning_rate": 9.877148934427037e-06,
+      "loss": 44.2945,
+      "step": 20250
+    },
+    {
+      "epoch": 0.08185296363482104,
+      "grad_norm": 779.4917602539062,
+      "learning_rate": 9.87683802168385e-06,
+      "loss": 57.1076,
+      "step": 20260
+    },
+    {
+      "epoch": 0.08189336490018867,
+      "grad_norm": 602.3546142578125,
+      "learning_rate": 9.876526720913045e-06,
+      "loss": 51.2614,
+      "step": 20270
+    },
+    {
+      "epoch": 0.08193376616555631,
+      "grad_norm": 657.62890625,
+      "learning_rate": 9.87621503213939e-06,
+      "loss": 65.6103,
+      "step": 20280
+    },
+    {
+      "epoch": 0.08197416743092394,
+      "grad_norm": 487.8164978027344,
+      "learning_rate": 9.875902955387682e-06,
+      "loss": 47.4494,
+      "step": 20290
+    },
+    {
+      "epoch": 0.08201456869629156,
+      "grad_norm": 679.0338745117188,
+      "learning_rate": 9.875590490682754e-06,
+      "loss": 41.2857,
+      "step": 20300
+    },
+    {
+      "epoch": 0.0820549699616592,
+      "grad_norm": 840.4469604492188,
+      "learning_rate": 9.875277638049466e-06,
+      "loss": 71.3646,
+      "step": 20310
+    },
+    {
+      "epoch": 0.08209537122702683,
+      "grad_norm": 874.2022705078125,
+      "learning_rate": 9.87496439751271e-06,
+      "loss": 40.4883,
+      "step": 20320
+    },
+    {
+      "epoch": 0.08213577249239447,
+      "grad_norm": 795.24853515625,
+      "learning_rate": 9.87465076909741e-06,
+      "loss": 60.0892,
+      "step": 20330
+    },
+    {
+      "epoch": 0.0821761737577621,
+      "grad_norm": 517.8695678710938,
+      "learning_rate": 9.874336752828523e-06,
+      "loss": 40.2634,
+      "step": 20340
+    },
+    {
+      "epoch": 0.08221657502312972,
+      "grad_norm": 2106.507080078125,
+      "learning_rate": 9.87402234873103e-06,
+      "loss": 57.6708,
+      "step": 20350
+    },
+    {
+      "epoch": 0.08225697628849736,
+      "grad_norm": 570.333251953125,
+      "learning_rate": 9.873707556829945e-06,
+      "loss": 45.0013,
+      "step": 20360
+    },
+    {
+      "epoch": 0.08229737755386499,
+      "grad_norm": 633.8534545898438,
+      "learning_rate": 9.873392377150318e-06,
+      "loss": 73.9094,
+      "step": 20370
+    },
+    {
+      "epoch": 0.08233777881923261,
+      "grad_norm": 831.1759643554688,
+      "learning_rate": 9.873076809717226e-06,
+      "loss": 43.2788,
+      "step": 20380
+    },
+    {
+      "epoch": 0.08237818008460025,
+      "grad_norm": 975.9166870117188,
+      "learning_rate": 9.872760854555776e-06,
+      "loss": 57.6913,
+      "step": 20390
+    },
+    {
+      "epoch": 0.08241858134996788,
+      "grad_norm": 687.0908203125,
+      "learning_rate": 9.872444511691108e-06,
+      "loss": 54.1361,
+      "step": 20400
+    },
+    {
+      "epoch": 0.08245898261533552,
+      "grad_norm": 812.077880859375,
+      "learning_rate": 9.872127781148392e-06,
+      "loss": 46.2579,
+      "step": 20410
+    },
+    {
+      "epoch": 0.08249938388070314,
+      "grad_norm": 1039.45947265625,
+      "learning_rate": 9.871810662952828e-06,
+      "loss": 60.9721,
+      "step": 20420
+    },
+    {
+      "epoch": 0.08253978514607077,
+      "grad_norm": 350.81365966796875,
+      "learning_rate": 9.87149315712965e-06,
+      "loss": 54.0255,
+      "step": 20430
+    },
+    {
+      "epoch": 0.08258018641143841,
+      "grad_norm": 1040.960205078125,
+      "learning_rate": 9.871175263704116e-06,
+      "loss": 47.519,
+      "step": 20440
+    },
+    {
+      "epoch": 0.08262058767680604,
+      "grad_norm": 903.642578125,
+      "learning_rate": 9.870856982701522e-06,
+      "loss": 42.4366,
+      "step": 20450
+    },
+    {
+      "epoch": 0.08266098894217366,
+      "grad_norm": 268.2828674316406,
+      "learning_rate": 9.870538314147194e-06,
+      "loss": 65.6971,
+      "step": 20460
+    },
+    {
+      "epoch": 0.0827013902075413,
+      "grad_norm": 728.4773559570312,
+      "learning_rate": 9.870219258066485e-06,
+      "loss": 49.7082,
+      "step": 20470
+    },
+    {
+      "epoch": 0.08274179147290893,
+      "grad_norm": 579.0664672851562,
+      "learning_rate": 9.86989981448478e-06,
+      "loss": 44.6814,
+      "step": 20480
+    },
+    {
+      "epoch": 0.08278219273827657,
+      "grad_norm": 944.3462524414062,
+      "learning_rate": 9.869579983427497e-06,
+      "loss": 68.3373,
+      "step": 20490
+    },
+    {
+      "epoch": 0.0828225940036442,
+      "grad_norm": 1739.9251708984375,
+      "learning_rate": 9.869259764920081e-06,
+      "loss": 61.5346,
+      "step": 20500
+    },
+    {
+      "epoch": 0.08286299526901182,
+      "grad_norm": 323.2250061035156,
+      "learning_rate": 9.868939158988016e-06,
+      "loss": 66.8795,
+      "step": 20510
+    },
+    {
+      "epoch": 0.08290339653437946,
+      "grad_norm": 610.7361450195312,
+      "learning_rate": 9.868618165656805e-06,
+      "loss": 44.2779,
+      "step": 20520
+    },
+    {
+      "epoch": 0.08294379779974709,
+      "grad_norm": 766.6193237304688,
+      "learning_rate": 9.868296784951992e-06,
+      "loss": 68.555,
+      "step": 20530
+    },
+    {
+      "epoch": 0.08298419906511471,
+      "grad_norm": 251.73240661621094,
+      "learning_rate": 9.867975016899145e-06,
+      "loss": 51.9702,
+      "step": 20540
+    },
+    {
+      "epoch": 0.08302460033048235,
+      "grad_norm": 753.54345703125,
+      "learning_rate": 9.867652861523866e-06,
+      "loss": 48.5123,
+      "step": 20550
+    },
+    {
+      "epoch": 0.08306500159584998,
+      "grad_norm": 655.2874145507812,
+      "learning_rate": 9.86733031885179e-06,
+      "loss": 56.5569,
+      "step": 20560
+    },
+    {
+      "epoch": 0.08310540286121762,
+      "grad_norm": 675.6414184570312,
+      "learning_rate": 9.867007388908579e-06,
+      "loss": 29.6619,
+      "step": 20570
+    },
+    {
+      "epoch": 0.08314580412658525,
+      "grad_norm": 1051.6103515625,
+      "learning_rate": 9.866684071719926e-06,
+      "loss": 71.0548,
+      "step": 20580
+    },
+    {
+      "epoch": 0.08318620539195287,
+      "grad_norm": 482.6436767578125,
+      "learning_rate": 9.866360367311557e-06,
+      "loss": 53.7929,
+      "step": 20590
+    },
+    {
+      "epoch": 0.08322660665732051,
+      "grad_norm": 714.1856689453125,
+      "learning_rate": 9.866036275709226e-06,
+      "loss": 81.5517,
+      "step": 20600
+    },
+    {
+      "epoch": 0.08326700792268814,
+      "grad_norm": 697.500244140625,
+      "learning_rate": 9.86571179693872e-06,
+      "loss": 92.779,
+      "step": 20610
+    },
+    {
+      "epoch": 0.08330740918805576,
+      "grad_norm": 888.16455078125,
+      "learning_rate": 9.865386931025858e-06,
+      "loss": 70.6248,
+      "step": 20620
+    },
+    {
+      "epoch": 0.0833478104534234,
+      "grad_norm": 1028.4215087890625,
+      "learning_rate": 9.865061677996487e-06,
+      "loss": 59.6395,
+      "step": 20630
+    },
+    {
+      "epoch": 0.08338821171879103,
+      "grad_norm": 930.8424072265625,
+      "learning_rate": 9.864736037876487e-06,
+      "loss": 47.4096,
+      "step": 20640
+    },
+    {
+      "epoch": 0.08342861298415867,
+      "grad_norm": 644.8805541992188,
+      "learning_rate": 9.864410010691766e-06,
+      "loss": 41.8741,
+      "step": 20650
+    },
+    {
+      "epoch": 0.0834690142495263,
+      "grad_norm": 1114.886474609375,
+      "learning_rate": 9.864083596468263e-06,
+      "loss": 70.8659,
+      "step": 20660
+    },
+    {
+      "epoch": 0.08350941551489392,
+      "grad_norm": 840.799072265625,
+      "learning_rate": 9.863756795231953e-06,
+      "loss": 73.1389,
+      "step": 20670
+    },
+    {
+      "epoch": 0.08354981678026156,
+      "grad_norm": 654.5409545898438,
+      "learning_rate": 9.863429607008837e-06,
+      "loss": 47.1239,
+      "step": 20680
+    },
+    {
+      "epoch": 0.08359021804562919,
+      "grad_norm": 833.7312622070312,
+      "learning_rate": 9.863102031824946e-06,
+      "loss": 60.6132,
+      "step": 20690
+    },
+    {
+      "epoch": 0.08363061931099681,
+      "grad_norm": 735.7879028320312,
+      "learning_rate": 9.862774069706346e-06,
+      "loss": 62.9169,
+      "step": 20700
+    },
+    {
+      "epoch": 0.08367102057636445,
+      "grad_norm": 428.26934814453125,
+      "learning_rate": 9.86244572067913e-06,
+      "loss": 50.549,
+      "step": 20710
+    },
+    {
+      "epoch": 0.08371142184173208,
+      "grad_norm": 561.2461547851562,
+      "learning_rate": 9.862116984769424e-06,
+      "loss": 56.5235,
+      "step": 20720
+    },
+    {
+      "epoch": 0.08375182310709972,
+      "grad_norm": 728.88525390625,
+      "learning_rate": 9.861787862003384e-06,
+      "loss": 70.6504,
+      "step": 20730
+    },
+    {
+      "epoch": 0.08379222437246735,
+      "grad_norm": 2484.103271484375,
+      "learning_rate": 9.861458352407196e-06,
+      "loss": 63.5527,
+      "step": 20740
+    },
+    {
+      "epoch": 0.08383262563783497,
+      "grad_norm": 1101.0321044921875,
+      "learning_rate": 9.861128456007076e-06,
+      "loss": 29.5614,
+      "step": 20750
+    },
+    {
+      "epoch": 0.08387302690320261,
+      "grad_norm": 739.0653076171875,
+      "learning_rate": 9.860798172829277e-06,
+      "loss": 60.2193,
+      "step": 20760
+    },
+    {
+      "epoch": 0.08391342816857024,
+      "grad_norm": 771.9661865234375,
+      "learning_rate": 9.860467502900076e-06,
+      "loss": 41.4003,
+      "step": 20770
+    },
+    {
+      "epoch": 0.08395382943393787,
+      "grad_norm": 412.1797790527344,
+      "learning_rate": 9.860136446245779e-06,
+      "loss": 47.9847,
+      "step": 20780
+    },
+    {
+      "epoch": 0.0839942306993055,
+      "grad_norm": 547.4696044921875,
+      "learning_rate": 9.859805002892733e-06,
+      "loss": 60.2853,
+      "step": 20790
+    },
+    {
+      "epoch": 0.08403463196467313,
+      "grad_norm": 2845.0302734375,
+      "learning_rate": 9.859473172867304e-06,
+      "loss": 90.4173,
+      "step": 20800
+    },
+    {
+      "epoch": 0.08407503323004077,
+      "grad_norm": 758.51171875,
+      "learning_rate": 9.859140956195898e-06,
+      "loss": 50.2199,
+      "step": 20810
+    },
+    {
+      "epoch": 0.0841154344954084,
+      "grad_norm": 598.7674560546875,
+      "learning_rate": 9.858808352904946e-06,
+      "loss": 56.4163,
+      "step": 20820
+    },
+    {
+      "epoch": 0.08415583576077602,
+      "grad_norm": 595.4998168945312,
+      "learning_rate": 9.858475363020913e-06,
+      "loss": 50.6997,
+      "step": 20830
+    },
+    {
+      "epoch": 0.08419623702614366,
+      "grad_norm": 992.6197509765625,
+      "learning_rate": 9.858141986570294e-06,
+      "loss": 60.4768,
+      "step": 20840
+    },
+    {
+      "epoch": 0.08423663829151129,
+      "grad_norm": 451.0296325683594,
+      "learning_rate": 9.85780822357961e-06,
+      "loss": 34.9621,
+      "step": 20850
+    },
+    {
+      "epoch": 0.08427703955687892,
+      "grad_norm": 731.828857421875,
+      "learning_rate": 9.857474074075422e-06,
+      "loss": 40.2803,
+      "step": 20860
+    },
+    {
+      "epoch": 0.08431744082224656,
+      "grad_norm": 371.8586730957031,
+      "learning_rate": 9.857139538084313e-06,
+      "loss": 45.1926,
+      "step": 20870
+    },
+    {
+      "epoch": 0.08435784208761418,
+      "grad_norm": 822.1978149414062,
+      "learning_rate": 9.856804615632904e-06,
+      "loss": 62.567,
+      "step": 20880
+    },
+    {
+      "epoch": 0.08439824335298182,
+      "grad_norm": 592.7335205078125,
+      "learning_rate": 9.85646930674784e-06,
+      "loss": 47.0134,
+      "step": 20890
+    },
+    {
+      "epoch": 0.08443864461834945,
+      "grad_norm": 1128.1734619140625,
+      "learning_rate": 9.856133611455802e-06,
+      "loss": 58.2488,
+      "step": 20900
+    },
+    {
+      "epoch": 0.08447904588371707,
+      "grad_norm": 974.8399658203125,
+      "learning_rate": 9.855797529783499e-06,
+      "loss": 73.3392,
+      "step": 20910
+    },
+    {
+      "epoch": 0.08451944714908471,
+      "grad_norm": 734.9163818359375,
+      "learning_rate": 9.855461061757673e-06,
+      "loss": 55.1066,
+      "step": 20920
+    },
+    {
+      "epoch": 0.08455984841445234,
+      "grad_norm": 572.8345336914062,
+      "learning_rate": 9.855124207405093e-06,
+      "loss": 47.7488,
+      "step": 20930
+    },
+    {
+      "epoch": 0.08460024967981997,
+      "grad_norm": 506.1241149902344,
+      "learning_rate": 9.854786966752561e-06,
+      "loss": 49.7169,
+      "step": 20940
+    },
+    {
+      "epoch": 0.0846406509451876,
+      "grad_norm": 776.8680419921875,
+      "learning_rate": 9.854449339826912e-06,
+      "loss": 59.9915,
+      "step": 20950
+    },
+    {
+      "epoch": 0.08468105221055523,
+      "grad_norm": 680.7813110351562,
+      "learning_rate": 9.854111326655006e-06,
+      "loss": 34.6824,
+      "step": 20960
+    },
+    {
+      "epoch": 0.08472145347592287,
+      "grad_norm": 672.206787109375,
+      "learning_rate": 9.85377292726374e-06,
+      "loss": 62.063,
+      "step": 20970
+    },
+    {
+      "epoch": 0.0847618547412905,
+      "grad_norm": 1285.8265380859375,
+      "learning_rate": 9.85343414168004e-06,
+      "loss": 40.8714,
+      "step": 20980
+    },
+    {
+      "epoch": 0.08480225600665812,
+      "grad_norm": 404.3912353515625,
+      "learning_rate": 9.853094969930857e-06,
+      "loss": 55.0305,
+      "step": 20990
+    },
+    {
+      "epoch": 0.08484265727202576,
+      "grad_norm": 620.1910400390625,
+      "learning_rate": 9.85275541204318e-06,
+      "loss": 46.7199,
+      "step": 21000
+    },
+    {
+      "epoch": 0.08488305853739339,
+      "grad_norm": 4275.12841796875,
+      "learning_rate": 9.852415468044027e-06,
+      "loss": 65.9217,
+      "step": 21010
+    },
+    {
+      "epoch": 0.08492345980276102,
+      "grad_norm": 619.6917114257812,
+      "learning_rate": 9.852075137960446e-06,
+      "loss": 59.9521,
+      "step": 21020
+    },
+    {
+      "epoch": 0.08496386106812866,
+      "grad_norm": 161.13560485839844,
+      "learning_rate": 9.851734421819511e-06,
+      "loss": 46.9062,
+      "step": 21030
+    },
+    {
+      "epoch": 0.08500426233349628,
+      "grad_norm": 729.6267700195312,
+      "learning_rate": 9.851393319648338e-06,
+      "loss": 45.645,
+      "step": 21040
+    },
+    {
+      "epoch": 0.08504466359886392,
+      "grad_norm": 1013.0626220703125,
+      "learning_rate": 9.851051831474062e-06,
+      "loss": 53.9054,
+      "step": 21050
+    },
+    {
+      "epoch": 0.08508506486423155,
+      "grad_norm": 613.8585205078125,
+      "learning_rate": 9.850709957323855e-06,
+      "loss": 36.5241,
+      "step": 21060
+    },
+    {
+      "epoch": 0.08512546612959918,
+      "grad_norm": 339.6336364746094,
+      "learning_rate": 9.85036769722492e-06,
+      "loss": 62.6872,
+      "step": 21070
+    },
+    {
+      "epoch": 0.08516586739496682,
+      "grad_norm": 334.4331359863281,
+      "learning_rate": 9.850025051204484e-06,
+      "loss": 48.3355,
+      "step": 21080
+    },
+    {
+      "epoch": 0.08520626866033444,
+      "grad_norm": 987.1373901367188,
+      "learning_rate": 9.849682019289816e-06,
+      "loss": 55.5567,
+      "step": 21090
+    },
+    {
+      "epoch": 0.08524666992570207,
+      "grad_norm": 737.2757568359375,
+      "learning_rate": 9.849338601508204e-06,
+      "loss": 49.2449,
+      "step": 21100
+    },
+    {
+      "epoch": 0.08528707119106971,
+      "grad_norm": 780.7937622070312,
+      "learning_rate": 9.848994797886978e-06,
+      "loss": 51.7602,
+      "step": 21110
+    },
+    {
+      "epoch": 0.08532747245643733,
+      "grad_norm": 505.43365478515625,
+      "learning_rate": 9.84865060845349e-06,
+      "loss": 48.2497,
+      "step": 21120
+    },
+    {
+      "epoch": 0.08536787372180497,
+      "grad_norm": 416.4381103515625,
+      "learning_rate": 9.848306033235123e-06,
+      "loss": 49.8744,
+      "step": 21130
+    },
+    {
+      "epoch": 0.0854082749871726,
+      "grad_norm": 1224.279541015625,
+      "learning_rate": 9.847961072259298e-06,
+      "loss": 51.9309,
+      "step": 21140
+    },
+    {
+      "epoch": 0.08544867625254023,
+      "grad_norm": 855.0914306640625,
+      "learning_rate": 9.847615725553457e-06,
+      "loss": 45.4371,
+      "step": 21150
+    },
+    {
+      "epoch": 0.08548907751790787,
+      "grad_norm": 518.2168579101562,
+      "learning_rate": 9.847269993145082e-06,
+      "loss": 50.2552,
+      "step": 21160
+    },
+    {
+      "epoch": 0.08552947878327549,
+      "grad_norm": 387.5550231933594,
+      "learning_rate": 9.84692387506168e-06,
+      "loss": 51.4854,
+      "step": 21170
+    },
+    {
+      "epoch": 0.08556988004864312,
+      "grad_norm": 325.9864501953125,
+      "learning_rate": 9.846577371330788e-06,
+      "loss": 48.9065,
+      "step": 21180
+    },
+    {
+      "epoch": 0.08561028131401076,
+      "grad_norm": 643.303466796875,
+      "learning_rate": 9.846230481979978e-06,
+      "loss": 61.2373,
+      "step": 21190
+    },
+    {
+      "epoch": 0.08565068257937838,
+      "grad_norm": 1041.485107421875,
+      "learning_rate": 9.84588320703685e-06,
+      "loss": 52.6828,
+      "step": 21200
+    },
+    {
+      "epoch": 0.08569108384474602,
+      "grad_norm": 945.1188354492188,
+      "learning_rate": 9.845535546529036e-06,
+      "loss": 59.1794,
+      "step": 21210
+    },
+    {
+      "epoch": 0.08573148511011365,
+      "grad_norm": 1870.1568603515625,
+      "learning_rate": 9.845187500484194e-06,
+      "loss": 79.922,
+      "step": 21220
+    },
+    {
+      "epoch": 0.08577188637548128,
+      "grad_norm": 1248.2161865234375,
+      "learning_rate": 9.844839068930021e-06,
+      "loss": 58.7054,
+      "step": 21230
+    },
+    {
+      "epoch": 0.08581228764084892,
+      "grad_norm": 326.83563232421875,
+      "learning_rate": 9.844490251894237e-06,
+      "loss": 46.6798,
+      "step": 21240
+    },
+    {
+      "epoch": 0.08585268890621654,
+      "grad_norm": 860.9239501953125,
+      "learning_rate": 9.844141049404598e-06,
+      "loss": 50.0194,
+      "step": 21250
+    },
+    {
+      "epoch": 0.08589309017158417,
+      "grad_norm": 877.5747680664062,
+      "learning_rate": 9.843791461488887e-06,
+      "loss": 50.6901,
+      "step": 21260
+    },
+    {
+      "epoch": 0.08593349143695181,
+      "grad_norm": 942.8595581054688,
+      "learning_rate": 9.843441488174918e-06,
+      "loss": 49.5507,
+      "step": 21270
+    },
+    {
+      "epoch": 0.08597389270231943,
+      "grad_norm": 1001.177734375,
+      "learning_rate": 9.843091129490539e-06,
+      "loss": 65.9635,
+      "step": 21280
+    },
+    {
+      "epoch": 0.08601429396768706,
+      "grad_norm": 583.5938110351562,
+      "learning_rate": 9.842740385463628e-06,
+      "loss": 69.3124,
+      "step": 21290
+    },
+    {
+      "epoch": 0.0860546952330547,
+      "grad_norm": 808.9459838867188,
+      "learning_rate": 9.842389256122086e-06,
+      "loss": 42.665,
+      "step": 21300
+    },
+    {
+      "epoch": 0.08609509649842233,
+      "grad_norm": 1406.5587158203125,
+      "learning_rate": 9.842037741493856e-06,
+      "loss": 49.0096,
+      "step": 21310
+    },
+    {
+      "epoch": 0.08613549776378997,
+      "grad_norm": 591.5014038085938,
+      "learning_rate": 9.841685841606905e-06,
+      "loss": 48.2358,
+      "step": 21320
+    },
+    {
+      "epoch": 0.0861758990291576,
+      "grad_norm": 405.5210876464844,
+      "learning_rate": 9.841333556489232e-06,
+      "loss": 42.8229,
+      "step": 21330
+    },
+    {
+      "epoch": 0.08621630029452522,
+      "grad_norm": 945.532470703125,
+      "learning_rate": 9.840980886168866e-06,
+      "loss": 55.2916,
+      "step": 21340
+    },
+    {
+      "epoch": 0.08625670155989286,
+      "grad_norm": 670.8292236328125,
+      "learning_rate": 9.840627830673867e-06,
+      "loss": 61.8779,
+      "step": 21350
+    },
+    {
+      "epoch": 0.08629710282526049,
+      "grad_norm": 547.29345703125,
+      "learning_rate": 9.84027439003233e-06,
+      "loss": 37.4086,
+      "step": 21360
+    },
+    {
+      "epoch": 0.08633750409062811,
+      "grad_norm": 583.8782958984375,
+      "learning_rate": 9.839920564272372e-06,
+      "loss": 38.6236,
+      "step": 21370
+    },
+    {
+      "epoch": 0.08637790535599575,
+      "grad_norm": 1493.264404296875,
+      "learning_rate": 9.839566353422148e-06,
+      "loss": 52.7415,
+      "step": 21380
+    },
+    {
+      "epoch": 0.08641830662136338,
+      "grad_norm": 1489.428466796875,
+      "learning_rate": 9.839211757509838e-06,
+      "loss": 44.0712,
+      "step": 21390
+    },
+    {
+      "epoch": 0.08645870788673102,
+      "grad_norm": 1181.785400390625,
+      "learning_rate": 9.83885677656366e-06,
+      "loss": 79.1896,
+      "step": 21400
+    },
+    {
+      "epoch": 0.08649910915209864,
+      "grad_norm": 346.3993225097656,
+      "learning_rate": 9.838501410611852e-06,
+      "loss": 66.8448,
+      "step": 21410
+    },
+    {
+      "epoch": 0.08653951041746627,
+      "grad_norm": 749.7047119140625,
+      "learning_rate": 9.838145659682695e-06,
+      "loss": 46.5892,
+      "step": 21420
+    },
+    {
+      "epoch": 0.08657991168283391,
+      "grad_norm": 698.231201171875,
+      "learning_rate": 9.837789523804491e-06,
+      "loss": 46.5881,
+      "step": 21430
+    },
+    {
+      "epoch": 0.08662031294820154,
+      "grad_norm": 788.7677612304688,
+      "learning_rate": 9.837433003005578e-06,
+      "loss": 73.2478,
+      "step": 21440
+    },
+    {
+      "epoch": 0.08666071421356916,
+      "grad_norm": 494.36346435546875,
+      "learning_rate": 9.83707609731432e-06,
+      "loss": 43.4364,
+      "step": 21450
+    },
+    {
+      "epoch": 0.0867011154789368,
+      "grad_norm": 1333.8416748046875,
+      "learning_rate": 9.836718806759119e-06,
+      "loss": 59.773,
+      "step": 21460
+    },
+    {
+      "epoch": 0.08674151674430443,
+      "grad_norm": 10082.6943359375,
+      "learning_rate": 9.836361131368398e-06,
+      "loss": 79.0009,
+      "step": 21470
+    },
+    {
+      "epoch": 0.08678191800967207,
+      "grad_norm": 478.3102722167969,
+      "learning_rate": 9.836003071170617e-06,
+      "loss": 30.4878,
+      "step": 21480
+    },
+    {
+      "epoch": 0.0868223192750397,
+      "grad_norm": 1929.5330810546875,
+      "learning_rate": 9.835644626194268e-06,
+      "loss": 86.2987,
+      "step": 21490
+    },
+    {
+      "epoch": 0.08686272054040732,
+      "grad_norm": 860.5169677734375,
+      "learning_rate": 9.835285796467867e-06,
+      "loss": 46.1074,
+      "step": 21500
+    },
+    {
+      "epoch": 0.08690312180577496,
+      "grad_norm": 410.758056640625,
+      "learning_rate": 9.834926582019968e-06,
+      "loss": 51.9143,
+      "step": 21510
+    },
+    {
+      "epoch": 0.08694352307114259,
+      "grad_norm": 547.5929565429688,
+      "learning_rate": 9.834566982879149e-06,
+      "loss": 43.2907,
+      "step": 21520
+    },
+    {
+      "epoch": 0.08698392433651021,
+      "grad_norm": 878.74560546875,
+      "learning_rate": 9.83420699907402e-06,
+      "loss": 57.618,
+      "step": 21530
+    },
+    {
+      "epoch": 0.08702432560187785,
+      "grad_norm": 1142.8790283203125,
+      "learning_rate": 9.83384663063323e-06,
+      "loss": 43.2136,
+      "step": 21540
+    },
+    {
+      "epoch": 0.08706472686724548,
+      "grad_norm": 483.8853454589844,
+      "learning_rate": 9.833485877585447e-06,
+      "loss": 59.4731,
+      "step": 21550
+    },
+    {
+      "epoch": 0.08710512813261312,
+      "grad_norm": 716.4909057617188,
+      "learning_rate": 9.833124739959375e-06,
+      "loss": 54.2022,
+      "step": 21560
+    },
+    {
+      "epoch": 0.08714552939798074,
+      "grad_norm": 863.1256713867188,
+      "learning_rate": 9.83276321778375e-06,
+      "loss": 29.776,
+      "step": 21570
+    },
+    {
+      "epoch": 0.08718593066334837,
+      "grad_norm": 319.7196350097656,
+      "learning_rate": 9.832401311087334e-06,
+      "loss": 71.2109,
+      "step": 21580
+    },
+    {
+      "epoch": 0.08722633192871601,
+      "grad_norm": 491.4449462890625,
+      "learning_rate": 9.832039019898922e-06,
+      "loss": 42.9584,
+      "step": 21590
+    },
+    {
+      "epoch": 0.08726673319408364,
+      "grad_norm": 962.9393310546875,
+      "learning_rate": 9.831676344247343e-06,
+      "loss": 68.2823,
+      "step": 21600
+    },
+    {
+      "epoch": 0.08730713445945126,
+      "grad_norm": 371.64996337890625,
+      "learning_rate": 9.831313284161452e-06,
+      "loss": 44.8427,
+      "step": 21610
+    },
+    {
+      "epoch": 0.0873475357248189,
+      "grad_norm": 706.4290161132812,
+      "learning_rate": 9.830949839670134e-06,
+      "loss": 61.4159,
+      "step": 21620
+    },
+    {
+      "epoch": 0.08738793699018653,
+      "grad_norm": 251.71263122558594,
+      "learning_rate": 9.83058601080231e-06,
+      "loss": 54.2862,
+      "step": 21630
+    },
+    {
+      "epoch": 0.08742833825555417,
+      "grad_norm": 636.603759765625,
+      "learning_rate": 9.830221797586925e-06,
+      "loss": 40.2427,
+      "step": 21640
+    },
+    {
+      "epoch": 0.0874687395209218,
+      "grad_norm": 1012.7183837890625,
+      "learning_rate": 9.829857200052961e-06,
+      "loss": 46.5919,
+      "step": 21650
+    },
+    {
+      "epoch": 0.08750914078628942,
+      "grad_norm": 537.0764770507812,
+      "learning_rate": 9.829492218229426e-06,
+      "loss": 48.8234,
+      "step": 21660
+    },
+    {
+      "epoch": 0.08754954205165706,
+      "grad_norm": 236.4057159423828,
+      "learning_rate": 9.829126852145357e-06,
+      "loss": 56.6592,
+      "step": 21670
+    },
+    {
+      "epoch": 0.08758994331702469,
+      "grad_norm": 500.1606140136719,
+      "learning_rate": 9.82876110182983e-06,
+      "loss": 41.0559,
+      "step": 21680
+    },
+    {
+      "epoch": 0.08763034458239231,
+      "grad_norm": 633.2708740234375,
+      "learning_rate": 9.82839496731194e-06,
+      "loss": 49.9588,
+      "step": 21690
+    },
+    {
+      "epoch": 0.08767074584775995,
+      "grad_norm": 324.40264892578125,
+      "learning_rate": 9.828028448620824e-06,
+      "loss": 56.6327,
+      "step": 21700
+    },
+    {
+      "epoch": 0.08771114711312758,
+      "grad_norm": 1261.2022705078125,
+      "learning_rate": 9.827661545785641e-06,
+      "loss": 57.7509,
+      "step": 21710
+    },
+    {
+      "epoch": 0.08775154837849522,
+      "grad_norm": 438.5520324707031,
+      "learning_rate": 9.827294258835584e-06,
+      "loss": 53.7419,
+      "step": 21720
+    },
+    {
+      "epoch": 0.08779194964386285,
+      "grad_norm": 446.0879211425781,
+      "learning_rate": 9.82692658779988e-06,
+      "loss": 39.9361,
+      "step": 21730
+    },
+    {
+      "epoch": 0.08783235090923047,
+      "grad_norm": 1023.3678588867188,
+      "learning_rate": 9.826558532707777e-06,
+      "loss": 37.8205,
+      "step": 21740
+    },
+    {
+      "epoch": 0.08787275217459811,
+      "grad_norm": 331.857666015625,
+      "learning_rate": 9.826190093588564e-06,
+      "loss": 43.6651,
+      "step": 21750
+    },
+    {
+      "epoch": 0.08791315343996574,
+      "grad_norm": 654.5843505859375,
+      "learning_rate": 9.825821270471555e-06,
+      "loss": 43.0031,
+      "step": 21760
+    },
+    {
+      "epoch": 0.08795355470533336,
+      "grad_norm": 286.42913818359375,
+      "learning_rate": 9.825452063386094e-06,
+      "loss": 41.2635,
+      "step": 21770
+    },
+    {
+      "epoch": 0.087993955970701,
+      "grad_norm": 411.1146240234375,
+      "learning_rate": 9.825082472361558e-06,
+      "loss": 57.5872,
+      "step": 21780
+    },
+    {
+      "epoch": 0.08803435723606863,
+      "grad_norm": 326.46392822265625,
+      "learning_rate": 9.824712497427354e-06,
+      "loss": 31.2261,
+      "step": 21790
+    },
+    {
+      "epoch": 0.08807475850143627,
+      "grad_norm": 820.8731079101562,
+      "learning_rate": 9.824342138612918e-06,
+      "loss": 55.781,
+      "step": 21800
+    },
+    {
+      "epoch": 0.0881151597668039,
+      "grad_norm": 576.8391723632812,
+      "learning_rate": 9.823971395947723e-06,
+      "loss": 42.2346,
+      "step": 21810
+    },
+    {
+      "epoch": 0.08815556103217152,
+      "grad_norm": 830.1489868164062,
+      "learning_rate": 9.823600269461259e-06,
+      "loss": 31.6791,
+      "step": 21820
+    },
+    {
+      "epoch": 0.08819596229753916,
+      "grad_norm": 1346.4920654296875,
+      "learning_rate": 9.823228759183058e-06,
+      "loss": 57.8082,
+      "step": 21830
+    },
+    {
+      "epoch": 0.08823636356290679,
+      "grad_norm": 488.5693054199219,
+      "learning_rate": 9.822856865142683e-06,
+      "loss": 56.2668,
+      "step": 21840
+    },
+    {
+      "epoch": 0.08827676482827441,
+      "grad_norm": 417.81512451171875,
+      "learning_rate": 9.822484587369721e-06,
+      "loss": 35.235,
+      "step": 21850
+    },
+    {
+      "epoch": 0.08831716609364205,
+      "grad_norm": 837.1052856445312,
+      "learning_rate": 9.822111925893792e-06,
+      "loss": 57.6994,
+      "step": 21860
+    },
+    {
+      "epoch": 0.08835756735900968,
+      "grad_norm": 380.0332946777344,
+      "learning_rate": 9.821738880744549e-06,
+      "loss": 50.7504,
+      "step": 21870
+    },
+    {
+      "epoch": 0.08839796862437732,
+      "grad_norm": 821.6565551757812,
+      "learning_rate": 9.82136545195167e-06,
+      "loss": 46.5275,
+      "step": 21880
+    },
+    {
+      "epoch": 0.08843836988974495,
+      "grad_norm": 1301.0826416015625,
+      "learning_rate": 9.82099163954487e-06,
+      "loss": 41.9282,
+      "step": 21890
+    },
+    {
+      "epoch": 0.08847877115511257,
+      "grad_norm": 526.063232421875,
+      "learning_rate": 9.820617443553889e-06,
+      "loss": 57.149,
+      "step": 21900
+    },
+    {
+      "epoch": 0.08851917242048021,
+      "grad_norm": 743.5262451171875,
+      "learning_rate": 9.820242864008503e-06,
+      "loss": 61.0266,
+      "step": 21910
+    },
+    {
+      "epoch": 0.08855957368584784,
+      "grad_norm": 1047.2117919921875,
+      "learning_rate": 9.819867900938514e-06,
+      "loss": 39.4777,
+      "step": 21920
+    },
+    {
+      "epoch": 0.08859997495121547,
+      "grad_norm": 1059.58447265625,
+      "learning_rate": 9.819492554373758e-06,
+      "loss": 51.7157,
+      "step": 21930
+    },
+    {
+      "epoch": 0.0886403762165831,
+      "grad_norm": 539.9701538085938,
+      "learning_rate": 9.819116824344095e-06,
+      "loss": 71.8611,
+      "step": 21940
+    },
+    {
+      "epoch": 0.08868077748195073,
+      "grad_norm": 1279.98193359375,
+      "learning_rate": 9.818740710879424e-06,
+      "loss": 58.2858,
+      "step": 21950
+    },
+    {
+      "epoch": 0.08872117874731837,
+      "grad_norm": 972.816162109375,
+      "learning_rate": 9.81836421400967e-06,
+      "loss": 60.5523,
+      "step": 21960
+    },
+    {
+      "epoch": 0.088761580012686,
+      "grad_norm": 512.8277587890625,
+      "learning_rate": 9.81798733376479e-06,
+      "loss": 62.7751,
+      "step": 21970
+    },
+    {
+      "epoch": 0.08880198127805362,
+      "grad_norm": 531.3379516601562,
+      "learning_rate": 9.817610070174768e-06,
+      "loss": 38.6099,
+      "step": 21980
+    },
+    {
+      "epoch": 0.08884238254342126,
+      "grad_norm": 1373.3828125,
+      "learning_rate": 9.817232423269622e-06,
+      "loss": 55.1818,
+      "step": 21990
+    },
+    {
+      "epoch": 0.08888278380878889,
+      "grad_norm": 410.4630126953125,
+      "learning_rate": 9.816854393079402e-06,
+      "loss": 43.0107,
+      "step": 22000
+    },
+    {
+      "epoch": 0.08892318507415652,
+      "grad_norm": 502.2277526855469,
+      "learning_rate": 9.816475979634183e-06,
+      "loss": 52.2099,
+      "step": 22010
+    },
+    {
+      "epoch": 0.08896358633952416,
+      "grad_norm": 706.2998657226562,
+      "learning_rate": 9.816097182964076e-06,
+      "loss": 42.4107,
+      "step": 22020
+    },
+    {
+      "epoch": 0.08900398760489178,
+      "grad_norm": 916.6030883789062,
+      "learning_rate": 9.81571800309922e-06,
+      "loss": 59.517,
+      "step": 22030
+    },
+    {
+      "epoch": 0.08904438887025942,
+      "grad_norm": 514.5048828125,
+      "learning_rate": 9.815338440069782e-06,
+      "loss": 44.6423,
+      "step": 22040
+    },
+    {
+      "epoch": 0.08908479013562705,
+      "grad_norm": 859.5369262695312,
+      "learning_rate": 9.814958493905962e-06,
+      "loss": 46.8622,
+      "step": 22050
+    },
+    {
+      "epoch": 0.08912519140099467,
+      "grad_norm": 538.0739135742188,
+      "learning_rate": 9.814578164637996e-06,
+      "loss": 56.694,
+      "step": 22060
+    },
+    {
+      "epoch": 0.08916559266636231,
+      "grad_norm": 470.18182373046875,
+      "learning_rate": 9.81419745229614e-06,
+      "loss": 36.0551,
+      "step": 22070
+    },
+    {
+      "epoch": 0.08920599393172994,
+      "grad_norm": 1527.13720703125,
+      "learning_rate": 9.813816356910685e-06,
+      "loss": 46.0836,
+      "step": 22080
+    },
+    {
+      "epoch": 0.08924639519709757,
+      "grad_norm": 1404.5830078125,
+      "learning_rate": 9.813434878511956e-06,
+      "loss": 59.701,
+      "step": 22090
+    },
+    {
+      "epoch": 0.0892867964624652,
+      "grad_norm": 439.1456298828125,
+      "learning_rate": 9.813053017130305e-06,
+      "loss": 50.6273,
+      "step": 22100
+    },
+    {
+      "epoch": 0.08932719772783283,
+      "grad_norm": 3546.374267578125,
+      "learning_rate": 9.812670772796113e-06,
+      "loss": 63.3059,
+      "step": 22110
+    },
+    {
+      "epoch": 0.08936759899320047,
+      "grad_norm": 836.252197265625,
+      "learning_rate": 9.812288145539796e-06,
+      "loss": 47.2886,
+      "step": 22120
+    },
+    {
+      "epoch": 0.0894080002585681,
+      "grad_norm": 1181.7552490234375,
+      "learning_rate": 9.811905135391796e-06,
+      "loss": 54.5237,
+      "step": 22130
+    },
+    {
+      "epoch": 0.08944840152393572,
+      "grad_norm": 661.9832763671875,
+      "learning_rate": 9.81152174238259e-06,
+      "loss": 38.4662,
+      "step": 22140
+    },
+    {
+      "epoch": 0.08948880278930336,
+      "grad_norm": 883.0792846679688,
+      "learning_rate": 9.81113796654268e-06,
+      "loss": 57.6954,
+      "step": 22150
+    },
+    {
+      "epoch": 0.08952920405467099,
+      "grad_norm": 748.608642578125,
+      "learning_rate": 9.810753807902603e-06,
+      "loss": 36.5849,
+      "step": 22160
+    },
+    {
+      "epoch": 0.08956960532003862,
+      "grad_norm": 627.541259765625,
+      "learning_rate": 9.81036926649292e-06,
+      "loss": 53.9005,
+      "step": 22170
+    },
+    {
+      "epoch": 0.08961000658540626,
+      "grad_norm": 563.03662109375,
+      "learning_rate": 9.809984342344234e-06,
+      "loss": 41.9332,
+      "step": 22180
+    },
+    {
+      "epoch": 0.08965040785077388,
+      "grad_norm": 908.7540283203125,
+      "learning_rate": 9.80959903548717e-06,
+      "loss": 73.8175,
+      "step": 22190
+    },
+    {
+      "epoch": 0.08969080911614152,
+      "grad_norm": 650.3843994140625,
+      "learning_rate": 9.80921334595238e-06,
+      "loss": 59.4379,
+      "step": 22200
+    },
+    {
+      "epoch": 0.08973121038150915,
+      "grad_norm": 443.2868347167969,
+      "learning_rate": 9.808827273770558e-06,
+      "loss": 47.3863,
+      "step": 22210
+    },
+    {
+      "epoch": 0.08977161164687678,
+      "grad_norm": 411.5378112792969,
+      "learning_rate": 9.80844081897242e-06,
+      "loss": 82.0899,
+      "step": 22220
+    },
+    {
+      "epoch": 0.08981201291224442,
+      "grad_norm": 210.01197814941406,
+      "learning_rate": 9.808053981588712e-06,
+      "loss": 31.6324,
+      "step": 22230
+    },
+    {
+      "epoch": 0.08985241417761204,
+      "grad_norm": 722.1740112304688,
+      "learning_rate": 9.807666761650215e-06,
+      "loss": 69.5212,
+      "step": 22240
+    },
+    {
+      "epoch": 0.08989281544297967,
+      "grad_norm": 947.6121215820312,
+      "learning_rate": 9.80727915918774e-06,
+      "loss": 59.0137,
+      "step": 22250
+    },
+    {
+      "epoch": 0.08993321670834731,
+      "grad_norm": 697.5064086914062,
+      "learning_rate": 9.806891174232122e-06,
+      "loss": 49.3849,
+      "step": 22260
+    },
+    {
+      "epoch": 0.08997361797371493,
+      "grad_norm": 555.270263671875,
+      "learning_rate": 9.806502806814236e-06,
+      "loss": 44.8319,
+      "step": 22270
+    },
+    {
+      "epoch": 0.09001401923908257,
+      "grad_norm": 414.53692626953125,
+      "learning_rate": 9.806114056964977e-06,
+      "loss": 41.9408,
+      "step": 22280
+    },
+    {
+      "epoch": 0.0900544205044502,
+      "grad_norm": 324.5998840332031,
+      "learning_rate": 9.805724924715283e-06,
+      "loss": 65.8233,
+      "step": 22290
+    },
+    {
+      "epoch": 0.09009482176981783,
+      "grad_norm": 705.962646484375,
+      "learning_rate": 9.80533541009611e-06,
+      "loss": 64.0755,
+      "step": 22300
+    },
+    {
+      "epoch": 0.09013522303518547,
+      "grad_norm": 465.8612976074219,
+      "learning_rate": 9.804945513138454e-06,
+      "loss": 44.9652,
+      "step": 22310
+    },
+    {
+      "epoch": 0.09017562430055309,
+      "grad_norm": 360.5195617675781,
+      "learning_rate": 9.804555233873335e-06,
+      "loss": 64.9888,
+      "step": 22320
+    },
+    {
+      "epoch": 0.09021602556592072,
+      "grad_norm": 339.8669128417969,
+      "learning_rate": 9.804164572331804e-06,
+      "loss": 52.764,
+      "step": 22330
+    },
+    {
+      "epoch": 0.09025642683128836,
+      "grad_norm": 821.659423828125,
+      "learning_rate": 9.80377352854495e-06,
+      "loss": 51.951,
+      "step": 22340
+    },
+    {
+      "epoch": 0.09029682809665598,
+      "grad_norm": 384.43115234375,
+      "learning_rate": 9.80338210254388e-06,
+      "loss": 50.9012,
+      "step": 22350
+    },
+    {
+      "epoch": 0.09033722936202362,
+      "grad_norm": 520.3011474609375,
+      "learning_rate": 9.80299029435974e-06,
+      "loss": 53.0986,
+      "step": 22360
+    },
+    {
+      "epoch": 0.09037763062739125,
+      "grad_norm": 806.8153076171875,
+      "learning_rate": 9.802598104023706e-06,
+      "loss": 67.7043,
+      "step": 22370
+    },
+    {
+      "epoch": 0.09041803189275888,
+      "grad_norm": 443.0950927734375,
+      "learning_rate": 9.80220553156698e-06,
+      "loss": 42.6991,
+      "step": 22380
+    },
+    {
+      "epoch": 0.09045843315812652,
+      "grad_norm": 601.2862548828125,
+      "learning_rate": 9.801812577020802e-06,
+      "loss": 61.9289,
+      "step": 22390
+    },
+    {
+      "epoch": 0.09049883442349414,
+      "grad_norm": 611.3983764648438,
+      "learning_rate": 9.801419240416432e-06,
+      "loss": 43.7061,
+      "step": 22400
+    },
+    {
+      "epoch": 0.09053923568886177,
+      "grad_norm": 507.5821533203125,
+      "learning_rate": 9.80102552178517e-06,
+      "loss": 49.6816,
+      "step": 22410
+    },
+    {
+      "epoch": 0.09057963695422941,
+      "grad_norm": 262.9966735839844,
+      "learning_rate": 9.800631421158341e-06,
+      "loss": 39.2033,
+      "step": 22420
+    },
+    {
+      "epoch": 0.09062003821959703,
+      "grad_norm": 320.47772216796875,
+      "learning_rate": 9.800236938567302e-06,
+      "loss": 48.6889,
+      "step": 22430
+    },
+    {
+      "epoch": 0.09066043948496467,
+      "grad_norm": 678.9915771484375,
+      "learning_rate": 9.799842074043438e-06,
+      "loss": 48.05,
+      "step": 22440
+    },
+    {
+      "epoch": 0.0907008407503323,
+      "grad_norm": 647.835693359375,
+      "learning_rate": 9.799446827618172e-06,
+      "loss": 59.4923,
+      "step": 22450
+    },
+    {
+      "epoch": 0.09074124201569993,
+      "grad_norm": 1000.3504028320312,
+      "learning_rate": 9.799051199322944e-06,
+      "loss": 53.8829,
+      "step": 22460
+    },
+    {
+      "epoch": 0.09078164328106757,
+      "grad_norm": 543.1785278320312,
+      "learning_rate": 9.798655189189239e-06,
+      "loss": 53.5274,
+      "step": 22470
+    },
+    {
+      "epoch": 0.0908220445464352,
+      "grad_norm": 206.3665771484375,
+      "learning_rate": 9.798258797248563e-06,
+      "loss": 49.3775,
+      "step": 22480
+    },
+    {
+      "epoch": 0.09086244581180282,
+      "grad_norm": 499.0605773925781,
+      "learning_rate": 9.797862023532457e-06,
+      "loss": 53.9699,
+      "step": 22490
+    },
+    {
+      "epoch": 0.09090284707717046,
+      "grad_norm": 654.6769409179688,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 46.1002,
+      "step": 22500
+    },
+    {
+      "epoch": 0.09094324834253809,
+      "grad_norm": 800.8095703125,
+      "learning_rate": 9.797067330900256e-06,
+      "loss": 43.4725,
+      "step": 22510
+    },
+    {
+      "epoch": 0.09098364960790573,
+      "grad_norm": 574.1755981445312,
+      "learning_rate": 9.796669412047392e-06,
+      "loss": 44.0503,
+      "step": 22520
+    },
+    {
+      "epoch": 0.09102405087327335,
+      "grad_norm": 986.699462890625,
+      "learning_rate": 9.796271111545559e-06,
+      "loss": 43.2854,
+      "step": 22530
+    },
+    {
+      "epoch": 0.09106445213864098,
+      "grad_norm": 1443.898193359375,
+      "learning_rate": 9.795872429426443e-06,
+      "loss": 61.5884,
+      "step": 22540
+    },
+    {
+      "epoch": 0.09110485340400862,
+      "grad_norm": 523.7254638671875,
+      "learning_rate": 9.79547336572177e-06,
+      "loss": 40.8429,
+      "step": 22550
+    },
+    {
+      "epoch": 0.09114525466937624,
+      "grad_norm": 681.1943359375,
+      "learning_rate": 9.795073920463289e-06,
+      "loss": 68.2972,
+      "step": 22560
+    },
+    {
+      "epoch": 0.09118565593474387,
+      "grad_norm": 564.5675048828125,
+      "learning_rate": 9.794674093682781e-06,
+      "loss": 63.7268,
+      "step": 22570
+    },
+    {
+      "epoch": 0.09122605720011151,
+      "grad_norm": 379.54693603515625,
+      "learning_rate": 9.79427388541206e-06,
+      "loss": 57.5344,
+      "step": 22580
+    },
+    {
+      "epoch": 0.09126645846547914,
+      "grad_norm": 368.9071044921875,
+      "learning_rate": 9.79387329568297e-06,
+      "loss": 44.1385,
+      "step": 22590
+    },
+    {
+      "epoch": 0.09130685973084678,
+      "grad_norm": 672.96240234375,
+      "learning_rate": 9.793472324527383e-06,
+      "loss": 65.3485,
+      "step": 22600
+    },
+    {
+      "epoch": 0.0913472609962144,
+      "grad_norm": 553.6704711914062,
+      "learning_rate": 9.793070971977203e-06,
+      "loss": 51.0067,
+      "step": 22610
+    },
+    {
+      "epoch": 0.09138766226158203,
+      "grad_norm": 699.769287109375,
+      "learning_rate": 9.79266923806436e-06,
+      "loss": 56.594,
+      "step": 22620
+    },
+    {
+      "epoch": 0.09142806352694967,
+      "grad_norm": 565.623291015625,
+      "learning_rate": 9.792267122820823e-06,
+      "loss": 38.9469,
+      "step": 22630
+    },
+    {
+      "epoch": 0.0914684647923173,
+      "grad_norm": 319.3910217285156,
+      "learning_rate": 9.791864626278584e-06,
+      "loss": 57.7853,
+      "step": 22640
+    },
+    {
+      "epoch": 0.09150886605768492,
+      "grad_norm": 328.24212646484375,
+      "learning_rate": 9.791461748469669e-06,
+      "loss": 44.6982,
+      "step": 22650
+    },
+    {
+      "epoch": 0.09154926732305256,
+      "grad_norm": 479.53546142578125,
+      "learning_rate": 9.791058489426134e-06,
+      "loss": 51.2582,
+      "step": 22660
+    },
+    {
+      "epoch": 0.09158966858842019,
+      "grad_norm": 621.2388305664062,
+      "learning_rate": 9.790654849180059e-06,
+      "loss": 40.81,
+      "step": 22670
+    },
+    {
+      "epoch": 0.09163006985378783,
+      "grad_norm": 640.1640625,
+      "learning_rate": 9.790250827763566e-06,
+      "loss": 51.9004,
+      "step": 22680
+    },
+    {
+      "epoch": 0.09167047111915545,
+      "grad_norm": 623.7976684570312,
+      "learning_rate": 9.7898464252088e-06,
+      "loss": 40.9685,
+      "step": 22690
+    },
+    {
+      "epoch": 0.09171087238452308,
+      "grad_norm": 611.9249267578125,
+      "learning_rate": 9.789441641547935e-06,
+      "loss": 60.35,
+      "step": 22700
+    },
+    {
+      "epoch": 0.09175127364989072,
+      "grad_norm": 923.3186645507812,
+      "learning_rate": 9.789036476813178e-06,
+      "loss": 49.1127,
+      "step": 22710
+    },
+    {
+      "epoch": 0.09179167491525834,
+      "grad_norm": 357.7413635253906,
+      "learning_rate": 9.788630931036769e-06,
+      "loss": 44.903,
+      "step": 22720
+    },
+    {
+      "epoch": 0.09183207618062597,
+      "grad_norm": 742.0269775390625,
+      "learning_rate": 9.788225004250974e-06,
+      "loss": 63.4893,
+      "step": 22730
+    },
+    {
+      "epoch": 0.09187247744599361,
+      "grad_norm": 815.6424560546875,
+      "learning_rate": 9.78781869648809e-06,
+      "loss": 48.6374,
+      "step": 22740
+    },
+    {
+      "epoch": 0.09191287871136124,
+      "grad_norm": 599.6590576171875,
+      "learning_rate": 9.787412007780445e-06,
+      "loss": 32.8504,
+      "step": 22750
+    },
+    {
+      "epoch": 0.09195327997672888,
+      "grad_norm": 545.2987670898438,
+      "learning_rate": 9.787004938160398e-06,
+      "loss": 58.4387,
+      "step": 22760
+    },
+    {
+      "epoch": 0.0919936812420965,
+      "grad_norm": 643.601806640625,
+      "learning_rate": 9.786597487660336e-06,
+      "loss": 54.0817,
+      "step": 22770
+    },
+    {
+      "epoch": 0.09203408250746413,
+      "grad_norm": 852.6585693359375,
+      "learning_rate": 9.78618965631268e-06,
+      "loss": 53.0852,
+      "step": 22780
+    },
+    {
+      "epoch": 0.09207448377283177,
+      "grad_norm": 1306.4365234375,
+      "learning_rate": 9.785781444149883e-06,
+      "loss": 34.2664,
+      "step": 22790
+    },
+    {
+      "epoch": 0.0921148850381994,
+      "grad_norm": 850.7540893554688,
+      "learning_rate": 9.785372851204415e-06,
+      "loss": 52.8783,
+      "step": 22800
+    },
+    {
+      "epoch": 0.09215528630356702,
+      "grad_norm": 1039.580810546875,
+      "learning_rate": 9.784963877508794e-06,
+      "loss": 49.9097,
+      "step": 22810
+    },
+    {
+      "epoch": 0.09219568756893466,
+      "grad_norm": 544.6483764648438,
+      "learning_rate": 9.784554523095554e-06,
+      "loss": 49.3491,
+      "step": 22820
+    },
+    {
+      "epoch": 0.09223608883430229,
+      "grad_norm": 694.8217163085938,
+      "learning_rate": 9.784144787997272e-06,
+      "loss": 43.204,
+      "step": 22830
+    },
+    {
+      "epoch": 0.09227649009966993,
+      "grad_norm": 836.0192260742188,
+      "learning_rate": 9.783734672246545e-06,
+      "loss": 43.338,
+      "step": 22840
+    },
+    {
+      "epoch": 0.09231689136503755,
+      "grad_norm": 443.9679870605469,
+      "learning_rate": 9.783324175876004e-06,
+      "loss": 46.913,
+      "step": 22850
+    },
+    {
+      "epoch": 0.09235729263040518,
+      "grad_norm": 754.3963623046875,
+      "learning_rate": 9.782913298918311e-06,
+      "loss": 53.1269,
+      "step": 22860
+    },
+    {
+      "epoch": 0.09239769389577282,
+      "grad_norm": 625.8417358398438,
+      "learning_rate": 9.782502041406157e-06,
+      "loss": 43.964,
+      "step": 22870
+    },
+    {
+      "epoch": 0.09243809516114045,
+      "grad_norm": 405.47113037109375,
+      "learning_rate": 9.782090403372263e-06,
+      "loss": 37.2729,
+      "step": 22880
+    },
+    {
+      "epoch": 0.09247849642650807,
+      "grad_norm": 881.507568359375,
+      "learning_rate": 9.781678384849385e-06,
+      "loss": 58.9882,
+      "step": 22890
+    },
+    {
+      "epoch": 0.09251889769187571,
+      "grad_norm": 546.2739868164062,
+      "learning_rate": 9.7812659858703e-06,
+      "loss": 44.2559,
+      "step": 22900
+    },
+    {
+      "epoch": 0.09255929895724334,
+      "grad_norm": 428.82354736328125,
+      "learning_rate": 9.780853206467826e-06,
+      "loss": 59.4935,
+      "step": 22910
+    },
+    {
+      "epoch": 0.09259970022261098,
+      "grad_norm": 784.0477294921875,
+      "learning_rate": 9.780440046674803e-06,
+      "loss": 56.4658,
+      "step": 22920
+    },
+    {
+      "epoch": 0.0926401014879786,
+      "grad_norm": 428.0048828125,
+      "learning_rate": 9.780026506524106e-06,
+      "loss": 58.3504,
+      "step": 22930
+    },
+    {
+      "epoch": 0.09268050275334623,
+      "grad_norm": 1084.146728515625,
+      "learning_rate": 9.779612586048635e-06,
+      "loss": 54.8185,
+      "step": 22940
+    },
+    {
+      "epoch": 0.09272090401871387,
+      "grad_norm": 569.2861938476562,
+      "learning_rate": 9.779198285281326e-06,
+      "loss": 53.3557,
+      "step": 22950
+    },
+    {
+      "epoch": 0.0927613052840815,
+      "grad_norm": 486.71405029296875,
+      "learning_rate": 9.778783604255145e-06,
+      "loss": 48.3609,
+      "step": 22960
+    },
+    {
+      "epoch": 0.09280170654944912,
+      "grad_norm": 504.45513916015625,
+      "learning_rate": 9.778368543003083e-06,
+      "loss": 55.2413,
+      "step": 22970
+    },
+    {
+      "epoch": 0.09284210781481676,
+      "grad_norm": 940.0879516601562,
+      "learning_rate": 9.777953101558164e-06,
+      "loss": 53.3876,
+      "step": 22980
+    },
+    {
+      "epoch": 0.09288250908018439,
+      "grad_norm": 656.7711181640625,
+      "learning_rate": 9.777537279953448e-06,
+      "loss": 58.7076,
+      "step": 22990
+    },
+    {
+      "epoch": 0.09292291034555203,
+      "grad_norm": 744.0709838867188,
+      "learning_rate": 9.777121078222015e-06,
+      "loss": 45.4177,
+      "step": 23000
+    },
+    {
+      "epoch": 0.09296331161091966,
+      "grad_norm": 823.136474609375,
+      "learning_rate": 9.77670449639698e-06,
+      "loss": 47.3975,
+      "step": 23010
+    },
+    {
+      "epoch": 0.09300371287628728,
+      "grad_norm": 625.787353515625,
+      "learning_rate": 9.776287534511492e-06,
+      "loss": 31.936,
+      "step": 23020
+    },
+    {
+      "epoch": 0.09304411414165492,
+      "grad_norm": 562.3572387695312,
+      "learning_rate": 9.775870192598726e-06,
+      "loss": 68.5727,
+      "step": 23030
+    },
+    {
+      "epoch": 0.09308451540702255,
+      "grad_norm": 584.5991821289062,
+      "learning_rate": 9.775452470691886e-06,
+      "loss": 56.3662,
+      "step": 23040
+    },
+    {
+      "epoch": 0.09312491667239017,
+      "grad_norm": 476.9333801269531,
+      "learning_rate": 9.77503436882421e-06,
+      "loss": 121.5844,
+      "step": 23050
+    },
+    {
+      "epoch": 0.09316531793775781,
+      "grad_norm": 380.27105712890625,
+      "learning_rate": 9.774615887028964e-06,
+      "loss": 45.8855,
+      "step": 23060
+    },
+    {
+      "epoch": 0.09320571920312544,
+      "grad_norm": 820.0983276367188,
+      "learning_rate": 9.774197025339442e-06,
+      "loss": 53.6872,
+      "step": 23070
+    },
+    {
+      "epoch": 0.09324612046849308,
+      "grad_norm": 1121.6595458984375,
+      "learning_rate": 9.773777783788976e-06,
+      "loss": 65.0373,
+      "step": 23080
+    },
+    {
+      "epoch": 0.0932865217338607,
+      "grad_norm": 942.5689697265625,
+      "learning_rate": 9.77335816241092e-06,
+      "loss": 57.7903,
+      "step": 23090
+    },
+    {
+      "epoch": 0.09332692299922833,
+      "grad_norm": 469.630859375,
+      "learning_rate": 9.77293816123866e-06,
+      "loss": 44.2538,
+      "step": 23100
+    },
+    {
+      "epoch": 0.09336732426459597,
+      "grad_norm": 305.6590270996094,
+      "learning_rate": 9.772517780305618e-06,
+      "loss": 54.9539,
+      "step": 23110
+    },
+    {
+      "epoch": 0.0934077255299636,
+      "grad_norm": 469.95758056640625,
+      "learning_rate": 9.772097019645236e-06,
+      "loss": 39.7449,
+      "step": 23120
+    },
+    {
+      "epoch": 0.09344812679533122,
+      "grad_norm": 896.3937377929688,
+      "learning_rate": 9.771675879290998e-06,
+      "loss": 53.7042,
+      "step": 23130
+    },
+    {
+      "epoch": 0.09348852806069886,
+      "grad_norm": 543.0301513671875,
+      "learning_rate": 9.771254359276407e-06,
+      "loss": 44.7906,
+      "step": 23140
+    },
+    {
+      "epoch": 0.09352892932606649,
+      "grad_norm": 766.5215454101562,
+      "learning_rate": 9.770832459635004e-06,
+      "loss": 79.5679,
+      "step": 23150
+    },
+    {
+      "epoch": 0.09356933059143413,
+      "grad_norm": 1159.5113525390625,
+      "learning_rate": 9.77041018040036e-06,
+      "loss": 51.9176,
+      "step": 23160
+    },
+    {
+      "epoch": 0.09360973185680176,
+      "grad_norm": 884.1097412109375,
+      "learning_rate": 9.769987521606068e-06,
+      "loss": 41.639,
+      "step": 23170
+    },
+    {
+      "epoch": 0.09365013312216938,
+      "grad_norm": 527.7691040039062,
+      "learning_rate": 9.769564483285761e-06,
+      "loss": 43.4502,
+      "step": 23180
+    },
+    {
+      "epoch": 0.09369053438753702,
+      "grad_norm": 605.3202514648438,
+      "learning_rate": 9.769141065473099e-06,
+      "loss": 53.9118,
+      "step": 23190
+    },
+    {
+      "epoch": 0.09373093565290465,
+      "grad_norm": 1220.0347900390625,
+      "learning_rate": 9.768717268201768e-06,
+      "loss": 59.4184,
+      "step": 23200
+    },
+    {
+      "epoch": 0.09377133691827227,
+      "grad_norm": 1246.2518310546875,
+      "learning_rate": 9.768293091505491e-06,
+      "loss": 68.2398,
+      "step": 23210
+    },
+    {
+      "epoch": 0.09381173818363991,
+      "grad_norm": 608.1319580078125,
+      "learning_rate": 9.767868535418014e-06,
+      "loss": 61.0542,
+      "step": 23220
+    },
+    {
+      "epoch": 0.09385213944900754,
+      "grad_norm": 839.668701171875,
+      "learning_rate": 9.767443599973122e-06,
+      "loss": 125.2062,
+      "step": 23230
+    },
+    {
+      "epoch": 0.09389254071437518,
+      "grad_norm": 303.346435546875,
+      "learning_rate": 9.76701828520462e-06,
+      "loss": 43.7648,
+      "step": 23240
+    },
+    {
+      "epoch": 0.0939329419797428,
+      "grad_norm": 724.7330932617188,
+      "learning_rate": 9.766592591146353e-06,
+      "loss": 26.6217,
+      "step": 23250
+    },
+    {
+      "epoch": 0.09397334324511043,
+      "grad_norm": 236.2427978515625,
+      "learning_rate": 9.766166517832188e-06,
+      "loss": 45.8864,
+      "step": 23260
+    },
+    {
+      "epoch": 0.09401374451047807,
+      "grad_norm": 245.2400665283203,
+      "learning_rate": 9.765740065296025e-06,
+      "loss": 35.4692,
+      "step": 23270
+    },
+    {
+      "epoch": 0.0940541457758457,
+      "grad_norm": 612.867431640625,
+      "learning_rate": 9.765313233571798e-06,
+      "loss": 48.1673,
+      "step": 23280
+    },
+    {
+      "epoch": 0.09409454704121333,
+      "grad_norm": 440.6855773925781,
+      "learning_rate": 9.76488602269347e-06,
+      "loss": 39.7642,
+      "step": 23290
+    },
+    {
+      "epoch": 0.09413494830658097,
+      "grad_norm": 500.4783935546875,
+      "learning_rate": 9.764458432695026e-06,
+      "loss": 44.2278,
+      "step": 23300
+    },
+    {
+      "epoch": 0.09417534957194859,
+      "grad_norm": 243.58364868164062,
+      "learning_rate": 9.76403046361049e-06,
+      "loss": 31.7886,
+      "step": 23310
+    },
+    {
+      "epoch": 0.09421575083731623,
+      "grad_norm": 1071.7940673828125,
+      "learning_rate": 9.763602115473914e-06,
+      "loss": 66.7946,
+      "step": 23320
+    },
+    {
+      "epoch": 0.09425615210268386,
+      "grad_norm": 483.341552734375,
+      "learning_rate": 9.763173388319381e-06,
+      "loss": 43.0986,
+      "step": 23330
+    },
+    {
+      "epoch": 0.09429655336805148,
+      "grad_norm": 346.4289245605469,
+      "learning_rate": 9.762744282181e-06,
+      "loss": 50.6579,
+      "step": 23340
+    },
+    {
+      "epoch": 0.09433695463341912,
+      "grad_norm": 708.1380004882812,
+      "learning_rate": 9.762314797092916e-06,
+      "loss": 38.8699,
+      "step": 23350
+    },
+    {
+      "epoch": 0.09437735589878675,
+      "grad_norm": 536.1306762695312,
+      "learning_rate": 9.761884933089301e-06,
+      "loss": 43.5115,
+      "step": 23360
+    },
+    {
+      "epoch": 0.09441775716415438,
+      "grad_norm": 487.45355224609375,
+      "learning_rate": 9.761454690204352e-06,
+      "loss": 44.2896,
+      "step": 23370
+    },
+    {
+      "epoch": 0.09445815842952202,
+      "grad_norm": 896.3792724609375,
+      "learning_rate": 9.76102406847231e-06,
+      "loss": 54.5288,
+      "step": 23380
+    },
+    {
+      "epoch": 0.09449855969488964,
+      "grad_norm": 830.8819580078125,
+      "learning_rate": 9.760593067927428e-06,
+      "loss": 97.8602,
+      "step": 23390
+    },
+    {
+      "epoch": 0.09453896096025728,
+      "grad_norm": 769.9625244140625,
+      "learning_rate": 9.760161688604008e-06,
+      "loss": 96.4636,
+      "step": 23400
+    },
+    {
+      "epoch": 0.09457936222562491,
+      "grad_norm": 772.0631103515625,
+      "learning_rate": 9.759729930536367e-06,
+      "loss": 55.2128,
+      "step": 23410
+    },
+    {
+      "epoch": 0.09461976349099253,
+      "grad_norm": 713.3424072265625,
+      "learning_rate": 9.75929779375886e-06,
+      "loss": 43.9229,
+      "step": 23420
+    },
+    {
+      "epoch": 0.09466016475636017,
+      "grad_norm": 606.2196655273438,
+      "learning_rate": 9.75886527830587e-06,
+      "loss": 30.7504,
+      "step": 23430
+    },
+    {
+      "epoch": 0.0947005660217278,
+      "grad_norm": 384.1034240722656,
+      "learning_rate": 9.75843238421181e-06,
+      "loss": 47.3888,
+      "step": 23440
+    },
+    {
+      "epoch": 0.09474096728709543,
+      "grad_norm": 678.3400268554688,
+      "learning_rate": 9.757999111511121e-06,
+      "loss": 52.3418,
+      "step": 23450
+    },
+    {
+      "epoch": 0.09478136855246307,
+      "grad_norm": 493.89691162109375,
+      "learning_rate": 9.757565460238281e-06,
+      "loss": 48.9043,
+      "step": 23460
+    },
+    {
+      "epoch": 0.09482176981783069,
+      "grad_norm": 785.3358764648438,
+      "learning_rate": 9.757131430427791e-06,
+      "loss": 40.5328,
+      "step": 23470
+    },
+    {
+      "epoch": 0.09486217108319833,
+      "grad_norm": 525.2156372070312,
+      "learning_rate": 9.756697022114185e-06,
+      "loss": 35.2212,
+      "step": 23480
+    },
+    {
+      "epoch": 0.09490257234856596,
+      "grad_norm": 1074.95556640625,
+      "learning_rate": 9.756262235332029e-06,
+      "loss": 63.1825,
+      "step": 23490
+    },
+    {
+      "epoch": 0.09494297361393358,
+      "grad_norm": 756.4033203125,
+      "learning_rate": 9.755827070115915e-06,
+      "loss": 60.1457,
+      "step": 23500
+    },
+    {
+      "epoch": 0.09498337487930122,
+      "grad_norm": 655.587890625,
+      "learning_rate": 9.755391526500466e-06,
+      "loss": 58.1549,
+      "step": 23510
+    },
+    {
+      "epoch": 0.09502377614466885,
+      "grad_norm": 876.078857421875,
+      "learning_rate": 9.75495560452034e-06,
+      "loss": 50.9512,
+      "step": 23520
+    },
+    {
+      "epoch": 0.09506417741003648,
+      "grad_norm": 643.4996337890625,
+      "learning_rate": 9.754519304210214e-06,
+      "loss": 45.0397,
+      "step": 23530
+    },
+    {
+      "epoch": 0.09510457867540412,
+      "grad_norm": 799.6237182617188,
+      "learning_rate": 9.754082625604812e-06,
+      "loss": 36.293,
+      "step": 23540
+    },
+    {
+      "epoch": 0.09514497994077174,
+      "grad_norm": 1540.503173828125,
+      "learning_rate": 9.753645568738872e-06,
+      "loss": 61.4238,
+      "step": 23550
+    },
+    {
+      "epoch": 0.09518538120613938,
+      "grad_norm": 634.837890625,
+      "learning_rate": 9.75320813364717e-06,
+      "loss": 48.8427,
+      "step": 23560
+    },
+    {
+      "epoch": 0.09522578247150701,
+      "grad_norm": 1066.7530517578125,
+      "learning_rate": 9.752770320364512e-06,
+      "loss": 45.408,
+      "step": 23570
+    },
+    {
+      "epoch": 0.09526618373687464,
+      "grad_norm": 431.2601623535156,
+      "learning_rate": 9.752332128925732e-06,
+      "loss": 45.2199,
+      "step": 23580
+    },
+    {
+      "epoch": 0.09530658500224228,
+      "grad_norm": 505.9089660644531,
+      "learning_rate": 9.751893559365693e-06,
+      "loss": 38.7723,
+      "step": 23590
+    },
+    {
+      "epoch": 0.0953469862676099,
+      "grad_norm": 540.0280151367188,
+      "learning_rate": 9.751454611719294e-06,
+      "loss": 70.2385,
+      "step": 23600
+    },
+    {
+      "epoch": 0.09538738753297753,
+      "grad_norm": 233.38671875,
+      "learning_rate": 9.751015286021455e-06,
+      "loss": 48.6521,
+      "step": 23610
+    },
+    {
+      "epoch": 0.09542778879834517,
+      "grad_norm": 429.7454833984375,
+      "learning_rate": 9.750575582307136e-06,
+      "loss": 63.9905,
+      "step": 23620
+    },
+    {
+      "epoch": 0.0954681900637128,
+      "grad_norm": 808.2809448242188,
+      "learning_rate": 9.75013550061132e-06,
+      "loss": 67.099,
+      "step": 23630
+    },
+    {
+      "epoch": 0.09550859132908043,
+      "grad_norm": 462.7481384277344,
+      "learning_rate": 9.749695040969022e-06,
+      "loss": 38.5078,
+      "step": 23640
+    },
+    {
+      "epoch": 0.09554899259444806,
+      "grad_norm": 1053.72607421875,
+      "learning_rate": 9.749254203415288e-06,
+      "loss": 60.8307,
+      "step": 23650
+    },
+    {
+      "epoch": 0.09558939385981569,
+      "grad_norm": 527.76513671875,
+      "learning_rate": 9.748812987985193e-06,
+      "loss": 49.8133,
+      "step": 23660
+    },
+    {
+      "epoch": 0.09562979512518333,
+      "grad_norm": 660.1892700195312,
+      "learning_rate": 9.748371394713842e-06,
+      "loss": 42.2738,
+      "step": 23670
+    },
+    {
+      "epoch": 0.09567019639055095,
+      "grad_norm": 435.3009948730469,
+      "learning_rate": 9.747929423636372e-06,
+      "loss": 60.2794,
+      "step": 23680
+    },
+    {
+      "epoch": 0.09571059765591858,
+      "grad_norm": 705.1487426757812,
+      "learning_rate": 9.74748707478795e-06,
+      "loss": 48.4879,
+      "step": 23690
+    },
+    {
+      "epoch": 0.09575099892128622,
+      "grad_norm": 624.763916015625,
+      "learning_rate": 9.747044348203766e-06,
+      "loss": 55.8783,
+      "step": 23700
+    },
+    {
+      "epoch": 0.09579140018665384,
+      "grad_norm": 243.1126708984375,
+      "learning_rate": 9.74660124391905e-06,
+      "loss": 55.0685,
+      "step": 23710
+    },
+    {
+      "epoch": 0.09583180145202148,
+      "grad_norm": 553.0624389648438,
+      "learning_rate": 9.746157761969058e-06,
+      "loss": 64.8828,
+      "step": 23720
+    },
+    {
+      "epoch": 0.09587220271738911,
+      "grad_norm": 3757.382080078125,
+      "learning_rate": 9.745713902389074e-06,
+      "loss": 74.2841,
+      "step": 23730
+    },
+    {
+      "epoch": 0.09591260398275674,
+      "grad_norm": 869.4652709960938,
+      "learning_rate": 9.745269665214415e-06,
+      "loss": 73.5666,
+      "step": 23740
+    },
+    {
+      "epoch": 0.09595300524812438,
+      "grad_norm": 885.2520751953125,
+      "learning_rate": 9.744825050480425e-06,
+      "loss": 54.041,
+      "step": 23750
+    },
+    {
+      "epoch": 0.095993406513492,
+      "grad_norm": 602.6681518554688,
+      "learning_rate": 9.744380058222483e-06,
+      "loss": 58.2484,
+      "step": 23760
+    },
+    {
+      "epoch": 0.09603380777885963,
+      "grad_norm": 327.1231384277344,
+      "learning_rate": 9.743934688475994e-06,
+      "loss": 53.0578,
+      "step": 23770
+    },
+    {
+      "epoch": 0.09607420904422727,
+      "grad_norm": 537.817626953125,
+      "learning_rate": 9.743488941276394e-06,
+      "loss": 45.4333,
+      "step": 23780
+    },
+    {
+      "epoch": 0.0961146103095949,
+      "grad_norm": 657.9298706054688,
+      "learning_rate": 9.743042816659147e-06,
+      "loss": 70.0273,
+      "step": 23790
+    },
+    {
+      "epoch": 0.09615501157496253,
+      "grad_norm": 439.4751281738281,
+      "learning_rate": 9.742596314659751e-06,
+      "loss": 57.6084,
+      "step": 23800
+    },
+    {
+      "epoch": 0.09619541284033016,
+      "grad_norm": 246.55235290527344,
+      "learning_rate": 9.742149435313732e-06,
+      "loss": 47.0762,
+      "step": 23810
+    },
+    {
+      "epoch": 0.09623581410569779,
+      "grad_norm": 485.2574768066406,
+      "learning_rate": 9.741702178656647e-06,
+      "loss": 47.7125,
+      "step": 23820
+    },
+    {
+      "epoch": 0.09627621537106543,
+      "grad_norm": 946.404052734375,
+      "learning_rate": 9.74125454472408e-06,
+      "loss": 64.874,
+      "step": 23830
+    },
+    {
+      "epoch": 0.09631661663643305,
+      "grad_norm": 368.9576110839844,
+      "learning_rate": 9.740806533551647e-06,
+      "loss": 58.668,
+      "step": 23840
+    },
+    {
+      "epoch": 0.09635701790180068,
+      "grad_norm": 882.6640014648438,
+      "learning_rate": 9.740358145174999e-06,
+      "loss": 36.2069,
+      "step": 23850
+    },
+    {
+      "epoch": 0.09639741916716832,
+      "grad_norm": 496.58367919921875,
+      "learning_rate": 9.739909379629805e-06,
+      "loss": 49.835,
+      "step": 23860
+    },
+    {
+      "epoch": 0.09643782043253595,
+      "grad_norm": 400.39715576171875,
+      "learning_rate": 9.739460236951778e-06,
+      "loss": 50.0755,
+      "step": 23870
+    },
+    {
+      "epoch": 0.09647822169790359,
+      "grad_norm": 576.3343505859375,
+      "learning_rate": 9.739010717176649e-06,
+      "loss": 38.722,
+      "step": 23880
+    },
+    {
+      "epoch": 0.09651862296327121,
+      "grad_norm": 463.68853759765625,
+      "learning_rate": 9.738560820340189e-06,
+      "loss": 48.1985,
+      "step": 23890
+    },
+    {
+      "epoch": 0.09655902422863884,
+      "grad_norm": 504.19940185546875,
+      "learning_rate": 9.738110546478188e-06,
+      "loss": 59.8322,
+      "step": 23900
+    },
+    {
+      "epoch": 0.09659942549400648,
+      "grad_norm": 1040.739501953125,
+      "learning_rate": 9.737659895626478e-06,
+      "loss": 63.9173,
+      "step": 23910
+    },
+    {
+      "epoch": 0.0966398267593741,
+      "grad_norm": 671.8014526367188,
+      "learning_rate": 9.737208867820914e-06,
+      "loss": 38.5287,
+      "step": 23920
+    },
+    {
+      "epoch": 0.09668022802474173,
+      "grad_norm": 643.8292236328125,
+      "learning_rate": 9.736757463097378e-06,
+      "loss": 57.7698,
+      "step": 23930
+    },
+    {
+      "epoch": 0.09672062929010937,
+      "grad_norm": 479.278564453125,
+      "learning_rate": 9.736305681491792e-06,
+      "loss": 53.5585,
+      "step": 23940
+    },
+    {
+      "epoch": 0.096761030555477,
+      "grad_norm": 839.4384765625,
+      "learning_rate": 9.735853523040098e-06,
+      "loss": 57.4403,
+      "step": 23950
+    },
+    {
+      "epoch": 0.09680143182084464,
+      "grad_norm": 667.2028198242188,
+      "learning_rate": 9.735400987778274e-06,
+      "loss": 50.5064,
+      "step": 23960
+    },
+    {
+      "epoch": 0.09684183308621226,
+      "grad_norm": 474.6343078613281,
+      "learning_rate": 9.734948075742328e-06,
+      "loss": 65.1481,
+      "step": 23970
+    },
+    {
+      "epoch": 0.09688223435157989,
+      "grad_norm": 284.0242614746094,
+      "learning_rate": 9.734494786968293e-06,
+      "loss": 38.4558,
+      "step": 23980
+    },
+    {
+      "epoch": 0.09692263561694753,
+      "grad_norm": 335.21466064453125,
+      "learning_rate": 9.734041121492235e-06,
+      "loss": 51.5069,
+      "step": 23990
+    },
+    {
+      "epoch": 0.09696303688231515,
+      "grad_norm": 1233.53369140625,
+      "learning_rate": 9.733587079350254e-06,
+      "loss": 58.2334,
+      "step": 24000
+    },
+    {
+      "epoch": 0.09700343814768278,
+      "grad_norm": 846.1797485351562,
+      "learning_rate": 9.73313266057847e-06,
+      "loss": 57.5334,
+      "step": 24010
+    },
+    {
+      "epoch": 0.09704383941305042,
+      "grad_norm": 553.3931884765625,
+      "learning_rate": 9.732677865213044e-06,
+      "loss": 69.9946,
+      "step": 24020
+    },
+    {
+      "epoch": 0.09708424067841805,
+      "grad_norm": 881.1492309570312,
+      "learning_rate": 9.73222269329016e-06,
+      "loss": 55.2486,
+      "step": 24030
+    },
+    {
+      "epoch": 0.09712464194378569,
+      "grad_norm": 423.26806640625,
+      "learning_rate": 9.731767144846034e-06,
+      "loss": 43.9426,
+      "step": 24040
+    },
+    {
+      "epoch": 0.09716504320915331,
+      "grad_norm": 849.846923828125,
+      "learning_rate": 9.731311219916912e-06,
+      "loss": 45.18,
+      "step": 24050
+    },
+    {
+      "epoch": 0.09720544447452094,
+      "grad_norm": 550.5770874023438,
+      "learning_rate": 9.730854918539072e-06,
+      "loss": 48.6627,
+      "step": 24060
+    },
+    {
+      "epoch": 0.09724584573988858,
+      "grad_norm": 213.45596313476562,
+      "learning_rate": 9.730398240748816e-06,
+      "loss": 28.9641,
+      "step": 24070
+    },
+    {
+      "epoch": 0.0972862470052562,
+      "grad_norm": 736.7731323242188,
+      "learning_rate": 9.729941186582482e-06,
+      "loss": 56.5135,
+      "step": 24080
+    },
+    {
+      "epoch": 0.09732664827062383,
+      "grad_norm": 574.6141967773438,
+      "learning_rate": 9.729483756076436e-06,
+      "loss": 39.7891,
+      "step": 24090
+    },
+    {
+      "epoch": 0.09736704953599147,
+      "grad_norm": 418.90606689453125,
+      "learning_rate": 9.729025949267072e-06,
+      "loss": 61.8591,
+      "step": 24100
+    },
+    {
+      "epoch": 0.0974074508013591,
+      "grad_norm": 603.3834838867188,
+      "learning_rate": 9.728567766190817e-06,
+      "loss": 51.6418,
+      "step": 24110
+    },
+    {
+      "epoch": 0.09744785206672674,
+      "grad_norm": 576.2706298828125,
+      "learning_rate": 9.728109206884125e-06,
+      "loss": 48.2781,
+      "step": 24120
+    },
+    {
+      "epoch": 0.09748825333209436,
+      "grad_norm": 609.32861328125,
+      "learning_rate": 9.727650271383485e-06,
+      "loss": 57.3191,
+      "step": 24130
+    },
+    {
+      "epoch": 0.09752865459746199,
+      "grad_norm": 1344.430908203125,
+      "learning_rate": 9.727190959725407e-06,
+      "loss": 51.7622,
+      "step": 24140
+    },
+    {
+      "epoch": 0.09756905586282963,
+      "grad_norm": 996.9224853515625,
+      "learning_rate": 9.72673127194644e-06,
+      "loss": 49.3176,
+      "step": 24150
+    },
+    {
+      "epoch": 0.09760945712819726,
+      "grad_norm": 469.8564147949219,
+      "learning_rate": 9.72627120808316e-06,
+      "loss": 50.8814,
+      "step": 24160
+    },
+    {
+      "epoch": 0.09764985839356488,
+      "grad_norm": 656.7633666992188,
+      "learning_rate": 9.725810768172169e-06,
+      "loss": 42.3984,
+      "step": 24170
+    },
+    {
+      "epoch": 0.09769025965893252,
+      "grad_norm": 842.8346557617188,
+      "learning_rate": 9.725349952250105e-06,
+      "loss": 68.7092,
+      "step": 24180
+    },
+    {
+      "epoch": 0.09773066092430015,
+      "grad_norm": 724.8569946289062,
+      "learning_rate": 9.724888760353631e-06,
+      "loss": 54.5133,
+      "step": 24190
+    },
+    {
+      "epoch": 0.09777106218966779,
+      "grad_norm": 1168.2147216796875,
+      "learning_rate": 9.72442719251944e-06,
+      "loss": 44.4517,
+      "step": 24200
+    },
+    {
+      "epoch": 0.09781146345503541,
+      "grad_norm": 905.7360229492188,
+      "learning_rate": 9.723965248784264e-06,
+      "loss": 67.6947,
+      "step": 24210
+    },
+    {
+      "epoch": 0.09785186472040304,
+      "grad_norm": 591.1112670898438,
+      "learning_rate": 9.723502929184851e-06,
+      "loss": 45.0093,
+      "step": 24220
+    },
+    {
+      "epoch": 0.09789226598577068,
+      "grad_norm": 1430.4012451171875,
+      "learning_rate": 9.723040233757987e-06,
+      "loss": 71.5288,
+      "step": 24230
+    },
+    {
+      "epoch": 0.0979326672511383,
+      "grad_norm": 675.6157836914062,
+      "learning_rate": 9.722577162540489e-06,
+      "loss": 46.933,
+      "step": 24240
+    },
+    {
+      "epoch": 0.09797306851650593,
+      "grad_norm": 528.8775024414062,
+      "learning_rate": 9.7221137155692e-06,
+      "loss": 50.8753,
+      "step": 24250
+    },
+    {
+      "epoch": 0.09801346978187357,
+      "grad_norm": 880.9699096679688,
+      "learning_rate": 9.721649892880995e-06,
+      "loss": 69.2766,
+      "step": 24260
+    },
+    {
+      "epoch": 0.0980538710472412,
+      "grad_norm": 1034.8077392578125,
+      "learning_rate": 9.721185694512776e-06,
+      "loss": 41.0543,
+      "step": 24270
+    },
+    {
+      "epoch": 0.09809427231260884,
+      "grad_norm": 1290.8067626953125,
+      "learning_rate": 9.720721120501478e-06,
+      "loss": 41.7326,
+      "step": 24280
+    },
+    {
+      "epoch": 0.09813467357797646,
+      "grad_norm": 617.8194580078125,
+      "learning_rate": 9.720256170884066e-06,
+      "loss": 45.3977,
+      "step": 24290
+    },
+    {
+      "epoch": 0.09817507484334409,
+      "grad_norm": 480.2484436035156,
+      "learning_rate": 9.719790845697534e-06,
+      "loss": 37.6541,
+      "step": 24300
+    },
+    {
+      "epoch": 0.09821547610871173,
+      "grad_norm": 569.1683959960938,
+      "learning_rate": 9.719325144978907e-06,
+      "loss": 50.1147,
+      "step": 24310
+    },
+    {
+      "epoch": 0.09825587737407936,
+      "grad_norm": 638.9716796875,
+      "learning_rate": 9.718859068765234e-06,
+      "loss": 56.0363,
+      "step": 24320
+    },
+    {
+      "epoch": 0.09829627863944698,
+      "grad_norm": 638.63525390625,
+      "learning_rate": 9.718392617093602e-06,
+      "loss": 44.4817,
+      "step": 24330
+    },
+    {
+      "epoch": 0.09833667990481462,
+      "grad_norm": 538.0601806640625,
+      "learning_rate": 9.717925790001125e-06,
+      "loss": 35.1493,
+      "step": 24340
+    },
+    {
+      "epoch": 0.09837708117018225,
+      "grad_norm": 655.0274047851562,
+      "learning_rate": 9.717458587524946e-06,
+      "loss": 35.8556,
+      "step": 24350
+    },
+    {
+      "epoch": 0.09841748243554987,
+      "grad_norm": 2356.542724609375,
+      "learning_rate": 9.716991009702236e-06,
+      "loss": 77.9995,
+      "step": 24360
+    },
+    {
+      "epoch": 0.09845788370091751,
+      "grad_norm": 423.50372314453125,
+      "learning_rate": 9.7165230565702e-06,
+      "loss": 52.0112,
+      "step": 24370
+    },
+    {
+      "epoch": 0.09849828496628514,
+      "grad_norm": 596.177001953125,
+      "learning_rate": 9.71605472816607e-06,
+      "loss": 68.7233,
+      "step": 24380
+    },
+    {
+      "epoch": 0.09853868623165278,
+      "grad_norm": 848.5178833007812,
+      "learning_rate": 9.71558602452711e-06,
+      "loss": 56.1421,
+      "step": 24390
+    },
+    {
+      "epoch": 0.0985790874970204,
+      "grad_norm": 548.5185546875,
+      "learning_rate": 9.71511694569061e-06,
+      "loss": 57.7374,
+      "step": 24400
+    },
+    {
+      "epoch": 0.09861948876238803,
+      "grad_norm": 959.719970703125,
+      "learning_rate": 9.714647491693897e-06,
+      "loss": 48.1545,
+      "step": 24410
+    },
+    {
+      "epoch": 0.09865989002775567,
+      "grad_norm": 1168.14794921875,
+      "learning_rate": 9.714177662574316e-06,
+      "loss": 60.9899,
+      "step": 24420
+    },
+    {
+      "epoch": 0.0987002912931233,
+      "grad_norm": 834.342041015625,
+      "learning_rate": 9.713707458369258e-06,
+      "loss": 37.8764,
+      "step": 24430
+    },
+    {
+      "epoch": 0.09874069255849093,
+      "grad_norm": 506.5158386230469,
+      "learning_rate": 9.713236879116127e-06,
+      "loss": 63.5037,
+      "step": 24440
+    },
+    {
+      "epoch": 0.09878109382385857,
+      "grad_norm": 651.78759765625,
+      "learning_rate": 9.71276592485237e-06,
+      "loss": 72.3737,
+      "step": 24450
+    },
+    {
+      "epoch": 0.09882149508922619,
+      "grad_norm": 760.8204345703125,
+      "learning_rate": 9.712294595615458e-06,
+      "loss": 52.8254,
+      "step": 24460
+    },
+    {
+      "epoch": 0.09886189635459383,
+      "grad_norm": 928.4266967773438,
+      "learning_rate": 9.711822891442887e-06,
+      "loss": 47.5478,
+      "step": 24470
+    },
+    {
+      "epoch": 0.09890229761996146,
+      "grad_norm": 553.4059448242188,
+      "learning_rate": 9.711350812372198e-06,
+      "loss": 35.1253,
+      "step": 24480
+    },
+    {
+      "epoch": 0.09894269888532908,
+      "grad_norm": 774.9642944335938,
+      "learning_rate": 9.710878358440945e-06,
+      "loss": 90.113,
+      "step": 24490
+    },
+    {
+      "epoch": 0.09898310015069672,
+      "grad_norm": 762.3937377929688,
+      "learning_rate": 9.710405529686722e-06,
+      "loss": 60.1524,
+      "step": 24500
+    },
+    {
+      "epoch": 0.09902350141606435,
+      "grad_norm": 754.8513793945312,
+      "learning_rate": 9.709932326147147e-06,
+      "loss": 57.3479,
+      "step": 24510
+    },
+    {
+      "epoch": 0.09906390268143198,
+      "grad_norm": 2144.4775390625,
+      "learning_rate": 9.709458747859874e-06,
+      "loss": 57.9094,
+      "step": 24520
+    },
+    {
+      "epoch": 0.09910430394679962,
+      "grad_norm": 404.76617431640625,
+      "learning_rate": 9.708984794862581e-06,
+      "loss": 50.33,
+      "step": 24530
+    },
+    {
+      "epoch": 0.09914470521216724,
+      "grad_norm": 587.0454711914062,
+      "learning_rate": 9.708510467192981e-06,
+      "loss": 42.9969,
+      "step": 24540
+    },
+    {
+      "epoch": 0.09918510647753488,
+      "grad_norm": 457.4508056640625,
+      "learning_rate": 9.70803576488881e-06,
+      "loss": 47.6615,
+      "step": 24550
+    },
+    {
+      "epoch": 0.09922550774290251,
+      "grad_norm": 676.8587646484375,
+      "learning_rate": 9.707560687987843e-06,
+      "loss": 48.9946,
+      "step": 24560
+    },
+    {
+      "epoch": 0.09926590900827013,
+      "grad_norm": 1258.223388671875,
+      "learning_rate": 9.707085236527873e-06,
+      "loss": 67.3204,
+      "step": 24570
+    },
+    {
+      "epoch": 0.09930631027363777,
+      "grad_norm": 503.6045837402344,
+      "learning_rate": 9.706609410546736e-06,
+      "loss": 40.1334,
+      "step": 24580
+    },
+    {
+      "epoch": 0.0993467115390054,
+      "grad_norm": 786.1532592773438,
+      "learning_rate": 9.706133210082288e-06,
+      "loss": 54.2325,
+      "step": 24590
+    },
+    {
+      "epoch": 0.09938711280437303,
+      "grad_norm": 456.151123046875,
+      "learning_rate": 9.705656635172418e-06,
+      "loss": 62.9975,
+      "step": 24600
+    },
+    {
+      "epoch": 0.09942751406974067,
+      "grad_norm": 651.8788452148438,
+      "learning_rate": 9.705179685855048e-06,
+      "loss": 44.2353,
+      "step": 24610
+    },
+    {
+      "epoch": 0.09946791533510829,
+      "grad_norm": 685.383056640625,
+      "learning_rate": 9.704702362168121e-06,
+      "loss": 38.0583,
+      "step": 24620
+    },
+    {
+      "epoch": 0.09950831660047593,
+      "grad_norm": 366.0917053222656,
+      "learning_rate": 9.704224664149621e-06,
+      "loss": 37.4004,
+      "step": 24630
+    },
+    {
+      "epoch": 0.09954871786584356,
+      "grad_norm": 546.1963500976562,
+      "learning_rate": 9.703746591837552e-06,
+      "loss": 50.105,
+      "step": 24640
+    },
+    {
+      "epoch": 0.09958911913121118,
+      "grad_norm": 707.4985961914062,
+      "learning_rate": 9.703268145269957e-06,
+      "loss": 45.0067,
+      "step": 24650
+    },
+    {
+      "epoch": 0.09962952039657882,
+      "grad_norm": 918.8980712890625,
+      "learning_rate": 9.702789324484898e-06,
+      "loss": 38.0209,
+      "step": 24660
+    },
+    {
+      "epoch": 0.09966992166194645,
+      "grad_norm": 1020.766845703125,
+      "learning_rate": 9.702310129520476e-06,
+      "loss": 62.4222,
+      "step": 24670
+    },
+    {
+      "epoch": 0.09971032292731408,
+      "grad_norm": 1279.9139404296875,
+      "learning_rate": 9.701830560414817e-06,
+      "loss": 49.9268,
+      "step": 24680
+    },
+    {
+      "epoch": 0.09975072419268172,
+      "grad_norm": 896.4984130859375,
+      "learning_rate": 9.701350617206081e-06,
+      "loss": 60.7759,
+      "step": 24690
+    },
+    {
+      "epoch": 0.09979112545804934,
+      "grad_norm": 583.581298828125,
+      "learning_rate": 9.700870299932453e-06,
+      "loss": 33.8629,
+      "step": 24700
+    },
+    {
+      "epoch": 0.09983152672341698,
+      "grad_norm": 683.1466674804688,
+      "learning_rate": 9.700389608632146e-06,
+      "loss": 65.077,
+      "step": 24710
+    },
+    {
+      "epoch": 0.09987192798878461,
+      "grad_norm": 742.3271484375,
+      "learning_rate": 9.699908543343413e-06,
+      "loss": 44.0271,
+      "step": 24720
+    },
+    {
+      "epoch": 0.09991232925415224,
+      "grad_norm": 280.29315185546875,
+      "learning_rate": 9.699427104104525e-06,
+      "loss": 53.0423,
+      "step": 24730
+    },
+    {
+      "epoch": 0.09995273051951988,
+      "grad_norm": 834.8994750976562,
+      "learning_rate": 9.698945290953789e-06,
+      "loss": 65.0682,
+      "step": 24740
+    },
+    {
+      "epoch": 0.0999931317848875,
+      "grad_norm": 1069.6033935546875,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 61.3382,
+      "step": 24750
+    },
+    {
+      "epoch": 0.10003353305025513,
+      "grad_norm": 301.9818115234375,
+      "learning_rate": 9.69798054307015e-06,
+      "loss": 42.7463,
+      "step": 24760
+    },
+    {
+      "epoch": 0.10007393431562277,
+      "grad_norm": 563.1431884765625,
+      "learning_rate": 9.697497608414007e-06,
+      "loss": 42.7955,
+      "step": 24770
+    },
+    {
+      "epoch": 0.1001143355809904,
+      "grad_norm": 715.0123291015625,
+      "learning_rate": 9.697014299999536e-06,
+      "loss": 37.4068,
+      "step": 24780
+    },
+    {
+      "epoch": 0.10015473684635803,
+      "grad_norm": 946.2361450195312,
+      "learning_rate": 9.696530617865197e-06,
+      "loss": 61.5986,
+      "step": 24790
+    },
+    {
+      "epoch": 0.10019513811172566,
+      "grad_norm": 630.42626953125,
+      "learning_rate": 9.696046562049469e-06,
+      "loss": 56.6305,
+      "step": 24800
+    },
+    {
+      "epoch": 0.10023553937709329,
+      "grad_norm": 725.95947265625,
+      "learning_rate": 9.695562132590865e-06,
+      "loss": 60.9256,
+      "step": 24810
+    },
+    {
+      "epoch": 0.10027594064246093,
+      "grad_norm": 868.1839599609375,
+      "learning_rate": 9.695077329527936e-06,
+      "loss": 53.6544,
+      "step": 24820
+    },
+    {
+      "epoch": 0.10031634190782855,
+      "grad_norm": 817.20751953125,
+      "learning_rate": 9.694592152899249e-06,
+      "loss": 89.8513,
+      "step": 24830
+    },
+    {
+      "epoch": 0.10035674317319618,
+      "grad_norm": 603.2654418945312,
+      "learning_rate": 9.694106602743411e-06,
+      "loss": 62.9391,
+      "step": 24840
+    },
+    {
+      "epoch": 0.10039714443856382,
+      "grad_norm": 190.94393920898438,
+      "learning_rate": 9.693620679099055e-06,
+      "loss": 50.3173,
+      "step": 24850
+    },
+    {
+      "epoch": 0.10043754570393144,
+      "grad_norm": 761.8203735351562,
+      "learning_rate": 9.693134382004839e-06,
+      "loss": 74.0534,
+      "step": 24860
+    },
+    {
+      "epoch": 0.10047794696929908,
+      "grad_norm": 860.0404663085938,
+      "learning_rate": 9.69264771149946e-06,
+      "loss": 58.5714,
+      "step": 24870
+    },
+    {
+      "epoch": 0.10051834823466671,
+      "grad_norm": 734.0036010742188,
+      "learning_rate": 9.692160667621639e-06,
+      "loss": 52.3132,
+      "step": 24880
+    },
+    {
+      "epoch": 0.10055874950003434,
+      "grad_norm": 894.4503784179688,
+      "learning_rate": 9.69167325041013e-06,
+      "loss": 28.4199,
+      "step": 24890
+    },
+    {
+      "epoch": 0.10059915076540198,
+      "grad_norm": 394.6111145019531,
+      "learning_rate": 9.69118545990371e-06,
+      "loss": 49.8315,
+      "step": 24900
+    },
+    {
+      "epoch": 0.1006395520307696,
+      "grad_norm": 600.8136596679688,
+      "learning_rate": 9.690697296141194e-06,
+      "loss": 40.0911,
+      "step": 24910
+    },
+    {
+      "epoch": 0.10067995329613723,
+      "grad_norm": 710.9945678710938,
+      "learning_rate": 9.690208759161418e-06,
+      "loss": 42.2445,
+      "step": 24920
+    },
+    {
+      "epoch": 0.10072035456150487,
+      "grad_norm": 516.5250244140625,
+      "learning_rate": 9.689719849003261e-06,
+      "loss": 40.4775,
+      "step": 24930
+    },
+    {
+      "epoch": 0.1007607558268725,
+      "grad_norm": 948.2924194335938,
+      "learning_rate": 9.689230565705617e-06,
+      "loss": 54.2192,
+      "step": 24940
+    },
+    {
+      "epoch": 0.10080115709224013,
+      "grad_norm": 692.30712890625,
+      "learning_rate": 9.688740909307416e-06,
+      "loss": 42.3075,
+      "step": 24950
+    },
+    {
+      "epoch": 0.10084155835760776,
+      "grad_norm": 279.9989013671875,
+      "learning_rate": 9.68825087984762e-06,
+      "loss": 44.2146,
+      "step": 24960
+    },
+    {
+      "epoch": 0.10088195962297539,
+      "grad_norm": 425.18505859375,
+      "learning_rate": 9.687760477365217e-06,
+      "loss": 26.8136,
+      "step": 24970
+    },
+    {
+      "epoch": 0.10092236088834303,
+      "grad_norm": 1031.68115234375,
+      "learning_rate": 9.687269701899228e-06,
+      "loss": 73.9425,
+      "step": 24980
+    },
+    {
+      "epoch": 0.10096276215371065,
+      "grad_norm": 873.32763671875,
+      "learning_rate": 9.6867785534887e-06,
+      "loss": 50.5244,
+      "step": 24990
+    },
+    {
+      "epoch": 0.10100316341907828,
+      "grad_norm": 332.6380920410156,
+      "learning_rate": 9.686287032172712e-06,
+      "loss": 49.424,
+      "step": 25000
+    },
+    {
+      "epoch": 0.10104356468444592,
+      "grad_norm": 6846.8583984375,
+      "learning_rate": 9.685795137990372e-06,
+      "loss": 97.2676,
+      "step": 25010
+    },
+    {
+      "epoch": 0.10108396594981355,
+      "grad_norm": 860.862548828125,
+      "learning_rate": 9.685302870980819e-06,
+      "loss": 80.9092,
+      "step": 25020
+    },
+    {
+      "epoch": 0.10112436721518119,
+      "grad_norm": 712.3943481445312,
+      "learning_rate": 9.684810231183218e-06,
+      "loss": 48.4841,
+      "step": 25030
+    },
+    {
+      "epoch": 0.10116476848054881,
+      "grad_norm": 820.1293334960938,
+      "learning_rate": 9.684317218636767e-06,
+      "loss": 42.9877,
+      "step": 25040
+    },
+    {
+      "epoch": 0.10120516974591644,
+      "grad_norm": 782.7633056640625,
+      "learning_rate": 9.683823833380692e-06,
+      "loss": 59.5399,
+      "step": 25050
+    },
+    {
+      "epoch": 0.10124557101128408,
+      "grad_norm": 718.3815307617188,
+      "learning_rate": 9.683330075454252e-06,
+      "loss": 53.6291,
+      "step": 25060
+    },
+    {
+      "epoch": 0.1012859722766517,
+      "grad_norm": 488.6388854980469,
+      "learning_rate": 9.68283594489673e-06,
+      "loss": 53.4475,
+      "step": 25070
+    },
+    {
+      "epoch": 0.10132637354201933,
+      "grad_norm": 604.8165283203125,
+      "learning_rate": 9.682341441747446e-06,
+      "loss": 48.3369,
+      "step": 25080
+    },
+    {
+      "epoch": 0.10136677480738697,
+      "grad_norm": 1349.2950439453125,
+      "learning_rate": 9.68184656604574e-06,
+      "loss": 58.4348,
+      "step": 25090
+    },
+    {
+      "epoch": 0.1014071760727546,
+      "grad_norm": 1390.267333984375,
+      "learning_rate": 9.681351317830991e-06,
+      "loss": 47.6931,
+      "step": 25100
+    },
+    {
+      "epoch": 0.10144757733812224,
+      "grad_norm": 280.08746337890625,
+      "learning_rate": 9.680855697142601e-06,
+      "loss": 64.118,
+      "step": 25110
+    },
+    {
+      "epoch": 0.10148797860348986,
+      "grad_norm": 441.5658264160156,
+      "learning_rate": 9.680359704020005e-06,
+      "loss": 48.1863,
+      "step": 25120
+    },
+    {
+      "epoch": 0.10152837986885749,
+      "grad_norm": 354.169921875,
+      "learning_rate": 9.67986333850267e-06,
+      "loss": 45.0124,
+      "step": 25130
+    },
+    {
+      "epoch": 0.10156878113422513,
+      "grad_norm": 281.8157043457031,
+      "learning_rate": 9.679366600630085e-06,
+      "loss": 74.9676,
+      "step": 25140
+    },
+    {
+      "epoch": 0.10160918239959275,
+      "grad_norm": 355.419921875,
+      "learning_rate": 9.678869490441775e-06,
+      "loss": 57.2231,
+      "step": 25150
+    },
+    {
+      "epoch": 0.10164958366496038,
+      "grad_norm": 700.468505859375,
+      "learning_rate": 9.678372007977292e-06,
+      "loss": 58.7536,
+      "step": 25160
+    },
+    {
+      "epoch": 0.10168998493032802,
+      "grad_norm": 935.990478515625,
+      "learning_rate": 9.67787415327622e-06,
+      "loss": 60.8175,
+      "step": 25170
+    },
+    {
+      "epoch": 0.10173038619569565,
+      "grad_norm": 727.43994140625,
+      "learning_rate": 9.67737592637817e-06,
+      "loss": 61.9682,
+      "step": 25180
+    },
+    {
+      "epoch": 0.10177078746106329,
+      "grad_norm": 507.2603759765625,
+      "learning_rate": 9.676877327322785e-06,
+      "loss": 67.5783,
+      "step": 25190
+    },
+    {
+      "epoch": 0.10181118872643091,
+      "grad_norm": 652.1765747070312,
+      "learning_rate": 9.676378356149733e-06,
+      "loss": 41.8282,
+      "step": 25200
+    },
+    {
+      "epoch": 0.10185158999179854,
+      "grad_norm": 751.2033081054688,
+      "learning_rate": 9.675879012898719e-06,
+      "loss": 55.6442,
+      "step": 25210
+    },
+    {
+      "epoch": 0.10189199125716618,
+      "grad_norm": 429.15814208984375,
+      "learning_rate": 9.67537929760947e-06,
+      "loss": 31.6516,
+      "step": 25220
+    },
+    {
+      "epoch": 0.1019323925225338,
+      "grad_norm": 612.541015625,
+      "learning_rate": 9.674879210321747e-06,
+      "loss": 58.0954,
+      "step": 25230
+    },
+    {
+      "epoch": 0.10197279378790143,
+      "grad_norm": 3285.99853515625,
+      "learning_rate": 9.67437875107534e-06,
+      "loss": 90.851,
+      "step": 25240
+    },
+    {
+      "epoch": 0.10201319505326907,
+      "grad_norm": 482.5252990722656,
+      "learning_rate": 9.673877919910069e-06,
+      "loss": 43.1322,
+      "step": 25250
+    },
+    {
+      "epoch": 0.1020535963186367,
+      "grad_norm": 661.02294921875,
+      "learning_rate": 9.673376716865781e-06,
+      "loss": 59.0004,
+      "step": 25260
+    },
+    {
+      "epoch": 0.10209399758400434,
+      "grad_norm": 734.8023681640625,
+      "learning_rate": 9.672875141982358e-06,
+      "loss": 55.4186,
+      "step": 25270
+    },
+    {
+      "epoch": 0.10213439884937196,
+      "grad_norm": 605.2401123046875,
+      "learning_rate": 9.672373195299704e-06,
+      "loss": 43.0386,
+      "step": 25280
+    },
+    {
+      "epoch": 0.10217480011473959,
+      "grad_norm": 489.87982177734375,
+      "learning_rate": 9.67187087685776e-06,
+      "loss": 48.6595,
+      "step": 25290
+    },
+    {
+      "epoch": 0.10221520138010723,
+      "grad_norm": 468.3460693359375,
+      "learning_rate": 9.671368186696488e-06,
+      "loss": 44.0622,
+      "step": 25300
+    },
+    {
+      "epoch": 0.10225560264547486,
+      "grad_norm": 1122.7103271484375,
+      "learning_rate": 9.670865124855889e-06,
+      "loss": 53.3322,
+      "step": 25310
+    },
+    {
+      "epoch": 0.10229600391084248,
+      "grad_norm": 437.603515625,
+      "learning_rate": 9.67036169137599e-06,
+      "loss": 58.5753,
+      "step": 25320
+    },
+    {
+      "epoch": 0.10233640517621012,
+      "grad_norm": 671.1591796875,
+      "learning_rate": 9.669857886296842e-06,
+      "loss": 52.2499,
+      "step": 25330
+    },
+    {
+      "epoch": 0.10237680644157775,
+      "grad_norm": 412.7358703613281,
+      "learning_rate": 9.669353709658537e-06,
+      "loss": 75.7357,
+      "step": 25340
+    },
+    {
+      "epoch": 0.10241720770694539,
+      "grad_norm": 769.44775390625,
+      "learning_rate": 9.668849161501186e-06,
+      "loss": 50.5775,
+      "step": 25350
+    },
+    {
+      "epoch": 0.10245760897231301,
+      "grad_norm": 321.61749267578125,
+      "learning_rate": 9.668344241864934e-06,
+      "loss": 42.0402,
+      "step": 25360
+    },
+    {
+      "epoch": 0.10249801023768064,
+      "grad_norm": 645.4732055664062,
+      "learning_rate": 9.667838950789957e-06,
+      "loss": 77.0858,
+      "step": 25370
+    },
+    {
+      "epoch": 0.10253841150304828,
+      "grad_norm": 922.1859130859375,
+      "learning_rate": 9.667333288316454e-06,
+      "loss": 69.5223,
+      "step": 25380
+    },
+    {
+      "epoch": 0.1025788127684159,
+      "grad_norm": 683.8822631835938,
+      "learning_rate": 9.666827254484663e-06,
+      "loss": 43.2432,
+      "step": 25390
+    },
+    {
+      "epoch": 0.10261921403378353,
+      "grad_norm": 686.84423828125,
+      "learning_rate": 9.666320849334846e-06,
+      "loss": 67.6858,
+      "step": 25400
+    },
+    {
+      "epoch": 0.10265961529915117,
+      "grad_norm": 376.72894287109375,
+      "learning_rate": 9.665814072907293e-06,
+      "loss": 45.7562,
+      "step": 25410
+    },
+    {
+      "epoch": 0.1027000165645188,
+      "grad_norm": 690.8438720703125,
+      "learning_rate": 9.665306925242329e-06,
+      "loss": 50.5314,
+      "step": 25420
+    },
+    {
+      "epoch": 0.10274041782988644,
+      "grad_norm": 519.7843017578125,
+      "learning_rate": 9.664799406380302e-06,
+      "loss": 35.6388,
+      "step": 25430
+    },
+    {
+      "epoch": 0.10278081909525406,
+      "grad_norm": 739.2352294921875,
+      "learning_rate": 9.664291516361597e-06,
+      "loss": 56.1373,
+      "step": 25440
+    },
+    {
+      "epoch": 0.10282122036062169,
+      "grad_norm": 564.990966796875,
+      "learning_rate": 9.663783255226622e-06,
+      "loss": 71.4753,
+      "step": 25450
+    },
+    {
+      "epoch": 0.10286162162598933,
+      "grad_norm": 643.6279907226562,
+      "learning_rate": 9.663274623015816e-06,
+      "loss": 59.0967,
+      "step": 25460
+    },
+    {
+      "epoch": 0.10290202289135696,
+      "grad_norm": 794.2958984375,
+      "learning_rate": 9.662765619769651e-06,
+      "loss": 58.3527,
+      "step": 25470
+    },
+    {
+      "epoch": 0.10294242415672458,
+      "grad_norm": 868.2996826171875,
+      "learning_rate": 9.662256245528622e-06,
+      "loss": 43.9523,
+      "step": 25480
+    },
+    {
+      "epoch": 0.10298282542209222,
+      "grad_norm": 301.7526550292969,
+      "learning_rate": 9.661746500333265e-06,
+      "loss": 36.3696,
+      "step": 25490
+    },
+    {
+      "epoch": 0.10302322668745985,
+      "grad_norm": 802.8497924804688,
+      "learning_rate": 9.66123638422413e-06,
+      "loss": 58.9308,
+      "step": 25500
+    },
+    {
+      "epoch": 0.10306362795282749,
+      "grad_norm": 747.5589599609375,
+      "learning_rate": 9.66072589724181e-06,
+      "loss": 45.8074,
+      "step": 25510
+    },
+    {
+      "epoch": 0.10310402921819511,
+      "grad_norm": 474.6321716308594,
+      "learning_rate": 9.66021503942692e-06,
+      "loss": 58.1529,
+      "step": 25520
+    },
+    {
+      "epoch": 0.10314443048356274,
+      "grad_norm": 289.06378173828125,
+      "learning_rate": 9.659703810820105e-06,
+      "loss": 57.4549,
+      "step": 25530
+    },
+    {
+      "epoch": 0.10318483174893038,
+      "grad_norm": 699.4983520507812,
+      "learning_rate": 9.659192211462043e-06,
+      "loss": 72.7736,
+      "step": 25540
+    },
+    {
+      "epoch": 0.103225233014298,
+      "grad_norm": 823.0641479492188,
+      "learning_rate": 9.658680241393441e-06,
+      "loss": 53.6581,
+      "step": 25550
+    },
+    {
+      "epoch": 0.10326563427966563,
+      "grad_norm": 486.10650634765625,
+      "learning_rate": 9.658167900655032e-06,
+      "loss": 36.218,
+      "step": 25560
+    },
+    {
+      "epoch": 0.10330603554503327,
+      "grad_norm": 566.1380004882812,
+      "learning_rate": 9.657655189287582e-06,
+      "loss": 44.0325,
+      "step": 25570
+    },
+    {
+      "epoch": 0.1033464368104009,
+      "grad_norm": 264.7056884765625,
+      "learning_rate": 9.657142107331883e-06,
+      "loss": 41.3755,
+      "step": 25580
+    },
+    {
+      "epoch": 0.10338683807576854,
+      "grad_norm": 781.30712890625,
+      "learning_rate": 9.65662865482876e-06,
+      "loss": 51.5806,
+      "step": 25590
+    },
+    {
+      "epoch": 0.10342723934113617,
+      "grad_norm": 468.07794189453125,
+      "learning_rate": 9.656114831819067e-06,
+      "loss": 49.1324,
+      "step": 25600
+    },
+    {
+      "epoch": 0.10346764060650379,
+      "grad_norm": 287.0177917480469,
+      "learning_rate": 9.655600638343685e-06,
+      "loss": 51.1372,
+      "step": 25610
+    },
+    {
+      "epoch": 0.10350804187187143,
+      "grad_norm": 979.0823364257812,
+      "learning_rate": 9.655086074443527e-06,
+      "loss": 46.1757,
+      "step": 25620
+    },
+    {
+      "epoch": 0.10354844313723906,
+      "grad_norm": 790.9038696289062,
+      "learning_rate": 9.654571140159534e-06,
+      "loss": 60.9249,
+      "step": 25630
+    },
+    {
+      "epoch": 0.10358884440260668,
+      "grad_norm": 796.7022705078125,
+      "learning_rate": 9.654055835532676e-06,
+      "loss": 71.7788,
+      "step": 25640
+    },
+    {
+      "epoch": 0.10362924566797432,
+      "grad_norm": 573.487548828125,
+      "learning_rate": 9.653540160603956e-06,
+      "loss": 61.6155,
+      "step": 25650
+    },
+    {
+      "epoch": 0.10366964693334195,
+      "grad_norm": 654.7489013671875,
+      "learning_rate": 9.653024115414402e-06,
+      "loss": 46.4036,
+      "step": 25660
+    },
+    {
+      "epoch": 0.10371004819870959,
+      "grad_norm": 797.578369140625,
+      "learning_rate": 9.652507700005072e-06,
+      "loss": 45.1721,
+      "step": 25670
+    },
+    {
+      "epoch": 0.10375044946407722,
+      "grad_norm": 274.95050048828125,
+      "learning_rate": 9.651990914417057e-06,
+      "loss": 34.9745,
+      "step": 25680
+    },
+    {
+      "epoch": 0.10379085072944484,
+      "grad_norm": 552.0946044921875,
+      "learning_rate": 9.651473758691477e-06,
+      "loss": 53.0471,
+      "step": 25690
+    },
+    {
+      "epoch": 0.10383125199481248,
+      "grad_norm": 955.2523803710938,
+      "learning_rate": 9.650956232869475e-06,
+      "loss": 52.053,
+      "step": 25700
+    },
+    {
+      "epoch": 0.10387165326018011,
+      "grad_norm": 625.6551513671875,
+      "learning_rate": 9.650438336992231e-06,
+      "loss": 56.149,
+      "step": 25710
+    },
+    {
+      "epoch": 0.10391205452554773,
+      "grad_norm": 296.4748229980469,
+      "learning_rate": 9.64992007110095e-06,
+      "loss": 41.5693,
+      "step": 25720
+    },
+    {
+      "epoch": 0.10395245579091537,
+      "grad_norm": 962.1800537109375,
+      "learning_rate": 9.64940143523687e-06,
+      "loss": 86.3123,
+      "step": 25730
+    },
+    {
+      "epoch": 0.103992857056283,
+      "grad_norm": 618.9432373046875,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 49.0561,
+      "step": 25740
+    },
+    {
+      "epoch": 0.10403325832165064,
+      "grad_norm": 846.5286254882812,
+      "learning_rate": 9.648363053755406e-06,
+      "loss": 54.6641,
+      "step": 25750
+    },
+    {
+      "epoch": 0.10407365958701827,
+      "grad_norm": 963.1185913085938,
+      "learning_rate": 9.647843308220636e-06,
+      "loss": 55.3922,
+      "step": 25760
+    },
+    {
+      "epoch": 0.10411406085238589,
+      "grad_norm": 1572.095703125,
+      "learning_rate": 9.647323192878306e-06,
+      "loss": 59.8284,
+      "step": 25770
+    },
+    {
+      "epoch": 0.10415446211775353,
+      "grad_norm": 809.3494262695312,
+      "learning_rate": 9.646802707769798e-06,
+      "loss": 42.1801,
+      "step": 25780
+    },
+    {
+      "epoch": 0.10419486338312116,
+      "grad_norm": 745.4800415039062,
+      "learning_rate": 9.646281852936525e-06,
+      "loss": 42.9003,
+      "step": 25790
+    },
+    {
+      "epoch": 0.10423526464848878,
+      "grad_norm": 335.8274841308594,
+      "learning_rate": 9.64576062841993e-06,
+      "loss": 51.5139,
+      "step": 25800
+    },
+    {
+      "epoch": 0.10427566591385642,
+      "grad_norm": 529.6254272460938,
+      "learning_rate": 9.64523903426148e-06,
+      "loss": 40.3054,
+      "step": 25810
+    },
+    {
+      "epoch": 0.10431606717922405,
+      "grad_norm": 672.9226684570312,
+      "learning_rate": 9.64471707050268e-06,
+      "loss": 52.9082,
+      "step": 25820
+    },
+    {
+      "epoch": 0.10435646844459169,
+      "grad_norm": 506.335693359375,
+      "learning_rate": 9.644194737185058e-06,
+      "loss": 62.1664,
+      "step": 25830
+    },
+    {
+      "epoch": 0.10439686970995932,
+      "grad_norm": 320.349853515625,
+      "learning_rate": 9.643672034350177e-06,
+      "loss": 56.1261,
+      "step": 25840
+    },
+    {
+      "epoch": 0.10443727097532694,
+      "grad_norm": 425.2008056640625,
+      "learning_rate": 9.643148962039622e-06,
+      "loss": 46.4004,
+      "step": 25850
+    },
+    {
+      "epoch": 0.10447767224069458,
+      "grad_norm": 435.9482727050781,
+      "learning_rate": 9.642625520295014e-06,
+      "loss": 46.6332,
+      "step": 25860
+    },
+    {
+      "epoch": 0.10451807350606221,
+      "grad_norm": 808.4017333984375,
+      "learning_rate": 9.642101709158001e-06,
+      "loss": 50.6643,
+      "step": 25870
+    },
+    {
+      "epoch": 0.10455847477142984,
+      "grad_norm": 464.9044494628906,
+      "learning_rate": 9.641577528670257e-06,
+      "loss": 55.6681,
+      "step": 25880
+    },
+    {
+      "epoch": 0.10459887603679748,
+      "grad_norm": 979.295166015625,
+      "learning_rate": 9.641052978873494e-06,
+      "loss": 68.0048,
+      "step": 25890
+    },
+    {
+      "epoch": 0.1046392773021651,
+      "grad_norm": 732.1350708007812,
+      "learning_rate": 9.640528059809442e-06,
+      "loss": 57.266,
+      "step": 25900
+    },
+    {
+      "epoch": 0.10467967856753274,
+      "grad_norm": 1971.1300048828125,
+      "learning_rate": 9.640002771519872e-06,
+      "loss": 54.0411,
+      "step": 25910
+    },
+    {
+      "epoch": 0.10472007983290037,
+      "grad_norm": 994.9387817382812,
+      "learning_rate": 9.639477114046575e-06,
+      "loss": 42.02,
+      "step": 25920
+    },
+    {
+      "epoch": 0.104760481098268,
+      "grad_norm": 358.152587890625,
+      "learning_rate": 9.638951087431376e-06,
+      "loss": 40.5045,
+      "step": 25930
+    },
+    {
+      "epoch": 0.10480088236363563,
+      "grad_norm": 386.483154296875,
+      "learning_rate": 9.638424691716129e-06,
+      "loss": 66.2506,
+      "step": 25940
+    },
+    {
+      "epoch": 0.10484128362900326,
+      "grad_norm": 737.9490356445312,
+      "learning_rate": 9.637897926942716e-06,
+      "loss": 69.654,
+      "step": 25950
+    },
+    {
+      "epoch": 0.10488168489437089,
+      "grad_norm": 597.8994750976562,
+      "learning_rate": 9.637370793153051e-06,
+      "loss": 50.6889,
+      "step": 25960
+    },
+    {
+      "epoch": 0.10492208615973853,
+      "grad_norm": 912.8892822265625,
+      "learning_rate": 9.636843290389076e-06,
+      "loss": 53.4189,
+      "step": 25970
+    },
+    {
+      "epoch": 0.10496248742510615,
+      "grad_norm": 1086.555419921875,
+      "learning_rate": 9.636315418692759e-06,
+      "loss": 62.2072,
+      "step": 25980
+    },
+    {
+      "epoch": 0.10500288869047379,
+      "grad_norm": 701.8748779296875,
+      "learning_rate": 9.635787178106102e-06,
+      "loss": 48.1612,
+      "step": 25990
+    },
+    {
+      "epoch": 0.10504328995584142,
+      "grad_norm": 594.7928466796875,
+      "learning_rate": 9.635258568671135e-06,
+      "loss": 39.6997,
+      "step": 26000
+    },
+    {
+      "epoch": 0.10508369122120904,
+      "grad_norm": 475.304443359375,
+      "learning_rate": 9.634729590429917e-06,
+      "loss": 42.5213,
+      "step": 26010
+    },
+    {
+      "epoch": 0.10512409248657668,
+      "grad_norm": 553.9869995117188,
+      "learning_rate": 9.634200243424535e-06,
+      "loss": 40.3878,
+      "step": 26020
+    },
+    {
+      "epoch": 0.10516449375194431,
+      "grad_norm": 477.8350830078125,
+      "learning_rate": 9.633670527697108e-06,
+      "loss": 52.4336,
+      "step": 26030
+    },
+    {
+      "epoch": 0.10520489501731194,
+      "grad_norm": 876.521484375,
+      "learning_rate": 9.633140443289784e-06,
+      "loss": 49.6513,
+      "step": 26040
+    },
+    {
+      "epoch": 0.10524529628267958,
+      "grad_norm": 918.1558227539062,
+      "learning_rate": 9.632609990244737e-06,
+      "loss": 45.0479,
+      "step": 26050
+    },
+    {
+      "epoch": 0.1052856975480472,
+      "grad_norm": 1112.887939453125,
+      "learning_rate": 9.632079168604175e-06,
+      "loss": 48.0344,
+      "step": 26060
+    },
+    {
+      "epoch": 0.10532609881341484,
+      "grad_norm": 494.50140380859375,
+      "learning_rate": 9.63154797841033e-06,
+      "loss": 43.9376,
+      "step": 26070
+    },
+    {
+      "epoch": 0.10536650007878247,
+      "grad_norm": 639.7578735351562,
+      "learning_rate": 9.63101641970547e-06,
+      "loss": 57.9903,
+      "step": 26080
+    },
+    {
+      "epoch": 0.1054069013441501,
+      "grad_norm": 959.657470703125,
+      "learning_rate": 9.630484492531886e-06,
+      "loss": 50.5194,
+      "step": 26090
+    },
+    {
+      "epoch": 0.10544730260951773,
+      "grad_norm": 541.671875,
+      "learning_rate": 9.629952196931902e-06,
+      "loss": 29.0569,
+      "step": 26100
+    },
+    {
+      "epoch": 0.10548770387488536,
+      "grad_norm": 867.1376342773438,
+      "learning_rate": 9.629419532947872e-06,
+      "loss": 45.2668,
+      "step": 26110
+    },
+    {
+      "epoch": 0.10552810514025299,
+      "grad_norm": 859.1763305664062,
+      "learning_rate": 9.628886500622174e-06,
+      "loss": 38.3891,
+      "step": 26120
+    },
+    {
+      "epoch": 0.10556850640562063,
+      "grad_norm": 656.4653930664062,
+      "learning_rate": 9.62835309999722e-06,
+      "loss": 39.2543,
+      "step": 26130
+    },
+    {
+      "epoch": 0.10560890767098825,
+      "grad_norm": 630.9413452148438,
+      "learning_rate": 9.627819331115453e-06,
+      "loss": 45.4982,
+      "step": 26140
+    },
+    {
+      "epoch": 0.10564930893635589,
+      "grad_norm": 902.5018310546875,
+      "learning_rate": 9.627285194019342e-06,
+      "loss": 76.9234,
+      "step": 26150
+    },
+    {
+      "epoch": 0.10568971020172352,
+      "grad_norm": 460.1519775390625,
+      "learning_rate": 9.626750688751382e-06,
+      "loss": 49.7342,
+      "step": 26160
+    },
+    {
+      "epoch": 0.10573011146709115,
+      "grad_norm": 1033.299560546875,
+      "learning_rate": 9.626215815354104e-06,
+      "loss": 61.9341,
+      "step": 26170
+    },
+    {
+      "epoch": 0.10577051273245879,
+      "grad_norm": 726.0341186523438,
+      "learning_rate": 9.625680573870067e-06,
+      "loss": 42.1929,
+      "step": 26180
+    },
+    {
+      "epoch": 0.10581091399782641,
+      "grad_norm": 466.17462158203125,
+      "learning_rate": 9.625144964341853e-06,
+      "loss": 42.927,
+      "step": 26190
+    },
+    {
+      "epoch": 0.10585131526319404,
+      "grad_norm": 163.19650268554688,
+      "learning_rate": 9.624608986812082e-06,
+      "loss": 50.3991,
+      "step": 26200
+    },
+    {
+      "epoch": 0.10589171652856168,
+      "grad_norm": 844.4330444335938,
+      "learning_rate": 9.624072641323398e-06,
+      "loss": 47.5053,
+      "step": 26210
+    },
+    {
+      "epoch": 0.1059321177939293,
+      "grad_norm": 477.6076354980469,
+      "learning_rate": 9.623535927918474e-06,
+      "loss": 54.595,
+      "step": 26220
+    },
+    {
+      "epoch": 0.10597251905929694,
+      "grad_norm": 629.3731079101562,
+      "learning_rate": 9.622998846640018e-06,
+      "loss": 51.9192,
+      "step": 26230
+    },
+    {
+      "epoch": 0.10601292032466457,
+      "grad_norm": 1155.002197265625,
+      "learning_rate": 9.62246139753076e-06,
+      "loss": 39.1533,
+      "step": 26240
+    },
+    {
+      "epoch": 0.1060533215900322,
+      "grad_norm": 685.2892456054688,
+      "learning_rate": 9.621923580633462e-06,
+      "loss": 25.9385,
+      "step": 26250
+    },
+    {
+      "epoch": 0.10609372285539984,
+      "grad_norm": 351.75091552734375,
+      "learning_rate": 9.621385395990915e-06,
+      "loss": 33.7549,
+      "step": 26260
+    },
+    {
+      "epoch": 0.10613412412076746,
+      "grad_norm": 994.1544189453125,
+      "learning_rate": 9.620846843645944e-06,
+      "loss": 65.0634,
+      "step": 26270
+    },
+    {
+      "epoch": 0.10617452538613509,
+      "grad_norm": 448.69256591796875,
+      "learning_rate": 9.620307923641395e-06,
+      "loss": 33.9385,
+      "step": 26280
+    },
+    {
+      "epoch": 0.10621492665150273,
+      "grad_norm": 653.5578002929688,
+      "learning_rate": 9.61976863602015e-06,
+      "loss": 64.1324,
+      "step": 26290
+    },
+    {
+      "epoch": 0.10625532791687035,
+      "grad_norm": 1237.69140625,
+      "learning_rate": 9.619228980825114e-06,
+      "loss": 57.0205,
+      "step": 26300
+    },
+    {
+      "epoch": 0.106295729182238,
+      "grad_norm": 518.8150634765625,
+      "learning_rate": 9.61868895809923e-06,
+      "loss": 42.3386,
+      "step": 26310
+    },
+    {
+      "epoch": 0.10633613044760562,
+      "grad_norm": 1088.1500244140625,
+      "learning_rate": 9.618148567885462e-06,
+      "loss": 38.555,
+      "step": 26320
+    },
+    {
+      "epoch": 0.10637653171297325,
+      "grad_norm": 643.849365234375,
+      "learning_rate": 9.617607810226806e-06,
+      "loss": 61.6261,
+      "step": 26330
+    },
+    {
+      "epoch": 0.10641693297834089,
+      "grad_norm": 997.7747192382812,
+      "learning_rate": 9.61706668516629e-06,
+      "loss": 51.5593,
+      "step": 26340
+    },
+    {
+      "epoch": 0.10645733424370851,
+      "grad_norm": 939.482177734375,
+      "learning_rate": 9.616525192746965e-06,
+      "loss": 62.9643,
+      "step": 26350
+    },
+    {
+      "epoch": 0.10649773550907614,
+      "grad_norm": 840.9242553710938,
+      "learning_rate": 9.61598333301192e-06,
+      "loss": 49.4016,
+      "step": 26360
+    },
+    {
+      "epoch": 0.10653813677444378,
+      "grad_norm": 824.6075439453125,
+      "learning_rate": 9.615441106004264e-06,
+      "loss": 40.5007,
+      "step": 26370
+    },
+    {
+      "epoch": 0.1065785380398114,
+      "grad_norm": 383.22613525390625,
+      "learning_rate": 9.614898511767142e-06,
+      "loss": 42.2523,
+      "step": 26380
+    },
+    {
+      "epoch": 0.10661893930517904,
+      "grad_norm": 478.3547668457031,
+      "learning_rate": 9.614355550343724e-06,
+      "loss": 56.1983,
+      "step": 26390
+    },
+    {
+      "epoch": 0.10665934057054667,
+      "grad_norm": 608.23291015625,
+      "learning_rate": 9.613812221777212e-06,
+      "loss": 46.3041,
+      "step": 26400
+    },
+    {
+      "epoch": 0.1066997418359143,
+      "grad_norm": 820.3971557617188,
+      "learning_rate": 9.613268526110838e-06,
+      "loss": 48.1949,
+      "step": 26410
+    },
+    {
+      "epoch": 0.10674014310128194,
+      "grad_norm": 712.67529296875,
+      "learning_rate": 9.612724463387857e-06,
+      "loss": 67.0135,
+      "step": 26420
+    },
+    {
+      "epoch": 0.10678054436664956,
+      "grad_norm": 780.4419555664062,
+      "learning_rate": 9.612180033651561e-06,
+      "loss": 49.2101,
+      "step": 26430
+    },
+    {
+      "epoch": 0.10682094563201719,
+      "grad_norm": 1020.588623046875,
+      "learning_rate": 9.611635236945267e-06,
+      "loss": 65.4931,
+      "step": 26440
+    },
+    {
+      "epoch": 0.10686134689738483,
+      "grad_norm": 933.3222045898438,
+      "learning_rate": 9.61109007331232e-06,
+      "loss": 60.496,
+      "step": 26450
+    },
+    {
+      "epoch": 0.10690174816275246,
+      "grad_norm": 775.654296875,
+      "learning_rate": 9.610544542796101e-06,
+      "loss": 54.8142,
+      "step": 26460
+    },
+    {
+      "epoch": 0.1069421494281201,
+      "grad_norm": 244.994384765625,
+      "learning_rate": 9.609998645440011e-06,
+      "loss": 37.4775,
+      "step": 26470
+    },
+    {
+      "epoch": 0.10698255069348772,
+      "grad_norm": 1276.1300048828125,
+      "learning_rate": 9.609452381287486e-06,
+      "loss": 52.7323,
+      "step": 26480
+    },
+    {
+      "epoch": 0.10702295195885535,
+      "grad_norm": 626.2915649414062,
+      "learning_rate": 9.608905750381988e-06,
+      "loss": 65.2879,
+      "step": 26490
+    },
+    {
+      "epoch": 0.10706335322422299,
+      "grad_norm": 544.8095092773438,
+      "learning_rate": 9.608358752767013e-06,
+      "loss": 41.748,
+      "step": 26500
+    },
+    {
+      "epoch": 0.10710375448959061,
+      "grad_norm": 863.2129516601562,
+      "learning_rate": 9.60781138848608e-06,
+      "loss": 68.7763,
+      "step": 26510
+    },
+    {
+      "epoch": 0.10714415575495824,
+      "grad_norm": 262.70263671875,
+      "learning_rate": 9.607263657582744e-06,
+      "loss": 39.4715,
+      "step": 26520
+    },
+    {
+      "epoch": 0.10718455702032588,
+      "grad_norm": 399.425537109375,
+      "learning_rate": 9.60671556010058e-06,
+      "loss": 50.2638,
+      "step": 26530
+    },
+    {
+      "epoch": 0.1072249582856935,
+      "grad_norm": 637.1754150390625,
+      "learning_rate": 9.606167096083205e-06,
+      "loss": 51.3568,
+      "step": 26540
+    },
+    {
+      "epoch": 0.10726535955106115,
+      "grad_norm": 692.915283203125,
+      "learning_rate": 9.60561826557425e-06,
+      "loss": 43.2627,
+      "step": 26550
+    },
+    {
+      "epoch": 0.10730576081642877,
+      "grad_norm": 634.8587646484375,
+      "learning_rate": 9.60506906861739e-06,
+      "loss": 37.3307,
+      "step": 26560
+    },
+    {
+      "epoch": 0.1073461620817964,
+      "grad_norm": 641.59228515625,
+      "learning_rate": 9.604519505256316e-06,
+      "loss": 57.2253,
+      "step": 26570
+    },
+    {
+      "epoch": 0.10738656334716404,
+      "grad_norm": 755.43115234375,
+      "learning_rate": 9.603969575534757e-06,
+      "loss": 65.6552,
+      "step": 26580
+    },
+    {
+      "epoch": 0.10742696461253166,
+      "grad_norm": 508.3285827636719,
+      "learning_rate": 9.60341927949647e-06,
+      "loss": 62.2325,
+      "step": 26590
+    },
+    {
+      "epoch": 0.10746736587789929,
+      "grad_norm": 1276.345703125,
+      "learning_rate": 9.602868617185238e-06,
+      "loss": 53.7871,
+      "step": 26600
+    },
+    {
+      "epoch": 0.10750776714326693,
+      "grad_norm": 620.2528686523438,
+      "learning_rate": 9.602317588644872e-06,
+      "loss": 59.2076,
+      "step": 26610
+    },
+    {
+      "epoch": 0.10754816840863456,
+      "grad_norm": 284.98577880859375,
+      "learning_rate": 9.601766193919217e-06,
+      "loss": 34.0635,
+      "step": 26620
+    },
+    {
+      "epoch": 0.1075885696740022,
+      "grad_norm": 777.9571533203125,
+      "learning_rate": 9.601214433052147e-06,
+      "loss": 50.2766,
+      "step": 26630
+    },
+    {
+      "epoch": 0.10762897093936982,
+      "grad_norm": 818.9276733398438,
+      "learning_rate": 9.600662306087562e-06,
+      "loss": 62.3097,
+      "step": 26640
+    },
+    {
+      "epoch": 0.10766937220473745,
+      "grad_norm": 866.0750122070312,
+      "learning_rate": 9.600109813069389e-06,
+      "loss": 35.7204,
+      "step": 26650
+    },
+    {
+      "epoch": 0.10770977347010509,
+      "grad_norm": 968.46484375,
+      "learning_rate": 9.599556954041591e-06,
+      "loss": 57.4643,
+      "step": 26660
+    },
+    {
+      "epoch": 0.10775017473547271,
+      "grad_norm": 1058.4351806640625,
+      "learning_rate": 9.599003729048157e-06,
+      "loss": 48.7317,
+      "step": 26670
+    },
+    {
+      "epoch": 0.10779057600084034,
+      "grad_norm": 965.6939086914062,
+      "learning_rate": 9.598450138133101e-06,
+      "loss": 36.7754,
+      "step": 26680
+    },
+    {
+      "epoch": 0.10783097726620798,
+      "grad_norm": 713.4266967773438,
+      "learning_rate": 9.597896181340471e-06,
+      "loss": 44.2337,
+      "step": 26690
+    },
+    {
+      "epoch": 0.1078713785315756,
+      "grad_norm": 731.5632934570312,
+      "learning_rate": 9.597341858714344e-06,
+      "loss": 56.622,
+      "step": 26700
+    },
+    {
+      "epoch": 0.10791177979694325,
+      "grad_norm": 828.4664306640625,
+      "learning_rate": 9.596787170298824e-06,
+      "loss": 45.5558,
+      "step": 26710
+    },
+    {
+      "epoch": 0.10795218106231087,
+      "grad_norm": 523.5830688476562,
+      "learning_rate": 9.596232116138047e-06,
+      "loss": 45.0617,
+      "step": 26720
+    },
+    {
+      "epoch": 0.1079925823276785,
+      "grad_norm": 1065.1573486328125,
+      "learning_rate": 9.595676696276173e-06,
+      "loss": 50.7657,
+      "step": 26730
+    },
+    {
+      "epoch": 0.10803298359304614,
+      "grad_norm": 664.0277709960938,
+      "learning_rate": 9.595120910757396e-06,
+      "loss": 63.9061,
+      "step": 26740
+    },
+    {
+      "epoch": 0.10807338485841377,
+      "grad_norm": 518.4659423828125,
+      "learning_rate": 9.594564759625936e-06,
+      "loss": 48.3799,
+      "step": 26750
+    },
+    {
+      "epoch": 0.10811378612378139,
+      "grad_norm": 591.0075073242188,
+      "learning_rate": 9.594008242926046e-06,
+      "loss": 52.9573,
+      "step": 26760
+    },
+    {
+      "epoch": 0.10815418738914903,
+      "grad_norm": 374.7685852050781,
+      "learning_rate": 9.593451360702003e-06,
+      "loss": 49.838,
+      "step": 26770
+    },
+    {
+      "epoch": 0.10819458865451666,
+      "grad_norm": 893.8427734375,
+      "learning_rate": 9.592894112998115e-06,
+      "loss": 61.3094,
+      "step": 26780
+    },
+    {
+      "epoch": 0.1082349899198843,
+      "grad_norm": 234.3647003173828,
+      "learning_rate": 9.592336499858721e-06,
+      "loss": 64.3124,
+      "step": 26790
+    },
+    {
+      "epoch": 0.10827539118525192,
+      "grad_norm": 341.9205627441406,
+      "learning_rate": 9.59177852132819e-06,
+      "loss": 52.1276,
+      "step": 26800
+    },
+    {
+      "epoch": 0.10831579245061955,
+      "grad_norm": 509.6340026855469,
+      "learning_rate": 9.591220177450912e-06,
+      "loss": 51.8163,
+      "step": 26810
+    },
+    {
+      "epoch": 0.10835619371598719,
+      "grad_norm": 896.3446655273438,
+      "learning_rate": 9.590661468271319e-06,
+      "loss": 49.8318,
+      "step": 26820
+    },
+    {
+      "epoch": 0.10839659498135482,
+      "grad_norm": 545.89794921875,
+      "learning_rate": 9.59010239383386e-06,
+      "loss": 56.6707,
+      "step": 26830
+    },
+    {
+      "epoch": 0.10843699624672244,
+      "grad_norm": 472.1811828613281,
+      "learning_rate": 9.589542954183018e-06,
+      "loss": 39.5817,
+      "step": 26840
+    },
+    {
+      "epoch": 0.10847739751209008,
+      "grad_norm": 374.705322265625,
+      "learning_rate": 9.588983149363307e-06,
+      "loss": 47.7198,
+      "step": 26850
+    },
+    {
+      "epoch": 0.10851779877745771,
+      "grad_norm": 537.0018310546875,
+      "learning_rate": 9.588422979419267e-06,
+      "loss": 59.9525,
+      "step": 26860
+    },
+    {
+      "epoch": 0.10855820004282535,
+      "grad_norm": 441.2084655761719,
+      "learning_rate": 9.587862444395471e-06,
+      "loss": 37.375,
+      "step": 26870
+    },
+    {
+      "epoch": 0.10859860130819297,
+      "grad_norm": 827.1460571289062,
+      "learning_rate": 9.587301544336513e-06,
+      "loss": 51.7789,
+      "step": 26880
+    },
+    {
+      "epoch": 0.1086390025735606,
+      "grad_norm": 621.7218627929688,
+      "learning_rate": 9.586740279287024e-06,
+      "loss": 48.793,
+      "step": 26890
+    },
+    {
+      "epoch": 0.10867940383892824,
+      "grad_norm": 1153.9176025390625,
+      "learning_rate": 9.586178649291664e-06,
+      "loss": 74.7823,
+      "step": 26900
+    },
+    {
+      "epoch": 0.10871980510429587,
+      "grad_norm": 1106.1214599609375,
+      "learning_rate": 9.585616654395113e-06,
+      "loss": 70.2931,
+      "step": 26910
+    },
+    {
+      "epoch": 0.10876020636966349,
+      "grad_norm": 258.2606201171875,
+      "learning_rate": 9.585054294642093e-06,
+      "loss": 47.435,
+      "step": 26920
+    },
+    {
+      "epoch": 0.10880060763503113,
+      "grad_norm": 509.2825622558594,
+      "learning_rate": 9.584491570077343e-06,
+      "loss": 59.2019,
+      "step": 26930
+    },
+    {
+      "epoch": 0.10884100890039876,
+      "grad_norm": 454.8273620605469,
+      "learning_rate": 9.58392848074564e-06,
+      "loss": 40.4965,
+      "step": 26940
+    },
+    {
+      "epoch": 0.1088814101657664,
+      "grad_norm": 216.57943725585938,
+      "learning_rate": 9.583365026691785e-06,
+      "loss": 51.4199,
+      "step": 26950
+    },
+    {
+      "epoch": 0.10892181143113402,
+      "grad_norm": 636.304931640625,
+      "learning_rate": 9.58280120796061e-06,
+      "loss": 36.1534,
+      "step": 26960
+    },
+    {
+      "epoch": 0.10896221269650165,
+      "grad_norm": 385.84832763671875,
+      "learning_rate": 9.582237024596974e-06,
+      "loss": 53.6006,
+      "step": 26970
+    },
+    {
+      "epoch": 0.10900261396186929,
+      "grad_norm": 586.0780639648438,
+      "learning_rate": 9.581672476645768e-06,
+      "loss": 50.0682,
+      "step": 26980
+    },
+    {
+      "epoch": 0.10904301522723692,
+      "grad_norm": 331.61419677734375,
+      "learning_rate": 9.58110756415191e-06,
+      "loss": 54.5409,
+      "step": 26990
+    },
+    {
+      "epoch": 0.10908341649260454,
+      "grad_norm": 237.11599731445312,
+      "learning_rate": 9.580542287160348e-06,
+      "loss": 39.9448,
+      "step": 27000
+    },
+    {
+      "epoch": 0.10912381775797218,
+      "grad_norm": 481.5954284667969,
+      "learning_rate": 9.579976645716058e-06,
+      "loss": 61.8992,
+      "step": 27010
+    },
+    {
+      "epoch": 0.10916421902333981,
+      "grad_norm": 654.479248046875,
+      "learning_rate": 9.579410639864046e-06,
+      "loss": 67.2557,
+      "step": 27020
+    },
+    {
+      "epoch": 0.10920462028870745,
+      "grad_norm": 524.7271728515625,
+      "learning_rate": 9.578844269649345e-06,
+      "loss": 66.595,
+      "step": 27030
+    },
+    {
+      "epoch": 0.10924502155407508,
+      "grad_norm": 451.2859802246094,
+      "learning_rate": 9.578277535117022e-06,
+      "loss": 47.5412,
+      "step": 27040
+    },
+    {
+      "epoch": 0.1092854228194427,
+      "grad_norm": 448.099853515625,
+      "learning_rate": 9.577710436312164e-06,
+      "loss": 58.6692,
+      "step": 27050
+    },
+    {
+      "epoch": 0.10932582408481034,
+      "grad_norm": 505.60125732421875,
+      "learning_rate": 9.577142973279896e-06,
+      "loss": 42.8677,
+      "step": 27060
+    },
+    {
+      "epoch": 0.10936622535017797,
+      "grad_norm": 724.5872192382812,
+      "learning_rate": 9.576575146065369e-06,
+      "loss": 61.896,
+      "step": 27070
+    },
+    {
+      "epoch": 0.1094066266155456,
+      "grad_norm": 1077.6009521484375,
+      "learning_rate": 9.576006954713762e-06,
+      "loss": 60.5044,
+      "step": 27080
+    },
+    {
+      "epoch": 0.10944702788091323,
+      "grad_norm": 604.5072631835938,
+      "learning_rate": 9.57543839927028e-06,
+      "loss": 44.5489,
+      "step": 27090
+    },
+    {
+      "epoch": 0.10948742914628086,
+      "grad_norm": 524.4970703125,
+      "learning_rate": 9.574869479780165e-06,
+      "loss": 85.2106,
+      "step": 27100
+    },
+    {
+      "epoch": 0.1095278304116485,
+      "grad_norm": 732.091796875,
+      "learning_rate": 9.57430019628868e-06,
+      "loss": 36.05,
+      "step": 27110
+    },
+    {
+      "epoch": 0.10956823167701613,
+      "grad_norm": 695.1461181640625,
+      "learning_rate": 9.573730548841122e-06,
+      "loss": 35.9356,
+      "step": 27120
+    },
+    {
+      "epoch": 0.10960863294238375,
+      "grad_norm": 511.49029541015625,
+      "learning_rate": 9.573160537482816e-06,
+      "loss": 46.0325,
+      "step": 27130
+    },
+    {
+      "epoch": 0.10964903420775139,
+      "grad_norm": 2797.2470703125,
+      "learning_rate": 9.572590162259112e-06,
+      "loss": 56.4708,
+      "step": 27140
+    },
+    {
+      "epoch": 0.10968943547311902,
+      "grad_norm": 478.7276611328125,
+      "learning_rate": 9.572019423215395e-06,
+      "loss": 52.5807,
+      "step": 27150
+    },
+    {
+      "epoch": 0.10972983673848664,
+      "grad_norm": 350.75634765625,
+      "learning_rate": 9.571448320397076e-06,
+      "loss": 68.0859,
+      "step": 27160
+    },
+    {
+      "epoch": 0.10977023800385428,
+      "grad_norm": 709.3019409179688,
+      "learning_rate": 9.570876853849593e-06,
+      "loss": 68.0508,
+      "step": 27170
+    },
+    {
+      "epoch": 0.10981063926922191,
+      "grad_norm": 629.4264526367188,
+      "learning_rate": 9.570305023618417e-06,
+      "loss": 42.6662,
+      "step": 27180
+    },
+    {
+      "epoch": 0.10985104053458955,
+      "grad_norm": 747.0570678710938,
+      "learning_rate": 9.569732829749045e-06,
+      "loss": 39.6871,
+      "step": 27190
+    },
+    {
+      "epoch": 0.10989144179995718,
+      "grad_norm": 605.722412109375,
+      "learning_rate": 9.569160272287003e-06,
+      "loss": 40.4398,
+      "step": 27200
+    },
+    {
+      "epoch": 0.1099318430653248,
+      "grad_norm": 266.65997314453125,
+      "learning_rate": 9.56858735127785e-06,
+      "loss": 42.8571,
+      "step": 27210
+    },
+    {
+      "epoch": 0.10997224433069244,
+      "grad_norm": 1407.9671630859375,
+      "learning_rate": 9.568014066767166e-06,
+      "loss": 67.4599,
+      "step": 27220
+    },
+    {
+      "epoch": 0.11001264559606007,
+      "grad_norm": 347.4352722167969,
+      "learning_rate": 9.567440418800569e-06,
+      "loss": 39.2598,
+      "step": 27230
+    },
+    {
+      "epoch": 0.1100530468614277,
+      "grad_norm": 843.9002685546875,
+      "learning_rate": 9.566866407423698e-06,
+      "loss": 48.4884,
+      "step": 27240
+    },
+    {
+      "epoch": 0.11009344812679533,
+      "grad_norm": 428.3054504394531,
+      "learning_rate": 9.566292032682228e-06,
+      "loss": 36.8349,
+      "step": 27250
+    },
+    {
+      "epoch": 0.11013384939216296,
+      "grad_norm": 707.8931274414062,
+      "learning_rate": 9.565717294621856e-06,
+      "loss": 59.8093,
+      "step": 27260
+    },
+    {
+      "epoch": 0.1101742506575306,
+      "grad_norm": 828.3231811523438,
+      "learning_rate": 9.565142193288313e-06,
+      "loss": 51.0172,
+      "step": 27270
+    },
+    {
+      "epoch": 0.11021465192289823,
+      "grad_norm": 589.5343627929688,
+      "learning_rate": 9.564566728727358e-06,
+      "loss": 40.5171,
+      "step": 27280
+    },
+    {
+      "epoch": 0.11025505318826585,
+      "grad_norm": 1542.44140625,
+      "learning_rate": 9.563990900984775e-06,
+      "loss": 40.179,
+      "step": 27290
+    },
+    {
+      "epoch": 0.11029545445363349,
+      "grad_norm": 592.8526611328125,
+      "learning_rate": 9.563414710106382e-06,
+      "loss": 53.8067,
+      "step": 27300
+    },
+    {
+      "epoch": 0.11033585571900112,
+      "grad_norm": 588.4796752929688,
+      "learning_rate": 9.562838156138025e-06,
+      "loss": 41.335,
+      "step": 27310
+    },
+    {
+      "epoch": 0.11037625698436875,
+      "grad_norm": 491.3796081542969,
+      "learning_rate": 9.562261239125575e-06,
+      "loss": 64.4173,
+      "step": 27320
+    },
+    {
+      "epoch": 0.11041665824973639,
+      "grad_norm": 712.91796875,
+      "learning_rate": 9.561683959114938e-06,
+      "loss": 45.0702,
+      "step": 27330
+    },
+    {
+      "epoch": 0.11045705951510401,
+      "grad_norm": 844.9070434570312,
+      "learning_rate": 9.561106316152043e-06,
+      "loss": 64.8769,
+      "step": 27340
+    },
+    {
+      "epoch": 0.11049746078047164,
+      "grad_norm": 886.7588500976562,
+      "learning_rate": 9.56052831028285e-06,
+      "loss": 68.3311,
+      "step": 27350
+    },
+    {
+      "epoch": 0.11053786204583928,
+      "grad_norm": 494.68353271484375,
+      "learning_rate": 9.559949941553351e-06,
+      "loss": 100.2485,
+      "step": 27360
+    },
+    {
+      "epoch": 0.1105782633112069,
+      "grad_norm": 414.14691162109375,
+      "learning_rate": 9.559371210009562e-06,
+      "loss": 59.7353,
+      "step": 27370
+    },
+    {
+      "epoch": 0.11061866457657454,
+      "grad_norm": 317.25750732421875,
+      "learning_rate": 9.55879211569753e-06,
+      "loss": 42.0364,
+      "step": 27380
+    },
+    {
+      "epoch": 0.11065906584194217,
+      "grad_norm": 559.4423828125,
+      "learning_rate": 9.55821265866333e-06,
+      "loss": 57.4018,
+      "step": 27390
+    },
+    {
+      "epoch": 0.1106994671073098,
+      "grad_norm": 521.9277954101562,
+      "learning_rate": 9.55763283895307e-06,
+      "loss": 73.1083,
+      "step": 27400
+    },
+    {
+      "epoch": 0.11073986837267744,
+      "grad_norm": 671.8775634765625,
+      "learning_rate": 9.557052656612882e-06,
+      "loss": 60.0441,
+      "step": 27410
+    },
+    {
+      "epoch": 0.11078026963804506,
+      "grad_norm": 3136.1943359375,
+      "learning_rate": 9.556472111688928e-06,
+      "loss": 67.2099,
+      "step": 27420
+    },
+    {
+      "epoch": 0.11082067090341269,
+      "grad_norm": 810.9188842773438,
+      "learning_rate": 9.555891204227399e-06,
+      "loss": 47.5018,
+      "step": 27430
+    },
+    {
+      "epoch": 0.11086107216878033,
+      "grad_norm": 692.3886108398438,
+      "learning_rate": 9.555309934274515e-06,
+      "loss": 39.6191,
+      "step": 27440
+    },
+    {
+      "epoch": 0.11090147343414795,
+      "grad_norm": 672.2152099609375,
+      "learning_rate": 9.554728301876525e-06,
+      "loss": 56.8934,
+      "step": 27450
+    },
+    {
+      "epoch": 0.1109418746995156,
+      "grad_norm": 684.3619995117188,
+      "learning_rate": 9.554146307079711e-06,
+      "loss": 54.4783,
+      "step": 27460
+    },
+    {
+      "epoch": 0.11098227596488322,
+      "grad_norm": 448.3148498535156,
+      "learning_rate": 9.553563949930374e-06,
+      "loss": 53.4588,
+      "step": 27470
+    },
+    {
+      "epoch": 0.11102267723025085,
+      "grad_norm": 957.7836303710938,
+      "learning_rate": 9.552981230474849e-06,
+      "loss": 58.2424,
+      "step": 27480
+    },
+    {
+      "epoch": 0.11106307849561849,
+      "grad_norm": 363.869384765625,
+      "learning_rate": 9.552398148759506e-06,
+      "loss": 39.7609,
+      "step": 27490
+    },
+    {
+      "epoch": 0.11110347976098611,
+      "grad_norm": 365.90118408203125,
+      "learning_rate": 9.551814704830734e-06,
+      "loss": 36.9783,
+      "step": 27500
+    },
+    {
+      "epoch": 0.11114388102635374,
+      "grad_norm": 503.6572570800781,
+      "learning_rate": 9.551230898734955e-06,
+      "loss": 30.8108,
+      "step": 27510
+    },
+    {
+      "epoch": 0.11118428229172138,
+      "grad_norm": 128.0505828857422,
+      "learning_rate": 9.550646730518623e-06,
+      "loss": 48.5108,
+      "step": 27520
+    },
+    {
+      "epoch": 0.111224683557089,
+      "grad_norm": 435.7357177734375,
+      "learning_rate": 9.550062200228214e-06,
+      "loss": 35.7215,
+      "step": 27530
+    },
+    {
+      "epoch": 0.11126508482245664,
+      "grad_norm": 975.533447265625,
+      "learning_rate": 9.549477307910238e-06,
+      "loss": 49.5548,
+      "step": 27540
+    },
+    {
+      "epoch": 0.11130548608782427,
+      "grad_norm": 533.999267578125,
+      "learning_rate": 9.548892053611232e-06,
+      "loss": 67.928,
+      "step": 27550
+    },
+    {
+      "epoch": 0.1113458873531919,
+      "grad_norm": 621.736328125,
+      "learning_rate": 9.54830643737776e-06,
+      "loss": 35.6264,
+      "step": 27560
+    },
+    {
+      "epoch": 0.11138628861855954,
+      "grad_norm": 805.3984985351562,
+      "learning_rate": 9.54772045925642e-06,
+      "loss": 59.4797,
+      "step": 27570
+    },
+    {
+      "epoch": 0.11142668988392716,
+      "grad_norm": 538.3480224609375,
+      "learning_rate": 9.547134119293835e-06,
+      "loss": 51.2988,
+      "step": 27580
+    },
+    {
+      "epoch": 0.11146709114929479,
+      "grad_norm": 699.3734741210938,
+      "learning_rate": 9.546547417536656e-06,
+      "loss": 43.1402,
+      "step": 27590
+    },
+    {
+      "epoch": 0.11150749241466243,
+      "grad_norm": 773.6231689453125,
+      "learning_rate": 9.545960354031564e-06,
+      "loss": 69.3904,
+      "step": 27600
+    },
+    {
+      "epoch": 0.11154789368003006,
+      "grad_norm": 1388.7728271484375,
+      "learning_rate": 9.545372928825271e-06,
+      "loss": 58.2387,
+      "step": 27610
+    },
+    {
+      "epoch": 0.1115882949453977,
+      "grad_norm": 633.869384765625,
+      "learning_rate": 9.544785141964514e-06,
+      "loss": 72.711,
+      "step": 27620
+    },
+    {
+      "epoch": 0.11162869621076532,
+      "grad_norm": 1325.7919921875,
+      "learning_rate": 9.544196993496062e-06,
+      "loss": 52.6202,
+      "step": 27630
+    },
+    {
+      "epoch": 0.11166909747613295,
+      "grad_norm": 556.158935546875,
+      "learning_rate": 9.54360848346671e-06,
+      "loss": 35.6353,
+      "step": 27640
+    },
+    {
+      "epoch": 0.11170949874150059,
+      "grad_norm": 634.7527465820312,
+      "learning_rate": 9.543019611923283e-06,
+      "loss": 57.7024,
+      "step": 27650
+    },
+    {
+      "epoch": 0.11174990000686821,
+      "grad_norm": 603.2935791015625,
+      "learning_rate": 9.542430378912634e-06,
+      "loss": 54.0695,
+      "step": 27660
+    },
+    {
+      "epoch": 0.11179030127223584,
+      "grad_norm": 494.1707458496094,
+      "learning_rate": 9.541840784481648e-06,
+      "loss": 36.8178,
+      "step": 27670
+    },
+    {
+      "epoch": 0.11183070253760348,
+      "grad_norm": 764.1178588867188,
+      "learning_rate": 9.541250828677235e-06,
+      "loss": 47.1555,
+      "step": 27680
+    },
+    {
+      "epoch": 0.1118711038029711,
+      "grad_norm": 681.1707153320312,
+      "learning_rate": 9.540660511546335e-06,
+      "loss": 42.6035,
+      "step": 27690
+    },
+    {
+      "epoch": 0.11191150506833875,
+      "grad_norm": 1050.41015625,
+      "learning_rate": 9.540069833135917e-06,
+      "loss": 57.2844,
+      "step": 27700
+    },
+    {
+      "epoch": 0.11195190633370637,
+      "grad_norm": 494.7498779296875,
+      "learning_rate": 9.539478793492978e-06,
+      "loss": 44.8998,
+      "step": 27710
+    },
+    {
+      "epoch": 0.111992307599074,
+      "grad_norm": 657.7085571289062,
+      "learning_rate": 9.538887392664544e-06,
+      "loss": 35.1478,
+      "step": 27720
+    },
+    {
+      "epoch": 0.11203270886444164,
+      "grad_norm": 864.594970703125,
+      "learning_rate": 9.53829563069767e-06,
+      "loss": 54.5924,
+      "step": 27730
+    },
+    {
+      "epoch": 0.11207311012980926,
+      "grad_norm": 439.203857421875,
+      "learning_rate": 9.537703507639444e-06,
+      "loss": 49.3219,
+      "step": 27740
+    },
+    {
+      "epoch": 0.11211351139517689,
+      "grad_norm": 404.7133483886719,
+      "learning_rate": 9.537111023536973e-06,
+      "loss": 55.6897,
+      "step": 27750
+    },
+    {
+      "epoch": 0.11215391266054453,
+      "grad_norm": 669.194091796875,
+      "learning_rate": 9.536518178437402e-06,
+      "loss": 39.5461,
+      "step": 27760
+    },
+    {
+      "epoch": 0.11219431392591216,
+      "grad_norm": 591.7159423828125,
+      "learning_rate": 9.535924972387898e-06,
+      "loss": 34.4681,
+      "step": 27770
+    },
+    {
+      "epoch": 0.1122347151912798,
+      "grad_norm": 478.49212646484375,
+      "learning_rate": 9.535331405435662e-06,
+      "loss": 40.2554,
+      "step": 27780
+    },
+    {
+      "epoch": 0.11227511645664742,
+      "grad_norm": 549.56787109375,
+      "learning_rate": 9.534737477627918e-06,
+      "loss": 28.9227,
+      "step": 27790
+    },
+    {
+      "epoch": 0.11231551772201505,
+      "grad_norm": 686.5111083984375,
+      "learning_rate": 9.534143189011928e-06,
+      "loss": 63.8506,
+      "step": 27800
+    },
+    {
+      "epoch": 0.11235591898738269,
+      "grad_norm": 730.6181030273438,
+      "learning_rate": 9.533548539634971e-06,
+      "loss": 65.5437,
+      "step": 27810
+    },
+    {
+      "epoch": 0.11239632025275031,
+      "grad_norm": 625.7398681640625,
+      "learning_rate": 9.532953529544365e-06,
+      "loss": 49.1752,
+      "step": 27820
+    },
+    {
+      "epoch": 0.11243672151811794,
+      "grad_norm": 332.5801696777344,
+      "learning_rate": 9.532358158787446e-06,
+      "loss": 50.4811,
+      "step": 27830
+    },
+    {
+      "epoch": 0.11247712278348558,
+      "grad_norm": 855.0786743164062,
+      "learning_rate": 9.531762427411592e-06,
+      "loss": 32.4726,
+      "step": 27840
+    },
+    {
+      "epoch": 0.1125175240488532,
+      "grad_norm": 464.85125732421875,
+      "learning_rate": 9.531166335464198e-06,
+      "loss": 57.4719,
+      "step": 27850
+    },
+    {
+      "epoch": 0.11255792531422085,
+      "grad_norm": 464.2534484863281,
+      "learning_rate": 9.530569882992698e-06,
+      "loss": 56.7898,
+      "step": 27860
+    },
+    {
+      "epoch": 0.11259832657958847,
+      "grad_norm": 338.19952392578125,
+      "learning_rate": 9.52997307004454e-06,
+      "loss": 45.8968,
+      "step": 27870
+    },
+    {
+      "epoch": 0.1126387278449561,
+      "grad_norm": 974.372314453125,
+      "learning_rate": 9.529375896667218e-06,
+      "loss": 58.9688,
+      "step": 27880
+    },
+    {
+      "epoch": 0.11267912911032374,
+      "grad_norm": 679.4461059570312,
+      "learning_rate": 9.528778362908241e-06,
+      "loss": 39.7818,
+      "step": 27890
+    },
+    {
+      "epoch": 0.11271953037569137,
+      "grad_norm": 752.9938354492188,
+      "learning_rate": 9.528180468815155e-06,
+      "loss": 52.0521,
+      "step": 27900
+    },
+    {
+      "epoch": 0.11275993164105899,
+      "grad_norm": 577.4219970703125,
+      "learning_rate": 9.527582214435531e-06,
+      "loss": 35.0712,
+      "step": 27910
+    },
+    {
+      "epoch": 0.11280033290642663,
+      "grad_norm": 1068.46337890625,
+      "learning_rate": 9.526983599816968e-06,
+      "loss": 53.9641,
+      "step": 27920
+    },
+    {
+      "epoch": 0.11284073417179426,
+      "grad_norm": 313.15789794921875,
+      "learning_rate": 9.526384625007096e-06,
+      "loss": 46.0763,
+      "step": 27930
+    },
+    {
+      "epoch": 0.1128811354371619,
+      "grad_norm": 842.06591796875,
+      "learning_rate": 9.525785290053573e-06,
+      "loss": 47.3168,
+      "step": 27940
+    },
+    {
+      "epoch": 0.11292153670252952,
+      "grad_norm": 699.9207153320312,
+      "learning_rate": 9.525185595004085e-06,
+      "loss": 90.293,
+      "step": 27950
+    },
+    {
+      "epoch": 0.11296193796789715,
+      "grad_norm": 576.01220703125,
+      "learning_rate": 9.524585539906345e-06,
+      "loss": 45.2152,
+      "step": 27960
+    },
+    {
+      "epoch": 0.11300233923326479,
+      "grad_norm": 606.6934204101562,
+      "learning_rate": 9.523985124808102e-06,
+      "loss": 55.5013,
+      "step": 27970
+    },
+    {
+      "epoch": 0.11304274049863242,
+      "grad_norm": 891.949462890625,
+      "learning_rate": 9.523384349757123e-06,
+      "loss": 43.9636,
+      "step": 27980
+    },
+    {
+      "epoch": 0.11308314176400004,
+      "grad_norm": 659.2472534179688,
+      "learning_rate": 9.522783214801213e-06,
+      "loss": 71.8418,
+      "step": 27990
+    },
+    {
+      "epoch": 0.11312354302936768,
+      "grad_norm": 873.6649169921875,
+      "learning_rate": 9.522181719988196e-06,
+      "loss": 47.9444,
+      "step": 28000
+    },
+    {
+      "epoch": 0.11316394429473531,
+      "grad_norm": 907.7310180664062,
+      "learning_rate": 9.521579865365935e-06,
+      "loss": 52.8025,
+      "step": 28010
+    },
+    {
+      "epoch": 0.11320434556010295,
+      "grad_norm": 656.29248046875,
+      "learning_rate": 9.520977650982316e-06,
+      "loss": 43.1285,
+      "step": 28020
+    },
+    {
+      "epoch": 0.11324474682547057,
+      "grad_norm": 368.1136474609375,
+      "learning_rate": 9.520375076885253e-06,
+      "loss": 57.7966,
+      "step": 28030
+    },
+    {
+      "epoch": 0.1132851480908382,
+      "grad_norm": 934.730224609375,
+      "learning_rate": 9.519772143122691e-06,
+      "loss": 64.5761,
+      "step": 28040
+    },
+    {
+      "epoch": 0.11332554935620584,
+      "grad_norm": 1095.819580078125,
+      "learning_rate": 9.519168849742603e-06,
+      "loss": 67.1447,
+      "step": 28050
+    },
+    {
+      "epoch": 0.11336595062157347,
+      "grad_norm": 500.99053955078125,
+      "learning_rate": 9.51856519679299e-06,
+      "loss": 38.3834,
+      "step": 28060
+    },
+    {
+      "epoch": 0.11340635188694109,
+      "grad_norm": 684.0264892578125,
+      "learning_rate": 9.517961184321882e-06,
+      "loss": 59.6902,
+      "step": 28070
+    },
+    {
+      "epoch": 0.11344675315230873,
+      "grad_norm": 419.83099365234375,
+      "learning_rate": 9.517356812377336e-06,
+      "loss": 43.5778,
+      "step": 28080
+    },
+    {
+      "epoch": 0.11348715441767636,
+      "grad_norm": 859.154052734375,
+      "learning_rate": 9.516752081007441e-06,
+      "loss": 60.8287,
+      "step": 28090
+    },
+    {
+      "epoch": 0.113527555683044,
+      "grad_norm": 688.739501953125,
+      "learning_rate": 9.51614699026031e-06,
+      "loss": 51.3651,
+      "step": 28100
+    },
+    {
+      "epoch": 0.11356795694841162,
+      "grad_norm": 646.2841796875,
+      "learning_rate": 9.515541540184093e-06,
+      "loss": 49.2372,
+      "step": 28110
+    },
+    {
+      "epoch": 0.11360835821377925,
+      "grad_norm": 857.5767211914062,
+      "learning_rate": 9.514935730826957e-06,
+      "loss": 46.2163,
+      "step": 28120
+    },
+    {
+      "epoch": 0.11364875947914689,
+      "grad_norm": 553.1755981445312,
+      "learning_rate": 9.514329562237107e-06,
+      "loss": 42.8402,
+      "step": 28130
+    },
+    {
+      "epoch": 0.11368916074451452,
+      "grad_norm": 509.92474365234375,
+      "learning_rate": 9.51372303446277e-06,
+      "loss": 44.156,
+      "step": 28140
+    },
+    {
+      "epoch": 0.11372956200988214,
+      "grad_norm": 554.966552734375,
+      "learning_rate": 9.513116147552207e-06,
+      "loss": 42.856,
+      "step": 28150
+    },
+    {
+      "epoch": 0.11376996327524978,
+      "grad_norm": 784.25830078125,
+      "learning_rate": 9.512508901553703e-06,
+      "loss": 67.9408,
+      "step": 28160
+    },
+    {
+      "epoch": 0.11381036454061741,
+      "grad_norm": 406.4463806152344,
+      "learning_rate": 9.511901296515578e-06,
+      "loss": 57.0736,
+      "step": 28170
+    },
+    {
+      "epoch": 0.11385076580598505,
+      "grad_norm": 957.32080078125,
+      "learning_rate": 9.511293332486172e-06,
+      "loss": 55.6416,
+      "step": 28180
+    },
+    {
+      "epoch": 0.11389116707135268,
+      "grad_norm": 840.68310546875,
+      "learning_rate": 9.51068500951386e-06,
+      "loss": 56.1937,
+      "step": 28190
+    },
+    {
+      "epoch": 0.1139315683367203,
+      "grad_norm": 607.9652709960938,
+      "learning_rate": 9.510076327647043e-06,
+      "loss": 50.4717,
+      "step": 28200
+    },
+    {
+      "epoch": 0.11397196960208794,
+      "grad_norm": 921.6870727539062,
+      "learning_rate": 9.509467286934151e-06,
+      "loss": 35.7569,
+      "step": 28210
+    },
+    {
+      "epoch": 0.11401237086745557,
+      "grad_norm": 740.13232421875,
+      "learning_rate": 9.508857887423644e-06,
+      "loss": 46.1539,
+      "step": 28220
+    },
+    {
+      "epoch": 0.1140527721328232,
+      "grad_norm": 872.6014404296875,
+      "learning_rate": 9.508248129164006e-06,
+      "loss": 51.5815,
+      "step": 28230
+    },
+    {
+      "epoch": 0.11409317339819083,
+      "grad_norm": 684.9937744140625,
+      "learning_rate": 9.507638012203755e-06,
+      "loss": 55.3623,
+      "step": 28240
+    },
+    {
+      "epoch": 0.11413357466355846,
+      "grad_norm": 487.1953430175781,
+      "learning_rate": 9.507027536591436e-06,
+      "loss": 37.2326,
+      "step": 28250
+    },
+    {
+      "epoch": 0.1141739759289261,
+      "grad_norm": 896.9238891601562,
+      "learning_rate": 9.506416702375618e-06,
+      "loss": 45.9022,
+      "step": 28260
+    },
+    {
+      "epoch": 0.11421437719429373,
+      "grad_norm": 454.0743408203125,
+      "learning_rate": 9.505805509604906e-06,
+      "loss": 57.2897,
+      "step": 28270
+    },
+    {
+      "epoch": 0.11425477845966135,
+      "grad_norm": 1185.238525390625,
+      "learning_rate": 9.505193958327927e-06,
+      "loss": 74.1116,
+      "step": 28280
+    },
+    {
+      "epoch": 0.11429517972502899,
+      "grad_norm": 518.6088256835938,
+      "learning_rate": 9.504582048593343e-06,
+      "loss": 52.5233,
+      "step": 28290
+    },
+    {
+      "epoch": 0.11433558099039662,
+      "grad_norm": 625.5014038085938,
+      "learning_rate": 9.503969780449838e-06,
+      "loss": 32.3718,
+      "step": 28300
+    },
+    {
+      "epoch": 0.11437598225576424,
+      "grad_norm": 1069.5562744140625,
+      "learning_rate": 9.503357153946126e-06,
+      "loss": 50.4978,
+      "step": 28310
+    },
+    {
+      "epoch": 0.11441638352113188,
+      "grad_norm": 333.4046325683594,
+      "learning_rate": 9.502744169130955e-06,
+      "loss": 50.467,
+      "step": 28320
+    },
+    {
+      "epoch": 0.11445678478649951,
+      "grad_norm": 344.731201171875,
+      "learning_rate": 9.502130826053095e-06,
+      "loss": 49.283,
+      "step": 28330
+    },
+    {
+      "epoch": 0.11449718605186715,
+      "grad_norm": 707.1812744140625,
+      "learning_rate": 9.501517124761347e-06,
+      "loss": 35.1441,
+      "step": 28340
+    },
+    {
+      "epoch": 0.11453758731723478,
+      "grad_norm": 1012.0341796875,
+      "learning_rate": 9.50090306530454e-06,
+      "loss": 61.0239,
+      "step": 28350
+    },
+    {
+      "epoch": 0.1145779885826024,
+      "grad_norm": 489.1926574707031,
+      "learning_rate": 9.500288647731533e-06,
+      "loss": 40.3379,
+      "step": 28360
+    },
+    {
+      "epoch": 0.11461838984797004,
+      "grad_norm": 698.6495361328125,
+      "learning_rate": 9.49967387209121e-06,
+      "loss": 53.1061,
+      "step": 28370
+    },
+    {
+      "epoch": 0.11465879111333767,
+      "grad_norm": 1728.1700439453125,
+      "learning_rate": 9.499058738432492e-06,
+      "loss": 51.7011,
+      "step": 28380
+    },
+    {
+      "epoch": 0.1146991923787053,
+      "grad_norm": 505.17584228515625,
+      "learning_rate": 9.498443246804314e-06,
+      "loss": 56.5864,
+      "step": 28390
+    },
+    {
+      "epoch": 0.11473959364407293,
+      "grad_norm": 463.1519470214844,
+      "learning_rate": 9.497827397255655e-06,
+      "loss": 53.7967,
+      "step": 28400
+    },
+    {
+      "epoch": 0.11477999490944056,
+      "grad_norm": 1365.9278564453125,
+      "learning_rate": 9.49721118983551e-06,
+      "loss": 63.5547,
+      "step": 28410
+    },
+    {
+      "epoch": 0.1148203961748082,
+      "grad_norm": 1068.4659423828125,
+      "learning_rate": 9.49659462459291e-06,
+      "loss": 46.5801,
+      "step": 28420
+    },
+    {
+      "epoch": 0.11486079744017583,
+      "grad_norm": 507.3717956542969,
+      "learning_rate": 9.495977701576913e-06,
+      "loss": 38.8723,
+      "step": 28430
+    },
+    {
+      "epoch": 0.11490119870554345,
+      "grad_norm": 637.6323852539062,
+      "learning_rate": 9.495360420836603e-06,
+      "loss": 63.2256,
+      "step": 28440
+    },
+    {
+      "epoch": 0.11494159997091109,
+      "grad_norm": 1089.265869140625,
+      "learning_rate": 9.494742782421099e-06,
+      "loss": 51.0123,
+      "step": 28450
+    },
+    {
+      "epoch": 0.11498200123627872,
+      "grad_norm": 891.3598022460938,
+      "learning_rate": 9.494124786379535e-06,
+      "loss": 43.423,
+      "step": 28460
+    },
+    {
+      "epoch": 0.11502240250164635,
+      "grad_norm": 604.9508056640625,
+      "learning_rate": 9.49350643276109e-06,
+      "loss": 46.5848,
+      "step": 28470
+    },
+    {
+      "epoch": 0.11506280376701399,
+      "grad_norm": 635.8272094726562,
+      "learning_rate": 9.49288772161496e-06,
+      "loss": 37.0862,
+      "step": 28480
+    },
+    {
+      "epoch": 0.11510320503238161,
+      "grad_norm": 371.99444580078125,
+      "learning_rate": 9.492268652990374e-06,
+      "loss": 61.0252,
+      "step": 28490
+    },
+    {
+      "epoch": 0.11514360629774925,
+      "grad_norm": 503.8235778808594,
+      "learning_rate": 9.491649226936586e-06,
+      "loss": 56.4696,
+      "step": 28500
+    },
+    {
+      "epoch": 0.11518400756311688,
+      "grad_norm": 692.4237060546875,
+      "learning_rate": 9.491029443502884e-06,
+      "loss": 56.2257,
+      "step": 28510
+    },
+    {
+      "epoch": 0.1152244088284845,
+      "grad_norm": 224.27877807617188,
+      "learning_rate": 9.490409302738582e-06,
+      "loss": 39.3771,
+      "step": 28520
+    },
+    {
+      "epoch": 0.11526481009385214,
+      "grad_norm": 678.6007080078125,
+      "learning_rate": 9.489788804693017e-06,
+      "loss": 42.1856,
+      "step": 28530
+    },
+    {
+      "epoch": 0.11530521135921977,
+      "grad_norm": 1114.4183349609375,
+      "learning_rate": 9.489167949415563e-06,
+      "loss": 58.9925,
+      "step": 28540
+    },
+    {
+      "epoch": 0.1153456126245874,
+      "grad_norm": 557.2171630859375,
+      "learning_rate": 9.48854673695562e-06,
+      "loss": 55.2478,
+      "step": 28550
+    },
+    {
+      "epoch": 0.11538601388995504,
+      "grad_norm": 998.3175048828125,
+      "learning_rate": 9.48792516736261e-06,
+      "loss": 35.496,
+      "step": 28560
+    },
+    {
+      "epoch": 0.11542641515532266,
+      "grad_norm": 794.63232421875,
+      "learning_rate": 9.487303240685992e-06,
+      "loss": 55.0467,
+      "step": 28570
+    },
+    {
+      "epoch": 0.1154668164206903,
+      "grad_norm": 455.69879150390625,
+      "learning_rate": 9.48668095697525e-06,
+      "loss": 41.8241,
+      "step": 28580
+    },
+    {
+      "epoch": 0.11550721768605793,
+      "grad_norm": 712.5789794921875,
+      "learning_rate": 9.486058316279894e-06,
+      "loss": 59.8318,
+      "step": 28590
+    },
+    {
+      "epoch": 0.11554761895142555,
+      "grad_norm": 707.5386962890625,
+      "learning_rate": 9.485435318649468e-06,
+      "loss": 50.9216,
+      "step": 28600
+    },
+    {
+      "epoch": 0.1155880202167932,
+      "grad_norm": 722.671875,
+      "learning_rate": 9.484811964133537e-06,
+      "loss": 65.316,
+      "step": 28610
+    },
+    {
+      "epoch": 0.11562842148216082,
+      "grad_norm": 509.6846923828125,
+      "learning_rate": 9.484188252781701e-06,
+      "loss": 50.3926,
+      "step": 28620
+    },
+    {
+      "epoch": 0.11566882274752845,
+      "grad_norm": 1625.15869140625,
+      "learning_rate": 9.483564184643586e-06,
+      "loss": 37.5687,
+      "step": 28630
+    },
+    {
+      "epoch": 0.11570922401289609,
+      "grad_norm": 425.3319396972656,
+      "learning_rate": 9.482939759768845e-06,
+      "loss": 37.632,
+      "step": 28640
+    },
+    {
+      "epoch": 0.11574962527826371,
+      "grad_norm": 526.8452758789062,
+      "learning_rate": 9.48231497820716e-06,
+      "loss": 55.4761,
+      "step": 28650
+    },
+    {
+      "epoch": 0.11579002654363135,
+      "grad_norm": 524.2207641601562,
+      "learning_rate": 9.481689840008246e-06,
+      "loss": 54.3112,
+      "step": 28660
+    },
+    {
+      "epoch": 0.11583042780899898,
+      "grad_norm": 793.691162109375,
+      "learning_rate": 9.481064345221838e-06,
+      "loss": 60.8318,
+      "step": 28670
+    },
+    {
+      "epoch": 0.1158708290743666,
+      "grad_norm": 453.846435546875,
+      "learning_rate": 9.480438493897707e-06,
+      "loss": 56.1913,
+      "step": 28680
+    },
+    {
+      "epoch": 0.11591123033973424,
+      "grad_norm": 836.8286743164062,
+      "learning_rate": 9.479812286085645e-06,
+      "loss": 49.0991,
+      "step": 28690
+    },
+    {
+      "epoch": 0.11595163160510187,
+      "grad_norm": 656.8587036132812,
+      "learning_rate": 9.47918572183548e-06,
+      "loss": 66.3525,
+      "step": 28700
+    },
+    {
+      "epoch": 0.1159920328704695,
+      "grad_norm": 918.6471557617188,
+      "learning_rate": 9.478558801197065e-06,
+      "loss": 54.7832,
+      "step": 28710
+    },
+    {
+      "epoch": 0.11603243413583714,
+      "grad_norm": 702.2949829101562,
+      "learning_rate": 9.47793152422028e-06,
+      "loss": 43.806,
+      "step": 28720
+    },
+    {
+      "epoch": 0.11607283540120476,
+      "grad_norm": 683.9447021484375,
+      "learning_rate": 9.477303890955032e-06,
+      "loss": 41.7837,
+      "step": 28730
+    },
+    {
+      "epoch": 0.1161132366665724,
+      "grad_norm": 881.1528930664062,
+      "learning_rate": 9.476675901451264e-06,
+      "loss": 62.3749,
+      "step": 28740
+    },
+    {
+      "epoch": 0.11615363793194003,
+      "grad_norm": 235.79928588867188,
+      "learning_rate": 9.476047555758938e-06,
+      "loss": 34.0945,
+      "step": 28750
+    },
+    {
+      "epoch": 0.11619403919730766,
+      "grad_norm": 1202.2252197265625,
+      "learning_rate": 9.475418853928051e-06,
+      "loss": 40.956,
+      "step": 28760
+    },
+    {
+      "epoch": 0.1162344404626753,
+      "grad_norm": 723.0114135742188,
+      "learning_rate": 9.474789796008625e-06,
+      "loss": 59.2273,
+      "step": 28770
+    },
+    {
+      "epoch": 0.11627484172804292,
+      "grad_norm": 948.4896850585938,
+      "learning_rate": 9.474160382050711e-06,
+      "loss": 43.0877,
+      "step": 28780
+    },
+    {
+      "epoch": 0.11631524299341055,
+      "grad_norm": 665.72998046875,
+      "learning_rate": 9.47353061210439e-06,
+      "loss": 58.8115,
+      "step": 28790
+    },
+    {
+      "epoch": 0.11635564425877819,
+      "grad_norm": 928.5210571289062,
+      "learning_rate": 9.47290048621977e-06,
+      "loss": 72.7469,
+      "step": 28800
+    },
+    {
+      "epoch": 0.11639604552414581,
+      "grad_norm": 837.2612915039062,
+      "learning_rate": 9.472270004446984e-06,
+      "loss": 62.3605,
+      "step": 28810
+    },
+    {
+      "epoch": 0.11643644678951345,
+      "grad_norm": 404.30841064453125,
+      "learning_rate": 9.4716391668362e-06,
+      "loss": 43.8358,
+      "step": 28820
+    },
+    {
+      "epoch": 0.11647684805488108,
+      "grad_norm": 522.2630615234375,
+      "learning_rate": 9.471007973437607e-06,
+      "loss": 35.7472,
+      "step": 28830
+    },
+    {
+      "epoch": 0.1165172493202487,
+      "grad_norm": 826.51708984375,
+      "learning_rate": 9.470376424301432e-06,
+      "loss": 68.2935,
+      "step": 28840
+    },
+    {
+      "epoch": 0.11655765058561635,
+      "grad_norm": 969.8021240234375,
+      "learning_rate": 9.46974451947792e-06,
+      "loss": 44.8838,
+      "step": 28850
+    },
+    {
+      "epoch": 0.11659805185098397,
+      "grad_norm": 907.9220581054688,
+      "learning_rate": 9.469112259017349e-06,
+      "loss": 37.908,
+      "step": 28860
+    },
+    {
+      "epoch": 0.1166384531163516,
+      "grad_norm": 1271.998779296875,
+      "learning_rate": 9.468479642970027e-06,
+      "loss": 65.6431,
+      "step": 28870
+    },
+    {
+      "epoch": 0.11667885438171924,
+      "grad_norm": 616.036865234375,
+      "learning_rate": 9.467846671386287e-06,
+      "loss": 39.8803,
+      "step": 28880
+    },
+    {
+      "epoch": 0.11671925564708686,
+      "grad_norm": 800.3985595703125,
+      "learning_rate": 9.467213344316493e-06,
+      "loss": 60.458,
+      "step": 28890
+    },
+    {
+      "epoch": 0.1167596569124545,
+      "grad_norm": 982.2677612304688,
+      "learning_rate": 9.466579661811032e-06,
+      "loss": 75.691,
+      "step": 28900
+    },
+    {
+      "epoch": 0.11680005817782213,
+      "grad_norm": 649.3153076171875,
+      "learning_rate": 9.46594562392033e-06,
+      "loss": 44.2065,
+      "step": 28910
+    },
+    {
+      "epoch": 0.11684045944318976,
+      "grad_norm": 982.4515380859375,
+      "learning_rate": 9.465311230694828e-06,
+      "loss": 45.287,
+      "step": 28920
+    },
+    {
+      "epoch": 0.1168808607085574,
+      "grad_norm": 727.6909790039062,
+      "learning_rate": 9.464676482185005e-06,
+      "loss": 30.1195,
+      "step": 28930
+    },
+    {
+      "epoch": 0.11692126197392502,
+      "grad_norm": 301.429931640625,
+      "learning_rate": 9.464041378441365e-06,
+      "loss": 58.9985,
+      "step": 28940
+    },
+    {
+      "epoch": 0.11696166323929265,
+      "grad_norm": 854.9701538085938,
+      "learning_rate": 9.46340591951444e-06,
+      "loss": 72.6457,
+      "step": 28950
+    },
+    {
+      "epoch": 0.11700206450466029,
+      "grad_norm": 518.9381713867188,
+      "learning_rate": 9.462770105454789e-06,
+      "loss": 51.8344,
+      "step": 28960
+    },
+    {
+      "epoch": 0.11704246577002791,
+      "grad_norm": 741.2862548828125,
+      "learning_rate": 9.462133936313002e-06,
+      "loss": 40.7278,
+      "step": 28970
+    },
+    {
+      "epoch": 0.11708286703539555,
+      "grad_norm": 752.0146484375,
+      "learning_rate": 9.461497412139697e-06,
+      "loss": 63.2356,
+      "step": 28980
+    },
+    {
+      "epoch": 0.11712326830076318,
+      "grad_norm": 656.5631103515625,
+      "learning_rate": 9.46086053298552e-06,
+      "loss": 55.1799,
+      "step": 28990
+    },
+    {
+      "epoch": 0.11716366956613081,
+      "grad_norm": 655.6858520507812,
+      "learning_rate": 9.460223298901138e-06,
+      "loss": 46.3702,
+      "step": 29000
+    },
+    {
+      "epoch": 0.11720407083149845,
+      "grad_norm": 367.1216735839844,
+      "learning_rate": 9.459585709937262e-06,
+      "loss": 82.6018,
+      "step": 29010
+    },
+    {
+      "epoch": 0.11724447209686607,
+      "grad_norm": 822.807373046875,
+      "learning_rate": 9.458947766144617e-06,
+      "loss": 87.3932,
+      "step": 29020
+    },
+    {
+      "epoch": 0.1172848733622337,
+      "grad_norm": 728.6187133789062,
+      "learning_rate": 9.458309467573963e-06,
+      "loss": 40.4143,
+      "step": 29030
+    },
+    {
+      "epoch": 0.11732527462760134,
+      "grad_norm": 690.3950805664062,
+      "learning_rate": 9.457670814276083e-06,
+      "loss": 53.4531,
+      "step": 29040
+    },
+    {
+      "epoch": 0.11736567589296897,
+      "grad_norm": 833.6300048828125,
+      "learning_rate": 9.457031806301795e-06,
+      "loss": 49.5981,
+      "step": 29050
+    },
+    {
+      "epoch": 0.1174060771583366,
+      "grad_norm": 394.2637939453125,
+      "learning_rate": 9.456392443701943e-06,
+      "loss": 28.4206,
+      "step": 29060
+    },
+    {
+      "epoch": 0.11744647842370423,
+      "grad_norm": 447.1683349609375,
+      "learning_rate": 9.455752726527395e-06,
+      "loss": 42.3745,
+      "step": 29070
+    },
+    {
+      "epoch": 0.11748687968907186,
+      "grad_norm": 747.0037841796875,
+      "learning_rate": 9.45511265482905e-06,
+      "loss": 67.5624,
+      "step": 29080
+    },
+    {
+      "epoch": 0.1175272809544395,
+      "grad_norm": 619.459228515625,
+      "learning_rate": 9.454472228657841e-06,
+      "loss": 49.5098,
+      "step": 29090
+    },
+    {
+      "epoch": 0.11756768221980712,
+      "grad_norm": 461.28607177734375,
+      "learning_rate": 9.453831448064717e-06,
+      "loss": 34.5057,
+      "step": 29100
+    },
+    {
+      "epoch": 0.11760808348517475,
+      "grad_norm": 478.06207275390625,
+      "learning_rate": 9.453190313100666e-06,
+      "loss": 60.1882,
+      "step": 29110
+    },
+    {
+      "epoch": 0.11764848475054239,
+      "grad_norm": 740.0570068359375,
+      "learning_rate": 9.4525488238167e-06,
+      "loss": 45.3488,
+      "step": 29120
+    },
+    {
+      "epoch": 0.11768888601591002,
+      "grad_norm": 622.1351928710938,
+      "learning_rate": 9.451906980263857e-06,
+      "loss": 40.4671,
+      "step": 29130
+    },
+    {
+      "epoch": 0.11772928728127766,
+      "grad_norm": 646.7392578125,
+      "learning_rate": 9.451264782493208e-06,
+      "loss": 51.3332,
+      "step": 29140
+    },
+    {
+      "epoch": 0.11776968854664528,
+      "grad_norm": 364.55999755859375,
+      "learning_rate": 9.450622230555849e-06,
+      "loss": 50.0434,
+      "step": 29150
+    },
+    {
+      "epoch": 0.11781008981201291,
+      "grad_norm": 586.4746704101562,
+      "learning_rate": 9.449979324502905e-06,
+      "loss": 33.2186,
+      "step": 29160
+    },
+    {
+      "epoch": 0.11785049107738055,
+      "grad_norm": 899.0317993164062,
+      "learning_rate": 9.449336064385529e-06,
+      "loss": 62.2745,
+      "step": 29170
+    },
+    {
+      "epoch": 0.11789089234274817,
+      "grad_norm": 466.4114990234375,
+      "learning_rate": 9.4486924502549e-06,
+      "loss": 57.8692,
+      "step": 29180
+    },
+    {
+      "epoch": 0.1179312936081158,
+      "grad_norm": 708.7315063476562,
+      "learning_rate": 9.448048482162231e-06,
+      "loss": 109.0333,
+      "step": 29190
+    },
+    {
+      "epoch": 0.11797169487348344,
+      "grad_norm": 617.0620727539062,
+      "learning_rate": 9.447404160158758e-06,
+      "loss": 37.3076,
+      "step": 29200
+    },
+    {
+      "epoch": 0.11801209613885107,
+      "grad_norm": 1074.2401123046875,
+      "learning_rate": 9.446759484295745e-06,
+      "loss": 63.679,
+      "step": 29210
+    },
+    {
+      "epoch": 0.1180524974042187,
+      "grad_norm": 859.2569580078125,
+      "learning_rate": 9.44611445462449e-06,
+      "loss": 47.2913,
+      "step": 29220
+    },
+    {
+      "epoch": 0.11809289866958633,
+      "grad_norm": 357.2909851074219,
+      "learning_rate": 9.445469071196312e-06,
+      "loss": 63.1842,
+      "step": 29230
+    },
+    {
+      "epoch": 0.11813329993495396,
+      "grad_norm": 916.20068359375,
+      "learning_rate": 9.444823334062562e-06,
+      "loss": 60.4991,
+      "step": 29240
+    },
+    {
+      "epoch": 0.1181737012003216,
+      "grad_norm": 238.49969482421875,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 62.8598,
+      "step": 29250
+    },
+    {
+      "epoch": 0.11821410246568922,
+      "grad_norm": 761.7315063476562,
+      "learning_rate": 9.443530798883887e-06,
+      "loss": 39.2142,
+      "step": 29260
+    },
+    {
+      "epoch": 0.11825450373105685,
+      "grad_norm": 2124.98779296875,
+      "learning_rate": 9.442884000941803e-06,
+      "loss": 51.3082,
+      "step": 29270
+    },
+    {
+      "epoch": 0.11829490499642449,
+      "grad_norm": 922.1412353515625,
+      "learning_rate": 9.44223684949983e-06,
+      "loss": 51.5363,
+      "step": 29280
+    },
+    {
+      "epoch": 0.11833530626179212,
+      "grad_norm": 718.6259155273438,
+      "learning_rate": 9.441589344609457e-06,
+      "loss": 65.476,
+      "step": 29290
+    },
+    {
+      "epoch": 0.11837570752715976,
+      "grad_norm": 972.5775146484375,
+      "learning_rate": 9.440941486322205e-06,
+      "loss": 65.9795,
+      "step": 29300
+    },
+    {
+      "epoch": 0.11841610879252738,
+      "grad_norm": 566.7423706054688,
+      "learning_rate": 9.44029327468962e-06,
+      "loss": 45.4314,
+      "step": 29310
+    },
+    {
+      "epoch": 0.11845651005789501,
+      "grad_norm": 269.9037780761719,
+      "learning_rate": 9.439644709763276e-06,
+      "loss": 57.1088,
+      "step": 29320
+    },
+    {
+      "epoch": 0.11849691132326265,
+      "grad_norm": 1288.29150390625,
+      "learning_rate": 9.43899579159478e-06,
+      "loss": 40.9538,
+      "step": 29330
+    },
+    {
+      "epoch": 0.11853731258863028,
+      "grad_norm": 2228.189453125,
+      "learning_rate": 9.438346520235759e-06,
+      "loss": 73.1681,
+      "step": 29340
+    },
+    {
+      "epoch": 0.1185777138539979,
+      "grad_norm": 754.686767578125,
+      "learning_rate": 9.437696895737876e-06,
+      "loss": 80.8364,
+      "step": 29350
+    },
+    {
+      "epoch": 0.11861811511936554,
+      "grad_norm": 423.5769958496094,
+      "learning_rate": 9.437046918152817e-06,
+      "loss": 43.7477,
+      "step": 29360
+    },
+    {
+      "epoch": 0.11865851638473317,
+      "grad_norm": 539.15234375,
+      "learning_rate": 9.436396587532297e-06,
+      "loss": 44.1189,
+      "step": 29370
+    },
+    {
+      "epoch": 0.11869891765010081,
+      "grad_norm": 809.685791015625,
+      "learning_rate": 9.435745903928062e-06,
+      "loss": 50.4817,
+      "step": 29380
+    },
+    {
+      "epoch": 0.11873931891546843,
+      "grad_norm": 692.5665283203125,
+      "learning_rate": 9.435094867391881e-06,
+      "loss": 58.9802,
+      "step": 29390
+    },
+    {
+      "epoch": 0.11877972018083606,
+      "grad_norm": 363.3084411621094,
+      "learning_rate": 9.434443477975557e-06,
+      "loss": 38.1004,
+      "step": 29400
+    },
+    {
+      "epoch": 0.1188201214462037,
+      "grad_norm": 835.03515625,
+      "learning_rate": 9.433791735730917e-06,
+      "loss": 44.2224,
+      "step": 29410
+    },
+    {
+      "epoch": 0.11886052271157133,
+      "grad_norm": 734.1571044921875,
+      "learning_rate": 9.433139640709817e-06,
+      "loss": 53.3648,
+      "step": 29420
+    },
+    {
+      "epoch": 0.11890092397693895,
+      "grad_norm": 669.4501953125,
+      "learning_rate": 9.432487192964142e-06,
+      "loss": 58.705,
+      "step": 29430
+    },
+    {
+      "epoch": 0.11894132524230659,
+      "grad_norm": 521.8321533203125,
+      "learning_rate": 9.431834392545803e-06,
+      "loss": 58.5444,
+      "step": 29440
+    },
+    {
+      "epoch": 0.11898172650767422,
+      "grad_norm": 370.6477355957031,
+      "learning_rate": 9.43118123950674e-06,
+      "loss": 46.7026,
+      "step": 29450
+    },
+    {
+      "epoch": 0.11902212777304186,
+      "grad_norm": 558.437255859375,
+      "learning_rate": 9.430527733898922e-06,
+      "loss": 47.0624,
+      "step": 29460
+    },
+    {
+      "epoch": 0.11906252903840948,
+      "grad_norm": 691.9668579101562,
+      "learning_rate": 9.429873875774344e-06,
+      "loss": 46.8082,
+      "step": 29470
+    },
+    {
+      "epoch": 0.11910293030377711,
+      "grad_norm": 739.7124633789062,
+      "learning_rate": 9.429219665185034e-06,
+      "loss": 57.6623,
+      "step": 29480
+    },
+    {
+      "epoch": 0.11914333156914475,
+      "grad_norm": 1001.8974609375,
+      "learning_rate": 9.428565102183043e-06,
+      "loss": 46.2761,
+      "step": 29490
+    },
+    {
+      "epoch": 0.11918373283451238,
+      "grad_norm": 427.9935607910156,
+      "learning_rate": 9.42791018682045e-06,
+      "loss": 78.2117,
+      "step": 29500
+    },
+    {
+      "epoch": 0.11922413409988,
+      "grad_norm": 1094.6697998046875,
+      "learning_rate": 9.427254919149367e-06,
+      "loss": 68.6234,
+      "step": 29510
+    },
+    {
+      "epoch": 0.11926453536524764,
+      "grad_norm": 436.7558898925781,
+      "learning_rate": 9.426599299221925e-06,
+      "loss": 45.9419,
+      "step": 29520
+    },
+    {
+      "epoch": 0.11930493663061527,
+      "grad_norm": 570.2569580078125,
+      "learning_rate": 9.425943327090295e-06,
+      "loss": 61.1487,
+      "step": 29530
+    },
+    {
+      "epoch": 0.11934533789598291,
+      "grad_norm": 1047.4046630859375,
+      "learning_rate": 9.425287002806666e-06,
+      "loss": 60.9063,
+      "step": 29540
+    },
+    {
+      "epoch": 0.11938573916135053,
+      "grad_norm": 661.2189331054688,
+      "learning_rate": 9.42463032642326e-06,
+      "loss": 44.0193,
+      "step": 29550
+    },
+    {
+      "epoch": 0.11942614042671816,
+      "grad_norm": 543.251708984375,
+      "learning_rate": 9.423973297992324e-06,
+      "loss": 41.5051,
+      "step": 29560
+    },
+    {
+      "epoch": 0.1194665416920858,
+      "grad_norm": 870.4523315429688,
+      "learning_rate": 9.423315917566137e-06,
+      "loss": 47.8213,
+      "step": 29570
+    },
+    {
+      "epoch": 0.11950694295745343,
+      "grad_norm": 733.741943359375,
+      "learning_rate": 9.422658185197002e-06,
+      "loss": 44.6389,
+      "step": 29580
+    },
+    {
+      "epoch": 0.11954734422282105,
+      "grad_norm": 424.02923583984375,
+      "learning_rate": 9.422000100937253e-06,
+      "loss": 53.1652,
+      "step": 29590
+    },
+    {
+      "epoch": 0.1195877454881887,
+      "grad_norm": 812.8602905273438,
+      "learning_rate": 9.42134166483925e-06,
+      "loss": 36.4047,
+      "step": 29600
+    },
+    {
+      "epoch": 0.11962814675355632,
+      "grad_norm": 1044.6739501953125,
+      "learning_rate": 9.420682876955382e-06,
+      "loss": 63.9026,
+      "step": 29610
+    },
+    {
+      "epoch": 0.11966854801892396,
+      "grad_norm": 752.5007934570312,
+      "learning_rate": 9.420023737338065e-06,
+      "loss": 54.0017,
+      "step": 29620
+    },
+    {
+      "epoch": 0.11970894928429159,
+      "grad_norm": 322.84503173828125,
+      "learning_rate": 9.419364246039745e-06,
+      "loss": 60.8512,
+      "step": 29630
+    },
+    {
+      "epoch": 0.11974935054965921,
+      "grad_norm": 407.95428466796875,
+      "learning_rate": 9.418704403112894e-06,
+      "loss": 31.9523,
+      "step": 29640
+    },
+    {
+      "epoch": 0.11978975181502685,
+      "grad_norm": 855.3703002929688,
+      "learning_rate": 9.418044208610013e-06,
+      "loss": 85.5561,
+      "step": 29650
+    },
+    {
+      "epoch": 0.11983015308039448,
+      "grad_norm": 801.6728515625,
+      "learning_rate": 9.41738366258363e-06,
+      "loss": 40.399,
+      "step": 29660
+    },
+    {
+      "epoch": 0.1198705543457621,
+      "grad_norm": 874.2803344726562,
+      "learning_rate": 9.416722765086304e-06,
+      "loss": 50.6003,
+      "step": 29670
+    },
+    {
+      "epoch": 0.11991095561112974,
+      "grad_norm": 1156.8819580078125,
+      "learning_rate": 9.416061516170615e-06,
+      "loss": 49.4536,
+      "step": 29680
+    },
+    {
+      "epoch": 0.11995135687649737,
+      "grad_norm": 627.3614501953125,
+      "learning_rate": 9.415399915889179e-06,
+      "loss": 45.6941,
+      "step": 29690
+    },
+    {
+      "epoch": 0.11999175814186501,
+      "grad_norm": 559.5956420898438,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 56.8244,
+      "step": 29700
+    },
+    {
+      "epoch": 0.12003215940723264,
+      "grad_norm": 653.4930419921875,
+      "learning_rate": 9.414075661439653e-06,
+      "loss": 57.2013,
+      "step": 29710
+    },
+    {
+      "epoch": 0.12007256067260026,
+      "grad_norm": 3804.873291015625,
+      "learning_rate": 9.413413007376928e-06,
+      "loss": 67.4335,
+      "step": 29720
+    },
+    {
+      "epoch": 0.1201129619379679,
+      "grad_norm": 703.4122924804688,
+      "learning_rate": 9.412750002159186e-06,
+      "loss": 60.5386,
+      "step": 29730
+    },
+    {
+      "epoch": 0.12015336320333553,
+      "grad_norm": 396.68994140625,
+      "learning_rate": 9.412086645839177e-06,
+      "loss": 50.3618,
+      "step": 29740
+    },
+    {
+      "epoch": 0.12019376446870315,
+      "grad_norm": 1265.5760498046875,
+      "learning_rate": 9.411422938469683e-06,
+      "loss": 47.9156,
+      "step": 29750
+    },
+    {
+      "epoch": 0.1202341657340708,
+      "grad_norm": 402.34283447265625,
+      "learning_rate": 9.41075888010351e-06,
+      "loss": 51.1713,
+      "step": 29760
+    },
+    {
+      "epoch": 0.12027456699943842,
+      "grad_norm": 712.1184692382812,
+      "learning_rate": 9.410094470793497e-06,
+      "loss": 52.6493,
+      "step": 29770
+    },
+    {
+      "epoch": 0.12031496826480606,
+      "grad_norm": 632.7574462890625,
+      "learning_rate": 9.409429710592505e-06,
+      "loss": 46.334,
+      "step": 29780
+    },
+    {
+      "epoch": 0.12035536953017369,
+      "grad_norm": 186.35232543945312,
+      "learning_rate": 9.408764599553429e-06,
+      "loss": 43.2821,
+      "step": 29790
+    },
+    {
+      "epoch": 0.12039577079554131,
+      "grad_norm": 747.8989868164062,
+      "learning_rate": 9.408099137729188e-06,
+      "loss": 45.6112,
+      "step": 29800
+    },
+    {
+      "epoch": 0.12043617206090895,
+      "grad_norm": 1416.9267578125,
+      "learning_rate": 9.407433325172727e-06,
+      "loss": 39.188,
+      "step": 29810
+    },
+    {
+      "epoch": 0.12047657332627658,
+      "grad_norm": 594.1754760742188,
+      "learning_rate": 9.406767161937025e-06,
+      "loss": 47.1414,
+      "step": 29820
+    },
+    {
+      "epoch": 0.1205169745916442,
+      "grad_norm": 420.6256103515625,
+      "learning_rate": 9.406100648075084e-06,
+      "loss": 47.8546,
+      "step": 29830
+    },
+    {
+      "epoch": 0.12055737585701184,
+      "grad_norm": 911.7869262695312,
+      "learning_rate": 9.405433783639936e-06,
+      "loss": 43.1674,
+      "step": 29840
+    },
+    {
+      "epoch": 0.12059777712237947,
+      "grad_norm": 920.2177124023438,
+      "learning_rate": 9.40476656868464e-06,
+      "loss": 44.3151,
+      "step": 29850
+    },
+    {
+      "epoch": 0.12063817838774711,
+      "grad_norm": 648.54833984375,
+      "learning_rate": 9.404099003262282e-06,
+      "loss": 35.8031,
+      "step": 29860
+    },
+    {
+      "epoch": 0.12067857965311474,
+      "grad_norm": 658.5719604492188,
+      "learning_rate": 9.40343108742598e-06,
+      "loss": 46.2772,
+      "step": 29870
+    },
+    {
+      "epoch": 0.12071898091848236,
+      "grad_norm": 662.2303466796875,
+      "learning_rate": 9.402762821228875e-06,
+      "loss": 45.1511,
+      "step": 29880
+    },
+    {
+      "epoch": 0.12075938218385,
+      "grad_norm": 829.085205078125,
+      "learning_rate": 9.402094204724138e-06,
+      "loss": 56.8606,
+      "step": 29890
+    },
+    {
+      "epoch": 0.12079978344921763,
+      "grad_norm": 356.94500732421875,
+      "learning_rate": 9.401425237964966e-06,
+      "loss": 64.053,
+      "step": 29900
+    },
+    {
+      "epoch": 0.12084018471458526,
+      "grad_norm": 726.9462890625,
+      "learning_rate": 9.400755921004592e-06,
+      "loss": 55.208,
+      "step": 29910
+    },
+    {
+      "epoch": 0.1208805859799529,
+      "grad_norm": 389.7232666015625,
+      "learning_rate": 9.400086253896264e-06,
+      "loss": 54.952,
+      "step": 29920
+    },
+    {
+      "epoch": 0.12092098724532052,
+      "grad_norm": 2380.118896484375,
+      "learning_rate": 9.399416236693264e-06,
+      "loss": 80.5505,
+      "step": 29930
+    },
+    {
+      "epoch": 0.12096138851068816,
+      "grad_norm": 451.5255432128906,
+      "learning_rate": 9.398745869448909e-06,
+      "loss": 44.9507,
+      "step": 29940
+    },
+    {
+      "epoch": 0.12100178977605579,
+      "grad_norm": 728.6965942382812,
+      "learning_rate": 9.39807515221653e-06,
+      "loss": 76.1519,
+      "step": 29950
+    },
+    {
+      "epoch": 0.12104219104142341,
+      "grad_norm": 1228.5821533203125,
+      "learning_rate": 9.397404085049496e-06,
+      "loss": 64.1188,
+      "step": 29960
+    },
+    {
+      "epoch": 0.12108259230679105,
+      "grad_norm": 668.8536987304688,
+      "learning_rate": 9.3967326680012e-06,
+      "loss": 56.7463,
+      "step": 29970
+    },
+    {
+      "epoch": 0.12112299357215868,
+      "grad_norm": 1020.1240234375,
+      "learning_rate": 9.396060901125064e-06,
+      "loss": 52.0907,
+      "step": 29980
+    },
+    {
+      "epoch": 0.1211633948375263,
+      "grad_norm": 1008.449951171875,
+      "learning_rate": 9.395388784474538e-06,
+      "loss": 38.816,
+      "step": 29990
+    },
+    {
+      "epoch": 0.12120379610289395,
+      "grad_norm": 828.15234375,
+      "learning_rate": 9.394716318103098e-06,
+      "loss": 67.8823,
+      "step": 30000
+    },
+    {
+      "epoch": 0.12124419736826157,
+      "grad_norm": 242.62425231933594,
+      "learning_rate": 9.394043502064249e-06,
+      "loss": 47.3324,
+      "step": 30010
+    },
+    {
+      "epoch": 0.12128459863362921,
+      "grad_norm": 615.1395263671875,
+      "learning_rate": 9.393370336411527e-06,
+      "loss": 63.4642,
+      "step": 30020
+    },
+    {
+      "epoch": 0.12132499989899684,
+      "grad_norm": 610.5171508789062,
+      "learning_rate": 9.392696821198488e-06,
+      "loss": 53.2339,
+      "step": 30030
+    },
+    {
+      "epoch": 0.12136540116436446,
+      "grad_norm": 685.2667846679688,
+      "learning_rate": 9.392022956478724e-06,
+      "loss": 47.9081,
+      "step": 30040
+    },
+    {
+      "epoch": 0.1214058024297321,
+      "grad_norm": 585.4192504882812,
+      "learning_rate": 9.391348742305849e-06,
+      "loss": 66.8327,
+      "step": 30050
+    },
+    {
+      "epoch": 0.12144620369509973,
+      "grad_norm": 737.0886840820312,
+      "learning_rate": 9.390674178733508e-06,
+      "loss": 51.1965,
+      "step": 30060
+    },
+    {
+      "epoch": 0.12148660496046736,
+      "grad_norm": 450.2093505859375,
+      "learning_rate": 9.389999265815373e-06,
+      "loss": 59.4211,
+      "step": 30070
+    },
+    {
+      "epoch": 0.121527006225835,
+      "grad_norm": 134.6776885986328,
+      "learning_rate": 9.389324003605144e-06,
+      "loss": 67.8788,
+      "step": 30080
+    },
+    {
+      "epoch": 0.12156740749120262,
+      "grad_norm": 995.7404174804688,
+      "learning_rate": 9.388648392156547e-06,
+      "loss": 49.0113,
+      "step": 30090
+    },
+    {
+      "epoch": 0.12160780875657026,
+      "grad_norm": 561.1322021484375,
+      "learning_rate": 9.387972431523341e-06,
+      "loss": 39.4004,
+      "step": 30100
+    },
+    {
+      "epoch": 0.12164821002193789,
+      "grad_norm": 746.4854125976562,
+      "learning_rate": 9.387296121759305e-06,
+      "loss": 45.6675,
+      "step": 30110
+    },
+    {
+      "epoch": 0.12168861128730551,
+      "grad_norm": 412.273193359375,
+      "learning_rate": 9.386619462918254e-06,
+      "loss": 47.4065,
+      "step": 30120
+    },
+    {
+      "epoch": 0.12172901255267315,
+      "grad_norm": 2498.973876953125,
+      "learning_rate": 9.385942455054022e-06,
+      "loss": 54.9776,
+      "step": 30130
+    },
+    {
+      "epoch": 0.12176941381804078,
+      "grad_norm": 584.4258422851562,
+      "learning_rate": 9.385265098220478e-06,
+      "loss": 47.1616,
+      "step": 30140
+    },
+    {
+      "epoch": 0.12180981508340841,
+      "grad_norm": 768.3419799804688,
+      "learning_rate": 9.384587392471516e-06,
+      "loss": 50.8677,
+      "step": 30150
+    },
+    {
+      "epoch": 0.12185021634877605,
+      "grad_norm": 722.3310546875,
+      "learning_rate": 9.383909337861058e-06,
+      "loss": 68.8301,
+      "step": 30160
+    },
+    {
+      "epoch": 0.12189061761414367,
+      "grad_norm": 347.5932922363281,
+      "learning_rate": 9.383230934443053e-06,
+      "loss": 51.4054,
+      "step": 30170
+    },
+    {
+      "epoch": 0.12193101887951131,
+      "grad_norm": 373.2593688964844,
+      "learning_rate": 9.382552182271478e-06,
+      "loss": 49.6567,
+      "step": 30180
+    },
+    {
+      "epoch": 0.12197142014487894,
+      "grad_norm": 1203.6734619140625,
+      "learning_rate": 9.38187308140034e-06,
+      "loss": 56.8176,
+      "step": 30190
+    },
+    {
+      "epoch": 0.12201182141024657,
+      "grad_norm": 727.6742553710938,
+      "learning_rate": 9.381193631883672e-06,
+      "loss": 39.6272,
+      "step": 30200
+    },
+    {
+      "epoch": 0.1220522226756142,
+      "grad_norm": 619.8937377929688,
+      "learning_rate": 9.380513833775531e-06,
+      "loss": 44.3137,
+      "step": 30210
+    },
+    {
+      "epoch": 0.12209262394098183,
+      "grad_norm": 540.3925170898438,
+      "learning_rate": 9.37983368713001e-06,
+      "loss": 43.6342,
+      "step": 30220
+    },
+    {
+      "epoch": 0.12213302520634946,
+      "grad_norm": 5220.17626953125,
+      "learning_rate": 9.379153192001223e-06,
+      "loss": 59.757,
+      "step": 30230
+    },
+    {
+      "epoch": 0.1221734264717171,
+      "grad_norm": 963.0521850585938,
+      "learning_rate": 9.378472348443315e-06,
+      "loss": 34.9574,
+      "step": 30240
+    },
+    {
+      "epoch": 0.12221382773708472,
+      "grad_norm": 399.6465759277344,
+      "learning_rate": 9.377791156510456e-06,
+      "loss": 35.0663,
+      "step": 30250
+    },
+    {
+      "epoch": 0.12225422900245236,
+      "grad_norm": 524.9255981445312,
+      "learning_rate": 9.377109616256846e-06,
+      "loss": 53.069,
+      "step": 30260
+    },
+    {
+      "epoch": 0.12229463026781999,
+      "grad_norm": 573.6826782226562,
+      "learning_rate": 9.37642772773671e-06,
+      "loss": 31.281,
+      "step": 30270
+    },
+    {
+      "epoch": 0.12233503153318762,
+      "grad_norm": 430.6109313964844,
+      "learning_rate": 9.375745491004307e-06,
+      "loss": 35.5072,
+      "step": 30280
+    },
+    {
+      "epoch": 0.12237543279855526,
+      "grad_norm": 309.88275146484375,
+      "learning_rate": 9.375062906113916e-06,
+      "loss": 47.0233,
+      "step": 30290
+    },
+    {
+      "epoch": 0.12241583406392288,
+      "grad_norm": 873.5589599609375,
+      "learning_rate": 9.37437997311985e-06,
+      "loss": 49.2769,
+      "step": 30300
+    },
+    {
+      "epoch": 0.12245623532929051,
+      "grad_norm": 958.1603393554688,
+      "learning_rate": 9.373696692076446e-06,
+      "loss": 64.9906,
+      "step": 30310
+    },
+    {
+      "epoch": 0.12249663659465815,
+      "grad_norm": 935.3135375976562,
+      "learning_rate": 9.373013063038066e-06,
+      "loss": 46.1228,
+      "step": 30320
+    },
+    {
+      "epoch": 0.12253703786002577,
+      "grad_norm": 632.3682861328125,
+      "learning_rate": 9.372329086059108e-06,
+      "loss": 61.2088,
+      "step": 30330
+    },
+    {
+      "epoch": 0.12257743912539341,
+      "grad_norm": 914.8970336914062,
+      "learning_rate": 9.37164476119399e-06,
+      "loss": 58.5277,
+      "step": 30340
+    },
+    {
+      "epoch": 0.12261784039076104,
+      "grad_norm": 900.7513427734375,
+      "learning_rate": 9.370960088497162e-06,
+      "loss": 45.0407,
+      "step": 30350
+    },
+    {
+      "epoch": 0.12265824165612867,
+      "grad_norm": 491.3097229003906,
+      "learning_rate": 9.370275068023097e-06,
+      "loss": 64.0002,
+      "step": 30360
+    },
+    {
+      "epoch": 0.1226986429214963,
+      "grad_norm": 579.8344116210938,
+      "learning_rate": 9.369589699826306e-06,
+      "loss": 60.9695,
+      "step": 30370
+    },
+    {
+      "epoch": 0.12273904418686393,
+      "grad_norm": 410.3919677734375,
+      "learning_rate": 9.368903983961315e-06,
+      "loss": 43.6325,
+      "step": 30380
+    },
+    {
+      "epoch": 0.12277944545223156,
+      "grad_norm": 201.71817016601562,
+      "learning_rate": 9.368217920482684e-06,
+      "loss": 53.0613,
+      "step": 30390
+    },
+    {
+      "epoch": 0.1228198467175992,
+      "grad_norm": 408.74169921875,
+      "learning_rate": 9.367531509445001e-06,
+      "loss": 55.3284,
+      "step": 30400
+    },
+    {
+      "epoch": 0.12286024798296682,
+      "grad_norm": 355.0393981933594,
+      "learning_rate": 9.366844750902878e-06,
+      "loss": 43.9765,
+      "step": 30410
+    },
+    {
+      "epoch": 0.12290064924833445,
+      "grad_norm": 513.59765625,
+      "learning_rate": 9.36615764491096e-06,
+      "loss": 34.4743,
+      "step": 30420
+    },
+    {
+      "epoch": 0.12294105051370209,
+      "grad_norm": 894.3479614257812,
+      "learning_rate": 9.365470191523917e-06,
+      "loss": 41.8912,
+      "step": 30430
+    },
+    {
+      "epoch": 0.12298145177906972,
+      "grad_norm": 601.699462890625,
+      "learning_rate": 9.364782390796446e-06,
+      "loss": 49.0976,
+      "step": 30440
+    },
+    {
+      "epoch": 0.12302185304443736,
+      "grad_norm": 399.9588317871094,
+      "learning_rate": 9.364094242783272e-06,
+      "loss": 34.56,
+      "step": 30450
+    },
+    {
+      "epoch": 0.12306225430980498,
+      "grad_norm": 1018.5534057617188,
+      "learning_rate": 9.363405747539147e-06,
+      "loss": 50.4698,
+      "step": 30460
+    },
+    {
+      "epoch": 0.12310265557517261,
+      "grad_norm": 898.0135498046875,
+      "learning_rate": 9.362716905118851e-06,
+      "loss": 53.0371,
+      "step": 30470
+    },
+    {
+      "epoch": 0.12314305684054025,
+      "grad_norm": 941.5891723632812,
+      "learning_rate": 9.362027715577195e-06,
+      "loss": 57.5847,
+      "step": 30480
+    },
+    {
+      "epoch": 0.12318345810590788,
+      "grad_norm": 507.96929931640625,
+      "learning_rate": 9.361338178969012e-06,
+      "loss": 60.7366,
+      "step": 30490
+    },
+    {
+      "epoch": 0.1232238593712755,
+      "grad_norm": 668.412353515625,
+      "learning_rate": 9.360648295349165e-06,
+      "loss": 59.3322,
+      "step": 30500
+    },
+    {
+      "epoch": 0.12326426063664314,
+      "grad_norm": 1075.7337646484375,
+      "learning_rate": 9.359958064772547e-06,
+      "loss": 59.0285,
+      "step": 30510
+    },
+    {
+      "epoch": 0.12330466190201077,
+      "grad_norm": 760.0520629882812,
+      "learning_rate": 9.359267487294075e-06,
+      "loss": 51.6215,
+      "step": 30520
+    },
+    {
+      "epoch": 0.12334506316737841,
+      "grad_norm": 1032.9842529296875,
+      "learning_rate": 9.358576562968695e-06,
+      "loss": 45.0181,
+      "step": 30530
+    },
+    {
+      "epoch": 0.12338546443274603,
+      "grad_norm": 369.5265197753906,
+      "learning_rate": 9.357885291851382e-06,
+      "loss": 50.1647,
+      "step": 30540
+    },
+    {
+      "epoch": 0.12342586569811366,
+      "grad_norm": 518.7932739257812,
+      "learning_rate": 9.357193673997133e-06,
+      "loss": 58.2824,
+      "step": 30550
+    },
+    {
+      "epoch": 0.1234662669634813,
+      "grad_norm": 828.255859375,
+      "learning_rate": 9.356501709460984e-06,
+      "loss": 39.197,
+      "step": 30560
+    },
+    {
+      "epoch": 0.12350666822884893,
+      "grad_norm": 1110.9344482421875,
+      "learning_rate": 9.355809398297986e-06,
+      "loss": 52.1149,
+      "step": 30570
+    },
+    {
+      "epoch": 0.12354706949421655,
+      "grad_norm": 973.9660034179688,
+      "learning_rate": 9.355116740563225e-06,
+      "loss": 37.8704,
+      "step": 30580
+    },
+    {
+      "epoch": 0.12358747075958419,
+      "grad_norm": 919.6453247070312,
+      "learning_rate": 9.354423736311813e-06,
+      "loss": 43.1774,
+      "step": 30590
+    },
+    {
+      "epoch": 0.12362787202495182,
+      "grad_norm": 295.9122009277344,
+      "learning_rate": 9.353730385598887e-06,
+      "loss": 49.3843,
+      "step": 30600
+    },
+    {
+      "epoch": 0.12366827329031946,
+      "grad_norm": 513.4649047851562,
+      "learning_rate": 9.353036688479615e-06,
+      "loss": 57.7642,
+      "step": 30610
+    },
+    {
+      "epoch": 0.12370867455568708,
+      "grad_norm": 777.9099731445312,
+      "learning_rate": 9.352342645009193e-06,
+      "loss": 55.8895,
+      "step": 30620
+    },
+    {
+      "epoch": 0.12374907582105471,
+      "grad_norm": 917.2174072265625,
+      "learning_rate": 9.35164825524284e-06,
+      "loss": 59.0916,
+      "step": 30630
+    },
+    {
+      "epoch": 0.12378947708642235,
+      "grad_norm": 1081.7359619140625,
+      "learning_rate": 9.350953519235807e-06,
+      "loss": 52.1535,
+      "step": 30640
+    },
+    {
+      "epoch": 0.12382987835178998,
+      "grad_norm": 960.4583740234375,
+      "learning_rate": 9.35025843704337e-06,
+      "loss": 46.3987,
+      "step": 30650
+    },
+    {
+      "epoch": 0.1238702796171576,
+      "grad_norm": 569.8470458984375,
+      "learning_rate": 9.349563008720836e-06,
+      "loss": 42.4655,
+      "step": 30660
+    },
+    {
+      "epoch": 0.12391068088252524,
+      "grad_norm": 456.7442932128906,
+      "learning_rate": 9.348867234323534e-06,
+      "loss": 35.9404,
+      "step": 30670
+    },
+    {
+      "epoch": 0.12395108214789287,
+      "grad_norm": 580.541748046875,
+      "learning_rate": 9.348171113906826e-06,
+      "loss": 40.7233,
+      "step": 30680
+    },
+    {
+      "epoch": 0.12399148341326051,
+      "grad_norm": 542.5335083007812,
+      "learning_rate": 9.347474647526095e-06,
+      "loss": 63.2249,
+      "step": 30690
+    },
+    {
+      "epoch": 0.12403188467862813,
+      "grad_norm": 822.7742919921875,
+      "learning_rate": 9.34677783523676e-06,
+      "loss": 46.3272,
+      "step": 30700
+    },
+    {
+      "epoch": 0.12407228594399576,
+      "grad_norm": 1118.610595703125,
+      "learning_rate": 9.346080677094262e-06,
+      "loss": 42.0597,
+      "step": 30710
+    },
+    {
+      "epoch": 0.1241126872093634,
+      "grad_norm": 1029.7388916015625,
+      "learning_rate": 9.345383173154072e-06,
+      "loss": 45.3306,
+      "step": 30720
+    },
+    {
+      "epoch": 0.12415308847473103,
+      "grad_norm": 314.6160583496094,
+      "learning_rate": 9.344685323471682e-06,
+      "loss": 24.6263,
+      "step": 30730
+    },
+    {
+      "epoch": 0.12419348974009865,
+      "grad_norm": 2012.316650390625,
+      "learning_rate": 9.343987128102624e-06,
+      "loss": 65.5848,
+      "step": 30740
+    },
+    {
+      "epoch": 0.1242338910054663,
+      "grad_norm": 1248.8795166015625,
+      "learning_rate": 9.343288587102444e-06,
+      "loss": 42.0409,
+      "step": 30750
+    },
+    {
+      "epoch": 0.12427429227083392,
+      "grad_norm": 509.92193603515625,
+      "learning_rate": 9.342589700526725e-06,
+      "loss": 36.0369,
+      "step": 30760
+    },
+    {
+      "epoch": 0.12431469353620156,
+      "grad_norm": 649.0388793945312,
+      "learning_rate": 9.341890468431072e-06,
+      "loss": 41.3814,
+      "step": 30770
+    },
+    {
+      "epoch": 0.12435509480156919,
+      "grad_norm": 743.8023681640625,
+      "learning_rate": 9.341190890871123e-06,
+      "loss": 47.9775,
+      "step": 30780
+    },
+    {
+      "epoch": 0.12439549606693681,
+      "grad_norm": 1114.2352294921875,
+      "learning_rate": 9.340490967902535e-06,
+      "loss": 50.7062,
+      "step": 30790
+    },
+    {
+      "epoch": 0.12443589733230445,
+      "grad_norm": 940.1904907226562,
+      "learning_rate": 9.339790699581004e-06,
+      "loss": 63.2531,
+      "step": 30800
+    },
+    {
+      "epoch": 0.12447629859767208,
+      "grad_norm": 1059.23193359375,
+      "learning_rate": 9.339090085962244e-06,
+      "loss": 39.3221,
+      "step": 30810
+    },
+    {
+      "epoch": 0.1245166998630397,
+      "grad_norm": 397.91900634765625,
+      "learning_rate": 9.338389127101998e-06,
+      "loss": 44.4031,
+      "step": 30820
+    },
+    {
+      "epoch": 0.12455710112840734,
+      "grad_norm": 765.754638671875,
+      "learning_rate": 9.337687823056041e-06,
+      "loss": 44.932,
+      "step": 30830
+    },
+    {
+      "epoch": 0.12459750239377497,
+      "grad_norm": 554.5352172851562,
+      "learning_rate": 9.336986173880169e-06,
+      "loss": 54.2205,
+      "step": 30840
+    },
+    {
+      "epoch": 0.12463790365914261,
+      "grad_norm": 546.3801879882812,
+      "learning_rate": 9.336284179630215e-06,
+      "loss": 37.6779,
+      "step": 30850
+    },
+    {
+      "epoch": 0.12467830492451024,
+      "grad_norm": 84.68031311035156,
+      "learning_rate": 9.335581840362026e-06,
+      "loss": 53.455,
+      "step": 30860
+    },
+    {
+      "epoch": 0.12471870618987786,
+      "grad_norm": 656.2481689453125,
+      "learning_rate": 9.33487915613149e-06,
+      "loss": 42.2726,
+      "step": 30870
+    },
+    {
+      "epoch": 0.1247591074552455,
+      "grad_norm": 810.590087890625,
+      "learning_rate": 9.334176126994512e-06,
+      "loss": 40.2841,
+      "step": 30880
+    },
+    {
+      "epoch": 0.12479950872061313,
+      "grad_norm": 514.6560668945312,
+      "learning_rate": 9.333472753007031e-06,
+      "loss": 91.8756,
+      "step": 30890
+    },
+    {
+      "epoch": 0.12483990998598075,
+      "grad_norm": 838.5552368164062,
+      "learning_rate": 9.332769034225012e-06,
+      "loss": 38.2255,
+      "step": 30900
+    },
+    {
+      "epoch": 0.1248803112513484,
+      "grad_norm": 335.7632141113281,
+      "learning_rate": 9.332064970704445e-06,
+      "loss": 49.4308,
+      "step": 30910
+    },
+    {
+      "epoch": 0.12492071251671602,
+      "grad_norm": 540.861572265625,
+      "learning_rate": 9.33136056250135e-06,
+      "loss": 39.3426,
+      "step": 30920
+    },
+    {
+      "epoch": 0.12496111378208366,
+      "grad_norm": 622.8423461914062,
+      "learning_rate": 9.330655809671773e-06,
+      "loss": 47.4215,
+      "step": 30930
+    },
+    {
+      "epoch": 0.12500151504745127,
+      "grad_norm": 360.4191589355469,
+      "learning_rate": 9.32995071227179e-06,
+      "loss": 50.474,
+      "step": 30940
+    },
+    {
+      "epoch": 0.1250419163128189,
+      "grad_norm": 407.1855773925781,
+      "learning_rate": 9.3292452703575e-06,
+      "loss": 39.8224,
+      "step": 30950
+    },
+    {
+      "epoch": 0.12508231757818655,
+      "grad_norm": 773.7116088867188,
+      "learning_rate": 9.328539483985031e-06,
+      "loss": 110.1445,
+      "step": 30960
+    },
+    {
+      "epoch": 0.1251227188435542,
+      "grad_norm": 305.62103271484375,
+      "learning_rate": 9.327833353210541e-06,
+      "loss": 53.3745,
+      "step": 30970
+    },
+    {
+      "epoch": 0.1251631201089218,
+      "grad_norm": 477.38482666015625,
+      "learning_rate": 9.327126878090214e-06,
+      "loss": 42.7223,
+      "step": 30980
+    },
+    {
+      "epoch": 0.12520352137428944,
+      "grad_norm": 635.5834350585938,
+      "learning_rate": 9.32642005868026e-06,
+      "loss": 47.8754,
+      "step": 30990
+    },
+    {
+      "epoch": 0.12524392263965708,
+      "grad_norm": 721.0254516601562,
+      "learning_rate": 9.325712895036916e-06,
+      "loss": 57.4999,
+      "step": 31000
+    },
+    {
+      "epoch": 0.1252843239050247,
+      "grad_norm": 443.795166015625,
+      "learning_rate": 9.32500538721645e-06,
+      "loss": 41.7979,
+      "step": 31010
+    },
+    {
+      "epoch": 0.12532472517039234,
+      "grad_norm": 863.4866333007812,
+      "learning_rate": 9.324297535275156e-06,
+      "loss": 60.9883,
+      "step": 31020
+    },
+    {
+      "epoch": 0.12536512643575998,
+      "grad_norm": 582.2689208984375,
+      "learning_rate": 9.323589339269352e-06,
+      "loss": 53.3317,
+      "step": 31030
+    },
+    {
+      "epoch": 0.1254055277011276,
+      "grad_norm": 799.9541625976562,
+      "learning_rate": 9.322880799255385e-06,
+      "loss": 73.9565,
+      "step": 31040
+    },
+    {
+      "epoch": 0.12544592896649523,
+      "grad_norm": 1019.82861328125,
+      "learning_rate": 9.322171915289635e-06,
+      "loss": 37.8343,
+      "step": 31050
+    },
+    {
+      "epoch": 0.12548633023186287,
+      "grad_norm": 759.5281372070312,
+      "learning_rate": 9.321462687428499e-06,
+      "loss": 47.9594,
+      "step": 31060
+    },
+    {
+      "epoch": 0.12552673149723048,
+      "grad_norm": 686.42822265625,
+      "learning_rate": 9.320753115728413e-06,
+      "loss": 49.3588,
+      "step": 31070
+    },
+    {
+      "epoch": 0.12556713276259812,
+      "grad_norm": 761.8621215820312,
+      "learning_rate": 9.320043200245829e-06,
+      "loss": 59.4951,
+      "step": 31080
+    },
+    {
+      "epoch": 0.12560753402796576,
+      "grad_norm": 1079.365234375,
+      "learning_rate": 9.319332941037235e-06,
+      "loss": 71.4847,
+      "step": 31090
+    },
+    {
+      "epoch": 0.12564793529333337,
+      "grad_norm": 302.6757507324219,
+      "learning_rate": 9.31862233815914e-06,
+      "loss": 51.828,
+      "step": 31100
+    },
+    {
+      "epoch": 0.125688336558701,
+      "grad_norm": 644.4342651367188,
+      "learning_rate": 9.317911391668087e-06,
+      "loss": 43.484,
+      "step": 31110
+    },
+    {
+      "epoch": 0.12572873782406865,
+      "grad_norm": 829.1041259765625,
+      "learning_rate": 9.317200101620641e-06,
+      "loss": 44.3326,
+      "step": 31120
+    },
+    {
+      "epoch": 0.1257691390894363,
+      "grad_norm": 398.044921875,
+      "learning_rate": 9.316488468073397e-06,
+      "loss": 50.6066,
+      "step": 31130
+    },
+    {
+      "epoch": 0.1258095403548039,
+      "grad_norm": 592.14208984375,
+      "learning_rate": 9.315776491082973e-06,
+      "loss": 52.6301,
+      "step": 31140
+    },
+    {
+      "epoch": 0.12584994162017155,
+      "grad_norm": 201.01861572265625,
+      "learning_rate": 9.315064170706023e-06,
+      "loss": 48.0798,
+      "step": 31150
+    },
+    {
+      "epoch": 0.12589034288553919,
+      "grad_norm": 738.100830078125,
+      "learning_rate": 9.31435150699922e-06,
+      "loss": 51.338,
+      "step": 31160
+    },
+    {
+      "epoch": 0.1259307441509068,
+      "grad_norm": 756.7727661132812,
+      "learning_rate": 9.313638500019267e-06,
+      "loss": 54.9391,
+      "step": 31170
+    },
+    {
+      "epoch": 0.12597114541627444,
+      "grad_norm": 287.8023986816406,
+      "learning_rate": 9.312925149822895e-06,
+      "loss": 28.0471,
+      "step": 31180
+    },
+    {
+      "epoch": 0.12601154668164208,
+      "grad_norm": 723.8349609375,
+      "learning_rate": 9.312211456466862e-06,
+      "loss": 36.3156,
+      "step": 31190
+    },
+    {
+      "epoch": 0.1260519479470097,
+      "grad_norm": 878.0177612304688,
+      "learning_rate": 9.311497420007955e-06,
+      "loss": 61.6736,
+      "step": 31200
+    },
+    {
+      "epoch": 0.12609234921237733,
+      "grad_norm": 1783.608154296875,
+      "learning_rate": 9.310783040502987e-06,
+      "loss": 45.5743,
+      "step": 31210
+    },
+    {
+      "epoch": 0.12613275047774497,
+      "grad_norm": 1379.0848388671875,
+      "learning_rate": 9.310068318008794e-06,
+      "loss": 72.9638,
+      "step": 31220
+    },
+    {
+      "epoch": 0.12617315174311258,
+      "grad_norm": 477.46966552734375,
+      "learning_rate": 9.309353252582246e-06,
+      "loss": 45.4072,
+      "step": 31230
+    },
+    {
+      "epoch": 0.12621355300848022,
+      "grad_norm": 495.3040466308594,
+      "learning_rate": 9.308637844280236e-06,
+      "loss": 34.8518,
+      "step": 31240
+    },
+    {
+      "epoch": 0.12625395427384786,
+      "grad_norm": 834.5982055664062,
+      "learning_rate": 9.307922093159688e-06,
+      "loss": 53.4758,
+      "step": 31250
+    },
+    {
+      "epoch": 0.12629435553921547,
+      "grad_norm": 2230.323486328125,
+      "learning_rate": 9.30720599927755e-06,
+      "loss": 78.3272,
+      "step": 31260
+    },
+    {
+      "epoch": 0.12633475680458311,
+      "grad_norm": 433.3248291015625,
+      "learning_rate": 9.306489562690797e-06,
+      "loss": 35.0247,
+      "step": 31270
+    },
+    {
+      "epoch": 0.12637515806995075,
+      "grad_norm": 868.4600830078125,
+      "learning_rate": 9.305772783456435e-06,
+      "loss": 35.0469,
+      "step": 31280
+    },
+    {
+      "epoch": 0.1264155593353184,
+      "grad_norm": 406.2452697753906,
+      "learning_rate": 9.305055661631493e-06,
+      "loss": 76.8342,
+      "step": 31290
+    },
+    {
+      "epoch": 0.126455960600686,
+      "grad_norm": 443.42449951171875,
+      "learning_rate": 9.304338197273029e-06,
+      "loss": 39.4005,
+      "step": 31300
+    },
+    {
+      "epoch": 0.12649636186605365,
+      "grad_norm": 713.2732543945312,
+      "learning_rate": 9.303620390438128e-06,
+      "loss": 55.5563,
+      "step": 31310
+    },
+    {
+      "epoch": 0.1265367631314213,
+      "grad_norm": 710.0195922851562,
+      "learning_rate": 9.302902241183905e-06,
+      "loss": 45.8039,
+      "step": 31320
+    },
+    {
+      "epoch": 0.1265771643967889,
+      "grad_norm": 808.3165283203125,
+      "learning_rate": 9.302183749567498e-06,
+      "loss": 31.8709,
+      "step": 31330
+    },
+    {
+      "epoch": 0.12661756566215654,
+      "grad_norm": 558.5325927734375,
+      "learning_rate": 9.301464915646074e-06,
+      "loss": 40.9993,
+      "step": 31340
+    },
+    {
+      "epoch": 0.12665796692752418,
+      "grad_norm": 374.66131591796875,
+      "learning_rate": 9.30074573947683e-06,
+      "loss": 44.0041,
+      "step": 31350
+    },
+    {
+      "epoch": 0.1266983681928918,
+      "grad_norm": 549.226318359375,
+      "learning_rate": 9.30002622111698e-06,
+      "loss": 44.0095,
+      "step": 31360
+    },
+    {
+      "epoch": 0.12673876945825943,
+      "grad_norm": 198.41749572753906,
+      "learning_rate": 9.299306360623782e-06,
+      "loss": 51.1259,
+      "step": 31370
+    },
+    {
+      "epoch": 0.12677917072362707,
+      "grad_norm": 907.3374633789062,
+      "learning_rate": 9.298586158054508e-06,
+      "loss": 62.2416,
+      "step": 31380
+    },
+    {
+      "epoch": 0.12681957198899468,
+      "grad_norm": 1143.6590576171875,
+      "learning_rate": 9.297865613466459e-06,
+      "loss": 57.0977,
+      "step": 31390
+    },
+    {
+      "epoch": 0.12685997325436232,
+      "grad_norm": 301.8861389160156,
+      "learning_rate": 9.29714472691697e-06,
+      "loss": 37.5276,
+      "step": 31400
+    },
+    {
+      "epoch": 0.12690037451972996,
+      "grad_norm": 1299.40087890625,
+      "learning_rate": 9.296423498463396e-06,
+      "loss": 64.7403,
+      "step": 31410
+    },
+    {
+      "epoch": 0.12694077578509758,
+      "grad_norm": 624.0057983398438,
+      "learning_rate": 9.29570192816312e-06,
+      "loss": 52.6827,
+      "step": 31420
+    },
+    {
+      "epoch": 0.12698117705046522,
+      "grad_norm": 185.71693420410156,
+      "learning_rate": 9.29498001607356e-06,
+      "loss": 52.4407,
+      "step": 31430
+    },
+    {
+      "epoch": 0.12702157831583286,
+      "grad_norm": 527.6602783203125,
+      "learning_rate": 9.294257762252148e-06,
+      "loss": 37.781,
+      "step": 31440
+    },
+    {
+      "epoch": 0.1270619795812005,
+      "grad_norm": 1054.0302734375,
+      "learning_rate": 9.293535166756356e-06,
+      "loss": 66.2366,
+      "step": 31450
+    },
+    {
+      "epoch": 0.1271023808465681,
+      "grad_norm": 1055.7821044921875,
+      "learning_rate": 9.292812229643674e-06,
+      "loss": 45.0993,
+      "step": 31460
+    },
+    {
+      "epoch": 0.12714278211193575,
+      "grad_norm": 1077.979736328125,
+      "learning_rate": 9.292088950971624e-06,
+      "loss": 54.6229,
+      "step": 31470
+    },
+    {
+      "epoch": 0.1271831833773034,
+      "grad_norm": 678.5938110351562,
+      "learning_rate": 9.291365330797755e-06,
+      "loss": 47.38,
+      "step": 31480
+    },
+    {
+      "epoch": 0.127223584642671,
+      "grad_norm": 712.6802978515625,
+      "learning_rate": 9.290641369179643e-06,
+      "loss": 48.0058,
+      "step": 31490
+    },
+    {
+      "epoch": 0.12726398590803864,
+      "grad_norm": 626.124267578125,
+      "learning_rate": 9.289917066174887e-06,
+      "loss": 52.2015,
+      "step": 31500
+    },
+    {
+      "epoch": 0.12730438717340628,
+      "grad_norm": 573.7828979492188,
+      "learning_rate": 9.289192421841116e-06,
+      "loss": 70.4851,
+      "step": 31510
+    },
+    {
+      "epoch": 0.1273447884387739,
+      "grad_norm": 888.5181884765625,
+      "learning_rate": 9.288467436235992e-06,
+      "loss": 40.2558,
+      "step": 31520
+    },
+    {
+      "epoch": 0.12738518970414153,
+      "grad_norm": 995.8738403320312,
+      "learning_rate": 9.287742109417194e-06,
+      "loss": 48.9089,
+      "step": 31530
+    },
+    {
+      "epoch": 0.12742559096950917,
+      "grad_norm": 302.162353515625,
+      "learning_rate": 9.287016441442435e-06,
+      "loss": 60.6778,
+      "step": 31540
+    },
+    {
+      "epoch": 0.12746599223487678,
+      "grad_norm": 576.8596801757812,
+      "learning_rate": 9.28629043236945e-06,
+      "loss": 48.8662,
+      "step": 31550
+    },
+    {
+      "epoch": 0.12750639350024442,
+      "grad_norm": 474.55120849609375,
+      "learning_rate": 9.285564082256011e-06,
+      "loss": 53.2105,
+      "step": 31560
+    },
+    {
+      "epoch": 0.12754679476561206,
+      "grad_norm": 593.5822143554688,
+      "learning_rate": 9.284837391159904e-06,
+      "loss": 61.1313,
+      "step": 31570
+    },
+    {
+      "epoch": 0.12758719603097968,
+      "grad_norm": 434.08734130859375,
+      "learning_rate": 9.284110359138951e-06,
+      "loss": 51.8877,
+      "step": 31580
+    },
+    {
+      "epoch": 0.12762759729634732,
+      "grad_norm": 1331.6153564453125,
+      "learning_rate": 9.283382986250997e-06,
+      "loss": 52.297,
+      "step": 31590
+    },
+    {
+      "epoch": 0.12766799856171496,
+      "grad_norm": 464.4027099609375,
+      "learning_rate": 9.282655272553917e-06,
+      "loss": 39.4069,
+      "step": 31600
+    },
+    {
+      "epoch": 0.1277083998270826,
+      "grad_norm": 354.9674377441406,
+      "learning_rate": 9.281927218105613e-06,
+      "loss": 48.6394,
+      "step": 31610
+    },
+    {
+      "epoch": 0.1277488010924502,
+      "grad_norm": 1142.4324951171875,
+      "learning_rate": 9.281198822964011e-06,
+      "loss": 54.2571,
+      "step": 31620
+    },
+    {
+      "epoch": 0.12778920235781785,
+      "grad_norm": 416.7544250488281,
+      "learning_rate": 9.280470087187066e-06,
+      "loss": 53.5044,
+      "step": 31630
+    },
+    {
+      "epoch": 0.1278296036231855,
+      "grad_norm": 765.3726806640625,
+      "learning_rate": 9.279741010832761e-06,
+      "loss": 56.0496,
+      "step": 31640
+    },
+    {
+      "epoch": 0.1278700048885531,
+      "grad_norm": 884.37109375,
+      "learning_rate": 9.279011593959107e-06,
+      "loss": 75.0785,
+      "step": 31650
+    },
+    {
+      "epoch": 0.12791040615392074,
+      "grad_norm": 718.0591430664062,
+      "learning_rate": 9.278281836624137e-06,
+      "loss": 68.1496,
+      "step": 31660
+    },
+    {
+      "epoch": 0.12795080741928838,
+      "grad_norm": 598.039794921875,
+      "learning_rate": 9.277551738885915e-06,
+      "loss": 49.5493,
+      "step": 31670
+    },
+    {
+      "epoch": 0.127991208684656,
+      "grad_norm": 669.0643920898438,
+      "learning_rate": 9.276821300802535e-06,
+      "loss": 44.912,
+      "step": 31680
+    },
+    {
+      "epoch": 0.12803160995002363,
+      "grad_norm": 1183.2327880859375,
+      "learning_rate": 9.276090522432109e-06,
+      "loss": 53.1531,
+      "step": 31690
+    },
+    {
+      "epoch": 0.12807201121539127,
+      "grad_norm": 278.2745666503906,
+      "learning_rate": 9.275359403832787e-06,
+      "loss": 57.979,
+      "step": 31700
+    },
+    {
+      "epoch": 0.12811241248075889,
+      "grad_norm": 545.3709106445312,
+      "learning_rate": 9.274627945062738e-06,
+      "loss": 42.9828,
+      "step": 31710
+    },
+    {
+      "epoch": 0.12815281374612653,
+      "grad_norm": 426.2801818847656,
+      "learning_rate": 9.27389614618016e-06,
+      "loss": 69.7885,
+      "step": 31720
+    },
+    {
+      "epoch": 0.12819321501149417,
+      "grad_norm": 691.3192749023438,
+      "learning_rate": 9.273164007243281e-06,
+      "loss": 50.9559,
+      "step": 31730
+    },
+    {
+      "epoch": 0.12823361627686178,
+      "grad_norm": 3655.259033203125,
+      "learning_rate": 9.272431528310354e-06,
+      "loss": 90.1712,
+      "step": 31740
+    },
+    {
+      "epoch": 0.12827401754222942,
+      "grad_norm": 812.1537475585938,
+      "learning_rate": 9.271698709439658e-06,
+      "loss": 52.1837,
+      "step": 31750
+    },
+    {
+      "epoch": 0.12831441880759706,
+      "grad_norm": 490.5054931640625,
+      "learning_rate": 9.2709655506895e-06,
+      "loss": 64.8262,
+      "step": 31760
+    },
+    {
+      "epoch": 0.1283548200729647,
+      "grad_norm": 1356.9622802734375,
+      "learning_rate": 9.270232052118214e-06,
+      "loss": 61.913,
+      "step": 31770
+    },
+    {
+      "epoch": 0.1283952213383323,
+      "grad_norm": 413.27093505859375,
+      "learning_rate": 9.26949821378416e-06,
+      "loss": 35.5495,
+      "step": 31780
+    },
+    {
+      "epoch": 0.12843562260369995,
+      "grad_norm": 477.1658935546875,
+      "learning_rate": 9.268764035745727e-06,
+      "loss": 49.7717,
+      "step": 31790
+    },
+    {
+      "epoch": 0.1284760238690676,
+      "grad_norm": 755.6824951171875,
+      "learning_rate": 9.268029518061335e-06,
+      "loss": 42.3219,
+      "step": 31800
+    },
+    {
+      "epoch": 0.1285164251344352,
+      "grad_norm": 637.2919921875,
+      "learning_rate": 9.267294660789417e-06,
+      "loss": 60.4771,
+      "step": 31810
+    },
+    {
+      "epoch": 0.12855682639980284,
+      "grad_norm": 363.55499267578125,
+      "learning_rate": 9.26655946398845e-06,
+      "loss": 60.0659,
+      "step": 31820
+    },
+    {
+      "epoch": 0.12859722766517048,
+      "grad_norm": 745.2439575195312,
+      "learning_rate": 9.265823927716927e-06,
+      "loss": 40.7839,
+      "step": 31830
+    },
+    {
+      "epoch": 0.1286376289305381,
+      "grad_norm": 627.36865234375,
+      "learning_rate": 9.26508805203337e-06,
+      "loss": 48.2277,
+      "step": 31840
+    },
+    {
+      "epoch": 0.12867803019590573,
+      "grad_norm": 484.4584655761719,
+      "learning_rate": 9.264351836996332e-06,
+      "loss": 45.7992,
+      "step": 31850
+    },
+    {
+      "epoch": 0.12871843146127337,
+      "grad_norm": 292.0769958496094,
+      "learning_rate": 9.26361528266439e-06,
+      "loss": 47.8684,
+      "step": 31860
+    },
+    {
+      "epoch": 0.128758832726641,
+      "grad_norm": 441.6422424316406,
+      "learning_rate": 9.262878389096147e-06,
+      "loss": 29.7647,
+      "step": 31870
+    },
+    {
+      "epoch": 0.12879923399200863,
+      "grad_norm": 736.0634765625,
+      "learning_rate": 9.262141156350233e-06,
+      "loss": 37.1215,
+      "step": 31880
+    },
+    {
+      "epoch": 0.12883963525737627,
+      "grad_norm": 552.0074462890625,
+      "learning_rate": 9.261403584485308e-06,
+      "loss": 50.1483,
+      "step": 31890
+    },
+    {
+      "epoch": 0.12888003652274388,
+      "grad_norm": 629.8731689453125,
+      "learning_rate": 9.260665673560058e-06,
+      "loss": 47.3221,
+      "step": 31900
+    },
+    {
+      "epoch": 0.12892043778811152,
+      "grad_norm": 1326.4381103515625,
+      "learning_rate": 9.259927423633193e-06,
+      "loss": 36.9234,
+      "step": 31910
+    },
+    {
+      "epoch": 0.12896083905347916,
+      "grad_norm": 526.2960815429688,
+      "learning_rate": 9.259188834763455e-06,
+      "loss": 43.9766,
+      "step": 31920
+    },
+    {
+      "epoch": 0.1290012403188468,
+      "grad_norm": 1196.8380126953125,
+      "learning_rate": 9.258449907009607e-06,
+      "loss": 65.413,
+      "step": 31930
+    },
+    {
+      "epoch": 0.1290416415842144,
+      "grad_norm": 277.989990234375,
+      "learning_rate": 9.257710640430444e-06,
+      "loss": 36.3897,
+      "step": 31940
+    },
+    {
+      "epoch": 0.12908204284958205,
+      "grad_norm": 544.9930419921875,
+      "learning_rate": 9.256971035084786e-06,
+      "loss": 53.2941,
+      "step": 31950
+    },
+    {
+      "epoch": 0.1291224441149497,
+      "grad_norm": 517.1935424804688,
+      "learning_rate": 9.256231091031477e-06,
+      "loss": 39.4429,
+      "step": 31960
+    },
+    {
+      "epoch": 0.1291628453803173,
+      "grad_norm": 396.2089538574219,
+      "learning_rate": 9.255490808329397e-06,
+      "loss": 43.8766,
+      "step": 31970
+    },
+    {
+      "epoch": 0.12920324664568494,
+      "grad_norm": 330.9097900390625,
+      "learning_rate": 9.254750187037443e-06,
+      "loss": 31.7952,
+      "step": 31980
+    },
+    {
+      "epoch": 0.12924364791105258,
+      "grad_norm": 805.3584594726562,
+      "learning_rate": 9.254009227214543e-06,
+      "loss": 52.6609,
+      "step": 31990
+    },
+    {
+      "epoch": 0.1292840491764202,
+      "grad_norm": 981.2225341796875,
+      "learning_rate": 9.253267928919652e-06,
+      "loss": 40.1239,
+      "step": 32000
+    },
+    {
+      "epoch": 0.12932445044178784,
+      "grad_norm": 765.0771484375,
+      "learning_rate": 9.25252629221175e-06,
+      "loss": 63.7369,
+      "step": 32010
+    },
+    {
+      "epoch": 0.12936485170715548,
+      "grad_norm": 1077.512451171875,
+      "learning_rate": 9.251784317149848e-06,
+      "loss": 62.9928,
+      "step": 32020
+    },
+    {
+      "epoch": 0.1294052529725231,
+      "grad_norm": 1018.3309326171875,
+      "learning_rate": 9.251042003792983e-06,
+      "loss": 51.2373,
+      "step": 32030
+    },
+    {
+      "epoch": 0.12944565423789073,
+      "grad_norm": 554.8124389648438,
+      "learning_rate": 9.250299352200214e-06,
+      "loss": 54.4565,
+      "step": 32040
+    },
+    {
+      "epoch": 0.12948605550325837,
+      "grad_norm": 843.983642578125,
+      "learning_rate": 9.249556362430631e-06,
+      "loss": 41.286,
+      "step": 32050
+    },
+    {
+      "epoch": 0.12952645676862598,
+      "grad_norm": 2317.033447265625,
+      "learning_rate": 9.248813034543353e-06,
+      "loss": 48.9243,
+      "step": 32060
+    },
+    {
+      "epoch": 0.12956685803399362,
+      "grad_norm": 656.9749755859375,
+      "learning_rate": 9.24806936859752e-06,
+      "loss": 43.4692,
+      "step": 32070
+    },
+    {
+      "epoch": 0.12960725929936126,
+      "grad_norm": 491.047119140625,
+      "learning_rate": 9.247325364652304e-06,
+      "loss": 66.3645,
+      "step": 32080
+    },
+    {
+      "epoch": 0.1296476605647289,
+      "grad_norm": 646.3383178710938,
+      "learning_rate": 9.2465810227669e-06,
+      "loss": 52.3016,
+      "step": 32090
+    },
+    {
+      "epoch": 0.1296880618300965,
+      "grad_norm": 765.2969360351562,
+      "learning_rate": 9.245836343000534e-06,
+      "loss": 47.1274,
+      "step": 32100
+    },
+    {
+      "epoch": 0.12972846309546415,
+      "grad_norm": 600.1698608398438,
+      "learning_rate": 9.245091325412456e-06,
+      "loss": 34.7515,
+      "step": 32110
+    },
+    {
+      "epoch": 0.1297688643608318,
+      "grad_norm": 986.0180053710938,
+      "learning_rate": 9.244345970061944e-06,
+      "loss": 56.5286,
+      "step": 32120
+    },
+    {
+      "epoch": 0.1298092656261994,
+      "grad_norm": 521.9662475585938,
+      "learning_rate": 9.243600277008301e-06,
+      "loss": 45.0921,
+      "step": 32130
+    },
+    {
+      "epoch": 0.12984966689156704,
+      "grad_norm": 212.4668731689453,
+      "learning_rate": 9.24285424631086e-06,
+      "loss": 28.9113,
+      "step": 32140
+    },
+    {
+      "epoch": 0.12989006815693468,
+      "grad_norm": 526.3524169921875,
+      "learning_rate": 9.242107878028978e-06,
+      "loss": 39.1954,
+      "step": 32150
+    },
+    {
+      "epoch": 0.1299304694223023,
+      "grad_norm": 538.303955078125,
+      "learning_rate": 9.241361172222043e-06,
+      "loss": 54.2439,
+      "step": 32160
+    },
+    {
+      "epoch": 0.12997087068766994,
+      "grad_norm": 860.3079223632812,
+      "learning_rate": 9.240614128949463e-06,
+      "loss": 64.1403,
+      "step": 32170
+    },
+    {
+      "epoch": 0.13001127195303758,
+      "grad_norm": 777.4006958007812,
+      "learning_rate": 9.239866748270679e-06,
+      "loss": 43.971,
+      "step": 32180
+    },
+    {
+      "epoch": 0.1300516732184052,
+      "grad_norm": 841.7437744140625,
+      "learning_rate": 9.239119030245156e-06,
+      "loss": 59.4323,
+      "step": 32190
+    },
+    {
+      "epoch": 0.13009207448377283,
+      "grad_norm": 711.6234741210938,
+      "learning_rate": 9.238370974932387e-06,
+      "loss": 57.6824,
+      "step": 32200
+    },
+    {
+      "epoch": 0.13013247574914047,
+      "grad_norm": 512.6424560546875,
+      "learning_rate": 9.23762258239189e-06,
+      "loss": 60.2203,
+      "step": 32210
+    },
+    {
+      "epoch": 0.13017287701450808,
+      "grad_norm": 710.1792602539062,
+      "learning_rate": 9.236873852683213e-06,
+      "loss": 59.2537,
+      "step": 32220
+    },
+    {
+      "epoch": 0.13021327827987572,
+      "grad_norm": 533.3241577148438,
+      "learning_rate": 9.23612478586593e-06,
+      "loss": 62.686,
+      "step": 32230
+    },
+    {
+      "epoch": 0.13025367954524336,
+      "grad_norm": 2313.620849609375,
+      "learning_rate": 9.235375381999636e-06,
+      "loss": 56.1435,
+      "step": 32240
+    },
+    {
+      "epoch": 0.130294080810611,
+      "grad_norm": 292.7706298828125,
+      "learning_rate": 9.234625641143962e-06,
+      "loss": 49.0173,
+      "step": 32250
+    },
+    {
+      "epoch": 0.1303344820759786,
+      "grad_norm": 811.357666015625,
+      "learning_rate": 9.233875563358559e-06,
+      "loss": 35.5564,
+      "step": 32260
+    },
+    {
+      "epoch": 0.13037488334134625,
+      "grad_norm": 1006.41796875,
+      "learning_rate": 9.23312514870311e-06,
+      "loss": 35.7152,
+      "step": 32270
+    },
+    {
+      "epoch": 0.1304152846067139,
+      "grad_norm": 709.2618408203125,
+      "learning_rate": 9.232374397237318e-06,
+      "loss": 53.3593,
+      "step": 32280
+    },
+    {
+      "epoch": 0.1304556858720815,
+      "grad_norm": 870.2730712890625,
+      "learning_rate": 9.231623309020922e-06,
+      "loss": 65.6405,
+      "step": 32290
+    },
+    {
+      "epoch": 0.13049608713744915,
+      "grad_norm": 475.84686279296875,
+      "learning_rate": 9.230871884113679e-06,
+      "loss": 32.081,
+      "step": 32300
+    },
+    {
+      "epoch": 0.13053648840281679,
+      "grad_norm": 588.6025390625,
+      "learning_rate": 9.230120122575376e-06,
+      "loss": 57.8686,
+      "step": 32310
+    },
+    {
+      "epoch": 0.1305768896681844,
+      "grad_norm": 481.3664245605469,
+      "learning_rate": 9.22936802446583e-06,
+      "loss": 56.8029,
+      "step": 32320
+    },
+    {
+      "epoch": 0.13061729093355204,
+      "grad_norm": 1488.2110595703125,
+      "learning_rate": 9.228615589844879e-06,
+      "loss": 60.244,
+      "step": 32330
+    },
+    {
+      "epoch": 0.13065769219891968,
+      "grad_norm": 422.4720153808594,
+      "learning_rate": 9.227862818772392e-06,
+      "loss": 35.8967,
+      "step": 32340
+    },
+    {
+      "epoch": 0.1306980934642873,
+      "grad_norm": 587.341064453125,
+      "learning_rate": 9.227109711308265e-06,
+      "loss": 60.7431,
+      "step": 32350
+    },
+    {
+      "epoch": 0.13073849472965493,
+      "grad_norm": 770.119384765625,
+      "learning_rate": 9.226356267512417e-06,
+      "loss": 67.5445,
+      "step": 32360
+    },
+    {
+      "epoch": 0.13077889599502257,
+      "grad_norm": 426.0884704589844,
+      "learning_rate": 9.225602487444799e-06,
+      "loss": 34.8377,
+      "step": 32370
+    },
+    {
+      "epoch": 0.13081929726039018,
+      "grad_norm": 339.0352783203125,
+      "learning_rate": 9.224848371165382e-06,
+      "loss": 45.3635,
+      "step": 32380
+    },
+    {
+      "epoch": 0.13085969852575782,
+      "grad_norm": 0.0,
+      "learning_rate": 9.224093918734172e-06,
+      "loss": 38.837,
+      "step": 32390
+    },
+    {
+      "epoch": 0.13090009979112546,
+      "grad_norm": 488.4760437011719,
+      "learning_rate": 9.223339130211194e-06,
+      "loss": 50.7496,
+      "step": 32400
+    },
+    {
+      "epoch": 0.1309405010564931,
+      "grad_norm": 940.7568969726562,
+      "learning_rate": 9.222584005656501e-06,
+      "loss": 65.9121,
+      "step": 32410
+    },
+    {
+      "epoch": 0.13098090232186071,
+      "grad_norm": 796.3623657226562,
+      "learning_rate": 9.22182854513018e-06,
+      "loss": 47.539,
+      "step": 32420
+    },
+    {
+      "epoch": 0.13102130358722835,
+      "grad_norm": 684.649169921875,
+      "learning_rate": 9.221072748692336e-06,
+      "loss": 42.9674,
+      "step": 32430
+    },
+    {
+      "epoch": 0.131061704852596,
+      "grad_norm": 555.63232421875,
+      "learning_rate": 9.220316616403109e-06,
+      "loss": 41.5548,
+      "step": 32440
+    },
+    {
+      "epoch": 0.1311021061179636,
+      "grad_norm": 801.7620239257812,
+      "learning_rate": 9.219560148322655e-06,
+      "loss": 45.9207,
+      "step": 32450
+    },
+    {
+      "epoch": 0.13114250738333125,
+      "grad_norm": 624.9111938476562,
+      "learning_rate": 9.218803344511165e-06,
+      "loss": 59.801,
+      "step": 32460
+    },
+    {
+      "epoch": 0.1311829086486989,
+      "grad_norm": 673.0691528320312,
+      "learning_rate": 9.218046205028854e-06,
+      "loss": 57.1818,
+      "step": 32470
+    },
+    {
+      "epoch": 0.1312233099140665,
+      "grad_norm": 817.8880615234375,
+      "learning_rate": 9.217288729935966e-06,
+      "loss": 40.0779,
+      "step": 32480
+    },
+    {
+      "epoch": 0.13126371117943414,
+      "grad_norm": 759.4141845703125,
+      "learning_rate": 9.216530919292768e-06,
+      "loss": 50.3778,
+      "step": 32490
+    },
+    {
+      "epoch": 0.13130411244480178,
+      "grad_norm": 936.8798217773438,
+      "learning_rate": 9.215772773159556e-06,
+      "loss": 69.4996,
+      "step": 32500
+    },
+    {
+      "epoch": 0.1313445137101694,
+      "grad_norm": 995.8248901367188,
+      "learning_rate": 9.215014291596653e-06,
+      "loss": 55.1168,
+      "step": 32510
+    },
+    {
+      "epoch": 0.13138491497553703,
+      "grad_norm": 858.1106567382812,
+      "learning_rate": 9.214255474664405e-06,
+      "loss": 45.871,
+      "step": 32520
+    },
+    {
+      "epoch": 0.13142531624090467,
+      "grad_norm": 720.53955078125,
+      "learning_rate": 9.213496322423193e-06,
+      "loss": 63.9526,
+      "step": 32530
+    },
+    {
+      "epoch": 0.13146571750627228,
+      "grad_norm": 2278.630615234375,
+      "learning_rate": 9.212736834933413e-06,
+      "loss": 76.4255,
+      "step": 32540
+    },
+    {
+      "epoch": 0.13150611877163992,
+      "grad_norm": 560.2059936523438,
+      "learning_rate": 9.211977012255497e-06,
+      "loss": 43.5413,
+      "step": 32550
+    },
+    {
+      "epoch": 0.13154652003700756,
+      "grad_norm": 1519.486572265625,
+      "learning_rate": 9.211216854449903e-06,
+      "loss": 77.558,
+      "step": 32560
+    },
+    {
+      "epoch": 0.1315869213023752,
+      "grad_norm": 337.37933349609375,
+      "learning_rate": 9.210456361577109e-06,
+      "loss": 51.9883,
+      "step": 32570
+    },
+    {
+      "epoch": 0.13162732256774282,
+      "grad_norm": 601.6719360351562,
+      "learning_rate": 9.209695533697624e-06,
+      "loss": 40.5055,
+      "step": 32580
+    },
+    {
+      "epoch": 0.13166772383311046,
+      "grad_norm": 541.4317626953125,
+      "learning_rate": 9.208934370871989e-06,
+      "loss": 50.7977,
+      "step": 32590
+    },
+    {
+      "epoch": 0.1317081250984781,
+      "grad_norm": 1253.9888916015625,
+      "learning_rate": 9.20817287316076e-06,
+      "loss": 45.5538,
+      "step": 32600
+    },
+    {
+      "epoch": 0.1317485263638457,
+      "grad_norm": 489.9399108886719,
+      "learning_rate": 9.20741104062453e-06,
+      "loss": 53.2811,
+      "step": 32610
+    },
+    {
+      "epoch": 0.13178892762921335,
+      "grad_norm": 561.935791015625,
+      "learning_rate": 9.206648873323912e-06,
+      "loss": 49.4477,
+      "step": 32620
+    },
+    {
+      "epoch": 0.131829328894581,
+      "grad_norm": 408.33892822265625,
+      "learning_rate": 9.205886371319548e-06,
+      "loss": 50.0402,
+      "step": 32630
+    },
+    {
+      "epoch": 0.1318697301599486,
+      "grad_norm": 391.8433532714844,
+      "learning_rate": 9.20512353467211e-06,
+      "loss": 48.5688,
+      "step": 32640
+    },
+    {
+      "epoch": 0.13191013142531624,
+      "grad_norm": 710.4462890625,
+      "learning_rate": 9.204360363442288e-06,
+      "loss": 56.6522,
+      "step": 32650
+    },
+    {
+      "epoch": 0.13195053269068388,
+      "grad_norm": 644.66943359375,
+      "learning_rate": 9.20359685769081e-06,
+      "loss": 57.3297,
+      "step": 32660
+    },
+    {
+      "epoch": 0.1319909339560515,
+      "grad_norm": 622.2406616210938,
+      "learning_rate": 9.202833017478421e-06,
+      "loss": 36.6121,
+      "step": 32670
+    },
+    {
+      "epoch": 0.13203133522141913,
+      "grad_norm": 559.0179443359375,
+      "learning_rate": 9.2020688428659e-06,
+      "loss": 43.6517,
+      "step": 32680
+    },
+    {
+      "epoch": 0.13207173648678677,
+      "grad_norm": 543.2446899414062,
+      "learning_rate": 9.201304333914042e-06,
+      "loss": 45.9316,
+      "step": 32690
+    },
+    {
+      "epoch": 0.13211213775215438,
+      "grad_norm": 493.87835693359375,
+      "learning_rate": 9.200539490683682e-06,
+      "loss": 48.7048,
+      "step": 32700
+    },
+    {
+      "epoch": 0.13215253901752202,
+      "grad_norm": 705.3010864257812,
+      "learning_rate": 9.19977431323567e-06,
+      "loss": 43.7614,
+      "step": 32710
+    },
+    {
+      "epoch": 0.13219294028288966,
+      "grad_norm": 530.714111328125,
+      "learning_rate": 9.199008801630893e-06,
+      "loss": 58.9209,
+      "step": 32720
+    },
+    {
+      "epoch": 0.1322333415482573,
+      "grad_norm": 529.1024780273438,
+      "learning_rate": 9.198242955930257e-06,
+      "loss": 51.1792,
+      "step": 32730
+    },
+    {
+      "epoch": 0.13227374281362492,
+      "grad_norm": 714.8733520507812,
+      "learning_rate": 9.197476776194693e-06,
+      "loss": 38.4417,
+      "step": 32740
+    },
+    {
+      "epoch": 0.13231414407899256,
+      "grad_norm": 903.4959106445312,
+      "learning_rate": 9.196710262485168e-06,
+      "loss": 58.4088,
+      "step": 32750
+    },
+    {
+      "epoch": 0.1323545453443602,
+      "grad_norm": 694.1265869140625,
+      "learning_rate": 9.195943414862667e-06,
+      "loss": 62.1965,
+      "step": 32760
+    },
+    {
+      "epoch": 0.1323949466097278,
+      "grad_norm": 653.4110107421875,
+      "learning_rate": 9.195176233388206e-06,
+      "loss": 50.7297,
+      "step": 32770
+    },
+    {
+      "epoch": 0.13243534787509545,
+      "grad_norm": 956.4697265625,
+      "learning_rate": 9.194408718122825e-06,
+      "loss": 36.8561,
+      "step": 32780
+    },
+    {
+      "epoch": 0.1324757491404631,
+      "grad_norm": 761.3573608398438,
+      "learning_rate": 9.193640869127592e-06,
+      "loss": 66.8244,
+      "step": 32790
+    },
+    {
+      "epoch": 0.1325161504058307,
+      "grad_norm": 444.17547607421875,
+      "learning_rate": 9.192872686463601e-06,
+      "loss": 44.2107,
+      "step": 32800
+    },
+    {
+      "epoch": 0.13255655167119834,
+      "grad_norm": 1537.195556640625,
+      "learning_rate": 9.192104170191973e-06,
+      "loss": 69.7521,
+      "step": 32810
+    },
+    {
+      "epoch": 0.13259695293656598,
+      "grad_norm": 513.7176513671875,
+      "learning_rate": 9.191335320373856e-06,
+      "loss": 45.5239,
+      "step": 32820
+    },
+    {
+      "epoch": 0.1326373542019336,
+      "grad_norm": 545.9681396484375,
+      "learning_rate": 9.190566137070422e-06,
+      "loss": 50.4583,
+      "step": 32830
+    },
+    {
+      "epoch": 0.13267775546730123,
+      "grad_norm": 1057.2578125,
+      "learning_rate": 9.189796620342875e-06,
+      "loss": 64.9317,
+      "step": 32840
+    },
+    {
+      "epoch": 0.13271815673266887,
+      "grad_norm": 528.858642578125,
+      "learning_rate": 9.189026770252437e-06,
+      "loss": 65.4341,
+      "step": 32850
+    },
+    {
+      "epoch": 0.13275855799803649,
+      "grad_norm": 465.4150695800781,
+      "learning_rate": 9.188256586860365e-06,
+      "loss": 63.703,
+      "step": 32860
+    },
+    {
+      "epoch": 0.13279895926340413,
+      "grad_norm": 471.5633850097656,
+      "learning_rate": 9.187486070227938e-06,
+      "loss": 44.7681,
+      "step": 32870
+    },
+    {
+      "epoch": 0.13283936052877177,
+      "grad_norm": 1108.5830078125,
+      "learning_rate": 9.186715220416463e-06,
+      "loss": 55.0163,
+      "step": 32880
+    },
+    {
+      "epoch": 0.1328797617941394,
+      "grad_norm": 829.570068359375,
+      "learning_rate": 9.185944037487271e-06,
+      "loss": 71.4634,
+      "step": 32890
+    },
+    {
+      "epoch": 0.13292016305950702,
+      "grad_norm": 404.9236755371094,
+      "learning_rate": 9.185172521501723e-06,
+      "loss": 39.8276,
+      "step": 32900
+    },
+    {
+      "epoch": 0.13296056432487466,
+      "grad_norm": 623.8037109375,
+      "learning_rate": 9.184400672521204e-06,
+      "loss": 44.3128,
+      "step": 32910
+    },
+    {
+      "epoch": 0.1330009655902423,
+      "grad_norm": 622.3230590820312,
+      "learning_rate": 9.183628490607129e-06,
+      "loss": 38.3063,
+      "step": 32920
+    },
+    {
+      "epoch": 0.1330413668556099,
+      "grad_norm": 629.4878540039062,
+      "learning_rate": 9.182855975820934e-06,
+      "loss": 61.7241,
+      "step": 32930
+    },
+    {
+      "epoch": 0.13308176812097755,
+      "grad_norm": 456.10931396484375,
+      "learning_rate": 9.182083128224086e-06,
+      "loss": 64.5606,
+      "step": 32940
+    },
+    {
+      "epoch": 0.1331221693863452,
+      "grad_norm": 600.0465087890625,
+      "learning_rate": 9.181309947878077e-06,
+      "loss": 50.6354,
+      "step": 32950
+    },
+    {
+      "epoch": 0.1331625706517128,
+      "grad_norm": 352.6278381347656,
+      "learning_rate": 9.180536434844426e-06,
+      "loss": 49.3764,
+      "step": 32960
+    },
+    {
+      "epoch": 0.13320297191708044,
+      "grad_norm": 474.4338073730469,
+      "learning_rate": 9.179762589184676e-06,
+      "loss": 47.952,
+      "step": 32970
+    },
+    {
+      "epoch": 0.13324337318244808,
+      "grad_norm": 722.0818481445312,
+      "learning_rate": 9.1789884109604e-06,
+      "loss": 46.6468,
+      "step": 32980
+    },
+    {
+      "epoch": 0.1332837744478157,
+      "grad_norm": 687.3217163085938,
+      "learning_rate": 9.178213900233193e-06,
+      "loss": 44.9744,
+      "step": 32990
+    },
+    {
+      "epoch": 0.13332417571318333,
+      "grad_norm": 494.89044189453125,
+      "learning_rate": 9.177439057064684e-06,
+      "loss": 53.1962,
+      "step": 33000
+    },
+    {
+      "epoch": 0.13336457697855097,
+      "grad_norm": 262.19580078125,
+      "learning_rate": 9.17666388151652e-06,
+      "loss": 42.9505,
+      "step": 33010
+    },
+    {
+      "epoch": 0.1334049782439186,
+      "grad_norm": 502.8430480957031,
+      "learning_rate": 9.175888373650377e-06,
+      "loss": 51.0244,
+      "step": 33020
+    },
+    {
+      "epoch": 0.13344537950928623,
+      "grad_norm": 749.8898315429688,
+      "learning_rate": 9.175112533527963e-06,
+      "loss": 40.869,
+      "step": 33030
+    },
+    {
+      "epoch": 0.13348578077465387,
+      "grad_norm": 364.7505187988281,
+      "learning_rate": 9.174336361211007e-06,
+      "loss": 63.2581,
+      "step": 33040
+    },
+    {
+      "epoch": 0.1335261820400215,
+      "grad_norm": 688.9202880859375,
+      "learning_rate": 9.173559856761262e-06,
+      "loss": 39.9341,
+      "step": 33050
+    },
+    {
+      "epoch": 0.13356658330538912,
+      "grad_norm": 1022.2750854492188,
+      "learning_rate": 9.172783020240514e-06,
+      "loss": 59.776,
+      "step": 33060
+    },
+    {
+      "epoch": 0.13360698457075676,
+      "grad_norm": 461.04437255859375,
+      "learning_rate": 9.172005851710573e-06,
+      "loss": 40.5875,
+      "step": 33070
+    },
+    {
+      "epoch": 0.1336473858361244,
+      "grad_norm": 552.5075073242188,
+      "learning_rate": 9.171228351233272e-06,
+      "loss": 63.339,
+      "step": 33080
+    },
+    {
+      "epoch": 0.133687787101492,
+      "grad_norm": 204.60736083984375,
+      "learning_rate": 9.170450518870475e-06,
+      "loss": 53.6323,
+      "step": 33090
+    },
+    {
+      "epoch": 0.13372818836685965,
+      "grad_norm": 597.3870239257812,
+      "learning_rate": 9.169672354684069e-06,
+      "loss": 58.8157,
+      "step": 33100
+    },
+    {
+      "epoch": 0.1337685896322273,
+      "grad_norm": 626.3027954101562,
+      "learning_rate": 9.168893858735972e-06,
+      "loss": 53.9851,
+      "step": 33110
+    },
+    {
+      "epoch": 0.1338089908975949,
+      "grad_norm": 631.9942626953125,
+      "learning_rate": 9.168115031088122e-06,
+      "loss": 52.9761,
+      "step": 33120
+    },
+    {
+      "epoch": 0.13384939216296254,
+      "grad_norm": 863.2987670898438,
+      "learning_rate": 9.167335871802488e-06,
+      "loss": 47.3985,
+      "step": 33130
+    },
+    {
+      "epoch": 0.13388979342833018,
+      "grad_norm": 734.11181640625,
+      "learning_rate": 9.166556380941063e-06,
+      "loss": 53.5019,
+      "step": 33140
+    },
+    {
+      "epoch": 0.1339301946936978,
+      "grad_norm": 669.5849609375,
+      "learning_rate": 9.16577655856587e-06,
+      "loss": 42.8189,
+      "step": 33150
+    },
+    {
+      "epoch": 0.13397059595906544,
+      "grad_norm": 447.07562255859375,
+      "learning_rate": 9.164996404738955e-06,
+      "loss": 31.3628,
+      "step": 33160
+    },
+    {
+      "epoch": 0.13401099722443308,
+      "grad_norm": 822.84716796875,
+      "learning_rate": 9.16421591952239e-06,
+      "loss": 49.1292,
+      "step": 33170
+    },
+    {
+      "epoch": 0.1340513984898007,
+      "grad_norm": 489.0323486328125,
+      "learning_rate": 9.163435102978276e-06,
+      "loss": 40.1574,
+      "step": 33180
+    },
+    {
+      "epoch": 0.13409179975516833,
+      "grad_norm": 808.0147094726562,
+      "learning_rate": 9.162653955168739e-06,
+      "loss": 48.2493,
+      "step": 33190
+    },
+    {
+      "epoch": 0.13413220102053597,
+      "grad_norm": 679.6731567382812,
+      "learning_rate": 9.161872476155929e-06,
+      "loss": 44.6656,
+      "step": 33200
+    },
+    {
+      "epoch": 0.1341726022859036,
+      "grad_norm": 703.5978393554688,
+      "learning_rate": 9.161090666002029e-06,
+      "loss": 33.7541,
+      "step": 33210
+    },
+    {
+      "epoch": 0.13421300355127122,
+      "grad_norm": 405.0184020996094,
+      "learning_rate": 9.16030852476924e-06,
+      "loss": 35.5508,
+      "step": 33220
+    },
+    {
+      "epoch": 0.13425340481663886,
+      "grad_norm": 326.9285583496094,
+      "learning_rate": 9.159526052519794e-06,
+      "loss": 48.6239,
+      "step": 33230
+    },
+    {
+      "epoch": 0.1342938060820065,
+      "grad_norm": 1562.007080078125,
+      "learning_rate": 9.15874324931595e-06,
+      "loss": 53.0464,
+      "step": 33240
+    },
+    {
+      "epoch": 0.1343342073473741,
+      "grad_norm": 1214.361328125,
+      "learning_rate": 9.157960115219993e-06,
+      "loss": 54.1673,
+      "step": 33250
+    },
+    {
+      "epoch": 0.13437460861274175,
+      "grad_norm": 589.89013671875,
+      "learning_rate": 9.157176650294231e-06,
+      "loss": 39.7849,
+      "step": 33260
+    },
+    {
+      "epoch": 0.1344150098781094,
+      "grad_norm": 561.1547241210938,
+      "learning_rate": 9.156392854601001e-06,
+      "loss": 60.4532,
+      "step": 33270
+    },
+    {
+      "epoch": 0.134455411143477,
+      "grad_norm": 407.2488708496094,
+      "learning_rate": 9.155608728202669e-06,
+      "loss": 46.0791,
+      "step": 33280
+    },
+    {
+      "epoch": 0.13449581240884464,
+      "grad_norm": 546.0625,
+      "learning_rate": 9.154824271161621e-06,
+      "loss": 47.7987,
+      "step": 33290
+    },
+    {
+      "epoch": 0.13453621367421228,
+      "grad_norm": 709.1144409179688,
+      "learning_rate": 9.154039483540273e-06,
+      "loss": 62.7406,
+      "step": 33300
+    },
+    {
+      "epoch": 0.1345766149395799,
+      "grad_norm": 706.6863403320312,
+      "learning_rate": 9.153254365401069e-06,
+      "loss": 36.7554,
+      "step": 33310
+    },
+    {
+      "epoch": 0.13461701620494754,
+      "grad_norm": 619.6150512695312,
+      "learning_rate": 9.152468916806477e-06,
+      "loss": 60.1805,
+      "step": 33320
+    },
+    {
+      "epoch": 0.13465741747031518,
+      "grad_norm": 244.01255798339844,
+      "learning_rate": 9.151683137818989e-06,
+      "loss": 51.3976,
+      "step": 33330
+    },
+    {
+      "epoch": 0.1346978187356828,
+      "grad_norm": 3886.03857421875,
+      "learning_rate": 9.150897028501126e-06,
+      "loss": 81.1586,
+      "step": 33340
+    },
+    {
+      "epoch": 0.13473822000105043,
+      "grad_norm": 556.2538452148438,
+      "learning_rate": 9.15011058891544e-06,
+      "loss": 50.1933,
+      "step": 33350
+    },
+    {
+      "epoch": 0.13477862126641807,
+      "grad_norm": 807.3983764648438,
+      "learning_rate": 9.149323819124498e-06,
+      "loss": 41.1168,
+      "step": 33360
+    },
+    {
+      "epoch": 0.1348190225317857,
+      "grad_norm": 325.5909423828125,
+      "learning_rate": 9.148536719190904e-06,
+      "loss": 41.2033,
+      "step": 33370
+    },
+    {
+      "epoch": 0.13485942379715332,
+      "grad_norm": 613.8952026367188,
+      "learning_rate": 9.147749289177282e-06,
+      "loss": 41.9996,
+      "step": 33380
+    },
+    {
+      "epoch": 0.13489982506252096,
+      "grad_norm": 361.23492431640625,
+      "learning_rate": 9.146961529146285e-06,
+      "loss": 42.0668,
+      "step": 33390
+    },
+    {
+      "epoch": 0.1349402263278886,
+      "grad_norm": 648.5516357421875,
+      "learning_rate": 9.146173439160591e-06,
+      "loss": 41.9859,
+      "step": 33400
+    },
+    {
+      "epoch": 0.1349806275932562,
+      "grad_norm": 296.4998474121094,
+      "learning_rate": 9.145385019282904e-06,
+      "loss": 39.537,
+      "step": 33410
+    },
+    {
+      "epoch": 0.13502102885862385,
+      "grad_norm": 1153.4014892578125,
+      "learning_rate": 9.144596269575957e-06,
+      "loss": 61.5977,
+      "step": 33420
+    },
+    {
+      "epoch": 0.1350614301239915,
+      "grad_norm": 532.4719848632812,
+      "learning_rate": 9.143807190102504e-06,
+      "loss": 58.7296,
+      "step": 33430
+    },
+    {
+      "epoch": 0.1351018313893591,
+      "grad_norm": 764.9578247070312,
+      "learning_rate": 9.143017780925331e-06,
+      "loss": 52.383,
+      "step": 33440
+    },
+    {
+      "epoch": 0.13514223265472675,
+      "grad_norm": 722.291748046875,
+      "learning_rate": 9.142228042107248e-06,
+      "loss": 49.525,
+      "step": 33450
+    },
+    {
+      "epoch": 0.13518263392009439,
+      "grad_norm": 0.0,
+      "learning_rate": 9.141437973711092e-06,
+      "loss": 36.1941,
+      "step": 33460
+    },
+    {
+      "epoch": 0.135223035185462,
+      "grad_norm": 920.4172973632812,
+      "learning_rate": 9.14064757579972e-06,
+      "loss": 45.6598,
+      "step": 33470
+    },
+    {
+      "epoch": 0.13526343645082964,
+      "grad_norm": 359.2383117675781,
+      "learning_rate": 9.139856848436023e-06,
+      "loss": 37.3868,
+      "step": 33480
+    },
+    {
+      "epoch": 0.13530383771619728,
+      "grad_norm": 668.7049560546875,
+      "learning_rate": 9.139065791682916e-06,
+      "loss": 39.9217,
+      "step": 33490
+    },
+    {
+      "epoch": 0.1353442389815649,
+      "grad_norm": 674.8594360351562,
+      "learning_rate": 9.138274405603342e-06,
+      "loss": 33.1557,
+      "step": 33500
+    },
+    {
+      "epoch": 0.13538464024693253,
+      "grad_norm": 396.17352294921875,
+      "learning_rate": 9.137482690260265e-06,
+      "loss": 64.2327,
+      "step": 33510
+    },
+    {
+      "epoch": 0.13542504151230017,
+      "grad_norm": 560.4313354492188,
+      "learning_rate": 9.13669064571668e-06,
+      "loss": 71.6313,
+      "step": 33520
+    },
+    {
+      "epoch": 0.13546544277766778,
+      "grad_norm": 537.1707153320312,
+      "learning_rate": 9.135898272035601e-06,
+      "loss": 67.0521,
+      "step": 33530
+    },
+    {
+      "epoch": 0.13550584404303542,
+      "grad_norm": 557.8136596679688,
+      "learning_rate": 9.13510556928008e-06,
+      "loss": 37.5089,
+      "step": 33540
+    },
+    {
+      "epoch": 0.13554624530840306,
+      "grad_norm": 407.9730529785156,
+      "learning_rate": 9.134312537513188e-06,
+      "loss": 34.4554,
+      "step": 33550
+    },
+    {
+      "epoch": 0.1355866465737707,
+      "grad_norm": 1424.865478515625,
+      "learning_rate": 9.133519176798021e-06,
+      "loss": 39.1332,
+      "step": 33560
+    },
+    {
+      "epoch": 0.13562704783913831,
+      "grad_norm": 669.2245483398438,
+      "learning_rate": 9.132725487197701e-06,
+      "loss": 43.7718,
+      "step": 33570
+    },
+    {
+      "epoch": 0.13566744910450595,
+      "grad_norm": 712.7960815429688,
+      "learning_rate": 9.131931468775382e-06,
+      "loss": 54.1853,
+      "step": 33580
+    },
+    {
+      "epoch": 0.1357078503698736,
+      "grad_norm": 473.095458984375,
+      "learning_rate": 9.131137121594239e-06,
+      "loss": 52.0267,
+      "step": 33590
+    },
+    {
+      "epoch": 0.1357482516352412,
+      "grad_norm": 494.1900329589844,
+      "learning_rate": 9.130342445717474e-06,
+      "loss": 49.4793,
+      "step": 33600
+    },
+    {
+      "epoch": 0.13578865290060885,
+      "grad_norm": 597.4051513671875,
+      "learning_rate": 9.129547441208317e-06,
+      "loss": 67.9233,
+      "step": 33610
+    },
+    {
+      "epoch": 0.1358290541659765,
+      "grad_norm": 254.22824096679688,
+      "learning_rate": 9.128752108130022e-06,
+      "loss": 62.3962,
+      "step": 33620
+    },
+    {
+      "epoch": 0.1358694554313441,
+      "grad_norm": 312.43621826171875,
+      "learning_rate": 9.12795644654587e-06,
+      "loss": 43.4682,
+      "step": 33630
+    },
+    {
+      "epoch": 0.13590985669671174,
+      "grad_norm": 676.4065551757812,
+      "learning_rate": 9.127160456519168e-06,
+      "loss": 45.085,
+      "step": 33640
+    },
+    {
+      "epoch": 0.13595025796207938,
+      "grad_norm": 594.743408203125,
+      "learning_rate": 9.126364138113251e-06,
+      "loss": 106.4617,
+      "step": 33650
+    },
+    {
+      "epoch": 0.135990659227447,
+      "grad_norm": 800.3324584960938,
+      "learning_rate": 9.125567491391476e-06,
+      "loss": 49.7073,
+      "step": 33660
+    },
+    {
+      "epoch": 0.13603106049281463,
+      "grad_norm": 819.260498046875,
+      "learning_rate": 9.12477051641723e-06,
+      "loss": 53.5647,
+      "step": 33670
+    },
+    {
+      "epoch": 0.13607146175818227,
+      "grad_norm": 1157.8453369140625,
+      "learning_rate": 9.123973213253923e-06,
+      "loss": 52.8083,
+      "step": 33680
+    },
+    {
+      "epoch": 0.13611186302354988,
+      "grad_norm": 471.611572265625,
+      "learning_rate": 9.123175581964995e-06,
+      "loss": 50.6554,
+      "step": 33690
+    },
+    {
+      "epoch": 0.13615226428891752,
+      "grad_norm": 428.2491149902344,
+      "learning_rate": 9.122377622613909e-06,
+      "loss": 39.169,
+      "step": 33700
+    },
+    {
+      "epoch": 0.13619266555428516,
+      "grad_norm": 852.0259399414062,
+      "learning_rate": 9.121579335264155e-06,
+      "loss": 35.4563,
+      "step": 33710
+    },
+    {
+      "epoch": 0.1362330668196528,
+      "grad_norm": 266.2728271484375,
+      "learning_rate": 9.120780719979248e-06,
+      "loss": 63.9524,
+      "step": 33720
+    },
+    {
+      "epoch": 0.13627346808502042,
+      "grad_norm": 562.6665649414062,
+      "learning_rate": 9.11998177682273e-06,
+      "loss": 79.5502,
+      "step": 33730
+    },
+    {
+      "epoch": 0.13631386935038806,
+      "grad_norm": 866.7197875976562,
+      "learning_rate": 9.11918250585817e-06,
+      "loss": 48.4737,
+      "step": 33740
+    },
+    {
+      "epoch": 0.1363542706157557,
+      "grad_norm": 410.7391662597656,
+      "learning_rate": 9.118382907149164e-06,
+      "loss": 38.7582,
+      "step": 33750
+    },
+    {
+      "epoch": 0.1363946718811233,
+      "grad_norm": 1614.3094482421875,
+      "learning_rate": 9.117582980759332e-06,
+      "loss": 59.5404,
+      "step": 33760
+    },
+    {
+      "epoch": 0.13643507314649095,
+      "grad_norm": 583.3519897460938,
+      "learning_rate": 9.116782726752317e-06,
+      "loss": 49.7827,
+      "step": 33770
+    },
+    {
+      "epoch": 0.1364754744118586,
+      "grad_norm": 873.2832641601562,
+      "learning_rate": 9.115982145191796e-06,
+      "loss": 51.7136,
+      "step": 33780
+    },
+    {
+      "epoch": 0.1365158756772262,
+      "grad_norm": 726.3242797851562,
+      "learning_rate": 9.115181236141463e-06,
+      "loss": 35.4949,
+      "step": 33790
+    },
+    {
+      "epoch": 0.13655627694259384,
+      "grad_norm": 571.6852416992188,
+      "learning_rate": 9.114379999665047e-06,
+      "loss": 51.8902,
+      "step": 33800
+    },
+    {
+      "epoch": 0.13659667820796148,
+      "grad_norm": 685.9703369140625,
+      "learning_rate": 9.113578435826295e-06,
+      "loss": 78.0631,
+      "step": 33810
+    },
+    {
+      "epoch": 0.1366370794733291,
+      "grad_norm": 728.8062744140625,
+      "learning_rate": 9.112776544688988e-06,
+      "loss": 41.2744,
+      "step": 33820
+    },
+    {
+      "epoch": 0.13667748073869673,
+      "grad_norm": 941.1890869140625,
+      "learning_rate": 9.111974326316926e-06,
+      "loss": 61.4087,
+      "step": 33830
+    },
+    {
+      "epoch": 0.13671788200406437,
+      "grad_norm": 472.9835205078125,
+      "learning_rate": 9.111171780773938e-06,
+      "loss": 57.9784,
+      "step": 33840
+    },
+    {
+      "epoch": 0.13675828326943198,
+      "grad_norm": 248.1824188232422,
+      "learning_rate": 9.110368908123878e-06,
+      "loss": 48.1912,
+      "step": 33850
+    },
+    {
+      "epoch": 0.13679868453479962,
+      "grad_norm": 577.3104858398438,
+      "learning_rate": 9.10956570843063e-06,
+      "loss": 45.3517,
+      "step": 33860
+    },
+    {
+      "epoch": 0.13683908580016726,
+      "grad_norm": 400.1105041503906,
+      "learning_rate": 9.108762181758096e-06,
+      "loss": 35.0138,
+      "step": 33870
+    },
+    {
+      "epoch": 0.1368794870655349,
+      "grad_norm": 369.667236328125,
+      "learning_rate": 9.107958328170215e-06,
+      "loss": 57.6431,
+      "step": 33880
+    },
+    {
+      "epoch": 0.13691988833090252,
+      "grad_norm": 403.0173645019531,
+      "learning_rate": 9.10715414773094e-06,
+      "loss": 48.3835,
+      "step": 33890
+    },
+    {
+      "epoch": 0.13696028959627016,
+      "grad_norm": 383.2605895996094,
+      "learning_rate": 9.10634964050426e-06,
+      "loss": 43.6177,
+      "step": 33900
+    },
+    {
+      "epoch": 0.1370006908616378,
+      "grad_norm": 677.1846313476562,
+      "learning_rate": 9.105544806554184e-06,
+      "loss": 57.0482,
+      "step": 33910
+    },
+    {
+      "epoch": 0.1370410921270054,
+      "grad_norm": 1978.128662109375,
+      "learning_rate": 9.104739645944752e-06,
+      "loss": 59.6453,
+      "step": 33920
+    },
+    {
+      "epoch": 0.13708149339237305,
+      "grad_norm": 668.5293579101562,
+      "learning_rate": 9.103934158740023e-06,
+      "loss": 37.7574,
+      "step": 33930
+    },
+    {
+      "epoch": 0.1371218946577407,
+      "grad_norm": 763.6205444335938,
+      "learning_rate": 9.10312834500409e-06,
+      "loss": 61.1796,
+      "step": 33940
+    },
+    {
+      "epoch": 0.1371622959231083,
+      "grad_norm": 504.1656494140625,
+      "learning_rate": 9.102322204801062e-06,
+      "loss": 50.8536,
+      "step": 33950
+    },
+    {
+      "epoch": 0.13720269718847594,
+      "grad_norm": 1202.7027587890625,
+      "learning_rate": 9.101515738195084e-06,
+      "loss": 70.192,
+      "step": 33960
+    },
+    {
+      "epoch": 0.13724309845384358,
+      "grad_norm": 633.1737060546875,
+      "learning_rate": 9.100708945250322e-06,
+      "loss": 62.8866,
+      "step": 33970
+    },
+    {
+      "epoch": 0.1372834997192112,
+      "grad_norm": 1090.31103515625,
+      "learning_rate": 9.099901826030969e-06,
+      "loss": 55.2263,
+      "step": 33980
+    },
+    {
+      "epoch": 0.13732390098457883,
+      "grad_norm": 678.004150390625,
+      "learning_rate": 9.099094380601244e-06,
+      "loss": 67.2974,
+      "step": 33990
+    },
+    {
+      "epoch": 0.13736430224994647,
+      "grad_norm": 1015.7597045898438,
+      "learning_rate": 9.098286609025392e-06,
+      "loss": 61.169,
+      "step": 34000
+    },
+    {
+      "epoch": 0.13740470351531409,
+      "grad_norm": 460.1115417480469,
+      "learning_rate": 9.097478511367682e-06,
+      "loss": 41.7142,
+      "step": 34010
+    },
+    {
+      "epoch": 0.13744510478068173,
+      "grad_norm": 558.1769409179688,
+      "learning_rate": 9.096670087692413e-06,
+      "loss": 62.2378,
+      "step": 34020
+    },
+    {
+      "epoch": 0.13748550604604937,
+      "grad_norm": 420.15423583984375,
+      "learning_rate": 9.095861338063906e-06,
+      "loss": 33.5995,
+      "step": 34030
+    },
+    {
+      "epoch": 0.137525907311417,
+      "grad_norm": 855.2376708984375,
+      "learning_rate": 9.09505226254651e-06,
+      "loss": 41.2862,
+      "step": 34040
+    },
+    {
+      "epoch": 0.13756630857678462,
+      "grad_norm": 668.4373168945312,
+      "learning_rate": 9.094242861204598e-06,
+      "loss": 51.963,
+      "step": 34050
+    },
+    {
+      "epoch": 0.13760670984215226,
+      "grad_norm": 1038.2156982421875,
+      "learning_rate": 9.093433134102572e-06,
+      "loss": 49.4119,
+      "step": 34060
+    },
+    {
+      "epoch": 0.1376471111075199,
+      "grad_norm": 781.8707275390625,
+      "learning_rate": 9.09262308130486e-06,
+      "loss": 46.6765,
+      "step": 34070
+    },
+    {
+      "epoch": 0.1376875123728875,
+      "grad_norm": 443.0128173828125,
+      "learning_rate": 9.091812702875908e-06,
+      "loss": 45.1318,
+      "step": 34080
+    },
+    {
+      "epoch": 0.13772791363825515,
+      "grad_norm": 2211.4814453125,
+      "learning_rate": 9.0910019988802e-06,
+      "loss": 53.6052,
+      "step": 34090
+    },
+    {
+      "epoch": 0.1377683149036228,
+      "grad_norm": 769.3303833007812,
+      "learning_rate": 9.09019096938224e-06,
+      "loss": 41.2156,
+      "step": 34100
+    },
+    {
+      "epoch": 0.1378087161689904,
+      "grad_norm": 540.2257080078125,
+      "learning_rate": 9.089379614446554e-06,
+      "loss": 53.6039,
+      "step": 34110
+    },
+    {
+      "epoch": 0.13784911743435804,
+      "grad_norm": 684.0350341796875,
+      "learning_rate": 9.0885679341377e-06,
+      "loss": 43.7098,
+      "step": 34120
+    },
+    {
+      "epoch": 0.13788951869972568,
+      "grad_norm": 524.279296875,
+      "learning_rate": 9.08775592852026e-06,
+      "loss": 38.2165,
+      "step": 34130
+    },
+    {
+      "epoch": 0.1379299199650933,
+      "grad_norm": 805.431396484375,
+      "learning_rate": 9.08694359765884e-06,
+      "loss": 83.9917,
+      "step": 34140
+    },
+    {
+      "epoch": 0.13797032123046093,
+      "grad_norm": 611.6764526367188,
+      "learning_rate": 9.086130941618075e-06,
+      "loss": 41.6229,
+      "step": 34150
+    },
+    {
+      "epoch": 0.13801072249582857,
+      "grad_norm": 606.640869140625,
+      "learning_rate": 9.085317960462625e-06,
+      "loss": 59.065,
+      "step": 34160
+    },
+    {
+      "epoch": 0.1380511237611962,
+      "grad_norm": 598.1301879882812,
+      "learning_rate": 9.084504654257173e-06,
+      "loss": 49.4473,
+      "step": 34170
+    },
+    {
+      "epoch": 0.13809152502656383,
+      "grad_norm": 430.78424072265625,
+      "learning_rate": 9.08369102306643e-06,
+      "loss": 50.9787,
+      "step": 34180
+    },
+    {
+      "epoch": 0.13813192629193147,
+      "grad_norm": 1345.611572265625,
+      "learning_rate": 9.082877066955135e-06,
+      "loss": 65.4843,
+      "step": 34190
+    },
+    {
+      "epoch": 0.1381723275572991,
+      "grad_norm": 874.1762084960938,
+      "learning_rate": 9.08206278598805e-06,
+      "loss": 46.0913,
+      "step": 34200
+    },
+    {
+      "epoch": 0.13821272882266672,
+      "grad_norm": 1135.8431396484375,
+      "learning_rate": 9.081248180229963e-06,
+      "loss": 73.032,
+      "step": 34210
+    },
+    {
+      "epoch": 0.13825313008803436,
+      "grad_norm": 507.51470947265625,
+      "learning_rate": 9.080433249745688e-06,
+      "loss": 39.2776,
+      "step": 34220
+    },
+    {
+      "epoch": 0.138293531353402,
+      "grad_norm": 789.2049560546875,
+      "learning_rate": 9.079617994600066e-06,
+      "loss": 37.8734,
+      "step": 34230
+    },
+    {
+      "epoch": 0.1383339326187696,
+      "grad_norm": 470.8572998046875,
+      "learning_rate": 9.078802414857963e-06,
+      "loss": 43.4763,
+      "step": 34240
+    },
+    {
+      "epoch": 0.13837433388413725,
+      "grad_norm": 899.4437866210938,
+      "learning_rate": 9.077986510584273e-06,
+      "loss": 48.1022,
+      "step": 34250
+    },
+    {
+      "epoch": 0.1384147351495049,
+      "grad_norm": 511.8283386230469,
+      "learning_rate": 9.07717028184391e-06,
+      "loss": 89.8688,
+      "step": 34260
+    },
+    {
+      "epoch": 0.1384551364148725,
+      "grad_norm": 485.81951904296875,
+      "learning_rate": 9.07635372870182e-06,
+      "loss": 38.4757,
+      "step": 34270
+    },
+    {
+      "epoch": 0.13849553768024014,
+      "grad_norm": 507.5504455566406,
+      "learning_rate": 9.07553685122297e-06,
+      "loss": 48.0852,
+      "step": 34280
+    },
+    {
+      "epoch": 0.13853593894560778,
+      "grad_norm": 752.1076049804688,
+      "learning_rate": 9.074719649472358e-06,
+      "loss": 40.384,
+      "step": 34290
+    },
+    {
+      "epoch": 0.1385763402109754,
+      "grad_norm": 714.6204833984375,
+      "learning_rate": 9.073902123515005e-06,
+      "loss": 63.1098,
+      "step": 34300
+    },
+    {
+      "epoch": 0.13861674147634304,
+      "grad_norm": 1068.350341796875,
+      "learning_rate": 9.073084273415956e-06,
+      "loss": 63.3068,
+      "step": 34310
+    },
+    {
+      "epoch": 0.13865714274171068,
+      "grad_norm": 1284.7296142578125,
+      "learning_rate": 9.072266099240286e-06,
+      "loss": 47.1397,
+      "step": 34320
+    },
+    {
+      "epoch": 0.1386975440070783,
+      "grad_norm": 1022.5322875976562,
+      "learning_rate": 9.07144760105309e-06,
+      "loss": 56.532,
+      "step": 34330
+    },
+    {
+      "epoch": 0.13873794527244593,
+      "grad_norm": 382.51300048828125,
+      "learning_rate": 9.070628778919493e-06,
+      "loss": 53.3356,
+      "step": 34340
+    },
+    {
+      "epoch": 0.13877834653781357,
+      "grad_norm": 430.1619873046875,
+      "learning_rate": 9.069809632904647e-06,
+      "loss": 39.5561,
+      "step": 34350
+    },
+    {
+      "epoch": 0.1388187478031812,
+      "grad_norm": 416.1811218261719,
+      "learning_rate": 9.068990163073726e-06,
+      "loss": 53.2887,
+      "step": 34360
+    },
+    {
+      "epoch": 0.13885914906854882,
+      "grad_norm": 787.1919555664062,
+      "learning_rate": 9.068170369491932e-06,
+      "loss": 57.9072,
+      "step": 34370
+    },
+    {
+      "epoch": 0.13889955033391646,
+      "grad_norm": 498.4957580566406,
+      "learning_rate": 9.067350252224491e-06,
+      "loss": 61.3324,
+      "step": 34380
+    },
+    {
+      "epoch": 0.1389399515992841,
+      "grad_norm": 690.4000244140625,
+      "learning_rate": 9.066529811336658e-06,
+      "loss": 52.2355,
+      "step": 34390
+    },
+    {
+      "epoch": 0.1389803528646517,
+      "grad_norm": 523.7412719726562,
+      "learning_rate": 9.06570904689371e-06,
+      "loss": 44.2429,
+      "step": 34400
+    },
+    {
+      "epoch": 0.13902075413001935,
+      "grad_norm": 558.5323486328125,
+      "learning_rate": 9.064887958960953e-06,
+      "loss": 47.6126,
+      "step": 34410
+    },
+    {
+      "epoch": 0.139061155395387,
+      "grad_norm": 838.606201171875,
+      "learning_rate": 9.064066547603716e-06,
+      "loss": 48.0207,
+      "step": 34420
+    },
+    {
+      "epoch": 0.1391015566607546,
+      "grad_norm": 698.4739379882812,
+      "learning_rate": 9.063244812887357e-06,
+      "loss": 55.4509,
+      "step": 34430
+    },
+    {
+      "epoch": 0.13914195792612225,
+      "grad_norm": 639.7809448242188,
+      "learning_rate": 9.062422754877253e-06,
+      "loss": 60.6359,
+      "step": 34440
+    },
+    {
+      "epoch": 0.13918235919148988,
+      "grad_norm": 806.6522827148438,
+      "learning_rate": 9.061600373638816e-06,
+      "loss": 36.7169,
+      "step": 34450
+    },
+    {
+      "epoch": 0.1392227604568575,
+      "grad_norm": 737.7337036132812,
+      "learning_rate": 9.06077766923748e-06,
+      "loss": 43.2489,
+      "step": 34460
+    },
+    {
+      "epoch": 0.13926316172222514,
+      "grad_norm": 441.1222839355469,
+      "learning_rate": 9.059954641738697e-06,
+      "loss": 71.4044,
+      "step": 34470
+    },
+    {
+      "epoch": 0.13930356298759278,
+      "grad_norm": 800.96337890625,
+      "learning_rate": 9.059131291207958e-06,
+      "loss": 50.1446,
+      "step": 34480
+    },
+    {
+      "epoch": 0.1393439642529604,
+      "grad_norm": 669.9575805664062,
+      "learning_rate": 9.058307617710771e-06,
+      "loss": 50.5971,
+      "step": 34490
+    },
+    {
+      "epoch": 0.13938436551832803,
+      "grad_norm": 544.849609375,
+      "learning_rate": 9.057483621312671e-06,
+      "loss": 40.0475,
+      "step": 34500
+    },
+    {
+      "epoch": 0.13942476678369567,
+      "grad_norm": 424.7563781738281,
+      "learning_rate": 9.056659302079222e-06,
+      "loss": 38.7654,
+      "step": 34510
+    },
+    {
+      "epoch": 0.1394651680490633,
+      "grad_norm": 240.59954833984375,
+      "learning_rate": 9.055834660076008e-06,
+      "loss": 55.3523,
+      "step": 34520
+    },
+    {
+      "epoch": 0.13950556931443092,
+      "grad_norm": 877.2344360351562,
+      "learning_rate": 9.055009695368646e-06,
+      "loss": 54.841,
+      "step": 34530
+    },
+    {
+      "epoch": 0.13954597057979856,
+      "grad_norm": 714.7890014648438,
+      "learning_rate": 9.054184408022772e-06,
+      "loss": 43.7608,
+      "step": 34540
+    },
+    {
+      "epoch": 0.1395863718451662,
+      "grad_norm": 710.421142578125,
+      "learning_rate": 9.05335879810405e-06,
+      "loss": 78.2367,
+      "step": 34550
+    },
+    {
+      "epoch": 0.13962677311053381,
+      "grad_norm": 763.3213500976562,
+      "learning_rate": 9.052532865678171e-06,
+      "loss": 53.5301,
+      "step": 34560
+    },
+    {
+      "epoch": 0.13966717437590145,
+      "grad_norm": 262.8904724121094,
+      "learning_rate": 9.05170661081085e-06,
+      "loss": 34.3858,
+      "step": 34570
+    },
+    {
+      "epoch": 0.1397075756412691,
+      "grad_norm": 391.6971740722656,
+      "learning_rate": 9.050880033567831e-06,
+      "loss": 41.5089,
+      "step": 34580
+    },
+    {
+      "epoch": 0.1397479769066367,
+      "grad_norm": 320.8378601074219,
+      "learning_rate": 9.050053134014878e-06,
+      "loss": 47.4074,
+      "step": 34590
+    },
+    {
+      "epoch": 0.13978837817200435,
+      "grad_norm": 656.7279663085938,
+      "learning_rate": 9.049225912217782e-06,
+      "loss": 59.3808,
+      "step": 34600
+    },
+    {
+      "epoch": 0.13982877943737199,
+      "grad_norm": 976.0907592773438,
+      "learning_rate": 9.048398368242365e-06,
+      "loss": 39.1846,
+      "step": 34610
+    },
+    {
+      "epoch": 0.1398691807027396,
+      "grad_norm": 622.8963623046875,
+      "learning_rate": 9.047570502154471e-06,
+      "loss": 65.1156,
+      "step": 34620
+    },
+    {
+      "epoch": 0.13990958196810724,
+      "grad_norm": 553.8403930664062,
+      "learning_rate": 9.046742314019968e-06,
+      "loss": 30.1408,
+      "step": 34630
+    },
+    {
+      "epoch": 0.13994998323347488,
+      "grad_norm": 690.28076171875,
+      "learning_rate": 9.045913803904748e-06,
+      "loss": 61.2043,
+      "step": 34640
+    },
+    {
+      "epoch": 0.1399903844988425,
+      "grad_norm": 546.177001953125,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 38.7078,
+      "step": 34650
+    },
+    {
+      "epoch": 0.14003078576421013,
+      "grad_norm": 677.779296875,
+      "learning_rate": 9.04425581799588e-06,
+      "loss": 66.1053,
+      "step": 34660
+    },
+    {
+      "epoch": 0.14007118702957777,
+      "grad_norm": 377.52117919921875,
+      "learning_rate": 9.043426342334147e-06,
+      "loss": 43.3946,
+      "step": 34670
+    },
+    {
+      "epoch": 0.1401115882949454,
+      "grad_norm": 1007.6997680664062,
+      "learning_rate": 9.042596544955538e-06,
+      "loss": 55.9989,
+      "step": 34680
+    },
+    {
+      "epoch": 0.14015198956031302,
+      "grad_norm": 1100.5362548828125,
+      "learning_rate": 9.041766425926073e-06,
+      "loss": 60.7844,
+      "step": 34690
+    },
+    {
+      "epoch": 0.14019239082568066,
+      "grad_norm": 1078.2347412109375,
+      "learning_rate": 9.040935985311804e-06,
+      "loss": 50.4251,
+      "step": 34700
+    },
+    {
+      "epoch": 0.1402327920910483,
+      "grad_norm": 951.633056640625,
+      "learning_rate": 9.040105223178803e-06,
+      "loss": 49.7869,
+      "step": 34710
+    },
+    {
+      "epoch": 0.14027319335641592,
+      "grad_norm": 499.31390380859375,
+      "learning_rate": 9.039274139593173e-06,
+      "loss": 61.8163,
+      "step": 34720
+    },
+    {
+      "epoch": 0.14031359462178356,
+      "grad_norm": 720.83544921875,
+      "learning_rate": 9.038442734621034e-06,
+      "loss": 49.9191,
+      "step": 34730
+    },
+    {
+      "epoch": 0.1403539958871512,
+      "grad_norm": 591.9390258789062,
+      "learning_rate": 9.037611008328544e-06,
+      "loss": 60.4522,
+      "step": 34740
+    },
+    {
+      "epoch": 0.1403943971525188,
+      "grad_norm": 1439.417724609375,
+      "learning_rate": 9.036778960781874e-06,
+      "loss": 65.0005,
+      "step": 34750
+    },
+    {
+      "epoch": 0.14043479841788645,
+      "grad_norm": 1044.05517578125,
+      "learning_rate": 9.03594659204723e-06,
+      "loss": 53.8747,
+      "step": 34760
+    },
+    {
+      "epoch": 0.1404751996832541,
+      "grad_norm": 708.0410766601562,
+      "learning_rate": 9.035113902190838e-06,
+      "loss": 41.8351,
+      "step": 34770
+    },
+    {
+      "epoch": 0.1405156009486217,
+      "grad_norm": 660.4917602539062,
+      "learning_rate": 9.03428089127895e-06,
+      "loss": 52.9228,
+      "step": 34780
+    },
+    {
+      "epoch": 0.14055600221398934,
+      "grad_norm": 654.495361328125,
+      "learning_rate": 9.033447559377847e-06,
+      "loss": 54.6773,
+      "step": 34790
+    },
+    {
+      "epoch": 0.14059640347935698,
+      "grad_norm": 417.1293029785156,
+      "learning_rate": 9.032613906553833e-06,
+      "loss": 44.8771,
+      "step": 34800
+    },
+    {
+      "epoch": 0.1406368047447246,
+      "grad_norm": 663.234375,
+      "learning_rate": 9.031779932873238e-06,
+      "loss": 46.9563,
+      "step": 34810
+    },
+    {
+      "epoch": 0.14067720601009223,
+      "grad_norm": 714.1453857421875,
+      "learning_rate": 9.030945638402415e-06,
+      "loss": 51.3227,
+      "step": 34820
+    },
+    {
+      "epoch": 0.14071760727545987,
+      "grad_norm": 971.0957641601562,
+      "learning_rate": 9.030111023207751e-06,
+      "loss": 59.2087,
+      "step": 34830
+    },
+    {
+      "epoch": 0.1407580085408275,
+      "grad_norm": 619.7679443359375,
+      "learning_rate": 9.029276087355646e-06,
+      "loss": 54.8381,
+      "step": 34840
+    },
+    {
+      "epoch": 0.14079840980619512,
+      "grad_norm": 895.6705932617188,
+      "learning_rate": 9.028440830912536e-06,
+      "loss": 53.2132,
+      "step": 34850
+    },
+    {
+      "epoch": 0.14083881107156276,
+      "grad_norm": 1039.65576171875,
+      "learning_rate": 9.027605253944874e-06,
+      "loss": 42.7459,
+      "step": 34860
+    },
+    {
+      "epoch": 0.1408792123369304,
+      "grad_norm": 603.7433471679688,
+      "learning_rate": 9.026769356519149e-06,
+      "loss": 41.1029,
+      "step": 34870
+    },
+    {
+      "epoch": 0.14091961360229802,
+      "grad_norm": 770.4713134765625,
+      "learning_rate": 9.025933138701865e-06,
+      "loss": 42.8621,
+      "step": 34880
+    },
+    {
+      "epoch": 0.14096001486766566,
+      "grad_norm": 511.4858093261719,
+      "learning_rate": 9.02509660055956e-06,
+      "loss": 56.1045,
+      "step": 34890
+    },
+    {
+      "epoch": 0.1410004161330333,
+      "grad_norm": 436.2823486328125,
+      "learning_rate": 9.02425974215879e-06,
+      "loss": 42.6911,
+      "step": 34900
+    },
+    {
+      "epoch": 0.1410408173984009,
+      "grad_norm": 537.3641967773438,
+      "learning_rate": 9.02342256356614e-06,
+      "loss": 53.8692,
+      "step": 34910
+    },
+    {
+      "epoch": 0.14108121866376855,
+      "grad_norm": 832.6757202148438,
+      "learning_rate": 9.022585064848222e-06,
+      "loss": 40.6428,
+      "step": 34920
+    },
+    {
+      "epoch": 0.1411216199291362,
+      "grad_norm": 300.685546875,
+      "learning_rate": 9.021747246071673e-06,
+      "loss": 37.7268,
+      "step": 34930
+    },
+    {
+      "epoch": 0.1411620211945038,
+      "grad_norm": 360.01593017578125,
+      "learning_rate": 9.020909107303152e-06,
+      "loss": 67.776,
+      "step": 34940
+    },
+    {
+      "epoch": 0.14120242245987144,
+      "grad_norm": 817.0067138671875,
+      "learning_rate": 9.020070648609347e-06,
+      "loss": 54.2647,
+      "step": 34950
+    },
+    {
+      "epoch": 0.14124282372523908,
+      "grad_norm": 697.2791137695312,
+      "learning_rate": 9.01923187005697e-06,
+      "loss": 42.1943,
+      "step": 34960
+    },
+    {
+      "epoch": 0.1412832249906067,
+      "grad_norm": 492.3368835449219,
+      "learning_rate": 9.018392771712758e-06,
+      "loss": 38.4684,
+      "step": 34970
+    },
+    {
+      "epoch": 0.14132362625597433,
+      "grad_norm": 611.7986450195312,
+      "learning_rate": 9.017553353643479e-06,
+      "loss": 61.7044,
+      "step": 34980
+    },
+    {
+      "epoch": 0.14136402752134197,
+      "grad_norm": 373.13775634765625,
+      "learning_rate": 9.016713615915913e-06,
+      "loss": 48.7291,
+      "step": 34990
+    },
+    {
+      "epoch": 0.1414044287867096,
+      "grad_norm": 854.9056396484375,
+      "learning_rate": 9.01587355859688e-06,
+      "loss": 141.1521,
+      "step": 35000
+    },
+    {
+      "epoch": 0.14144483005207723,
+      "grad_norm": 270.6610107421875,
+      "learning_rate": 9.015033181753219e-06,
+      "loss": 38.4525,
+      "step": 35010
+    },
+    {
+      "epoch": 0.14148523131744487,
+      "grad_norm": 889.5081787109375,
+      "learning_rate": 9.014192485451794e-06,
+      "loss": 59.1941,
+      "step": 35020
+    },
+    {
+      "epoch": 0.1415256325828125,
+      "grad_norm": 612.941650390625,
+      "learning_rate": 9.013351469759497e-06,
+      "loss": 64.6319,
+      "step": 35030
+    },
+    {
+      "epoch": 0.14156603384818012,
+      "grad_norm": 803.6441040039062,
+      "learning_rate": 9.01251013474324e-06,
+      "loss": 48.9887,
+      "step": 35040
+    },
+    {
+      "epoch": 0.14160643511354776,
+      "grad_norm": 948.8998413085938,
+      "learning_rate": 9.011668480469969e-06,
+      "loss": 51.4555,
+      "step": 35050
+    },
+    {
+      "epoch": 0.1416468363789154,
+      "grad_norm": 1026.9730224609375,
+      "learning_rate": 9.010826507006644e-06,
+      "loss": 49.5063,
+      "step": 35060
+    },
+    {
+      "epoch": 0.141687237644283,
+      "grad_norm": 800.8943481445312,
+      "learning_rate": 9.009984214420265e-06,
+      "loss": 37.3596,
+      "step": 35070
+    },
+    {
+      "epoch": 0.14172763890965065,
+      "grad_norm": 725.1683349609375,
+      "learning_rate": 9.009141602777845e-06,
+      "loss": 80.1707,
+      "step": 35080
+    },
+    {
+      "epoch": 0.1417680401750183,
+      "grad_norm": 860.2307739257812,
+      "learning_rate": 9.008298672146425e-06,
+      "loss": 46.548,
+      "step": 35090
+    },
+    {
+      "epoch": 0.1418084414403859,
+      "grad_norm": 493.4331359863281,
+      "learning_rate": 9.007455422593077e-06,
+      "loss": 55.3125,
+      "step": 35100
+    },
+    {
+      "epoch": 0.14184884270575354,
+      "grad_norm": 725.03662109375,
+      "learning_rate": 9.006611854184893e-06,
+      "loss": 49.1633,
+      "step": 35110
+    },
+    {
+      "epoch": 0.14188924397112118,
+      "grad_norm": 1479.2064208984375,
+      "learning_rate": 9.00576796698899e-06,
+      "loss": 63.2753,
+      "step": 35120
+    },
+    {
+      "epoch": 0.1419296452364888,
+      "grad_norm": 584.6494140625,
+      "learning_rate": 9.004923761072515e-06,
+      "loss": 49.4957,
+      "step": 35130
+    },
+    {
+      "epoch": 0.14197004650185643,
+      "grad_norm": 293.61279296875,
+      "learning_rate": 9.004079236502636e-06,
+      "loss": 51.4557,
+      "step": 35140
+    },
+    {
+      "epoch": 0.14201044776722407,
+      "grad_norm": 397.71746826171875,
+      "learning_rate": 9.00323439334655e-06,
+      "loss": 52.3445,
+      "step": 35150
+    },
+    {
+      "epoch": 0.14205084903259171,
+      "grad_norm": 522.1874389648438,
+      "learning_rate": 9.002389231671474e-06,
+      "loss": 50.3419,
+      "step": 35160
+    },
+    {
+      "epoch": 0.14209125029795933,
+      "grad_norm": 985.80419921875,
+      "learning_rate": 9.001543751544654e-06,
+      "loss": 61.9348,
+      "step": 35170
+    },
+    {
+      "epoch": 0.14213165156332697,
+      "grad_norm": 755.8569946289062,
+      "learning_rate": 9.000697953033364e-06,
+      "loss": 43.5234,
+      "step": 35180
+    },
+    {
+      "epoch": 0.1421720528286946,
+      "grad_norm": 329.2603454589844,
+      "learning_rate": 8.999851836204901e-06,
+      "loss": 90.2732,
+      "step": 35190
+    },
+    {
+      "epoch": 0.14221245409406222,
+      "grad_norm": 514.22119140625,
+      "learning_rate": 8.99900540112658e-06,
+      "loss": 54.4841,
+      "step": 35200
+    },
+    {
+      "epoch": 0.14225285535942986,
+      "grad_norm": 372.5816955566406,
+      "learning_rate": 8.998158647865753e-06,
+      "loss": 31.2375,
+      "step": 35210
+    },
+    {
+      "epoch": 0.1422932566247975,
+      "grad_norm": 625.9616088867188,
+      "learning_rate": 8.997311576489793e-06,
+      "loss": 58.9871,
+      "step": 35220
+    },
+    {
+      "epoch": 0.1423336578901651,
+      "grad_norm": 702.8914184570312,
+      "learning_rate": 8.996464187066096e-06,
+      "loss": 44.5763,
+      "step": 35230
+    },
+    {
+      "epoch": 0.14237405915553275,
+      "grad_norm": 953.0204467773438,
+      "learning_rate": 8.995616479662084e-06,
+      "loss": 56.9604,
+      "step": 35240
+    },
+    {
+      "epoch": 0.1424144604209004,
+      "grad_norm": 1026.1070556640625,
+      "learning_rate": 8.994768454345207e-06,
+      "loss": 46.73,
+      "step": 35250
+    },
+    {
+      "epoch": 0.142454861686268,
+      "grad_norm": 630.1910400390625,
+      "learning_rate": 8.993920111182937e-06,
+      "loss": 43.647,
+      "step": 35260
+    },
+    {
+      "epoch": 0.14249526295163564,
+      "grad_norm": 320.8638000488281,
+      "learning_rate": 8.993071450242775e-06,
+      "loss": 50.5799,
+      "step": 35270
+    },
+    {
+      "epoch": 0.14253566421700328,
+      "grad_norm": 939.0507202148438,
+      "learning_rate": 8.99222247159224e-06,
+      "loss": 50.8784,
+      "step": 35280
+    },
+    {
+      "epoch": 0.1425760654823709,
+      "grad_norm": 1079.3389892578125,
+      "learning_rate": 8.991373175298887e-06,
+      "loss": 47.8058,
+      "step": 35290
+    },
+    {
+      "epoch": 0.14261646674773854,
+      "grad_norm": 784.1708374023438,
+      "learning_rate": 8.99052356143029e-06,
+      "loss": 43.6722,
+      "step": 35300
+    },
+    {
+      "epoch": 0.14265686801310618,
+      "grad_norm": 781.5846557617188,
+      "learning_rate": 8.989673630054044e-06,
+      "loss": 52.9857,
+      "step": 35310
+    },
+    {
+      "epoch": 0.14269726927847382,
+      "grad_norm": 1505.8087158203125,
+      "learning_rate": 8.988823381237778e-06,
+      "loss": 57.9318,
+      "step": 35320
+    },
+    {
+      "epoch": 0.14273767054384143,
+      "grad_norm": 427.637451171875,
+      "learning_rate": 8.987972815049144e-06,
+      "loss": 51.4073,
+      "step": 35330
+    },
+    {
+      "epoch": 0.14277807180920907,
+      "grad_norm": 641.6649780273438,
+      "learning_rate": 8.987121931555814e-06,
+      "loss": 59.3153,
+      "step": 35340
+    },
+    {
+      "epoch": 0.1428184730745767,
+      "grad_norm": 840.8336181640625,
+      "learning_rate": 8.986270730825489e-06,
+      "loss": 46.8198,
+      "step": 35350
+    },
+    {
+      "epoch": 0.14285887433994432,
+      "grad_norm": 585.4122924804688,
+      "learning_rate": 8.985419212925898e-06,
+      "loss": 54.9574,
+      "step": 35360
+    },
+    {
+      "epoch": 0.14289927560531196,
+      "grad_norm": 700.5527954101562,
+      "learning_rate": 8.98456737792479e-06,
+      "loss": 42.6561,
+      "step": 35370
+    },
+    {
+      "epoch": 0.1429396768706796,
+      "grad_norm": 518.1506958007812,
+      "learning_rate": 8.983715225889942e-06,
+      "loss": 50.0722,
+      "step": 35380
+    },
+    {
+      "epoch": 0.1429800781360472,
+      "grad_norm": 812.709228515625,
+      "learning_rate": 8.982862756889158e-06,
+      "loss": 66.0057,
+      "step": 35390
+    },
+    {
+      "epoch": 0.14302047940141485,
+      "grad_norm": 660.6160888671875,
+      "learning_rate": 8.982009970990262e-06,
+      "loss": 55.8418,
+      "step": 35400
+    },
+    {
+      "epoch": 0.1430608806667825,
+      "grad_norm": 636.583740234375,
+      "learning_rate": 8.98115686826111e-06,
+      "loss": 54.9813,
+      "step": 35410
+    },
+    {
+      "epoch": 0.1431012819321501,
+      "grad_norm": 570.1336059570312,
+      "learning_rate": 8.980303448769574e-06,
+      "loss": 56.2874,
+      "step": 35420
+    },
+    {
+      "epoch": 0.14314168319751774,
+      "grad_norm": 587.9059448242188,
+      "learning_rate": 8.979449712583562e-06,
+      "loss": 67.4113,
+      "step": 35430
+    },
+    {
+      "epoch": 0.14318208446288538,
+      "grad_norm": 845.802001953125,
+      "learning_rate": 8.978595659770997e-06,
+      "loss": 54.5296,
+      "step": 35440
+    },
+    {
+      "epoch": 0.143222485728253,
+      "grad_norm": 873.96044921875,
+      "learning_rate": 8.977741290399836e-06,
+      "loss": 57.2999,
+      "step": 35450
+    },
+    {
+      "epoch": 0.14326288699362064,
+      "grad_norm": 992.7213134765625,
+      "learning_rate": 8.976886604538055e-06,
+      "loss": 52.2296,
+      "step": 35460
+    },
+    {
+      "epoch": 0.14330328825898828,
+      "grad_norm": 628.52587890625,
+      "learning_rate": 8.976031602253661e-06,
+      "loss": 45.6108,
+      "step": 35470
+    },
+    {
+      "epoch": 0.14334368952435592,
+      "grad_norm": 736.7313232421875,
+      "learning_rate": 8.975176283614677e-06,
+      "loss": 52.8563,
+      "step": 35480
+    },
+    {
+      "epoch": 0.14338409078972353,
+      "grad_norm": 614.1516723632812,
+      "learning_rate": 8.97432064868916e-06,
+      "loss": 58.3903,
+      "step": 35490
+    },
+    {
+      "epoch": 0.14342449205509117,
+      "grad_norm": 695.76513671875,
+      "learning_rate": 8.973464697545191e-06,
+      "loss": 45.3606,
+      "step": 35500
+    },
+    {
+      "epoch": 0.1434648933204588,
+      "grad_norm": 527.5994873046875,
+      "learning_rate": 8.97260843025087e-06,
+      "loss": 36.107,
+      "step": 35510
+    },
+    {
+      "epoch": 0.14350529458582642,
+      "grad_norm": 721.7711791992188,
+      "learning_rate": 8.971751846874329e-06,
+      "loss": 47.1207,
+      "step": 35520
+    },
+    {
+      "epoch": 0.14354569585119406,
+      "grad_norm": 338.0601806640625,
+      "learning_rate": 8.97089494748372e-06,
+      "loss": 46.7291,
+      "step": 35530
+    },
+    {
+      "epoch": 0.1435860971165617,
+      "grad_norm": 1139.4949951171875,
+      "learning_rate": 8.970037732147226e-06,
+      "loss": 55.5211,
+      "step": 35540
+    },
+    {
+      "epoch": 0.1436264983819293,
+      "grad_norm": 633.7717895507812,
+      "learning_rate": 8.969180200933048e-06,
+      "loss": 60.3075,
+      "step": 35550
+    },
+    {
+      "epoch": 0.14366689964729695,
+      "grad_norm": 490.70562744140625,
+      "learning_rate": 8.968322353909417e-06,
+      "loss": 86.5231,
+      "step": 35560
+    },
+    {
+      "epoch": 0.1437073009126646,
+      "grad_norm": 558.4280395507812,
+      "learning_rate": 8.96746419114459e-06,
+      "loss": 42.8284,
+      "step": 35570
+    },
+    {
+      "epoch": 0.1437477021780322,
+      "grad_norm": 636.939453125,
+      "learning_rate": 8.966605712706844e-06,
+      "loss": 45.6461,
+      "step": 35580
+    },
+    {
+      "epoch": 0.14378810344339985,
+      "grad_norm": 507.1147766113281,
+      "learning_rate": 8.965746918664486e-06,
+      "loss": 44.9939,
+      "step": 35590
+    },
+    {
+      "epoch": 0.14382850470876749,
+      "grad_norm": 2863.917724609375,
+      "learning_rate": 8.964887809085846e-06,
+      "loss": 66.9657,
+      "step": 35600
+    },
+    {
+      "epoch": 0.1438689059741351,
+      "grad_norm": 489.3071594238281,
+      "learning_rate": 8.96402838403928e-06,
+      "loss": 43.2519,
+      "step": 35610
+    },
+    {
+      "epoch": 0.14390930723950274,
+      "grad_norm": 783.5155639648438,
+      "learning_rate": 8.96316864359317e-06,
+      "loss": 49.5435,
+      "step": 35620
+    },
+    {
+      "epoch": 0.14394970850487038,
+      "grad_norm": 283.0666198730469,
+      "learning_rate": 8.962308587815916e-06,
+      "loss": 41.0137,
+      "step": 35630
+    },
+    {
+      "epoch": 0.14399010977023802,
+      "grad_norm": 1222.13037109375,
+      "learning_rate": 8.961448216775955e-06,
+      "loss": 45.5039,
+      "step": 35640
+    },
+    {
+      "epoch": 0.14403051103560563,
+      "grad_norm": 1376.6578369140625,
+      "learning_rate": 8.960587530541737e-06,
+      "loss": 41.2508,
+      "step": 35650
+    },
+    {
+      "epoch": 0.14407091230097327,
+      "grad_norm": 634.9231567382812,
+      "learning_rate": 8.959726529181748e-06,
+      "loss": 45.087,
+      "step": 35660
+    },
+    {
+      "epoch": 0.1441113135663409,
+      "grad_norm": 559.2318725585938,
+      "learning_rate": 8.95886521276449e-06,
+      "loss": 70.1771,
+      "step": 35670
+    },
+    {
+      "epoch": 0.14415171483170852,
+      "grad_norm": 302.1412048339844,
+      "learning_rate": 8.958003581358498e-06,
+      "loss": 41.9856,
+      "step": 35680
+    },
+    {
+      "epoch": 0.14419211609707616,
+      "grad_norm": 543.0943603515625,
+      "learning_rate": 8.957141635032325e-06,
+      "loss": 47.1019,
+      "step": 35690
+    },
+    {
+      "epoch": 0.1442325173624438,
+      "grad_norm": 298.73309326171875,
+      "learning_rate": 8.956279373854553e-06,
+      "loss": 31.6609,
+      "step": 35700
+    },
+    {
+      "epoch": 0.14427291862781141,
+      "grad_norm": 1010.9532470703125,
+      "learning_rate": 8.955416797893787e-06,
+      "loss": 77.5079,
+      "step": 35710
+    },
+    {
+      "epoch": 0.14431331989317905,
+      "grad_norm": 515.0786743164062,
+      "learning_rate": 8.95455390721866e-06,
+      "loss": 50.1513,
+      "step": 35720
+    },
+    {
+      "epoch": 0.1443537211585467,
+      "grad_norm": 577.6357421875,
+      "learning_rate": 8.953690701897827e-06,
+      "loss": 35.7391,
+      "step": 35730
+    },
+    {
+      "epoch": 0.1443941224239143,
+      "grad_norm": 648.1744995117188,
+      "learning_rate": 8.952827181999973e-06,
+      "loss": 46.5073,
+      "step": 35740
+    },
+    {
+      "epoch": 0.14443452368928195,
+      "grad_norm": 412.1001281738281,
+      "learning_rate": 8.951963347593797e-06,
+      "loss": 65.8637,
+      "step": 35750
+    },
+    {
+      "epoch": 0.1444749249546496,
+      "grad_norm": 602.5824584960938,
+      "learning_rate": 8.951099198748036e-06,
+      "loss": 59.1762,
+      "step": 35760
+    },
+    {
+      "epoch": 0.1445153262200172,
+      "grad_norm": 550.7648315429688,
+      "learning_rate": 8.950234735531445e-06,
+      "loss": 45.6496,
+      "step": 35770
+    },
+    {
+      "epoch": 0.14455572748538484,
+      "grad_norm": 650.7678833007812,
+      "learning_rate": 8.949369958012806e-06,
+      "loss": 57.3394,
+      "step": 35780
+    },
+    {
+      "epoch": 0.14459612875075248,
+      "grad_norm": 167.07852172851562,
+      "learning_rate": 8.948504866260924e-06,
+      "loss": 64.0975,
+      "step": 35790
+    },
+    {
+      "epoch": 0.14463653001612012,
+      "grad_norm": 541.3006591796875,
+      "learning_rate": 8.94763946034463e-06,
+      "loss": 77.2448,
+      "step": 35800
+    },
+    {
+      "epoch": 0.14467693128148773,
+      "grad_norm": 476.4756774902344,
+      "learning_rate": 8.946773740332781e-06,
+      "loss": 57.2655,
+      "step": 35810
+    },
+    {
+      "epoch": 0.14471733254685537,
+      "grad_norm": 255.90960693359375,
+      "learning_rate": 8.945907706294262e-06,
+      "loss": 51.7911,
+      "step": 35820
+    },
+    {
+      "epoch": 0.144757733812223,
+      "grad_norm": 429.4159240722656,
+      "learning_rate": 8.945041358297973e-06,
+      "loss": 46.7109,
+      "step": 35830
+    },
+    {
+      "epoch": 0.14479813507759062,
+      "grad_norm": 570.712646484375,
+      "learning_rate": 8.94417469641285e-06,
+      "loss": 45.2215,
+      "step": 35840
+    },
+    {
+      "epoch": 0.14483853634295826,
+      "grad_norm": 549.2108154296875,
+      "learning_rate": 8.943307720707846e-06,
+      "loss": 50.7016,
+      "step": 35850
+    },
+    {
+      "epoch": 0.1448789376083259,
+      "grad_norm": 1001.9048461914062,
+      "learning_rate": 8.942440431251947e-06,
+      "loss": 54.3763,
+      "step": 35860
+    },
+    {
+      "epoch": 0.14491933887369352,
+      "grad_norm": 418.1084899902344,
+      "learning_rate": 8.941572828114154e-06,
+      "loss": 40.7367,
+      "step": 35870
+    },
+    {
+      "epoch": 0.14495974013906116,
+      "grad_norm": 402.5210876464844,
+      "learning_rate": 8.9407049113635e-06,
+      "loss": 87.0937,
+      "step": 35880
+    },
+    {
+      "epoch": 0.1450001414044288,
+      "grad_norm": 763.7208862304688,
+      "learning_rate": 8.939836681069042e-06,
+      "loss": 50.3678,
+      "step": 35890
+    },
+    {
+      "epoch": 0.1450405426697964,
+      "grad_norm": 1092.267333984375,
+      "learning_rate": 8.938968137299861e-06,
+      "loss": 50.2257,
+      "step": 35900
+    },
+    {
+      "epoch": 0.14508094393516405,
+      "grad_norm": 605.1162109375,
+      "learning_rate": 8.938099280125064e-06,
+      "loss": 44.4844,
+      "step": 35910
+    },
+    {
+      "epoch": 0.1451213452005317,
+      "grad_norm": 440.0594787597656,
+      "learning_rate": 8.937230109613778e-06,
+      "loss": 46.008,
+      "step": 35920
+    },
+    {
+      "epoch": 0.1451617464658993,
+      "grad_norm": 1271.16259765625,
+      "learning_rate": 8.936360625835164e-06,
+      "loss": 55.5143,
+      "step": 35930
+    },
+    {
+      "epoch": 0.14520214773126694,
+      "grad_norm": 631.4669799804688,
+      "learning_rate": 8.935490828858399e-06,
+      "loss": 49.6649,
+      "step": 35940
+    },
+    {
+      "epoch": 0.14524254899663458,
+      "grad_norm": 765.8115234375,
+      "learning_rate": 8.934620718752691e-06,
+      "loss": 36.8788,
+      "step": 35950
+    },
+    {
+      "epoch": 0.14528295026200222,
+      "grad_norm": 1090.13916015625,
+      "learning_rate": 8.933750295587269e-06,
+      "loss": 36.7493,
+      "step": 35960
+    },
+    {
+      "epoch": 0.14532335152736983,
+      "grad_norm": 830.1979370117188,
+      "learning_rate": 8.932879559431392e-06,
+      "loss": 41.6039,
+      "step": 35970
+    },
+    {
+      "epoch": 0.14536375279273747,
+      "grad_norm": 925.3457641601562,
+      "learning_rate": 8.932008510354336e-06,
+      "loss": 50.1517,
+      "step": 35980
+    },
+    {
+      "epoch": 0.1454041540581051,
+      "grad_norm": 359.84588623046875,
+      "learning_rate": 8.931137148425407e-06,
+      "loss": 53.3795,
+      "step": 35990
+    },
+    {
+      "epoch": 0.14544455532347272,
+      "grad_norm": 807.447021484375,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 58.659,
+      "step": 36000
+    },
+    {
+      "epoch": 0.14548495658884036,
+      "grad_norm": 303.23321533203125,
+      "learning_rate": 8.929393486289283e-06,
+      "loss": 36.5493,
+      "step": 36010
+    },
+    {
+      "epoch": 0.145525357854208,
+      "grad_norm": 621.742431640625,
+      "learning_rate": 8.928521186220822e-06,
+      "loss": 46.3721,
+      "step": 36020
+    },
+    {
+      "epoch": 0.14556575911957562,
+      "grad_norm": 442.9004211425781,
+      "learning_rate": 8.92764857357796e-06,
+      "loss": 36.3563,
+      "step": 36030
+    },
+    {
+      "epoch": 0.14560616038494326,
+      "grad_norm": 357.2429504394531,
+      "learning_rate": 8.926775648430124e-06,
+      "loss": 53.3133,
+      "step": 36040
+    },
+    {
+      "epoch": 0.1456465616503109,
+      "grad_norm": 805.5604858398438,
+      "learning_rate": 8.925902410846774e-06,
+      "loss": 49.645,
+      "step": 36050
+    },
+    {
+      "epoch": 0.1456869629156785,
+      "grad_norm": 658.550537109375,
+      "learning_rate": 8.925028860897384e-06,
+      "loss": 61.7202,
+      "step": 36060
+    },
+    {
+      "epoch": 0.14572736418104615,
+      "grad_norm": 505.9190979003906,
+      "learning_rate": 8.924154998651461e-06,
+      "loss": 49.1229,
+      "step": 36070
+    },
+    {
+      "epoch": 0.1457677654464138,
+      "grad_norm": 418.9546203613281,
+      "learning_rate": 8.923280824178538e-06,
+      "loss": 39.7103,
+      "step": 36080
+    },
+    {
+      "epoch": 0.1458081667117814,
+      "grad_norm": 642.6342163085938,
+      "learning_rate": 8.922406337548162e-06,
+      "loss": 46.5422,
+      "step": 36090
+    },
+    {
+      "epoch": 0.14584856797714904,
+      "grad_norm": 412.61297607421875,
+      "learning_rate": 8.921531538829917e-06,
+      "loss": 45.7388,
+      "step": 36100
+    },
+    {
+      "epoch": 0.14588896924251668,
+      "grad_norm": 1076.3936767578125,
+      "learning_rate": 8.920656428093403e-06,
+      "loss": 37.6155,
+      "step": 36110
+    },
+    {
+      "epoch": 0.14592937050788432,
+      "grad_norm": 515.2711181640625,
+      "learning_rate": 8.919781005408251e-06,
+      "loss": 66.2909,
+      "step": 36120
+    },
+    {
+      "epoch": 0.14596977177325193,
+      "grad_norm": 1106.967529296875,
+      "learning_rate": 8.918905270844113e-06,
+      "loss": 51.8518,
+      "step": 36130
+    },
+    {
+      "epoch": 0.14601017303861957,
+      "grad_norm": 619.3397216796875,
+      "learning_rate": 8.918029224470671e-06,
+      "loss": 53.6405,
+      "step": 36140
+    },
+    {
+      "epoch": 0.1460505743039872,
+      "grad_norm": 633.2496948242188,
+      "learning_rate": 8.917152866357621e-06,
+      "loss": 42.5888,
+      "step": 36150
+    },
+    {
+      "epoch": 0.14609097556935483,
+      "grad_norm": 764.9520263671875,
+      "learning_rate": 8.916276196574698e-06,
+      "loss": 35.7627,
+      "step": 36160
+    },
+    {
+      "epoch": 0.14613137683472247,
+      "grad_norm": 1141.3271484375,
+      "learning_rate": 8.91539921519165e-06,
+      "loss": 48.7545,
+      "step": 36170
+    },
+    {
+      "epoch": 0.1461717781000901,
+      "grad_norm": 1107.6766357421875,
+      "learning_rate": 8.914521922278255e-06,
+      "loss": 39.765,
+      "step": 36180
+    },
+    {
+      "epoch": 0.14621217936545772,
+      "grad_norm": 597.0726318359375,
+      "learning_rate": 8.913644317904317e-06,
+      "loss": 74.6474,
+      "step": 36190
+    },
+    {
+      "epoch": 0.14625258063082536,
+      "grad_norm": 713.440185546875,
+      "learning_rate": 8.912766402139662e-06,
+      "loss": 44.1024,
+      "step": 36200
+    },
+    {
+      "epoch": 0.146292981896193,
+      "grad_norm": 466.7696533203125,
+      "learning_rate": 8.91188817505414e-06,
+      "loss": 40.7422,
+      "step": 36210
+    },
+    {
+      "epoch": 0.1463333831615606,
+      "grad_norm": 993.3900756835938,
+      "learning_rate": 8.91100963671763e-06,
+      "loss": 53.3883,
+      "step": 36220
+    },
+    {
+      "epoch": 0.14637378442692825,
+      "grad_norm": 736.5969848632812,
+      "learning_rate": 8.910130787200032e-06,
+      "loss": 50.1694,
+      "step": 36230
+    },
+    {
+      "epoch": 0.1464141856922959,
+      "grad_norm": 2283.6953125,
+      "learning_rate": 8.909251626571273e-06,
+      "loss": 57.6746,
+      "step": 36240
+    },
+    {
+      "epoch": 0.1464545869576635,
+      "grad_norm": 379.3497619628906,
+      "learning_rate": 8.908372154901302e-06,
+      "loss": 38.0152,
+      "step": 36250
+    },
+    {
+      "epoch": 0.14649498822303114,
+      "grad_norm": 441.6664123535156,
+      "learning_rate": 8.907492372260096e-06,
+      "loss": 48.2084,
+      "step": 36260
+    },
+    {
+      "epoch": 0.14653538948839878,
+      "grad_norm": 816.5291137695312,
+      "learning_rate": 8.906612278717657e-06,
+      "loss": 56.6522,
+      "step": 36270
+    },
+    {
+      "epoch": 0.14657579075376642,
+      "grad_norm": 851.9935913085938,
+      "learning_rate": 8.905731874344005e-06,
+      "loss": 66.8219,
+      "step": 36280
+    },
+    {
+      "epoch": 0.14661619201913403,
+      "grad_norm": 553.1090087890625,
+      "learning_rate": 8.904851159209193e-06,
+      "loss": 50.3202,
+      "step": 36290
+    },
+    {
+      "epoch": 0.14665659328450167,
+      "grad_norm": 752.8753662109375,
+      "learning_rate": 8.903970133383297e-06,
+      "loss": 37.1439,
+      "step": 36300
+    },
+    {
+      "epoch": 0.14669699454986931,
+      "grad_norm": 837.05859375,
+      "learning_rate": 8.903088796936414e-06,
+      "loss": 62.281,
+      "step": 36310
+    },
+    {
+      "epoch": 0.14673739581523693,
+      "grad_norm": 162.0035858154297,
+      "learning_rate": 8.902207149938667e-06,
+      "loss": 44.1739,
+      "step": 36320
+    },
+    {
+      "epoch": 0.14677779708060457,
+      "grad_norm": 979.8157958984375,
+      "learning_rate": 8.901325192460206e-06,
+      "loss": 64.2416,
+      "step": 36330
+    },
+    {
+      "epoch": 0.1468181983459722,
+      "grad_norm": 603.8624267578125,
+      "learning_rate": 8.900442924571204e-06,
+      "loss": 44.7346,
+      "step": 36340
+    },
+    {
+      "epoch": 0.14685859961133982,
+      "grad_norm": 530.7752685546875,
+      "learning_rate": 8.89956034634186e-06,
+      "loss": 47.8523,
+      "step": 36350
+    },
+    {
+      "epoch": 0.14689900087670746,
+      "grad_norm": 816.960205078125,
+      "learning_rate": 8.898677457842394e-06,
+      "loss": 50.7503,
+      "step": 36360
+    },
+    {
+      "epoch": 0.1469394021420751,
+      "grad_norm": 1267.8363037109375,
+      "learning_rate": 8.897794259143057e-06,
+      "loss": 59.9809,
+      "step": 36370
+    },
+    {
+      "epoch": 0.1469798034074427,
+      "grad_norm": 404.8268127441406,
+      "learning_rate": 8.896910750314118e-06,
+      "loss": 37.5561,
+      "step": 36380
+    },
+    {
+      "epoch": 0.14702020467281035,
+      "grad_norm": 235.9456024169922,
+      "learning_rate": 8.896026931425876e-06,
+      "loss": 65.6808,
+      "step": 36390
+    },
+    {
+      "epoch": 0.147060605938178,
+      "grad_norm": 507.3228454589844,
+      "learning_rate": 8.895142802548653e-06,
+      "loss": 40.5951,
+      "step": 36400
+    },
+    {
+      "epoch": 0.1471010072035456,
+      "grad_norm": 842.0059204101562,
+      "learning_rate": 8.89425836375279e-06,
+      "loss": 66.0475,
+      "step": 36410
+    },
+    {
+      "epoch": 0.14714140846891324,
+      "grad_norm": 366.0971984863281,
+      "learning_rate": 8.893373615108663e-06,
+      "loss": 37.7263,
+      "step": 36420
+    },
+    {
+      "epoch": 0.14718180973428088,
+      "grad_norm": 629.5830688476562,
+      "learning_rate": 8.892488556686665e-06,
+      "loss": 46.7629,
+      "step": 36430
+    },
+    {
+      "epoch": 0.14722221099964852,
+      "grad_norm": 944.98291015625,
+      "learning_rate": 8.891603188557218e-06,
+      "loss": 48.1869,
+      "step": 36440
+    },
+    {
+      "epoch": 0.14726261226501614,
+      "grad_norm": 791.6475830078125,
+      "learning_rate": 8.890717510790763e-06,
+      "loss": 55.2396,
+      "step": 36450
+    },
+    {
+      "epoch": 0.14730301353038378,
+      "grad_norm": 631.4806518554688,
+      "learning_rate": 8.889831523457773e-06,
+      "loss": 55.5351,
+      "step": 36460
+    },
+    {
+      "epoch": 0.14734341479575142,
+      "grad_norm": 740.2923583984375,
+      "learning_rate": 8.888945226628742e-06,
+      "loss": 45.0742,
+      "step": 36470
+    },
+    {
+      "epoch": 0.14738381606111903,
+      "grad_norm": 294.7581787109375,
+      "learning_rate": 8.888058620374185e-06,
+      "loss": 43.5822,
+      "step": 36480
+    },
+    {
+      "epoch": 0.14742421732648667,
+      "grad_norm": 1752.1192626953125,
+      "learning_rate": 8.887171704764647e-06,
+      "loss": 54.3944,
+      "step": 36490
+    },
+    {
+      "epoch": 0.1474646185918543,
+      "grad_norm": 292.4616394042969,
+      "learning_rate": 8.8862844798707e-06,
+      "loss": 32.088,
+      "step": 36500
+    },
+    {
+      "epoch": 0.14750501985722192,
+      "grad_norm": 411.1035461425781,
+      "learning_rate": 8.885396945762928e-06,
+      "loss": 52.1857,
+      "step": 36510
+    },
+    {
+      "epoch": 0.14754542112258956,
+      "grad_norm": 854.386474609375,
+      "learning_rate": 8.884509102511956e-06,
+      "loss": 65.9514,
+      "step": 36520
+    },
+    {
+      "epoch": 0.1475858223879572,
+      "grad_norm": 438.49725341796875,
+      "learning_rate": 8.883620950188422e-06,
+      "loss": 42.1549,
+      "step": 36530
+    },
+    {
+      "epoch": 0.1476262236533248,
+      "grad_norm": 482.4796447753906,
+      "learning_rate": 8.882732488862988e-06,
+      "loss": 49.8962,
+      "step": 36540
+    },
+    {
+      "epoch": 0.14766662491869245,
+      "grad_norm": 671.1175537109375,
+      "learning_rate": 8.881843718606353e-06,
+      "loss": 66.4913,
+      "step": 36550
+    },
+    {
+      "epoch": 0.1477070261840601,
+      "grad_norm": 496.26824951171875,
+      "learning_rate": 8.880954639489227e-06,
+      "loss": 36.7379,
+      "step": 36560
+    },
+    {
+      "epoch": 0.1477474274494277,
+      "grad_norm": 560.552490234375,
+      "learning_rate": 8.880065251582354e-06,
+      "loss": 66.3091,
+      "step": 36570
+    },
+    {
+      "epoch": 0.14778782871479534,
+      "grad_norm": 558.4953002929688,
+      "learning_rate": 8.879175554956495e-06,
+      "loss": 60.9963,
+      "step": 36580
+    },
+    {
+      "epoch": 0.14782822998016298,
+      "grad_norm": 400.2424621582031,
+      "learning_rate": 8.87828554968244e-06,
+      "loss": 37.0019,
+      "step": 36590
+    },
+    {
+      "epoch": 0.1478686312455306,
+      "grad_norm": 1002.6381225585938,
+      "learning_rate": 8.877395235831002e-06,
+      "loss": 51.9359,
+      "step": 36600
+    },
+    {
+      "epoch": 0.14790903251089824,
+      "grad_norm": 455.9054260253906,
+      "learning_rate": 8.876504613473019e-06,
+      "loss": 66.0401,
+      "step": 36610
+    },
+    {
+      "epoch": 0.14794943377626588,
+      "grad_norm": 345.1482849121094,
+      "learning_rate": 8.875613682679356e-06,
+      "loss": 60.0617,
+      "step": 36620
+    },
+    {
+      "epoch": 0.14798983504163352,
+      "grad_norm": 840.730224609375,
+      "learning_rate": 8.874722443520898e-06,
+      "loss": 56.1831,
+      "step": 36630
+    },
+    {
+      "epoch": 0.14803023630700113,
+      "grad_norm": 566.4497680664062,
+      "learning_rate": 8.873830896068559e-06,
+      "loss": 44.0531,
+      "step": 36640
+    },
+    {
+      "epoch": 0.14807063757236877,
+      "grad_norm": 720.5826416015625,
+      "learning_rate": 8.872939040393274e-06,
+      "loss": 76.8549,
+      "step": 36650
+    },
+    {
+      "epoch": 0.1481110388377364,
+      "grad_norm": 1977.6263427734375,
+      "learning_rate": 8.872046876566003e-06,
+      "loss": 65.1014,
+      "step": 36660
+    },
+    {
+      "epoch": 0.14815144010310402,
+      "grad_norm": 617.149658203125,
+      "learning_rate": 8.871154404657734e-06,
+      "loss": 55.5341,
+      "step": 36670
+    },
+    {
+      "epoch": 0.14819184136847166,
+      "grad_norm": 433.0403137207031,
+      "learning_rate": 8.870261624739474e-06,
+      "loss": 46.8973,
+      "step": 36680
+    },
+    {
+      "epoch": 0.1482322426338393,
+      "grad_norm": 637.2809448242188,
+      "learning_rate": 8.869368536882258e-06,
+      "loss": 40.903,
+      "step": 36690
+    },
+    {
+      "epoch": 0.1482726438992069,
+      "grad_norm": 505.6259460449219,
+      "learning_rate": 8.868475141157146e-06,
+      "loss": 47.5848,
+      "step": 36700
+    },
+    {
+      "epoch": 0.14831304516457455,
+      "grad_norm": 929.6765747070312,
+      "learning_rate": 8.867581437635221e-06,
+      "loss": 64.581,
+      "step": 36710
+    },
+    {
+      "epoch": 0.1483534464299422,
+      "grad_norm": 293.0738525390625,
+      "learning_rate": 8.866687426387592e-06,
+      "loss": 50.6497,
+      "step": 36720
+    },
+    {
+      "epoch": 0.1483938476953098,
+      "grad_norm": 758.82666015625,
+      "learning_rate": 8.86579310748539e-06,
+      "loss": 45.4429,
+      "step": 36730
+    },
+    {
+      "epoch": 0.14843424896067745,
+      "grad_norm": 482.9668884277344,
+      "learning_rate": 8.86489848099977e-06,
+      "loss": 52.9492,
+      "step": 36740
+    },
+    {
+      "epoch": 0.14847465022604509,
+      "grad_norm": 637.962158203125,
+      "learning_rate": 8.864003547001916e-06,
+      "loss": 34.2872,
+      "step": 36750
+    },
+    {
+      "epoch": 0.1485150514914127,
+      "grad_norm": 288.27679443359375,
+      "learning_rate": 8.863108305563035e-06,
+      "loss": 75.5196,
+      "step": 36760
+    },
+    {
+      "epoch": 0.14855545275678034,
+      "grad_norm": 463.9593200683594,
+      "learning_rate": 8.862212756754354e-06,
+      "loss": 45.1242,
+      "step": 36770
+    },
+    {
+      "epoch": 0.14859585402214798,
+      "grad_norm": 1444.18798828125,
+      "learning_rate": 8.861316900647129e-06,
+      "loss": 52.0546,
+      "step": 36780
+    },
+    {
+      "epoch": 0.14863625528751562,
+      "grad_norm": 2431.045654296875,
+      "learning_rate": 8.860420737312638e-06,
+      "loss": 79.4979,
+      "step": 36790
+    },
+    {
+      "epoch": 0.14867665655288323,
+      "grad_norm": 451.2615051269531,
+      "learning_rate": 8.859524266822188e-06,
+      "loss": 36.7135,
+      "step": 36800
+    },
+    {
+      "epoch": 0.14871705781825087,
+      "grad_norm": 350.17462158203125,
+      "learning_rate": 8.858627489247105e-06,
+      "loss": 46.1026,
+      "step": 36810
+    },
+    {
+      "epoch": 0.1487574590836185,
+      "grad_norm": 908.991455078125,
+      "learning_rate": 8.85773040465874e-06,
+      "loss": 46.8176,
+      "step": 36820
+    },
+    {
+      "epoch": 0.14879786034898612,
+      "grad_norm": 1497.7669677734375,
+      "learning_rate": 8.856833013128472e-06,
+      "loss": 60.939,
+      "step": 36830
+    },
+    {
+      "epoch": 0.14883826161435376,
+      "grad_norm": 740.0812377929688,
+      "learning_rate": 8.855935314727702e-06,
+      "loss": 65.2613,
+      "step": 36840
+    },
+    {
+      "epoch": 0.1488786628797214,
+      "grad_norm": 394.499755859375,
+      "learning_rate": 8.855037309527854e-06,
+      "loss": 34.5515,
+      "step": 36850
+    },
+    {
+      "epoch": 0.14891906414508901,
+      "grad_norm": 317.7518005371094,
+      "learning_rate": 8.854138997600382e-06,
+      "loss": 42.0422,
+      "step": 36860
+    },
+    {
+      "epoch": 0.14895946541045665,
+      "grad_norm": 445.2290344238281,
+      "learning_rate": 8.853240379016757e-06,
+      "loss": 70.6691,
+      "step": 36870
+    },
+    {
+      "epoch": 0.1489998666758243,
+      "grad_norm": 675.0117797851562,
+      "learning_rate": 8.852341453848477e-06,
+      "loss": 55.443,
+      "step": 36880
+    },
+    {
+      "epoch": 0.1490402679411919,
+      "grad_norm": 491.013671875,
+      "learning_rate": 8.851442222167068e-06,
+      "loss": 49.8648,
+      "step": 36890
+    },
+    {
+      "epoch": 0.14908066920655955,
+      "grad_norm": 708.0944213867188,
+      "learning_rate": 8.850542684044078e-06,
+      "loss": 28.9624,
+      "step": 36900
+    },
+    {
+      "epoch": 0.1491210704719272,
+      "grad_norm": 384.8024597167969,
+      "learning_rate": 8.849642839551079e-06,
+      "loss": 40.8231,
+      "step": 36910
+    },
+    {
+      "epoch": 0.1491614717372948,
+      "grad_norm": 462.9109191894531,
+      "learning_rate": 8.848742688759666e-06,
+      "loss": 46.5803,
+      "step": 36920
+    },
+    {
+      "epoch": 0.14920187300266244,
+      "grad_norm": 207.1934814453125,
+      "learning_rate": 8.847842231741462e-06,
+      "loss": 41.873,
+      "step": 36930
+    },
+    {
+      "epoch": 0.14924227426803008,
+      "grad_norm": 603.2200317382812,
+      "learning_rate": 8.846941468568108e-06,
+      "loss": 48.2413,
+      "step": 36940
+    },
+    {
+      "epoch": 0.14928267553339772,
+      "grad_norm": 399.51690673828125,
+      "learning_rate": 8.846040399311278e-06,
+      "loss": 48.9574,
+      "step": 36950
+    },
+    {
+      "epoch": 0.14932307679876533,
+      "grad_norm": 1572.202392578125,
+      "learning_rate": 8.845139024042664e-06,
+      "loss": 68.8805,
+      "step": 36960
+    },
+    {
+      "epoch": 0.14936347806413297,
+      "grad_norm": 1128.0,
+      "learning_rate": 8.844237342833985e-06,
+      "loss": 54.9947,
+      "step": 36970
+    },
+    {
+      "epoch": 0.1494038793295006,
+      "grad_norm": 360.4505920410156,
+      "learning_rate": 8.843335355756983e-06,
+      "loss": 53.7954,
+      "step": 36980
+    },
+    {
+      "epoch": 0.14944428059486822,
+      "grad_norm": 4446.03125,
+      "learning_rate": 8.842433062883427e-06,
+      "loss": 73.6097,
+      "step": 36990
+    },
+    {
+      "epoch": 0.14948468186023586,
+      "grad_norm": 595.9097900390625,
+      "learning_rate": 8.841530464285105e-06,
+      "loss": 47.5093,
+      "step": 37000
+    },
+    {
+      "epoch": 0.1495250831256035,
+      "grad_norm": 248.24876403808594,
+      "learning_rate": 8.840627560033833e-06,
+      "loss": 46.4306,
+      "step": 37010
+    },
+    {
+      "epoch": 0.14956548439097112,
+      "grad_norm": 907.060546875,
+      "learning_rate": 8.839724350201452e-06,
+      "loss": 54.754,
+      "step": 37020
+    },
+    {
+      "epoch": 0.14960588565633876,
+      "grad_norm": 925.5053100585938,
+      "learning_rate": 8.838820834859829e-06,
+      "loss": 60.8077,
+      "step": 37030
+    },
+    {
+      "epoch": 0.1496462869217064,
+      "grad_norm": 633.4927978515625,
+      "learning_rate": 8.837917014080849e-06,
+      "loss": 43.2474,
+      "step": 37040
+    },
+    {
+      "epoch": 0.149686688187074,
+      "grad_norm": 612.7200317382812,
+      "learning_rate": 8.837012887936426e-06,
+      "loss": 44.9851,
+      "step": 37050
+    },
+    {
+      "epoch": 0.14972708945244165,
+      "grad_norm": 738.667724609375,
+      "learning_rate": 8.836108456498497e-06,
+      "loss": 55.44,
+      "step": 37060
+    },
+    {
+      "epoch": 0.1497674907178093,
+      "grad_norm": 415.4226379394531,
+      "learning_rate": 8.835203719839024e-06,
+      "loss": 52.5291,
+      "step": 37070
+    },
+    {
+      "epoch": 0.1498078919831769,
+      "grad_norm": 367.70318603515625,
+      "learning_rate": 8.834298678029988e-06,
+      "loss": 43.8007,
+      "step": 37080
+    },
+    {
+      "epoch": 0.14984829324854454,
+      "grad_norm": 418.97137451171875,
+      "learning_rate": 8.833393331143409e-06,
+      "loss": 30.1141,
+      "step": 37090
+    },
+    {
+      "epoch": 0.14988869451391218,
+      "grad_norm": 1005.45263671875,
+      "learning_rate": 8.832487679251311e-06,
+      "loss": 50.5335,
+      "step": 37100
+    },
+    {
+      "epoch": 0.14992909577927982,
+      "grad_norm": 639.8682250976562,
+      "learning_rate": 8.831581722425761e-06,
+      "loss": 49.5506,
+      "step": 37110
+    },
+    {
+      "epoch": 0.14996949704464743,
+      "grad_norm": 1013.299072265625,
+      "learning_rate": 8.830675460738835e-06,
+      "loss": 45.148,
+      "step": 37120
+    },
+    {
+      "epoch": 0.15000989831001507,
+      "grad_norm": 595.0233154296875,
+      "learning_rate": 8.829768894262644e-06,
+      "loss": 65.652,
+      "step": 37130
+    },
+    {
+      "epoch": 0.1500502995753827,
+      "grad_norm": 849.735595703125,
+      "learning_rate": 8.82886202306932e-06,
+      "loss": 50.526,
+      "step": 37140
+    },
+    {
+      "epoch": 0.15009070084075032,
+      "grad_norm": 346.1163635253906,
+      "learning_rate": 8.827954847231016e-06,
+      "loss": 59.2245,
+      "step": 37150
+    },
+    {
+      "epoch": 0.15013110210611796,
+      "grad_norm": 610.0313720703125,
+      "learning_rate": 8.82704736681991e-06,
+      "loss": 53.6018,
+      "step": 37160
+    },
+    {
+      "epoch": 0.1501715033714856,
+      "grad_norm": 753.264892578125,
+      "learning_rate": 8.826139581908211e-06,
+      "loss": 50.999,
+      "step": 37170
+    },
+    {
+      "epoch": 0.15021190463685322,
+      "grad_norm": 397.34228515625,
+      "learning_rate": 8.825231492568146e-06,
+      "loss": 39.9716,
+      "step": 37180
+    },
+    {
+      "epoch": 0.15025230590222086,
+      "grad_norm": 787.6273803710938,
+      "learning_rate": 8.824323098871966e-06,
+      "loss": 59.7436,
+      "step": 37190
+    },
+    {
+      "epoch": 0.1502927071675885,
+      "grad_norm": 794.3360595703125,
+      "learning_rate": 8.823414400891948e-06,
+      "loss": 55.8264,
+      "step": 37200
+    },
+    {
+      "epoch": 0.1503331084329561,
+      "grad_norm": 556.7963256835938,
+      "learning_rate": 8.822505398700395e-06,
+      "loss": 62.2807,
+      "step": 37210
+    },
+    {
+      "epoch": 0.15037350969832375,
+      "grad_norm": 652.981201171875,
+      "learning_rate": 8.821596092369627e-06,
+      "loss": 63.0537,
+      "step": 37220
+    },
+    {
+      "epoch": 0.1504139109636914,
+      "grad_norm": 525.3092041015625,
+      "learning_rate": 8.820686481971998e-06,
+      "loss": 40.2604,
+      "step": 37230
+    },
+    {
+      "epoch": 0.150454312229059,
+      "grad_norm": 302.7001647949219,
+      "learning_rate": 8.81977656757988e-06,
+      "loss": 34.4532,
+      "step": 37240
+    },
+    {
+      "epoch": 0.15049471349442664,
+      "grad_norm": 541.8635864257812,
+      "learning_rate": 8.81886634926567e-06,
+      "loss": 55.9025,
+      "step": 37250
+    },
+    {
+      "epoch": 0.15053511475979428,
+      "grad_norm": 504.7498474121094,
+      "learning_rate": 8.817955827101794e-06,
+      "loss": 59.6146,
+      "step": 37260
+    },
+    {
+      "epoch": 0.15057551602516192,
+      "grad_norm": 657.9691772460938,
+      "learning_rate": 8.817045001160693e-06,
+      "loss": 59.897,
+      "step": 37270
+    },
+    {
+      "epoch": 0.15061591729052953,
+      "grad_norm": 662.6056518554688,
+      "learning_rate": 8.816133871514838e-06,
+      "loss": 48.5258,
+      "step": 37280
+    },
+    {
+      "epoch": 0.15065631855589717,
+      "grad_norm": 480.95965576171875,
+      "learning_rate": 8.815222438236726e-06,
+      "loss": 62.4884,
+      "step": 37290
+    },
+    {
+      "epoch": 0.1506967198212648,
+      "grad_norm": 1077.7459716796875,
+      "learning_rate": 8.814310701398873e-06,
+      "loss": 38.3544,
+      "step": 37300
+    },
+    {
+      "epoch": 0.15073712108663243,
+      "grad_norm": 389.59796142578125,
+      "learning_rate": 8.813398661073823e-06,
+      "loss": 38.4094,
+      "step": 37310
+    },
+    {
+      "epoch": 0.15077752235200007,
+      "grad_norm": 521.3574829101562,
+      "learning_rate": 8.812486317334145e-06,
+      "loss": 30.9333,
+      "step": 37320
+    },
+    {
+      "epoch": 0.1508179236173677,
+      "grad_norm": 846.1622314453125,
+      "learning_rate": 8.811573670252426e-06,
+      "loss": 60.0239,
+      "step": 37330
+    },
+    {
+      "epoch": 0.15085832488273532,
+      "grad_norm": 554.4835815429688,
+      "learning_rate": 8.810660719901283e-06,
+      "loss": 33.6676,
+      "step": 37340
+    },
+    {
+      "epoch": 0.15089872614810296,
+      "grad_norm": 316.88494873046875,
+      "learning_rate": 8.809747466353356e-06,
+      "loss": 40.518,
+      "step": 37350
+    },
+    {
+      "epoch": 0.1509391274134706,
+      "grad_norm": 601.1395874023438,
+      "learning_rate": 8.808833909681305e-06,
+      "loss": 40.8812,
+      "step": 37360
+    },
+    {
+      "epoch": 0.1509795286788382,
+      "grad_norm": 449.86737060546875,
+      "learning_rate": 8.80792004995782e-06,
+      "loss": 49.2103,
+      "step": 37370
+    },
+    {
+      "epoch": 0.15101992994420585,
+      "grad_norm": 998.6974487304688,
+      "learning_rate": 8.807005887255615e-06,
+      "loss": 40.9746,
+      "step": 37380
+    },
+    {
+      "epoch": 0.1510603312095735,
+      "grad_norm": 662.0831909179688,
+      "learning_rate": 8.806091421647423e-06,
+      "loss": 42.3315,
+      "step": 37390
+    },
+    {
+      "epoch": 0.1511007324749411,
+      "grad_norm": 697.000244140625,
+      "learning_rate": 8.805176653206004e-06,
+      "loss": 57.2413,
+      "step": 37400
+    },
+    {
+      "epoch": 0.15114113374030874,
+      "grad_norm": 287.0083923339844,
+      "learning_rate": 8.80426158200414e-06,
+      "loss": 57.0611,
+      "step": 37410
+    },
+    {
+      "epoch": 0.15118153500567638,
+      "grad_norm": 656.7665405273438,
+      "learning_rate": 8.803346208114643e-06,
+      "loss": 59.7992,
+      "step": 37420
+    },
+    {
+      "epoch": 0.15122193627104402,
+      "grad_norm": 535.0011596679688,
+      "learning_rate": 8.802430531610344e-06,
+      "loss": 37.0664,
+      "step": 37430
+    },
+    {
+      "epoch": 0.15126233753641163,
+      "grad_norm": 401.31475830078125,
+      "learning_rate": 8.801514552564097e-06,
+      "loss": 41.4795,
+      "step": 37440
+    },
+    {
+      "epoch": 0.15130273880177927,
+      "grad_norm": 525.552490234375,
+      "learning_rate": 8.800598271048784e-06,
+      "loss": 42.3817,
+      "step": 37450
+    },
+    {
+      "epoch": 0.15134314006714691,
+      "grad_norm": 858.5873413085938,
+      "learning_rate": 8.799681687137309e-06,
+      "loss": 56.0704,
+      "step": 37460
+    },
+    {
+      "epoch": 0.15138354133251453,
+      "grad_norm": 531.9202270507812,
+      "learning_rate": 8.7987648009026e-06,
+      "loss": 44.5954,
+      "step": 37470
+    },
+    {
+      "epoch": 0.15142394259788217,
+      "grad_norm": 686.9660034179688,
+      "learning_rate": 8.79784761241761e-06,
+      "loss": 29.8809,
+      "step": 37480
+    },
+    {
+      "epoch": 0.1514643438632498,
+      "grad_norm": 616.3722534179688,
+      "learning_rate": 8.796930121755315e-06,
+      "loss": 51.981,
+      "step": 37490
+    },
+    {
+      "epoch": 0.15150474512861742,
+      "grad_norm": 538.9678344726562,
+      "learning_rate": 8.796012328988716e-06,
+      "loss": 63.4237,
+      "step": 37500
+    },
+    {
+      "epoch": 0.15154514639398506,
+      "grad_norm": 567.8744506835938,
+      "learning_rate": 8.795094234190837e-06,
+      "loss": 41.3883,
+      "step": 37510
+    },
+    {
+      "epoch": 0.1515855476593527,
+      "grad_norm": 544.1000366210938,
+      "learning_rate": 8.794175837434729e-06,
+      "loss": 55.9233,
+      "step": 37520
+    },
+    {
+      "epoch": 0.1516259489247203,
+      "grad_norm": 735.0166625976562,
+      "learning_rate": 8.79325713879346e-06,
+      "loss": 51.2997,
+      "step": 37530
+    },
+    {
+      "epoch": 0.15166635019008795,
+      "grad_norm": 494.5010681152344,
+      "learning_rate": 8.792338138340131e-06,
+      "loss": 39.8792,
+      "step": 37540
+    },
+    {
+      "epoch": 0.1517067514554556,
+      "grad_norm": 971.3145141601562,
+      "learning_rate": 8.791418836147858e-06,
+      "loss": 38.6304,
+      "step": 37550
+    },
+    {
+      "epoch": 0.1517471527208232,
+      "grad_norm": 639.4276733398438,
+      "learning_rate": 8.790499232289793e-06,
+      "loss": 49.2647,
+      "step": 37560
+    },
+    {
+      "epoch": 0.15178755398619084,
+      "grad_norm": 260.048095703125,
+      "learning_rate": 8.789579326839097e-06,
+      "loss": 55.1944,
+      "step": 37570
+    },
+    {
+      "epoch": 0.15182795525155848,
+      "grad_norm": 534.6347045898438,
+      "learning_rate": 8.788659119868966e-06,
+      "loss": 40.731,
+      "step": 37580
+    },
+    {
+      "epoch": 0.15186835651692612,
+      "grad_norm": 695.5957641601562,
+      "learning_rate": 8.787738611452616e-06,
+      "loss": 42.3443,
+      "step": 37590
+    },
+    {
+      "epoch": 0.15190875778229374,
+      "grad_norm": 640.8202514648438,
+      "learning_rate": 8.78681780166329e-06,
+      "loss": 53.9494,
+      "step": 37600
+    },
+    {
+      "epoch": 0.15194915904766138,
+      "grad_norm": 815.8932495117188,
+      "learning_rate": 8.785896690574248e-06,
+      "loss": 39.8566,
+      "step": 37610
+    },
+    {
+      "epoch": 0.15198956031302902,
+      "grad_norm": 292.79315185546875,
+      "learning_rate": 8.784975278258783e-06,
+      "loss": 40.0339,
+      "step": 37620
+    },
+    {
+      "epoch": 0.15202996157839663,
+      "grad_norm": 973.7109375,
+      "learning_rate": 8.784053564790205e-06,
+      "loss": 60.2069,
+      "step": 37630
+    },
+    {
+      "epoch": 0.15207036284376427,
+      "grad_norm": 748.5640258789062,
+      "learning_rate": 8.783131550241853e-06,
+      "loss": 64.9437,
+      "step": 37640
+    },
+    {
+      "epoch": 0.1521107641091319,
+      "grad_norm": 1388.9613037109375,
+      "learning_rate": 8.782209234687083e-06,
+      "loss": 55.6427,
+      "step": 37650
+    },
+    {
+      "epoch": 0.15215116537449952,
+      "grad_norm": 592.8505249023438,
+      "learning_rate": 8.781286618199285e-06,
+      "loss": 45.0461,
+      "step": 37660
+    },
+    {
+      "epoch": 0.15219156663986716,
+      "grad_norm": 513.772705078125,
+      "learning_rate": 8.780363700851863e-06,
+      "loss": 42.5973,
+      "step": 37670
+    },
+    {
+      "epoch": 0.1522319679052348,
+      "grad_norm": 502.7109069824219,
+      "learning_rate": 8.779440482718251e-06,
+      "loss": 50.6042,
+      "step": 37680
+    },
+    {
+      "epoch": 0.1522723691706024,
+      "grad_norm": 385.12823486328125,
+      "learning_rate": 8.778516963871904e-06,
+      "loss": 30.2378,
+      "step": 37690
+    },
+    {
+      "epoch": 0.15231277043597005,
+      "grad_norm": 677.427734375,
+      "learning_rate": 8.777593144386305e-06,
+      "loss": 53.152,
+      "step": 37700
+    },
+    {
+      "epoch": 0.1523531717013377,
+      "grad_norm": 401.56939697265625,
+      "learning_rate": 8.776669024334955e-06,
+      "loss": 43.2506,
+      "step": 37710
+    },
+    {
+      "epoch": 0.1523935729667053,
+      "grad_norm": 674.8109741210938,
+      "learning_rate": 8.775744603791385e-06,
+      "loss": 44.2603,
+      "step": 37720
+    },
+    {
+      "epoch": 0.15243397423207294,
+      "grad_norm": 442.9088439941406,
+      "learning_rate": 8.774819882829144e-06,
+      "loss": 30.5279,
+      "step": 37730
+    },
+    {
+      "epoch": 0.15247437549744058,
+      "grad_norm": 689.316650390625,
+      "learning_rate": 8.77389486152181e-06,
+      "loss": 45.8222,
+      "step": 37740
+    },
+    {
+      "epoch": 0.15251477676280822,
+      "grad_norm": 490.8953552246094,
+      "learning_rate": 8.772969539942981e-06,
+      "loss": 41.8513,
+      "step": 37750
+    },
+    {
+      "epoch": 0.15255517802817584,
+      "grad_norm": 1018.9712524414062,
+      "learning_rate": 8.772043918166282e-06,
+      "loss": 43.3342,
+      "step": 37760
+    },
+    {
+      "epoch": 0.15259557929354348,
+      "grad_norm": 679.5645141601562,
+      "learning_rate": 8.771117996265358e-06,
+      "loss": 87.8444,
+      "step": 37770
+    },
+    {
+      "epoch": 0.15263598055891112,
+      "grad_norm": 623.6113891601562,
+      "learning_rate": 8.770191774313883e-06,
+      "loss": 48.4947,
+      "step": 37780
+    },
+    {
+      "epoch": 0.15267638182427873,
+      "grad_norm": 605.280517578125,
+      "learning_rate": 8.769265252385552e-06,
+      "loss": 47.7166,
+      "step": 37790
+    },
+    {
+      "epoch": 0.15271678308964637,
+      "grad_norm": 575.60791015625,
+      "learning_rate": 8.768338430554083e-06,
+      "loss": 29.7103,
+      "step": 37800
+    },
+    {
+      "epoch": 0.152757184355014,
+      "grad_norm": 729.2935180664062,
+      "learning_rate": 8.76741130889322e-06,
+      "loss": 44.2859,
+      "step": 37810
+    },
+    {
+      "epoch": 0.15279758562038162,
+      "grad_norm": 816.4547729492188,
+      "learning_rate": 8.766483887476727e-06,
+      "loss": 52.9964,
+      "step": 37820
+    },
+    {
+      "epoch": 0.15283798688574926,
+      "grad_norm": 1206.9049072265625,
+      "learning_rate": 8.7655561663784e-06,
+      "loss": 63.2851,
+      "step": 37830
+    },
+    {
+      "epoch": 0.1528783881511169,
+      "grad_norm": 947.2355346679688,
+      "learning_rate": 8.764628145672048e-06,
+      "loss": 60.9178,
+      "step": 37840
+    },
+    {
+      "epoch": 0.1529187894164845,
+      "grad_norm": 1392.46923828125,
+      "learning_rate": 8.763699825431513e-06,
+      "loss": 52.3686,
+      "step": 37850
+    },
+    {
+      "epoch": 0.15295919068185215,
+      "grad_norm": 374.6267395019531,
+      "learning_rate": 8.762771205730656e-06,
+      "loss": 32.7278,
+      "step": 37860
+    },
+    {
+      "epoch": 0.1529995919472198,
+      "grad_norm": 583.3077392578125,
+      "learning_rate": 8.761842286643362e-06,
+      "loss": 42.1567,
+      "step": 37870
+    },
+    {
+      "epoch": 0.1530399932125874,
+      "grad_norm": 671.8289184570312,
+      "learning_rate": 8.760913068243542e-06,
+      "loss": 56.332,
+      "step": 37880
+    },
+    {
+      "epoch": 0.15308039447795505,
+      "grad_norm": 514.0142211914062,
+      "learning_rate": 8.759983550605132e-06,
+      "loss": 37.4075,
+      "step": 37890
+    },
+    {
+      "epoch": 0.15312079574332269,
+      "grad_norm": 417.3742980957031,
+      "learning_rate": 8.759053733802083e-06,
+      "loss": 32.1351,
+      "step": 37900
+    },
+    {
+      "epoch": 0.15316119700869033,
+      "grad_norm": 363.85223388671875,
+      "learning_rate": 8.758123617908383e-06,
+      "loss": 43.9274,
+      "step": 37910
+    },
+    {
+      "epoch": 0.15320159827405794,
+      "grad_norm": 392.4961242675781,
+      "learning_rate": 8.757193202998033e-06,
+      "loss": 40.1141,
+      "step": 37920
+    },
+    {
+      "epoch": 0.15324199953942558,
+      "grad_norm": 572.553955078125,
+      "learning_rate": 8.756262489145061e-06,
+      "loss": 47.0318,
+      "step": 37930
+    },
+    {
+      "epoch": 0.15328240080479322,
+      "grad_norm": 440.2536926269531,
+      "learning_rate": 8.755331476423526e-06,
+      "loss": 45.7451,
+      "step": 37940
+    },
+    {
+      "epoch": 0.15332280207016083,
+      "grad_norm": 506.631591796875,
+      "learning_rate": 8.754400164907496e-06,
+      "loss": 45.7129,
+      "step": 37950
+    },
+    {
+      "epoch": 0.15336320333552847,
+      "grad_norm": 315.4403991699219,
+      "learning_rate": 8.753468554671078e-06,
+      "loss": 45.1952,
+      "step": 37960
+    },
+    {
+      "epoch": 0.1534036046008961,
+      "grad_norm": 469.86474609375,
+      "learning_rate": 8.752536645788391e-06,
+      "loss": 57.6503,
+      "step": 37970
+    },
+    {
+      "epoch": 0.15344400586626372,
+      "grad_norm": 456.01617431640625,
+      "learning_rate": 8.751604438333587e-06,
+      "loss": 53.0288,
+      "step": 37980
+    },
+    {
+      "epoch": 0.15348440713163136,
+      "grad_norm": 560.4326782226562,
+      "learning_rate": 8.750671932380834e-06,
+      "loss": 49.6529,
+      "step": 37990
+    },
+    {
+      "epoch": 0.153524808396999,
+      "grad_norm": 843.6622924804688,
+      "learning_rate": 8.749739128004329e-06,
+      "loss": 34.5586,
+      "step": 38000
+    },
+    {
+      "epoch": 0.15356520966236661,
+      "grad_norm": 478.7063293457031,
+      "learning_rate": 8.748806025278292e-06,
+      "loss": 36.0544,
+      "step": 38010
+    },
+    {
+      "epoch": 0.15360561092773425,
+      "grad_norm": 455.8020324707031,
+      "learning_rate": 8.747872624276963e-06,
+      "loss": 47.236,
+      "step": 38020
+    },
+    {
+      "epoch": 0.1536460121931019,
+      "grad_norm": 915.6499633789062,
+      "learning_rate": 8.746938925074609e-06,
+      "loss": 45.514,
+      "step": 38030
+    },
+    {
+      "epoch": 0.1536864134584695,
+      "grad_norm": 931.0418701171875,
+      "learning_rate": 8.746004927745522e-06,
+      "loss": 44.9785,
+      "step": 38040
+    },
+    {
+      "epoch": 0.15372681472383715,
+      "grad_norm": 626.0626831054688,
+      "learning_rate": 8.745070632364014e-06,
+      "loss": 39.1456,
+      "step": 38050
+    },
+    {
+      "epoch": 0.1537672159892048,
+      "grad_norm": 537.35400390625,
+      "learning_rate": 8.744136039004422e-06,
+      "loss": 55.098,
+      "step": 38060
+    },
+    {
+      "epoch": 0.15380761725457243,
+      "grad_norm": 492.8907470703125,
+      "learning_rate": 8.743201147741112e-06,
+      "loss": 29.6499,
+      "step": 38070
+    },
+    {
+      "epoch": 0.15384801851994004,
+      "grad_norm": 386.3055114746094,
+      "learning_rate": 8.742265958648464e-06,
+      "loss": 55.5486,
+      "step": 38080
+    },
+    {
+      "epoch": 0.15388841978530768,
+      "grad_norm": 189.93057250976562,
+      "learning_rate": 8.741330471800888e-06,
+      "loss": 56.8159,
+      "step": 38090
+    },
+    {
+      "epoch": 0.15392882105067532,
+      "grad_norm": 689.0518188476562,
+      "learning_rate": 8.740394687272817e-06,
+      "loss": 57.4512,
+      "step": 38100
+    },
+    {
+      "epoch": 0.15396922231604293,
+      "grad_norm": 759.5892944335938,
+      "learning_rate": 8.739458605138706e-06,
+      "loss": 45.8897,
+      "step": 38110
+    },
+    {
+      "epoch": 0.15400962358141057,
+      "grad_norm": 470.87738037109375,
+      "learning_rate": 8.738522225473036e-06,
+      "loss": 57.6412,
+      "step": 38120
+    },
+    {
+      "epoch": 0.1540500248467782,
+      "grad_norm": 711.9561157226562,
+      "learning_rate": 8.737585548350312e-06,
+      "loss": 45.1267,
+      "step": 38130
+    },
+    {
+      "epoch": 0.15409042611214582,
+      "grad_norm": 525.4124755859375,
+      "learning_rate": 8.736648573845057e-06,
+      "loss": 47.2233,
+      "step": 38140
+    },
+    {
+      "epoch": 0.15413082737751346,
+      "grad_norm": 572.8314819335938,
+      "learning_rate": 8.735711302031824e-06,
+      "loss": 47.1612,
+      "step": 38150
+    },
+    {
+      "epoch": 0.1541712286428811,
+      "grad_norm": 466.9100341796875,
+      "learning_rate": 8.734773732985186e-06,
+      "loss": 41.9769,
+      "step": 38160
+    },
+    {
+      "epoch": 0.15421162990824872,
+      "grad_norm": 582.3335571289062,
+      "learning_rate": 8.733835866779745e-06,
+      "loss": 53.4043,
+      "step": 38170
+    },
+    {
+      "epoch": 0.15425203117361636,
+      "grad_norm": 561.9642944335938,
+      "learning_rate": 8.73289770349012e-06,
+      "loss": 38.1644,
+      "step": 38180
+    },
+    {
+      "epoch": 0.154292432438984,
+      "grad_norm": 542.4348754882812,
+      "learning_rate": 8.731959243190955e-06,
+      "loss": 42.0566,
+      "step": 38190
+    },
+    {
+      "epoch": 0.1543328337043516,
+      "grad_norm": 735.38330078125,
+      "learning_rate": 8.73102048595692e-06,
+      "loss": 31.1532,
+      "step": 38200
+    },
+    {
+      "epoch": 0.15437323496971925,
+      "grad_norm": 585.3547973632812,
+      "learning_rate": 8.730081431862709e-06,
+      "loss": 49.3706,
+      "step": 38210
+    },
+    {
+      "epoch": 0.1544136362350869,
+      "grad_norm": 439.5715637207031,
+      "learning_rate": 8.729142080983037e-06,
+      "loss": 43.1502,
+      "step": 38220
+    },
+    {
+      "epoch": 0.15445403750045453,
+      "grad_norm": 453.2962646484375,
+      "learning_rate": 8.728202433392645e-06,
+      "loss": 43.9217,
+      "step": 38230
+    },
+    {
+      "epoch": 0.15449443876582214,
+      "grad_norm": 836.9139404296875,
+      "learning_rate": 8.727262489166295e-06,
+      "loss": 78.5818,
+      "step": 38240
+    },
+    {
+      "epoch": 0.15453484003118978,
+      "grad_norm": 1474.2198486328125,
+      "learning_rate": 8.726322248378775e-06,
+      "loss": 72.0549,
+      "step": 38250
+    },
+    {
+      "epoch": 0.15457524129655742,
+      "grad_norm": 603.5780029296875,
+      "learning_rate": 8.725381711104894e-06,
+      "loss": 52.8777,
+      "step": 38260
+    },
+    {
+      "epoch": 0.15461564256192503,
+      "grad_norm": 850.0660400390625,
+      "learning_rate": 8.724440877419487e-06,
+      "loss": 76.4468,
+      "step": 38270
+    },
+    {
+      "epoch": 0.15465604382729267,
+      "grad_norm": 1414.240478515625,
+      "learning_rate": 8.723499747397415e-06,
+      "loss": 46.7855,
+      "step": 38280
+    },
+    {
+      "epoch": 0.1546964450926603,
+      "grad_norm": 704.088134765625,
+      "learning_rate": 8.722558321113555e-06,
+      "loss": 59.9468,
+      "step": 38290
+    },
+    {
+      "epoch": 0.15473684635802792,
+      "grad_norm": 369.7774963378906,
+      "learning_rate": 8.721616598642812e-06,
+      "loss": 39.3916,
+      "step": 38300
+    },
+    {
+      "epoch": 0.15477724762339556,
+      "grad_norm": 648.994384765625,
+      "learning_rate": 8.720674580060117e-06,
+      "loss": 48.405,
+      "step": 38310
+    },
+    {
+      "epoch": 0.1548176488887632,
+      "grad_norm": 367.8060607910156,
+      "learning_rate": 8.719732265440423e-06,
+      "loss": 65.3975,
+      "step": 38320
+    },
+    {
+      "epoch": 0.15485805015413082,
+      "grad_norm": 682.9230346679688,
+      "learning_rate": 8.718789654858702e-06,
+      "loss": 41.5378,
+      "step": 38330
+    },
+    {
+      "epoch": 0.15489845141949846,
+      "grad_norm": 341.1941223144531,
+      "learning_rate": 8.717846748389956e-06,
+      "loss": 40.6628,
+      "step": 38340
+    },
+    {
+      "epoch": 0.1549388526848661,
+      "grad_norm": 508.8382263183594,
+      "learning_rate": 8.716903546109208e-06,
+      "loss": 38.0144,
+      "step": 38350
+    },
+    {
+      "epoch": 0.1549792539502337,
+      "grad_norm": 480.63970947265625,
+      "learning_rate": 8.715960048091502e-06,
+      "loss": 45.1181,
+      "step": 38360
+    },
+    {
+      "epoch": 0.15501965521560135,
+      "grad_norm": 730.0436401367188,
+      "learning_rate": 8.715016254411908e-06,
+      "loss": 43.1608,
+      "step": 38370
+    },
+    {
+      "epoch": 0.155060056480969,
+      "grad_norm": 744.2866821289062,
+      "learning_rate": 8.714072165145521e-06,
+      "loss": 55.8971,
+      "step": 38380
+    },
+    {
+      "epoch": 0.15510045774633663,
+      "grad_norm": 422.5709533691406,
+      "learning_rate": 8.713127780367458e-06,
+      "loss": 47.1621,
+      "step": 38390
+    },
+    {
+      "epoch": 0.15514085901170424,
+      "grad_norm": 428.45452880859375,
+      "learning_rate": 8.712183100152858e-06,
+      "loss": 42.7035,
+      "step": 38400
+    },
+    {
+      "epoch": 0.15518126027707188,
+      "grad_norm": 491.593505859375,
+      "learning_rate": 8.711238124576884e-06,
+      "loss": 93.1383,
+      "step": 38410
+    },
+    {
+      "epoch": 0.15522166154243952,
+      "grad_norm": 1070.5885009765625,
+      "learning_rate": 8.710292853714726e-06,
+      "loss": 30.7639,
+      "step": 38420
+    },
+    {
+      "epoch": 0.15526206280780713,
+      "grad_norm": 1554.32666015625,
+      "learning_rate": 8.709347287641593e-06,
+      "loss": 53.9866,
+      "step": 38430
+    },
+    {
+      "epoch": 0.15530246407317477,
+      "grad_norm": 472.5458068847656,
+      "learning_rate": 8.70840142643272e-06,
+      "loss": 55.8533,
+      "step": 38440
+    },
+    {
+      "epoch": 0.1553428653385424,
+      "grad_norm": 448.1315612792969,
+      "learning_rate": 8.707455270163365e-06,
+      "loss": 63.6653,
+      "step": 38450
+    },
+    {
+      "epoch": 0.15538326660391003,
+      "grad_norm": 704.42138671875,
+      "learning_rate": 8.70650881890881e-06,
+      "loss": 42.9613,
+      "step": 38460
+    },
+    {
+      "epoch": 0.15542366786927767,
+      "grad_norm": 246.81031799316406,
+      "learning_rate": 8.705562072744358e-06,
+      "loss": 47.2647,
+      "step": 38470
+    },
+    {
+      "epoch": 0.1554640691346453,
+      "grad_norm": 652.4033203125,
+      "learning_rate": 8.704615031745337e-06,
+      "loss": 66.3134,
+      "step": 38480
+    },
+    {
+      "epoch": 0.15550447040001292,
+      "grad_norm": 790.2584838867188,
+      "learning_rate": 8.703667695987102e-06,
+      "loss": 58.0497,
+      "step": 38490
+    },
+    {
+      "epoch": 0.15554487166538056,
+      "grad_norm": 718.5376586914062,
+      "learning_rate": 8.702720065545024e-06,
+      "loss": 63.9357,
+      "step": 38500
+    },
+    {
+      "epoch": 0.1555852729307482,
+      "grad_norm": 754.3954467773438,
+      "learning_rate": 8.701772140494504e-06,
+      "loss": 62.0079,
+      "step": 38510
+    },
+    {
+      "epoch": 0.1556256741961158,
+      "grad_norm": 531.1658325195312,
+      "learning_rate": 8.700823920910964e-06,
+      "loss": 58.6769,
+      "step": 38520
+    },
+    {
+      "epoch": 0.15566607546148345,
+      "grad_norm": 1036.1522216796875,
+      "learning_rate": 8.699875406869848e-06,
+      "loss": 40.6459,
+      "step": 38530
+    },
+    {
+      "epoch": 0.1557064767268511,
+      "grad_norm": 411.3389587402344,
+      "learning_rate": 8.69892659844663e-06,
+      "loss": 58.2496,
+      "step": 38540
+    },
+    {
+      "epoch": 0.15574687799221873,
+      "grad_norm": 317.8341979980469,
+      "learning_rate": 8.697977495716794e-06,
+      "loss": 42.8371,
+      "step": 38550
+    },
+    {
+      "epoch": 0.15578727925758634,
+      "grad_norm": 837.8154907226562,
+      "learning_rate": 8.697028098755863e-06,
+      "loss": 47.0967,
+      "step": 38560
+    },
+    {
+      "epoch": 0.15582768052295398,
+      "grad_norm": 408.35845947265625,
+      "learning_rate": 8.69607840763937e-06,
+      "loss": 59.6451,
+      "step": 38570
+    },
+    {
+      "epoch": 0.15586808178832162,
+      "grad_norm": 421.7928771972656,
+      "learning_rate": 8.695128422442882e-06,
+      "loss": 37.1028,
+      "step": 38580
+    },
+    {
+      "epoch": 0.15590848305368923,
+      "grad_norm": 1320.11962890625,
+      "learning_rate": 8.694178143241984e-06,
+      "loss": 51.9049,
+      "step": 38590
+    },
+    {
+      "epoch": 0.15594888431905687,
+      "grad_norm": 702.5349731445312,
+      "learning_rate": 8.693227570112285e-06,
+      "loss": 59.5722,
+      "step": 38600
+    },
+    {
+      "epoch": 0.15598928558442451,
+      "grad_norm": 827.35888671875,
+      "learning_rate": 8.692276703129421e-06,
+      "loss": 55.1379,
+      "step": 38610
+    },
+    {
+      "epoch": 0.15602968684979213,
+      "grad_norm": 387.04266357421875,
+      "learning_rate": 8.691325542369041e-06,
+      "loss": 52.5193,
+      "step": 38620
+    },
+    {
+      "epoch": 0.15607008811515977,
+      "grad_norm": 271.24432373046875,
+      "learning_rate": 8.69037408790683e-06,
+      "loss": 55.1027,
+      "step": 38630
+    },
+    {
+      "epoch": 0.1561104893805274,
+      "grad_norm": 448.82666015625,
+      "learning_rate": 8.689422339818489e-06,
+      "loss": 38.2809,
+      "step": 38640
+    },
+    {
+      "epoch": 0.15615089064589502,
+      "grad_norm": 531.73828125,
+      "learning_rate": 8.688470298179746e-06,
+      "loss": 46.1902,
+      "step": 38650
+    },
+    {
+      "epoch": 0.15619129191126266,
+      "grad_norm": 103.58868408203125,
+      "learning_rate": 8.687517963066347e-06,
+      "loss": 37.2087,
+      "step": 38660
+    },
+    {
+      "epoch": 0.1562316931766303,
+      "grad_norm": 845.83251953125,
+      "learning_rate": 8.686565334554069e-06,
+      "loss": 67.7298,
+      "step": 38670
+    },
+    {
+      "epoch": 0.1562720944419979,
+      "grad_norm": 475.2627258300781,
+      "learning_rate": 8.685612412718704e-06,
+      "loss": 57.5004,
+      "step": 38680
+    },
+    {
+      "epoch": 0.15631249570736555,
+      "grad_norm": 812.0715942382812,
+      "learning_rate": 8.684659197636076e-06,
+      "loss": 46.925,
+      "step": 38690
+    },
+    {
+      "epoch": 0.1563528969727332,
+      "grad_norm": 531.0429077148438,
+      "learning_rate": 8.683705689382025e-06,
+      "loss": 59.5186,
+      "step": 38700
+    },
+    {
+      "epoch": 0.15639329823810083,
+      "grad_norm": 593.1111450195312,
+      "learning_rate": 8.682751888032419e-06,
+      "loss": 39.3045,
+      "step": 38710
+    },
+    {
+      "epoch": 0.15643369950346844,
+      "grad_norm": 720.40869140625,
+      "learning_rate": 8.681797793663147e-06,
+      "loss": 32.7965,
+      "step": 38720
+    },
+    {
+      "epoch": 0.15647410076883608,
+      "grad_norm": 1218.6812744140625,
+      "learning_rate": 8.680843406350122e-06,
+      "loss": 53.3814,
+      "step": 38730
+    },
+    {
+      "epoch": 0.15651450203420372,
+      "grad_norm": 1180.31884765625,
+      "learning_rate": 8.679888726169277e-06,
+      "loss": 56.8287,
+      "step": 38740
+    },
+    {
+      "epoch": 0.15655490329957134,
+      "grad_norm": 377.81390380859375,
+      "learning_rate": 8.678933753196577e-06,
+      "loss": 58.477,
+      "step": 38750
+    },
+    {
+      "epoch": 0.15659530456493898,
+      "grad_norm": 1002.2963256835938,
+      "learning_rate": 8.677978487508002e-06,
+      "loss": 44.0508,
+      "step": 38760
+    },
+    {
+      "epoch": 0.15663570583030662,
+      "grad_norm": 768.4991455078125,
+      "learning_rate": 8.677022929179558e-06,
+      "loss": 57.5769,
+      "step": 38770
+    },
+    {
+      "epoch": 0.15667610709567423,
+      "grad_norm": 1618.297607421875,
+      "learning_rate": 8.676067078287276e-06,
+      "loss": 39.0754,
+      "step": 38780
+    },
+    {
+      "epoch": 0.15671650836104187,
+      "grad_norm": 723.4669189453125,
+      "learning_rate": 8.675110934907206e-06,
+      "loss": 45.2615,
+      "step": 38790
+    },
+    {
+      "epoch": 0.1567569096264095,
+      "grad_norm": 397.9026184082031,
+      "learning_rate": 8.674154499115426e-06,
+      "loss": 37.6675,
+      "step": 38800
+    },
+    {
+      "epoch": 0.15679731089177712,
+      "grad_norm": 943.8479614257812,
+      "learning_rate": 8.673197770988034e-06,
+      "loss": 47.4713,
+      "step": 38810
+    },
+    {
+      "epoch": 0.15683771215714476,
+      "grad_norm": 940.014892578125,
+      "learning_rate": 8.672240750601152e-06,
+      "loss": 60.0122,
+      "step": 38820
+    },
+    {
+      "epoch": 0.1568781134225124,
+      "grad_norm": 605.8595581054688,
+      "learning_rate": 8.67128343803093e-06,
+      "loss": 48.2564,
+      "step": 38830
+    },
+    {
+      "epoch": 0.15691851468788,
+      "grad_norm": 613.501708984375,
+      "learning_rate": 8.670325833353532e-06,
+      "loss": 36.4502,
+      "step": 38840
+    },
+    {
+      "epoch": 0.15695891595324765,
+      "grad_norm": 357.8223876953125,
+      "learning_rate": 8.669367936645152e-06,
+      "loss": 45.7456,
+      "step": 38850
+    },
+    {
+      "epoch": 0.1569993172186153,
+      "grad_norm": 934.835693359375,
+      "learning_rate": 8.668409747982005e-06,
+      "loss": 43.5836,
+      "step": 38860
+    },
+    {
+      "epoch": 0.15703971848398293,
+      "grad_norm": 421.37188720703125,
+      "learning_rate": 8.667451267440332e-06,
+      "loss": 53.3812,
+      "step": 38870
+    },
+    {
+      "epoch": 0.15708011974935054,
+      "grad_norm": 176.40232849121094,
+      "learning_rate": 8.666492495096391e-06,
+      "loss": 49.2198,
+      "step": 38880
+    },
+    {
+      "epoch": 0.15712052101471818,
+      "grad_norm": 688.4175415039062,
+      "learning_rate": 8.66553343102647e-06,
+      "loss": 37.7146,
+      "step": 38890
+    },
+    {
+      "epoch": 0.15716092228008582,
+      "grad_norm": 498.5278015136719,
+      "learning_rate": 8.664574075306876e-06,
+      "loss": 37.7639,
+      "step": 38900
+    },
+    {
+      "epoch": 0.15720132354545344,
+      "grad_norm": 554.88232421875,
+      "learning_rate": 8.66361442801394e-06,
+      "loss": 41.1534,
+      "step": 38910
+    },
+    {
+      "epoch": 0.15724172481082108,
+      "grad_norm": 591.5156860351562,
+      "learning_rate": 8.662654489224018e-06,
+      "loss": 55.6884,
+      "step": 38920
+    },
+    {
+      "epoch": 0.15728212607618872,
+      "grad_norm": 915.4189453125,
+      "learning_rate": 8.661694259013489e-06,
+      "loss": 40.387,
+      "step": 38930
+    },
+    {
+      "epoch": 0.15732252734155633,
+      "grad_norm": 844.4616088867188,
+      "learning_rate": 8.660733737458751e-06,
+      "loss": 63.7767,
+      "step": 38940
+    },
+    {
+      "epoch": 0.15736292860692397,
+      "grad_norm": 492.8940124511719,
+      "learning_rate": 8.659772924636232e-06,
+      "loss": 48.9174,
+      "step": 38950
+    },
+    {
+      "epoch": 0.1574033298722916,
+      "grad_norm": 746.3994140625,
+      "learning_rate": 8.658811820622376e-06,
+      "loss": 42.9431,
+      "step": 38960
+    },
+    {
+      "epoch": 0.15744373113765922,
+      "grad_norm": 796.9805908203125,
+      "learning_rate": 8.657850425493656e-06,
+      "loss": 60.7916,
+      "step": 38970
+    },
+    {
+      "epoch": 0.15748413240302686,
+      "grad_norm": 392.3251647949219,
+      "learning_rate": 8.656888739326564e-06,
+      "loss": 51.2107,
+      "step": 38980
+    },
+    {
+      "epoch": 0.1575245336683945,
+      "grad_norm": 887.1876831054688,
+      "learning_rate": 8.65592676219762e-06,
+      "loss": 53.673,
+      "step": 38990
+    },
+    {
+      "epoch": 0.1575649349337621,
+      "grad_norm": 435.13690185546875,
+      "learning_rate": 8.65496449418336e-06,
+      "loss": 57.3104,
+      "step": 39000
+    },
+    {
+      "epoch": 0.15760533619912975,
+      "grad_norm": 769.52099609375,
+      "learning_rate": 8.654001935360349e-06,
+      "loss": 53.4194,
+      "step": 39010
+    },
+    {
+      "epoch": 0.1576457374644974,
+      "grad_norm": 691.1365356445312,
+      "learning_rate": 8.653039085805174e-06,
+      "loss": 28.5163,
+      "step": 39020
+    },
+    {
+      "epoch": 0.15768613872986503,
+      "grad_norm": 1318.233642578125,
+      "learning_rate": 8.652075945594444e-06,
+      "loss": 63.2581,
+      "step": 39030
+    },
+    {
+      "epoch": 0.15772653999523265,
+      "grad_norm": 338.80902099609375,
+      "learning_rate": 8.651112514804793e-06,
+      "loss": 51.1088,
+      "step": 39040
+    },
+    {
+      "epoch": 0.15776694126060029,
+      "grad_norm": 690.9937133789062,
+      "learning_rate": 8.650148793512874e-06,
+      "loss": 54.7363,
+      "step": 39050
+    },
+    {
+      "epoch": 0.15780734252596793,
+      "grad_norm": 672.234375,
+      "learning_rate": 8.649184781795367e-06,
+      "loss": 50.9707,
+      "step": 39060
+    },
+    {
+      "epoch": 0.15784774379133554,
+      "grad_norm": 353.51617431640625,
+      "learning_rate": 8.648220479728976e-06,
+      "loss": 67.0655,
+      "step": 39070
+    },
+    {
+      "epoch": 0.15788814505670318,
+      "grad_norm": 481.1822814941406,
+      "learning_rate": 8.647255887390425e-06,
+      "loss": 35.7537,
+      "step": 39080
+    },
+    {
+      "epoch": 0.15792854632207082,
+      "grad_norm": 845.3275756835938,
+      "learning_rate": 8.64629100485646e-06,
+      "loss": 74.3657,
+      "step": 39090
+    },
+    {
+      "epoch": 0.15796894758743843,
+      "grad_norm": 400.9262390136719,
+      "learning_rate": 8.645325832203855e-06,
+      "loss": 35.4863,
+      "step": 39100
+    },
+    {
+      "epoch": 0.15800934885280607,
+      "grad_norm": 611.3833618164062,
+      "learning_rate": 8.644360369509403e-06,
+      "loss": 50.1126,
+      "step": 39110
+    },
+    {
+      "epoch": 0.1580497501181737,
+      "grad_norm": 635.46142578125,
+      "learning_rate": 8.64339461684992e-06,
+      "loss": 50.9561,
+      "step": 39120
+    },
+    {
+      "epoch": 0.15809015138354132,
+      "grad_norm": 695.3146362304688,
+      "learning_rate": 8.64242857430225e-06,
+      "loss": 42.4911,
+      "step": 39130
+    },
+    {
+      "epoch": 0.15813055264890896,
+      "grad_norm": 320.0533142089844,
+      "learning_rate": 8.641462241943255e-06,
+      "loss": 42.3086,
+      "step": 39140
+    },
+    {
+      "epoch": 0.1581709539142766,
+      "grad_norm": 606.509521484375,
+      "learning_rate": 8.640495619849821e-06,
+      "loss": 45.3936,
+      "step": 39150
+    },
+    {
+      "epoch": 0.15821135517964421,
+      "grad_norm": 566.2433471679688,
+      "learning_rate": 8.639528708098858e-06,
+      "loss": 44.108,
+      "step": 39160
+    },
+    {
+      "epoch": 0.15825175644501185,
+      "grad_norm": 869.4197998046875,
+      "learning_rate": 8.6385615067673e-06,
+      "loss": 34.082,
+      "step": 39170
+    },
+    {
+      "epoch": 0.1582921577103795,
+      "grad_norm": 335.2037048339844,
+      "learning_rate": 8.6375940159321e-06,
+      "loss": 41.5822,
+      "step": 39180
+    },
+    {
+      "epoch": 0.15833255897574713,
+      "grad_norm": 629.8877563476562,
+      "learning_rate": 8.63662623567024e-06,
+      "loss": 31.4693,
+      "step": 39190
+    },
+    {
+      "epoch": 0.15837296024111475,
+      "grad_norm": 1557.7572021484375,
+      "learning_rate": 8.63565816605872e-06,
+      "loss": 41.352,
+      "step": 39200
+    },
+    {
+      "epoch": 0.1584133615064824,
+      "grad_norm": 666.0990600585938,
+      "learning_rate": 8.634689807174564e-06,
+      "loss": 47.8496,
+      "step": 39210
+    },
+    {
+      "epoch": 0.15845376277185003,
+      "grad_norm": 704.407958984375,
+      "learning_rate": 8.633721159094823e-06,
+      "loss": 46.6244,
+      "step": 39220
+    },
+    {
+      "epoch": 0.15849416403721764,
+      "grad_norm": 390.741455078125,
+      "learning_rate": 8.632752221896562e-06,
+      "loss": 45.0767,
+      "step": 39230
+    },
+    {
+      "epoch": 0.15853456530258528,
+      "grad_norm": 470.3848876953125,
+      "learning_rate": 8.631782995656884e-06,
+      "loss": 53.1645,
+      "step": 39240
+    },
+    {
+      "epoch": 0.15857496656795292,
+      "grad_norm": 357.0032043457031,
+      "learning_rate": 8.630813480452898e-06,
+      "loss": 34.2059,
+      "step": 39250
+    },
+    {
+      "epoch": 0.15861536783332053,
+      "grad_norm": 483.2372131347656,
+      "learning_rate": 8.629843676361747e-06,
+      "loss": 46.7471,
+      "step": 39260
+    },
+    {
+      "epoch": 0.15865576909868817,
+      "grad_norm": 889.9521484375,
+      "learning_rate": 8.628873583460593e-06,
+      "loss": 48.7918,
+      "step": 39270
+    },
+    {
+      "epoch": 0.1586961703640558,
+      "grad_norm": 338.2070007324219,
+      "learning_rate": 8.627903201826622e-06,
+      "loss": 45.2854,
+      "step": 39280
+    },
+    {
+      "epoch": 0.15873657162942342,
+      "grad_norm": 810.640380859375,
+      "learning_rate": 8.626932531537042e-06,
+      "loss": 32.9458,
+      "step": 39290
+    },
+    {
+      "epoch": 0.15877697289479106,
+      "grad_norm": 1532.687255859375,
+      "learning_rate": 8.625961572669087e-06,
+      "loss": 59.3802,
+      "step": 39300
+    },
+    {
+      "epoch": 0.1588173741601587,
+      "grad_norm": 657.8353271484375,
+      "learning_rate": 8.62499032530001e-06,
+      "loss": 37.4358,
+      "step": 39310
+    },
+    {
+      "epoch": 0.15885777542552632,
+      "grad_norm": 960.59521484375,
+      "learning_rate": 8.624018789507091e-06,
+      "loss": 40.7411,
+      "step": 39320
+    },
+    {
+      "epoch": 0.15889817669089396,
+      "grad_norm": 439.2926025390625,
+      "learning_rate": 8.62304696536763e-06,
+      "loss": 40.3107,
+      "step": 39330
+    },
+    {
+      "epoch": 0.1589385779562616,
+      "grad_norm": 1047.429931640625,
+      "learning_rate": 8.622074852958946e-06,
+      "loss": 44.9087,
+      "step": 39340
+    },
+    {
+      "epoch": 0.15897897922162924,
+      "grad_norm": 1334.8177490234375,
+      "learning_rate": 8.621102452358393e-06,
+      "loss": 64.5529,
+      "step": 39350
+    },
+    {
+      "epoch": 0.15901938048699685,
+      "grad_norm": 996.2518920898438,
+      "learning_rate": 8.620129763643333e-06,
+      "loss": 56.1161,
+      "step": 39360
+    },
+    {
+      "epoch": 0.1590597817523645,
+      "grad_norm": 610.1491088867188,
+      "learning_rate": 8.619156786891162e-06,
+      "loss": 60.381,
+      "step": 39370
+    },
+    {
+      "epoch": 0.15910018301773213,
+      "grad_norm": 994.11328125,
+      "learning_rate": 8.618183522179295e-06,
+      "loss": 42.3613,
+      "step": 39380
+    },
+    {
+      "epoch": 0.15914058428309974,
+      "grad_norm": 455.6463317871094,
+      "learning_rate": 8.617209969585171e-06,
+      "loss": 55.0342,
+      "step": 39390
+    },
+    {
+      "epoch": 0.15918098554846738,
+      "grad_norm": 661.022216796875,
+      "learning_rate": 8.616236129186252e-06,
+      "loss": 43.6494,
+      "step": 39400
+    },
+    {
+      "epoch": 0.15922138681383502,
+      "grad_norm": 139.87229919433594,
+      "learning_rate": 8.615262001060019e-06,
+      "loss": 48.6457,
+      "step": 39410
+    },
+    {
+      "epoch": 0.15926178807920263,
+      "grad_norm": 856.7398071289062,
+      "learning_rate": 8.61428758528398e-06,
+      "loss": 50.8592,
+      "step": 39420
+    },
+    {
+      "epoch": 0.15930218934457027,
+      "grad_norm": 848.9993896484375,
+      "learning_rate": 8.613312881935667e-06,
+      "loss": 69.9591,
+      "step": 39430
+    },
+    {
+      "epoch": 0.1593425906099379,
+      "grad_norm": 313.719482421875,
+      "learning_rate": 8.61233789109263e-06,
+      "loss": 43.3973,
+      "step": 39440
+    },
+    {
+      "epoch": 0.15938299187530552,
+      "grad_norm": 494.93194580078125,
+      "learning_rate": 8.611362612832445e-06,
+      "loss": 52.3588,
+      "step": 39450
+    },
+    {
+      "epoch": 0.15942339314067316,
+      "grad_norm": 493.6256103515625,
+      "learning_rate": 8.610387047232711e-06,
+      "loss": 68.9036,
+      "step": 39460
+    },
+    {
+      "epoch": 0.1594637944060408,
+      "grad_norm": 767.64208984375,
+      "learning_rate": 8.609411194371049e-06,
+      "loss": 58.1112,
+      "step": 39470
+    },
+    {
+      "epoch": 0.15950419567140842,
+      "grad_norm": 1218.093017578125,
+      "learning_rate": 8.608435054325103e-06,
+      "loss": 48.0329,
+      "step": 39480
+    },
+    {
+      "epoch": 0.15954459693677606,
+      "grad_norm": 615.8851318359375,
+      "learning_rate": 8.60745862717254e-06,
+      "loss": 44.5316,
+      "step": 39490
+    },
+    {
+      "epoch": 0.1595849982021437,
+      "grad_norm": 380.9693908691406,
+      "learning_rate": 8.606481912991052e-06,
+      "loss": 33.8241,
+      "step": 39500
+    },
+    {
+      "epoch": 0.1596253994675113,
+      "grad_norm": 859.0559692382812,
+      "learning_rate": 8.605504911858347e-06,
+      "loss": 44.5087,
+      "step": 39510
+    },
+    {
+      "epoch": 0.15966580073287895,
+      "grad_norm": 492.1607360839844,
+      "learning_rate": 8.604527623852165e-06,
+      "loss": 41.605,
+      "step": 39520
+    },
+    {
+      "epoch": 0.1597062019982466,
+      "grad_norm": 611.9906616210938,
+      "learning_rate": 8.603550049050262e-06,
+      "loss": 55.0675,
+      "step": 39530
+    },
+    {
+      "epoch": 0.15974660326361423,
+      "grad_norm": 469.29364013671875,
+      "learning_rate": 8.602572187530421e-06,
+      "loss": 44.7947,
+      "step": 39540
+    },
+    {
+      "epoch": 0.15978700452898184,
+      "grad_norm": 398.1039123535156,
+      "learning_rate": 8.601594039370441e-06,
+      "loss": 49.7298,
+      "step": 39550
+    },
+    {
+      "epoch": 0.15982740579434948,
+      "grad_norm": 601.5389404296875,
+      "learning_rate": 8.600615604648155e-06,
+      "loss": 50.9114,
+      "step": 39560
+    },
+    {
+      "epoch": 0.15986780705971712,
+      "grad_norm": 342.42767333984375,
+      "learning_rate": 8.599636883441408e-06,
+      "loss": 52.5872,
+      "step": 39570
+    },
+    {
+      "epoch": 0.15990820832508473,
+      "grad_norm": 857.3191528320312,
+      "learning_rate": 8.598657875828078e-06,
+      "loss": 77.0436,
+      "step": 39580
+    },
+    {
+      "epoch": 0.15994860959045237,
+      "grad_norm": 543.1047973632812,
+      "learning_rate": 8.597678581886055e-06,
+      "loss": 45.0541,
+      "step": 39590
+    },
+    {
+      "epoch": 0.15998901085582,
+      "grad_norm": 712.693115234375,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 42.4967,
+      "step": 39600
+    },
+    {
+      "epoch": 0.16002941212118763,
+      "grad_norm": 691.2177734375,
+      "learning_rate": 8.595719135327627e-06,
+      "loss": 50.3935,
+      "step": 39610
+    },
+    {
+      "epoch": 0.16006981338655527,
+      "grad_norm": 345.3578186035156,
+      "learning_rate": 8.594738982867126e-06,
+      "loss": 61.9182,
+      "step": 39620
+    },
+    {
+      "epoch": 0.1601102146519229,
+      "grad_norm": 494.4696960449219,
+      "learning_rate": 8.593758544389743e-06,
+      "loss": 57.9511,
+      "step": 39630
+    },
+    {
+      "epoch": 0.16015061591729052,
+      "grad_norm": 1123.7225341796875,
+      "learning_rate": 8.592777819973486e-06,
+      "loss": 59.5531,
+      "step": 39640
+    },
+    {
+      "epoch": 0.16019101718265816,
+      "grad_norm": 641.5360107421875,
+      "learning_rate": 8.591796809696386e-06,
+      "loss": 46.7559,
+      "step": 39650
+    },
+    {
+      "epoch": 0.1602314184480258,
+      "grad_norm": 862.7567138671875,
+      "learning_rate": 8.590815513636498e-06,
+      "loss": 47.3019,
+      "step": 39660
+    },
+    {
+      "epoch": 0.1602718197133934,
+      "grad_norm": 354.60186767578125,
+      "learning_rate": 8.5898339318719e-06,
+      "loss": 50.8752,
+      "step": 39670
+    },
+    {
+      "epoch": 0.16031222097876105,
+      "grad_norm": 875.8455810546875,
+      "learning_rate": 8.58885206448069e-06,
+      "loss": 41.4256,
+      "step": 39680
+    },
+    {
+      "epoch": 0.1603526222441287,
+      "grad_norm": 362.48443603515625,
+      "learning_rate": 8.587869911540993e-06,
+      "loss": 42.8521,
+      "step": 39690
+    },
+    {
+      "epoch": 0.16039302350949633,
+      "grad_norm": 679.5111694335938,
+      "learning_rate": 8.586887473130951e-06,
+      "loss": 34.4303,
+      "step": 39700
+    },
+    {
+      "epoch": 0.16043342477486394,
+      "grad_norm": 457.1158447265625,
+      "learning_rate": 8.585904749328736e-06,
+      "loss": 33.436,
+      "step": 39710
+    },
+    {
+      "epoch": 0.16047382604023158,
+      "grad_norm": 583.4326782226562,
+      "learning_rate": 8.584921740212537e-06,
+      "loss": 87.6192,
+      "step": 39720
+    },
+    {
+      "epoch": 0.16051422730559922,
+      "grad_norm": 542.652587890625,
+      "learning_rate": 8.583938445860569e-06,
+      "loss": 42.0307,
+      "step": 39730
+    },
+    {
+      "epoch": 0.16055462857096683,
+      "grad_norm": 989.8584594726562,
+      "learning_rate": 8.582954866351065e-06,
+      "loss": 49.7806,
+      "step": 39740
+    },
+    {
+      "epoch": 0.16059502983633447,
+      "grad_norm": 821.0096435546875,
+      "learning_rate": 8.581971001762287e-06,
+      "loss": 69.7385,
+      "step": 39750
+    },
+    {
+      "epoch": 0.16063543110170211,
+      "grad_norm": 461.29443359375,
+      "learning_rate": 8.580986852172514e-06,
+      "loss": 41.6745,
+      "step": 39760
+    },
+    {
+      "epoch": 0.16067583236706973,
+      "grad_norm": 520.6972045898438,
+      "learning_rate": 8.580002417660054e-06,
+      "loss": 64.0036,
+      "step": 39770
+    },
+    {
+      "epoch": 0.16071623363243737,
+      "grad_norm": 730.748046875,
+      "learning_rate": 8.579017698303228e-06,
+      "loss": 63.5609,
+      "step": 39780
+    },
+    {
+      "epoch": 0.160756634897805,
+      "grad_norm": 1039.7550048828125,
+      "learning_rate": 8.578032694180394e-06,
+      "loss": 39.672,
+      "step": 39790
+    },
+    {
+      "epoch": 0.16079703616317262,
+      "grad_norm": 965.9579467773438,
+      "learning_rate": 8.577047405369916e-06,
+      "loss": 39.7766,
+      "step": 39800
+    },
+    {
+      "epoch": 0.16083743742854026,
+      "grad_norm": 559.5202026367188,
+      "learning_rate": 8.576061831950193e-06,
+      "loss": 45.9991,
+      "step": 39810
+    },
+    {
+      "epoch": 0.1608778386939079,
+      "grad_norm": 724.7987060546875,
+      "learning_rate": 8.575075973999642e-06,
+      "loss": 48.9104,
+      "step": 39820
+    },
+    {
+      "epoch": 0.1609182399592755,
+      "grad_norm": 732.09326171875,
+      "learning_rate": 8.574089831596703e-06,
+      "loss": 72.9903,
+      "step": 39830
+    },
+    {
+      "epoch": 0.16095864122464315,
+      "grad_norm": 351.67486572265625,
+      "learning_rate": 8.57310340481984e-06,
+      "loss": 53.9298,
+      "step": 39840
+    },
+    {
+      "epoch": 0.1609990424900108,
+      "grad_norm": 646.2775268554688,
+      "learning_rate": 8.572116693747537e-06,
+      "loss": 30.1154,
+      "step": 39850
+    },
+    {
+      "epoch": 0.16103944375537843,
+      "grad_norm": 576.468017578125,
+      "learning_rate": 8.571129698458302e-06,
+      "loss": 40.7335,
+      "step": 39860
+    },
+    {
+      "epoch": 0.16107984502074604,
+      "grad_norm": 874.0496826171875,
+      "learning_rate": 8.570142419030668e-06,
+      "loss": 60.7948,
+      "step": 39870
+    },
+    {
+      "epoch": 0.16112024628611368,
+      "grad_norm": 466.7607727050781,
+      "learning_rate": 8.569154855543184e-06,
+      "loss": 44.0453,
+      "step": 39880
+    },
+    {
+      "epoch": 0.16116064755148132,
+      "grad_norm": 450.4464111328125,
+      "learning_rate": 8.56816700807443e-06,
+      "loss": 52.5402,
+      "step": 39890
+    },
+    {
+      "epoch": 0.16120104881684894,
+      "grad_norm": 416.64697265625,
+      "learning_rate": 8.567178876703002e-06,
+      "loss": 50.5467,
+      "step": 39900
+    },
+    {
+      "epoch": 0.16124145008221658,
+      "grad_norm": 388.1261901855469,
+      "learning_rate": 8.566190461507521e-06,
+      "loss": 57.1608,
+      "step": 39910
+    },
+    {
+      "epoch": 0.16128185134758422,
+      "grad_norm": 348.6123046875,
+      "learning_rate": 8.565201762566632e-06,
+      "loss": 46.7085,
+      "step": 39920
+    },
+    {
+      "epoch": 0.16132225261295183,
+      "grad_norm": 862.7435913085938,
+      "learning_rate": 8.564212779959003e-06,
+      "loss": 67.4059,
+      "step": 39930
+    },
+    {
+      "epoch": 0.16136265387831947,
+      "grad_norm": 422.3407287597656,
+      "learning_rate": 8.563223513763319e-06,
+      "loss": 46.8044,
+      "step": 39940
+    },
+    {
+      "epoch": 0.1614030551436871,
+      "grad_norm": 502.47418212890625,
+      "learning_rate": 8.562233964058294e-06,
+      "loss": 53.1609,
+      "step": 39950
+    },
+    {
+      "epoch": 0.16144345640905472,
+      "grad_norm": 851.7847290039062,
+      "learning_rate": 8.561244130922658e-06,
+      "loss": 69.4814,
+      "step": 39960
+    },
+    {
+      "epoch": 0.16148385767442236,
+      "grad_norm": 962.0199584960938,
+      "learning_rate": 8.560254014435172e-06,
+      "loss": 62.8515,
+      "step": 39970
+    },
+    {
+      "epoch": 0.16152425893979,
+      "grad_norm": 794.7582397460938,
+      "learning_rate": 8.559263614674615e-06,
+      "loss": 62.2661,
+      "step": 39980
+    },
+    {
+      "epoch": 0.1615646602051576,
+      "grad_norm": 484.3645324707031,
+      "learning_rate": 8.558272931719785e-06,
+      "loss": 62.7066,
+      "step": 39990
+    },
+    {
+      "epoch": 0.16160506147052525,
+      "grad_norm": 574.9622802734375,
+      "learning_rate": 8.557281965649508e-06,
+      "loss": 43.5384,
+      "step": 40000
+    },
+    {
+      "epoch": 0.1616454627358929,
+      "grad_norm": 381.67010498046875,
+      "learning_rate": 8.556290716542632e-06,
+      "loss": 43.6,
+      "step": 40010
+    },
+    {
+      "epoch": 0.16168586400126053,
+      "grad_norm": 632.470947265625,
+      "learning_rate": 8.555299184478026e-06,
+      "loss": 45.3281,
+      "step": 40020
+    },
+    {
+      "epoch": 0.16172626526662814,
+      "grad_norm": 614.8099975585938,
+      "learning_rate": 8.554307369534577e-06,
+      "loss": 41.8744,
+      "step": 40030
+    },
+    {
+      "epoch": 0.16176666653199578,
+      "grad_norm": 612.2963256835938,
+      "learning_rate": 8.553315271791207e-06,
+      "loss": 46.7159,
+      "step": 40040
+    },
+    {
+      "epoch": 0.16180706779736342,
+      "grad_norm": 912.1803588867188,
+      "learning_rate": 8.552322891326846e-06,
+      "loss": 61.35,
+      "step": 40050
+    },
+    {
+      "epoch": 0.16184746906273104,
+      "grad_norm": 866.2830810546875,
+      "learning_rate": 8.551330228220454e-06,
+      "loss": 44.3065,
+      "step": 40060
+    },
+    {
+      "epoch": 0.16188787032809868,
+      "grad_norm": 272.114013671875,
+      "learning_rate": 8.550337282551016e-06,
+      "loss": 36.4487,
+      "step": 40070
+    },
+    {
+      "epoch": 0.16192827159346632,
+      "grad_norm": 1175.4171142578125,
+      "learning_rate": 8.549344054397533e-06,
+      "loss": 50.1568,
+      "step": 40080
+    },
+    {
+      "epoch": 0.16196867285883393,
+      "grad_norm": 626.5860595703125,
+      "learning_rate": 8.548350543839034e-06,
+      "loss": 43.3123,
+      "step": 40090
+    },
+    {
+      "epoch": 0.16200907412420157,
+      "grad_norm": 618.5504760742188,
+      "learning_rate": 8.547356750954568e-06,
+      "loss": 39.5396,
+      "step": 40100
+    },
+    {
+      "epoch": 0.1620494753895692,
+      "grad_norm": 537.752685546875,
+      "learning_rate": 8.546362675823204e-06,
+      "loss": 45.1189,
+      "step": 40110
+    },
+    {
+      "epoch": 0.16208987665493682,
+      "grad_norm": 390.0348205566406,
+      "learning_rate": 8.545368318524036e-06,
+      "loss": 50.7311,
+      "step": 40120
+    },
+    {
+      "epoch": 0.16213027792030446,
+      "grad_norm": 645.032958984375,
+      "learning_rate": 8.544373679136184e-06,
+      "loss": 60.2118,
+      "step": 40130
+    },
+    {
+      "epoch": 0.1621706791856721,
+      "grad_norm": 706.8973388671875,
+      "learning_rate": 8.543378757738785e-06,
+      "loss": 86.3639,
+      "step": 40140
+    },
+    {
+      "epoch": 0.1622110804510397,
+      "grad_norm": 612.7260131835938,
+      "learning_rate": 8.542383554411e-06,
+      "loss": 48.4019,
+      "step": 40150
+    },
+    {
+      "epoch": 0.16225148171640735,
+      "grad_norm": 763.453857421875,
+      "learning_rate": 8.541388069232012e-06,
+      "loss": 62.7933,
+      "step": 40160
+    },
+    {
+      "epoch": 0.162291882981775,
+      "grad_norm": 622.7904663085938,
+      "learning_rate": 8.54039230228103e-06,
+      "loss": 66.0159,
+      "step": 40170
+    },
+    {
+      "epoch": 0.16233228424714263,
+      "grad_norm": 405.7510070800781,
+      "learning_rate": 8.53939625363728e-06,
+      "loss": 47.5229,
+      "step": 40180
+    },
+    {
+      "epoch": 0.16237268551251025,
+      "grad_norm": 812.6790771484375,
+      "learning_rate": 8.538399923380011e-06,
+      "loss": 49.0495,
+      "step": 40190
+    },
+    {
+      "epoch": 0.16241308677787789,
+      "grad_norm": 485.32135009765625,
+      "learning_rate": 8.537403311588502e-06,
+      "loss": 53.9039,
+      "step": 40200
+    },
+    {
+      "epoch": 0.16245348804324553,
+      "grad_norm": 724.760986328125,
+      "learning_rate": 8.536406418342044e-06,
+      "loss": 56.8525,
+      "step": 40210
+    },
+    {
+      "epoch": 0.16249388930861314,
+      "grad_norm": 400.77783203125,
+      "learning_rate": 8.53540924371996e-06,
+      "loss": 46.7015,
+      "step": 40220
+    },
+    {
+      "epoch": 0.16253429057398078,
+      "grad_norm": 319.41314697265625,
+      "learning_rate": 8.534411787801586e-06,
+      "loss": 40.1834,
+      "step": 40230
+    },
+    {
+      "epoch": 0.16257469183934842,
+      "grad_norm": 486.13262939453125,
+      "learning_rate": 8.533414050666287e-06,
+      "loss": 47.1607,
+      "step": 40240
+    },
+    {
+      "epoch": 0.16261509310471603,
+      "grad_norm": 579.2314453125,
+      "learning_rate": 8.532416032393447e-06,
+      "loss": 34.5006,
+      "step": 40250
+    },
+    {
+      "epoch": 0.16265549437008367,
+      "grad_norm": 368.7862548828125,
+      "learning_rate": 8.531417733062476e-06,
+      "loss": 34.5283,
+      "step": 40260
+    },
+    {
+      "epoch": 0.1626958956354513,
+      "grad_norm": 631.860595703125,
+      "learning_rate": 8.530419152752804e-06,
+      "loss": 80.3774,
+      "step": 40270
+    },
+    {
+      "epoch": 0.16273629690081892,
+      "grad_norm": 405.42877197265625,
+      "learning_rate": 8.529420291543882e-06,
+      "loss": 51.1724,
+      "step": 40280
+    },
+    {
+      "epoch": 0.16277669816618656,
+      "grad_norm": 282.8262634277344,
+      "learning_rate": 8.528421149515185e-06,
+      "loss": 41.0309,
+      "step": 40290
+    },
+    {
+      "epoch": 0.1628170994315542,
+      "grad_norm": 566.5156860351562,
+      "learning_rate": 8.52742172674621e-06,
+      "loss": 40.4973,
+      "step": 40300
+    },
+    {
+      "epoch": 0.16285750069692181,
+      "grad_norm": 911.9764404296875,
+      "learning_rate": 8.526422023316478e-06,
+      "loss": 55.5752,
+      "step": 40310
+    },
+    {
+      "epoch": 0.16289790196228945,
+      "grad_norm": 1154.181640625,
+      "learning_rate": 8.525422039305529e-06,
+      "loss": 44.3412,
+      "step": 40320
+    },
+    {
+      "epoch": 0.1629383032276571,
+      "grad_norm": 526.5676879882812,
+      "learning_rate": 8.524421774792926e-06,
+      "loss": 79.0259,
+      "step": 40330
+    },
+    {
+      "epoch": 0.16297870449302473,
+      "grad_norm": 1047.71875,
+      "learning_rate": 8.52342122985826e-06,
+      "loss": 59.4561,
+      "step": 40340
+    },
+    {
+      "epoch": 0.16301910575839235,
+      "grad_norm": 636.7918701171875,
+      "learning_rate": 8.522420404581135e-06,
+      "loss": 32.4132,
+      "step": 40350
+    },
+    {
+      "epoch": 0.16305950702376,
+      "grad_norm": 505.8265686035156,
+      "learning_rate": 8.521419299041185e-06,
+      "loss": 56.5309,
+      "step": 40360
+    },
+    {
+      "epoch": 0.16309990828912763,
+      "grad_norm": 757.0598754882812,
+      "learning_rate": 8.520417913318065e-06,
+      "loss": 36.9938,
+      "step": 40370
+    },
+    {
+      "epoch": 0.16314030955449524,
+      "grad_norm": 277.91778564453125,
+      "learning_rate": 8.519416247491445e-06,
+      "loss": 67.4497,
+      "step": 40380
+    },
+    {
+      "epoch": 0.16318071081986288,
+      "grad_norm": 335.56793212890625,
+      "learning_rate": 8.518414301641027e-06,
+      "loss": 44.4462,
+      "step": 40390
+    },
+    {
+      "epoch": 0.16322111208523052,
+      "grad_norm": 603.6694946289062,
+      "learning_rate": 8.517412075846529e-06,
+      "loss": 73.2141,
+      "step": 40400
+    },
+    {
+      "epoch": 0.16326151335059813,
+      "grad_norm": 638.0531005859375,
+      "learning_rate": 8.516409570187698e-06,
+      "loss": 34.8373,
+      "step": 40410
+    },
+    {
+      "epoch": 0.16330191461596577,
+      "grad_norm": 764.2570190429688,
+      "learning_rate": 8.515406784744294e-06,
+      "loss": 60.3965,
+      "step": 40420
+    },
+    {
+      "epoch": 0.1633423158813334,
+      "grad_norm": 740.9988403320312,
+      "learning_rate": 8.514403719596104e-06,
+      "loss": 46.767,
+      "step": 40430
+    },
+    {
+      "epoch": 0.16338271714670102,
+      "grad_norm": 835.2094116210938,
+      "learning_rate": 8.513400374822942e-06,
+      "loss": 68.2253,
+      "step": 40440
+    },
+    {
+      "epoch": 0.16342311841206866,
+      "grad_norm": 489.5925598144531,
+      "learning_rate": 8.512396750504635e-06,
+      "loss": 44.3674,
+      "step": 40450
+    },
+    {
+      "epoch": 0.1634635196774363,
+      "grad_norm": 527.5684204101562,
+      "learning_rate": 8.511392846721037e-06,
+      "loss": 38.6535,
+      "step": 40460
+    },
+    {
+      "epoch": 0.16350392094280392,
+      "grad_norm": 665.7122802734375,
+      "learning_rate": 8.510388663552027e-06,
+      "loss": 43.2655,
+      "step": 40470
+    },
+    {
+      "epoch": 0.16354432220817156,
+      "grad_norm": 463.0517883300781,
+      "learning_rate": 8.509384201077502e-06,
+      "loss": 51.3164,
+      "step": 40480
+    },
+    {
+      "epoch": 0.1635847234735392,
+      "grad_norm": 660.7234497070312,
+      "learning_rate": 8.508379459377381e-06,
+      "loss": 46.4135,
+      "step": 40490
+    },
+    {
+      "epoch": 0.16362512473890684,
+      "grad_norm": 2250.38330078125,
+      "learning_rate": 8.507374438531606e-06,
+      "loss": 85.4677,
+      "step": 40500
+    },
+    {
+      "epoch": 0.16366552600427445,
+      "grad_norm": 371.8035888671875,
+      "learning_rate": 8.506369138620148e-06,
+      "loss": 23.6598,
+      "step": 40510
+    },
+    {
+      "epoch": 0.1637059272696421,
+      "grad_norm": 377.6024475097656,
+      "learning_rate": 8.505363559722985e-06,
+      "loss": 60.3731,
+      "step": 40520
+    },
+    {
+      "epoch": 0.16374632853500973,
+      "grad_norm": 391.7501525878906,
+      "learning_rate": 8.504357701920134e-06,
+      "loss": 57.3929,
+      "step": 40530
+    },
+    {
+      "epoch": 0.16378672980037734,
+      "grad_norm": 557.2824096679688,
+      "learning_rate": 8.503351565291622e-06,
+      "loss": 35.8355,
+      "step": 40540
+    },
+    {
+      "epoch": 0.16382713106574498,
+      "grad_norm": 527.2345581054688,
+      "learning_rate": 8.502345149917506e-06,
+      "loss": 64.468,
+      "step": 40550
+    },
+    {
+      "epoch": 0.16386753233111262,
+      "grad_norm": 543.4524536132812,
+      "learning_rate": 8.501338455877859e-06,
+      "loss": 55.0868,
+      "step": 40560
+    },
+    {
+      "epoch": 0.16390793359648023,
+      "grad_norm": 664.7778930664062,
+      "learning_rate": 8.50033148325278e-06,
+      "loss": 45.5921,
+      "step": 40570
+    },
+    {
+      "epoch": 0.16394833486184787,
+      "grad_norm": 690.2338256835938,
+      "learning_rate": 8.499324232122389e-06,
+      "loss": 39.4778,
+      "step": 40580
+    },
+    {
+      "epoch": 0.1639887361272155,
+      "grad_norm": 1139.7445068359375,
+      "learning_rate": 8.498316702566828e-06,
+      "loss": 67.1272,
+      "step": 40590
+    },
+    {
+      "epoch": 0.16402913739258312,
+      "grad_norm": 1078.213623046875,
+      "learning_rate": 8.497308894666263e-06,
+      "loss": 47.1635,
+      "step": 40600
+    },
+    {
+      "epoch": 0.16406953865795076,
+      "grad_norm": 528.3421020507812,
+      "learning_rate": 8.496300808500878e-06,
+      "loss": 54.4817,
+      "step": 40610
+    },
+    {
+      "epoch": 0.1641099399233184,
+      "grad_norm": 1129.3258056640625,
+      "learning_rate": 8.495292444150887e-06,
+      "loss": 49.5488,
+      "step": 40620
+    },
+    {
+      "epoch": 0.16415034118868602,
+      "grad_norm": 995.45361328125,
+      "learning_rate": 8.494283801696514e-06,
+      "loss": 90.1325,
+      "step": 40630
+    },
+    {
+      "epoch": 0.16419074245405366,
+      "grad_norm": 287.7896423339844,
+      "learning_rate": 8.493274881218017e-06,
+      "loss": 50.7454,
+      "step": 40640
+    },
+    {
+      "epoch": 0.1642311437194213,
+      "grad_norm": 800.5263671875,
+      "learning_rate": 8.49226568279567e-06,
+      "loss": 72.1485,
+      "step": 40650
+    },
+    {
+      "epoch": 0.16427154498478894,
+      "grad_norm": 760.0538940429688,
+      "learning_rate": 8.49125620650977e-06,
+      "loss": 55.6863,
+      "step": 40660
+    },
+    {
+      "epoch": 0.16431194625015655,
+      "grad_norm": 778.16357421875,
+      "learning_rate": 8.490246452440636e-06,
+      "loss": 34.3954,
+      "step": 40670
+    },
+    {
+      "epoch": 0.1643523475155242,
+      "grad_norm": 500.4969482421875,
+      "learning_rate": 8.48923642066861e-06,
+      "loss": 48.982,
+      "step": 40680
+    },
+    {
+      "epoch": 0.16439274878089183,
+      "grad_norm": 604.9345092773438,
+      "learning_rate": 8.488226111274055e-06,
+      "loss": 50.3341,
+      "step": 40690
+    },
+    {
+      "epoch": 0.16443315004625944,
+      "grad_norm": 712.126708984375,
+      "learning_rate": 8.487215524337357e-06,
+      "loss": 57.6024,
+      "step": 40700
+    },
+    {
+      "epoch": 0.16447355131162708,
+      "grad_norm": 1261.656494140625,
+      "learning_rate": 8.486204659938924e-06,
+      "loss": 52.6965,
+      "step": 40710
+    },
+    {
+      "epoch": 0.16451395257699472,
+      "grad_norm": 830.56591796875,
+      "learning_rate": 8.485193518159186e-06,
+      "loss": 45.7652,
+      "step": 40720
+    },
+    {
+      "epoch": 0.16455435384236233,
+      "grad_norm": 352.6490478515625,
+      "learning_rate": 8.484182099078596e-06,
+      "loss": 67.0111,
+      "step": 40730
+    },
+    {
+      "epoch": 0.16459475510772997,
+      "grad_norm": 185.65826416015625,
+      "learning_rate": 8.483170402777624e-06,
+      "loss": 67.0452,
+      "step": 40740
+    },
+    {
+      "epoch": 0.1646351563730976,
+      "grad_norm": 894.4474487304688,
+      "learning_rate": 8.482158429336769e-06,
+      "loss": 41.1207,
+      "step": 40750
+    },
+    {
+      "epoch": 0.16467555763846523,
+      "grad_norm": 683.1304321289062,
+      "learning_rate": 8.48114617883655e-06,
+      "loss": 51.1379,
+      "step": 40760
+    },
+    {
+      "epoch": 0.16471595890383287,
+      "grad_norm": 625.189453125,
+      "learning_rate": 8.480133651357507e-06,
+      "loss": 39.9019,
+      "step": 40770
+    },
+    {
+      "epoch": 0.1647563601692005,
+      "grad_norm": 395.3323059082031,
+      "learning_rate": 8.479120846980197e-06,
+      "loss": 45.6717,
+      "step": 40780
+    },
+    {
+      "epoch": 0.16479676143456812,
+      "grad_norm": 786.5029907226562,
+      "learning_rate": 8.478107765785212e-06,
+      "loss": 45.9822,
+      "step": 40790
+    },
+    {
+      "epoch": 0.16483716269993576,
+      "grad_norm": 693.7678833007812,
+      "learning_rate": 8.477094407853153e-06,
+      "loss": 32.4612,
+      "step": 40800
+    },
+    {
+      "epoch": 0.1648775639653034,
+      "grad_norm": 445.7911071777344,
+      "learning_rate": 8.47608077326465e-06,
+      "loss": 71.7955,
+      "step": 40810
+    },
+    {
+      "epoch": 0.16491796523067104,
+      "grad_norm": 655.4782104492188,
+      "learning_rate": 8.475066862100352e-06,
+      "loss": 56.675,
+      "step": 40820
+    },
+    {
+      "epoch": 0.16495836649603865,
+      "grad_norm": 741.1582641601562,
+      "learning_rate": 8.474052674440934e-06,
+      "loss": 60.2743,
+      "step": 40830
+    },
+    {
+      "epoch": 0.1649987677614063,
+      "grad_norm": 646.4039916992188,
+      "learning_rate": 8.473038210367086e-06,
+      "loss": 48.6313,
+      "step": 40840
+    },
+    {
+      "epoch": 0.16503916902677393,
+      "grad_norm": 405.1526794433594,
+      "learning_rate": 8.47202346995953e-06,
+      "loss": 29.084,
+      "step": 40850
+    },
+    {
+      "epoch": 0.16507957029214154,
+      "grad_norm": 559.633544921875,
+      "learning_rate": 8.471008453298998e-06,
+      "loss": 48.7567,
+      "step": 40860
+    },
+    {
+      "epoch": 0.16511997155750918,
+      "grad_norm": 612.960205078125,
+      "learning_rate": 8.469993160466254e-06,
+      "loss": 33.431,
+      "step": 40870
+    },
+    {
+      "epoch": 0.16516037282287682,
+      "grad_norm": 855.2415161132812,
+      "learning_rate": 8.46897759154208e-06,
+      "loss": 50.173,
+      "step": 40880
+    },
+    {
+      "epoch": 0.16520077408824443,
+      "grad_norm": 730.0927124023438,
+      "learning_rate": 8.467961746607279e-06,
+      "loss": 57.6002,
+      "step": 40890
+    },
+    {
+      "epoch": 0.16524117535361207,
+      "grad_norm": 523.8104858398438,
+      "learning_rate": 8.466945625742678e-06,
+      "loss": 29.2724,
+      "step": 40900
+    },
+    {
+      "epoch": 0.16528157661897971,
+      "grad_norm": 572.157470703125,
+      "learning_rate": 8.465929229029124e-06,
+      "loss": 51.1289,
+      "step": 40910
+    },
+    {
+      "epoch": 0.16532197788434733,
+      "grad_norm": 851.248779296875,
+      "learning_rate": 8.464912556547486e-06,
+      "loss": 64.4108,
+      "step": 40920
+    },
+    {
+      "epoch": 0.16536237914971497,
+      "grad_norm": 1449.7430419921875,
+      "learning_rate": 8.46389560837866e-06,
+      "loss": 75.141,
+      "step": 40930
+    },
+    {
+      "epoch": 0.1654027804150826,
+      "grad_norm": 608.1896362304688,
+      "learning_rate": 8.462878384603558e-06,
+      "loss": 57.1169,
+      "step": 40940
+    },
+    {
+      "epoch": 0.16544318168045022,
+      "grad_norm": 887.8237915039062,
+      "learning_rate": 8.461860885303116e-06,
+      "loss": 43.4107,
+      "step": 40950
+    },
+    {
+      "epoch": 0.16548358294581786,
+      "grad_norm": 716.5380249023438,
+      "learning_rate": 8.460843110558287e-06,
+      "loss": 48.526,
+      "step": 40960
+    },
+    {
+      "epoch": 0.1655239842111855,
+      "grad_norm": 552.896240234375,
+      "learning_rate": 8.459825060450058e-06,
+      "loss": 37.3961,
+      "step": 40970
+    },
+    {
+      "epoch": 0.16556438547655314,
+      "grad_norm": 841.8020629882812,
+      "learning_rate": 8.458806735059428e-06,
+      "loss": 47.3296,
+      "step": 40980
+    },
+    {
+      "epoch": 0.16560478674192075,
+      "grad_norm": 617.2305908203125,
+      "learning_rate": 8.45778813446742e-06,
+      "loss": 32.6389,
+      "step": 40990
+    },
+    {
+      "epoch": 0.1656451880072884,
+      "grad_norm": 457.9773864746094,
+      "learning_rate": 8.456769258755078e-06,
+      "loss": 40.7037,
+      "step": 41000
+    },
+    {
+      "epoch": 0.16568558927265603,
+      "grad_norm": 453.1566162109375,
+      "learning_rate": 8.455750108003468e-06,
+      "loss": 43.0002,
+      "step": 41010
+    },
+    {
+      "epoch": 0.16572599053802364,
+      "grad_norm": 782.6980590820312,
+      "learning_rate": 8.454730682293686e-06,
+      "loss": 42.0216,
+      "step": 41020
+    },
+    {
+      "epoch": 0.16576639180339128,
+      "grad_norm": 697.3536376953125,
+      "learning_rate": 8.453710981706838e-06,
+      "loss": 31.8444,
+      "step": 41030
+    },
+    {
+      "epoch": 0.16580679306875892,
+      "grad_norm": 289.1650390625,
+      "learning_rate": 8.452691006324055e-06,
+      "loss": 56.3204,
+      "step": 41040
+    },
+    {
+      "epoch": 0.16584719433412654,
+      "grad_norm": 595.3594970703125,
+      "learning_rate": 8.451670756226496e-06,
+      "loss": 60.4295,
+      "step": 41050
+    },
+    {
+      "epoch": 0.16588759559949418,
+      "grad_norm": 486.653076171875,
+      "learning_rate": 8.450650231495336e-06,
+      "loss": 35.8233,
+      "step": 41060
+    },
+    {
+      "epoch": 0.16592799686486182,
+      "grad_norm": 399.5616455078125,
+      "learning_rate": 8.449629432211774e-06,
+      "loss": 57.242,
+      "step": 41070
+    },
+    {
+      "epoch": 0.16596839813022943,
+      "grad_norm": 421.82159423828125,
+      "learning_rate": 8.44860835845703e-06,
+      "loss": 41.3208,
+      "step": 41080
+    },
+    {
+      "epoch": 0.16600879939559707,
+      "grad_norm": 376.24420166015625,
+      "learning_rate": 8.447587010312343e-06,
+      "loss": 47.7066,
+      "step": 41090
+    },
+    {
+      "epoch": 0.1660492006609647,
+      "grad_norm": 383.0332336425781,
+      "learning_rate": 8.44656538785898e-06,
+      "loss": 42.9964,
+      "step": 41100
+    },
+    {
+      "epoch": 0.16608960192633232,
+      "grad_norm": 395.98724365234375,
+      "learning_rate": 8.44554349117823e-06,
+      "loss": 33.3155,
+      "step": 41110
+    },
+    {
+      "epoch": 0.16613000319169996,
+      "grad_norm": 660.3027954101562,
+      "learning_rate": 8.444521320351397e-06,
+      "loss": 88.4446,
+      "step": 41120
+    },
+    {
+      "epoch": 0.1661704044570676,
+      "grad_norm": 882.0906372070312,
+      "learning_rate": 8.44349887545981e-06,
+      "loss": 30.5517,
+      "step": 41130
+    },
+    {
+      "epoch": 0.16621080572243524,
+      "grad_norm": 308.0935363769531,
+      "learning_rate": 8.442476156584818e-06,
+      "loss": 36.8402,
+      "step": 41140
+    },
+    {
+      "epoch": 0.16625120698780285,
+      "grad_norm": 530.5660400390625,
+      "learning_rate": 8.4414531638078e-06,
+      "loss": 48.6802,
+      "step": 41150
+    },
+    {
+      "epoch": 0.1662916082531705,
+      "grad_norm": 313.61029052734375,
+      "learning_rate": 8.440429897210148e-06,
+      "loss": 57.5153,
+      "step": 41160
+    },
+    {
+      "epoch": 0.16633200951853813,
+      "grad_norm": 470.394775390625,
+      "learning_rate": 8.439406356873279e-06,
+      "loss": 38.5236,
+      "step": 41170
+    },
+    {
+      "epoch": 0.16637241078390574,
+      "grad_norm": 632.5592651367188,
+      "learning_rate": 8.43838254287863e-06,
+      "loss": 60.4704,
+      "step": 41180
+    },
+    {
+      "epoch": 0.16641281204927338,
+      "grad_norm": 387.6313781738281,
+      "learning_rate": 8.43735845530766e-06,
+      "loss": 48.1446,
+      "step": 41190
+    },
+    {
+      "epoch": 0.16645321331464102,
+      "grad_norm": 919.393310546875,
+      "learning_rate": 8.436334094241855e-06,
+      "loss": 70.1436,
+      "step": 41200
+    },
+    {
+      "epoch": 0.16649361458000864,
+      "grad_norm": 864.226806640625,
+      "learning_rate": 8.435309459762718e-06,
+      "loss": 51.2987,
+      "step": 41210
+    },
+    {
+      "epoch": 0.16653401584537628,
+      "grad_norm": 210.6269073486328,
+      "learning_rate": 8.434284551951772e-06,
+      "loss": 47.4699,
+      "step": 41220
+    },
+    {
+      "epoch": 0.16657441711074392,
+      "grad_norm": 479.52294921875,
+      "learning_rate": 8.433259370890565e-06,
+      "loss": 40.4656,
+      "step": 41230
+    },
+    {
+      "epoch": 0.16661481837611153,
+      "grad_norm": 975.6790771484375,
+      "learning_rate": 8.432233916660669e-06,
+      "loss": 46.843,
+      "step": 41240
+    },
+    {
+      "epoch": 0.16665521964147917,
+      "grad_norm": 774.6649780273438,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 40.471,
+      "step": 41250
+    },
+    {
+      "epoch": 0.1666956209068468,
+      "grad_norm": 423.52178955078125,
+      "learning_rate": 8.43018218902118e-06,
+      "loss": 48.6622,
+      "step": 41260
+    },
+    {
+      "epoch": 0.16673602217221442,
+      "grad_norm": 424.2727355957031,
+      "learning_rate": 8.429155915774839e-06,
+      "loss": 44.5995,
+      "step": 41270
+    },
+    {
+      "epoch": 0.16677642343758206,
+      "grad_norm": 1353.1103515625,
+      "learning_rate": 8.428129369686299e-06,
+      "loss": 107.302,
+      "step": 41280
+    },
+    {
+      "epoch": 0.1668168247029497,
+      "grad_norm": 476.1059265136719,
+      "learning_rate": 8.427102550837238e-06,
+      "loss": 52.2604,
+      "step": 41290
+    },
+    {
+      "epoch": 0.16685722596831734,
+      "grad_norm": 820.45068359375,
+      "learning_rate": 8.426075459309356e-06,
+      "loss": 59.7153,
+      "step": 41300
+    },
+    {
+      "epoch": 0.16689762723368495,
+      "grad_norm": 811.30322265625,
+      "learning_rate": 8.42504809518437e-06,
+      "loss": 43.2647,
+      "step": 41310
+    },
+    {
+      "epoch": 0.1669380284990526,
+      "grad_norm": 541.5391845703125,
+      "learning_rate": 8.42402045854403e-06,
+      "loss": 43.6814,
+      "step": 41320
+    },
+    {
+      "epoch": 0.16697842976442023,
+      "grad_norm": 931.9412231445312,
+      "learning_rate": 8.422992549470094e-06,
+      "loss": 61.0118,
+      "step": 41330
+    },
+    {
+      "epoch": 0.16701883102978785,
+      "grad_norm": 779.17724609375,
+      "learning_rate": 8.42196436804435e-06,
+      "loss": 53.9492,
+      "step": 41340
+    },
+    {
+      "epoch": 0.16705923229515549,
+      "grad_norm": 917.96142578125,
+      "learning_rate": 8.420935914348607e-06,
+      "loss": 48.7795,
+      "step": 41350
+    },
+    {
+      "epoch": 0.16709963356052313,
+      "grad_norm": 1356.6431884765625,
+      "learning_rate": 8.419907188464691e-06,
+      "loss": 62.2583,
+      "step": 41360
+    },
+    {
+      "epoch": 0.16714003482589074,
+      "grad_norm": 622.7725830078125,
+      "learning_rate": 8.418878190474459e-06,
+      "loss": 46.7867,
+      "step": 41370
+    },
+    {
+      "epoch": 0.16718043609125838,
+      "grad_norm": 498.8053894042969,
+      "learning_rate": 8.417848920459778e-06,
+      "loss": 52.3582,
+      "step": 41380
+    },
+    {
+      "epoch": 0.16722083735662602,
+      "grad_norm": 1115.1090087890625,
+      "learning_rate": 8.416819378502543e-06,
+      "loss": 61.8992,
+      "step": 41390
+    },
+    {
+      "epoch": 0.16726123862199363,
+      "grad_norm": 290.065185546875,
+      "learning_rate": 8.415789564684673e-06,
+      "loss": 49.5622,
+      "step": 41400
+    },
+    {
+      "epoch": 0.16730163988736127,
+      "grad_norm": 712.3739624023438,
+      "learning_rate": 8.414759479088102e-06,
+      "loss": 43.2107,
+      "step": 41410
+    },
+    {
+      "epoch": 0.1673420411527289,
+      "grad_norm": 562.4710083007812,
+      "learning_rate": 8.413729121794794e-06,
+      "loss": 41.9459,
+      "step": 41420
+    },
+    {
+      "epoch": 0.16738244241809652,
+      "grad_norm": 529.8401489257812,
+      "learning_rate": 8.412698492886723e-06,
+      "loss": 36.3189,
+      "step": 41430
+    },
+    {
+      "epoch": 0.16742284368346416,
+      "grad_norm": 653.1558227539062,
+      "learning_rate": 8.411667592445898e-06,
+      "loss": 67.4094,
+      "step": 41440
+    },
+    {
+      "epoch": 0.1674632449488318,
+      "grad_norm": 786.386962890625,
+      "learning_rate": 8.410636420554337e-06,
+      "loss": 54.185,
+      "step": 41450
+    },
+    {
+      "epoch": 0.16750364621419944,
+      "grad_norm": 522.7491455078125,
+      "learning_rate": 8.409604977294093e-06,
+      "loss": 45.0625,
+      "step": 41460
+    },
+    {
+      "epoch": 0.16754404747956705,
+      "grad_norm": 5738.32568359375,
+      "learning_rate": 8.408573262747225e-06,
+      "loss": 87.3419,
+      "step": 41470
+    },
+    {
+      "epoch": 0.1675844487449347,
+      "grad_norm": 768.86328125,
+      "learning_rate": 8.407541276995828e-06,
+      "loss": 50.3956,
+      "step": 41480
+    },
+    {
+      "epoch": 0.16762485001030233,
+      "grad_norm": 688.6503295898438,
+      "learning_rate": 8.40650902012201e-06,
+      "loss": 47.1603,
+      "step": 41490
+    },
+    {
+      "epoch": 0.16766525127566995,
+      "grad_norm": 650.5867309570312,
+      "learning_rate": 8.405476492207902e-06,
+      "loss": 39.5879,
+      "step": 41500
+    },
+    {
+      "epoch": 0.1677056525410376,
+      "grad_norm": 1079.1339111328125,
+      "learning_rate": 8.404443693335658e-06,
+      "loss": 45.5985,
+      "step": 41510
+    },
+    {
+      "epoch": 0.16774605380640523,
+      "grad_norm": 601.2516479492188,
+      "learning_rate": 8.403410623587454e-06,
+      "loss": 46.3612,
+      "step": 41520
+    },
+    {
+      "epoch": 0.16778645507177284,
+      "grad_norm": 781.9537963867188,
+      "learning_rate": 8.402377283045487e-06,
+      "loss": 60.7914,
+      "step": 41530
+    },
+    {
+      "epoch": 0.16782685633714048,
+      "grad_norm": 607.5199584960938,
+      "learning_rate": 8.401343671791974e-06,
+      "loss": 50.5169,
+      "step": 41540
+    },
+    {
+      "epoch": 0.16786725760250812,
+      "grad_norm": 413.96392822265625,
+      "learning_rate": 8.400309789909155e-06,
+      "loss": 57.9376,
+      "step": 41550
+    },
+    {
+      "epoch": 0.16790765886787573,
+      "grad_norm": 483.5220642089844,
+      "learning_rate": 8.399275637479291e-06,
+      "loss": 57.2534,
+      "step": 41560
+    },
+    {
+      "epoch": 0.16794806013324337,
+      "grad_norm": 1478.341796875,
+      "learning_rate": 8.398241214584666e-06,
+      "loss": 53.3458,
+      "step": 41570
+    },
+    {
+      "epoch": 0.167988461398611,
+      "grad_norm": 552.9909057617188,
+      "learning_rate": 8.397206521307584e-06,
+      "loss": 58.3804,
+      "step": 41580
+    },
+    {
+      "epoch": 0.16802886266397862,
+      "grad_norm": 608.0415649414062,
+      "learning_rate": 8.396171557730369e-06,
+      "loss": 62.6251,
+      "step": 41590
+    },
+    {
+      "epoch": 0.16806926392934626,
+      "grad_norm": 645.6259155273438,
+      "learning_rate": 8.39513632393537e-06,
+      "loss": 41.6197,
+      "step": 41600
+    },
+    {
+      "epoch": 0.1681096651947139,
+      "grad_norm": 648.4116821289062,
+      "learning_rate": 8.394100820004954e-06,
+      "loss": 62.4968,
+      "step": 41610
+    },
+    {
+      "epoch": 0.16815006646008154,
+      "grad_norm": 373.94134521484375,
+      "learning_rate": 8.393065046021513e-06,
+      "loss": 46.2327,
+      "step": 41620
+    },
+    {
+      "epoch": 0.16819046772544916,
+      "grad_norm": 190.22682189941406,
+      "learning_rate": 8.39202900206746e-06,
+      "loss": 50.5545,
+      "step": 41630
+    },
+    {
+      "epoch": 0.1682308689908168,
+      "grad_norm": 678.0693359375,
+      "learning_rate": 8.390992688225226e-06,
+      "loss": 41.0278,
+      "step": 41640
+    },
+    {
+      "epoch": 0.16827127025618444,
+      "grad_norm": 417.6565246582031,
+      "learning_rate": 8.389956104577265e-06,
+      "loss": 50.3119,
+      "step": 41650
+    },
+    {
+      "epoch": 0.16831167152155205,
+      "grad_norm": 270.62994384765625,
+      "learning_rate": 8.388919251206054e-06,
+      "loss": 57.1411,
+      "step": 41660
+    },
+    {
+      "epoch": 0.1683520727869197,
+      "grad_norm": 1314.5177001953125,
+      "learning_rate": 8.387882128194094e-06,
+      "loss": 52.9899,
+      "step": 41670
+    },
+    {
+      "epoch": 0.16839247405228733,
+      "grad_norm": 751.9255981445312,
+      "learning_rate": 8.3868447356239e-06,
+      "loss": 61.0519,
+      "step": 41680
+    },
+    {
+      "epoch": 0.16843287531765494,
+      "grad_norm": 831.6484375,
+      "learning_rate": 8.385807073578014e-06,
+      "loss": 43.6853,
+      "step": 41690
+    },
+    {
+      "epoch": 0.16847327658302258,
+      "grad_norm": 493.83502197265625,
+      "learning_rate": 8.384769142138998e-06,
+      "loss": 43.3712,
+      "step": 41700
+    },
+    {
+      "epoch": 0.16851367784839022,
+      "grad_norm": 663.5391845703125,
+      "learning_rate": 8.383730941389434e-06,
+      "loss": 58.759,
+      "step": 41710
+    },
+    {
+      "epoch": 0.16855407911375783,
+      "grad_norm": 201.29359436035156,
+      "learning_rate": 8.382692471411931e-06,
+      "loss": 47.675,
+      "step": 41720
+    },
+    {
+      "epoch": 0.16859448037912547,
+      "grad_norm": 660.2062377929688,
+      "learning_rate": 8.38165373228911e-06,
+      "loss": 69.8102,
+      "step": 41730
+    },
+    {
+      "epoch": 0.1686348816444931,
+      "grad_norm": 246.1966094970703,
+      "learning_rate": 8.380614724103622e-06,
+      "loss": 43.068,
+      "step": 41740
+    },
+    {
+      "epoch": 0.16867528290986072,
+      "grad_norm": 548.47998046875,
+      "learning_rate": 8.379575446938136e-06,
+      "loss": 57.0117,
+      "step": 41750
+    },
+    {
+      "epoch": 0.16871568417522836,
+      "grad_norm": 715.6472778320312,
+      "learning_rate": 8.37853590087534e-06,
+      "loss": 50.1982,
+      "step": 41760
+    },
+    {
+      "epoch": 0.168756085440596,
+      "grad_norm": 469.2695007324219,
+      "learning_rate": 8.377496085997949e-06,
+      "loss": 36.8721,
+      "step": 41770
+    },
+    {
+      "epoch": 0.16879648670596364,
+      "grad_norm": 609.3477783203125,
+      "learning_rate": 8.376456002388695e-06,
+      "loss": 51.9407,
+      "step": 41780
+    },
+    {
+      "epoch": 0.16883688797133126,
+      "grad_norm": 948.3466796875,
+      "learning_rate": 8.375415650130332e-06,
+      "loss": 39.1852,
+      "step": 41790
+    },
+    {
+      "epoch": 0.1688772892366989,
+      "grad_norm": 518.4425659179688,
+      "learning_rate": 8.37437502930564e-06,
+      "loss": 41.667,
+      "step": 41800
+    },
+    {
+      "epoch": 0.16891769050206654,
+      "grad_norm": 616.8099365234375,
+      "learning_rate": 8.373334139997409e-06,
+      "loss": 43.1468,
+      "step": 41810
+    },
+    {
+      "epoch": 0.16895809176743415,
+      "grad_norm": 675.777099609375,
+      "learning_rate": 8.372292982288463e-06,
+      "loss": 41.971,
+      "step": 41820
+    },
+    {
+      "epoch": 0.1689984930328018,
+      "grad_norm": 523.8914794921875,
+      "learning_rate": 8.371251556261642e-06,
+      "loss": 53.4606,
+      "step": 41830
+    },
+    {
+      "epoch": 0.16903889429816943,
+      "grad_norm": 511.80206298828125,
+      "learning_rate": 8.370209861999807e-06,
+      "loss": 47.4166,
+      "step": 41840
+    },
+    {
+      "epoch": 0.16907929556353704,
+      "grad_norm": 361.1617736816406,
+      "learning_rate": 8.36916789958584e-06,
+      "loss": 41.0987,
+      "step": 41850
+    },
+    {
+      "epoch": 0.16911969682890468,
+      "grad_norm": 6810.2587890625,
+      "learning_rate": 8.368125669102645e-06,
+      "loss": 76.2174,
+      "step": 41860
+    },
+    {
+      "epoch": 0.16916009809427232,
+      "grad_norm": 751.0748291015625,
+      "learning_rate": 8.36708317063315e-06,
+      "loss": 52.5193,
+      "step": 41870
+    },
+    {
+      "epoch": 0.16920049935963993,
+      "grad_norm": 572.6092529296875,
+      "learning_rate": 8.366040404260298e-06,
+      "loss": 48.3041,
+      "step": 41880
+    },
+    {
+      "epoch": 0.16924090062500757,
+      "grad_norm": 570.5328369140625,
+      "learning_rate": 8.36499737006706e-06,
+      "loss": 45.7522,
+      "step": 41890
+    },
+    {
+      "epoch": 0.1692813018903752,
+      "grad_norm": 529.4615478515625,
+      "learning_rate": 8.363954068136424e-06,
+      "loss": 53.588,
+      "step": 41900
+    },
+    {
+      "epoch": 0.16932170315574283,
+      "grad_norm": 1069.1943359375,
+      "learning_rate": 8.362910498551402e-06,
+      "loss": 77.0546,
+      "step": 41910
+    },
+    {
+      "epoch": 0.16936210442111047,
+      "grad_norm": 505.5280456542969,
+      "learning_rate": 8.361866661395024e-06,
+      "loss": 43.8829,
+      "step": 41920
+    },
+    {
+      "epoch": 0.1694025056864781,
+      "grad_norm": 902.7269897460938,
+      "learning_rate": 8.360822556750345e-06,
+      "loss": 42.4727,
+      "step": 41930
+    },
+    {
+      "epoch": 0.16944290695184575,
+      "grad_norm": 466.8572082519531,
+      "learning_rate": 8.35977818470044e-06,
+      "loss": 46.5572,
+      "step": 41940
+    },
+    {
+      "epoch": 0.16948330821721336,
+      "grad_norm": 894.6742553710938,
+      "learning_rate": 8.358733545328404e-06,
+      "loss": 58.5088,
+      "step": 41950
+    },
+    {
+      "epoch": 0.169523709482581,
+      "grad_norm": 285.3533935546875,
+      "learning_rate": 8.357688638717354e-06,
+      "loss": 54.5953,
+      "step": 41960
+    },
+    {
+      "epoch": 0.16956411074794864,
+      "grad_norm": 1186.5799560546875,
+      "learning_rate": 8.356643464950428e-06,
+      "loss": 54.535,
+      "step": 41970
+    },
+    {
+      "epoch": 0.16960451201331625,
+      "grad_norm": 445.8567199707031,
+      "learning_rate": 8.355598024110789e-06,
+      "loss": 41.7435,
+      "step": 41980
+    },
+    {
+      "epoch": 0.1696449132786839,
+      "grad_norm": 597.5289916992188,
+      "learning_rate": 8.354552316281613e-06,
+      "loss": 77.5292,
+      "step": 41990
+    },
+    {
+      "epoch": 0.16968531454405153,
+      "grad_norm": 226.13951110839844,
+      "learning_rate": 8.353506341546106e-06,
+      "loss": 45.1414,
+      "step": 42000
+    },
+    {
+      "epoch": 0.16972571580941914,
+      "grad_norm": 499.8537292480469,
+      "learning_rate": 8.352460099987488e-06,
+      "loss": 44.9113,
+      "step": 42010
+    },
+    {
+      "epoch": 0.16976611707478678,
+      "grad_norm": 264.3408508300781,
+      "learning_rate": 8.351413591689007e-06,
+      "loss": 32.2931,
+      "step": 42020
+    },
+    {
+      "epoch": 0.16980651834015442,
+      "grad_norm": 928.5385131835938,
+      "learning_rate": 8.350366816733927e-06,
+      "loss": 64.2158,
+      "step": 42030
+    },
+    {
+      "epoch": 0.16984691960552203,
+      "grad_norm": 859.626953125,
+      "learning_rate": 8.349319775205536e-06,
+      "loss": 43.5746,
+      "step": 42040
+    },
+    {
+      "epoch": 0.16988732087088967,
+      "grad_norm": 419.03350830078125,
+      "learning_rate": 8.34827246718714e-06,
+      "loss": 41.5845,
+      "step": 42050
+    },
+    {
+      "epoch": 0.16992772213625731,
+      "grad_norm": 453.50531005859375,
+      "learning_rate": 8.347224892762072e-06,
+      "loss": 34.3044,
+      "step": 42060
+    },
+    {
+      "epoch": 0.16996812340162493,
+      "grad_norm": 500.0800476074219,
+      "learning_rate": 8.346177052013681e-06,
+      "loss": 49.3354,
+      "step": 42070
+    },
+    {
+      "epoch": 0.17000852466699257,
+      "grad_norm": 579.340576171875,
+      "learning_rate": 8.345128945025338e-06,
+      "loss": 49.3049,
+      "step": 42080
+    },
+    {
+      "epoch": 0.1700489259323602,
+      "grad_norm": 755.8071899414062,
+      "learning_rate": 8.344080571880438e-06,
+      "loss": 53.1518,
+      "step": 42090
+    },
+    {
+      "epoch": 0.17008932719772785,
+      "grad_norm": 605.3427124023438,
+      "learning_rate": 8.343031932662394e-06,
+      "loss": 38.2326,
+      "step": 42100
+    },
+    {
+      "epoch": 0.17012972846309546,
+      "grad_norm": 600.05615234375,
+      "learning_rate": 8.341983027454641e-06,
+      "loss": 31.4536,
+      "step": 42110
+    },
+    {
+      "epoch": 0.1701701297284631,
+      "grad_norm": 986.0425415039062,
+      "learning_rate": 8.340933856340637e-06,
+      "loss": 52.7103,
+      "step": 42120
+    },
+    {
+      "epoch": 0.17021053099383074,
+      "grad_norm": 737.8534545898438,
+      "learning_rate": 8.339884419403857e-06,
+      "loss": 55.2349,
+      "step": 42130
+    },
+    {
+      "epoch": 0.17025093225919835,
+      "grad_norm": 334.7961730957031,
+      "learning_rate": 8.338834716727801e-06,
+      "loss": 44.5493,
+      "step": 42140
+    },
+    {
+      "epoch": 0.170291333524566,
+      "grad_norm": 610.7578125,
+      "learning_rate": 8.337784748395992e-06,
+      "loss": 48.5967,
+      "step": 42150
+    },
+    {
+      "epoch": 0.17033173478993363,
+      "grad_norm": 839.372802734375,
+      "learning_rate": 8.336734514491968e-06,
+      "loss": 45.0206,
+      "step": 42160
+    },
+    {
+      "epoch": 0.17037213605530124,
+      "grad_norm": 1020.9551391601562,
+      "learning_rate": 8.335684015099294e-06,
+      "loss": 55.0339,
+      "step": 42170
+    },
+    {
+      "epoch": 0.17041253732066888,
+      "grad_norm": 823.2727661132812,
+      "learning_rate": 8.33463325030155e-06,
+      "loss": 51.4678,
+      "step": 42180
+    },
+    {
+      "epoch": 0.17045293858603652,
+      "grad_norm": 524.025146484375,
+      "learning_rate": 8.333582220182344e-06,
+      "loss": 46.4598,
+      "step": 42190
+    },
+    {
+      "epoch": 0.17049333985140414,
+      "grad_norm": 473.0232849121094,
+      "learning_rate": 8.332530924825297e-06,
+      "loss": 57.2351,
+      "step": 42200
+    },
+    {
+      "epoch": 0.17053374111677178,
+      "grad_norm": 1434.7774658203125,
+      "learning_rate": 8.33147936431406e-06,
+      "loss": 56.9665,
+      "step": 42210
+    },
+    {
+      "epoch": 0.17057414238213942,
+      "grad_norm": 647.9461669921875,
+      "learning_rate": 8.3304275387323e-06,
+      "loss": 60.3315,
+      "step": 42220
+    },
+    {
+      "epoch": 0.17061454364750703,
+      "grad_norm": 1044.830078125,
+      "learning_rate": 8.329375448163703e-06,
+      "loss": 40.1567,
+      "step": 42230
+    },
+    {
+      "epoch": 0.17065494491287467,
+      "grad_norm": 894.5424194335938,
+      "learning_rate": 8.328323092691985e-06,
+      "loss": 46.116,
+      "step": 42240
+    },
+    {
+      "epoch": 0.1706953461782423,
+      "grad_norm": 698.2495727539062,
+      "learning_rate": 8.32727047240087e-06,
+      "loss": 50.6057,
+      "step": 42250
+    },
+    {
+      "epoch": 0.17073574744360995,
+      "grad_norm": 1016.3301391601562,
+      "learning_rate": 8.326217587374115e-06,
+      "loss": 59.295,
+      "step": 42260
+    },
+    {
+      "epoch": 0.17077614870897756,
+      "grad_norm": 549.875732421875,
+      "learning_rate": 8.325164437695493e-06,
+      "loss": 57.8668,
+      "step": 42270
+    },
+    {
+      "epoch": 0.1708165499743452,
+      "grad_norm": 565.3195190429688,
+      "learning_rate": 8.324111023448795e-06,
+      "loss": 48.7341,
+      "step": 42280
+    },
+    {
+      "epoch": 0.17085695123971284,
+      "grad_norm": 885.8643188476562,
+      "learning_rate": 8.32305734471784e-06,
+      "loss": 45.0074,
+      "step": 42290
+    },
+    {
+      "epoch": 0.17089735250508045,
+      "grad_norm": 1960.1605224609375,
+      "learning_rate": 8.322003401586463e-06,
+      "loss": 68.4631,
+      "step": 42300
+    },
+    {
+      "epoch": 0.1709377537704481,
+      "grad_norm": 509.7738037109375,
+      "learning_rate": 8.32094919413852e-06,
+      "loss": 57.2811,
+      "step": 42310
+    },
+    {
+      "epoch": 0.17097815503581573,
+      "grad_norm": 491.28338623046875,
+      "learning_rate": 8.319894722457892e-06,
+      "loss": 45.6266,
+      "step": 42320
+    },
+    {
+      "epoch": 0.17101855630118334,
+      "grad_norm": 430.2795104980469,
+      "learning_rate": 8.318839986628477e-06,
+      "loss": 43.9842,
+      "step": 42330
+    },
+    {
+      "epoch": 0.17105895756655098,
+      "grad_norm": 236.9319305419922,
+      "learning_rate": 8.317784986734194e-06,
+      "loss": 45.3445,
+      "step": 42340
+    },
+    {
+      "epoch": 0.17109935883191862,
+      "grad_norm": 1582.1502685546875,
+      "learning_rate": 8.316729722858987e-06,
+      "loss": 60.803,
+      "step": 42350
+    },
+    {
+      "epoch": 0.17113976009728624,
+      "grad_norm": 422.8695983886719,
+      "learning_rate": 8.31567419508682e-06,
+      "loss": 59.6711,
+      "step": 42360
+    },
+    {
+      "epoch": 0.17118016136265388,
+      "grad_norm": 1095.4122314453125,
+      "learning_rate": 8.31461840350167e-06,
+      "loss": 59.5847,
+      "step": 42370
+    },
+    {
+      "epoch": 0.17122056262802152,
+      "grad_norm": 346.7009582519531,
+      "learning_rate": 8.313562348187549e-06,
+      "loss": 60.8975,
+      "step": 42380
+    },
+    {
+      "epoch": 0.17126096389338913,
+      "grad_norm": 632.48974609375,
+      "learning_rate": 8.312506029228478e-06,
+      "loss": 47.6719,
+      "step": 42390
+    },
+    {
+      "epoch": 0.17130136515875677,
+      "grad_norm": 419.339111328125,
+      "learning_rate": 8.311449446708506e-06,
+      "loss": 49.8142,
+      "step": 42400
+    },
+    {
+      "epoch": 0.1713417664241244,
+      "grad_norm": 518.4583740234375,
+      "learning_rate": 8.310392600711698e-06,
+      "loss": 63.9603,
+      "step": 42410
+    },
+    {
+      "epoch": 0.17138216768949205,
+      "grad_norm": 645.0120239257812,
+      "learning_rate": 8.309335491322143e-06,
+      "loss": 54.8118,
+      "step": 42420
+    },
+    {
+      "epoch": 0.17142256895485966,
+      "grad_norm": 926.7328491210938,
+      "learning_rate": 8.30827811862395e-06,
+      "loss": 45.266,
+      "step": 42430
+    },
+    {
+      "epoch": 0.1714629702202273,
+      "grad_norm": 267.48138427734375,
+      "learning_rate": 8.307220482701251e-06,
+      "loss": 55.2687,
+      "step": 42440
+    },
+    {
+      "epoch": 0.17150337148559494,
+      "grad_norm": 754.5764770507812,
+      "learning_rate": 8.306162583638197e-06,
+      "loss": 39.9945,
+      "step": 42450
+    },
+    {
+      "epoch": 0.17154377275096255,
+      "grad_norm": 396.7652893066406,
+      "learning_rate": 8.305104421518959e-06,
+      "loss": 61.9433,
+      "step": 42460
+    },
+    {
+      "epoch": 0.1715841740163302,
+      "grad_norm": 1039.3521728515625,
+      "learning_rate": 8.30404599642773e-06,
+      "loss": 47.3649,
+      "step": 42470
+    },
+    {
+      "epoch": 0.17162457528169783,
+      "grad_norm": 678.3857421875,
+      "learning_rate": 8.302987308448724e-06,
+      "loss": 43.0866,
+      "step": 42480
+    },
+    {
+      "epoch": 0.17166497654706545,
+      "grad_norm": 368.931396484375,
+      "learning_rate": 8.301928357666178e-06,
+      "loss": 28.1968,
+      "step": 42490
+    },
+    {
+      "epoch": 0.17170537781243309,
+      "grad_norm": 455.97747802734375,
+      "learning_rate": 8.300869144164346e-06,
+      "loss": 72.2107,
+      "step": 42500
+    },
+    {
+      "epoch": 0.17174577907780073,
+      "grad_norm": 470.565673828125,
+      "learning_rate": 8.299809668027505e-06,
+      "loss": 51.0551,
+      "step": 42510
+    },
+    {
+      "epoch": 0.17178618034316834,
+      "grad_norm": 944.7191772460938,
+      "learning_rate": 8.298749929339953e-06,
+      "loss": 52.9875,
+      "step": 42520
+    },
+    {
+      "epoch": 0.17182658160853598,
+      "grad_norm": 566.98193359375,
+      "learning_rate": 8.297689928186009e-06,
+      "loss": 38.8439,
+      "step": 42530
+    },
+    {
+      "epoch": 0.17186698287390362,
+      "grad_norm": 336.9647521972656,
+      "learning_rate": 8.29662966465001e-06,
+      "loss": 59.2766,
+      "step": 42540
+    },
+    {
+      "epoch": 0.17190738413927123,
+      "grad_norm": 551.6116943359375,
+      "learning_rate": 8.295569138816319e-06,
+      "loss": 52.5348,
+      "step": 42550
+    },
+    {
+      "epoch": 0.17194778540463887,
+      "grad_norm": 659.5198974609375,
+      "learning_rate": 8.294508350769315e-06,
+      "loss": 34.3035,
+      "step": 42560
+    },
+    {
+      "epoch": 0.1719881866700065,
+      "grad_norm": 840.8699340820312,
+      "learning_rate": 8.293447300593402e-06,
+      "loss": 46.1301,
+      "step": 42570
+    },
+    {
+      "epoch": 0.17202858793537412,
+      "grad_norm": 1800.16748046875,
+      "learning_rate": 8.292385988373005e-06,
+      "loss": 56.3059,
+      "step": 42580
+    },
+    {
+      "epoch": 0.17206898920074176,
+      "grad_norm": 462.5638122558594,
+      "learning_rate": 8.29132441419256e-06,
+      "loss": 44.6035,
+      "step": 42590
+    },
+    {
+      "epoch": 0.1721093904661094,
+      "grad_norm": 684.3241577148438,
+      "learning_rate": 8.290262578136541e-06,
+      "loss": 58.0993,
+      "step": 42600
+    },
+    {
+      "epoch": 0.17214979173147704,
+      "grad_norm": 450.2313537597656,
+      "learning_rate": 8.289200480289426e-06,
+      "loss": 67.3204,
+      "step": 42610
+    },
+    {
+      "epoch": 0.17219019299684465,
+      "grad_norm": 498.5903015136719,
+      "learning_rate": 8.288138120735726e-06,
+      "loss": 28.4738,
+      "step": 42620
+    },
+    {
+      "epoch": 0.1722305942622123,
+      "grad_norm": 758.5321655273438,
+      "learning_rate": 8.287075499559965e-06,
+      "loss": 60.2802,
+      "step": 42630
+    },
+    {
+      "epoch": 0.17227099552757993,
+      "grad_norm": 351.36859130859375,
+      "learning_rate": 8.286012616846693e-06,
+      "loss": 52.4755,
+      "step": 42640
+    },
+    {
+      "epoch": 0.17231139679294755,
+      "grad_norm": 539.5321655273438,
+      "learning_rate": 8.284949472680477e-06,
+      "loss": 56.0628,
+      "step": 42650
+    },
+    {
+      "epoch": 0.1723517980583152,
+      "grad_norm": 2152.857177734375,
+      "learning_rate": 8.283886067145908e-06,
+      "loss": 56.3268,
+      "step": 42660
+    },
+    {
+      "epoch": 0.17239219932368283,
+      "grad_norm": 395.3343811035156,
+      "learning_rate": 8.282822400327595e-06,
+      "loss": 44.1626,
+      "step": 42670
+    },
+    {
+      "epoch": 0.17243260058905044,
+      "grad_norm": 594.5778198242188,
+      "learning_rate": 8.28175847231017e-06,
+      "loss": 51.8114,
+      "step": 42680
+    },
+    {
+      "epoch": 0.17247300185441808,
+      "grad_norm": 803.4881591796875,
+      "learning_rate": 8.280694283178285e-06,
+      "loss": 41.6613,
+      "step": 42690
+    },
+    {
+      "epoch": 0.17251340311978572,
+      "grad_norm": 1555.0865478515625,
+      "learning_rate": 8.27962983301661e-06,
+      "loss": 65.1994,
+      "step": 42700
+    },
+    {
+      "epoch": 0.17255380438515333,
+      "grad_norm": 695.3710327148438,
+      "learning_rate": 8.278565121909845e-06,
+      "loss": 66.7208,
+      "step": 42710
+    },
+    {
+      "epoch": 0.17259420565052097,
+      "grad_norm": 540.1356811523438,
+      "learning_rate": 8.277500149942697e-06,
+      "loss": 43.4715,
+      "step": 42720
+    },
+    {
+      "epoch": 0.1726346069158886,
+      "grad_norm": 683.9020385742188,
+      "learning_rate": 8.276434917199904e-06,
+      "loss": 37.1363,
+      "step": 42730
+    },
+    {
+      "epoch": 0.17267500818125622,
+      "grad_norm": 2300.76220703125,
+      "learning_rate": 8.275369423766222e-06,
+      "loss": 47.8163,
+      "step": 42740
+    },
+    {
+      "epoch": 0.17271540944662386,
+      "grad_norm": 752.5711059570312,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 39.8013,
+      "step": 42750
+    },
+    {
+      "epoch": 0.1727558107119915,
+      "grad_norm": 830.5774536132812,
+      "learning_rate": 8.273237655165314e-06,
+      "loss": 61.424,
+      "step": 42760
+    },
+    {
+      "epoch": 0.17279621197735914,
+      "grad_norm": 748.5219116210938,
+      "learning_rate": 8.272171380167705e-06,
+      "loss": 62.0189,
+      "step": 42770
+    },
+    {
+      "epoch": 0.17283661324272676,
+      "grad_norm": 423.2668762207031,
+      "learning_rate": 8.271104844818436e-06,
+      "loss": 46.4281,
+      "step": 42780
+    },
+    {
+      "epoch": 0.1728770145080944,
+      "grad_norm": 359.3676452636719,
+      "learning_rate": 8.270038049202366e-06,
+      "loss": 53.1651,
+      "step": 42790
+    },
+    {
+      "epoch": 0.17291741577346204,
+      "grad_norm": 264.7219543457031,
+      "learning_rate": 8.268970993404377e-06,
+      "loss": 46.8862,
+      "step": 42800
+    },
+    {
+      "epoch": 0.17295781703882965,
+      "grad_norm": 677.9569702148438,
+      "learning_rate": 8.267903677509368e-06,
+      "loss": 47.7671,
+      "step": 42810
+    },
+    {
+      "epoch": 0.1729982183041973,
+      "grad_norm": 898.0774536132812,
+      "learning_rate": 8.266836101602263e-06,
+      "loss": 52.6813,
+      "step": 42820
+    },
+    {
+      "epoch": 0.17303861956956493,
+      "grad_norm": 387.7664489746094,
+      "learning_rate": 8.265768265767999e-06,
+      "loss": 48.4607,
+      "step": 42830
+    },
+    {
+      "epoch": 0.17307902083493254,
+      "grad_norm": 914.6298828125,
+      "learning_rate": 8.264700170091543e-06,
+      "loss": 76.5356,
+      "step": 42840
+    },
+    {
+      "epoch": 0.17311942210030018,
+      "grad_norm": 419.8518981933594,
+      "learning_rate": 8.263631814657879e-06,
+      "loss": 33.402,
+      "step": 42850
+    },
+    {
+      "epoch": 0.17315982336566782,
+      "grad_norm": 523.0492553710938,
+      "learning_rate": 8.262563199552007e-06,
+      "loss": 41.1054,
+      "step": 42860
+    },
+    {
+      "epoch": 0.17320022463103543,
+      "grad_norm": 1040.5003662109375,
+      "learning_rate": 8.261494324858956e-06,
+      "loss": 54.1803,
+      "step": 42870
+    },
+    {
+      "epoch": 0.17324062589640307,
+      "grad_norm": 591.4171142578125,
+      "learning_rate": 8.26042519066377e-06,
+      "loss": 43.2964,
+      "step": 42880
+    },
+    {
+      "epoch": 0.1732810271617707,
+      "grad_norm": 359.90386962890625,
+      "learning_rate": 8.259355797051515e-06,
+      "loss": 28.9607,
+      "step": 42890
+    },
+    {
+      "epoch": 0.17332142842713832,
+      "grad_norm": 656.0526733398438,
+      "learning_rate": 8.258286144107277e-06,
+      "loss": 27.343,
+      "step": 42900
+    },
+    {
+      "epoch": 0.17336182969250596,
+      "grad_norm": 810.3823852539062,
+      "learning_rate": 8.257216231916162e-06,
+      "loss": 71.7967,
+      "step": 42910
+    },
+    {
+      "epoch": 0.1734022309578736,
+      "grad_norm": 662.1875610351562,
+      "learning_rate": 8.256146060563304e-06,
+      "loss": 46.7155,
+      "step": 42920
+    },
+    {
+      "epoch": 0.17344263222324124,
+      "grad_norm": 268.1656494140625,
+      "learning_rate": 8.255075630133847e-06,
+      "loss": 53.0825,
+      "step": 42930
+    },
+    {
+      "epoch": 0.17348303348860886,
+      "grad_norm": 458.50628662109375,
+      "learning_rate": 8.254004940712958e-06,
+      "loss": 52.152,
+      "step": 42940
+    },
+    {
+      "epoch": 0.1735234347539765,
+      "grad_norm": 639.825439453125,
+      "learning_rate": 8.252933992385833e-06,
+      "loss": 63.8295,
+      "step": 42950
+    },
+    {
+      "epoch": 0.17356383601934414,
+      "grad_norm": 506.6952209472656,
+      "learning_rate": 8.251862785237676e-06,
+      "loss": 54.3871,
+      "step": 42960
+    },
+    {
+      "epoch": 0.17360423728471175,
+      "grad_norm": 168.96426391601562,
+      "learning_rate": 8.250791319353723e-06,
+      "loss": 31.244,
+      "step": 42970
+    },
+    {
+      "epoch": 0.1736446385500794,
+      "grad_norm": 333.39715576171875,
+      "learning_rate": 8.249719594819225e-06,
+      "loss": 51.3396,
+      "step": 42980
+    },
+    {
+      "epoch": 0.17368503981544703,
+      "grad_norm": 512.88134765625,
+      "learning_rate": 8.248647611719452e-06,
+      "loss": 52.5945,
+      "step": 42990
+    },
+    {
+      "epoch": 0.17372544108081464,
+      "grad_norm": 451.7750549316406,
+      "learning_rate": 8.247575370139695e-06,
+      "loss": 36.7286,
+      "step": 43000
+    },
+    {
+      "epoch": 0.17376584234618228,
+      "grad_norm": 875.9497680664062,
+      "learning_rate": 8.246502870165273e-06,
+      "loss": 45.5384,
+      "step": 43010
+    },
+    {
+      "epoch": 0.17380624361154992,
+      "grad_norm": 1002.91455078125,
+      "learning_rate": 8.245430111881519e-06,
+      "loss": 47.9184,
+      "step": 43020
+    },
+    {
+      "epoch": 0.17384664487691753,
+      "grad_norm": 727.3613891601562,
+      "learning_rate": 8.244357095373783e-06,
+      "loss": 57.4964,
+      "step": 43030
+    },
+    {
+      "epoch": 0.17388704614228517,
+      "grad_norm": 462.3572082519531,
+      "learning_rate": 8.243283820727441e-06,
+      "loss": 37.5159,
+      "step": 43040
+    },
+    {
+      "epoch": 0.1739274474076528,
+      "grad_norm": 738.8742065429688,
+      "learning_rate": 8.242210288027893e-06,
+      "loss": 57.068,
+      "step": 43050
+    },
+    {
+      "epoch": 0.17396784867302043,
+      "grad_norm": 593.364990234375,
+      "learning_rate": 8.241136497360552e-06,
+      "loss": 61.0344,
+      "step": 43060
+    },
+    {
+      "epoch": 0.17400824993838807,
+      "grad_norm": 892.4779052734375,
+      "learning_rate": 8.240062448810853e-06,
+      "loss": 48.9156,
+      "step": 43070
+    },
+    {
+      "epoch": 0.1740486512037557,
+      "grad_norm": 783.9124755859375,
+      "learning_rate": 8.238988142464254e-06,
+      "loss": 53.5666,
+      "step": 43080
+    },
+    {
+      "epoch": 0.17408905246912335,
+      "grad_norm": 389.2962951660156,
+      "learning_rate": 8.237913578406236e-06,
+      "loss": 40.7518,
+      "step": 43090
+    },
+    {
+      "epoch": 0.17412945373449096,
+      "grad_norm": 536.7239990234375,
+      "learning_rate": 8.236838756722294e-06,
+      "loss": 53.6748,
+      "step": 43100
+    },
+    {
+      "epoch": 0.1741698549998586,
+      "grad_norm": 1080.3280029296875,
+      "learning_rate": 8.235763677497945e-06,
+      "loss": 151.0997,
+      "step": 43110
+    },
+    {
+      "epoch": 0.17421025626522624,
+      "grad_norm": 554.4981689453125,
+      "learning_rate": 8.234688340818732e-06,
+      "loss": 61.6212,
+      "step": 43120
+    },
+    {
+      "epoch": 0.17425065753059385,
+      "grad_norm": 828.6658935546875,
+      "learning_rate": 8.233612746770214e-06,
+      "loss": 55.1259,
+      "step": 43130
+    },
+    {
+      "epoch": 0.1742910587959615,
+      "grad_norm": 249.32948303222656,
+      "learning_rate": 8.232536895437968e-06,
+      "loss": 52.4912,
+      "step": 43140
+    },
+    {
+      "epoch": 0.17433146006132913,
+      "grad_norm": 770.260009765625,
+      "learning_rate": 8.231460786907597e-06,
+      "loss": 35.9362,
+      "step": 43150
+    },
+    {
+      "epoch": 0.17437186132669674,
+      "grad_norm": 637.1809692382812,
+      "learning_rate": 8.230384421264722e-06,
+      "loss": 55.1247,
+      "step": 43160
+    },
+    {
+      "epoch": 0.17441226259206438,
+      "grad_norm": 587.2365112304688,
+      "learning_rate": 8.229307798594985e-06,
+      "loss": 57.6803,
+      "step": 43170
+    },
+    {
+      "epoch": 0.17445266385743202,
+      "grad_norm": 431.0023498535156,
+      "learning_rate": 8.228230918984046e-06,
+      "loss": 40.9327,
+      "step": 43180
+    },
+    {
+      "epoch": 0.17449306512279963,
+      "grad_norm": 471.2975158691406,
+      "learning_rate": 8.22715378251759e-06,
+      "loss": 43.1537,
+      "step": 43190
+    },
+    {
+      "epoch": 0.17453346638816727,
+      "grad_norm": 738.6006469726562,
+      "learning_rate": 8.226076389281316e-06,
+      "loss": 68.729,
+      "step": 43200
+    },
+    {
+      "epoch": 0.17457386765353491,
+      "grad_norm": 496.95184326171875,
+      "learning_rate": 8.22499873936095e-06,
+      "loss": 46.558,
+      "step": 43210
+    },
+    {
+      "epoch": 0.17461426891890253,
+      "grad_norm": 473.88580322265625,
+      "learning_rate": 8.223920832842236e-06,
+      "loss": 139.2572,
+      "step": 43220
+    },
+    {
+      "epoch": 0.17465467018427017,
+      "grad_norm": 989.8582763671875,
+      "learning_rate": 8.222842669810936e-06,
+      "loss": 38.1966,
+      "step": 43230
+    },
+    {
+      "epoch": 0.1746950714496378,
+      "grad_norm": 934.4788208007812,
+      "learning_rate": 8.221764250352835e-06,
+      "loss": 48.2463,
+      "step": 43240
+    },
+    {
+      "epoch": 0.17473547271500545,
+      "grad_norm": 654.1959228515625,
+      "learning_rate": 8.220685574553739e-06,
+      "loss": 50.7117,
+      "step": 43250
+    },
+    {
+      "epoch": 0.17477587398037306,
+      "grad_norm": 736.326171875,
+      "learning_rate": 8.219606642499474e-06,
+      "loss": 47.1028,
+      "step": 43260
+    },
+    {
+      "epoch": 0.1748162752457407,
+      "grad_norm": 892.4651489257812,
+      "learning_rate": 8.218527454275884e-06,
+      "loss": 57.5517,
+      "step": 43270
+    },
+    {
+      "epoch": 0.17485667651110834,
+      "grad_norm": 749.7070922851562,
+      "learning_rate": 8.217448009968834e-06,
+      "loss": 63.03,
+      "step": 43280
+    },
+    {
+      "epoch": 0.17489707777647595,
+      "grad_norm": 809.4559326171875,
+      "learning_rate": 8.216368309664213e-06,
+      "loss": 46.0392,
+      "step": 43290
+    },
+    {
+      "epoch": 0.1749374790418436,
+      "grad_norm": 685.7556762695312,
+      "learning_rate": 8.215288353447927e-06,
+      "loss": 40.1804,
+      "step": 43300
+    },
+    {
+      "epoch": 0.17497788030721123,
+      "grad_norm": 751.0911865234375,
+      "learning_rate": 8.214208141405903e-06,
+      "loss": 42.3572,
+      "step": 43310
+    },
+    {
+      "epoch": 0.17501828157257884,
+      "grad_norm": 504.02471923828125,
+      "learning_rate": 8.213127673624088e-06,
+      "loss": 47.8609,
+      "step": 43320
+    },
+    {
+      "epoch": 0.17505868283794648,
+      "grad_norm": 544.6509399414062,
+      "learning_rate": 8.212046950188451e-06,
+      "loss": 60.7416,
+      "step": 43330
+    },
+    {
+      "epoch": 0.17509908410331412,
+      "grad_norm": 190.57473754882812,
+      "learning_rate": 8.21096597118498e-06,
+      "loss": 47.4481,
+      "step": 43340
+    },
+    {
+      "epoch": 0.17513948536868174,
+      "grad_norm": 336.1890869140625,
+      "learning_rate": 8.209884736699681e-06,
+      "loss": 43.6772,
+      "step": 43350
+    },
+    {
+      "epoch": 0.17517988663404938,
+      "grad_norm": 1088.29736328125,
+      "learning_rate": 8.208803246818586e-06,
+      "loss": 57.5273,
+      "step": 43360
+    },
+    {
+      "epoch": 0.17522028789941702,
+      "grad_norm": 1386.392578125,
+      "learning_rate": 8.207721501627743e-06,
+      "loss": 57.2031,
+      "step": 43370
+    },
+    {
+      "epoch": 0.17526068916478463,
+      "grad_norm": 499.1271667480469,
+      "learning_rate": 8.20663950121322e-06,
+      "loss": 42.4209,
+      "step": 43380
+    },
+    {
+      "epoch": 0.17530109043015227,
+      "grad_norm": 660.9666748046875,
+      "learning_rate": 8.20555724566111e-06,
+      "loss": 54.9783,
+      "step": 43390
+    },
+    {
+      "epoch": 0.1753414916955199,
+      "grad_norm": 267.87103271484375,
+      "learning_rate": 8.204474735057522e-06,
+      "loss": 52.6259,
+      "step": 43400
+    },
+    {
+      "epoch": 0.17538189296088755,
+      "grad_norm": 808.8639526367188,
+      "learning_rate": 8.203391969488586e-06,
+      "loss": 40.2631,
+      "step": 43410
+    },
+    {
+      "epoch": 0.17542229422625516,
+      "grad_norm": 353.7890625,
+      "learning_rate": 8.20230894904045e-06,
+      "loss": 37.8534,
+      "step": 43420
+    },
+    {
+      "epoch": 0.1754626954916228,
+      "grad_norm": 859.0033569335938,
+      "learning_rate": 8.20122567379929e-06,
+      "loss": 48.4326,
+      "step": 43430
+    },
+    {
+      "epoch": 0.17550309675699044,
+      "grad_norm": 704.4515991210938,
+      "learning_rate": 8.200142143851295e-06,
+      "loss": 46.7167,
+      "step": 43440
+    },
+    {
+      "epoch": 0.17554349802235805,
+      "grad_norm": 651.6087646484375,
+      "learning_rate": 8.199058359282675e-06,
+      "loss": 52.8648,
+      "step": 43450
+    },
+    {
+      "epoch": 0.1755838992877257,
+      "grad_norm": 514.385498046875,
+      "learning_rate": 8.197974320179664e-06,
+      "loss": 49.6813,
+      "step": 43460
+    },
+    {
+      "epoch": 0.17562430055309333,
+      "grad_norm": 477.1377258300781,
+      "learning_rate": 8.19689002662851e-06,
+      "loss": 58.1234,
+      "step": 43470
+    },
+    {
+      "epoch": 0.17566470181846094,
+      "grad_norm": 570.377685546875,
+      "learning_rate": 8.195805478715492e-06,
+      "loss": 33.3943,
+      "step": 43480
+    },
+    {
+      "epoch": 0.17570510308382858,
+      "grad_norm": 436.5005798339844,
+      "learning_rate": 8.194720676526898e-06,
+      "loss": 42.9418,
+      "step": 43490
+    },
+    {
+      "epoch": 0.17574550434919622,
+      "grad_norm": 737.1203002929688,
+      "learning_rate": 8.193635620149041e-06,
+      "loss": 45.1717,
+      "step": 43500
+    },
+    {
+      "epoch": 0.17578590561456384,
+      "grad_norm": 1095.8653564453125,
+      "learning_rate": 8.192550309668254e-06,
+      "loss": 53.5207,
+      "step": 43510
+    },
+    {
+      "epoch": 0.17582630687993148,
+      "grad_norm": 298.76434326171875,
+      "learning_rate": 8.191464745170892e-06,
+      "loss": 33.359,
+      "step": 43520
+    },
+    {
+      "epoch": 0.17586670814529912,
+      "grad_norm": 531.721435546875,
+      "learning_rate": 8.190378926743327e-06,
+      "loss": 42.1844,
+      "step": 43530
+    },
+    {
+      "epoch": 0.17590710941066673,
+      "grad_norm": 881.6190185546875,
+      "learning_rate": 8.189292854471953e-06,
+      "loss": 49.1135,
+      "step": 43540
+    },
+    {
+      "epoch": 0.17594751067603437,
+      "grad_norm": 239.04486083984375,
+      "learning_rate": 8.188206528443182e-06,
+      "loss": 34.2073,
+      "step": 43550
+    },
+    {
+      "epoch": 0.175987911941402,
+      "grad_norm": 390.4149169921875,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 59.7111,
+      "step": 43560
+    },
+    {
+      "epoch": 0.17602831320676965,
+      "grad_norm": 306.6148986816406,
+      "learning_rate": 8.186033115459211e-06,
+      "loss": 41.3283,
+      "step": 43570
+    },
+    {
+      "epoch": 0.17606871447213726,
+      "grad_norm": 480.1251220703125,
+      "learning_rate": 8.184946028676937e-06,
+      "loss": 85.4738,
+      "step": 43580
+    },
+    {
+      "epoch": 0.1761091157375049,
+      "grad_norm": 458.3759765625,
+      "learning_rate": 8.183858688483126e-06,
+      "loss": 36.9035,
+      "step": 43590
+    },
+    {
+      "epoch": 0.17614951700287254,
+      "grad_norm": 878.4601440429688,
+      "learning_rate": 8.182771094964292e-06,
+      "loss": 53.6184,
+      "step": 43600
+    },
+    {
+      "epoch": 0.17618991826824015,
+      "grad_norm": 918.716796875,
+      "learning_rate": 8.181683248206968e-06,
+      "loss": 46.9538,
+      "step": 43610
+    },
+    {
+      "epoch": 0.1762303195336078,
+      "grad_norm": 734.4132690429688,
+      "learning_rate": 8.180595148297709e-06,
+      "loss": 51.9942,
+      "step": 43620
+    },
+    {
+      "epoch": 0.17627072079897543,
+      "grad_norm": 648.378662109375,
+      "learning_rate": 8.179506795323092e-06,
+      "loss": 45.4542,
+      "step": 43630
+    },
+    {
+      "epoch": 0.17631112206434305,
+      "grad_norm": 629.5515747070312,
+      "learning_rate": 8.17841818936971e-06,
+      "loss": 58.7435,
+      "step": 43640
+    },
+    {
+      "epoch": 0.17635152332971069,
+      "grad_norm": 1205.067626953125,
+      "learning_rate": 8.177329330524182e-06,
+      "loss": 41.3752,
+      "step": 43650
+    },
+    {
+      "epoch": 0.17639192459507833,
+      "grad_norm": 569.39404296875,
+      "learning_rate": 8.17624021887314e-06,
+      "loss": 39.0071,
+      "step": 43660
+    },
+    {
+      "epoch": 0.17643232586044594,
+      "grad_norm": 829.2988891601562,
+      "learning_rate": 8.17515085450324e-06,
+      "loss": 44.5112,
+      "step": 43670
+    },
+    {
+      "epoch": 0.17647272712581358,
+      "grad_norm": 292.11273193359375,
+      "learning_rate": 8.174061237501159e-06,
+      "loss": 41.0113,
+      "step": 43680
+    },
+    {
+      "epoch": 0.17651312839118122,
+      "grad_norm": 632.9418334960938,
+      "learning_rate": 8.172971367953593e-06,
+      "loss": 38.4412,
+      "step": 43690
+    },
+    {
+      "epoch": 0.17655352965654883,
+      "grad_norm": 539.7322998046875,
+      "learning_rate": 8.171881245947257e-06,
+      "loss": 33.8846,
+      "step": 43700
+    },
+    {
+      "epoch": 0.17659393092191647,
+      "grad_norm": 755.9725341796875,
+      "learning_rate": 8.170790871568887e-06,
+      "loss": 53.129,
+      "step": 43710
+    },
+    {
+      "epoch": 0.1766343321872841,
+      "grad_norm": 616.8136596679688,
+      "learning_rate": 8.169700244905239e-06,
+      "loss": 47.7439,
+      "step": 43720
+    },
+    {
+      "epoch": 0.17667473345265175,
+      "grad_norm": 508.9373474121094,
+      "learning_rate": 8.168609366043089e-06,
+      "loss": 42.9057,
+      "step": 43730
+    },
+    {
+      "epoch": 0.17671513471801936,
+      "grad_norm": 416.2492980957031,
+      "learning_rate": 8.167518235069234e-06,
+      "loss": 42.0221,
+      "step": 43740
+    },
+    {
+      "epoch": 0.176755535983387,
+      "grad_norm": 648.8447265625,
+      "learning_rate": 8.16642685207049e-06,
+      "loss": 71.3161,
+      "step": 43750
+    },
+    {
+      "epoch": 0.17679593724875464,
+      "grad_norm": 587.8074951171875,
+      "learning_rate": 8.165335217133695e-06,
+      "loss": 48.2666,
+      "step": 43760
+    },
+    {
+      "epoch": 0.17683633851412225,
+      "grad_norm": 752.8903198242188,
+      "learning_rate": 8.164243330345702e-06,
+      "loss": 44.9774,
+      "step": 43770
+    },
+    {
+      "epoch": 0.1768767397794899,
+      "grad_norm": 984.27734375,
+      "learning_rate": 8.16315119179339e-06,
+      "loss": 44.6318,
+      "step": 43780
+    },
+    {
+      "epoch": 0.17691714104485753,
+      "grad_norm": 620.818359375,
+      "learning_rate": 8.162058801563652e-06,
+      "loss": 49.2427,
+      "step": 43790
+    },
+    {
+      "epoch": 0.17695754231022515,
+      "grad_norm": 289.87615966796875,
+      "learning_rate": 8.160966159743411e-06,
+      "loss": 43.1271,
+      "step": 43800
+    },
+    {
+      "epoch": 0.1769979435755928,
+      "grad_norm": 828.4949951171875,
+      "learning_rate": 8.159873266419598e-06,
+      "loss": 68.2872,
+      "step": 43810
+    },
+    {
+      "epoch": 0.17703834484096043,
+      "grad_norm": 859.2119750976562,
+      "learning_rate": 8.15878012167917e-06,
+      "loss": 36.2162,
+      "step": 43820
+    },
+    {
+      "epoch": 0.17707874610632804,
+      "grad_norm": 583.408935546875,
+      "learning_rate": 8.157686725609105e-06,
+      "loss": 40.5383,
+      "step": 43830
+    },
+    {
+      "epoch": 0.17711914737169568,
+      "grad_norm": 480.7803649902344,
+      "learning_rate": 8.1565930782964e-06,
+      "loss": 39.6614,
+      "step": 43840
+    },
+    {
+      "epoch": 0.17715954863706332,
+      "grad_norm": 730.824462890625,
+      "learning_rate": 8.155499179828068e-06,
+      "loss": 47.8291,
+      "step": 43850
+    },
+    {
+      "epoch": 0.17719994990243093,
+      "grad_norm": 521.2890014648438,
+      "learning_rate": 8.15440503029115e-06,
+      "loss": 44.0991,
+      "step": 43860
+    },
+    {
+      "epoch": 0.17724035116779857,
+      "grad_norm": 1121.678466796875,
+      "learning_rate": 8.153310629772702e-06,
+      "loss": 60.6592,
+      "step": 43870
+    },
+    {
+      "epoch": 0.1772807524331662,
+      "grad_norm": 459.24322509765625,
+      "learning_rate": 8.152215978359796e-06,
+      "loss": 62.694,
+      "step": 43880
+    },
+    {
+      "epoch": 0.17732115369853385,
+      "grad_norm": 771.0817260742188,
+      "learning_rate": 8.151121076139534e-06,
+      "loss": 49.6407,
+      "step": 43890
+    },
+    {
+      "epoch": 0.17736155496390146,
+      "grad_norm": 378.06890869140625,
+      "learning_rate": 8.150025923199027e-06,
+      "loss": 45.2549,
+      "step": 43900
+    },
+    {
+      "epoch": 0.1774019562292691,
+      "grad_norm": 655.2774047851562,
+      "learning_rate": 8.148930519625417e-06,
+      "loss": 33.8218,
+      "step": 43910
+    },
+    {
+      "epoch": 0.17744235749463674,
+      "grad_norm": 415.8019104003906,
+      "learning_rate": 8.147834865505855e-06,
+      "loss": 42.6838,
+      "step": 43920
+    },
+    {
+      "epoch": 0.17748275876000436,
+      "grad_norm": 436.5208740234375,
+      "learning_rate": 8.14673896092752e-06,
+      "loss": 42.1472,
+      "step": 43930
+    },
+    {
+      "epoch": 0.177523160025372,
+      "grad_norm": 785.6007080078125,
+      "learning_rate": 8.145642805977608e-06,
+      "loss": 50.6509,
+      "step": 43940
+    },
+    {
+      "epoch": 0.17756356129073964,
+      "grad_norm": 644.1143798828125,
+      "learning_rate": 8.144546400743334e-06,
+      "loss": 40.0659,
+      "step": 43950
+    },
+    {
+      "epoch": 0.17760396255610725,
+      "grad_norm": 288.6166687011719,
+      "learning_rate": 8.143449745311934e-06,
+      "loss": 65.2577,
+      "step": 43960
+    },
+    {
+      "epoch": 0.1776443638214749,
+      "grad_norm": 741.4886474609375,
+      "learning_rate": 8.142352839770663e-06,
+      "loss": 45.3209,
+      "step": 43970
+    },
+    {
+      "epoch": 0.17768476508684253,
+      "grad_norm": 541.5047607421875,
+      "learning_rate": 8.1412556842068e-06,
+      "loss": 44.5242,
+      "step": 43980
+    },
+    {
+      "epoch": 0.17772516635221014,
+      "grad_norm": 752.431640625,
+      "learning_rate": 8.140158278707637e-06,
+      "loss": 46.5009,
+      "step": 43990
+    },
+    {
+      "epoch": 0.17776556761757778,
+      "grad_norm": 816.7749633789062,
+      "learning_rate": 8.139060623360494e-06,
+      "loss": 48.2635,
+      "step": 44000
+    },
+    {
+      "epoch": 0.17780596888294542,
+      "grad_norm": 699.2485961914062,
+      "learning_rate": 8.1379627182527e-06,
+      "loss": 40.0845,
+      "step": 44010
+    },
+    {
+      "epoch": 0.17784637014831303,
+      "grad_norm": 676.007080078125,
+      "learning_rate": 8.136864563471617e-06,
+      "loss": 80.3815,
+      "step": 44020
+    },
+    {
+      "epoch": 0.17788677141368067,
+      "grad_norm": 200.27894592285156,
+      "learning_rate": 8.135766159104615e-06,
+      "loss": 48.3457,
+      "step": 44030
+    },
+    {
+      "epoch": 0.1779271726790483,
+      "grad_norm": 359.0489807128906,
+      "learning_rate": 8.134667505239092e-06,
+      "loss": 66.8701,
+      "step": 44040
+    },
+    {
+      "epoch": 0.17796757394441595,
+      "grad_norm": 601.3418579101562,
+      "learning_rate": 8.133568601962462e-06,
+      "loss": 22.5121,
+      "step": 44050
+    },
+    {
+      "epoch": 0.17800797520978356,
+      "grad_norm": 754.0125732421875,
+      "learning_rate": 8.132469449362158e-06,
+      "loss": 63.1259,
+      "step": 44060
+    },
+    {
+      "epoch": 0.1780483764751512,
+      "grad_norm": 632.214111328125,
+      "learning_rate": 8.131370047525637e-06,
+      "loss": 41.4233,
+      "step": 44070
+    },
+    {
+      "epoch": 0.17808877774051884,
+      "grad_norm": 370.5418701171875,
+      "learning_rate": 8.130270396540372e-06,
+      "loss": 58.6016,
+      "step": 44080
+    },
+    {
+      "epoch": 0.17812917900588646,
+      "grad_norm": 385.2149963378906,
+      "learning_rate": 8.129170496493857e-06,
+      "loss": 41.7259,
+      "step": 44090
+    },
+    {
+      "epoch": 0.1781695802712541,
+      "grad_norm": 492.43988037109375,
+      "learning_rate": 8.128070347473609e-06,
+      "loss": 140.3884,
+      "step": 44100
+    },
+    {
+      "epoch": 0.17820998153662174,
+      "grad_norm": 308.92840576171875,
+      "learning_rate": 8.126969949567157e-06,
+      "loss": 44.9308,
+      "step": 44110
+    },
+    {
+      "epoch": 0.17825038280198935,
+      "grad_norm": 822.5704345703125,
+      "learning_rate": 8.125869302862058e-06,
+      "loss": 40.5698,
+      "step": 44120
+    },
+    {
+      "epoch": 0.178290784067357,
+      "grad_norm": 576.1128540039062,
+      "learning_rate": 8.124768407445883e-06,
+      "loss": 65.7646,
+      "step": 44130
+    },
+    {
+      "epoch": 0.17833118533272463,
+      "grad_norm": 445.1959228515625,
+      "learning_rate": 8.123667263406228e-06,
+      "loss": 47.3017,
+      "step": 44140
+    },
+    {
+      "epoch": 0.17837158659809224,
+      "grad_norm": 853.5143432617188,
+      "learning_rate": 8.122565870830704e-06,
+      "loss": 48.5397,
+      "step": 44150
+    },
+    {
+      "epoch": 0.17841198786345988,
+      "grad_norm": 609.802734375,
+      "learning_rate": 8.121464229806944e-06,
+      "loss": 35.6042,
+      "step": 44160
+    },
+    {
+      "epoch": 0.17845238912882752,
+      "grad_norm": 701.1467895507812,
+      "learning_rate": 8.120362340422601e-06,
+      "loss": 45.621,
+      "step": 44170
+    },
+    {
+      "epoch": 0.17849279039419513,
+      "grad_norm": 717.2283325195312,
+      "learning_rate": 8.119260202765347e-06,
+      "loss": 39.1522,
+      "step": 44180
+    },
+    {
+      "epoch": 0.17853319165956277,
+      "grad_norm": 816.3579711914062,
+      "learning_rate": 8.118157816922874e-06,
+      "loss": 53.5569,
+      "step": 44190
+    },
+    {
+      "epoch": 0.1785735929249304,
+      "grad_norm": 711.2218627929688,
+      "learning_rate": 8.117055182982895e-06,
+      "loss": 35.55,
+      "step": 44200
+    },
+    {
+      "epoch": 0.17861399419029805,
+      "grad_norm": 971.1217651367188,
+      "learning_rate": 8.115952301033141e-06,
+      "loss": 62.0396,
+      "step": 44210
+    },
+    {
+      "epoch": 0.17865439545566567,
+      "grad_norm": 474.5082092285156,
+      "learning_rate": 8.11484917116136e-06,
+      "loss": 61.6075,
+      "step": 44220
+    },
+    {
+      "epoch": 0.1786947967210333,
+      "grad_norm": 424.6689453125,
+      "learning_rate": 8.113745793455328e-06,
+      "loss": 41.4968,
+      "step": 44230
+    },
+    {
+      "epoch": 0.17873519798640095,
+      "grad_norm": 1078.0762939453125,
+      "learning_rate": 8.112642168002831e-06,
+      "loss": 39.6164,
+      "step": 44240
+    },
+    {
+      "epoch": 0.17877559925176856,
+      "grad_norm": 1485.0335693359375,
+      "learning_rate": 8.111538294891684e-06,
+      "loss": 43.4463,
+      "step": 44250
+    },
+    {
+      "epoch": 0.1788160005171362,
+      "grad_norm": 733.3353881835938,
+      "learning_rate": 8.110434174209714e-06,
+      "loss": 58.078,
+      "step": 44260
+    },
+    {
+      "epoch": 0.17885640178250384,
+      "grad_norm": 1062.302001953125,
+      "learning_rate": 8.109329806044772e-06,
+      "loss": 78.6044,
+      "step": 44270
+    },
+    {
+      "epoch": 0.17889680304787145,
+      "grad_norm": 706.5615844726562,
+      "learning_rate": 8.108225190484728e-06,
+      "loss": 49.5474,
+      "step": 44280
+    },
+    {
+      "epoch": 0.1789372043132391,
+      "grad_norm": 834.2645263671875,
+      "learning_rate": 8.107120327617469e-06,
+      "loss": 33.1368,
+      "step": 44290
+    },
+    {
+      "epoch": 0.17897760557860673,
+      "grad_norm": 1417.78173828125,
+      "learning_rate": 8.106015217530906e-06,
+      "loss": 57.1304,
+      "step": 44300
+    },
+    {
+      "epoch": 0.17901800684397434,
+      "grad_norm": 302.16729736328125,
+      "learning_rate": 8.104909860312968e-06,
+      "loss": 39.4532,
+      "step": 44310
+    },
+    {
+      "epoch": 0.17905840810934198,
+      "grad_norm": 737.5609130859375,
+      "learning_rate": 8.1038042560516e-06,
+      "loss": 42.557,
+      "step": 44320
+    },
+    {
+      "epoch": 0.17909880937470962,
+      "grad_norm": 407.8067626953125,
+      "learning_rate": 8.102698404834773e-06,
+      "loss": 48.2338,
+      "step": 44330
+    },
+    {
+      "epoch": 0.17913921064007723,
+      "grad_norm": 584.7301025390625,
+      "learning_rate": 8.101592306750472e-06,
+      "loss": 50.2676,
+      "step": 44340
+    },
+    {
+      "epoch": 0.17917961190544487,
+      "grad_norm": 549.4971923828125,
+      "learning_rate": 8.100485961886707e-06,
+      "loss": 53.5906,
+      "step": 44350
+    },
+    {
+      "epoch": 0.17922001317081251,
+      "grad_norm": 341.3735656738281,
+      "learning_rate": 8.099379370331502e-06,
+      "loss": 41.9313,
+      "step": 44360
+    },
+    {
+      "epoch": 0.17926041443618015,
+      "grad_norm": 894.8284912109375,
+      "learning_rate": 8.098272532172906e-06,
+      "loss": 38.5997,
+      "step": 44370
+    },
+    {
+      "epoch": 0.17930081570154777,
+      "grad_norm": 509.7801513671875,
+      "learning_rate": 8.097165447498985e-06,
+      "loss": 53.9619,
+      "step": 44380
+    },
+    {
+      "epoch": 0.1793412169669154,
+      "grad_norm": 643.4464111328125,
+      "learning_rate": 8.09605811639782e-06,
+      "loss": 39.9498,
+      "step": 44390
+    },
+    {
+      "epoch": 0.17938161823228305,
+      "grad_norm": 721.2262573242188,
+      "learning_rate": 8.094950538957523e-06,
+      "loss": 33.5365,
+      "step": 44400
+    },
+    {
+      "epoch": 0.17942201949765066,
+      "grad_norm": 410.3100891113281,
+      "learning_rate": 8.093842715266214e-06,
+      "loss": 54.1215,
+      "step": 44410
+    },
+    {
+      "epoch": 0.1794624207630183,
+      "grad_norm": 634.4172973632812,
+      "learning_rate": 8.092734645412037e-06,
+      "loss": 61.592,
+      "step": 44420
+    },
+    {
+      "epoch": 0.17950282202838594,
+      "grad_norm": 538.3599243164062,
+      "learning_rate": 8.09162632948316e-06,
+      "loss": 39.2725,
+      "step": 44430
+    },
+    {
+      "epoch": 0.17954322329375355,
+      "grad_norm": 1388.5975341796875,
+      "learning_rate": 8.090517767567765e-06,
+      "loss": 61.4013,
+      "step": 44440
+    },
+    {
+      "epoch": 0.1795836245591212,
+      "grad_norm": 555.4326782226562,
+      "learning_rate": 8.089408959754055e-06,
+      "loss": 42.8655,
+      "step": 44450
+    },
+    {
+      "epoch": 0.17962402582448883,
+      "grad_norm": 727.263427734375,
+      "learning_rate": 8.088299906130252e-06,
+      "loss": 43.4522,
+      "step": 44460
+    },
+    {
+      "epoch": 0.17966442708985644,
+      "grad_norm": 462.22186279296875,
+      "learning_rate": 8.087190606784598e-06,
+      "loss": 48.6978,
+      "step": 44470
+    },
+    {
+      "epoch": 0.17970482835522408,
+      "grad_norm": 723.4546508789062,
+      "learning_rate": 8.086081061805357e-06,
+      "loss": 44.1117,
+      "step": 44480
+    },
+    {
+      "epoch": 0.17974522962059172,
+      "grad_norm": 670.451171875,
+      "learning_rate": 8.084971271280808e-06,
+      "loss": 67.2479,
+      "step": 44490
+    },
+    {
+      "epoch": 0.17978563088595934,
+      "grad_norm": 477.13946533203125,
+      "learning_rate": 8.083861235299253e-06,
+      "loss": 36.4865,
+      "step": 44500
+    },
+    {
+      "epoch": 0.17982603215132698,
+      "grad_norm": 683.2493286132812,
+      "learning_rate": 8.082750953949015e-06,
+      "loss": 52.2884,
+      "step": 44510
+    },
+    {
+      "epoch": 0.17986643341669462,
+      "grad_norm": 832.3070068359375,
+      "learning_rate": 8.081640427318429e-06,
+      "loss": 59.4732,
+      "step": 44520
+    },
+    {
+      "epoch": 0.17990683468206226,
+      "grad_norm": 610.156494140625,
+      "learning_rate": 8.080529655495856e-06,
+      "loss": 37.1259,
+      "step": 44530
+    },
+    {
+      "epoch": 0.17994723594742987,
+      "grad_norm": 602.242431640625,
+      "learning_rate": 8.079418638569679e-06,
+      "loss": 54.5028,
+      "step": 44540
+    },
+    {
+      "epoch": 0.1799876372127975,
+      "grad_norm": 884.65869140625,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 28.0929,
+      "step": 44550
+    },
+    {
+      "epoch": 0.18002803847816515,
+      "grad_norm": 436.0967102050781,
+      "learning_rate": 8.077195869760114e-06,
+      "loss": 45.9886,
+      "step": 44560
+    },
+    {
+      "epoch": 0.18006843974353276,
+      "grad_norm": 1172.73974609375,
+      "learning_rate": 8.076084118053584e-06,
+      "loss": 52.4016,
+      "step": 44570
+    },
+    {
+      "epoch": 0.1801088410089004,
+      "grad_norm": 488.72637939453125,
+      "learning_rate": 8.074972121597158e-06,
+      "loss": 56.7316,
+      "step": 44580
+    },
+    {
+      "epoch": 0.18014924227426804,
+      "grad_norm": 632.3367919921875,
+      "learning_rate": 8.073859880479314e-06,
+      "loss": 49.8892,
+      "step": 44590
+    },
+    {
+      "epoch": 0.18018964353963565,
+      "grad_norm": 302.1227722167969,
+      "learning_rate": 8.072747394788545e-06,
+      "loss": 34.2227,
+      "step": 44600
+    },
+    {
+      "epoch": 0.1802300448050033,
+      "grad_norm": 776.5565185546875,
+      "learning_rate": 8.071634664613367e-06,
+      "loss": 52.676,
+      "step": 44610
+    },
+    {
+      "epoch": 0.18027044607037093,
+      "grad_norm": 733.58251953125,
+      "learning_rate": 8.070521690042317e-06,
+      "loss": 60.9468,
+      "step": 44620
+    },
+    {
+      "epoch": 0.18031084733573854,
+      "grad_norm": 1145.7186279296875,
+      "learning_rate": 8.069408471163947e-06,
+      "loss": 65.979,
+      "step": 44630
+    },
+    {
+      "epoch": 0.18035124860110618,
+      "grad_norm": 509.70428466796875,
+      "learning_rate": 8.068295008066832e-06,
+      "loss": 40.7422,
+      "step": 44640
+    },
+    {
+      "epoch": 0.18039164986647382,
+      "grad_norm": 576.9971313476562,
+      "learning_rate": 8.067181300839565e-06,
+      "loss": 32.6827,
+      "step": 44650
+    },
+    {
+      "epoch": 0.18043205113184144,
+      "grad_norm": 710.1195678710938,
+      "learning_rate": 8.066067349570757e-06,
+      "loss": 58.6771,
+      "step": 44660
+    },
+    {
+      "epoch": 0.18047245239720908,
+      "grad_norm": 458.78106689453125,
+      "learning_rate": 8.064953154349042e-06,
+      "loss": 31.6616,
+      "step": 44670
+    },
+    {
+      "epoch": 0.18051285366257672,
+      "grad_norm": 411.84686279296875,
+      "learning_rate": 8.063838715263072e-06,
+      "loss": 33.9766,
+      "step": 44680
+    },
+    {
+      "epoch": 0.18055325492794436,
+      "grad_norm": 1309.62255859375,
+      "learning_rate": 8.062724032401515e-06,
+      "loss": 51.9341,
+      "step": 44690
+    },
+    {
+      "epoch": 0.18059365619331197,
+      "grad_norm": 537.8983154296875,
+      "learning_rate": 8.061609105853062e-06,
+      "loss": 46.2525,
+      "step": 44700
+    },
+    {
+      "epoch": 0.1806340574586796,
+      "grad_norm": 1111.2152099609375,
+      "learning_rate": 8.060493935706425e-06,
+      "loss": 55.7999,
+      "step": 44710
+    },
+    {
+      "epoch": 0.18067445872404725,
+      "grad_norm": 578.8212890625,
+      "learning_rate": 8.059378522050332e-06,
+      "loss": 64.9187,
+      "step": 44720
+    },
+    {
+      "epoch": 0.18071485998941486,
+      "grad_norm": 549.1094360351562,
+      "learning_rate": 8.05826286497353e-06,
+      "loss": 41.3403,
+      "step": 44730
+    },
+    {
+      "epoch": 0.1807552612547825,
+      "grad_norm": 641.1884765625,
+      "learning_rate": 8.057146964564786e-06,
+      "loss": 52.3945,
+      "step": 44740
+    },
+    {
+      "epoch": 0.18079566252015014,
+      "grad_norm": 886.13330078125,
+      "learning_rate": 8.05603082091289e-06,
+      "loss": 42.8315,
+      "step": 44750
+    },
+    {
+      "epoch": 0.18083606378551775,
+      "grad_norm": 639.1670532226562,
+      "learning_rate": 8.054914434106647e-06,
+      "loss": 56.5641,
+      "step": 44760
+    },
+    {
+      "epoch": 0.1808764650508854,
+      "grad_norm": 548.1455078125,
+      "learning_rate": 8.053797804234882e-06,
+      "loss": 52.2677,
+      "step": 44770
+    },
+    {
+      "epoch": 0.18091686631625303,
+      "grad_norm": 410.5792541503906,
+      "learning_rate": 8.052680931386441e-06,
+      "loss": 30.4637,
+      "step": 44780
+    },
+    {
+      "epoch": 0.18095726758162065,
+      "grad_norm": 578.6219482421875,
+      "learning_rate": 8.051563815650187e-06,
+      "loss": 35.4263,
+      "step": 44790
+    },
+    {
+      "epoch": 0.18099766884698829,
+      "grad_norm": 409.59832763671875,
+      "learning_rate": 8.050446457115005e-06,
+      "loss": 56.8702,
+      "step": 44800
+    },
+    {
+      "epoch": 0.18103807011235593,
+      "grad_norm": 621.2161865234375,
+      "learning_rate": 8.0493288558698e-06,
+      "loss": 45.1692,
+      "step": 44810
+    },
+    {
+      "epoch": 0.18107847137772354,
+      "grad_norm": 284.7547912597656,
+      "learning_rate": 8.04821101200349e-06,
+      "loss": 32.8964,
+      "step": 44820
+    },
+    {
+      "epoch": 0.18111887264309118,
+      "grad_norm": 200.6179962158203,
+      "learning_rate": 8.047092925605022e-06,
+      "loss": 51.5494,
+      "step": 44830
+    },
+    {
+      "epoch": 0.18115927390845882,
+      "grad_norm": 628.4204711914062,
+      "learning_rate": 8.045974596763352e-06,
+      "loss": 34.833,
+      "step": 44840
+    },
+    {
+      "epoch": 0.18119967517382646,
+      "grad_norm": 1149.173583984375,
+      "learning_rate": 8.044856025567464e-06,
+      "loss": 48.7898,
+      "step": 44850
+    },
+    {
+      "epoch": 0.18124007643919407,
+      "grad_norm": 605.9561157226562,
+      "learning_rate": 8.043737212106356e-06,
+      "loss": 46.0257,
+      "step": 44860
+    },
+    {
+      "epoch": 0.1812804777045617,
+      "grad_norm": 735.3001708984375,
+      "learning_rate": 8.042618156469045e-06,
+      "loss": 52.739,
+      "step": 44870
+    },
+    {
+      "epoch": 0.18132087896992935,
+      "grad_norm": 996.4307250976562,
+      "learning_rate": 8.041498858744572e-06,
+      "loss": 61.982,
+      "step": 44880
+    },
+    {
+      "epoch": 0.18136128023529696,
+      "grad_norm": 677.4853515625,
+      "learning_rate": 8.040379319021994e-06,
+      "loss": 51.9419,
+      "step": 44890
+    },
+    {
+      "epoch": 0.1814016815006646,
+      "grad_norm": 691.8176879882812,
+      "learning_rate": 8.039259537390388e-06,
+      "loss": 47.6419,
+      "step": 44900
+    },
+    {
+      "epoch": 0.18144208276603224,
+      "grad_norm": 1121.185302734375,
+      "learning_rate": 8.038139513938847e-06,
+      "loss": 31.3486,
+      "step": 44910
+    },
+    {
+      "epoch": 0.18148248403139985,
+      "grad_norm": 721.5740966796875,
+      "learning_rate": 8.037019248756488e-06,
+      "loss": 39.1443,
+      "step": 44920
+    },
+    {
+      "epoch": 0.1815228852967675,
+      "grad_norm": 791.95654296875,
+      "learning_rate": 8.035898741932447e-06,
+      "loss": 41.8861,
+      "step": 44930
+    },
+    {
+      "epoch": 0.18156328656213513,
+      "grad_norm": 499.211669921875,
+      "learning_rate": 8.034777993555875e-06,
+      "loss": 49.4517,
+      "step": 44940
+    },
+    {
+      "epoch": 0.18160368782750275,
+      "grad_norm": 342.6352844238281,
+      "learning_rate": 8.033657003715945e-06,
+      "loss": 26.7375,
+      "step": 44950
+    },
+    {
+      "epoch": 0.1816440890928704,
+      "grad_norm": 562.2408447265625,
+      "learning_rate": 8.032535772501851e-06,
+      "loss": 44.3456,
+      "step": 44960
+    },
+    {
+      "epoch": 0.18168449035823803,
+      "grad_norm": 707.8630981445312,
+      "learning_rate": 8.031414300002802e-06,
+      "loss": 37.7815,
+      "step": 44970
+    },
+    {
+      "epoch": 0.18172489162360564,
+      "grad_norm": 434.7294921875,
+      "learning_rate": 8.03029258630803e-06,
+      "loss": 38.9817,
+      "step": 44980
+    },
+    {
+      "epoch": 0.18176529288897328,
+      "grad_norm": 336.34130859375,
+      "learning_rate": 8.029170631506785e-06,
+      "loss": 37.1305,
+      "step": 44990
+    },
+    {
+      "epoch": 0.18180569415434092,
+      "grad_norm": 495.4858093261719,
+      "learning_rate": 8.028048435688333e-06,
+      "loss": 50.3649,
+      "step": 45000
+    },
+    {
+      "epoch": 0.18184609541970856,
+      "grad_norm": 708.4286499023438,
+      "learning_rate": 8.026925998941965e-06,
+      "loss": 54.8248,
+      "step": 45010
+    },
+    {
+      "epoch": 0.18188649668507617,
+      "grad_norm": 1438.5784912109375,
+      "learning_rate": 8.025803321356989e-06,
+      "loss": 61.7399,
+      "step": 45020
+    },
+    {
+      "epoch": 0.1819268979504438,
+      "grad_norm": 389.3792724609375,
+      "learning_rate": 8.024680403022726e-06,
+      "loss": 47.3041,
+      "step": 45030
+    },
+    {
+      "epoch": 0.18196729921581145,
+      "grad_norm": 544.342041015625,
+      "learning_rate": 8.023557244028526e-06,
+      "loss": 36.6487,
+      "step": 45040
+    },
+    {
+      "epoch": 0.18200770048117906,
+      "grad_norm": 765.8178100585938,
+      "learning_rate": 8.022433844463752e-06,
+      "loss": 46.9147,
+      "step": 45050
+    },
+    {
+      "epoch": 0.1820481017465467,
+      "grad_norm": 607.2188110351562,
+      "learning_rate": 8.02131020441779e-06,
+      "loss": 48.6501,
+      "step": 45060
+    },
+    {
+      "epoch": 0.18208850301191434,
+      "grad_norm": 337.20623779296875,
+      "learning_rate": 8.02018632398004e-06,
+      "loss": 74.1569,
+      "step": 45070
+    },
+    {
+      "epoch": 0.18212890427728196,
+      "grad_norm": 698.919677734375,
+      "learning_rate": 8.019062203239923e-06,
+      "loss": 51.9308,
+      "step": 45080
+    },
+    {
+      "epoch": 0.1821693055426496,
+      "grad_norm": 455.0402526855469,
+      "learning_rate": 8.017937842286882e-06,
+      "loss": 46.6393,
+      "step": 45090
+    },
+    {
+      "epoch": 0.18220970680801724,
+      "grad_norm": 697.3123779296875,
+      "learning_rate": 8.01681324121038e-06,
+      "loss": 57.6411,
+      "step": 45100
+    },
+    {
+      "epoch": 0.18225010807338485,
+      "grad_norm": 883.2453002929688,
+      "learning_rate": 8.015688400099893e-06,
+      "loss": 51.9202,
+      "step": 45110
+    },
+    {
+      "epoch": 0.1822905093387525,
+      "grad_norm": 489.3941345214844,
+      "learning_rate": 8.014563319044919e-06,
+      "loss": 64.1878,
+      "step": 45120
+    },
+    {
+      "epoch": 0.18233091060412013,
+      "grad_norm": 485.7003479003906,
+      "learning_rate": 8.013437998134978e-06,
+      "loss": 38.5387,
+      "step": 45130
+    },
+    {
+      "epoch": 0.18237131186948774,
+      "grad_norm": 423.7239074707031,
+      "learning_rate": 8.012312437459604e-06,
+      "loss": 28.8249,
+      "step": 45140
+    },
+    {
+      "epoch": 0.18241171313485538,
+      "grad_norm": 503.5894775390625,
+      "learning_rate": 8.011186637108354e-06,
+      "loss": 46.0242,
+      "step": 45150
+    },
+    {
+      "epoch": 0.18245211440022302,
+      "grad_norm": 731.257080078125,
+      "learning_rate": 8.010060597170805e-06,
+      "loss": 58.772,
+      "step": 45160
+    },
+    {
+      "epoch": 0.18249251566559066,
+      "grad_norm": 326.68988037109375,
+      "learning_rate": 8.008934317736546e-06,
+      "loss": 34.3258,
+      "step": 45170
+    },
+    {
+      "epoch": 0.18253291693095827,
+      "grad_norm": 654.277099609375,
+      "learning_rate": 8.007807798895195e-06,
+      "loss": 44.5783,
+      "step": 45180
+    },
+    {
+      "epoch": 0.1825733181963259,
+      "grad_norm": 456.9231872558594,
+      "learning_rate": 8.00668104073638e-06,
+      "loss": 48.7379,
+      "step": 45190
+    },
+    {
+      "epoch": 0.18261371946169355,
+      "grad_norm": 581.152099609375,
+      "learning_rate": 8.005554043349753e-06,
+      "loss": 69.1283,
+      "step": 45200
+    },
+    {
+      "epoch": 0.18265412072706116,
+      "grad_norm": 472.1055908203125,
+      "learning_rate": 8.004426806824985e-06,
+      "loss": 40.6944,
+      "step": 45210
+    },
+    {
+      "epoch": 0.1826945219924288,
+      "grad_norm": 305.08331298828125,
+      "learning_rate": 8.003299331251764e-06,
+      "loss": 26.3944,
+      "step": 45220
+    },
+    {
+      "epoch": 0.18273492325779644,
+      "grad_norm": 631.8370971679688,
+      "learning_rate": 8.002171616719798e-06,
+      "loss": 45.73,
+      "step": 45230
+    },
+    {
+      "epoch": 0.18277532452316406,
+      "grad_norm": 581.5281372070312,
+      "learning_rate": 8.001043663318815e-06,
+      "loss": 44.2141,
+      "step": 45240
+    },
+    {
+      "epoch": 0.1828157257885317,
+      "grad_norm": 546.5279541015625,
+      "learning_rate": 7.999915471138562e-06,
+      "loss": 32.7438,
+      "step": 45250
+    },
+    {
+      "epoch": 0.18285612705389934,
+      "grad_norm": 790.43701171875,
+      "learning_rate": 7.9987870402688e-06,
+      "loss": 39.4014,
+      "step": 45260
+    },
+    {
+      "epoch": 0.18289652831926695,
+      "grad_norm": 1440.0347900390625,
+      "learning_rate": 7.997658370799318e-06,
+      "loss": 59.1243,
+      "step": 45270
+    },
+    {
+      "epoch": 0.1829369295846346,
+      "grad_norm": 826.537841796875,
+      "learning_rate": 7.996529462819915e-06,
+      "loss": 52.4962,
+      "step": 45280
+    },
+    {
+      "epoch": 0.18297733085000223,
+      "grad_norm": 734.6253662109375,
+      "learning_rate": 7.995400316420416e-06,
+      "loss": 38.2799,
+      "step": 45290
+    },
+    {
+      "epoch": 0.18301773211536984,
+      "grad_norm": 674.3758544921875,
+      "learning_rate": 7.994270931690662e-06,
+      "loss": 47.8541,
+      "step": 45300
+    },
+    {
+      "epoch": 0.18305813338073748,
+      "grad_norm": 521.4122314453125,
+      "learning_rate": 7.993141308720511e-06,
+      "loss": 36.1015,
+      "step": 45310
+    },
+    {
+      "epoch": 0.18309853464610512,
+      "grad_norm": 738.7835083007812,
+      "learning_rate": 7.99201144759984e-06,
+      "loss": 44.7773,
+      "step": 45320
+    },
+    {
+      "epoch": 0.18313893591147276,
+      "grad_norm": 1291.3709716796875,
+      "learning_rate": 7.990881348418554e-06,
+      "loss": 53.2255,
+      "step": 45330
+    },
+    {
+      "epoch": 0.18317933717684037,
+      "grad_norm": 493.1036071777344,
+      "learning_rate": 7.989751011266565e-06,
+      "loss": 42.3629,
+      "step": 45340
+    },
+    {
+      "epoch": 0.183219738442208,
+      "grad_norm": 657.96826171875,
+      "learning_rate": 7.988620436233806e-06,
+      "loss": 50.5465,
+      "step": 45350
+    },
+    {
+      "epoch": 0.18326013970757565,
+      "grad_norm": 1039.5506591796875,
+      "learning_rate": 7.987489623410236e-06,
+      "loss": 66.5563,
+      "step": 45360
+    },
+    {
+      "epoch": 0.18330054097294327,
+      "grad_norm": 375.0739440917969,
+      "learning_rate": 7.986358572885828e-06,
+      "loss": 37.0097,
+      "step": 45370
+    },
+    {
+      "epoch": 0.1833409422383109,
+      "grad_norm": 509.0332946777344,
+      "learning_rate": 7.985227284750574e-06,
+      "loss": 42.3172,
+      "step": 45380
+    },
+    {
+      "epoch": 0.18338134350367855,
+      "grad_norm": 906.0650024414062,
+      "learning_rate": 7.984095759094485e-06,
+      "loss": 54.2499,
+      "step": 45390
+    },
+    {
+      "epoch": 0.18342174476904616,
+      "grad_norm": 832.644287109375,
+      "learning_rate": 7.982963996007591e-06,
+      "loss": 59.8838,
+      "step": 45400
+    },
+    {
+      "epoch": 0.1834621460344138,
+      "grad_norm": 371.72076416015625,
+      "learning_rate": 7.981831995579943e-06,
+      "loss": 48.7457,
+      "step": 45410
+    },
+    {
+      "epoch": 0.18350254729978144,
+      "grad_norm": 285.2592468261719,
+      "learning_rate": 7.980699757901607e-06,
+      "loss": 46.093,
+      "step": 45420
+    },
+    {
+      "epoch": 0.18354294856514905,
+      "grad_norm": 712.9854736328125,
+      "learning_rate": 7.97956728306267e-06,
+      "loss": 33.3265,
+      "step": 45430
+    },
+    {
+      "epoch": 0.1835833498305167,
+      "grad_norm": 175.07998657226562,
+      "learning_rate": 7.97843457115324e-06,
+      "loss": 33.9601,
+      "step": 45440
+    },
+    {
+      "epoch": 0.18362375109588433,
+      "grad_norm": 219.32000732421875,
+      "learning_rate": 7.97730162226344e-06,
+      "loss": 51.4527,
+      "step": 45450
+    },
+    {
+      "epoch": 0.18366415236125194,
+      "grad_norm": 399.9309997558594,
+      "learning_rate": 7.976168436483415e-06,
+      "loss": 66.9279,
+      "step": 45460
+    },
+    {
+      "epoch": 0.18370455362661958,
+      "grad_norm": 1032.3216552734375,
+      "learning_rate": 7.975035013903326e-06,
+      "loss": 46.2986,
+      "step": 45470
+    },
+    {
+      "epoch": 0.18374495489198722,
+      "grad_norm": 759.1256103515625,
+      "learning_rate": 7.973901354613353e-06,
+      "loss": 47.9423,
+      "step": 45480
+    },
+    {
+      "epoch": 0.18378535615735486,
+      "grad_norm": 829.003173828125,
+      "learning_rate": 7.972767458703697e-06,
+      "loss": 51.7219,
+      "step": 45490
+    },
+    {
+      "epoch": 0.18382575742272247,
+      "grad_norm": 728.3862915039062,
+      "learning_rate": 7.971633326264581e-06,
+      "loss": 64.3442,
+      "step": 45500
+    },
+    {
+      "epoch": 0.18386615868809011,
+      "grad_norm": 699.3826293945312,
+      "learning_rate": 7.970498957386237e-06,
+      "loss": 43.2976,
+      "step": 45510
+    },
+    {
+      "epoch": 0.18390655995345775,
+      "grad_norm": 634.3785400390625,
+      "learning_rate": 7.969364352158922e-06,
+      "loss": 40.1747,
+      "step": 45520
+    },
+    {
+      "epoch": 0.18394696121882537,
+      "grad_norm": 837.7490844726562,
+      "learning_rate": 7.968229510672915e-06,
+      "loss": 53.3326,
+      "step": 45530
+    },
+    {
+      "epoch": 0.183987362484193,
+      "grad_norm": 896.8665771484375,
+      "learning_rate": 7.967094433018508e-06,
+      "loss": 48.5009,
+      "step": 45540
+    },
+    {
+      "epoch": 0.18402776374956065,
+      "grad_norm": 616.2903442382812,
+      "learning_rate": 7.965959119286013e-06,
+      "loss": 52.8024,
+      "step": 45550
+    },
+    {
+      "epoch": 0.18406816501492826,
+      "grad_norm": 555.3331909179688,
+      "learning_rate": 7.964823569565765e-06,
+      "loss": 58.4551,
+      "step": 45560
+    },
+    {
+      "epoch": 0.1841085662802959,
+      "grad_norm": 592.5578002929688,
+      "learning_rate": 7.963687783948111e-06,
+      "loss": 42.5069,
+      "step": 45570
+    },
+    {
+      "epoch": 0.18414896754566354,
+      "grad_norm": 665.7092895507812,
+      "learning_rate": 7.96255176252342e-06,
+      "loss": 69.5754,
+      "step": 45580
+    },
+    {
+      "epoch": 0.18418936881103115,
+      "grad_norm": 947.595703125,
+      "learning_rate": 7.961415505382083e-06,
+      "loss": 39.7731,
+      "step": 45590
+    },
+    {
+      "epoch": 0.1842297700763988,
+      "grad_norm": 1249.362548828125,
+      "learning_rate": 7.960279012614508e-06,
+      "loss": 57.2685,
+      "step": 45600
+    },
+    {
+      "epoch": 0.18427017134176643,
+      "grad_norm": 501.44244384765625,
+      "learning_rate": 7.959142284311115e-06,
+      "loss": 41.9633,
+      "step": 45610
+    },
+    {
+      "epoch": 0.18431057260713404,
+      "grad_norm": 530.5111083984375,
+      "learning_rate": 7.958005320562349e-06,
+      "loss": 65.3412,
+      "step": 45620
+    },
+    {
+      "epoch": 0.18435097387250168,
+      "grad_norm": 804.0465087890625,
+      "learning_rate": 7.95686812145868e-06,
+      "loss": 76.4228,
+      "step": 45630
+    },
+    {
+      "epoch": 0.18439137513786932,
+      "grad_norm": 816.2349243164062,
+      "learning_rate": 7.955730687090582e-06,
+      "loss": 48.9861,
+      "step": 45640
+    },
+    {
+      "epoch": 0.18443177640323694,
+      "grad_norm": 346.8124084472656,
+      "learning_rate": 7.954593017548557e-06,
+      "loss": 43.3233,
+      "step": 45650
+    },
+    {
+      "epoch": 0.18447217766860458,
+      "grad_norm": 787.0435791015625,
+      "learning_rate": 7.953455112923127e-06,
+      "loss": 50.9048,
+      "step": 45660
+    },
+    {
+      "epoch": 0.18451257893397222,
+      "grad_norm": 842.18212890625,
+      "learning_rate": 7.952316973304828e-06,
+      "loss": 49.7624,
+      "step": 45670
+    },
+    {
+      "epoch": 0.18455298019933986,
+      "grad_norm": 945.1439208984375,
+      "learning_rate": 7.951178598784217e-06,
+      "loss": 34.3345,
+      "step": 45680
+    },
+    {
+      "epoch": 0.18459338146470747,
+      "grad_norm": 454.4792175292969,
+      "learning_rate": 7.950039989451868e-06,
+      "loss": 59.2897,
+      "step": 45690
+    },
+    {
+      "epoch": 0.1846337827300751,
+      "grad_norm": 581.51806640625,
+      "learning_rate": 7.948901145398376e-06,
+      "loss": 36.0325,
+      "step": 45700
+    },
+    {
+      "epoch": 0.18467418399544275,
+      "grad_norm": 453.9153747558594,
+      "learning_rate": 7.947762066714353e-06,
+      "loss": 39.5895,
+      "step": 45710
+    },
+    {
+      "epoch": 0.18471458526081036,
+      "grad_norm": 727.990234375,
+      "learning_rate": 7.946622753490433e-06,
+      "loss": 35.5361,
+      "step": 45720
+    },
+    {
+      "epoch": 0.184754986526178,
+      "grad_norm": 576.713623046875,
+      "learning_rate": 7.945483205817262e-06,
+      "loss": 36.7247,
+      "step": 45730
+    },
+    {
+      "epoch": 0.18479538779154564,
+      "grad_norm": 1433.50830078125,
+      "learning_rate": 7.94434342378551e-06,
+      "loss": 35.3859,
+      "step": 45740
+    },
+    {
+      "epoch": 0.18483578905691325,
+      "grad_norm": 90.16148376464844,
+      "learning_rate": 7.943203407485864e-06,
+      "loss": 49.1724,
+      "step": 45750
+    },
+    {
+      "epoch": 0.1848761903222809,
+      "grad_norm": 442.8776550292969,
+      "learning_rate": 7.942063157009033e-06,
+      "loss": 56.5173,
+      "step": 45760
+    },
+    {
+      "epoch": 0.18491659158764853,
+      "grad_norm": 409.0257568359375,
+      "learning_rate": 7.940922672445737e-06,
+      "loss": 37.2671,
+      "step": 45770
+    },
+    {
+      "epoch": 0.18495699285301614,
+      "grad_norm": 1186.24560546875,
+      "learning_rate": 7.939781953886722e-06,
+      "loss": 53.6409,
+      "step": 45780
+    },
+    {
+      "epoch": 0.18499739411838378,
+      "grad_norm": 218.8214111328125,
+      "learning_rate": 7.938641001422747e-06,
+      "loss": 32.4635,
+      "step": 45790
+    },
+    {
+      "epoch": 0.18503779538375142,
+      "grad_norm": 342.5951232910156,
+      "learning_rate": 7.937499815144597e-06,
+      "loss": 50.4769,
+      "step": 45800
+    },
+    {
+      "epoch": 0.18507819664911904,
+      "grad_norm": 891.3570556640625,
+      "learning_rate": 7.936358395143065e-06,
+      "loss": 67.8622,
+      "step": 45810
+    },
+    {
+      "epoch": 0.18511859791448668,
+      "grad_norm": 768.8310546875,
+      "learning_rate": 7.935216741508971e-06,
+      "loss": 57.9334,
+      "step": 45820
+    },
+    {
+      "epoch": 0.18515899917985432,
+      "grad_norm": 299.90533447265625,
+      "learning_rate": 7.934074854333153e-06,
+      "loss": 31.2045,
+      "step": 45830
+    },
+    {
+      "epoch": 0.18519940044522196,
+      "grad_norm": 343.8869323730469,
+      "learning_rate": 7.932932733706467e-06,
+      "loss": 50.6509,
+      "step": 45840
+    },
+    {
+      "epoch": 0.18523980171058957,
+      "grad_norm": 722.1644897460938,
+      "learning_rate": 7.931790379719781e-06,
+      "loss": 51.3919,
+      "step": 45850
+    },
+    {
+      "epoch": 0.1852802029759572,
+      "grad_norm": 559.4667358398438,
+      "learning_rate": 7.93064779246399e-06,
+      "loss": 51.729,
+      "step": 45860
+    },
+    {
+      "epoch": 0.18532060424132485,
+      "grad_norm": 623.2672119140625,
+      "learning_rate": 7.929504972030003e-06,
+      "loss": 36.6839,
+      "step": 45870
+    },
+    {
+      "epoch": 0.18536100550669246,
+      "grad_norm": 706.392578125,
+      "learning_rate": 7.928361918508752e-06,
+      "loss": 43.8657,
+      "step": 45880
+    },
+    {
+      "epoch": 0.1854014067720601,
+      "grad_norm": 499.6767578125,
+      "learning_rate": 7.927218631991182e-06,
+      "loss": 34.8604,
+      "step": 45890
+    },
+    {
+      "epoch": 0.18544180803742774,
+      "grad_norm": 524.6019897460938,
+      "learning_rate": 7.92607511256826e-06,
+      "loss": 39.3707,
+      "step": 45900
+    },
+    {
+      "epoch": 0.18548220930279535,
+      "grad_norm": 429.8359680175781,
+      "learning_rate": 7.924931360330968e-06,
+      "loss": 47.6919,
+      "step": 45910
+    },
+    {
+      "epoch": 0.185522610568163,
+      "grad_norm": 569.885498046875,
+      "learning_rate": 7.92378737537031e-06,
+      "loss": 44.3475,
+      "step": 45920
+    },
+    {
+      "epoch": 0.18556301183353063,
+      "grad_norm": 947.2518310546875,
+      "learning_rate": 7.922643157777314e-06,
+      "loss": 53.1495,
+      "step": 45930
+    },
+    {
+      "epoch": 0.18560341309889825,
+      "grad_norm": 860.892578125,
+      "learning_rate": 7.921498707643011e-06,
+      "loss": 60.5503,
+      "step": 45940
+    },
+    {
+      "epoch": 0.18564381436426589,
+      "grad_norm": 1184.7371826171875,
+      "learning_rate": 7.920354025058467e-06,
+      "loss": 54.1781,
+      "step": 45950
+    },
+    {
+      "epoch": 0.18568421562963353,
+      "grad_norm": 666.4442138671875,
+      "learning_rate": 7.919209110114752e-06,
+      "loss": 46.4685,
+      "step": 45960
+    },
+    {
+      "epoch": 0.18572461689500114,
+      "grad_norm": 800.7621459960938,
+      "learning_rate": 7.918063962902968e-06,
+      "loss": 43.6424,
+      "step": 45970
+    },
+    {
+      "epoch": 0.18576501816036878,
+      "grad_norm": 602.9782104492188,
+      "learning_rate": 7.916918583514227e-06,
+      "loss": 38.4737,
+      "step": 45980
+    },
+    {
+      "epoch": 0.18580541942573642,
+      "grad_norm": 361.55755615234375,
+      "learning_rate": 7.91577297203966e-06,
+      "loss": 46.0553,
+      "step": 45990
+    },
+    {
+      "epoch": 0.18584582069110406,
+      "grad_norm": 915.6522216796875,
+      "learning_rate": 7.91462712857042e-06,
+      "loss": 46.0892,
+      "step": 46000
+    },
+    {
+      "epoch": 0.18588622195647167,
+      "grad_norm": 965.8405151367188,
+      "learning_rate": 7.913481053197673e-06,
+      "loss": 33.8431,
+      "step": 46010
+    },
+    {
+      "epoch": 0.1859266232218393,
+      "grad_norm": 701.4907836914062,
+      "learning_rate": 7.912334746012613e-06,
+      "loss": 31.4115,
+      "step": 46020
+    },
+    {
+      "epoch": 0.18596702448720695,
+      "grad_norm": 470.70623779296875,
+      "learning_rate": 7.911188207106442e-06,
+      "loss": 29.5224,
+      "step": 46030
+    },
+    {
+      "epoch": 0.18600742575257456,
+      "grad_norm": 877.6990966796875,
+      "learning_rate": 7.910041436570386e-06,
+      "loss": 56.7888,
+      "step": 46040
+    },
+    {
+      "epoch": 0.1860478270179422,
+      "grad_norm": 787.839111328125,
+      "learning_rate": 7.90889443449569e-06,
+      "loss": 69.0265,
+      "step": 46050
+    },
+    {
+      "epoch": 0.18608822828330984,
+      "grad_norm": 740.5379638671875,
+      "learning_rate": 7.90774720097361e-06,
+      "loss": 49.0101,
+      "step": 46060
+    },
+    {
+      "epoch": 0.18612862954867745,
+      "grad_norm": 436.9485778808594,
+      "learning_rate": 7.906599736095433e-06,
+      "loss": 62.774,
+      "step": 46070
+    },
+    {
+      "epoch": 0.1861690308140451,
+      "grad_norm": 461.7222595214844,
+      "learning_rate": 7.905452039952453e-06,
+      "loss": 51.2547,
+      "step": 46080
+    },
+    {
+      "epoch": 0.18620943207941273,
+      "grad_norm": 967.5213623046875,
+      "learning_rate": 7.904304112635987e-06,
+      "loss": 46.2335,
+      "step": 46090
+    },
+    {
+      "epoch": 0.18624983334478035,
+      "grad_norm": 642.5325317382812,
+      "learning_rate": 7.903155954237375e-06,
+      "loss": 46.4207,
+      "step": 46100
+    },
+    {
+      "epoch": 0.186290234610148,
+      "grad_norm": 1148.9150390625,
+      "learning_rate": 7.902007564847967e-06,
+      "loss": 67.827,
+      "step": 46110
+    },
+    {
+      "epoch": 0.18633063587551563,
+      "grad_norm": 737.8858032226562,
+      "learning_rate": 7.900858944559133e-06,
+      "loss": 52.1405,
+      "step": 46120
+    },
+    {
+      "epoch": 0.18637103714088324,
+      "grad_norm": 762.6358642578125,
+      "learning_rate": 7.899710093462267e-06,
+      "loss": 38.3844,
+      "step": 46130
+    },
+    {
+      "epoch": 0.18641143840625088,
+      "grad_norm": 412.6938781738281,
+      "learning_rate": 7.898561011648777e-06,
+      "loss": 53.3843,
+      "step": 46140
+    },
+    {
+      "epoch": 0.18645183967161852,
+      "grad_norm": 606.4049072265625,
+      "learning_rate": 7.89741169921009e-06,
+      "loss": 40.317,
+      "step": 46150
+    },
+    {
+      "epoch": 0.18649224093698616,
+      "grad_norm": 684.5941772460938,
+      "learning_rate": 7.896262156237652e-06,
+      "loss": 53.5236,
+      "step": 46160
+    },
+    {
+      "epoch": 0.18653264220235377,
+      "grad_norm": 751.0098266601562,
+      "learning_rate": 7.895112382822925e-06,
+      "loss": 71.0752,
+      "step": 46170
+    },
+    {
+      "epoch": 0.1865730434677214,
+      "grad_norm": 555.09521484375,
+      "learning_rate": 7.893962379057393e-06,
+      "loss": 41.15,
+      "step": 46180
+    },
+    {
+      "epoch": 0.18661344473308905,
+      "grad_norm": 830.2692260742188,
+      "learning_rate": 7.892812145032557e-06,
+      "loss": 41.5861,
+      "step": 46190
+    },
+    {
+      "epoch": 0.18665384599845666,
+      "grad_norm": 775.268310546875,
+      "learning_rate": 7.891661680839932e-06,
+      "loss": 61.9445,
+      "step": 46200
+    },
+    {
+      "epoch": 0.1866942472638243,
+      "grad_norm": 706.5367431640625,
+      "learning_rate": 7.89051098657106e-06,
+      "loss": 78.4231,
+      "step": 46210
+    },
+    {
+      "epoch": 0.18673464852919194,
+      "grad_norm": 700.9517822265625,
+      "learning_rate": 7.889360062317495e-06,
+      "loss": 35.4704,
+      "step": 46220
+    },
+    {
+      "epoch": 0.18677504979455956,
+      "grad_norm": 944.4683837890625,
+      "learning_rate": 7.888208908170812e-06,
+      "loss": 51.0714,
+      "step": 46230
+    },
+    {
+      "epoch": 0.1868154510599272,
+      "grad_norm": 866.54443359375,
+      "learning_rate": 7.887057524222596e-06,
+      "loss": 59.0249,
+      "step": 46240
+    },
+    {
+      "epoch": 0.18685585232529484,
+      "grad_norm": 799.7461547851562,
+      "learning_rate": 7.885905910564466e-06,
+      "loss": 65.52,
+      "step": 46250
+    },
+    {
+      "epoch": 0.18689625359066245,
+      "grad_norm": 1012.56396484375,
+      "learning_rate": 7.884754067288047e-06,
+      "loss": 55.5126,
+      "step": 46260
+    },
+    {
+      "epoch": 0.1869366548560301,
+      "grad_norm": 729.2407836914062,
+      "learning_rate": 7.883601994484986e-06,
+      "loss": 47.3084,
+      "step": 46270
+    },
+    {
+      "epoch": 0.18697705612139773,
+      "grad_norm": 505.71856689453125,
+      "learning_rate": 7.882449692246948e-06,
+      "loss": 41.2005,
+      "step": 46280
+    },
+    {
+      "epoch": 0.18701745738676534,
+      "grad_norm": 1154.6778564453125,
+      "learning_rate": 7.881297160665616e-06,
+      "loss": 49.5151,
+      "step": 46290
+    },
+    {
+      "epoch": 0.18705785865213298,
+      "grad_norm": 525.9098510742188,
+      "learning_rate": 7.880144399832693e-06,
+      "loss": 42.0621,
+      "step": 46300
+    },
+    {
+      "epoch": 0.18709825991750062,
+      "grad_norm": 970.4302978515625,
+      "learning_rate": 7.878991409839897e-06,
+      "loss": 56.0448,
+      "step": 46310
+    },
+    {
+      "epoch": 0.18713866118286826,
+      "grad_norm": 808.189697265625,
+      "learning_rate": 7.87783819077897e-06,
+      "loss": 47.8887,
+      "step": 46320
+    },
+    {
+      "epoch": 0.18717906244823587,
+      "grad_norm": 357.2666015625,
+      "learning_rate": 7.876684742741665e-06,
+      "loss": 28.5624,
+      "step": 46330
+    },
+    {
+      "epoch": 0.1872194637136035,
+      "grad_norm": 428.8252868652344,
+      "learning_rate": 7.875531065819755e-06,
+      "loss": 31.8156,
+      "step": 46340
+    },
+    {
+      "epoch": 0.18725986497897115,
+      "grad_norm": 831.3470458984375,
+      "learning_rate": 7.874377160105037e-06,
+      "loss": 50.5764,
+      "step": 46350
+    },
+    {
+      "epoch": 0.18730026624433876,
+      "grad_norm": 724.0596313476562,
+      "learning_rate": 7.873223025689319e-06,
+      "loss": 50.731,
+      "step": 46360
+    },
+    {
+      "epoch": 0.1873406675097064,
+      "grad_norm": 865.5615844726562,
+      "learning_rate": 7.872068662664432e-06,
+      "loss": 58.5636,
+      "step": 46370
+    },
+    {
+      "epoch": 0.18738106877507404,
+      "grad_norm": 537.1841430664062,
+      "learning_rate": 7.870914071122222e-06,
+      "loss": 63.4505,
+      "step": 46380
+    },
+    {
+      "epoch": 0.18742147004044166,
+      "grad_norm": 715.6597900390625,
+      "learning_rate": 7.869759251154554e-06,
+      "loss": 45.5798,
+      "step": 46390
+    },
+    {
+      "epoch": 0.1874618713058093,
+      "grad_norm": 418.3079833984375,
+      "learning_rate": 7.868604202853314e-06,
+      "loss": 37.2996,
+      "step": 46400
+    },
+    {
+      "epoch": 0.18750227257117694,
+      "grad_norm": 661.2510375976562,
+      "learning_rate": 7.867448926310403e-06,
+      "loss": 75.2654,
+      "step": 46410
+    },
+    {
+      "epoch": 0.18754267383654455,
+      "grad_norm": 398.23077392578125,
+      "learning_rate": 7.866293421617741e-06,
+      "loss": 46.4417,
+      "step": 46420
+    },
+    {
+      "epoch": 0.1875830751019122,
+      "grad_norm": 1762.594482421875,
+      "learning_rate": 7.865137688867264e-06,
+      "loss": 45.2139,
+      "step": 46430
+    },
+    {
+      "epoch": 0.18762347636727983,
+      "grad_norm": 663.5921630859375,
+      "learning_rate": 7.86398172815093e-06,
+      "loss": 36.9223,
+      "step": 46440
+    },
+    {
+      "epoch": 0.18766387763264744,
+      "grad_norm": 613.1920776367188,
+      "learning_rate": 7.862825539560716e-06,
+      "loss": 46.3909,
+      "step": 46450
+    },
+    {
+      "epoch": 0.18770427889801508,
+      "grad_norm": 401.7522888183594,
+      "learning_rate": 7.861669123188613e-06,
+      "loss": 36.1008,
+      "step": 46460
+    },
+    {
+      "epoch": 0.18774468016338272,
+      "grad_norm": 319.8637390136719,
+      "learning_rate": 7.86051247912663e-06,
+      "loss": 51.4729,
+      "step": 46470
+    },
+    {
+      "epoch": 0.18778508142875036,
+      "grad_norm": 397.6440124511719,
+      "learning_rate": 7.859355607466797e-06,
+      "loss": 55.7406,
+      "step": 46480
+    },
+    {
+      "epoch": 0.18782548269411797,
+      "grad_norm": 306.9148254394531,
+      "learning_rate": 7.858198508301161e-06,
+      "loss": 47.0562,
+      "step": 46490
+    },
+    {
+      "epoch": 0.1878658839594856,
+      "grad_norm": 568.0166625976562,
+      "learning_rate": 7.857041181721788e-06,
+      "loss": 31.964,
+      "step": 46500
+    },
+    {
+      "epoch": 0.18790628522485325,
+      "grad_norm": 439.1131286621094,
+      "learning_rate": 7.855883627820757e-06,
+      "loss": 25.5811,
+      "step": 46510
+    },
+    {
+      "epoch": 0.18794668649022087,
+      "grad_norm": 988.5216064453125,
+      "learning_rate": 7.854725846690175e-06,
+      "loss": 40.9094,
+      "step": 46520
+    },
+    {
+      "epoch": 0.1879870877555885,
+      "grad_norm": 401.2980041503906,
+      "learning_rate": 7.85356783842216e-06,
+      "loss": 46.3652,
+      "step": 46530
+    },
+    {
+      "epoch": 0.18802748902095615,
+      "grad_norm": 349.0110168457031,
+      "learning_rate": 7.852409603108845e-06,
+      "loss": 33.5314,
+      "step": 46540
+    },
+    {
+      "epoch": 0.18806789028632376,
+      "grad_norm": 694.8333129882812,
+      "learning_rate": 7.85125114084239e-06,
+      "loss": 40.1035,
+      "step": 46550
+    },
+    {
+      "epoch": 0.1881082915516914,
+      "grad_norm": 521.3746337890625,
+      "learning_rate": 7.850092451714967e-06,
+      "loss": 33.4153,
+      "step": 46560
+    },
+    {
+      "epoch": 0.18814869281705904,
+      "grad_norm": 1033.23486328125,
+      "learning_rate": 7.84893353581877e-06,
+      "loss": 38.9485,
+      "step": 46570
+    },
+    {
+      "epoch": 0.18818909408242665,
+      "grad_norm": 380.1110534667969,
+      "learning_rate": 7.847774393246005e-06,
+      "loss": 73.8259,
+      "step": 46580
+    },
+    {
+      "epoch": 0.1882294953477943,
+      "grad_norm": 336.256103515625,
+      "learning_rate": 7.8466150240889e-06,
+      "loss": 43.2453,
+      "step": 46590
+    },
+    {
+      "epoch": 0.18826989661316193,
+      "grad_norm": 512.811279296875,
+      "learning_rate": 7.845455428439703e-06,
+      "loss": 59.6738,
+      "step": 46600
+    },
+    {
+      "epoch": 0.18831029787852954,
+      "grad_norm": 419.74517822265625,
+      "learning_rate": 7.844295606390675e-06,
+      "loss": 49.9523,
+      "step": 46610
+    },
+    {
+      "epoch": 0.18835069914389718,
+      "grad_norm": 469.2853088378906,
+      "learning_rate": 7.843135558034101e-06,
+      "loss": 31.1946,
+      "step": 46620
+    },
+    {
+      "epoch": 0.18839110040926482,
+      "grad_norm": 805.4540405273438,
+      "learning_rate": 7.841975283462278e-06,
+      "loss": 42.8343,
+      "step": 46630
+    },
+    {
+      "epoch": 0.18843150167463246,
+      "grad_norm": 423.8988342285156,
+      "learning_rate": 7.840814782767525e-06,
+      "loss": 45.9534,
+      "step": 46640
+    },
+    {
+      "epoch": 0.18847190294000007,
+      "grad_norm": 618.1028442382812,
+      "learning_rate": 7.839654056042176e-06,
+      "loss": 55.0553,
+      "step": 46650
+    },
+    {
+      "epoch": 0.18851230420536771,
+      "grad_norm": 100.71240234375,
+      "learning_rate": 7.838493103378588e-06,
+      "loss": 45.5022,
+      "step": 46660
+    },
+    {
+      "epoch": 0.18855270547073535,
+      "grad_norm": 350.67462158203125,
+      "learning_rate": 7.83733192486913e-06,
+      "loss": 41.2146,
+      "step": 46670
+    },
+    {
+      "epoch": 0.18859310673610297,
+      "grad_norm": 791.0911254882812,
+      "learning_rate": 7.836170520606191e-06,
+      "loss": 46.5223,
+      "step": 46680
+    },
+    {
+      "epoch": 0.1886335080014706,
+      "grad_norm": 696.6979370117188,
+      "learning_rate": 7.83500889068218e-06,
+      "loss": 33.1812,
+      "step": 46690
+    },
+    {
+      "epoch": 0.18867390926683825,
+      "grad_norm": 492.4786682128906,
+      "learning_rate": 7.833847035189524e-06,
+      "loss": 46.7815,
+      "step": 46700
+    },
+    {
+      "epoch": 0.18871431053220586,
+      "grad_norm": 376.7648620605469,
+      "learning_rate": 7.832684954220664e-06,
+      "loss": 69.6158,
+      "step": 46710
+    },
+    {
+      "epoch": 0.1887547117975735,
+      "grad_norm": 572.7025146484375,
+      "learning_rate": 7.831522647868064e-06,
+      "loss": 51.5028,
+      "step": 46720
+    },
+    {
+      "epoch": 0.18879511306294114,
+      "grad_norm": 703.757568359375,
+      "learning_rate": 7.8303601162242e-06,
+      "loss": 42.5955,
+      "step": 46730
+    },
+    {
+      "epoch": 0.18883551432830875,
+      "grad_norm": 538.8714599609375,
+      "learning_rate": 7.829197359381571e-06,
+      "loss": 33.9557,
+      "step": 46740
+    },
+    {
+      "epoch": 0.1888759155936764,
+      "grad_norm": 1617.576416015625,
+      "learning_rate": 7.828034377432694e-06,
+      "loss": 50.4682,
+      "step": 46750
+    },
+    {
+      "epoch": 0.18891631685904403,
+      "grad_norm": 376.0511474609375,
+      "learning_rate": 7.826871170470099e-06,
+      "loss": 36.5483,
+      "step": 46760
+    },
+    {
+      "epoch": 0.18895671812441164,
+      "grad_norm": 975.2518920898438,
+      "learning_rate": 7.82570773858634e-06,
+      "loss": 64.4105,
+      "step": 46770
+    },
+    {
+      "epoch": 0.18899711938977928,
+      "grad_norm": 1208.195556640625,
+      "learning_rate": 7.824544081873984e-06,
+      "loss": 49.3749,
+      "step": 46780
+    },
+    {
+      "epoch": 0.18903752065514692,
+      "grad_norm": 1309.72607421875,
+      "learning_rate": 7.823380200425618e-06,
+      "loss": 69.8779,
+      "step": 46790
+    },
+    {
+      "epoch": 0.18907792192051456,
+      "grad_norm": 560.5164794921875,
+      "learning_rate": 7.822216094333847e-06,
+      "loss": 38.4529,
+      "step": 46800
+    },
+    {
+      "epoch": 0.18911832318588218,
+      "grad_norm": 631.4193115234375,
+      "learning_rate": 7.821051763691293e-06,
+      "loss": 36.4182,
+      "step": 46810
+    },
+    {
+      "epoch": 0.18915872445124982,
+      "grad_norm": 1111.424560546875,
+      "learning_rate": 7.819887208590597e-06,
+      "loss": 45.4261,
+      "step": 46820
+    },
+    {
+      "epoch": 0.18919912571661746,
+      "grad_norm": 431.0059814453125,
+      "learning_rate": 7.818722429124418e-06,
+      "loss": 36.5693,
+      "step": 46830
+    },
+    {
+      "epoch": 0.18923952698198507,
+      "grad_norm": 721.0582885742188,
+      "learning_rate": 7.817557425385433e-06,
+      "loss": 54.937,
+      "step": 46840
+    },
+    {
+      "epoch": 0.1892799282473527,
+      "grad_norm": 179.35610961914062,
+      "learning_rate": 7.816392197466333e-06,
+      "loss": 49.9349,
+      "step": 46850
+    },
+    {
+      "epoch": 0.18932032951272035,
+      "grad_norm": 661.6287231445312,
+      "learning_rate": 7.815226745459831e-06,
+      "loss": 44.7497,
+      "step": 46860
+    },
+    {
+      "epoch": 0.18936073077808796,
+      "grad_norm": 802.9453125,
+      "learning_rate": 7.814061069458657e-06,
+      "loss": 42.5215,
+      "step": 46870
+    },
+    {
+      "epoch": 0.1894011320434556,
+      "grad_norm": 1228.8883056640625,
+      "learning_rate": 7.81289516955556e-06,
+      "loss": 59.6201,
+      "step": 46880
+    },
+    {
+      "epoch": 0.18944153330882324,
+      "grad_norm": 555.4932861328125,
+      "learning_rate": 7.811729045843303e-06,
+      "loss": 51.9421,
+      "step": 46890
+    },
+    {
+      "epoch": 0.18948193457419085,
+      "grad_norm": 495.4557189941406,
+      "learning_rate": 7.81056269841467e-06,
+      "loss": 82.952,
+      "step": 46900
+    },
+    {
+      "epoch": 0.1895223358395585,
+      "grad_norm": 354.2893371582031,
+      "learning_rate": 7.80939612736246e-06,
+      "loss": 43.3217,
+      "step": 46910
+    },
+    {
+      "epoch": 0.18956273710492613,
+      "grad_norm": 1203.706787109375,
+      "learning_rate": 7.808229332779496e-06,
+      "loss": 46.9553,
+      "step": 46920
+    },
+    {
+      "epoch": 0.18960313837029374,
+      "grad_norm": 460.4730224609375,
+      "learning_rate": 7.807062314758612e-06,
+      "loss": 33.5704,
+      "step": 46930
+    },
+    {
+      "epoch": 0.18964353963566138,
+      "grad_norm": 309.24981689453125,
+      "learning_rate": 7.80589507339266e-06,
+      "loss": 54.543,
+      "step": 46940
+    },
+    {
+      "epoch": 0.18968394090102902,
+      "grad_norm": 573.7108154296875,
+      "learning_rate": 7.804727608774516e-06,
+      "loss": 53.2966,
+      "step": 46950
+    },
+    {
+      "epoch": 0.18972434216639666,
+      "grad_norm": 988.41259765625,
+      "learning_rate": 7.803559920997067e-06,
+      "loss": 46.6455,
+      "step": 46960
+    },
+    {
+      "epoch": 0.18976474343176428,
+      "grad_norm": 534.046630859375,
+      "learning_rate": 7.802392010153223e-06,
+      "loss": 32.7703,
+      "step": 46970
+    },
+    {
+      "epoch": 0.18980514469713192,
+      "grad_norm": 544.7523803710938,
+      "learning_rate": 7.801223876335907e-06,
+      "loss": 63.0975,
+      "step": 46980
+    },
+    {
+      "epoch": 0.18984554596249956,
+      "grad_norm": 608.649169921875,
+      "learning_rate": 7.800055519638064e-06,
+      "loss": 49.6523,
+      "step": 46990
+    },
+    {
+      "epoch": 0.18988594722786717,
+      "grad_norm": 977.2344360351562,
+      "learning_rate": 7.798886940152654e-06,
+      "loss": 50.4927,
+      "step": 47000
+    },
+    {
+      "epoch": 0.1899263484932348,
+      "grad_norm": 653.7332153320312,
+      "learning_rate": 7.797718137972654e-06,
+      "loss": 56.2325,
+      "step": 47010
+    },
+    {
+      "epoch": 0.18996674975860245,
+      "grad_norm": 859.51806640625,
+      "learning_rate": 7.79654911319106e-06,
+      "loss": 46.1248,
+      "step": 47020
+    },
+    {
+      "epoch": 0.19000715102397006,
+      "grad_norm": 285.8470764160156,
+      "learning_rate": 7.795379865900892e-06,
+      "loss": 35.2613,
+      "step": 47030
+    },
+    {
+      "epoch": 0.1900475522893377,
+      "grad_norm": 723.26904296875,
+      "learning_rate": 7.794210396195175e-06,
+      "loss": 48.0343,
+      "step": 47040
+    },
+    {
+      "epoch": 0.19008795355470534,
+      "grad_norm": 846.8826904296875,
+      "learning_rate": 7.79304070416696e-06,
+      "loss": 44.9846,
+      "step": 47050
+    },
+    {
+      "epoch": 0.19012835482007295,
+      "grad_norm": 375.7879638671875,
+      "learning_rate": 7.791870789909315e-06,
+      "loss": 54.1298,
+      "step": 47060
+    },
+    {
+      "epoch": 0.1901687560854406,
+      "grad_norm": 944.4963989257812,
+      "learning_rate": 7.790700653515324e-06,
+      "loss": 40.4307,
+      "step": 47070
+    },
+    {
+      "epoch": 0.19020915735080823,
+      "grad_norm": 455.2538146972656,
+      "learning_rate": 7.789530295078089e-06,
+      "loss": 52.4162,
+      "step": 47080
+    },
+    {
+      "epoch": 0.19024955861617585,
+      "grad_norm": 548.3023071289062,
+      "learning_rate": 7.788359714690732e-06,
+      "loss": 35.7722,
+      "step": 47090
+    },
+    {
+      "epoch": 0.19028995988154349,
+      "grad_norm": 358.2738037109375,
+      "learning_rate": 7.787188912446389e-06,
+      "loss": 51.5841,
+      "step": 47100
+    },
+    {
+      "epoch": 0.19033036114691113,
+      "grad_norm": 1344.7706298828125,
+      "learning_rate": 7.786017888438214e-06,
+      "loss": 30.4217,
+      "step": 47110
+    },
+    {
+      "epoch": 0.19037076241227877,
+      "grad_norm": 593.27685546875,
+      "learning_rate": 7.784846642759383e-06,
+      "loss": 42.8882,
+      "step": 47120
+    },
+    {
+      "epoch": 0.19041116367764638,
+      "grad_norm": 757.25,
+      "learning_rate": 7.783675175503087e-06,
+      "loss": 60.5505,
+      "step": 47130
+    },
+    {
+      "epoch": 0.19045156494301402,
+      "grad_norm": 427.2942199707031,
+      "learning_rate": 7.78250348676253e-06,
+      "loss": 46.353,
+      "step": 47140
+    },
+    {
+      "epoch": 0.19049196620838166,
+      "grad_norm": 302.1863098144531,
+      "learning_rate": 7.781331576630941e-06,
+      "loss": 52.2239,
+      "step": 47150
+    },
+    {
+      "epoch": 0.19053236747374927,
+      "grad_norm": 444.2868347167969,
+      "learning_rate": 7.780159445201562e-06,
+      "loss": 52.4015,
+      "step": 47160
+    },
+    {
+      "epoch": 0.1905727687391169,
+      "grad_norm": 411.0354309082031,
+      "learning_rate": 7.778987092567658e-06,
+      "loss": 34.217,
+      "step": 47170
+    },
+    {
+      "epoch": 0.19061317000448455,
+      "grad_norm": 359.02069091796875,
+      "learning_rate": 7.777814518822504e-06,
+      "loss": 47.4867,
+      "step": 47180
+    },
+    {
+      "epoch": 0.19065357126985216,
+      "grad_norm": 673.830810546875,
+      "learning_rate": 7.776641724059398e-06,
+      "loss": 40.569,
+      "step": 47190
+    },
+    {
+      "epoch": 0.1906939725352198,
+      "grad_norm": 902.4163208007812,
+      "learning_rate": 7.77546870837165e-06,
+      "loss": 36.7457,
+      "step": 47200
+    },
+    {
+      "epoch": 0.19073437380058744,
+      "grad_norm": 663.3661499023438,
+      "learning_rate": 7.774295471852596e-06,
+      "loss": 66.2683,
+      "step": 47210
+    },
+    {
+      "epoch": 0.19077477506595505,
+      "grad_norm": 489.5060119628906,
+      "learning_rate": 7.773122014595584e-06,
+      "loss": 55.9361,
+      "step": 47220
+    },
+    {
+      "epoch": 0.1908151763313227,
+      "grad_norm": 572.3447875976562,
+      "learning_rate": 7.771948336693983e-06,
+      "loss": 58.0868,
+      "step": 47230
+    },
+    {
+      "epoch": 0.19085557759669033,
+      "grad_norm": 777.2191772460938,
+      "learning_rate": 7.770774438241168e-06,
+      "loss": 58.8669,
+      "step": 47240
+    },
+    {
+      "epoch": 0.19089597886205795,
+      "grad_norm": 754.6911010742188,
+      "learning_rate": 7.769600319330553e-06,
+      "loss": 47.5421,
+      "step": 47250
+    },
+    {
+      "epoch": 0.1909363801274256,
+      "grad_norm": 693.4794311523438,
+      "learning_rate": 7.768425980055548e-06,
+      "loss": 35.347,
+      "step": 47260
+    },
+    {
+      "epoch": 0.19097678139279323,
+      "grad_norm": 724.5881958007812,
+      "learning_rate": 7.767251420509593e-06,
+      "loss": 55.401,
+      "step": 47270
+    },
+    {
+      "epoch": 0.19101718265816087,
+      "grad_norm": 370.7134704589844,
+      "learning_rate": 7.766076640786145e-06,
+      "loss": 39.1298,
+      "step": 47280
+    },
+    {
+      "epoch": 0.19105758392352848,
+      "grad_norm": 688.7514038085938,
+      "learning_rate": 7.764901640978671e-06,
+      "loss": 57.3265,
+      "step": 47290
+    },
+    {
+      "epoch": 0.19109798518889612,
+      "grad_norm": 725.8629150390625,
+      "learning_rate": 7.763726421180664e-06,
+      "loss": 43.4557,
+      "step": 47300
+    },
+    {
+      "epoch": 0.19113838645426376,
+      "grad_norm": 408.5218811035156,
+      "learning_rate": 7.762550981485629e-06,
+      "loss": 48.6489,
+      "step": 47310
+    },
+    {
+      "epoch": 0.19117878771963137,
+      "grad_norm": 475.49114990234375,
+      "learning_rate": 7.76137532198709e-06,
+      "loss": 42.917,
+      "step": 47320
+    },
+    {
+      "epoch": 0.191219188984999,
+      "grad_norm": 921.5711059570312,
+      "learning_rate": 7.76019944277859e-06,
+      "loss": 35.9836,
+      "step": 47330
+    },
+    {
+      "epoch": 0.19125959025036665,
+      "grad_norm": 696.03369140625,
+      "learning_rate": 7.759023343953689e-06,
+      "loss": 56.2156,
+      "step": 47340
+    },
+    {
+      "epoch": 0.19129999151573426,
+      "grad_norm": 962.43603515625,
+      "learning_rate": 7.757847025605963e-06,
+      "loss": 54.9843,
+      "step": 47350
+    },
+    {
+      "epoch": 0.1913403927811019,
+      "grad_norm": 471.5603942871094,
+      "learning_rate": 7.756670487829005e-06,
+      "loss": 54.7021,
+      "step": 47360
+    },
+    {
+      "epoch": 0.19138079404646954,
+      "grad_norm": 202.61300659179688,
+      "learning_rate": 7.755493730716428e-06,
+      "loss": 46.6755,
+      "step": 47370
+    },
+    {
+      "epoch": 0.19142119531183716,
+      "grad_norm": 397.8414611816406,
+      "learning_rate": 7.75431675436186e-06,
+      "loss": 43.3254,
+      "step": 47380
+    },
+    {
+      "epoch": 0.1914615965772048,
+      "grad_norm": 439.717529296875,
+      "learning_rate": 7.753139558858949e-06,
+      "loss": 45.3347,
+      "step": 47390
+    },
+    {
+      "epoch": 0.19150199784257244,
+      "grad_norm": 431.6274108886719,
+      "learning_rate": 7.751962144301359e-06,
+      "loss": 44.974,
+      "step": 47400
+    },
+    {
+      "epoch": 0.19154239910794005,
+      "grad_norm": 301.6814880371094,
+      "learning_rate": 7.75078451078277e-06,
+      "loss": 36.2488,
+      "step": 47410
+    },
+    {
+      "epoch": 0.1915828003733077,
+      "grad_norm": 379.6395568847656,
+      "learning_rate": 7.749606658396883e-06,
+      "loss": 34.9849,
+      "step": 47420
+    },
+    {
+      "epoch": 0.19162320163867533,
+      "grad_norm": 1423.8985595703125,
+      "learning_rate": 7.748428587237412e-06,
+      "loss": 45.026,
+      "step": 47430
+    },
+    {
+      "epoch": 0.19166360290404297,
+      "grad_norm": 437.26409912109375,
+      "learning_rate": 7.747250297398092e-06,
+      "loss": 48.0475,
+      "step": 47440
+    },
+    {
+      "epoch": 0.19170400416941058,
+      "grad_norm": 1188.09619140625,
+      "learning_rate": 7.746071788972675e-06,
+      "loss": 63.0107,
+      "step": 47450
+    },
+    {
+      "epoch": 0.19174440543477822,
+      "grad_norm": 596.3973388671875,
+      "learning_rate": 7.744893062054928e-06,
+      "loss": 37.3981,
+      "step": 47460
+    },
+    {
+      "epoch": 0.19178480670014586,
+      "grad_norm": 1094.22021484375,
+      "learning_rate": 7.743714116738636e-06,
+      "loss": 51.8161,
+      "step": 47470
+    },
+    {
+      "epoch": 0.19182520796551347,
+      "grad_norm": 527.591552734375,
+      "learning_rate": 7.742534953117607e-06,
+      "loss": 63.8491,
+      "step": 47480
+    },
+    {
+      "epoch": 0.1918656092308811,
+      "grad_norm": 1503.0655517578125,
+      "learning_rate": 7.741355571285656e-06,
+      "loss": 64.103,
+      "step": 47490
+    },
+    {
+      "epoch": 0.19190601049624875,
+      "grad_norm": 335.1996154785156,
+      "learning_rate": 7.740175971336624e-06,
+      "loss": 33.4057,
+      "step": 47500
+    },
+    {
+      "epoch": 0.19194641176161636,
+      "grad_norm": 675.87451171875,
+      "learning_rate": 7.738996153364364e-06,
+      "loss": 48.068,
+      "step": 47510
+    },
+    {
+      "epoch": 0.191986813026984,
+      "grad_norm": 501.4929504394531,
+      "learning_rate": 7.737816117462752e-06,
+      "loss": 43.2299,
+      "step": 47520
+    },
+    {
+      "epoch": 0.19202721429235164,
+      "grad_norm": 358.48583984375,
+      "learning_rate": 7.736635863725677e-06,
+      "loss": 55.4011,
+      "step": 47530
+    },
+    {
+      "epoch": 0.19206761555771926,
+      "grad_norm": 241.45448303222656,
+      "learning_rate": 7.735455392247044e-06,
+      "loss": 61.9485,
+      "step": 47540
+    },
+    {
+      "epoch": 0.1921080168230869,
+      "grad_norm": 323.69732666015625,
+      "learning_rate": 7.73427470312078e-06,
+      "loss": 38.1733,
+      "step": 47550
+    },
+    {
+      "epoch": 0.19214841808845454,
+      "grad_norm": 330.3637390136719,
+      "learning_rate": 7.733093796440828e-06,
+      "loss": 46.7346,
+      "step": 47560
+    },
+    {
+      "epoch": 0.19218881935382215,
+      "grad_norm": 1063.1988525390625,
+      "learning_rate": 7.731912672301145e-06,
+      "loss": 50.6998,
+      "step": 47570
+    },
+    {
+      "epoch": 0.1922292206191898,
+      "grad_norm": 655.8283081054688,
+      "learning_rate": 7.730731330795707e-06,
+      "loss": 46.5782,
+      "step": 47580
+    },
+    {
+      "epoch": 0.19226962188455743,
+      "grad_norm": 961.4747314453125,
+      "learning_rate": 7.72954977201851e-06,
+      "loss": 59.9621,
+      "step": 47590
+    },
+    {
+      "epoch": 0.19231002314992507,
+      "grad_norm": 322.00726318359375,
+      "learning_rate": 7.728367996063566e-06,
+      "loss": 57.8829,
+      "step": 47600
+    },
+    {
+      "epoch": 0.19235042441529268,
+      "grad_norm": 896.306396484375,
+      "learning_rate": 7.727186003024902e-06,
+      "loss": 62.804,
+      "step": 47610
+    },
+    {
+      "epoch": 0.19239082568066032,
+      "grad_norm": 582.4077758789062,
+      "learning_rate": 7.726003792996562e-06,
+      "loss": 58.3848,
+      "step": 47620
+    },
+    {
+      "epoch": 0.19243122694602796,
+      "grad_norm": 1208.078125,
+      "learning_rate": 7.724821366072612e-06,
+      "loss": 54.5546,
+      "step": 47630
+    },
+    {
+      "epoch": 0.19247162821139557,
+      "grad_norm": 807.9841918945312,
+      "learning_rate": 7.723638722347132e-06,
+      "loss": 29.3268,
+      "step": 47640
+    },
+    {
+      "epoch": 0.1925120294767632,
+      "grad_norm": 1568.94580078125,
+      "learning_rate": 7.722455861914218e-06,
+      "loss": 67.2609,
+      "step": 47650
+    },
+    {
+      "epoch": 0.19255243074213085,
+      "grad_norm": 918.30224609375,
+      "learning_rate": 7.721272784867983e-06,
+      "loss": 45.9969,
+      "step": 47660
+    },
+    {
+      "epoch": 0.19259283200749847,
+      "grad_norm": 733.860595703125,
+      "learning_rate": 7.720089491302565e-06,
+      "loss": 43.838,
+      "step": 47670
+    },
+    {
+      "epoch": 0.1926332332728661,
+      "grad_norm": 403.791748046875,
+      "learning_rate": 7.718905981312108e-06,
+      "loss": 47.9569,
+      "step": 47680
+    },
+    {
+      "epoch": 0.19267363453823375,
+      "grad_norm": 344.37738037109375,
+      "learning_rate": 7.71772225499078e-06,
+      "loss": 43.2986,
+      "step": 47690
+    },
+    {
+      "epoch": 0.19271403580360136,
+      "grad_norm": 959.1226196289062,
+      "learning_rate": 7.716538312432767e-06,
+      "loss": 57.1505,
+      "step": 47700
+    },
+    {
+      "epoch": 0.192754437068969,
+      "grad_norm": 765.4625854492188,
+      "learning_rate": 7.715354153732265e-06,
+      "loss": 44.0606,
+      "step": 47710
+    },
+    {
+      "epoch": 0.19279483833433664,
+      "grad_norm": 910.7199096679688,
+      "learning_rate": 7.714169778983496e-06,
+      "loss": 40.9219,
+      "step": 47720
+    },
+    {
+      "epoch": 0.19283523959970425,
+      "grad_norm": 561.5447387695312,
+      "learning_rate": 7.712985188280694e-06,
+      "loss": 75.7906,
+      "step": 47730
+    },
+    {
+      "epoch": 0.1928756408650719,
+      "grad_norm": 406.99566650390625,
+      "learning_rate": 7.711800381718111e-06,
+      "loss": 40.4035,
+      "step": 47740
+    },
+    {
+      "epoch": 0.19291604213043953,
+      "grad_norm": 757.7989501953125,
+      "learning_rate": 7.710615359390018e-06,
+      "loss": 43.6714,
+      "step": 47750
+    },
+    {
+      "epoch": 0.19295644339580717,
+      "grad_norm": 532.935546875,
+      "learning_rate": 7.7094301213907e-06,
+      "loss": 40.9307,
+      "step": 47760
+    },
+    {
+      "epoch": 0.19299684466117478,
+      "grad_norm": 393.7739562988281,
+      "learning_rate": 7.708244667814463e-06,
+      "loss": 53.9688,
+      "step": 47770
+    },
+    {
+      "epoch": 0.19303724592654242,
+      "grad_norm": 952.9371948242188,
+      "learning_rate": 7.707058998755626e-06,
+      "loss": 40.4923,
+      "step": 47780
+    },
+    {
+      "epoch": 0.19307764719191006,
+      "grad_norm": 470.9456787109375,
+      "learning_rate": 7.705873114308529e-06,
+      "loss": 41.2027,
+      "step": 47790
+    },
+    {
+      "epoch": 0.19311804845727767,
+      "grad_norm": 593.4171142578125,
+      "learning_rate": 7.704687014567524e-06,
+      "loss": 36.4702,
+      "step": 47800
+    },
+    {
+      "epoch": 0.19315844972264531,
+      "grad_norm": 538.1138916015625,
+      "learning_rate": 7.703500699626988e-06,
+      "loss": 52.4323,
+      "step": 47810
+    },
+    {
+      "epoch": 0.19319885098801295,
+      "grad_norm": 861.8800048828125,
+      "learning_rate": 7.702314169581311e-06,
+      "loss": 26.6001,
+      "step": 47820
+    },
+    {
+      "epoch": 0.19323925225338057,
+      "grad_norm": 396.1697082519531,
+      "learning_rate": 7.701127424524894e-06,
+      "loss": 43.8624,
+      "step": 47830
+    },
+    {
+      "epoch": 0.1932796535187482,
+      "grad_norm": 665.240234375,
+      "learning_rate": 7.699940464552166e-06,
+      "loss": 72.4191,
+      "step": 47840
+    },
+    {
+      "epoch": 0.19332005478411585,
+      "grad_norm": 516.3006591796875,
+      "learning_rate": 7.698753289757565e-06,
+      "loss": 76.2827,
+      "step": 47850
+    },
+    {
+      "epoch": 0.19336045604948346,
+      "grad_norm": 3109.86767578125,
+      "learning_rate": 7.69756590023555e-06,
+      "loss": 81.7451,
+      "step": 47860
+    },
+    {
+      "epoch": 0.1934008573148511,
+      "grad_norm": 441.69775390625,
+      "learning_rate": 7.696378296080598e-06,
+      "loss": 42.5113,
+      "step": 47870
+    },
+    {
+      "epoch": 0.19344125858021874,
+      "grad_norm": 449.0380554199219,
+      "learning_rate": 7.6951904773872e-06,
+      "loss": 39.6808,
+      "step": 47880
+    },
+    {
+      "epoch": 0.19348165984558635,
+      "grad_norm": 702.9412841796875,
+      "learning_rate": 7.694002444249863e-06,
+      "loss": 50.4996,
+      "step": 47890
+    },
+    {
+      "epoch": 0.193522061110954,
+      "grad_norm": 923.0763549804688,
+      "learning_rate": 7.692814196763118e-06,
+      "loss": 49.8511,
+      "step": 47900
+    },
+    {
+      "epoch": 0.19356246237632163,
+      "grad_norm": 581.8436279296875,
+      "learning_rate": 7.691625735021505e-06,
+      "loss": 61.1597,
+      "step": 47910
+    },
+    {
+      "epoch": 0.19360286364168927,
+      "grad_norm": 906.2317504882812,
+      "learning_rate": 7.690437059119584e-06,
+      "loss": 52.1174,
+      "step": 47920
+    },
+    {
+      "epoch": 0.19364326490705688,
+      "grad_norm": 812.0711059570312,
+      "learning_rate": 7.689248169151935e-06,
+      "loss": 66.8462,
+      "step": 47930
+    },
+    {
+      "epoch": 0.19368366617242452,
+      "grad_norm": 762.5548095703125,
+      "learning_rate": 7.68805906521315e-06,
+      "loss": 42.8686,
+      "step": 47940
+    },
+    {
+      "epoch": 0.19372406743779216,
+      "grad_norm": 832.9835205078125,
+      "learning_rate": 7.686869747397843e-06,
+      "loss": 43.76,
+      "step": 47950
+    },
+    {
+      "epoch": 0.19376446870315978,
+      "grad_norm": 536.840576171875,
+      "learning_rate": 7.685680215800639e-06,
+      "loss": 52.4641,
+      "step": 47960
+    },
+    {
+      "epoch": 0.19380486996852742,
+      "grad_norm": 303.6435546875,
+      "learning_rate": 7.684490470516185e-06,
+      "loss": 42.1546,
+      "step": 47970
+    },
+    {
+      "epoch": 0.19384527123389506,
+      "grad_norm": 684.5233764648438,
+      "learning_rate": 7.683300511639149e-06,
+      "loss": 49.268,
+      "step": 47980
+    },
+    {
+      "epoch": 0.19388567249926267,
+      "grad_norm": 818.1359252929688,
+      "learning_rate": 7.682110339264203e-06,
+      "loss": 52.341,
+      "step": 47990
+    },
+    {
+      "epoch": 0.1939260737646303,
+      "grad_norm": 322.98992919921875,
+      "learning_rate": 7.680919953486047e-06,
+      "loss": 46.4746,
+      "step": 48000
+    },
+    {
+      "epoch": 0.19396647502999795,
+      "grad_norm": 484.1806640625,
+      "learning_rate": 7.679729354399395e-06,
+      "loss": 30.4402,
+      "step": 48010
+    },
+    {
+      "epoch": 0.19400687629536556,
+      "grad_norm": 594.4285278320312,
+      "learning_rate": 7.678538542098974e-06,
+      "loss": 46.43,
+      "step": 48020
+    },
+    {
+      "epoch": 0.1940472775607332,
+      "grad_norm": 2891.762939453125,
+      "learning_rate": 7.677347516679536e-06,
+      "loss": 73.9085,
+      "step": 48030
+    },
+    {
+      "epoch": 0.19408767882610084,
+      "grad_norm": 977.7078857421875,
+      "learning_rate": 7.676156278235845e-06,
+      "loss": 48.965,
+      "step": 48040
+    },
+    {
+      "epoch": 0.19412808009146845,
+      "grad_norm": 448.16082763671875,
+      "learning_rate": 7.674964826862679e-06,
+      "loss": 38.0282,
+      "step": 48050
+    },
+    {
+      "epoch": 0.1941684813568361,
+      "grad_norm": 597.8922119140625,
+      "learning_rate": 7.673773162654836e-06,
+      "loss": 62.7354,
+      "step": 48060
+    },
+    {
+      "epoch": 0.19420888262220373,
+      "grad_norm": 469.99371337890625,
+      "learning_rate": 7.672581285707135e-06,
+      "loss": 45.9692,
+      "step": 48070
+    },
+    {
+      "epoch": 0.19424928388757137,
+      "grad_norm": 415.48046875,
+      "learning_rate": 7.67138919611441e-06,
+      "loss": 62.1821,
+      "step": 48080
+    },
+    {
+      "epoch": 0.19428968515293898,
+      "grad_norm": 435.82470703125,
+      "learning_rate": 7.670196893971502e-06,
+      "loss": 46.7921,
+      "step": 48090
+    },
+    {
+      "epoch": 0.19433008641830662,
+      "grad_norm": 1045.2288818359375,
+      "learning_rate": 7.669004379373284e-06,
+      "loss": 54.4301,
+      "step": 48100
+    },
+    {
+      "epoch": 0.19437048768367426,
+      "grad_norm": 983.9075927734375,
+      "learning_rate": 7.667811652414637e-06,
+      "loss": 37.8054,
+      "step": 48110
+    },
+    {
+      "epoch": 0.19441088894904188,
+      "grad_norm": 609.9669799804688,
+      "learning_rate": 7.666618713190459e-06,
+      "loss": 35.3491,
+      "step": 48120
+    },
+    {
+      "epoch": 0.19445129021440952,
+      "grad_norm": 632.4072265625,
+      "learning_rate": 7.665425561795669e-06,
+      "loss": 35.028,
+      "step": 48130
+    },
+    {
+      "epoch": 0.19449169147977716,
+      "grad_norm": 749.6817626953125,
+      "learning_rate": 7.664232198325198e-06,
+      "loss": 45.446,
+      "step": 48140
+    },
+    {
+      "epoch": 0.19453209274514477,
+      "grad_norm": 1017.4483032226562,
+      "learning_rate": 7.663038622873999e-06,
+      "loss": 36.2746,
+      "step": 48150
+    },
+    {
+      "epoch": 0.1945724940105124,
+      "grad_norm": 710.2518310546875,
+      "learning_rate": 7.66184483553704e-06,
+      "loss": 86.899,
+      "step": 48160
+    },
+    {
+      "epoch": 0.19461289527588005,
+      "grad_norm": 1045.6429443359375,
+      "learning_rate": 7.660650836409302e-06,
+      "loss": 64.3826,
+      "step": 48170
+    },
+    {
+      "epoch": 0.19465329654124766,
+      "grad_norm": 784.1268310546875,
+      "learning_rate": 7.65945662558579e-06,
+      "loss": 36.2017,
+      "step": 48180
+    },
+    {
+      "epoch": 0.1946936978066153,
+      "grad_norm": 178.67416381835938,
+      "learning_rate": 7.658262203161517e-06,
+      "loss": 45.8245,
+      "step": 48190
+    },
+    {
+      "epoch": 0.19473409907198294,
+      "grad_norm": 812.579833984375,
+      "learning_rate": 7.65706756923152e-06,
+      "loss": 41.051,
+      "step": 48200
+    },
+    {
+      "epoch": 0.19477450033735055,
+      "grad_norm": 1162.462158203125,
+      "learning_rate": 7.655872723890854e-06,
+      "loss": 51.0171,
+      "step": 48210
+    },
+    {
+      "epoch": 0.1948149016027182,
+      "grad_norm": 877.551513671875,
+      "learning_rate": 7.654677667234582e-06,
+      "loss": 47.0945,
+      "step": 48220
+    },
+    {
+      "epoch": 0.19485530286808583,
+      "grad_norm": 325.6798400878906,
+      "learning_rate": 7.65348239935779e-06,
+      "loss": 59.1071,
+      "step": 48230
+    },
+    {
+      "epoch": 0.19489570413345347,
+      "grad_norm": 385.6723937988281,
+      "learning_rate": 7.652286920355583e-06,
+      "loss": 46.9059,
+      "step": 48240
+    },
+    {
+      "epoch": 0.19493610539882109,
+      "grad_norm": 684.6091918945312,
+      "learning_rate": 7.651091230323079e-06,
+      "loss": 48.4449,
+      "step": 48250
+    },
+    {
+      "epoch": 0.19497650666418873,
+      "grad_norm": 925.9144287109375,
+      "learning_rate": 7.649895329355411e-06,
+      "loss": 31.311,
+      "step": 48260
+    },
+    {
+      "epoch": 0.19501690792955637,
+      "grad_norm": 257.57537841796875,
+      "learning_rate": 7.648699217547733e-06,
+      "loss": 24.9668,
+      "step": 48270
+    },
+    {
+      "epoch": 0.19505730919492398,
+      "grad_norm": 833.0692138671875,
+      "learning_rate": 7.647502894995215e-06,
+      "loss": 44.589,
+      "step": 48280
+    },
+    {
+      "epoch": 0.19509771046029162,
+      "grad_norm": 995.8534545898438,
+      "learning_rate": 7.646306361793042e-06,
+      "loss": 49.3848,
+      "step": 48290
+    },
+    {
+      "epoch": 0.19513811172565926,
+      "grad_norm": 739.4403076171875,
+      "learning_rate": 7.645109618036416e-06,
+      "loss": 49.1146,
+      "step": 48300
+    },
+    {
+      "epoch": 0.19517851299102687,
+      "grad_norm": 772.0132446289062,
+      "learning_rate": 7.643912663820559e-06,
+      "loss": 54.9199,
+      "step": 48310
+    },
+    {
+      "epoch": 0.1952189142563945,
+      "grad_norm": 949.4467163085938,
+      "learning_rate": 7.642715499240702e-06,
+      "loss": 53.1441,
+      "step": 48320
+    },
+    {
+      "epoch": 0.19525931552176215,
+      "grad_norm": 283.2403564453125,
+      "learning_rate": 7.641518124392105e-06,
+      "loss": 34.6873,
+      "step": 48330
+    },
+    {
+      "epoch": 0.19529971678712976,
+      "grad_norm": 493.2290954589844,
+      "learning_rate": 7.640320539370032e-06,
+      "loss": 39.8798,
+      "step": 48340
+    },
+    {
+      "epoch": 0.1953401180524974,
+      "grad_norm": 636.8895874023438,
+      "learning_rate": 7.63912274426977e-06,
+      "loss": 68.3165,
+      "step": 48350
+    },
+    {
+      "epoch": 0.19538051931786504,
+      "grad_norm": 429.77777099609375,
+      "learning_rate": 7.637924739186624e-06,
+      "loss": 38.3885,
+      "step": 48360
+    },
+    {
+      "epoch": 0.19542092058323265,
+      "grad_norm": 761.0277709960938,
+      "learning_rate": 7.636726524215913e-06,
+      "loss": 50.0891,
+      "step": 48370
+    },
+    {
+      "epoch": 0.1954613218486003,
+      "grad_norm": 821.7488403320312,
+      "learning_rate": 7.635528099452974e-06,
+      "loss": 62.0171,
+      "step": 48380
+    },
+    {
+      "epoch": 0.19550172311396793,
+      "grad_norm": 724.60693359375,
+      "learning_rate": 7.634329464993158e-06,
+      "loss": 49.4184,
+      "step": 48390
+    },
+    {
+      "epoch": 0.19554212437933557,
+      "grad_norm": 890.350341796875,
+      "learning_rate": 7.633130620931837e-06,
+      "loss": 38.8696,
+      "step": 48400
+    },
+    {
+      "epoch": 0.1955825256447032,
+      "grad_norm": 762.0298461914062,
+      "learning_rate": 7.631931567364398e-06,
+      "loss": 45.8956,
+      "step": 48410
+    },
+    {
+      "epoch": 0.19562292691007083,
+      "grad_norm": 1499.270263671875,
+      "learning_rate": 7.630732304386244e-06,
+      "loss": 48.7371,
+      "step": 48420
+    },
+    {
+      "epoch": 0.19566332817543847,
+      "grad_norm": 606.8875122070312,
+      "learning_rate": 7.629532832092792e-06,
+      "loss": 35.9768,
+      "step": 48430
+    },
+    {
+      "epoch": 0.19570372944080608,
+      "grad_norm": 697.2933349609375,
+      "learning_rate": 7.62833315057948e-06,
+      "loss": 42.1523,
+      "step": 48440
+    },
+    {
+      "epoch": 0.19574413070617372,
+      "grad_norm": 1122.26123046875,
+      "learning_rate": 7.627133259941762e-06,
+      "loss": 43.5487,
+      "step": 48450
+    },
+    {
+      "epoch": 0.19578453197154136,
+      "grad_norm": 1514.467529296875,
+      "learning_rate": 7.625933160275109e-06,
+      "loss": 60.6185,
+      "step": 48460
+    },
+    {
+      "epoch": 0.19582493323690897,
+      "grad_norm": 891.0689086914062,
+      "learning_rate": 7.6247328516750055e-06,
+      "loss": 52.881,
+      "step": 48470
+    },
+    {
+      "epoch": 0.1958653345022766,
+      "grad_norm": 572.1576538085938,
+      "learning_rate": 7.623532334236954e-06,
+      "loss": 60.1336,
+      "step": 48480
+    },
+    {
+      "epoch": 0.19590573576764425,
+      "grad_norm": 1341.4376220703125,
+      "learning_rate": 7.622331608056474e-06,
+      "loss": 44.1682,
+      "step": 48490
+    },
+    {
+      "epoch": 0.19594613703301186,
+      "grad_norm": 665.36962890625,
+      "learning_rate": 7.621130673229105e-06,
+      "loss": 58.796,
+      "step": 48500
+    },
+    {
+      "epoch": 0.1959865382983795,
+      "grad_norm": 861.0709228515625,
+      "learning_rate": 7.619929529850397e-06,
+      "loss": 53.8948,
+      "step": 48510
+    },
+    {
+      "epoch": 0.19602693956374714,
+      "grad_norm": 489.127197265625,
+      "learning_rate": 7.618728178015919e-06,
+      "loss": 38.3744,
+      "step": 48520
+    },
+    {
+      "epoch": 0.19606734082911476,
+      "grad_norm": 653.3378295898438,
+      "learning_rate": 7.617526617821259e-06,
+      "loss": 35.1475,
+      "step": 48530
+    },
+    {
+      "epoch": 0.1961077420944824,
+      "grad_norm": 673.4270629882812,
+      "learning_rate": 7.616324849362019e-06,
+      "loss": 50.4059,
+      "step": 48540
+    },
+    {
+      "epoch": 0.19614814335985004,
+      "grad_norm": 505.6314392089844,
+      "learning_rate": 7.615122872733819e-06,
+      "loss": 54.5434,
+      "step": 48550
+    },
+    {
+      "epoch": 0.19618854462521768,
+      "grad_norm": 905.9286499023438,
+      "learning_rate": 7.613920688032293e-06,
+      "loss": 54.7946,
+      "step": 48560
+    },
+    {
+      "epoch": 0.1962289458905853,
+      "grad_norm": 450.91143798828125,
+      "learning_rate": 7.612718295353094e-06,
+      "loss": 33.1633,
+      "step": 48570
+    },
+    {
+      "epoch": 0.19626934715595293,
+      "grad_norm": 285.4925842285156,
+      "learning_rate": 7.61151569479189e-06,
+      "loss": 83.4821,
+      "step": 48580
+    },
+    {
+      "epoch": 0.19630974842132057,
+      "grad_norm": 1145.9517822265625,
+      "learning_rate": 7.610312886444369e-06,
+      "loss": 69.2726,
+      "step": 48590
+    },
+    {
+      "epoch": 0.19635014968668818,
+      "grad_norm": 343.5018615722656,
+      "learning_rate": 7.60910987040623e-06,
+      "loss": 45.3421,
+      "step": 48600
+    },
+    {
+      "epoch": 0.19639055095205582,
+      "grad_norm": 571.7493896484375,
+      "learning_rate": 7.607906646773195e-06,
+      "loss": 51.0718,
+      "step": 48610
+    },
+    {
+      "epoch": 0.19643095221742346,
+      "grad_norm": 1071.421875,
+      "learning_rate": 7.606703215640995e-06,
+      "loss": 51.5955,
+      "step": 48620
+    },
+    {
+      "epoch": 0.19647135348279107,
+      "grad_norm": 727.3692016601562,
+      "learning_rate": 7.605499577105382e-06,
+      "loss": 56.2631,
+      "step": 48630
+    },
+    {
+      "epoch": 0.1965117547481587,
+      "grad_norm": 555.5590209960938,
+      "learning_rate": 7.604295731262128e-06,
+      "loss": 47.5872,
+      "step": 48640
+    },
+    {
+      "epoch": 0.19655215601352635,
+      "grad_norm": 1017.7325439453125,
+      "learning_rate": 7.603091678207013e-06,
+      "loss": 43.9335,
+      "step": 48650
+    },
+    {
+      "epoch": 0.19659255727889396,
+      "grad_norm": 919.7353515625,
+      "learning_rate": 7.60188741803584e-06,
+      "loss": 57.0481,
+      "step": 48660
+    },
+    {
+      "epoch": 0.1966329585442616,
+      "grad_norm": 814.9369506835938,
+      "learning_rate": 7.600682950844428e-06,
+      "loss": 32.845,
+      "step": 48670
+    },
+    {
+      "epoch": 0.19667335980962924,
+      "grad_norm": 609.3458251953125,
+      "learning_rate": 7.599478276728607e-06,
+      "loss": 30.4059,
+      "step": 48680
+    },
+    {
+      "epoch": 0.19671376107499686,
+      "grad_norm": 458.807861328125,
+      "learning_rate": 7.5982733957842304e-06,
+      "loss": 29.1032,
+      "step": 48690
+    },
+    {
+      "epoch": 0.1967541623403645,
+      "grad_norm": 330.7550354003906,
+      "learning_rate": 7.597068308107165e-06,
+      "loss": 44.6583,
+      "step": 48700
+    },
+    {
+      "epoch": 0.19679456360573214,
+      "grad_norm": 788.6397705078125,
+      "learning_rate": 7.595863013793292e-06,
+      "loss": 63.0783,
+      "step": 48710
+    },
+    {
+      "epoch": 0.19683496487109975,
+      "grad_norm": 796.1050415039062,
+      "learning_rate": 7.594657512938513e-06,
+      "loss": 44.8516,
+      "step": 48720
+    },
+    {
+      "epoch": 0.1968753661364674,
+      "grad_norm": 555.8182983398438,
+      "learning_rate": 7.593451805638743e-06,
+      "loss": 41.593,
+      "step": 48730
+    },
+    {
+      "epoch": 0.19691576740183503,
+      "grad_norm": 864.0370483398438,
+      "learning_rate": 7.592245891989914e-06,
+      "loss": 40.4903,
+      "step": 48740
+    },
+    {
+      "epoch": 0.19695616866720267,
+      "grad_norm": 816.231689453125,
+      "learning_rate": 7.5910397720879785e-06,
+      "loss": 63.3931,
+      "step": 48750
+    },
+    {
+      "epoch": 0.19699656993257028,
+      "grad_norm": 503.22900390625,
+      "learning_rate": 7.589833446028898e-06,
+      "loss": 76.2282,
+      "step": 48760
+    },
+    {
+      "epoch": 0.19703697119793792,
+      "grad_norm": 488.8203430175781,
+      "learning_rate": 7.5886269139086565e-06,
+      "loss": 52.7377,
+      "step": 48770
+    },
+    {
+      "epoch": 0.19707737246330556,
+      "grad_norm": 693.25,
+      "learning_rate": 7.587420175823252e-06,
+      "loss": 46.3758,
+      "step": 48780
+    },
+    {
+      "epoch": 0.19711777372867317,
+      "grad_norm": 998.3317260742188,
+      "learning_rate": 7.586213231868699e-06,
+      "loss": 65.0064,
+      "step": 48790
+    },
+    {
+      "epoch": 0.1971581749940408,
+      "grad_norm": 2683.894775390625,
+      "learning_rate": 7.585006082141028e-06,
+      "loss": 71.4827,
+      "step": 48800
+    },
+    {
+      "epoch": 0.19719857625940845,
+      "grad_norm": 496.90478515625,
+      "learning_rate": 7.583798726736286e-06,
+      "loss": 35.5153,
+      "step": 48810
+    },
+    {
+      "epoch": 0.19723897752477607,
+      "grad_norm": 419.7237243652344,
+      "learning_rate": 7.5825911657505365e-06,
+      "loss": 60.3092,
+      "step": 48820
+    },
+    {
+      "epoch": 0.1972793787901437,
+      "grad_norm": 523.56982421875,
+      "learning_rate": 7.581383399279863e-06,
+      "loss": 61.9478,
+      "step": 48830
+    },
+    {
+      "epoch": 0.19731978005551135,
+      "grad_norm": 812.9863891601562,
+      "learning_rate": 7.580175427420358e-06,
+      "loss": 52.6858,
+      "step": 48840
+    },
+    {
+      "epoch": 0.19736018132087896,
+      "grad_norm": 430.5237731933594,
+      "learning_rate": 7.578967250268137e-06,
+      "loss": 36.8604,
+      "step": 48850
+    },
+    {
+      "epoch": 0.1974005825862466,
+      "grad_norm": 576.6433715820312,
+      "learning_rate": 7.577758867919325e-06,
+      "loss": 31.4436,
+      "step": 48860
+    },
+    {
+      "epoch": 0.19744098385161424,
+      "grad_norm": 369.1871032714844,
+      "learning_rate": 7.576550280470072e-06,
+      "loss": 37.8606,
+      "step": 48870
+    },
+    {
+      "epoch": 0.19748138511698185,
+      "grad_norm": 373.350830078125,
+      "learning_rate": 7.5753414880165365e-06,
+      "loss": 44.6778,
+      "step": 48880
+    },
+    {
+      "epoch": 0.1975217863823495,
+      "grad_norm": 330.2793273925781,
+      "learning_rate": 7.5741324906548996e-06,
+      "loss": 40.3745,
+      "step": 48890
+    },
+    {
+      "epoch": 0.19756218764771713,
+      "grad_norm": 365.4303283691406,
+      "learning_rate": 7.572923288481355e-06,
+      "loss": 43.1003,
+      "step": 48900
+    },
+    {
+      "epoch": 0.19760258891308477,
+      "grad_norm": 363.63287353515625,
+      "learning_rate": 7.571713881592109e-06,
+      "loss": 30.675,
+      "step": 48910
+    },
+    {
+      "epoch": 0.19764299017845238,
+      "grad_norm": 387.30242919921875,
+      "learning_rate": 7.570504270083394e-06,
+      "loss": 44.3049,
+      "step": 48920
+    },
+    {
+      "epoch": 0.19768339144382002,
+      "grad_norm": 514.7061157226562,
+      "learning_rate": 7.569294454051452e-06,
+      "loss": 42.9896,
+      "step": 48930
+    },
+    {
+      "epoch": 0.19772379270918766,
+      "grad_norm": 803.383544921875,
+      "learning_rate": 7.568084433592542e-06,
+      "loss": 51.6624,
+      "step": 48940
+    },
+    {
+      "epoch": 0.19776419397455527,
+      "grad_norm": 410.1513671875,
+      "learning_rate": 7.566874208802939e-06,
+      "loss": 44.9291,
+      "step": 48950
+    },
+    {
+      "epoch": 0.19780459523992291,
+      "grad_norm": 790.0667724609375,
+      "learning_rate": 7.5656637797789335e-06,
+      "loss": 65.2939,
+      "step": 48960
+    },
+    {
+      "epoch": 0.19784499650529055,
+      "grad_norm": 403.3978271484375,
+      "learning_rate": 7.564453146616837e-06,
+      "loss": 31.9904,
+      "step": 48970
+    },
+    {
+      "epoch": 0.19788539777065817,
+      "grad_norm": 1210.605224609375,
+      "learning_rate": 7.563242309412975e-06,
+      "loss": 58.8124,
+      "step": 48980
+    },
+    {
+      "epoch": 0.1979257990360258,
+      "grad_norm": 963.6305541992188,
+      "learning_rate": 7.562031268263686e-06,
+      "loss": 55.9536,
+      "step": 48990
+    },
+    {
+      "epoch": 0.19796620030139345,
+      "grad_norm": 704.73388671875,
+      "learning_rate": 7.5608200232653254e-06,
+      "loss": 63.4184,
+      "step": 49000
+    },
+    {
+      "epoch": 0.19800660156676106,
+      "grad_norm": 729.5773315429688,
+      "learning_rate": 7.5596085745142654e-06,
+      "loss": 50.0997,
+      "step": 49010
+    },
+    {
+      "epoch": 0.1980470028321287,
+      "grad_norm": 961.2758178710938,
+      "learning_rate": 7.558396922106903e-06,
+      "loss": 48.7712,
+      "step": 49020
+    },
+    {
+      "epoch": 0.19808740409749634,
+      "grad_norm": 398.1002197265625,
+      "learning_rate": 7.557185066139638e-06,
+      "loss": 53.2708,
+      "step": 49030
+    },
+    {
+      "epoch": 0.19812780536286395,
+      "grad_norm": 802.1510009765625,
+      "learning_rate": 7.555973006708892e-06,
+      "loss": 39.0903,
+      "step": 49040
+    },
+    {
+      "epoch": 0.1981682066282316,
+      "grad_norm": 373.4618835449219,
+      "learning_rate": 7.554760743911104e-06,
+      "loss": 61.9456,
+      "step": 49050
+    },
+    {
+      "epoch": 0.19820860789359923,
+      "grad_norm": 475.2350158691406,
+      "learning_rate": 7.553548277842729e-06,
+      "loss": 54.6043,
+      "step": 49060
+    },
+    {
+      "epoch": 0.19824900915896687,
+      "grad_norm": 274.4440002441406,
+      "learning_rate": 7.5523356086002364e-06,
+      "loss": 34.3625,
+      "step": 49070
+    },
+    {
+      "epoch": 0.19828941042433448,
+      "grad_norm": 786.3656616210938,
+      "learning_rate": 7.551122736280113e-06,
+      "loss": 44.7099,
+      "step": 49080
+    },
+    {
+      "epoch": 0.19832981168970212,
+      "grad_norm": 145.26461791992188,
+      "learning_rate": 7.549909660978863e-06,
+      "loss": 44.07,
+      "step": 49090
+    },
+    {
+      "epoch": 0.19837021295506976,
+      "grad_norm": 560.5723266601562,
+      "learning_rate": 7.548696382793002e-06,
+      "loss": 41.1856,
+      "step": 49100
+    },
+    {
+      "epoch": 0.19841061422043738,
+      "grad_norm": 351.9812316894531,
+      "learning_rate": 7.547482901819066e-06,
+      "loss": 30.168,
+      "step": 49110
+    },
+    {
+      "epoch": 0.19845101548580502,
+      "grad_norm": 607.218994140625,
+      "learning_rate": 7.5462692181536094e-06,
+      "loss": 34.1062,
+      "step": 49120
+    },
+    {
+      "epoch": 0.19849141675117266,
+      "grad_norm": 1311.45166015625,
+      "learning_rate": 7.545055331893195e-06,
+      "loss": 34.9294,
+      "step": 49130
+    },
+    {
+      "epoch": 0.19853181801654027,
+      "grad_norm": 1162.9915771484375,
+      "learning_rate": 7.543841243134409e-06,
+      "loss": 40.0276,
+      "step": 49140
+    },
+    {
+      "epoch": 0.1985722192819079,
+      "grad_norm": 785.1849975585938,
+      "learning_rate": 7.5426269519738495e-06,
+      "loss": 63.0314,
+      "step": 49150
+    },
+    {
+      "epoch": 0.19861262054727555,
+      "grad_norm": 686.400634765625,
+      "learning_rate": 7.541412458508133e-06,
+      "loss": 64.1595,
+      "step": 49160
+    },
+    {
+      "epoch": 0.19865302181264316,
+      "grad_norm": 589.6557006835938,
+      "learning_rate": 7.54019776283389e-06,
+      "loss": 48.8167,
+      "step": 49170
+    },
+    {
+      "epoch": 0.1986934230780108,
+      "grad_norm": 700.9690551757812,
+      "learning_rate": 7.53898286504777e-06,
+      "loss": 52.2654,
+      "step": 49180
+    },
+    {
+      "epoch": 0.19873382434337844,
+      "grad_norm": 361.7765197753906,
+      "learning_rate": 7.537767765246436e-06,
+      "loss": 28.0827,
+      "step": 49190
+    },
+    {
+      "epoch": 0.19877422560874605,
+      "grad_norm": 1053.419677734375,
+      "learning_rate": 7.536552463526565e-06,
+      "loss": 82.5294,
+      "step": 49200
+    },
+    {
+      "epoch": 0.1988146268741137,
+      "grad_norm": 401.47967529296875,
+      "learning_rate": 7.535336959984858e-06,
+      "loss": 40.8976,
+      "step": 49210
+    },
+    {
+      "epoch": 0.19885502813948133,
+      "grad_norm": 695.1689453125,
+      "learning_rate": 7.5341212547180246e-06,
+      "loss": 37.4906,
+      "step": 49220
+    },
+    {
+      "epoch": 0.19889542940484897,
+      "grad_norm": 522.3072509765625,
+      "learning_rate": 7.532905347822792e-06,
+      "loss": 56.5834,
+      "step": 49230
+    },
+    {
+      "epoch": 0.19893583067021658,
+      "grad_norm": 484.8719787597656,
+      "learning_rate": 7.5316892393959064e-06,
+      "loss": 83.0188,
+      "step": 49240
+    },
+    {
+      "epoch": 0.19897623193558422,
+      "grad_norm": 372.8064880371094,
+      "learning_rate": 7.530472929534126e-06,
+      "loss": 47.5193,
+      "step": 49250
+    },
+    {
+      "epoch": 0.19901663320095186,
+      "grad_norm": 275.8352355957031,
+      "learning_rate": 7.529256418334228e-06,
+      "loss": 48.3916,
+      "step": 49260
+    },
+    {
+      "epoch": 0.19905703446631948,
+      "grad_norm": 383.5005187988281,
+      "learning_rate": 7.528039705893006e-06,
+      "loss": 39.6332,
+      "step": 49270
+    },
+    {
+      "epoch": 0.19909743573168712,
+      "grad_norm": 810.511474609375,
+      "learning_rate": 7.5268227923072665e-06,
+      "loss": 42.8357,
+      "step": 49280
+    },
+    {
+      "epoch": 0.19913783699705476,
+      "grad_norm": 441.7340087890625,
+      "learning_rate": 7.525605677673831e-06,
+      "loss": 112.611,
+      "step": 49290
+    },
+    {
+      "epoch": 0.19917823826242237,
+      "grad_norm": 373.8868408203125,
+      "learning_rate": 7.524388362089545e-06,
+      "loss": 31.0369,
+      "step": 49300
+    },
+    {
+      "epoch": 0.19921863952779,
+      "grad_norm": 386.7220153808594,
+      "learning_rate": 7.523170845651263e-06,
+      "loss": 39.6656,
+      "step": 49310
+    },
+    {
+      "epoch": 0.19925904079315765,
+      "grad_norm": 604.7882690429688,
+      "learning_rate": 7.521953128455856e-06,
+      "loss": 37.7455,
+      "step": 49320
+    },
+    {
+      "epoch": 0.19929944205852526,
+      "grad_norm": 498.83953857421875,
+      "learning_rate": 7.520735210600213e-06,
+      "loss": 42.6832,
+      "step": 49330
+    },
+    {
+      "epoch": 0.1993398433238929,
+      "grad_norm": 430.9537353515625,
+      "learning_rate": 7.519517092181237e-06,
+      "loss": 56.7781,
+      "step": 49340
+    },
+    {
+      "epoch": 0.19938024458926054,
+      "grad_norm": 435.7265625,
+      "learning_rate": 7.518298773295849e-06,
+      "loss": 58.8235,
+      "step": 49350
+    },
+    {
+      "epoch": 0.19942064585462815,
+      "grad_norm": 358.5899353027344,
+      "learning_rate": 7.517080254040985e-06,
+      "loss": 37.2343,
+      "step": 49360
+    },
+    {
+      "epoch": 0.1994610471199958,
+      "grad_norm": 412.10986328125,
+      "learning_rate": 7.5158615345136e-06,
+      "loss": 29.2832,
+      "step": 49370
+    },
+    {
+      "epoch": 0.19950144838536343,
+      "grad_norm": 416.3394470214844,
+      "learning_rate": 7.514642614810655e-06,
+      "loss": 42.876,
+      "step": 49380
+    },
+    {
+      "epoch": 0.19954184965073107,
+      "grad_norm": 811.4172973632812,
+      "learning_rate": 7.51342349502914e-06,
+      "loss": 94.7614,
+      "step": 49390
+    },
+    {
+      "epoch": 0.19958225091609869,
+      "grad_norm": 792.0985717773438,
+      "learning_rate": 7.512204175266052e-06,
+      "loss": 45.6878,
+      "step": 49400
+    },
+    {
+      "epoch": 0.19962265218146633,
+      "grad_norm": 809.537353515625,
+      "learning_rate": 7.510984655618407e-06,
+      "loss": 37.4703,
+      "step": 49410
+    },
+    {
+      "epoch": 0.19966305344683397,
+      "grad_norm": 552.7601928710938,
+      "learning_rate": 7.509764936183237e-06,
+      "loss": 44.3597,
+      "step": 49420
+    },
+    {
+      "epoch": 0.19970345471220158,
+      "grad_norm": 576.9872436523438,
+      "learning_rate": 7.5085450170575876e-06,
+      "loss": 51.2273,
+      "step": 49430
+    },
+    {
+      "epoch": 0.19974385597756922,
+      "grad_norm": 422.26397705078125,
+      "learning_rate": 7.5073248983385265e-06,
+      "loss": 48.0467,
+      "step": 49440
+    },
+    {
+      "epoch": 0.19978425724293686,
+      "grad_norm": 221.23141479492188,
+      "learning_rate": 7.50610458012313e-06,
+      "loss": 38.9246,
+      "step": 49450
+    },
+    {
+      "epoch": 0.19982465850830447,
+      "grad_norm": 475.1904296875,
+      "learning_rate": 7.504884062508493e-06,
+      "loss": 47.4083,
+      "step": 49460
+    },
+    {
+      "epoch": 0.1998650597736721,
+      "grad_norm": 901.5618896484375,
+      "learning_rate": 7.503663345591726e-06,
+      "loss": 29.5857,
+      "step": 49470
+    },
+    {
+      "epoch": 0.19990546103903975,
+      "grad_norm": 417.44183349609375,
+      "learning_rate": 7.502442429469956e-06,
+      "loss": 38.4475,
+      "step": 49480
+    },
+    {
+      "epoch": 0.19994586230440736,
+      "grad_norm": 512.4078369140625,
+      "learning_rate": 7.501221314240329e-06,
+      "loss": 27.7243,
+      "step": 49490
+    },
+    {
+      "epoch": 0.199986263569775,
+      "grad_norm": 617.75390625,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 33.2255,
+      "step": 49500
+    },
+    {
+      "epoch": 0.20002666483514264,
+      "grad_norm": 1441.7674560546875,
+      "learning_rate": 7.4987784868461455e-06,
+      "loss": 67.9786,
+      "step": 49510
+    },
+    {
+      "epoch": 0.20006706610051025,
+      "grad_norm": 520.3792724609375,
+      "learning_rate": 7.497556774875953e-06,
+      "loss": 41.5277,
+      "step": 49520
+    },
+    {
+      "epoch": 0.2001074673658779,
+      "grad_norm": 852.6936645507812,
+      "learning_rate": 7.496334864186632e-06,
+      "loss": 34.2242,
+      "step": 49530
+    },
+    {
+      "epoch": 0.20014786863124553,
+      "grad_norm": 488.62506103515625,
+      "learning_rate": 7.4951127548754025e-06,
+      "loss": 52.5192,
+      "step": 49540
+    },
+    {
+      "epoch": 0.20018826989661317,
+      "grad_norm": 405.8353576660156,
+      "learning_rate": 7.4938904470395e-06,
+      "loss": 52.7841,
+      "step": 49550
+    },
+    {
+      "epoch": 0.2002286711619808,
+      "grad_norm": 660.2997436523438,
+      "learning_rate": 7.492667940776182e-06,
+      "loss": 64.3081,
+      "step": 49560
+    },
+    {
+      "epoch": 0.20026907242734843,
+      "grad_norm": 783.1737670898438,
+      "learning_rate": 7.491445236182715e-06,
+      "loss": 50.6159,
+      "step": 49570
+    },
+    {
+      "epoch": 0.20030947369271607,
+      "grad_norm": 392.0615234375,
+      "learning_rate": 7.490222333356384e-06,
+      "loss": 41.7744,
+      "step": 49580
+    },
+    {
+      "epoch": 0.20034987495808368,
+      "grad_norm": 486.9169616699219,
+      "learning_rate": 7.488999232394492e-06,
+      "loss": 37.5612,
+      "step": 49590
+    },
+    {
+      "epoch": 0.20039027622345132,
+      "grad_norm": 674.69775390625,
+      "learning_rate": 7.487775933394353e-06,
+      "loss": 58.0785,
+      "step": 49600
+    },
+    {
+      "epoch": 0.20043067748881896,
+      "grad_norm": 372.39263916015625,
+      "learning_rate": 7.4865524364533e-06,
+      "loss": 54.9935,
+      "step": 49610
+    },
+    {
+      "epoch": 0.20047107875418657,
+      "grad_norm": 583.2684936523438,
+      "learning_rate": 7.485328741668683e-06,
+      "loss": 52.8313,
+      "step": 49620
+    },
+    {
+      "epoch": 0.2005114800195542,
+      "grad_norm": 674.1826171875,
+      "learning_rate": 7.484104849137862e-06,
+      "loss": 45.282,
+      "step": 49630
+    },
+    {
+      "epoch": 0.20055188128492185,
+      "grad_norm": 478.70257568359375,
+      "learning_rate": 7.482880758958219e-06,
+      "loss": 40.7054,
+      "step": 49640
+    },
+    {
+      "epoch": 0.20059228255028946,
+      "grad_norm": 557.6510620117188,
+      "learning_rate": 7.48165647122715e-06,
+      "loss": 53.2588,
+      "step": 49650
+    },
+    {
+      "epoch": 0.2006326838156571,
+      "grad_norm": 708.6511840820312,
+      "learning_rate": 7.480431986042065e-06,
+      "loss": 48.5664,
+      "step": 49660
+    },
+    {
+      "epoch": 0.20067308508102474,
+      "grad_norm": 655.1844482421875,
+      "learning_rate": 7.47920730350039e-06,
+      "loss": 45.7343,
+      "step": 49670
+    },
+    {
+      "epoch": 0.20071348634639236,
+      "grad_norm": 1270.8260498046875,
+      "learning_rate": 7.477982423699568e-06,
+      "loss": 40.3143,
+      "step": 49680
+    },
+    {
+      "epoch": 0.20075388761176,
+      "grad_norm": 591.7367553710938,
+      "learning_rate": 7.476757346737057e-06,
+      "loss": 38.9331,
+      "step": 49690
+    },
+    {
+      "epoch": 0.20079428887712764,
+      "grad_norm": 591.572509765625,
+      "learning_rate": 7.47553207271033e-06,
+      "loss": 44.8786,
+      "step": 49700
+    },
+    {
+      "epoch": 0.20083469014249528,
+      "grad_norm": 402.4839172363281,
+      "learning_rate": 7.474306601716877e-06,
+      "loss": 51.7751,
+      "step": 49710
+    },
+    {
+      "epoch": 0.2008750914078629,
+      "grad_norm": 595.6873168945312,
+      "learning_rate": 7.473080933854205e-06,
+      "loss": 44.1173,
+      "step": 49720
+    },
+    {
+      "epoch": 0.20091549267323053,
+      "grad_norm": 1096.7359619140625,
+      "learning_rate": 7.471855069219831e-06,
+      "loss": 46.3273,
+      "step": 49730
+    },
+    {
+      "epoch": 0.20095589393859817,
+      "grad_norm": 590.0582275390625,
+      "learning_rate": 7.470629007911294e-06,
+      "loss": 54.4458,
+      "step": 49740
+    },
+    {
+      "epoch": 0.20099629520396578,
+      "grad_norm": 453.25115966796875,
+      "learning_rate": 7.469402750026147e-06,
+      "loss": 56.3232,
+      "step": 49750
+    },
+    {
+      "epoch": 0.20103669646933342,
+      "grad_norm": 566.2916870117188,
+      "learning_rate": 7.468176295661955e-06,
+      "loss": 55.3626,
+      "step": 49760
+    },
+    {
+      "epoch": 0.20107709773470106,
+      "grad_norm": 454.93035888671875,
+      "learning_rate": 7.466949644916301e-06,
+      "loss": 52.3334,
+      "step": 49770
+    },
+    {
+      "epoch": 0.20111749900006867,
+      "grad_norm": 637.5740356445312,
+      "learning_rate": 7.465722797886788e-06,
+      "loss": 51.3216,
+      "step": 49780
+    },
+    {
+      "epoch": 0.2011579002654363,
+      "grad_norm": 509.33929443359375,
+      "learning_rate": 7.464495754671027e-06,
+      "loss": 41.1044,
+      "step": 49790
+    },
+    {
+      "epoch": 0.20119830153080395,
+      "grad_norm": 600.6532592773438,
+      "learning_rate": 7.4632685153666505e-06,
+      "loss": 54.9956,
+      "step": 49800
+    },
+    {
+      "epoch": 0.20123870279617156,
+      "grad_norm": 800.8745727539062,
+      "learning_rate": 7.462041080071301e-06,
+      "loss": 78.0207,
+      "step": 49810
+    },
+    {
+      "epoch": 0.2012791040615392,
+      "grad_norm": 823.1187744140625,
+      "learning_rate": 7.460813448882643e-06,
+      "loss": 51.9847,
+      "step": 49820
+    },
+    {
+      "epoch": 0.20131950532690684,
+      "grad_norm": 647.9494018554688,
+      "learning_rate": 7.459585621898353e-06,
+      "loss": 35.7294,
+      "step": 49830
+    },
+    {
+      "epoch": 0.20135990659227446,
+      "grad_norm": 567.6367797851562,
+      "learning_rate": 7.4583575992161235e-06,
+      "loss": 35.5241,
+      "step": 49840
+    },
+    {
+      "epoch": 0.2014003078576421,
+      "grad_norm": 414.89678955078125,
+      "learning_rate": 7.457129380933662e-06,
+      "loss": 56.1163,
+      "step": 49850
+    },
+    {
+      "epoch": 0.20144070912300974,
+      "grad_norm": 270.87213134765625,
+      "learning_rate": 7.4559009671486906e-06,
+      "loss": 47.756,
+      "step": 49860
+    },
+    {
+      "epoch": 0.20148111038837738,
+      "grad_norm": 447.1685791015625,
+      "learning_rate": 7.454672357958951e-06,
+      "loss": 29.9517,
+      "step": 49870
+    },
+    {
+      "epoch": 0.201521511653745,
+      "grad_norm": 537.8594970703125,
+      "learning_rate": 7.453443553462198e-06,
+      "loss": 32.8862,
+      "step": 49880
+    },
+    {
+      "epoch": 0.20156191291911263,
+      "grad_norm": 505.7225036621094,
+      "learning_rate": 7.4522145537562015e-06,
+      "loss": 43.0821,
+      "step": 49890
+    },
+    {
+      "epoch": 0.20160231418448027,
+      "grad_norm": 372.20123291015625,
+      "learning_rate": 7.450985358938747e-06,
+      "loss": 36.8695,
+      "step": 49900
+    },
+    {
+      "epoch": 0.20164271544984788,
+      "grad_norm": 403.6573791503906,
+      "learning_rate": 7.449755969107635e-06,
+      "loss": 42.9503,
+      "step": 49910
+    },
+    {
+      "epoch": 0.20168311671521552,
+      "grad_norm": 328.0631408691406,
+      "learning_rate": 7.4485263843606835e-06,
+      "loss": 70.2635,
+      "step": 49920
+    },
+    {
+      "epoch": 0.20172351798058316,
+      "grad_norm": 356.9211120605469,
+      "learning_rate": 7.447296604795726e-06,
+      "loss": 40.2945,
+      "step": 49930
+    },
+    {
+      "epoch": 0.20176391924595077,
+      "grad_norm": 574.4561157226562,
+      "learning_rate": 7.4460666305106084e-06,
+      "loss": 40.958,
+      "step": 49940
+    },
+    {
+      "epoch": 0.2018043205113184,
+      "grad_norm": 480.0083923339844,
+      "learning_rate": 7.444836461603195e-06,
+      "loss": 66.8758,
+      "step": 49950
+    },
+    {
+      "epoch": 0.20184472177668605,
+      "grad_norm": 632.5971069335938,
+      "learning_rate": 7.443606098171363e-06,
+      "loss": 69.6456,
+      "step": 49960
+    },
+    {
+      "epoch": 0.20188512304205367,
+      "grad_norm": 394.3634033203125,
+      "learning_rate": 7.442375540313012e-06,
+      "loss": 42.7379,
+      "step": 49970
+    },
+    {
+      "epoch": 0.2019255243074213,
+      "grad_norm": 620.8236083984375,
+      "learning_rate": 7.441144788126045e-06,
+      "loss": 53.8149,
+      "step": 49980
+    },
+    {
+      "epoch": 0.20196592557278895,
+      "grad_norm": 620.8430786132812,
+      "learning_rate": 7.4399138417083925e-06,
+      "loss": 73.245,
+      "step": 49990
+    },
+    {
+      "epoch": 0.20200632683815656,
+      "grad_norm": 610.7430419921875,
+      "learning_rate": 7.438682701157993e-06,
+      "loss": 46.0496,
+      "step": 50000
+    },
+    {
+      "epoch": 0.2020467281035242,
+      "grad_norm": 587.0072021484375,
+      "learning_rate": 7.437451366572803e-06,
+      "loss": 49.8662,
+      "step": 50010
+    },
+    {
+      "epoch": 0.20208712936889184,
+      "grad_norm": 126.13793182373047,
+      "learning_rate": 7.436219838050793e-06,
+      "loss": 46.2343,
+      "step": 50020
+    },
+    {
+      "epoch": 0.20212753063425948,
+      "grad_norm": 680.1400146484375,
+      "learning_rate": 7.4349881156899525e-06,
+      "loss": 49.6154,
+      "step": 50030
+    },
+    {
+      "epoch": 0.2021679318996271,
+      "grad_norm": 800.9889526367188,
+      "learning_rate": 7.433756199588282e-06,
+      "loss": 63.2274,
+      "step": 50040
+    },
+    {
+      "epoch": 0.20220833316499473,
+      "grad_norm": 1124.2677001953125,
+      "learning_rate": 7.4325240898438e-06,
+      "loss": 52.991,
+      "step": 50050
+    },
+    {
+      "epoch": 0.20224873443036237,
+      "grad_norm": 466.38226318359375,
+      "learning_rate": 7.4312917865545406e-06,
+      "loss": 34.7683,
+      "step": 50060
+    },
+    {
+      "epoch": 0.20228913569572998,
+      "grad_norm": 294.4744567871094,
+      "learning_rate": 7.430059289818552e-06,
+      "loss": 38.1768,
+      "step": 50070
+    },
+    {
+      "epoch": 0.20232953696109762,
+      "grad_norm": 1064.6177978515625,
+      "learning_rate": 7.4288265997338985e-06,
+      "loss": 49.4151,
+      "step": 50080
+    },
+    {
+      "epoch": 0.20236993822646526,
+      "grad_norm": 502.5269775390625,
+      "learning_rate": 7.427593716398658e-06,
+      "loss": 42.8233,
+      "step": 50090
+    },
+    {
+      "epoch": 0.20241033949183287,
+      "grad_norm": 834.306884765625,
+      "learning_rate": 7.426360639910927e-06,
+      "loss": 46.4767,
+      "step": 50100
+    },
+    {
+      "epoch": 0.20245074075720051,
+      "grad_norm": 543.8235473632812,
+      "learning_rate": 7.425127370368815e-06,
+      "loss": 47.3622,
+      "step": 50110
+    },
+    {
+      "epoch": 0.20249114202256815,
+      "grad_norm": 519.2015380859375,
+      "learning_rate": 7.423893907870449e-06,
+      "loss": 65.9216,
+      "step": 50120
+    },
+    {
+      "epoch": 0.20253154328793577,
+      "grad_norm": 728.6759643554688,
+      "learning_rate": 7.422660252513969e-06,
+      "loss": 88.8377,
+      "step": 50130
+    },
+    {
+      "epoch": 0.2025719445533034,
+      "grad_norm": 733.7372436523438,
+      "learning_rate": 7.421426404397531e-06,
+      "loss": 52.0609,
+      "step": 50140
+    },
+    {
+      "epoch": 0.20261234581867105,
+      "grad_norm": 624.3734130859375,
+      "learning_rate": 7.420192363619305e-06,
+      "loss": 58.3511,
+      "step": 50150
+    },
+    {
+      "epoch": 0.20265274708403866,
+      "grad_norm": 634.50634765625,
+      "learning_rate": 7.418958130277483e-06,
+      "loss": 53.5587,
+      "step": 50160
+    },
+    {
+      "epoch": 0.2026931483494063,
+      "grad_norm": 273.2563781738281,
+      "learning_rate": 7.417723704470261e-06,
+      "loss": 44.3106,
+      "step": 50170
+    },
+    {
+      "epoch": 0.20273354961477394,
+      "grad_norm": 729.7420043945312,
+      "learning_rate": 7.4164890862958615e-06,
+      "loss": 38.8169,
+      "step": 50180
+    },
+    {
+      "epoch": 0.20277395088014158,
+      "grad_norm": 476.1905822753906,
+      "learning_rate": 7.415254275852515e-06,
+      "loss": 40.2199,
+      "step": 50190
+    },
+    {
+      "epoch": 0.2028143521455092,
+      "grad_norm": 646.1971435546875,
+      "learning_rate": 7.414019273238471e-06,
+      "loss": 39.1268,
+      "step": 50200
+    },
+    {
+      "epoch": 0.20285475341087683,
+      "grad_norm": 422.8984680175781,
+      "learning_rate": 7.4127840785519915e-06,
+      "loss": 40.8058,
+      "step": 50210
+    },
+    {
+      "epoch": 0.20289515467624447,
+      "grad_norm": 502.76092529296875,
+      "learning_rate": 7.411548691891357e-06,
+      "loss": 49.4966,
+      "step": 50220
+    },
+    {
+      "epoch": 0.20293555594161208,
+      "grad_norm": 832.3966674804688,
+      "learning_rate": 7.41031311335486e-06,
+      "loss": 85.8858,
+      "step": 50230
+    },
+    {
+      "epoch": 0.20297595720697972,
+      "grad_norm": 478.6716613769531,
+      "learning_rate": 7.409077343040809e-06,
+      "loss": 58.2753,
+      "step": 50240
+    },
+    {
+      "epoch": 0.20301635847234736,
+      "grad_norm": 610.2609252929688,
+      "learning_rate": 7.407841381047533e-06,
+      "loss": 53.1357,
+      "step": 50250
+    },
+    {
+      "epoch": 0.20305675973771498,
+      "grad_norm": 498.36273193359375,
+      "learning_rate": 7.406605227473367e-06,
+      "loss": 32.1867,
+      "step": 50260
+    },
+    {
+      "epoch": 0.20309716100308262,
+      "grad_norm": 374.0442810058594,
+      "learning_rate": 7.405368882416668e-06,
+      "loss": 48.0912,
+      "step": 50270
+    },
+    {
+      "epoch": 0.20313756226845026,
+      "grad_norm": 329.8494567871094,
+      "learning_rate": 7.404132345975806e-06,
+      "loss": 45.6471,
+      "step": 50280
+    },
+    {
+      "epoch": 0.20317796353381787,
+      "grad_norm": 340.72100830078125,
+      "learning_rate": 7.4028956182491665e-06,
+      "loss": 36.8876,
+      "step": 50290
+    },
+    {
+      "epoch": 0.2032183647991855,
+      "grad_norm": 695.551025390625,
+      "learning_rate": 7.401658699335151e-06,
+      "loss": 39.3887,
+      "step": 50300
+    },
+    {
+      "epoch": 0.20325876606455315,
+      "grad_norm": 879.5848999023438,
+      "learning_rate": 7.400421589332175e-06,
+      "loss": 46.7177,
+      "step": 50310
+    },
+    {
+      "epoch": 0.20329916732992076,
+      "grad_norm": 482.7911682128906,
+      "learning_rate": 7.39918428833867e-06,
+      "loss": 60.0693,
+      "step": 50320
+    },
+    {
+      "epoch": 0.2033395685952884,
+      "grad_norm": 692.8958129882812,
+      "learning_rate": 7.397946796453081e-06,
+      "loss": 52.0577,
+      "step": 50330
+    },
+    {
+      "epoch": 0.20337996986065604,
+      "grad_norm": 384.43829345703125,
+      "learning_rate": 7.39670911377387e-06,
+      "loss": 27.1583,
+      "step": 50340
+    },
+    {
+      "epoch": 0.20342037112602368,
+      "grad_norm": 122.59984588623047,
+      "learning_rate": 7.395471240399515e-06,
+      "loss": 50.0477,
+      "step": 50350
+    },
+    {
+      "epoch": 0.2034607723913913,
+      "grad_norm": 545.1742553710938,
+      "learning_rate": 7.394233176428508e-06,
+      "loss": 29.567,
+      "step": 50360
+    },
+    {
+      "epoch": 0.20350117365675893,
+      "grad_norm": 628.303955078125,
+      "learning_rate": 7.3929949219593545e-06,
+      "loss": 48.9455,
+      "step": 50370
+    },
+    {
+      "epoch": 0.20354157492212657,
+      "grad_norm": 540.7025756835938,
+      "learning_rate": 7.391756477090577e-06,
+      "loss": 43.8203,
+      "step": 50380
+    },
+    {
+      "epoch": 0.20358197618749418,
+      "grad_norm": 628.2692260742188,
+      "learning_rate": 7.3905178419207126e-06,
+      "loss": 34.8731,
+      "step": 50390
+    },
+    {
+      "epoch": 0.20362237745286182,
+      "grad_norm": 638.9990844726562,
+      "learning_rate": 7.3892790165483164e-06,
+      "loss": 36.1334,
+      "step": 50400
+    },
+    {
+      "epoch": 0.20366277871822946,
+      "grad_norm": 684.90380859375,
+      "learning_rate": 7.388040001071953e-06,
+      "loss": 53.3651,
+      "step": 50410
+    },
+    {
+      "epoch": 0.20370317998359708,
+      "grad_norm": 378.3131103515625,
+      "learning_rate": 7.386800795590208e-06,
+      "loss": 37.3771,
+      "step": 50420
+    },
+    {
+      "epoch": 0.20374358124896472,
+      "grad_norm": 987.3731079101562,
+      "learning_rate": 7.385561400201675e-06,
+      "loss": 36.6902,
+      "step": 50430
+    },
+    {
+      "epoch": 0.20378398251433236,
+      "grad_norm": 1104.762451171875,
+      "learning_rate": 7.384321815004971e-06,
+      "loss": 56.715,
+      "step": 50440
+    },
+    {
+      "epoch": 0.20382438377969997,
+      "grad_norm": 616.8389282226562,
+      "learning_rate": 7.383082040098723e-06,
+      "loss": 39.1386,
+      "step": 50450
+    },
+    {
+      "epoch": 0.2038647850450676,
+      "grad_norm": 618.7867431640625,
+      "learning_rate": 7.381842075581573e-06,
+      "loss": 40.5047,
+      "step": 50460
+    },
+    {
+      "epoch": 0.20390518631043525,
+      "grad_norm": 629.0042114257812,
+      "learning_rate": 7.380601921552181e-06,
+      "loss": 51.9558,
+      "step": 50470
+    },
+    {
+      "epoch": 0.20394558757580286,
+      "grad_norm": 1132.140869140625,
+      "learning_rate": 7.379361578109218e-06,
+      "loss": 62.1229,
+      "step": 50480
+    },
+    {
+      "epoch": 0.2039859888411705,
+      "grad_norm": 673.3324584960938,
+      "learning_rate": 7.378121045351378e-06,
+      "loss": 44.218,
+      "step": 50490
+    },
+    {
+      "epoch": 0.20402639010653814,
+      "grad_norm": 583.1757202148438,
+      "learning_rate": 7.376880323377357e-06,
+      "loss": 41.8356,
+      "step": 50500
+    },
+    {
+      "epoch": 0.20406679137190578,
+      "grad_norm": 732.2437744140625,
+      "learning_rate": 7.375639412285877e-06,
+      "loss": 41.8455,
+      "step": 50510
+    },
+    {
+      "epoch": 0.2041071926372734,
+      "grad_norm": 180.8105926513672,
+      "learning_rate": 7.374398312175674e-06,
+      "loss": 30.912,
+      "step": 50520
+    },
+    {
+      "epoch": 0.20414759390264103,
+      "grad_norm": 564.68603515625,
+      "learning_rate": 7.373157023145493e-06,
+      "loss": 50.1993,
+      "step": 50530
+    },
+    {
+      "epoch": 0.20418799516800867,
+      "grad_norm": 738.6881713867188,
+      "learning_rate": 7.371915545294098e-06,
+      "loss": 47.1647,
+      "step": 50540
+    },
+    {
+      "epoch": 0.20422839643337629,
+      "grad_norm": 524.3673706054688,
+      "learning_rate": 7.37067387872027e-06,
+      "loss": 62.6645,
+      "step": 50550
+    },
+    {
+      "epoch": 0.20426879769874393,
+      "grad_norm": 527.126953125,
+      "learning_rate": 7.369432023522801e-06,
+      "loss": 43.1103,
+      "step": 50560
+    },
+    {
+      "epoch": 0.20430919896411157,
+      "grad_norm": 677.6494140625,
+      "learning_rate": 7.3681899798005006e-06,
+      "loss": 43.2993,
+      "step": 50570
+    },
+    {
+      "epoch": 0.20434960022947918,
+      "grad_norm": 464.6173095703125,
+      "learning_rate": 7.366947747652191e-06,
+      "loss": 33.6024,
+      "step": 50580
+    },
+    {
+      "epoch": 0.20439000149484682,
+      "grad_norm": 548.3930053710938,
+      "learning_rate": 7.365705327176713e-06,
+      "loss": 36.8351,
+      "step": 50590
+    },
+    {
+      "epoch": 0.20443040276021446,
+      "grad_norm": 853.4623413085938,
+      "learning_rate": 7.364462718472919e-06,
+      "loss": 44.7392,
+      "step": 50600
+    },
+    {
+      "epoch": 0.20447080402558207,
+      "grad_norm": 466.6961364746094,
+      "learning_rate": 7.363219921639677e-06,
+      "loss": 43.5595,
+      "step": 50610
+    },
+    {
+      "epoch": 0.2045112052909497,
+      "grad_norm": 765.1495971679688,
+      "learning_rate": 7.361976936775872e-06,
+      "loss": 47.2941,
+      "step": 50620
+    },
+    {
+      "epoch": 0.20455160655631735,
+      "grad_norm": 338.2070007324219,
+      "learning_rate": 7.360733763980404e-06,
+      "loss": 40.3301,
+      "step": 50630
+    },
+    {
+      "epoch": 0.20459200782168496,
+      "grad_norm": 421.01458740234375,
+      "learning_rate": 7.3594904033521815e-06,
+      "loss": 54.3122,
+      "step": 50640
+    },
+    {
+      "epoch": 0.2046324090870526,
+      "grad_norm": 319.7207946777344,
+      "learning_rate": 7.358246854990138e-06,
+      "loss": 50.9342,
+      "step": 50650
+    },
+    {
+      "epoch": 0.20467281035242024,
+      "grad_norm": 311.44586181640625,
+      "learning_rate": 7.357003118993215e-06,
+      "loss": 34.9208,
+      "step": 50660
+    },
+    {
+      "epoch": 0.20471321161778788,
+      "grad_norm": 540.4470825195312,
+      "learning_rate": 7.355759195460371e-06,
+      "loss": 41.9387,
+      "step": 50670
+    },
+    {
+      "epoch": 0.2047536128831555,
+      "grad_norm": 697.3502807617188,
+      "learning_rate": 7.354515084490579e-06,
+      "loss": 33.2285,
+      "step": 50680
+    },
+    {
+      "epoch": 0.20479401414852313,
+      "grad_norm": 971.903076171875,
+      "learning_rate": 7.353270786182828e-06,
+      "loss": 69.069,
+      "step": 50690
+    },
+    {
+      "epoch": 0.20483441541389077,
+      "grad_norm": 821.0972900390625,
+      "learning_rate": 7.352026300636121e-06,
+      "loss": 66.8979,
+      "step": 50700
+    },
+    {
+      "epoch": 0.2048748166792584,
+      "grad_norm": 438.63519287109375,
+      "learning_rate": 7.350781627949475e-06,
+      "loss": 43.2084,
+      "step": 50710
+    },
+    {
+      "epoch": 0.20491521794462603,
+      "grad_norm": 492.3321533203125,
+      "learning_rate": 7.3495367682219236e-06,
+      "loss": 56.6399,
+      "step": 50720
+    },
+    {
+      "epoch": 0.20495561920999367,
+      "grad_norm": 241.5388946533203,
+      "learning_rate": 7.348291721552514e-06,
+      "loss": 31.3594,
+      "step": 50730
+    },
+    {
+      "epoch": 0.20499602047536128,
+      "grad_norm": 635.99609375,
+      "learning_rate": 7.3470464880403105e-06,
+      "loss": 36.4675,
+      "step": 50740
+    },
+    {
+      "epoch": 0.20503642174072892,
+      "grad_norm": 885.6656494140625,
+      "learning_rate": 7.345801067784388e-06,
+      "loss": 44.9877,
+      "step": 50750
+    },
+    {
+      "epoch": 0.20507682300609656,
+      "grad_norm": 323.6302185058594,
+      "learning_rate": 7.34455546088384e-06,
+      "loss": 42.1206,
+      "step": 50760
+    },
+    {
+      "epoch": 0.20511722427146417,
+      "grad_norm": 579.3381958007812,
+      "learning_rate": 7.343309667437775e-06,
+      "loss": 37.2963,
+      "step": 50770
+    },
+    {
+      "epoch": 0.2051576255368318,
+      "grad_norm": 773.6015014648438,
+      "learning_rate": 7.3420636875453135e-06,
+      "loss": 65.9994,
+      "step": 50780
+    },
+    {
+      "epoch": 0.20519802680219945,
+      "grad_norm": 685.1111450195312,
+      "learning_rate": 7.340817521305595e-06,
+      "loss": 51.553,
+      "step": 50790
+    },
+    {
+      "epoch": 0.20523842806756706,
+      "grad_norm": 333.6510009765625,
+      "learning_rate": 7.3395711688177676e-06,
+      "loss": 39.827,
+      "step": 50800
+    },
+    {
+      "epoch": 0.2052788293329347,
+      "grad_norm": 543.8275756835938,
+      "learning_rate": 7.3383246301809985e-06,
+      "loss": 52.7167,
+      "step": 50810
+    },
+    {
+      "epoch": 0.20531923059830234,
+      "grad_norm": 568.8641967773438,
+      "learning_rate": 7.337077905494472e-06,
+      "loss": 31.9992,
+      "step": 50820
+    },
+    {
+      "epoch": 0.20535963186366998,
+      "grad_norm": 643.5866088867188,
+      "learning_rate": 7.335830994857382e-06,
+      "loss": 38.0165,
+      "step": 50830
+    },
+    {
+      "epoch": 0.2054000331290376,
+      "grad_norm": 956.9198608398438,
+      "learning_rate": 7.334583898368939e-06,
+      "loss": 57.5262,
+      "step": 50840
+    },
+    {
+      "epoch": 0.20544043439440524,
+      "grad_norm": 383.0577697753906,
+      "learning_rate": 7.333336616128369e-06,
+      "loss": 52.8704,
+      "step": 50850
+    },
+    {
+      "epoch": 0.20548083565977288,
+      "grad_norm": 554.4602661132812,
+      "learning_rate": 7.332089148234913e-06,
+      "loss": 61.3882,
+      "step": 50860
+    },
+    {
+      "epoch": 0.2055212369251405,
+      "grad_norm": 623.8341064453125,
+      "learning_rate": 7.330841494787828e-06,
+      "loss": 51.7853,
+      "step": 50870
+    },
+    {
+      "epoch": 0.20556163819050813,
+      "grad_norm": 1747.639404296875,
+      "learning_rate": 7.329593655886382e-06,
+      "loss": 58.8053,
+      "step": 50880
+    },
+    {
+      "epoch": 0.20560203945587577,
+      "grad_norm": 613.093505859375,
+      "learning_rate": 7.3283456316298595e-06,
+      "loss": 45.6349,
+      "step": 50890
+    },
+    {
+      "epoch": 0.20564244072124338,
+      "grad_norm": 570.6395263671875,
+      "learning_rate": 7.32709742211756e-06,
+      "loss": 44.1308,
+      "step": 50900
+    },
+    {
+      "epoch": 0.20568284198661102,
+      "grad_norm": 717.2446899414062,
+      "learning_rate": 7.325849027448799e-06,
+      "loss": 54.2538,
+      "step": 50910
+    },
+    {
+      "epoch": 0.20572324325197866,
+      "grad_norm": 314.36724853515625,
+      "learning_rate": 7.324600447722907e-06,
+      "loss": 48.2092,
+      "step": 50920
+    },
+    {
+      "epoch": 0.20576364451734627,
+      "grad_norm": 372.06494140625,
+      "learning_rate": 7.323351683039224e-06,
+      "loss": 96.2132,
+      "step": 50930
+    },
+    {
+      "epoch": 0.2058040457827139,
+      "grad_norm": 483.8934326171875,
+      "learning_rate": 7.32210273349711e-06,
+      "loss": 47.4042,
+      "step": 50940
+    },
+    {
+      "epoch": 0.20584444704808155,
+      "grad_norm": 411.3906555175781,
+      "learning_rate": 7.32085359919594e-06,
+      "loss": 50.984,
+      "step": 50950
+    },
+    {
+      "epoch": 0.20588484831344916,
+      "grad_norm": 892.8008422851562,
+      "learning_rate": 7.3196042802350995e-06,
+      "loss": 59.2704,
+      "step": 50960
+    },
+    {
+      "epoch": 0.2059252495788168,
+      "grad_norm": 605.1826171875,
+      "learning_rate": 7.3183547767139916e-06,
+      "loss": 42.0492,
+      "step": 50970
+    },
+    {
+      "epoch": 0.20596565084418444,
+      "grad_norm": 1213.0029296875,
+      "learning_rate": 7.317105088732035e-06,
+      "loss": 38.7694,
+      "step": 50980
+    },
+    {
+      "epoch": 0.20600605210955208,
+      "grad_norm": 384.1051330566406,
+      "learning_rate": 7.31585521638866e-06,
+      "loss": 41.8644,
+      "step": 50990
+    },
+    {
+      "epoch": 0.2060464533749197,
+      "grad_norm": 767.3082885742188,
+      "learning_rate": 7.314605159783313e-06,
+      "loss": 47.898,
+      "step": 51000
+    },
+    {
+      "epoch": 0.20608685464028734,
+      "grad_norm": 593.4794921875,
+      "learning_rate": 7.313354919015457e-06,
+      "loss": 50.6159,
+      "step": 51010
+    },
+    {
+      "epoch": 0.20612725590565498,
+      "grad_norm": 833.2760009765625,
+      "learning_rate": 7.312104494184566e-06,
+      "loss": 42.8232,
+      "step": 51020
+    },
+    {
+      "epoch": 0.2061676571710226,
+      "grad_norm": 488.1883544921875,
+      "learning_rate": 7.310853885390133e-06,
+      "loss": 44.1013,
+      "step": 51030
+    },
+    {
+      "epoch": 0.20620805843639023,
+      "grad_norm": 771.5651245117188,
+      "learning_rate": 7.309603092731661e-06,
+      "loss": 33.7422,
+      "step": 51040
+    },
+    {
+      "epoch": 0.20624845970175787,
+      "grad_norm": 1035.781982421875,
+      "learning_rate": 7.30835211630867e-06,
+      "loss": 58.5417,
+      "step": 51050
+    },
+    {
+      "epoch": 0.20628886096712548,
+      "grad_norm": 504.0133056640625,
+      "learning_rate": 7.3071009562206965e-06,
+      "loss": 50.3386,
+      "step": 51060
+    },
+    {
+      "epoch": 0.20632926223249312,
+      "grad_norm": 500.3634338378906,
+      "learning_rate": 7.305849612567287e-06,
+      "loss": 47.1454,
+      "step": 51070
+    },
+    {
+      "epoch": 0.20636966349786076,
+      "grad_norm": 419.568603515625,
+      "learning_rate": 7.304598085448007e-06,
+      "loss": 39.7767,
+      "step": 51080
+    },
+    {
+      "epoch": 0.20641006476322837,
+      "grad_norm": 741.7049560546875,
+      "learning_rate": 7.303346374962433e-06,
+      "loss": 85.6226,
+      "step": 51090
+    },
+    {
+      "epoch": 0.206450466028596,
+      "grad_norm": 868.6483154296875,
+      "learning_rate": 7.302094481210159e-06,
+      "loss": 71.4091,
+      "step": 51100
+    },
+    {
+      "epoch": 0.20649086729396365,
+      "grad_norm": 810.5807495117188,
+      "learning_rate": 7.300842404290792e-06,
+      "loss": 33.1312,
+      "step": 51110
+    },
+    {
+      "epoch": 0.20653126855933127,
+      "grad_norm": 339.3719177246094,
+      "learning_rate": 7.2995901443039554e-06,
+      "loss": 48.107,
+      "step": 51120
+    },
+    {
+      "epoch": 0.2065716698246989,
+      "grad_norm": 1150.4307861328125,
+      "learning_rate": 7.298337701349285e-06,
+      "loss": 53.5217,
+      "step": 51130
+    },
+    {
+      "epoch": 0.20661207109006655,
+      "grad_norm": 590.6275634765625,
+      "learning_rate": 7.29708507552643e-06,
+      "loss": 59.1666,
+      "step": 51140
+    },
+    {
+      "epoch": 0.20665247235543419,
+      "grad_norm": 721.5498046875,
+      "learning_rate": 7.295832266935059e-06,
+      "loss": 43.1562,
+      "step": 51150
+    },
+    {
+      "epoch": 0.2066928736208018,
+      "grad_norm": 483.1606750488281,
+      "learning_rate": 7.2945792756748505e-06,
+      "loss": 41.8515,
+      "step": 51160
+    },
+    {
+      "epoch": 0.20673327488616944,
+      "grad_norm": 531.7870483398438,
+      "learning_rate": 7.2933261018455005e-06,
+      "loss": 46.7922,
+      "step": 51170
+    },
+    {
+      "epoch": 0.20677367615153708,
+      "grad_norm": 763.1092529296875,
+      "learning_rate": 7.292072745546716e-06,
+      "loss": 63.6227,
+      "step": 51180
+    },
+    {
+      "epoch": 0.2068140774169047,
+      "grad_norm": 611.9227294921875,
+      "learning_rate": 7.290819206878223e-06,
+      "loss": 41.897,
+      "step": 51190
+    },
+    {
+      "epoch": 0.20685447868227233,
+      "grad_norm": 729.5850830078125,
+      "learning_rate": 7.289565485939759e-06,
+      "loss": 55.8601,
+      "step": 51200
+    },
+    {
+      "epoch": 0.20689487994763997,
+      "grad_norm": 318.1732482910156,
+      "learning_rate": 7.288311582831078e-06,
+      "loss": 50.0635,
+      "step": 51210
+    },
+    {
+      "epoch": 0.20693528121300758,
+      "grad_norm": 309.36285400390625,
+      "learning_rate": 7.2870574976519455e-06,
+      "loss": 47.5563,
+      "step": 51220
+    },
+    {
+      "epoch": 0.20697568247837522,
+      "grad_norm": 418.387451171875,
+      "learning_rate": 7.2858032305021455e-06,
+      "loss": 73.4059,
+      "step": 51230
+    },
+    {
+      "epoch": 0.20701608374374286,
+      "grad_norm": 468.94757080078125,
+      "learning_rate": 7.28454878148147e-06,
+      "loss": 48.932,
+      "step": 51240
+    },
+    {
+      "epoch": 0.20705648500911047,
+      "grad_norm": 493.0873718261719,
+      "learning_rate": 7.283294150689735e-06,
+      "loss": 46.617,
+      "step": 51250
+    },
+    {
+      "epoch": 0.20709688627447811,
+      "grad_norm": 480.55999755859375,
+      "learning_rate": 7.282039338226763e-06,
+      "loss": 40.8503,
+      "step": 51260
+    },
+    {
+      "epoch": 0.20713728753984575,
+      "grad_norm": 458.3509521484375,
+      "learning_rate": 7.280784344192393e-06,
+      "loss": 51.6308,
+      "step": 51270
+    },
+    {
+      "epoch": 0.20717768880521337,
+      "grad_norm": 463.3767395019531,
+      "learning_rate": 7.279529168686481e-06,
+      "loss": 46.1649,
+      "step": 51280
+    },
+    {
+      "epoch": 0.207218090070581,
+      "grad_norm": 568.6404418945312,
+      "learning_rate": 7.278273811808894e-06,
+      "loss": 33.446,
+      "step": 51290
+    },
+    {
+      "epoch": 0.20725849133594865,
+      "grad_norm": 465.77655029296875,
+      "learning_rate": 7.2770182736595164e-06,
+      "loss": 53.7961,
+      "step": 51300
+    },
+    {
+      "epoch": 0.2072988926013163,
+      "grad_norm": 362.6551513671875,
+      "learning_rate": 7.275762554338244e-06,
+      "loss": 37.3343,
+      "step": 51310
+    },
+    {
+      "epoch": 0.2073392938666839,
+      "grad_norm": 717.495849609375,
+      "learning_rate": 7.2745066539449905e-06,
+      "loss": 47.863,
+      "step": 51320
+    },
+    {
+      "epoch": 0.20737969513205154,
+      "grad_norm": 368.6878662109375,
+      "learning_rate": 7.27325057257968e-06,
+      "loss": 40.8304,
+      "step": 51330
+    },
+    {
+      "epoch": 0.20742009639741918,
+      "grad_norm": 560.428955078125,
+      "learning_rate": 7.271994310342254e-06,
+      "loss": 50.8628,
+      "step": 51340
+    },
+    {
+      "epoch": 0.2074604976627868,
+      "grad_norm": 770.505615234375,
+      "learning_rate": 7.270737867332669e-06,
+      "loss": 37.3534,
+      "step": 51350
+    },
+    {
+      "epoch": 0.20750089892815443,
+      "grad_norm": 920.7933959960938,
+      "learning_rate": 7.2694812436508934e-06,
+      "loss": 41.0482,
+      "step": 51360
+    },
+    {
+      "epoch": 0.20754130019352207,
+      "grad_norm": 788.7203369140625,
+      "learning_rate": 7.268224439396909e-06,
+      "loss": 48.9918,
+      "step": 51370
+    },
+    {
+      "epoch": 0.20758170145888968,
+      "grad_norm": 1051.6204833984375,
+      "learning_rate": 7.266967454670717e-06,
+      "loss": 42.9526,
+      "step": 51380
+    },
+    {
+      "epoch": 0.20762210272425732,
+      "grad_norm": 474.6462707519531,
+      "learning_rate": 7.265710289572328e-06,
+      "loss": 50.7116,
+      "step": 51390
+    },
+    {
+      "epoch": 0.20766250398962496,
+      "grad_norm": 414.03656005859375,
+      "learning_rate": 7.264452944201771e-06,
+      "loss": 33.1895,
+      "step": 51400
+    },
+    {
+      "epoch": 0.20770290525499258,
+      "grad_norm": 550.8810424804688,
+      "learning_rate": 7.263195418659083e-06,
+      "loss": 68.3899,
+      "step": 51410
+    },
+    {
+      "epoch": 0.20774330652036022,
+      "grad_norm": 681.5443725585938,
+      "learning_rate": 7.261937713044325e-06,
+      "loss": 54.9852,
+      "step": 51420
+    },
+    {
+      "epoch": 0.20778370778572786,
+      "grad_norm": 647.8605346679688,
+      "learning_rate": 7.260679827457562e-06,
+      "loss": 48.1843,
+      "step": 51430
+    },
+    {
+      "epoch": 0.20782410905109547,
+      "grad_norm": 504.14276123046875,
+      "learning_rate": 7.259421761998881e-06,
+      "loss": 33.8833,
+      "step": 51440
+    },
+    {
+      "epoch": 0.2078645103164631,
+      "grad_norm": 750.7991943359375,
+      "learning_rate": 7.2581635167683805e-06,
+      "loss": 38.5598,
+      "step": 51450
+    },
+    {
+      "epoch": 0.20790491158183075,
+      "grad_norm": 728.8818359375,
+      "learning_rate": 7.256905091866171e-06,
+      "loss": 40.8459,
+      "step": 51460
+    },
+    {
+      "epoch": 0.2079453128471984,
+      "grad_norm": 1682.726318359375,
+      "learning_rate": 7.255646487392382e-06,
+      "loss": 72.4861,
+      "step": 51470
+    },
+    {
+      "epoch": 0.207985714112566,
+      "grad_norm": 847.2395629882812,
+      "learning_rate": 7.254387703447154e-06,
+      "loss": 41.9428,
+      "step": 51480
+    },
+    {
+      "epoch": 0.20802611537793364,
+      "grad_norm": 511.3044738769531,
+      "learning_rate": 7.2531287401306435e-06,
+      "loss": 60.2857,
+      "step": 51490
+    },
+    {
+      "epoch": 0.20806651664330128,
+      "grad_norm": 354.39312744140625,
+      "learning_rate": 7.251869597543019e-06,
+      "loss": 34.2236,
+      "step": 51500
+    },
+    {
+      "epoch": 0.2081069179086689,
+      "grad_norm": 611.3307495117188,
+      "learning_rate": 7.250610275784464e-06,
+      "loss": 45.4532,
+      "step": 51510
+    },
+    {
+      "epoch": 0.20814731917403653,
+      "grad_norm": 648.39404296875,
+      "learning_rate": 7.2493507749551795e-06,
+      "loss": 49.6291,
+      "step": 51520
+    },
+    {
+      "epoch": 0.20818772043940417,
+      "grad_norm": 706.2927856445312,
+      "learning_rate": 7.248091095155378e-06,
+      "loss": 64.6364,
+      "step": 51530
+    },
+    {
+      "epoch": 0.20822812170477178,
+      "grad_norm": 389.3920593261719,
+      "learning_rate": 7.246831236485283e-06,
+      "loss": 25.9292,
+      "step": 51540
+    },
+    {
+      "epoch": 0.20826852297013942,
+      "grad_norm": 664.6102294921875,
+      "learning_rate": 7.245571199045139e-06,
+      "loss": 45.7861,
+      "step": 51550
+    },
+    {
+      "epoch": 0.20830892423550706,
+      "grad_norm": 500.1164855957031,
+      "learning_rate": 7.244310982935202e-06,
+      "loss": 30.2852,
+      "step": 51560
+    },
+    {
+      "epoch": 0.20834932550087468,
+      "grad_norm": 460.7940368652344,
+      "learning_rate": 7.243050588255738e-06,
+      "loss": 42.1839,
+      "step": 51570
+    },
+    {
+      "epoch": 0.20838972676624232,
+      "grad_norm": 703.8712768554688,
+      "learning_rate": 7.241790015107034e-06,
+      "loss": 65.8007,
+      "step": 51580
+    },
+    {
+      "epoch": 0.20843012803160996,
+      "grad_norm": 368.80810546875,
+      "learning_rate": 7.240529263589386e-06,
+      "loss": 39.5731,
+      "step": 51590
+    },
+    {
+      "epoch": 0.20847052929697757,
+      "grad_norm": 783.1382446289062,
+      "learning_rate": 7.239268333803109e-06,
+      "loss": 37.1502,
+      "step": 51600
+    },
+    {
+      "epoch": 0.2085109305623452,
+      "grad_norm": 409.4545593261719,
+      "learning_rate": 7.2380072258485265e-06,
+      "loss": 55.8675,
+      "step": 51610
+    },
+    {
+      "epoch": 0.20855133182771285,
+      "grad_norm": 508.4222106933594,
+      "learning_rate": 7.2367459398259795e-06,
+      "loss": 33.9861,
+      "step": 51620
+    },
+    {
+      "epoch": 0.20859173309308046,
+      "grad_norm": 473.3620300292969,
+      "learning_rate": 7.2354844758358234e-06,
+      "loss": 52.3211,
+      "step": 51630
+    },
+    {
+      "epoch": 0.2086321343584481,
+      "grad_norm": 787.4571533203125,
+      "learning_rate": 7.234222833978427e-06,
+      "loss": 29.9313,
+      "step": 51640
+    },
+    {
+      "epoch": 0.20867253562381574,
+      "grad_norm": 279.2911071777344,
+      "learning_rate": 7.232961014354175e-06,
+      "loss": 37.1078,
+      "step": 51650
+    },
+    {
+      "epoch": 0.20871293688918338,
+      "grad_norm": 603.6970825195312,
+      "learning_rate": 7.23169901706346e-06,
+      "loss": 43.0854,
+      "step": 51660
+    },
+    {
+      "epoch": 0.208753338154551,
+      "grad_norm": 899.671875,
+      "learning_rate": 7.2304368422067e-06,
+      "loss": 48.4667,
+      "step": 51670
+    },
+    {
+      "epoch": 0.20879373941991863,
+      "grad_norm": 795.8191528320312,
+      "learning_rate": 7.2291744898843145e-06,
+      "loss": 27.7021,
+      "step": 51680
+    },
+    {
+      "epoch": 0.20883414068528627,
+      "grad_norm": 940.0831909179688,
+      "learning_rate": 7.227911960196746e-06,
+      "loss": 59.3487,
+      "step": 51690
+    },
+    {
+      "epoch": 0.20887454195065389,
+      "grad_norm": 430.9837341308594,
+      "learning_rate": 7.226649253244448e-06,
+      "loss": 39.4674,
+      "step": 51700
+    },
+    {
+      "epoch": 0.20891494321602153,
+      "grad_norm": 714.0769653320312,
+      "learning_rate": 7.225386369127886e-06,
+      "loss": 39.4765,
+      "step": 51710
+    },
+    {
+      "epoch": 0.20895534448138917,
+      "grad_norm": 369.2260437011719,
+      "learning_rate": 7.224123307947545e-06,
+      "loss": 37.7946,
+      "step": 51720
+    },
+    {
+      "epoch": 0.20899574574675678,
+      "grad_norm": 393.2871398925781,
+      "learning_rate": 7.2228600698039205e-06,
+      "loss": 47.0149,
+      "step": 51730
+    },
+    {
+      "epoch": 0.20903614701212442,
+      "grad_norm": 284.41650390625,
+      "learning_rate": 7.221596654797522e-06,
+      "loss": 48.9927,
+      "step": 51740
+    },
+    {
+      "epoch": 0.20907654827749206,
+      "grad_norm": 800.0281982421875,
+      "learning_rate": 7.2203330630288714e-06,
+      "loss": 64.5519,
+      "step": 51750
+    },
+    {
+      "epoch": 0.20911694954285967,
+      "grad_norm": 445.57330322265625,
+      "learning_rate": 7.21906929459851e-06,
+      "loss": 42.594,
+      "step": 51760
+    },
+    {
+      "epoch": 0.2091573508082273,
+      "grad_norm": 471.7644348144531,
+      "learning_rate": 7.217805349606988e-06,
+      "loss": 32.7521,
+      "step": 51770
+    },
+    {
+      "epoch": 0.20919775207359495,
+      "grad_norm": 541.0687866210938,
+      "learning_rate": 7.216541228154875e-06,
+      "loss": 111.9155,
+      "step": 51780
+    },
+    {
+      "epoch": 0.20923815333896256,
+      "grad_norm": 513.6289672851562,
+      "learning_rate": 7.215276930342747e-06,
+      "loss": 29.2734,
+      "step": 51790
+    },
+    {
+      "epoch": 0.2092785546043302,
+      "grad_norm": 402.09423828125,
+      "learning_rate": 7.214012456271202e-06,
+      "loss": 37.2545,
+      "step": 51800
+    },
+    {
+      "epoch": 0.20931895586969784,
+      "grad_norm": 878.948974609375,
+      "learning_rate": 7.212747806040845e-06,
+      "loss": 49.578,
+      "step": 51810
+    },
+    {
+      "epoch": 0.20935935713506548,
+      "grad_norm": 462.6136474609375,
+      "learning_rate": 7.211482979752302e-06,
+      "loss": 37.1633,
+      "step": 51820
+    },
+    {
+      "epoch": 0.2093997584004331,
+      "grad_norm": 1256.9906005859375,
+      "learning_rate": 7.210217977506207e-06,
+      "loss": 46.0968,
+      "step": 51830
+    },
+    {
+      "epoch": 0.20944015966580073,
+      "grad_norm": 496.5869445800781,
+      "learning_rate": 7.208952799403211e-06,
+      "loss": 48.2958,
+      "step": 51840
+    },
+    {
+      "epoch": 0.20948056093116837,
+      "grad_norm": 795.1804809570312,
+      "learning_rate": 7.207687445543977e-06,
+      "loss": 52.5797,
+      "step": 51850
+    },
+    {
+      "epoch": 0.209520962196536,
+      "grad_norm": 694.8826904296875,
+      "learning_rate": 7.206421916029187e-06,
+      "loss": 47.4546,
+      "step": 51860
+    },
+    {
+      "epoch": 0.20956136346190363,
+      "grad_norm": 905.2157592773438,
+      "learning_rate": 7.205156210959529e-06,
+      "loss": 40.945,
+      "step": 51870
+    },
+    {
+      "epoch": 0.20960176472727127,
+      "grad_norm": 572.3497924804688,
+      "learning_rate": 7.203890330435715e-06,
+      "loss": 57.2266,
+      "step": 51880
+    },
+    {
+      "epoch": 0.20964216599263888,
+      "grad_norm": 538.5952758789062,
+      "learning_rate": 7.202624274558458e-06,
+      "loss": 37.1721,
+      "step": 51890
+    },
+    {
+      "epoch": 0.20968256725800652,
+      "grad_norm": 330.7435607910156,
+      "learning_rate": 7.201358043428499e-06,
+      "loss": 32.4022,
+      "step": 51900
+    },
+    {
+      "epoch": 0.20972296852337416,
+      "grad_norm": 847.5732421875,
+      "learning_rate": 7.200091637146582e-06,
+      "loss": 60.8524,
+      "step": 51910
+    },
+    {
+      "epoch": 0.20976336978874177,
+      "grad_norm": 599.7764892578125,
+      "learning_rate": 7.198825055813471e-06,
+      "loss": 68.1129,
+      "step": 51920
+    },
+    {
+      "epoch": 0.2098037710541094,
+      "grad_norm": 442.0505065917969,
+      "learning_rate": 7.197558299529941e-06,
+      "loss": 42.2968,
+      "step": 51930
+    },
+    {
+      "epoch": 0.20984417231947705,
+      "grad_norm": 387.84210205078125,
+      "learning_rate": 7.196291368396784e-06,
+      "loss": 43.1034,
+      "step": 51940
+    },
+    {
+      "epoch": 0.20988457358484466,
+      "grad_norm": 896.5294189453125,
+      "learning_rate": 7.1950242625148e-06,
+      "loss": 55.5236,
+      "step": 51950
+    },
+    {
+      "epoch": 0.2099249748502123,
+      "grad_norm": 604.0323486328125,
+      "learning_rate": 7.1937569819848115e-06,
+      "loss": 40.1479,
+      "step": 51960
+    },
+    {
+      "epoch": 0.20996537611557994,
+      "grad_norm": 284.2196350097656,
+      "learning_rate": 7.192489526907646e-06,
+      "loss": 32.0391,
+      "step": 51970
+    },
+    {
+      "epoch": 0.21000577738094758,
+      "grad_norm": 549.9973754882812,
+      "learning_rate": 7.191221897384153e-06,
+      "loss": 37.4856,
+      "step": 51980
+    },
+    {
+      "epoch": 0.2100461786463152,
+      "grad_norm": 819.0157470703125,
+      "learning_rate": 7.189954093515189e-06,
+      "loss": 45.887,
+      "step": 51990
+    },
+    {
+      "epoch": 0.21008657991168284,
+      "grad_norm": 814.2695922851562,
+      "learning_rate": 7.188686115401628e-06,
+      "loss": 42.8499,
+      "step": 52000
+    },
+    {
+      "epoch": 0.21012698117705048,
+      "grad_norm": 803.1535034179688,
+      "learning_rate": 7.187417963144358e-06,
+      "loss": 40.0624,
+      "step": 52010
+    },
+    {
+      "epoch": 0.2101673824424181,
+      "grad_norm": 545.9307861328125,
+      "learning_rate": 7.18614963684428e-06,
+      "loss": 46.7121,
+      "step": 52020
+    },
+    {
+      "epoch": 0.21020778370778573,
+      "grad_norm": 747.6693115234375,
+      "learning_rate": 7.184881136602309e-06,
+      "loss": 38.1845,
+      "step": 52030
+    },
+    {
+      "epoch": 0.21024818497315337,
+      "grad_norm": 607.844970703125,
+      "learning_rate": 7.183612462519371e-06,
+      "loss": 48.4949,
+      "step": 52040
+    },
+    {
+      "epoch": 0.21028858623852098,
+      "grad_norm": 536.132568359375,
+      "learning_rate": 7.182343614696412e-06,
+      "loss": 39.4975,
+      "step": 52050
+    },
+    {
+      "epoch": 0.21032898750388862,
+      "grad_norm": 684.9125366210938,
+      "learning_rate": 7.181074593234387e-06,
+      "loss": 40.7165,
+      "step": 52060
+    },
+    {
+      "epoch": 0.21036938876925626,
+      "grad_norm": 539.2525024414062,
+      "learning_rate": 7.179805398234266e-06,
+      "loss": 42.7935,
+      "step": 52070
+    },
+    {
+      "epoch": 0.21040979003462387,
+      "grad_norm": 431.33050537109375,
+      "learning_rate": 7.178536029797035e-06,
+      "loss": 41.8031,
+      "step": 52080
+    },
+    {
+      "epoch": 0.2104501912999915,
+      "grad_norm": 742.8863525390625,
+      "learning_rate": 7.177266488023688e-06,
+      "loss": 49.8048,
+      "step": 52090
+    },
+    {
+      "epoch": 0.21049059256535915,
+      "grad_norm": 1108.1619873046875,
+      "learning_rate": 7.17599677301524e-06,
+      "loss": 64.1411,
+      "step": 52100
+    },
+    {
+      "epoch": 0.21053099383072676,
+      "grad_norm": 691.4251708984375,
+      "learning_rate": 7.174726884872716e-06,
+      "loss": 35.8404,
+      "step": 52110
+    },
+    {
+      "epoch": 0.2105713950960944,
+      "grad_norm": 752.3907470703125,
+      "learning_rate": 7.173456823697154e-06,
+      "loss": 43.8726,
+      "step": 52120
+    },
+    {
+      "epoch": 0.21061179636146204,
+      "grad_norm": 594.6886596679688,
+      "learning_rate": 7.172186589589607e-06,
+      "loss": 89.6082,
+      "step": 52130
+    },
+    {
+      "epoch": 0.21065219762682968,
+      "grad_norm": 550.98193359375,
+      "learning_rate": 7.170916182651141e-06,
+      "loss": 31.0973,
+      "step": 52140
+    },
+    {
+      "epoch": 0.2106925988921973,
+      "grad_norm": 527.2732543945312,
+      "learning_rate": 7.1696456029828386e-06,
+      "loss": 55.1561,
+      "step": 52150
+    },
+    {
+      "epoch": 0.21073300015756494,
+      "grad_norm": 1399.20166015625,
+      "learning_rate": 7.168374850685794e-06,
+      "loss": 52.6242,
+      "step": 52160
+    },
+    {
+      "epoch": 0.21077340142293258,
+      "grad_norm": 689.70751953125,
+      "learning_rate": 7.167103925861113e-06,
+      "loss": 67.5702,
+      "step": 52170
+    },
+    {
+      "epoch": 0.2108138026883002,
+      "grad_norm": 3195.75244140625,
+      "learning_rate": 7.165832828609918e-06,
+      "loss": 45.5313,
+      "step": 52180
+    },
+    {
+      "epoch": 0.21085420395366783,
+      "grad_norm": 767.225830078125,
+      "learning_rate": 7.164561559033344e-06,
+      "loss": 67.4989,
+      "step": 52190
+    },
+    {
+      "epoch": 0.21089460521903547,
+      "grad_norm": 933.3297119140625,
+      "learning_rate": 7.163290117232542e-06,
+      "loss": 40.6139,
+      "step": 52200
+    },
+    {
+      "epoch": 0.21093500648440308,
+      "grad_norm": 416.0960388183594,
+      "learning_rate": 7.162018503308674e-06,
+      "loss": 61.1358,
+      "step": 52210
+    },
+    {
+      "epoch": 0.21097540774977072,
+      "grad_norm": 527.6151733398438,
+      "learning_rate": 7.1607467173629145e-06,
+      "loss": 49.4484,
+      "step": 52220
+    },
+    {
+      "epoch": 0.21101580901513836,
+      "grad_norm": 259.9767761230469,
+      "learning_rate": 7.1594747594964564e-06,
+      "loss": 36.1799,
+      "step": 52230
+    },
+    {
+      "epoch": 0.21105621028050597,
+      "grad_norm": 416.2400207519531,
+      "learning_rate": 7.1582026298105e-06,
+      "loss": 40.1228,
+      "step": 52240
+    },
+    {
+      "epoch": 0.2110966115458736,
+      "grad_norm": 469.9295349121094,
+      "learning_rate": 7.156930328406268e-06,
+      "loss": 43.5409,
+      "step": 52250
+    },
+    {
+      "epoch": 0.21113701281124125,
+      "grad_norm": 446.4311828613281,
+      "learning_rate": 7.1556578553849875e-06,
+      "loss": 88.1864,
+      "step": 52260
+    },
+    {
+      "epoch": 0.21117741407660887,
+      "grad_norm": 576.50732421875,
+      "learning_rate": 7.154385210847905e-06,
+      "loss": 43.5277,
+      "step": 52270
+    },
+    {
+      "epoch": 0.2112178153419765,
+      "grad_norm": 395.7337341308594,
+      "learning_rate": 7.153112394896279e-06,
+      "loss": 34.336,
+      "step": 52280
+    },
+    {
+      "epoch": 0.21125821660734415,
+      "grad_norm": 319.70098876953125,
+      "learning_rate": 7.15183940763138e-06,
+      "loss": 69.1773,
+      "step": 52290
+    },
+    {
+      "epoch": 0.21129861787271179,
+      "grad_norm": 509.9462585449219,
+      "learning_rate": 7.150566249154496e-06,
+      "loss": 45.0828,
+      "step": 52300
+    },
+    {
+      "epoch": 0.2113390191380794,
+      "grad_norm": 244.54661560058594,
+      "learning_rate": 7.149292919566924e-06,
+      "loss": 38.4245,
+      "step": 52310
+    },
+    {
+      "epoch": 0.21137942040344704,
+      "grad_norm": 701.5598754882812,
+      "learning_rate": 7.148019418969979e-06,
+      "loss": 56.945,
+      "step": 52320
+    },
+    {
+      "epoch": 0.21141982166881468,
+      "grad_norm": 378.5613098144531,
+      "learning_rate": 7.146745747464987e-06,
+      "loss": 30.5057,
+      "step": 52330
+    },
+    {
+      "epoch": 0.2114602229341823,
+      "grad_norm": 755.4013671875,
+      "learning_rate": 7.145471905153288e-06,
+      "loss": 53.356,
+      "step": 52340
+    },
+    {
+      "epoch": 0.21150062419954993,
+      "grad_norm": 254.3358612060547,
+      "learning_rate": 7.1441978921362365e-06,
+      "loss": 40.9182,
+      "step": 52350
+    },
+    {
+      "epoch": 0.21154102546491757,
+      "grad_norm": 688.7052001953125,
+      "learning_rate": 7.142923708515199e-06,
+      "loss": 68.2775,
+      "step": 52360
+    },
+    {
+      "epoch": 0.21158142673028518,
+      "grad_norm": 329.7192077636719,
+      "learning_rate": 7.141649354391556e-06,
+      "loss": 55.6541,
+      "step": 52370
+    },
+    {
+      "epoch": 0.21162182799565282,
+      "grad_norm": 591.1115112304688,
+      "learning_rate": 7.140374829866703e-06,
+      "loss": 76.5433,
+      "step": 52380
+    },
+    {
+      "epoch": 0.21166222926102046,
+      "grad_norm": 1178.5982666015625,
+      "learning_rate": 7.1391001350420486e-06,
+      "loss": 49.616,
+      "step": 52390
+    },
+    {
+      "epoch": 0.21170263052638807,
+      "grad_norm": 694.3931884765625,
+      "learning_rate": 7.137825270019012e-06,
+      "loss": 42.3436,
+      "step": 52400
+    },
+    {
+      "epoch": 0.21174303179175571,
+      "grad_norm": 1042.828857421875,
+      "learning_rate": 7.1365502348990315e-06,
+      "loss": 49.1559,
+      "step": 52410
+    },
+    {
+      "epoch": 0.21178343305712335,
+      "grad_norm": 440.44696044921875,
+      "learning_rate": 7.135275029783554e-06,
+      "loss": 43.2997,
+      "step": 52420
+    },
+    {
+      "epoch": 0.21182383432249097,
+      "grad_norm": 441.6261291503906,
+      "learning_rate": 7.133999654774041e-06,
+      "loss": 52.5607,
+      "step": 52430
+    },
+    {
+      "epoch": 0.2118642355878586,
+      "grad_norm": 643.361328125,
+      "learning_rate": 7.13272410997197e-06,
+      "loss": 60.0243,
+      "step": 52440
+    },
+    {
+      "epoch": 0.21190463685322625,
+      "grad_norm": 840.8086547851562,
+      "learning_rate": 7.13144839547883e-06,
+      "loss": 37.9704,
+      "step": 52450
+    },
+    {
+      "epoch": 0.2119450381185939,
+      "grad_norm": 772.6046752929688,
+      "learning_rate": 7.130172511396123e-06,
+      "loss": 50.9666,
+      "step": 52460
+    },
+    {
+      "epoch": 0.2119854393839615,
+      "grad_norm": 465.1317138671875,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 33.6535,
+      "step": 52470
+    },
+    {
+      "epoch": 0.21202584064932914,
+      "grad_norm": 420.5824890136719,
+      "learning_rate": 7.127620234868085e-06,
+      "loss": 46.2685,
+      "step": 52480
+    },
+    {
+      "epoch": 0.21206624191469678,
+      "grad_norm": 675.41845703125,
+      "learning_rate": 7.126343842625828e-06,
+      "loss": 86.1336,
+      "step": 52490
+    },
+    {
+      "epoch": 0.2121066431800644,
+      "grad_norm": 523.8388671875,
+      "learning_rate": 7.1250672812001505e-06,
+      "loss": 40.6279,
+      "step": 52500
+    },
+    {
+      "epoch": 0.21214704444543203,
+      "grad_norm": 675.9490966796875,
+      "learning_rate": 7.123790550692624e-06,
+      "loss": 53.4468,
+      "step": 52510
+    },
+    {
+      "epoch": 0.21218744571079967,
+      "grad_norm": 550.1663208007812,
+      "learning_rate": 7.1225136512048275e-06,
+      "loss": 49.8968,
+      "step": 52520
+    },
+    {
+      "epoch": 0.21222784697616728,
+      "grad_norm": 504.02569580078125,
+      "learning_rate": 7.1212365828383615e-06,
+      "loss": 36.7957,
+      "step": 52530
+    },
+    {
+      "epoch": 0.21226824824153492,
+      "grad_norm": 462.62005615234375,
+      "learning_rate": 7.119959345694835e-06,
+      "loss": 38.4813,
+      "step": 52540
+    },
+    {
+      "epoch": 0.21230864950690256,
+      "grad_norm": 591.6526489257812,
+      "learning_rate": 7.118681939875875e-06,
+      "loss": 40.4088,
+      "step": 52550
+    },
+    {
+      "epoch": 0.21234905077227018,
+      "grad_norm": 949.3772583007812,
+      "learning_rate": 7.117404365483116e-06,
+      "loss": 53.9539,
+      "step": 52560
+    },
+    {
+      "epoch": 0.21238945203763782,
+      "grad_norm": 612.3878173828125,
+      "learning_rate": 7.116126622618207e-06,
+      "loss": 37.0243,
+      "step": 52570
+    },
+    {
+      "epoch": 0.21242985330300546,
+      "grad_norm": 2274.1650390625,
+      "learning_rate": 7.114848711382816e-06,
+      "loss": 68.6226,
+      "step": 52580
+    },
+    {
+      "epoch": 0.21247025456837307,
+      "grad_norm": 200.64561462402344,
+      "learning_rate": 7.1135706318786195e-06,
+      "loss": 48.2132,
+      "step": 52590
+    },
+    {
+      "epoch": 0.2125106558337407,
+      "grad_norm": 579.8665161132812,
+      "learning_rate": 7.112292384207306e-06,
+      "loss": 53.6652,
+      "step": 52600
+    },
+    {
+      "epoch": 0.21255105709910835,
+      "grad_norm": 1212.4814453125,
+      "learning_rate": 7.111013968470581e-06,
+      "loss": 52.715,
+      "step": 52610
+    },
+    {
+      "epoch": 0.212591458364476,
+      "grad_norm": 550.4876098632812,
+      "learning_rate": 7.109735384770166e-06,
+      "loss": 42.4386,
+      "step": 52620
+    },
+    {
+      "epoch": 0.2126318596298436,
+      "grad_norm": 1110.4034423828125,
+      "learning_rate": 7.108456633207787e-06,
+      "loss": 60.6928,
+      "step": 52630
+    },
+    {
+      "epoch": 0.21267226089521124,
+      "grad_norm": 907.2188110351562,
+      "learning_rate": 7.10717771388519e-06,
+      "loss": 60.7178,
+      "step": 52640
+    },
+    {
+      "epoch": 0.21271266216057888,
+      "grad_norm": 127.77151489257812,
+      "learning_rate": 7.105898626904134e-06,
+      "loss": 47.8759,
+      "step": 52650
+    },
+    {
+      "epoch": 0.2127530634259465,
+      "grad_norm": 1381.4952392578125,
+      "learning_rate": 7.104619372366387e-06,
+      "loss": 51.4232,
+      "step": 52660
+    },
+    {
+      "epoch": 0.21279346469131413,
+      "grad_norm": 756.6724243164062,
+      "learning_rate": 7.103339950373737e-06,
+      "loss": 81.3154,
+      "step": 52670
+    },
+    {
+      "epoch": 0.21283386595668177,
+      "grad_norm": 514.89453125,
+      "learning_rate": 7.102060361027981e-06,
+      "loss": 71.0428,
+      "step": 52680
+    },
+    {
+      "epoch": 0.21287426722204938,
+      "grad_norm": 542.7091674804688,
+      "learning_rate": 7.100780604430928e-06,
+      "loss": 49.2722,
+      "step": 52690
+    },
+    {
+      "epoch": 0.21291466848741702,
+      "grad_norm": 815.9537963867188,
+      "learning_rate": 7.099500680684404e-06,
+      "loss": 46.4009,
+      "step": 52700
+    },
+    {
+      "epoch": 0.21295506975278466,
+      "grad_norm": 361.59857177734375,
+      "learning_rate": 7.0982205898902444e-06,
+      "loss": 40.411,
+      "step": 52710
+    },
+    {
+      "epoch": 0.21299547101815228,
+      "grad_norm": 439.58587646484375,
+      "learning_rate": 7.096940332150305e-06,
+      "loss": 36.6399,
+      "step": 52720
+    },
+    {
+      "epoch": 0.21303587228351992,
+      "grad_norm": 663.2642822265625,
+      "learning_rate": 7.095659907566446e-06,
+      "loss": 57.5411,
+      "step": 52730
+    },
+    {
+      "epoch": 0.21307627354888756,
+      "grad_norm": 1188.1744384765625,
+      "learning_rate": 7.094379316240545e-06,
+      "loss": 46.7822,
+      "step": 52740
+    },
+    {
+      "epoch": 0.21311667481425517,
+      "grad_norm": 667.1629638671875,
+      "learning_rate": 7.093098558274494e-06,
+      "loss": 56.3298,
+      "step": 52750
+    },
+    {
+      "epoch": 0.2131570760796228,
+      "grad_norm": 797.8674926757812,
+      "learning_rate": 7.091817633770197e-06,
+      "loss": 43.1346,
+      "step": 52760
+    },
+    {
+      "epoch": 0.21319747734499045,
+      "grad_norm": 355.7433776855469,
+      "learning_rate": 7.090536542829571e-06,
+      "loss": 49.5407,
+      "step": 52770
+    },
+    {
+      "epoch": 0.2132378786103581,
+      "grad_norm": 712.264892578125,
+      "learning_rate": 7.089255285554546e-06,
+      "loss": 49.0901,
+      "step": 52780
+    },
+    {
+      "epoch": 0.2132782798757257,
+      "grad_norm": 755.2996215820312,
+      "learning_rate": 7.087973862047067e-06,
+      "loss": 55.9708,
+      "step": 52790
+    },
+    {
+      "epoch": 0.21331868114109334,
+      "grad_norm": 283.7481689453125,
+      "learning_rate": 7.08669227240909e-06,
+      "loss": 40.528,
+      "step": 52800
+    },
+    {
+      "epoch": 0.21335908240646098,
+      "grad_norm": 784.1795654296875,
+      "learning_rate": 7.085410516742586e-06,
+      "loss": 31.2034,
+      "step": 52810
+    },
+    {
+      "epoch": 0.2133994836718286,
+      "grad_norm": 790.679443359375,
+      "learning_rate": 7.084128595149538e-06,
+      "loss": 46.8797,
+      "step": 52820
+    },
+    {
+      "epoch": 0.21343988493719623,
+      "grad_norm": 773.3251953125,
+      "learning_rate": 7.082846507731942e-06,
+      "loss": 53.9314,
+      "step": 52830
+    },
+    {
+      "epoch": 0.21348028620256387,
+      "grad_norm": 443.99810791015625,
+      "learning_rate": 7.081564254591809e-06,
+      "loss": 41.7004,
+      "step": 52840
+    },
+    {
+      "epoch": 0.21352068746793149,
+      "grad_norm": 634.8325805664062,
+      "learning_rate": 7.08028183583116e-06,
+      "loss": 37.2924,
+      "step": 52850
+    },
+    {
+      "epoch": 0.21356108873329913,
+      "grad_norm": 728.5630493164062,
+      "learning_rate": 7.078999251552034e-06,
+      "loss": 31.9136,
+      "step": 52860
+    },
+    {
+      "epoch": 0.21360148999866677,
+      "grad_norm": 731.4481811523438,
+      "learning_rate": 7.077716501856478e-06,
+      "loss": 33.3455,
+      "step": 52870
+    },
+    {
+      "epoch": 0.21364189126403438,
+      "grad_norm": 861.4498901367188,
+      "learning_rate": 7.076433586846555e-06,
+      "loss": 64.4692,
+      "step": 52880
+    },
+    {
+      "epoch": 0.21368229252940202,
+      "grad_norm": 600.2234497070312,
+      "learning_rate": 7.075150506624342e-06,
+      "loss": 48.3308,
+      "step": 52890
+    },
+    {
+      "epoch": 0.21372269379476966,
+      "grad_norm": 527.2857666015625,
+      "learning_rate": 7.073867261291926e-06,
+      "loss": 50.0515,
+      "step": 52900
+    },
+    {
+      "epoch": 0.21376309506013727,
+      "grad_norm": 594.2874755859375,
+      "learning_rate": 7.0725838509514115e-06,
+      "loss": 39.6426,
+      "step": 52910
+    },
+    {
+      "epoch": 0.2138034963255049,
+      "grad_norm": 1075.2667236328125,
+      "learning_rate": 7.07130027570491e-06,
+      "loss": 68.3749,
+      "step": 52920
+    },
+    {
+      "epoch": 0.21384389759087255,
+      "grad_norm": 585.2786865234375,
+      "learning_rate": 7.070016535654551e-06,
+      "loss": 36.8384,
+      "step": 52930
+    },
+    {
+      "epoch": 0.2138842988562402,
+      "grad_norm": 740.2340087890625,
+      "learning_rate": 7.068732630902479e-06,
+      "loss": 55.1437,
+      "step": 52940
+    },
+    {
+      "epoch": 0.2139247001216078,
+      "grad_norm": 682.5702514648438,
+      "learning_rate": 7.067448561550844e-06,
+      "loss": 43.2231,
+      "step": 52950
+    },
+    {
+      "epoch": 0.21396510138697544,
+      "grad_norm": 727.7985229492188,
+      "learning_rate": 7.066164327701815e-06,
+      "loss": 58.4932,
+      "step": 52960
+    },
+    {
+      "epoch": 0.21400550265234308,
+      "grad_norm": 408.12554931640625,
+      "learning_rate": 7.064879929457573e-06,
+      "loss": 52.6824,
+      "step": 52970
+    },
+    {
+      "epoch": 0.2140459039177107,
+      "grad_norm": 678.7532348632812,
+      "learning_rate": 7.063595366920314e-06,
+      "loss": 43.5814,
+      "step": 52980
+    },
+    {
+      "epoch": 0.21408630518307833,
+      "grad_norm": 554.4685668945312,
+      "learning_rate": 7.062310640192239e-06,
+      "loss": 54.5951,
+      "step": 52990
+    },
+    {
+      "epoch": 0.21412670644844597,
+      "grad_norm": 759.0743408203125,
+      "learning_rate": 7.061025749375572e-06,
+      "loss": 40.8169,
+      "step": 53000
+    },
+    {
+      "epoch": 0.2141671077138136,
+      "grad_norm": 751.3245239257812,
+      "learning_rate": 7.059740694572545e-06,
+      "loss": 55.642,
+      "step": 53010
+    },
+    {
+      "epoch": 0.21420750897918123,
+      "grad_norm": 823.6691284179688,
+      "learning_rate": 7.058455475885405e-06,
+      "loss": 35.4595,
+      "step": 53020
+    },
+    {
+      "epoch": 0.21424791024454887,
+      "grad_norm": 808.4596557617188,
+      "learning_rate": 7.05717009341641e-06,
+      "loss": 43.8247,
+      "step": 53030
+    },
+    {
+      "epoch": 0.21428831150991648,
+      "grad_norm": 435.7613220214844,
+      "learning_rate": 7.05588454726783e-06,
+      "loss": 39.8654,
+      "step": 53040
+    },
+    {
+      "epoch": 0.21432871277528412,
+      "grad_norm": 761.8973999023438,
+      "learning_rate": 7.054598837541951e-06,
+      "loss": 43.7041,
+      "step": 53050
+    },
+    {
+      "epoch": 0.21436911404065176,
+      "grad_norm": 1002.57177734375,
+      "learning_rate": 7.053312964341075e-06,
+      "loss": 56.8311,
+      "step": 53060
+    },
+    {
+      "epoch": 0.21440951530601937,
+      "grad_norm": 725.5249633789062,
+      "learning_rate": 7.052026927767508e-06,
+      "loss": 53.9829,
+      "step": 53070
+    },
+    {
+      "epoch": 0.214449916571387,
+      "grad_norm": 832.4341430664062,
+      "learning_rate": 7.050740727923576e-06,
+      "loss": 42.1309,
+      "step": 53080
+    },
+    {
+      "epoch": 0.21449031783675465,
+      "grad_norm": 2103.50830078125,
+      "learning_rate": 7.049454364911615e-06,
+      "loss": 51.145,
+      "step": 53090
+    },
+    {
+      "epoch": 0.2145307191021223,
+      "grad_norm": 579.0474853515625,
+      "learning_rate": 7.048167838833977e-06,
+      "loss": 29.026,
+      "step": 53100
+    },
+    {
+      "epoch": 0.2145711203674899,
+      "grad_norm": 646.6517333984375,
+      "learning_rate": 7.046881149793026e-06,
+      "loss": 67.8125,
+      "step": 53110
+    },
+    {
+      "epoch": 0.21461152163285754,
+      "grad_norm": 900.6437377929688,
+      "learning_rate": 7.045594297891133e-06,
+      "loss": 46.3305,
+      "step": 53120
+    },
+    {
+      "epoch": 0.21465192289822518,
+      "grad_norm": 954.8676147460938,
+      "learning_rate": 7.04430728323069e-06,
+      "loss": 38.0878,
+      "step": 53130
+    },
+    {
+      "epoch": 0.2146923241635928,
+      "grad_norm": 607.4523315429688,
+      "learning_rate": 7.043020105914098e-06,
+      "loss": 40.7413,
+      "step": 53140
+    },
+    {
+      "epoch": 0.21473272542896044,
+      "grad_norm": 743.0084838867188,
+      "learning_rate": 7.041732766043775e-06,
+      "loss": 59.0576,
+      "step": 53150
+    },
+    {
+      "epoch": 0.21477312669432808,
+      "grad_norm": 422.3906555175781,
+      "learning_rate": 7.040445263722145e-06,
+      "loss": 42.1732,
+      "step": 53160
+    },
+    {
+      "epoch": 0.2148135279596957,
+      "grad_norm": 568.6240234375,
+      "learning_rate": 7.039157599051648e-06,
+      "loss": 33.5946,
+      "step": 53170
+    },
+    {
+      "epoch": 0.21485392922506333,
+      "grad_norm": 944.6744995117188,
+      "learning_rate": 7.037869772134741e-06,
+      "loss": 36.3536,
+      "step": 53180
+    },
+    {
+      "epoch": 0.21489433049043097,
+      "grad_norm": 779.457275390625,
+      "learning_rate": 7.036581783073888e-06,
+      "loss": 44.802,
+      "step": 53190
+    },
+    {
+      "epoch": 0.21493473175579858,
+      "grad_norm": 993.3648071289062,
+      "learning_rate": 7.035293631971569e-06,
+      "loss": 73.5553,
+      "step": 53200
+    },
+    {
+      "epoch": 0.21497513302116622,
+      "grad_norm": 738.565185546875,
+      "learning_rate": 7.034005318930277e-06,
+      "loss": 41.3091,
+      "step": 53210
+    },
+    {
+      "epoch": 0.21501553428653386,
+      "grad_norm": 818.487060546875,
+      "learning_rate": 7.032716844052517e-06,
+      "loss": 51.1636,
+      "step": 53220
+    },
+    {
+      "epoch": 0.21505593555190147,
+      "grad_norm": 591.581787109375,
+      "learning_rate": 7.031428207440807e-06,
+      "loss": 61.1763,
+      "step": 53230
+    },
+    {
+      "epoch": 0.2150963368172691,
+      "grad_norm": 590.1148681640625,
+      "learning_rate": 7.030139409197676e-06,
+      "loss": 36.4519,
+      "step": 53240
+    },
+    {
+      "epoch": 0.21513673808263675,
+      "grad_norm": 357.1758728027344,
+      "learning_rate": 7.02885044942567e-06,
+      "loss": 46.619,
+      "step": 53250
+    },
+    {
+      "epoch": 0.2151771393480044,
+      "grad_norm": 982.3209228515625,
+      "learning_rate": 7.027561328227345e-06,
+      "loss": 42.6811,
+      "step": 53260
+    },
+    {
+      "epoch": 0.215217540613372,
+      "grad_norm": 764.6962890625,
+      "learning_rate": 7.02627204570527e-06,
+      "loss": 30.1889,
+      "step": 53270
+    },
+    {
+      "epoch": 0.21525794187873964,
+      "grad_norm": 556.0413208007812,
+      "learning_rate": 7.024982601962027e-06,
+      "loss": 38.2327,
+      "step": 53280
+    },
+    {
+      "epoch": 0.21529834314410728,
+      "grad_norm": 604.3740844726562,
+      "learning_rate": 7.023692997100213e-06,
+      "loss": 52.7227,
+      "step": 53290
+    },
+    {
+      "epoch": 0.2153387444094749,
+      "grad_norm": 735.7214965820312,
+      "learning_rate": 7.0224032312224345e-06,
+      "loss": 49.6408,
+      "step": 53300
+    },
+    {
+      "epoch": 0.21537914567484254,
+      "grad_norm": 4784.7998046875,
+      "learning_rate": 7.021113304431313e-06,
+      "loss": 105.0257,
+      "step": 53310
+    },
+    {
+      "epoch": 0.21541954694021018,
+      "grad_norm": 563.2162475585938,
+      "learning_rate": 7.01982321682948e-06,
+      "loss": 47.3066,
+      "step": 53320
+    },
+    {
+      "epoch": 0.2154599482055778,
+      "grad_norm": 506.74755859375,
+      "learning_rate": 7.018532968519584e-06,
+      "loss": 62.6589,
+      "step": 53330
+    },
+    {
+      "epoch": 0.21550034947094543,
+      "grad_norm": 249.8182830810547,
+      "learning_rate": 7.0172425596042846e-06,
+      "loss": 29.9312,
+      "step": 53340
+    },
+    {
+      "epoch": 0.21554075073631307,
+      "grad_norm": 1026.1318359375,
+      "learning_rate": 7.0159519901862515e-06,
+      "loss": 55.5077,
+      "step": 53350
+    },
+    {
+      "epoch": 0.21558115200168068,
+      "grad_norm": 447.23590087890625,
+      "learning_rate": 7.014661260368171e-06,
+      "loss": 38.1175,
+      "step": 53360
+    },
+    {
+      "epoch": 0.21562155326704832,
+      "grad_norm": 608.2694702148438,
+      "learning_rate": 7.01337037025274e-06,
+      "loss": 38.778,
+      "step": 53370
+    },
+    {
+      "epoch": 0.21566195453241596,
+      "grad_norm": 430.6813659667969,
+      "learning_rate": 7.012079319942668e-06,
+      "loss": 45.4868,
+      "step": 53380
+    },
+    {
+      "epoch": 0.21570235579778357,
+      "grad_norm": 899.20556640625,
+      "learning_rate": 7.01078810954068e-06,
+      "loss": 56.9402,
+      "step": 53390
+    },
+    {
+      "epoch": 0.2157427570631512,
+      "grad_norm": 434.4818115234375,
+      "learning_rate": 7.0094967391495095e-06,
+      "loss": 47.1917,
+      "step": 53400
+    },
+    {
+      "epoch": 0.21578315832851885,
+      "grad_norm": 891.284423828125,
+      "learning_rate": 7.008205208871906e-06,
+      "loss": 65.6954,
+      "step": 53410
+    },
+    {
+      "epoch": 0.2158235595938865,
+      "grad_norm": 715.593017578125,
+      "learning_rate": 7.00691351881063e-06,
+      "loss": 55.7541,
+      "step": 53420
+    },
+    {
+      "epoch": 0.2158639608592541,
+      "grad_norm": 615.5816650390625,
+      "learning_rate": 7.005621669068456e-06,
+      "loss": 56.0196,
+      "step": 53430
+    },
+    {
+      "epoch": 0.21590436212462175,
+      "grad_norm": 655.7932739257812,
+      "learning_rate": 7.004329659748172e-06,
+      "loss": 38.692,
+      "step": 53440
+    },
+    {
+      "epoch": 0.21594476338998939,
+      "grad_norm": 279.7879638671875,
+      "learning_rate": 7.003037490952574e-06,
+      "loss": 51.2097,
+      "step": 53450
+    },
+    {
+      "epoch": 0.215985164655357,
+      "grad_norm": 533.7430419921875,
+      "learning_rate": 7.0017451627844765e-06,
+      "loss": 42.7369,
+      "step": 53460
+    },
+    {
+      "epoch": 0.21602556592072464,
+      "grad_norm": 509.7603759765625,
+      "learning_rate": 7.0004526753467004e-06,
+      "loss": 38.7973,
+      "step": 53470
+    },
+    {
+      "epoch": 0.21606596718609228,
+      "grad_norm": 376.4740295410156,
+      "learning_rate": 6.999160028742089e-06,
+      "loss": 50.9646,
+      "step": 53480
+    },
+    {
+      "epoch": 0.2161063684514599,
+      "grad_norm": 787.3605346679688,
+      "learning_rate": 6.997867223073487e-06,
+      "loss": 51.2587,
+      "step": 53490
+    },
+    {
+      "epoch": 0.21614676971682753,
+      "grad_norm": 595.9751586914062,
+      "learning_rate": 6.996574258443761e-06,
+      "loss": 45.6339,
+      "step": 53500
+    },
+    {
+      "epoch": 0.21618717098219517,
+      "grad_norm": 446.9659729003906,
+      "learning_rate": 6.995281134955784e-06,
+      "loss": 44.5907,
+      "step": 53510
+    },
+    {
+      "epoch": 0.21622757224756278,
+      "grad_norm": 366.5422058105469,
+      "learning_rate": 6.993987852712442e-06,
+      "loss": 33.3201,
+      "step": 53520
+    },
+    {
+      "epoch": 0.21626797351293042,
+      "grad_norm": 417.02825927734375,
+      "learning_rate": 6.992694411816638e-06,
+      "loss": 61.968,
+      "step": 53530
+    },
+    {
+      "epoch": 0.21630837477829806,
+      "grad_norm": 453.3203430175781,
+      "learning_rate": 6.991400812371287e-06,
+      "loss": 38.1036,
+      "step": 53540
+    },
+    {
+      "epoch": 0.21634877604366567,
+      "grad_norm": 592.3311767578125,
+      "learning_rate": 6.990107054479313e-06,
+      "loss": 53.7675,
+      "step": 53550
+    },
+    {
+      "epoch": 0.21638917730903331,
+      "grad_norm": 1016.1787109375,
+      "learning_rate": 6.988813138243652e-06,
+      "loss": 52.4604,
+      "step": 53560
+    },
+    {
+      "epoch": 0.21642957857440095,
+      "grad_norm": 586.8818359375,
+      "learning_rate": 6.987519063767257e-06,
+      "loss": 39.3914,
+      "step": 53570
+    },
+    {
+      "epoch": 0.2164699798397686,
+      "grad_norm": 458.52301025390625,
+      "learning_rate": 6.986224831153092e-06,
+      "loss": 41.3697,
+      "step": 53580
+    },
+    {
+      "epoch": 0.2165103811051362,
+      "grad_norm": 536.421630859375,
+      "learning_rate": 6.984930440504134e-06,
+      "loss": 52.9655,
+      "step": 53590
+    },
+    {
+      "epoch": 0.21655078237050385,
+      "grad_norm": 463.3578186035156,
+      "learning_rate": 6.9836358919233695e-06,
+      "loss": 45.0425,
+      "step": 53600
+    },
+    {
+      "epoch": 0.2165911836358715,
+      "grad_norm": 930.7780151367188,
+      "learning_rate": 6.982341185513799e-06,
+      "loss": 54.5087,
+      "step": 53610
+    },
+    {
+      "epoch": 0.2166315849012391,
+      "grad_norm": 346.7239074707031,
+      "learning_rate": 6.981046321378441e-06,
+      "loss": 46.0492,
+      "step": 53620
+    },
+    {
+      "epoch": 0.21667198616660674,
+      "grad_norm": 660.0195922851562,
+      "learning_rate": 6.979751299620318e-06,
+      "loss": 109.7135,
+      "step": 53630
+    },
+    {
+      "epoch": 0.21671238743197438,
+      "grad_norm": 440.25726318359375,
+      "learning_rate": 6.978456120342469e-06,
+      "loss": 37.689,
+      "step": 53640
+    },
+    {
+      "epoch": 0.216752788697342,
+      "grad_norm": 614.2511596679688,
+      "learning_rate": 6.977160783647947e-06,
+      "loss": 45.4165,
+      "step": 53650
+    },
+    {
+      "epoch": 0.21679318996270963,
+      "grad_norm": 1187.5201416015625,
+      "learning_rate": 6.975865289639815e-06,
+      "loss": 42.2823,
+      "step": 53660
+    },
+    {
+      "epoch": 0.21683359122807727,
+      "grad_norm": 571.8782348632812,
+      "learning_rate": 6.974569638421151e-06,
+      "loss": 59.7416,
+      "step": 53670
+    },
+    {
+      "epoch": 0.21687399249344488,
+      "grad_norm": 830.4623413085938,
+      "learning_rate": 6.973273830095042e-06,
+      "loss": 47.285,
+      "step": 53680
+    },
+    {
+      "epoch": 0.21691439375881252,
+      "grad_norm": 614.6996459960938,
+      "learning_rate": 6.971977864764591e-06,
+      "loss": 38.1775,
+      "step": 53690
+    },
+    {
+      "epoch": 0.21695479502418016,
+      "grad_norm": 789.254150390625,
+      "learning_rate": 6.970681742532911e-06,
+      "loss": 66.3674,
+      "step": 53700
+    },
+    {
+      "epoch": 0.21699519628954778,
+      "grad_norm": 991.6810302734375,
+      "learning_rate": 6.969385463503129e-06,
+      "loss": 47.4346,
+      "step": 53710
+    },
+    {
+      "epoch": 0.21703559755491542,
+      "grad_norm": 14252.0068359375,
+      "learning_rate": 6.968089027778384e-06,
+      "loss": 99.8792,
+      "step": 53720
+    },
+    {
+      "epoch": 0.21707599882028306,
+      "grad_norm": 790.77001953125,
+      "learning_rate": 6.9667924354618275e-06,
+      "loss": 46.2285,
+      "step": 53730
+    },
+    {
+      "epoch": 0.2171164000856507,
+      "grad_norm": 666.9031372070312,
+      "learning_rate": 6.965495686656623e-06,
+      "loss": 46.2733,
+      "step": 53740
+    },
+    {
+      "epoch": 0.2171568013510183,
+      "grad_norm": 669.7155151367188,
+      "learning_rate": 6.964198781465948e-06,
+      "loss": 41.1374,
+      "step": 53750
+    },
+    {
+      "epoch": 0.21719720261638595,
+      "grad_norm": 406.15411376953125,
+      "learning_rate": 6.962901719992989e-06,
+      "loss": 35.3804,
+      "step": 53760
+    },
+    {
+      "epoch": 0.2172376038817536,
+      "grad_norm": 755.256103515625,
+      "learning_rate": 6.961604502340949e-06,
+      "loss": 46.1026,
+      "step": 53770
+    },
+    {
+      "epoch": 0.2172780051471212,
+      "grad_norm": 810.091064453125,
+      "learning_rate": 6.960307128613042e-06,
+      "loss": 35.6879,
+      "step": 53780
+    },
+    {
+      "epoch": 0.21731840641248884,
+      "grad_norm": 615.5449829101562,
+      "learning_rate": 6.959009598912493e-06,
+      "loss": 61.5976,
+      "step": 53790
+    },
+    {
+      "epoch": 0.21735880767785648,
+      "grad_norm": 769.8095703125,
+      "learning_rate": 6.957711913342541e-06,
+      "loss": 56.504,
+      "step": 53800
+    },
+    {
+      "epoch": 0.2173992089432241,
+      "grad_norm": 325.61114501953125,
+      "learning_rate": 6.956414072006437e-06,
+      "loss": 53.6313,
+      "step": 53810
+    },
+    {
+      "epoch": 0.21743961020859173,
+      "grad_norm": 642.9010009765625,
+      "learning_rate": 6.955116075007443e-06,
+      "loss": 34.7719,
+      "step": 53820
+    },
+    {
+      "epoch": 0.21748001147395937,
+      "grad_norm": 524.3311157226562,
+      "learning_rate": 6.953817922448837e-06,
+      "loss": 64.898,
+      "step": 53830
+    },
+    {
+      "epoch": 0.21752041273932698,
+      "grad_norm": 912.242919921875,
+      "learning_rate": 6.9525196144339055e-06,
+      "loss": 70.4545,
+      "step": 53840
+    },
+    {
+      "epoch": 0.21756081400469462,
+      "grad_norm": 668.8618774414062,
+      "learning_rate": 6.951221151065948e-06,
+      "loss": 63.9754,
+      "step": 53850
+    },
+    {
+      "epoch": 0.21760121527006226,
+      "grad_norm": 320.6151428222656,
+      "learning_rate": 6.949922532448279e-06,
+      "loss": 30.3866,
+      "step": 53860
+    },
+    {
+      "epoch": 0.21764161653542988,
+      "grad_norm": 510.21746826171875,
+      "learning_rate": 6.948623758684223e-06,
+      "loss": 27.041,
+      "step": 53870
+    },
+    {
+      "epoch": 0.21768201780079752,
+      "grad_norm": 875.3880004882812,
+      "learning_rate": 6.9473248298771176e-06,
+      "loss": 60.386,
+      "step": 53880
+    },
+    {
+      "epoch": 0.21772241906616516,
+      "grad_norm": 181.17555236816406,
+      "learning_rate": 6.946025746130312e-06,
+      "loss": 47.359,
+      "step": 53890
+    },
+    {
+      "epoch": 0.2177628203315328,
+      "grad_norm": 909.2413940429688,
+      "learning_rate": 6.944726507547169e-06,
+      "loss": 55.6297,
+      "step": 53900
+    },
+    {
+      "epoch": 0.2178032215969004,
+      "grad_norm": 518.627685546875,
+      "learning_rate": 6.943427114231064e-06,
+      "loss": 30.3371,
+      "step": 53910
+    },
+    {
+      "epoch": 0.21784362286226805,
+      "grad_norm": 435.1603088378906,
+      "learning_rate": 6.942127566285382e-06,
+      "loss": 37.2483,
+      "step": 53920
+    },
+    {
+      "epoch": 0.2178840241276357,
+      "grad_norm": 759.238037109375,
+      "learning_rate": 6.940827863813523e-06,
+      "loss": 42.3018,
+      "step": 53930
+    },
+    {
+      "epoch": 0.2179244253930033,
+      "grad_norm": 2238.64794921875,
+      "learning_rate": 6.9395280069188964e-06,
+      "loss": 74.1954,
+      "step": 53940
+    },
+    {
+      "epoch": 0.21796482665837094,
+      "grad_norm": 242.63772583007812,
+      "learning_rate": 6.9382279957049295e-06,
+      "loss": 36.7179,
+      "step": 53950
+    },
+    {
+      "epoch": 0.21800522792373858,
+      "grad_norm": 550.920166015625,
+      "learning_rate": 6.936927830275055e-06,
+      "loss": 74.0684,
+      "step": 53960
+    },
+    {
+      "epoch": 0.2180456291891062,
+      "grad_norm": 537.9491577148438,
+      "learning_rate": 6.935627510732724e-06,
+      "loss": 33.9198,
+      "step": 53970
+    },
+    {
+      "epoch": 0.21808603045447383,
+      "grad_norm": 1171.784423828125,
+      "learning_rate": 6.934327037181394e-06,
+      "loss": 49.3926,
+      "step": 53980
+    },
+    {
+      "epoch": 0.21812643171984147,
+      "grad_norm": 811.305419921875,
+      "learning_rate": 6.933026409724538e-06,
+      "loss": 40.5747,
+      "step": 53990
+    },
+    {
+      "epoch": 0.21816683298520909,
+      "grad_norm": 492.3915710449219,
+      "learning_rate": 6.931725628465643e-06,
+      "loss": 34.5997,
+      "step": 54000
+    },
+    {
+      "epoch": 0.21820723425057673,
+      "grad_norm": 575.7453002929688,
+      "learning_rate": 6.9304246935082065e-06,
+      "loss": 62.0087,
+      "step": 54010
+    },
+    {
+      "epoch": 0.21824763551594437,
+      "grad_norm": 700.669921875,
+      "learning_rate": 6.929123604955735e-06,
+      "loss": 47.6949,
+      "step": 54020
+    },
+    {
+      "epoch": 0.21828803678131198,
+      "grad_norm": 573.3309326171875,
+      "learning_rate": 6.927822362911753e-06,
+      "loss": 36.7347,
+      "step": 54030
+    },
+    {
+      "epoch": 0.21832843804667962,
+      "grad_norm": 756.4100341796875,
+      "learning_rate": 6.926520967479791e-06,
+      "loss": 46.6211,
+      "step": 54040
+    },
+    {
+      "epoch": 0.21836883931204726,
+      "grad_norm": 755.5197143554688,
+      "learning_rate": 6.9252194187634e-06,
+      "loss": 72.3489,
+      "step": 54050
+    },
+    {
+      "epoch": 0.2184092405774149,
+      "grad_norm": 590.9158935546875,
+      "learning_rate": 6.923917716866133e-06,
+      "loss": 46.8761,
+      "step": 54060
+    },
+    {
+      "epoch": 0.2184496418427825,
+      "grad_norm": 490.1414794921875,
+      "learning_rate": 6.922615861891564e-06,
+      "loss": 35.4525,
+      "step": 54070
+    },
+    {
+      "epoch": 0.21849004310815015,
+      "grad_norm": 672.7666015625,
+      "learning_rate": 6.921313853943275e-06,
+      "loss": 45.0449,
+      "step": 54080
+    },
+    {
+      "epoch": 0.2185304443735178,
+      "grad_norm": 775.6009521484375,
+      "learning_rate": 6.9200116931248575e-06,
+      "loss": 55.573,
+      "step": 54090
+    },
+    {
+      "epoch": 0.2185708456388854,
+      "grad_norm": 469.2464294433594,
+      "learning_rate": 6.918709379539924e-06,
+      "loss": 53.0544,
+      "step": 54100
+    },
+    {
+      "epoch": 0.21861124690425304,
+      "grad_norm": 762.6869506835938,
+      "learning_rate": 6.917406913292089e-06,
+      "loss": 71.2748,
+      "step": 54110
+    },
+    {
+      "epoch": 0.21865164816962068,
+      "grad_norm": 407.9251708984375,
+      "learning_rate": 6.916104294484988e-06,
+      "loss": 70.6682,
+      "step": 54120
+    },
+    {
+      "epoch": 0.2186920494349883,
+      "grad_norm": 1123.1390380859375,
+      "learning_rate": 6.91480152322226e-06,
+      "loss": 67.6451,
+      "step": 54130
+    },
+    {
+      "epoch": 0.21873245070035593,
+      "grad_norm": 662.4995727539062,
+      "learning_rate": 6.913498599607563e-06,
+      "loss": 49.0049,
+      "step": 54140
+    },
+    {
+      "epoch": 0.21877285196572357,
+      "grad_norm": 684.1514892578125,
+      "learning_rate": 6.9121955237445644e-06,
+      "loss": 46.24,
+      "step": 54150
+    },
+    {
+      "epoch": 0.2188132532310912,
+      "grad_norm": 591.7760009765625,
+      "learning_rate": 6.910892295736944e-06,
+      "loss": 35.831,
+      "step": 54160
+    },
+    {
+      "epoch": 0.21885365449645883,
+      "grad_norm": 423.4569396972656,
+      "learning_rate": 6.9095889156883934e-06,
+      "loss": 33.5558,
+      "step": 54170
+    },
+    {
+      "epoch": 0.21889405576182647,
+      "grad_norm": 420.7939758300781,
+      "learning_rate": 6.908285383702617e-06,
+      "loss": 48.0798,
+      "step": 54180
+    },
+    {
+      "epoch": 0.21893445702719408,
+      "grad_norm": 716.4933471679688,
+      "learning_rate": 6.906981699883329e-06,
+      "loss": 57.6833,
+      "step": 54190
+    },
+    {
+      "epoch": 0.21897485829256172,
+      "grad_norm": 564.2745971679688,
+      "learning_rate": 6.90567786433426e-06,
+      "loss": 37.345,
+      "step": 54200
+    },
+    {
+      "epoch": 0.21901525955792936,
+      "grad_norm": 570.277099609375,
+      "learning_rate": 6.904373877159149e-06,
+      "loss": 46.2849,
+      "step": 54210
+    },
+    {
+      "epoch": 0.219055660823297,
+      "grad_norm": 768.31591796875,
+      "learning_rate": 6.903069738461749e-06,
+      "loss": 45.064,
+      "step": 54220
+    },
+    {
+      "epoch": 0.2190960620886646,
+      "grad_norm": 1730.7943115234375,
+      "learning_rate": 6.901765448345823e-06,
+      "loss": 54.5519,
+      "step": 54230
+    },
+    {
+      "epoch": 0.21913646335403225,
+      "grad_norm": 450.1281433105469,
+      "learning_rate": 6.900461006915149e-06,
+      "loss": 37.9456,
+      "step": 54240
+    },
+    {
+      "epoch": 0.2191768646193999,
+      "grad_norm": 929.2560424804688,
+      "learning_rate": 6.899156414273514e-06,
+      "loss": 59.381,
+      "step": 54250
+    },
+    {
+      "epoch": 0.2192172658847675,
+      "grad_norm": 986.7847290039062,
+      "learning_rate": 6.89785167052472e-06,
+      "loss": 47.791,
+      "step": 54260
+    },
+    {
+      "epoch": 0.21925766715013514,
+      "grad_norm": 658.82568359375,
+      "learning_rate": 6.896546775772577e-06,
+      "loss": 43.5379,
+      "step": 54270
+    },
+    {
+      "epoch": 0.21929806841550278,
+      "grad_norm": 1264.8040771484375,
+      "learning_rate": 6.8952417301209114e-06,
+      "loss": 50.3495,
+      "step": 54280
+    },
+    {
+      "epoch": 0.2193384696808704,
+      "grad_norm": 505.1601867675781,
+      "learning_rate": 6.893936533673561e-06,
+      "loss": 45.3253,
+      "step": 54290
+    },
+    {
+      "epoch": 0.21937887094623804,
+      "grad_norm": 1080.662353515625,
+      "learning_rate": 6.892631186534371e-06,
+      "loss": 44.8194,
+      "step": 54300
+    },
+    {
+      "epoch": 0.21941927221160568,
+      "grad_norm": 44.70790481567383,
+      "learning_rate": 6.891325688807204e-06,
+      "loss": 34.7908,
+      "step": 54310
+    },
+    {
+      "epoch": 0.2194596734769733,
+      "grad_norm": 584.1826171875,
+      "learning_rate": 6.890020040595932e-06,
+      "loss": 34.2276,
+      "step": 54320
+    },
+    {
+      "epoch": 0.21950007474234093,
+      "grad_norm": 531.162353515625,
+      "learning_rate": 6.88871424200444e-06,
+      "loss": 40.3016,
+      "step": 54330
+    },
+    {
+      "epoch": 0.21954047600770857,
+      "grad_norm": 662.1864624023438,
+      "learning_rate": 6.887408293136621e-06,
+      "loss": 42.8401,
+      "step": 54340
+    },
+    {
+      "epoch": 0.21958087727307618,
+      "grad_norm": 794.0258178710938,
+      "learning_rate": 6.886102194096389e-06,
+      "loss": 48.1658,
+      "step": 54350
+    },
+    {
+      "epoch": 0.21962127853844382,
+      "grad_norm": 871.9588623046875,
+      "learning_rate": 6.884795944987661e-06,
+      "loss": 37.4216,
+      "step": 54360
+    },
+    {
+      "epoch": 0.21966167980381146,
+      "grad_norm": 313.5548400878906,
+      "learning_rate": 6.8834895459143694e-06,
+      "loss": 59.3293,
+      "step": 54370
+    },
+    {
+      "epoch": 0.2197020810691791,
+      "grad_norm": 611.6298828125,
+      "learning_rate": 6.882182996980457e-06,
+      "loss": 49.3419,
+      "step": 54380
+    },
+    {
+      "epoch": 0.2197424823345467,
+      "grad_norm": 558.3862915039062,
+      "learning_rate": 6.880876298289885e-06,
+      "loss": 27.1308,
+      "step": 54390
+    },
+    {
+      "epoch": 0.21978288359991435,
+      "grad_norm": 360.9166259765625,
+      "learning_rate": 6.879569449946617e-06,
+      "loss": 33.9725,
+      "step": 54400
+    },
+    {
+      "epoch": 0.219823284865282,
+      "grad_norm": 1138.4752197265625,
+      "learning_rate": 6.878262452054632e-06,
+      "loss": 69.0244,
+      "step": 54410
+    },
+    {
+      "epoch": 0.2198636861306496,
+      "grad_norm": 871.0791015625,
+      "learning_rate": 6.876955304717925e-06,
+      "loss": 59.2687,
+      "step": 54420
+    },
+    {
+      "epoch": 0.21990408739601724,
+      "grad_norm": 590.9871215820312,
+      "learning_rate": 6.875648008040499e-06,
+      "loss": 55.6743,
+      "step": 54430
+    },
+    {
+      "epoch": 0.21994448866138488,
+      "grad_norm": 423.3692626953125,
+      "learning_rate": 6.874340562126368e-06,
+      "loss": 52.967,
+      "step": 54440
+    },
+    {
+      "epoch": 0.2199848899267525,
+      "grad_norm": 711.0861206054688,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 37.9035,
+      "step": 54450
+    },
+    {
+      "epoch": 0.22002529119212014,
+      "grad_norm": 636.9541015625,
+      "learning_rate": 6.871725223004118e-06,
+      "loss": 44.2598,
+      "step": 54460
+    },
+    {
+      "epoch": 0.22006569245748778,
+      "grad_norm": 857.3299560546875,
+      "learning_rate": 6.870417330004086e-06,
+      "loss": 40.3004,
+      "step": 54470
+    },
+    {
+      "epoch": 0.2201060937228554,
+      "grad_norm": 810.0281372070312,
+      "learning_rate": 6.869109288183534e-06,
+      "loss": 42.4852,
+      "step": 54480
+    },
+    {
+      "epoch": 0.22014649498822303,
+      "grad_norm": 1230.8558349609375,
+      "learning_rate": 6.867801097646534e-06,
+      "loss": 56.5736,
+      "step": 54490
+    },
+    {
+      "epoch": 0.22018689625359067,
+      "grad_norm": 435.6243896484375,
+      "learning_rate": 6.866492758497171e-06,
+      "loss": 42.9811,
+      "step": 54500
+    },
+    {
+      "epoch": 0.22022729751895828,
+      "grad_norm": 570.0313110351562,
+      "learning_rate": 6.865184270839546e-06,
+      "loss": 25.4508,
+      "step": 54510
+    },
+    {
+      "epoch": 0.22026769878432592,
+      "grad_norm": 658.4278564453125,
+      "learning_rate": 6.863875634777767e-06,
+      "loss": 46.7509,
+      "step": 54520
+    },
+    {
+      "epoch": 0.22030810004969356,
+      "grad_norm": 608.4541625976562,
+      "learning_rate": 6.86256685041596e-06,
+      "loss": 39.0254,
+      "step": 54530
+    },
+    {
+      "epoch": 0.2203485013150612,
+      "grad_norm": 216.2620086669922,
+      "learning_rate": 6.861257917858257e-06,
+      "loss": 35.8314,
+      "step": 54540
+    },
+    {
+      "epoch": 0.2203889025804288,
+      "grad_norm": 772.1763916015625,
+      "learning_rate": 6.859948837208802e-06,
+      "loss": 47.3703,
+      "step": 54550
+    },
+    {
+      "epoch": 0.22042930384579645,
+      "grad_norm": 872.4230346679688,
+      "learning_rate": 6.8586396085717536e-06,
+      "loss": 46.5018,
+      "step": 54560
+    },
+    {
+      "epoch": 0.2204697051111641,
+      "grad_norm": 907.5761108398438,
+      "learning_rate": 6.8573302320512836e-06,
+      "loss": 34.8815,
+      "step": 54570
+    },
+    {
+      "epoch": 0.2205101063765317,
+      "grad_norm": 423.8174743652344,
+      "learning_rate": 6.85602070775157e-06,
+      "loss": 33.9442,
+      "step": 54580
+    },
+    {
+      "epoch": 0.22055050764189935,
+      "grad_norm": 582.1429443359375,
+      "learning_rate": 6.854711035776806e-06,
+      "loss": 65.9389,
+      "step": 54590
+    },
+    {
+      "epoch": 0.22059090890726699,
+      "grad_norm": 550.8157348632812,
+      "learning_rate": 6.853401216231198e-06,
+      "loss": 72.7223,
+      "step": 54600
+    },
+    {
+      "epoch": 0.2206313101726346,
+      "grad_norm": 623.2466430664062,
+      "learning_rate": 6.8520912492189605e-06,
+      "loss": 37.4688,
+      "step": 54610
+    },
+    {
+      "epoch": 0.22067171143800224,
+      "grad_norm": 3190.148681640625,
+      "learning_rate": 6.850781134844323e-06,
+      "loss": 57.139,
+      "step": 54620
+    },
+    {
+      "epoch": 0.22071211270336988,
+      "grad_norm": 926.8209838867188,
+      "learning_rate": 6.8494708732115235e-06,
+      "loss": 46.188,
+      "step": 54630
+    },
+    {
+      "epoch": 0.2207525139687375,
+      "grad_norm": 757.6647338867188,
+      "learning_rate": 6.8481604644248155e-06,
+      "loss": 51.8081,
+      "step": 54640
+    },
+    {
+      "epoch": 0.22079291523410513,
+      "grad_norm": 351.1914367675781,
+      "learning_rate": 6.846849908588461e-06,
+      "loss": 47.889,
+      "step": 54650
+    },
+    {
+      "epoch": 0.22083331649947277,
+      "grad_norm": 605.3834838867188,
+      "learning_rate": 6.845539205806735e-06,
+      "loss": 36.6061,
+      "step": 54660
+    },
+    {
+      "epoch": 0.22087371776484038,
+      "grad_norm": 424.5836486816406,
+      "learning_rate": 6.844228356183924e-06,
+      "loss": 58.7972,
+      "step": 54670
+    },
+    {
+      "epoch": 0.22091411903020802,
+      "grad_norm": 1018.7999877929688,
+      "learning_rate": 6.842917359824326e-06,
+      "loss": 46.3858,
+      "step": 54680
+    },
+    {
+      "epoch": 0.22095452029557566,
+      "grad_norm": 740.5365600585938,
+      "learning_rate": 6.841606216832253e-06,
+      "loss": 37.1616,
+      "step": 54690
+    },
+    {
+      "epoch": 0.22099492156094327,
+      "grad_norm": 974.8652954101562,
+      "learning_rate": 6.840294927312024e-06,
+      "loss": 38.5104,
+      "step": 54700
+    },
+    {
+      "epoch": 0.22103532282631091,
+      "grad_norm": 816.814453125,
+      "learning_rate": 6.838983491367974e-06,
+      "loss": 47.8214,
+      "step": 54710
+    },
+    {
+      "epoch": 0.22107572409167855,
+      "grad_norm": 525.71044921875,
+      "learning_rate": 6.837671909104447e-06,
+      "loss": 48.3607,
+      "step": 54720
+    },
+    {
+      "epoch": 0.2211161253570462,
+      "grad_norm": 3194.286376953125,
+      "learning_rate": 6.836360180625801e-06,
+      "loss": 74.2742,
+      "step": 54730
+    },
+    {
+      "epoch": 0.2211565266224138,
+      "grad_norm": 414.290283203125,
+      "learning_rate": 6.835048306036404e-06,
+      "loss": 35.6531,
+      "step": 54740
+    },
+    {
+      "epoch": 0.22119692788778145,
+      "grad_norm": 409.2863464355469,
+      "learning_rate": 6.833736285440632e-06,
+      "loss": 44.9234,
+      "step": 54750
+    },
+    {
+      "epoch": 0.2212373291531491,
+      "grad_norm": 958.4154663085938,
+      "learning_rate": 6.832424118942881e-06,
+      "loss": 69.0475,
+      "step": 54760
+    },
+    {
+      "epoch": 0.2212777304185167,
+      "grad_norm": 655.8173828125,
+      "learning_rate": 6.831111806647552e-06,
+      "loss": 61.3611,
+      "step": 54770
+    },
+    {
+      "epoch": 0.22131813168388434,
+      "grad_norm": 481.6536865234375,
+      "learning_rate": 6.829799348659061e-06,
+      "loss": 33.9049,
+      "step": 54780
+    },
+    {
+      "epoch": 0.22135853294925198,
+      "grad_norm": 809.3758544921875,
+      "learning_rate": 6.828486745081835e-06,
+      "loss": 53.7308,
+      "step": 54790
+    },
+    {
+      "epoch": 0.2213989342146196,
+      "grad_norm": 624.48974609375,
+      "learning_rate": 6.8271739960203065e-06,
+      "loss": 45.1322,
+      "step": 54800
+    },
+    {
+      "epoch": 0.22143933547998723,
+      "grad_norm": 911.08203125,
+      "learning_rate": 6.825861101578931e-06,
+      "loss": 57.4674,
+      "step": 54810
+    },
+    {
+      "epoch": 0.22147973674535487,
+      "grad_norm": 1179.452392578125,
+      "learning_rate": 6.824548061862166e-06,
+      "loss": 36.5426,
+      "step": 54820
+    },
+    {
+      "epoch": 0.22152013801072248,
+      "grad_norm": 564.9418334960938,
+      "learning_rate": 6.823234876974489e-06,
+      "loss": 42.9665,
+      "step": 54830
+    },
+    {
+      "epoch": 0.22156053927609012,
+      "grad_norm": 409.09228515625,
+      "learning_rate": 6.8219215470203756e-06,
+      "loss": 51.0909,
+      "step": 54840
+    },
+    {
+      "epoch": 0.22160094054145776,
+      "grad_norm": 764.6719360351562,
+      "learning_rate": 6.820608072104329e-06,
+      "loss": 74.6919,
+      "step": 54850
+    },
+    {
+      "epoch": 0.22164134180682538,
+      "grad_norm": 267.8291931152344,
+      "learning_rate": 6.819294452330853e-06,
+      "loss": 35.3009,
+      "step": 54860
+    },
+    {
+      "epoch": 0.22168174307219302,
+      "grad_norm": 774.4483032226562,
+      "learning_rate": 6.817980687804467e-06,
+      "loss": 53.1735,
+      "step": 54870
+    },
+    {
+      "epoch": 0.22172214433756066,
+      "grad_norm": 427.7686767578125,
+      "learning_rate": 6.8166667786297e-06,
+      "loss": 48.7483,
+      "step": 54880
+    },
+    {
+      "epoch": 0.2217625456029283,
+      "grad_norm": 412.5601501464844,
+      "learning_rate": 6.815352724911095e-06,
+      "loss": 58.3952,
+      "step": 54890
+    },
+    {
+      "epoch": 0.2218029468682959,
+      "grad_norm": 472.51348876953125,
+      "learning_rate": 6.814038526753205e-06,
+      "loss": 37.2425,
+      "step": 54900
+    },
+    {
+      "epoch": 0.22184334813366355,
+      "grad_norm": 715.309814453125,
+      "learning_rate": 6.812724184260596e-06,
+      "loss": 39.9846,
+      "step": 54910
+    },
+    {
+      "epoch": 0.2218837493990312,
+      "grad_norm": 476.5248107910156,
+      "learning_rate": 6.811409697537843e-06,
+      "loss": 44.4196,
+      "step": 54920
+    },
+    {
+      "epoch": 0.2219241506643988,
+      "grad_norm": 695.730712890625,
+      "learning_rate": 6.810095066689533e-06,
+      "loss": 63.0258,
+      "step": 54930
+    },
+    {
+      "epoch": 0.22196455192976644,
+      "grad_norm": 694.9153442382812,
+      "learning_rate": 6.808780291820264e-06,
+      "loss": 60.7545,
+      "step": 54940
+    },
+    {
+      "epoch": 0.22200495319513408,
+      "grad_norm": 619.1652221679688,
+      "learning_rate": 6.80746537303465e-06,
+      "loss": 32.6867,
+      "step": 54950
+    },
+    {
+      "epoch": 0.2220453544605017,
+      "grad_norm": 453.65350341796875,
+      "learning_rate": 6.806150310437312e-06,
+      "loss": 47.0478,
+      "step": 54960
+    },
+    {
+      "epoch": 0.22208575572586933,
+      "grad_norm": 825.5439453125,
+      "learning_rate": 6.804835104132883e-06,
+      "loss": 45.5884,
+      "step": 54970
+    },
+    {
+      "epoch": 0.22212615699123697,
+      "grad_norm": 599.8566284179688,
+      "learning_rate": 6.803519754226007e-06,
+      "loss": 54.5817,
+      "step": 54980
+    },
+    {
+      "epoch": 0.22216655825660458,
+      "grad_norm": 888.5885620117188,
+      "learning_rate": 6.80220426082134e-06,
+      "loss": 25.1306,
+      "step": 54990
+    },
+    {
+      "epoch": 0.22220695952197222,
+      "grad_norm": 1242.8448486328125,
+      "learning_rate": 6.800888624023552e-06,
+      "loss": 76.5537,
+      "step": 55000
+    },
+    {
+      "epoch": 0.22224736078733986,
+      "grad_norm": 503.8450622558594,
+      "learning_rate": 6.799572843937322e-06,
+      "loss": 42.9535,
+      "step": 55010
+    },
+    {
+      "epoch": 0.22228776205270748,
+      "grad_norm": 819.0281372070312,
+      "learning_rate": 6.79825692066734e-06,
+      "loss": 45.8547,
+      "step": 55020
+    },
+    {
+      "epoch": 0.22232816331807512,
+      "grad_norm": 668.2685546875,
+      "learning_rate": 6.796940854318306e-06,
+      "loss": 52.2774,
+      "step": 55030
+    },
+    {
+      "epoch": 0.22236856458344276,
+      "grad_norm": 479.3298645019531,
+      "learning_rate": 6.795624644994936e-06,
+      "loss": 46.8034,
+      "step": 55040
+    },
+    {
+      "epoch": 0.2224089658488104,
+      "grad_norm": 128.8311309814453,
+      "learning_rate": 6.794308292801954e-06,
+      "loss": 54.5953,
+      "step": 55050
+    },
+    {
+      "epoch": 0.222449367114178,
+      "grad_norm": 621.6553955078125,
+      "learning_rate": 6.792991797844095e-06,
+      "loss": 46.0012,
+      "step": 55060
+    },
+    {
+      "epoch": 0.22248976837954565,
+      "grad_norm": 336.5859069824219,
+      "learning_rate": 6.791675160226109e-06,
+      "loss": 31.5495,
+      "step": 55070
+    },
+    {
+      "epoch": 0.2225301696449133,
+      "grad_norm": 518.2781372070312,
+      "learning_rate": 6.790358380052752e-06,
+      "loss": 42.9662,
+      "step": 55080
+    },
+    {
+      "epoch": 0.2225705709102809,
+      "grad_norm": 368.1739807128906,
+      "learning_rate": 6.789041457428796e-06,
+      "loss": 53.6045,
+      "step": 55090
+    },
+    {
+      "epoch": 0.22261097217564854,
+      "grad_norm": 631.5914916992188,
+      "learning_rate": 6.7877243924590205e-06,
+      "loss": 62.7045,
+      "step": 55100
+    },
+    {
+      "epoch": 0.22265137344101618,
+      "grad_norm": 674.2762451171875,
+      "learning_rate": 6.7864071852482205e-06,
+      "loss": 49.4443,
+      "step": 55110
+    },
+    {
+      "epoch": 0.2226917747063838,
+      "grad_norm": 569.96142578125,
+      "learning_rate": 6.7850898359012e-06,
+      "loss": 50.1245,
+      "step": 55120
+    },
+    {
+      "epoch": 0.22273217597175143,
+      "grad_norm": 821.7766723632812,
+      "learning_rate": 6.7837723445227724e-06,
+      "loss": 46.7975,
+      "step": 55130
+    },
+    {
+      "epoch": 0.22277257723711907,
+      "grad_norm": 724.6019897460938,
+      "learning_rate": 6.782454711217767e-06,
+      "loss": 43.0379,
+      "step": 55140
+    },
+    {
+      "epoch": 0.22281297850248669,
+      "grad_norm": 363.89599609375,
+      "learning_rate": 6.78113693609102e-06,
+      "loss": 66.6985,
+      "step": 55150
+    },
+    {
+      "epoch": 0.22285337976785433,
+      "grad_norm": 766.9359741210938,
+      "learning_rate": 6.77981901924738e-06,
+      "loss": 60.0962,
+      "step": 55160
+    },
+    {
+      "epoch": 0.22289378103322197,
+      "grad_norm": 478.49761962890625,
+      "learning_rate": 6.7785009607917095e-06,
+      "loss": 42.5678,
+      "step": 55170
+    },
+    {
+      "epoch": 0.22293418229858958,
+      "grad_norm": 531.2022705078125,
+      "learning_rate": 6.777182760828881e-06,
+      "loss": 45.1689,
+      "step": 55180
+    },
+    {
+      "epoch": 0.22297458356395722,
+      "grad_norm": 576.1405639648438,
+      "learning_rate": 6.7758644194637755e-06,
+      "loss": 78.0099,
+      "step": 55190
+    },
+    {
+      "epoch": 0.22301498482932486,
+      "grad_norm": 385.7039489746094,
+      "learning_rate": 6.774545936801289e-06,
+      "loss": 36.7261,
+      "step": 55200
+    },
+    {
+      "epoch": 0.2230553860946925,
+      "grad_norm": 606.2195434570312,
+      "learning_rate": 6.773227312946327e-06,
+      "loss": 31.0199,
+      "step": 55210
+    },
+    {
+      "epoch": 0.2230957873600601,
+      "grad_norm": 485.9703063964844,
+      "learning_rate": 6.771908548003803e-06,
+      "loss": 40.6115,
+      "step": 55220
+    },
+    {
+      "epoch": 0.22313618862542775,
+      "grad_norm": 896.8988037109375,
+      "learning_rate": 6.77058964207865e-06,
+      "loss": 41.6147,
+      "step": 55230
+    },
+    {
+      "epoch": 0.2231765898907954,
+      "grad_norm": 245.3710479736328,
+      "learning_rate": 6.769270595275804e-06,
+      "loss": 38.9666,
+      "step": 55240
+    },
+    {
+      "epoch": 0.223216991156163,
+      "grad_norm": 1829.5657958984375,
+      "learning_rate": 6.767951407700217e-06,
+      "loss": 51.5566,
+      "step": 55250
+    },
+    {
+      "epoch": 0.22325739242153064,
+      "grad_norm": 606.4932861328125,
+      "learning_rate": 6.766632079456852e-06,
+      "loss": 48.8662,
+      "step": 55260
+    },
+    {
+      "epoch": 0.22329779368689828,
+      "grad_norm": 446.11346435546875,
+      "learning_rate": 6.765312610650677e-06,
+      "loss": 33.2803,
+      "step": 55270
+    },
+    {
+      "epoch": 0.2233381949522659,
+      "grad_norm": 476.4100036621094,
+      "learning_rate": 6.763993001386681e-06,
+      "loss": 60.5214,
+      "step": 55280
+    },
+    {
+      "epoch": 0.22337859621763353,
+      "grad_norm": 462.5067443847656,
+      "learning_rate": 6.762673251769858e-06,
+      "loss": 40.2899,
+      "step": 55290
+    },
+    {
+      "epoch": 0.22341899748300117,
+      "grad_norm": 504.2951965332031,
+      "learning_rate": 6.761353361905214e-06,
+      "loss": 52.4844,
+      "step": 55300
+    },
+    {
+      "epoch": 0.2234593987483688,
+      "grad_norm": 783.8707885742188,
+      "learning_rate": 6.7600333318977655e-06,
+      "loss": 45.4845,
+      "step": 55310
+    },
+    {
+      "epoch": 0.22349980001373643,
+      "grad_norm": 938.263427734375,
+      "learning_rate": 6.758713161852541e-06,
+      "loss": 53.5728,
+      "step": 55320
+    },
+    {
+      "epoch": 0.22354020127910407,
+      "grad_norm": 319.70440673828125,
+      "learning_rate": 6.757392851874584e-06,
+      "loss": 26.493,
+      "step": 55330
+    },
+    {
+      "epoch": 0.22358060254447168,
+      "grad_norm": 903.2843627929688,
+      "learning_rate": 6.756072402068943e-06,
+      "loss": 64.1053,
+      "step": 55340
+    },
+    {
+      "epoch": 0.22362100380983932,
+      "grad_norm": 1832.6444091796875,
+      "learning_rate": 6.75475181254068e-06,
+      "loss": 66.4452,
+      "step": 55350
+    },
+    {
+      "epoch": 0.22366140507520696,
+      "grad_norm": 469.994140625,
+      "learning_rate": 6.753431083394868e-06,
+      "loss": 66.304,
+      "step": 55360
+    },
+    {
+      "epoch": 0.2237018063405746,
+      "grad_norm": 573.45068359375,
+      "learning_rate": 6.75211021473659e-06,
+      "loss": 40.0917,
+      "step": 55370
+    },
+    {
+      "epoch": 0.2237422076059422,
+      "grad_norm": 661.8993530273438,
+      "learning_rate": 6.750789206670945e-06,
+      "loss": 42.8067,
+      "step": 55380
+    },
+    {
+      "epoch": 0.22378260887130985,
+      "grad_norm": 445.6514892578125,
+      "learning_rate": 6.749468059303039e-06,
+      "loss": 43.0675,
+      "step": 55390
+    },
+    {
+      "epoch": 0.2238230101366775,
+      "grad_norm": 566.9107055664062,
+      "learning_rate": 6.748146772737988e-06,
+      "loss": 39.0637,
+      "step": 55400
+    },
+    {
+      "epoch": 0.2238634114020451,
+      "grad_norm": 346.462646484375,
+      "learning_rate": 6.7468253470809205e-06,
+      "loss": 31.7741,
+      "step": 55410
+    },
+    {
+      "epoch": 0.22390381266741274,
+      "grad_norm": 325.1048889160156,
+      "learning_rate": 6.745503782436976e-06,
+      "loss": 52.311,
+      "step": 55420
+    },
+    {
+      "epoch": 0.22394421393278038,
+      "grad_norm": 282.2523498535156,
+      "learning_rate": 6.7441820789113085e-06,
+      "loss": 39.3717,
+      "step": 55430
+    },
+    {
+      "epoch": 0.223984615198148,
+      "grad_norm": 498.2767639160156,
+      "learning_rate": 6.7428602366090764e-06,
+      "loss": 43.0168,
+      "step": 55440
+    },
+    {
+      "epoch": 0.22402501646351564,
+      "grad_norm": 548.7891845703125,
+      "learning_rate": 6.741538255635454e-06,
+      "loss": 42.5905,
+      "step": 55450
+    },
+    {
+      "epoch": 0.22406541772888328,
+      "grad_norm": 208.3527069091797,
+      "learning_rate": 6.740216136095626e-06,
+      "loss": 31.6547,
+      "step": 55460
+    },
+    {
+      "epoch": 0.2241058189942509,
+      "grad_norm": 321.1961669921875,
+      "learning_rate": 6.738893878094786e-06,
+      "loss": 45.5326,
+      "step": 55470
+    },
+    {
+      "epoch": 0.22414622025961853,
+      "grad_norm": 581.4257202148438,
+      "learning_rate": 6.737571481738141e-06,
+      "loss": 48.1782,
+      "step": 55480
+    },
+    {
+      "epoch": 0.22418662152498617,
+      "grad_norm": 536.7522583007812,
+      "learning_rate": 6.736248947130907e-06,
+      "loss": 41.7099,
+      "step": 55490
+    },
+    {
+      "epoch": 0.22422702279035378,
+      "grad_norm": 870.3583374023438,
+      "learning_rate": 6.734926274378313e-06,
+      "loss": 44.808,
+      "step": 55500
+    },
+    {
+      "epoch": 0.22426742405572142,
+      "grad_norm": 1111.5572509765625,
+      "learning_rate": 6.733603463585598e-06,
+      "loss": 54.9259,
+      "step": 55510
+    },
+    {
+      "epoch": 0.22430782532108906,
+      "grad_norm": 477.1290588378906,
+      "learning_rate": 6.73228051485801e-06,
+      "loss": 50.8869,
+      "step": 55520
+    },
+    {
+      "epoch": 0.2243482265864567,
+      "grad_norm": 615.2061157226562,
+      "learning_rate": 6.7309574283008125e-06,
+      "loss": 56.2442,
+      "step": 55530
+    },
+    {
+      "epoch": 0.2243886278518243,
+      "grad_norm": 728.2523803710938,
+      "learning_rate": 6.729634204019277e-06,
+      "loss": 53.0035,
+      "step": 55540
+    },
+    {
+      "epoch": 0.22442902911719195,
+      "grad_norm": 743.2389526367188,
+      "learning_rate": 6.7283108421186835e-06,
+      "loss": 62.9054,
+      "step": 55550
+    },
+    {
+      "epoch": 0.2244694303825596,
+      "grad_norm": 616.263916015625,
+      "learning_rate": 6.726987342704331e-06,
+      "loss": 40.6529,
+      "step": 55560
+    },
+    {
+      "epoch": 0.2245098316479272,
+      "grad_norm": 774.1998901367188,
+      "learning_rate": 6.72566370588152e-06,
+      "loss": 46.4587,
+      "step": 55570
+    },
+    {
+      "epoch": 0.22455023291329484,
+      "grad_norm": 532.6005859375,
+      "learning_rate": 6.724339931755568e-06,
+      "loss": 25.9825,
+      "step": 55580
+    },
+    {
+      "epoch": 0.22459063417866248,
+      "grad_norm": 672.2699584960938,
+      "learning_rate": 6.7230160204318e-06,
+      "loss": 65.6449,
+      "step": 55590
+    },
+    {
+      "epoch": 0.2246310354440301,
+      "grad_norm": 675.6005249023438,
+      "learning_rate": 6.721691972015557e-06,
+      "loss": 28.1062,
+      "step": 55600
+    },
+    {
+      "epoch": 0.22467143670939774,
+      "grad_norm": 1068.322265625,
+      "learning_rate": 6.720367786612185e-06,
+      "loss": 53.7093,
+      "step": 55610
+    },
+    {
+      "epoch": 0.22471183797476538,
+      "grad_norm": 431.6038818359375,
+      "learning_rate": 6.719043464327043e-06,
+      "loss": 29.514,
+      "step": 55620
+    },
+    {
+      "epoch": 0.224752239240133,
+      "grad_norm": 425.9440002441406,
+      "learning_rate": 6.717719005265502e-06,
+      "loss": 66.4312,
+      "step": 55630
+    },
+    {
+      "epoch": 0.22479264050550063,
+      "grad_norm": 509.307373046875,
+      "learning_rate": 6.716394409532944e-06,
+      "loss": 46.8041,
+      "step": 55640
+    },
+    {
+      "epoch": 0.22483304177086827,
+      "grad_norm": 439.1161804199219,
+      "learning_rate": 6.715069677234758e-06,
+      "loss": 41.741,
+      "step": 55650
+    },
+    {
+      "epoch": 0.22487344303623588,
+      "grad_norm": 896.5545654296875,
+      "learning_rate": 6.713744808476349e-06,
+      "loss": 54.8341,
+      "step": 55660
+    },
+    {
+      "epoch": 0.22491384430160352,
+      "grad_norm": 802.5460205078125,
+      "learning_rate": 6.712419803363132e-06,
+      "loss": 47.4479,
+      "step": 55670
+    },
+    {
+      "epoch": 0.22495424556697116,
+      "grad_norm": 182.49765014648438,
+      "learning_rate": 6.711094662000529e-06,
+      "loss": 47.8153,
+      "step": 55680
+    },
+    {
+      "epoch": 0.2249946468323388,
+      "grad_norm": 677.7386474609375,
+      "learning_rate": 6.709769384493978e-06,
+      "loss": 74.3336,
+      "step": 55690
+    },
+    {
+      "epoch": 0.2250350480977064,
+      "grad_norm": 2979.65283203125,
+      "learning_rate": 6.708443970948923e-06,
+      "loss": 56.4762,
+      "step": 55700
+    },
+    {
+      "epoch": 0.22507544936307405,
+      "grad_norm": 885.204345703125,
+      "learning_rate": 6.707118421470822e-06,
+      "loss": 51.7887,
+      "step": 55710
+    },
+    {
+      "epoch": 0.2251158506284417,
+      "grad_norm": 1178.65185546875,
+      "learning_rate": 6.705792736165142e-06,
+      "loss": 49.2251,
+      "step": 55720
+    },
+    {
+      "epoch": 0.2251562518938093,
+      "grad_norm": 518.1280517578125,
+      "learning_rate": 6.7044669151373645e-06,
+      "loss": 61.4405,
+      "step": 55730
+    },
+    {
+      "epoch": 0.22519665315917695,
+      "grad_norm": 538.97265625,
+      "learning_rate": 6.7031409584929765e-06,
+      "loss": 51.3991,
+      "step": 55740
+    },
+    {
+      "epoch": 0.22523705442454459,
+      "grad_norm": 790.5933227539062,
+      "learning_rate": 6.701814866337477e-06,
+      "loss": 51.5159,
+      "step": 55750
+    },
+    {
+      "epoch": 0.2252774556899122,
+      "grad_norm": 669.2061157226562,
+      "learning_rate": 6.700488638776379e-06,
+      "loss": 49.5673,
+      "step": 55760
+    },
+    {
+      "epoch": 0.22531785695527984,
+      "grad_norm": 428.3297119140625,
+      "learning_rate": 6.699162275915208e-06,
+      "loss": 37.612,
+      "step": 55770
+    },
+    {
+      "epoch": 0.22535825822064748,
+      "grad_norm": 211.5701904296875,
+      "learning_rate": 6.6978357778594896e-06,
+      "loss": 61.2684,
+      "step": 55780
+    },
+    {
+      "epoch": 0.2253986594860151,
+      "grad_norm": 550.987548828125,
+      "learning_rate": 6.69650914471477e-06,
+      "loss": 42.0971,
+      "step": 55790
+    },
+    {
+      "epoch": 0.22543906075138273,
+      "grad_norm": 529.25439453125,
+      "learning_rate": 6.695182376586603e-06,
+      "loss": 39.4691,
+      "step": 55800
+    },
+    {
+      "epoch": 0.22547946201675037,
+      "grad_norm": 201.97569274902344,
+      "learning_rate": 6.6938554735805565e-06,
+      "loss": 30.0844,
+      "step": 55810
+    },
+    {
+      "epoch": 0.22551986328211798,
+      "grad_norm": 409.1141357421875,
+      "learning_rate": 6.6925284358022035e-06,
+      "loss": 49.2336,
+      "step": 55820
+    },
+    {
+      "epoch": 0.22556026454748562,
+      "grad_norm": 649.8071899414062,
+      "learning_rate": 6.69120126335713e-06,
+      "loss": 44.6068,
+      "step": 55830
+    },
+    {
+      "epoch": 0.22560066581285326,
+      "grad_norm": 803.1749267578125,
+      "learning_rate": 6.689873956350932e-06,
+      "loss": 35.6512,
+      "step": 55840
+    },
+    {
+      "epoch": 0.2256410670782209,
+      "grad_norm": 536.291015625,
+      "learning_rate": 6.688546514889221e-06,
+      "loss": 40.9719,
+      "step": 55850
+    },
+    {
+      "epoch": 0.22568146834358851,
+      "grad_norm": 1277.4453125,
+      "learning_rate": 6.687218939077613e-06,
+      "loss": 51.9526,
+      "step": 55860
+    },
+    {
+      "epoch": 0.22572186960895615,
+      "grad_norm": 839.084228515625,
+      "learning_rate": 6.685891229021736e-06,
+      "loss": 47.9897,
+      "step": 55870
+    },
+    {
+      "epoch": 0.2257622708743238,
+      "grad_norm": 375.78631591796875,
+      "learning_rate": 6.6845633848272315e-06,
+      "loss": 46.8047,
+      "step": 55880
+    },
+    {
+      "epoch": 0.2258026721396914,
+      "grad_norm": 804.9295654296875,
+      "learning_rate": 6.68323540659975e-06,
+      "loss": 50.1948,
+      "step": 55890
+    },
+    {
+      "epoch": 0.22584307340505905,
+      "grad_norm": 969.8311767578125,
+      "learning_rate": 6.681907294444952e-06,
+      "loss": 44.3812,
+      "step": 55900
+    },
+    {
+      "epoch": 0.2258834746704267,
+      "grad_norm": 413.19512939453125,
+      "learning_rate": 6.6805790484685094e-06,
+      "loss": 40.0582,
+      "step": 55910
+    },
+    {
+      "epoch": 0.2259238759357943,
+      "grad_norm": 1078.072021484375,
+      "learning_rate": 6.679250668776105e-06,
+      "loss": 39.8081,
+      "step": 55920
+    },
+    {
+      "epoch": 0.22596427720116194,
+      "grad_norm": 622.490234375,
+      "learning_rate": 6.677922155473432e-06,
+      "loss": 51.5911,
+      "step": 55930
+    },
+    {
+      "epoch": 0.22600467846652958,
+      "grad_norm": 369.42498779296875,
+      "learning_rate": 6.676593508666192e-06,
+      "loss": 41.1671,
+      "step": 55940
+    },
+    {
+      "epoch": 0.2260450797318972,
+      "grad_norm": 433.25811767578125,
+      "learning_rate": 6.675264728460103e-06,
+      "loss": 37.081,
+      "step": 55950
+    },
+    {
+      "epoch": 0.22608548099726483,
+      "grad_norm": 568.7027587890625,
+      "learning_rate": 6.673935814960887e-06,
+      "loss": 57.7069,
+      "step": 55960
+    },
+    {
+      "epoch": 0.22612588226263247,
+      "grad_norm": 603.1103515625,
+      "learning_rate": 6.672606768274281e-06,
+      "loss": 40.4617,
+      "step": 55970
+    },
+    {
+      "epoch": 0.22616628352800008,
+      "grad_norm": 1139.7769775390625,
+      "learning_rate": 6.67127758850603e-06,
+      "loss": 40.6864,
+      "step": 55980
+    },
+    {
+      "epoch": 0.22620668479336772,
+      "grad_norm": 589.6168212890625,
+      "learning_rate": 6.669948275761893e-06,
+      "loss": 34.2788,
+      "step": 55990
+    },
+    {
+      "epoch": 0.22624708605873536,
+      "grad_norm": 230.50637817382812,
+      "learning_rate": 6.668618830147634e-06,
+      "loss": 41.0629,
+      "step": 56000
+    },
+    {
+      "epoch": 0.226287487324103,
+      "grad_norm": 468.4389953613281,
+      "learning_rate": 6.667289251769033e-06,
+      "loss": 41.7365,
+      "step": 56010
+    },
+    {
+      "epoch": 0.22632788858947062,
+      "grad_norm": 454.0594787597656,
+      "learning_rate": 6.6659595407318775e-06,
+      "loss": 41.983,
+      "step": 56020
+    },
+    {
+      "epoch": 0.22636828985483826,
+      "grad_norm": 830.2134399414062,
+      "learning_rate": 6.664629697141969e-06,
+      "loss": 38.8899,
+      "step": 56030
+    },
+    {
+      "epoch": 0.2264086911202059,
+      "grad_norm": 1085.2601318359375,
+      "learning_rate": 6.663299721105113e-06,
+      "loss": 42.8715,
+      "step": 56040
+    },
+    {
+      "epoch": 0.2264490923855735,
+      "grad_norm": 590.5885620117188,
+      "learning_rate": 6.661969612727133e-06,
+      "loss": 48.8924,
+      "step": 56050
+    },
+    {
+      "epoch": 0.22648949365094115,
+      "grad_norm": 850.6708374023438,
+      "learning_rate": 6.660639372113858e-06,
+      "loss": 61.1752,
+      "step": 56060
+    },
+    {
+      "epoch": 0.2265298949163088,
+      "grad_norm": 309.1644287109375,
+      "learning_rate": 6.65930899937113e-06,
+      "loss": 52.0125,
+      "step": 56070
+    },
+    {
+      "epoch": 0.2265702961816764,
+      "grad_norm": 535.253662109375,
+      "learning_rate": 6.657978494604799e-06,
+      "loss": 58.5695,
+      "step": 56080
+    },
+    {
+      "epoch": 0.22661069744704404,
+      "grad_norm": 688.1857299804688,
+      "learning_rate": 6.656647857920728e-06,
+      "loss": 49.479,
+      "step": 56090
+    },
+    {
+      "epoch": 0.22665109871241168,
+      "grad_norm": 718.3746948242188,
+      "learning_rate": 6.655317089424791e-06,
+      "loss": 48.7625,
+      "step": 56100
+    },
+    {
+      "epoch": 0.2266914999777793,
+      "grad_norm": 609.1416625976562,
+      "learning_rate": 6.6539861892228695e-06,
+      "loss": 36.7984,
+      "step": 56110
+    },
+    {
+      "epoch": 0.22673190124314693,
+      "grad_norm": 245.2581329345703,
+      "learning_rate": 6.652655157420859e-06,
+      "loss": 29.8116,
+      "step": 56120
+    },
+    {
+      "epoch": 0.22677230250851457,
+      "grad_norm": 521.8516845703125,
+      "learning_rate": 6.651323994124661e-06,
+      "loss": 46.0755,
+      "step": 56130
+    },
+    {
+      "epoch": 0.22681270377388218,
+      "grad_norm": 714.7499389648438,
+      "learning_rate": 6.649992699440191e-06,
+      "loss": 42.7365,
+      "step": 56140
+    },
+    {
+      "epoch": 0.22685310503924982,
+      "grad_norm": 560.3165283203125,
+      "learning_rate": 6.648661273473375e-06,
+      "loss": 35.3493,
+      "step": 56150
+    },
+    {
+      "epoch": 0.22689350630461746,
+      "grad_norm": 602.651611328125,
+      "learning_rate": 6.6473297163301485e-06,
+      "loss": 50.3261,
+      "step": 56160
+    },
+    {
+      "epoch": 0.2269339075699851,
+      "grad_norm": 1904.8004150390625,
+      "learning_rate": 6.645998028116455e-06,
+      "loss": 57.5763,
+      "step": 56170
+    },
+    {
+      "epoch": 0.22697430883535272,
+      "grad_norm": 1206.6541748046875,
+      "learning_rate": 6.6446662089382545e-06,
+      "loss": 41.4296,
+      "step": 56180
+    },
+    {
+      "epoch": 0.22701471010072036,
+      "grad_norm": 435.66119384765625,
+      "learning_rate": 6.643334258901511e-06,
+      "loss": 39.747,
+      "step": 56190
+    },
+    {
+      "epoch": 0.227055111366088,
+      "grad_norm": 762.4393310546875,
+      "learning_rate": 6.642002178112202e-06,
+      "loss": 57.2834,
+      "step": 56200
+    },
+    {
+      "epoch": 0.2270955126314556,
+      "grad_norm": 483.8197937011719,
+      "learning_rate": 6.640669966676316e-06,
+      "loss": 31.7781,
+      "step": 56210
+    },
+    {
+      "epoch": 0.22713591389682325,
+      "grad_norm": 1017.3893432617188,
+      "learning_rate": 6.6393376246998485e-06,
+      "loss": 52.463,
+      "step": 56220
+    },
+    {
+      "epoch": 0.2271763151621909,
+      "grad_norm": 728.10693359375,
+      "learning_rate": 6.638005152288811e-06,
+      "loss": 40.0086,
+      "step": 56230
+    },
+    {
+      "epoch": 0.2272167164275585,
+      "grad_norm": 703.792724609375,
+      "learning_rate": 6.636672549549221e-06,
+      "loss": 37.7492,
+      "step": 56240
+    },
+    {
+      "epoch": 0.22725711769292614,
+      "grad_norm": 445.5412292480469,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 40.1628,
+      "step": 56250
+    },
+    {
+      "epoch": 0.22729751895829378,
+      "grad_norm": 941.9268798828125,
+      "learning_rate": 6.634006953508512e-06,
+      "loss": 49.6389,
+      "step": 56260
+    },
+    {
+      "epoch": 0.2273379202236614,
+      "grad_norm": 877.1407470703125,
+      "learning_rate": 6.63267396041948e-06,
+      "loss": 44.4499,
+      "step": 56270
+    },
+    {
+      "epoch": 0.22737832148902903,
+      "grad_norm": 1341.0885009765625,
+      "learning_rate": 6.631340837426075e-06,
+      "loss": 47.847,
+      "step": 56280
+    },
+    {
+      "epoch": 0.22741872275439667,
+      "grad_norm": 495.6767578125,
+      "learning_rate": 6.630007584634366e-06,
+      "loss": 37.3601,
+      "step": 56290
+    },
+    {
+      "epoch": 0.22745912401976429,
+      "grad_norm": 737.677978515625,
+      "learning_rate": 6.628674202150434e-06,
+      "loss": 40.301,
+      "step": 56300
+    },
+    {
+      "epoch": 0.22749952528513193,
+      "grad_norm": 879.7401733398438,
+      "learning_rate": 6.627340690080371e-06,
+      "loss": 69.7933,
+      "step": 56310
+    },
+    {
+      "epoch": 0.22753992655049957,
+      "grad_norm": 583.7142944335938,
+      "learning_rate": 6.626007048530276e-06,
+      "loss": 39.7344,
+      "step": 56320
+    },
+    {
+      "epoch": 0.2275803278158672,
+      "grad_norm": 588.3863525390625,
+      "learning_rate": 6.624673277606264e-06,
+      "loss": 71.7574,
+      "step": 56330
+    },
+    {
+      "epoch": 0.22762072908123482,
+      "grad_norm": 739.7526245117188,
+      "learning_rate": 6.623339377414456e-06,
+      "loss": 31.9641,
+      "step": 56340
+    },
+    {
+      "epoch": 0.22766113034660246,
+      "grad_norm": 562.2548828125,
+      "learning_rate": 6.622005348060983e-06,
+      "loss": 31.7818,
+      "step": 56350
+    },
+    {
+      "epoch": 0.2277015316119701,
+      "grad_norm": 488.791748046875,
+      "learning_rate": 6.620671189651988e-06,
+      "loss": 41.5536,
+      "step": 56360
+    },
+    {
+      "epoch": 0.2277419328773377,
+      "grad_norm": 1060.41796875,
+      "learning_rate": 6.6193369022936245e-06,
+      "loss": 53.0142,
+      "step": 56370
+    },
+    {
+      "epoch": 0.22778233414270535,
+      "grad_norm": 1061.703369140625,
+      "learning_rate": 6.618002486092056e-06,
+      "loss": 36.131,
+      "step": 56380
+    },
+    {
+      "epoch": 0.227822735408073,
+      "grad_norm": 552.3648681640625,
+      "learning_rate": 6.616667941153456e-06,
+      "loss": 43.893,
+      "step": 56390
+    },
+    {
+      "epoch": 0.2278631366734406,
+      "grad_norm": 816.7445068359375,
+      "learning_rate": 6.615333267584007e-06,
+      "loss": 66.4245,
+      "step": 56400
+    },
+    {
+      "epoch": 0.22790353793880824,
+      "grad_norm": 527.3714599609375,
+      "learning_rate": 6.613998465489902e-06,
+      "loss": 42.1384,
+      "step": 56410
+    },
+    {
+      "epoch": 0.22794393920417588,
+      "grad_norm": 400.6829833984375,
+      "learning_rate": 6.612663534977347e-06,
+      "loss": 53.9582,
+      "step": 56420
+    },
+    {
+      "epoch": 0.2279843404695435,
+      "grad_norm": 1087.0968017578125,
+      "learning_rate": 6.611328476152557e-06,
+      "loss": 50.3561,
+      "step": 56430
+    },
+    {
+      "epoch": 0.22802474173491113,
+      "grad_norm": 498.5974426269531,
+      "learning_rate": 6.609993289121753e-06,
+      "loss": 54.2461,
+      "step": 56440
+    },
+    {
+      "epoch": 0.22806514300027877,
+      "grad_norm": 467.4459228515625,
+      "learning_rate": 6.608657973991172e-06,
+      "loss": 37.9477,
+      "step": 56450
+    },
+    {
+      "epoch": 0.2281055442656464,
+      "grad_norm": 851.5455322265625,
+      "learning_rate": 6.607322530867061e-06,
+      "loss": 49.9047,
+      "step": 56460
+    },
+    {
+      "epoch": 0.22814594553101403,
+      "grad_norm": 640.7950439453125,
+      "learning_rate": 6.605986959855672e-06,
+      "loss": 51.9378,
+      "step": 56470
+    },
+    {
+      "epoch": 0.22818634679638167,
+      "grad_norm": 726.4421997070312,
+      "learning_rate": 6.60465126106327e-06,
+      "loss": 43.164,
+      "step": 56480
+    },
+    {
+      "epoch": 0.2282267480617493,
+      "grad_norm": 347.83416748046875,
+      "learning_rate": 6.6033154345961314e-06,
+      "loss": 79.5957,
+      "step": 56490
+    },
+    {
+      "epoch": 0.22826714932711692,
+      "grad_norm": 665.018798828125,
+      "learning_rate": 6.601979480560543e-06,
+      "loss": 80.1648,
+      "step": 56500
+    },
+    {
+      "epoch": 0.22830755059248456,
+      "grad_norm": 478.4368896484375,
+      "learning_rate": 6.6006433990627985e-06,
+      "loss": 40.2869,
+      "step": 56510
+    },
+    {
+      "epoch": 0.2283479518578522,
+      "grad_norm": 442.74713134765625,
+      "learning_rate": 6.599307190209206e-06,
+      "loss": 47.7656,
+      "step": 56520
+    },
+    {
+      "epoch": 0.2283883531232198,
+      "grad_norm": 1278.633544921875,
+      "learning_rate": 6.5979708541060796e-06,
+      "loss": 51.5611,
+      "step": 56530
+    },
+    {
+      "epoch": 0.22842875438858745,
+      "grad_norm": 477.2268981933594,
+      "learning_rate": 6.596634390859745e-06,
+      "loss": 54.8893,
+      "step": 56540
+    },
+    {
+      "epoch": 0.2284691556539551,
+      "grad_norm": 531.1737060546875,
+      "learning_rate": 6.59529780057654e-06,
+      "loss": 32.3914,
+      "step": 56550
+    },
+    {
+      "epoch": 0.2285095569193227,
+      "grad_norm": 690.1337890625,
+      "learning_rate": 6.593961083362811e-06,
+      "loss": 44.5659,
+      "step": 56560
+    },
+    {
+      "epoch": 0.22854995818469034,
+      "grad_norm": 358.41119384765625,
+      "learning_rate": 6.592624239324914e-06,
+      "loss": 42.6093,
+      "step": 56570
+    },
+    {
+      "epoch": 0.22859035945005798,
+      "grad_norm": 794.7353515625,
+      "learning_rate": 6.591287268569215e-06,
+      "loss": 41.1062,
+      "step": 56580
+    },
+    {
+      "epoch": 0.2286307607154256,
+      "grad_norm": 623.95361328125,
+      "learning_rate": 6.589950171202092e-06,
+      "loss": 49.7269,
+      "step": 56590
+    },
+    {
+      "epoch": 0.22867116198079324,
+      "grad_norm": 727.10693359375,
+      "learning_rate": 6.588612947329929e-06,
+      "loss": 48.3992,
+      "step": 56600
+    },
+    {
+      "epoch": 0.22871156324616088,
+      "grad_norm": 829.1666259765625,
+      "learning_rate": 6.587275597059125e-06,
+      "loss": 25.1917,
+      "step": 56610
+    },
+    {
+      "epoch": 0.2287519645115285,
+      "grad_norm": 734.826416015625,
+      "learning_rate": 6.585938120496087e-06,
+      "loss": 44.4312,
+      "step": 56620
+    },
+    {
+      "epoch": 0.22879236577689613,
+      "grad_norm": 937.6222534179688,
+      "learning_rate": 6.584600517747232e-06,
+      "loss": 46.1989,
+      "step": 56630
+    },
+    {
+      "epoch": 0.22883276704226377,
+      "grad_norm": 714.8348388671875,
+      "learning_rate": 6.583262788918985e-06,
+      "loss": 47.9685,
+      "step": 56640
+    },
+    {
+      "epoch": 0.2288731683076314,
+      "grad_norm": 470.21832275390625,
+      "learning_rate": 6.581924934117783e-06,
+      "loss": 57.1887,
+      "step": 56650
+    },
+    {
+      "epoch": 0.22891356957299902,
+      "grad_norm": 594.2630615234375,
+      "learning_rate": 6.580586953450076e-06,
+      "loss": 41.4932,
+      "step": 56660
+    },
+    {
+      "epoch": 0.22895397083836666,
+      "grad_norm": 613.1303100585938,
+      "learning_rate": 6.579248847022317e-06,
+      "loss": 39.0782,
+      "step": 56670
+    },
+    {
+      "epoch": 0.2289943721037343,
+      "grad_norm": 756.616455078125,
+      "learning_rate": 6.577910614940978e-06,
+      "loss": 48.7595,
+      "step": 56680
+    },
+    {
+      "epoch": 0.2290347733691019,
+      "grad_norm": 333.4111633300781,
+      "learning_rate": 6.576572257312531e-06,
+      "loss": 41.8574,
+      "step": 56690
+    },
+    {
+      "epoch": 0.22907517463446955,
+      "grad_norm": 813.4775390625,
+      "learning_rate": 6.5752337742434644e-06,
+      "loss": 34.6469,
+      "step": 56700
+    },
+    {
+      "epoch": 0.2291155758998372,
+      "grad_norm": 498.3751220703125,
+      "learning_rate": 6.573895165840276e-06,
+      "loss": 47.3905,
+      "step": 56710
+    },
+    {
+      "epoch": 0.2291559771652048,
+      "grad_norm": 803.7112426757812,
+      "learning_rate": 6.5725564322094745e-06,
+      "loss": 59.7862,
+      "step": 56720
+    },
+    {
+      "epoch": 0.22919637843057244,
+      "grad_norm": 1082.3946533203125,
+      "learning_rate": 6.571217573457573e-06,
+      "loss": 53.9844,
+      "step": 56730
+    },
+    {
+      "epoch": 0.22923677969594008,
+      "grad_norm": 276.7662353515625,
+      "learning_rate": 6.569878589691101e-06,
+      "loss": 42.0922,
+      "step": 56740
+    },
+    {
+      "epoch": 0.2292771809613077,
+      "grad_norm": 1344.534912109375,
+      "learning_rate": 6.568539481016593e-06,
+      "loss": 47.5675,
+      "step": 56750
+    },
+    {
+      "epoch": 0.22931758222667534,
+      "grad_norm": 832.77490234375,
+      "learning_rate": 6.567200247540599e-06,
+      "loss": 68.7369,
+      "step": 56760
+    },
+    {
+      "epoch": 0.22935798349204298,
+      "grad_norm": 260.89385986328125,
+      "learning_rate": 6.5658608893696714e-06,
+      "loss": 45.8488,
+      "step": 56770
+    },
+    {
+      "epoch": 0.2293983847574106,
+      "grad_norm": 691.3558349609375,
+      "learning_rate": 6.564521406610382e-06,
+      "loss": 49.6963,
+      "step": 56780
+    },
+    {
+      "epoch": 0.22943878602277823,
+      "grad_norm": 866.0549926757812,
+      "learning_rate": 6.563181799369301e-06,
+      "loss": 32.6304,
+      "step": 56790
+    },
+    {
+      "epoch": 0.22947918728814587,
+      "grad_norm": 760.6442260742188,
+      "learning_rate": 6.561842067753021e-06,
+      "loss": 41.1299,
+      "step": 56800
+    },
+    {
+      "epoch": 0.2295195885535135,
+      "grad_norm": 1116.607666015625,
+      "learning_rate": 6.560502211868135e-06,
+      "loss": 56.4624,
+      "step": 56810
+    },
+    {
+      "epoch": 0.22955998981888112,
+      "grad_norm": 805.6516723632812,
+      "learning_rate": 6.55916223182125e-06,
+      "loss": 35.0971,
+      "step": 56820
+    },
+    {
+      "epoch": 0.22960039108424876,
+      "grad_norm": 378.9771728515625,
+      "learning_rate": 6.55782212771898e-06,
+      "loss": 43.8403,
+      "step": 56830
+    },
+    {
+      "epoch": 0.2296407923496164,
+      "grad_norm": 850.3655395507812,
+      "learning_rate": 6.5564818996679536e-06,
+      "loss": 50.1022,
+      "step": 56840
+    },
+    {
+      "epoch": 0.229681193614984,
+      "grad_norm": 827.5731811523438,
+      "learning_rate": 6.555141547774807e-06,
+      "loss": 66.2093,
+      "step": 56850
+    },
+    {
+      "epoch": 0.22972159488035165,
+      "grad_norm": 383.89361572265625,
+      "learning_rate": 6.553801072146184e-06,
+      "loss": 43.9942,
+      "step": 56860
+    },
+    {
+      "epoch": 0.2297619961457193,
+      "grad_norm": 634.6798095703125,
+      "learning_rate": 6.55246047288874e-06,
+      "loss": 63.0002,
+      "step": 56870
+    },
+    {
+      "epoch": 0.2298023974110869,
+      "grad_norm": 617.5804443359375,
+      "learning_rate": 6.551119750109142e-06,
+      "loss": 62.7401,
+      "step": 56880
+    },
+    {
+      "epoch": 0.22984279867645455,
+      "grad_norm": 757.3110961914062,
+      "learning_rate": 6.5497789039140635e-06,
+      "loss": 54.9512,
+      "step": 56890
+    },
+    {
+      "epoch": 0.22988319994182219,
+      "grad_norm": 897.7661743164062,
+      "learning_rate": 6.54843793441019e-06,
+      "loss": 42.9392,
+      "step": 56900
+    },
+    {
+      "epoch": 0.2299236012071898,
+      "grad_norm": 560.5343017578125,
+      "learning_rate": 6.547096841704217e-06,
+      "loss": 49.1479,
+      "step": 56910
+    },
+    {
+      "epoch": 0.22996400247255744,
+      "grad_norm": 169.0382537841797,
+      "learning_rate": 6.545755625902848e-06,
+      "loss": 40.0921,
+      "step": 56920
+    },
+    {
+      "epoch": 0.23000440373792508,
+      "grad_norm": 1067.6502685546875,
+      "learning_rate": 6.544414287112798e-06,
+      "loss": 62.6796,
+      "step": 56930
+    },
+    {
+      "epoch": 0.2300448050032927,
+      "grad_norm": 717.0487670898438,
+      "learning_rate": 6.54307282544079e-06,
+      "loss": 51.3947,
+      "step": 56940
+    },
+    {
+      "epoch": 0.23008520626866033,
+      "grad_norm": 572.951904296875,
+      "learning_rate": 6.5417312409935606e-06,
+      "loss": 52.8754,
+      "step": 56950
+    },
+    {
+      "epoch": 0.23012560753402797,
+      "grad_norm": 958.3705444335938,
+      "learning_rate": 6.540389533877852e-06,
+      "loss": 72.8596,
+      "step": 56960
+    },
+    {
+      "epoch": 0.2301660087993956,
+      "grad_norm": 515.6321411132812,
+      "learning_rate": 6.539047704200417e-06,
+      "loss": 47.7139,
+      "step": 56970
+    },
+    {
+      "epoch": 0.23020641006476322,
+      "grad_norm": 645.0633544921875,
+      "learning_rate": 6.53770575206802e-06,
+      "loss": 45.1984,
+      "step": 56980
+    },
+    {
+      "epoch": 0.23024681133013086,
+      "grad_norm": 382.80560302734375,
+      "learning_rate": 6.536363677587433e-06,
+      "loss": 42.9186,
+      "step": 56990
+    },
+    {
+      "epoch": 0.2302872125954985,
+      "grad_norm": 316.0374755859375,
+      "learning_rate": 6.535021480865439e-06,
+      "loss": 39.2293,
+      "step": 57000
+    },
+    {
+      "epoch": 0.23032761386086611,
+      "grad_norm": 455.7964782714844,
+      "learning_rate": 6.5336791620088306e-06,
+      "loss": 38.0706,
+      "step": 57010
+    },
+    {
+      "epoch": 0.23036801512623375,
+      "grad_norm": 625.7360229492188,
+      "learning_rate": 6.53233672112441e-06,
+      "loss": 38.7002,
+      "step": 57020
+    },
+    {
+      "epoch": 0.2304084163916014,
+      "grad_norm": 525.91845703125,
+      "learning_rate": 6.530994158318988e-06,
+      "loss": 46.5832,
+      "step": 57030
+    },
+    {
+      "epoch": 0.230448817656969,
+      "grad_norm": 578.9883422851562,
+      "learning_rate": 6.529651473699389e-06,
+      "loss": 35.1653,
+      "step": 57040
+    },
+    {
+      "epoch": 0.23048921892233665,
+      "grad_norm": 756.6240844726562,
+      "learning_rate": 6.528308667372441e-06,
+      "loss": 47.4836,
+      "step": 57050
+    },
+    {
+      "epoch": 0.2305296201877043,
+      "grad_norm": 375.4329528808594,
+      "learning_rate": 6.526965739444988e-06,
+      "loss": 49.6254,
+      "step": 57060
+    },
+    {
+      "epoch": 0.2305700214530719,
+      "grad_norm": 856.4522705078125,
+      "learning_rate": 6.525622690023878e-06,
+      "loss": 54.8425,
+      "step": 57070
+    },
+    {
+      "epoch": 0.23061042271843954,
+      "grad_norm": 648.9761352539062,
+      "learning_rate": 6.524279519215972e-06,
+      "loss": 49.5565,
+      "step": 57080
+    },
+    {
+      "epoch": 0.23065082398380718,
+      "grad_norm": 279.9421081542969,
+      "learning_rate": 6.522936227128139e-06,
+      "loss": 29.2129,
+      "step": 57090
+    },
+    {
+      "epoch": 0.2306912252491748,
+      "grad_norm": 434.59619140625,
+      "learning_rate": 6.521592813867261e-06,
+      "loss": 38.2154,
+      "step": 57100
+    },
+    {
+      "epoch": 0.23073162651454243,
+      "grad_norm": 904.1709594726562,
+      "learning_rate": 6.520249279540227e-06,
+      "loss": 48.9872,
+      "step": 57110
+    },
+    {
+      "epoch": 0.23077202777991007,
+      "grad_norm": 2435.330810546875,
+      "learning_rate": 6.5189056242539325e-06,
+      "loss": 60.082,
+      "step": 57120
+    },
+    {
+      "epoch": 0.2308124290452777,
+      "grad_norm": 279.1988525390625,
+      "learning_rate": 6.51756184811529e-06,
+      "loss": 40.6881,
+      "step": 57130
+    },
+    {
+      "epoch": 0.23085283031064532,
+      "grad_norm": 811.0878295898438,
+      "learning_rate": 6.516217951231215e-06,
+      "loss": 51.1541,
+      "step": 57140
+    },
+    {
+      "epoch": 0.23089323157601296,
+      "grad_norm": 386.1441955566406,
+      "learning_rate": 6.514873933708637e-06,
+      "loss": 33.2295,
+      "step": 57150
+    },
+    {
+      "epoch": 0.2309336328413806,
+      "grad_norm": 745.3314819335938,
+      "learning_rate": 6.513529795654493e-06,
+      "loss": 48.7001,
+      "step": 57160
+    },
+    {
+      "epoch": 0.23097403410674822,
+      "grad_norm": 761.8244018554688,
+      "learning_rate": 6.512185537175727e-06,
+      "loss": 34.5987,
+      "step": 57170
+    },
+    {
+      "epoch": 0.23101443537211586,
+      "grad_norm": 357.27984619140625,
+      "learning_rate": 6.5108411583793e-06,
+      "loss": 63.6459,
+      "step": 57180
+    },
+    {
+      "epoch": 0.2310548366374835,
+      "grad_norm": 698.4451293945312,
+      "learning_rate": 6.509496659372175e-06,
+      "loss": 31.1018,
+      "step": 57190
+    },
+    {
+      "epoch": 0.2310952379028511,
+      "grad_norm": 948.5643310546875,
+      "learning_rate": 6.508152040261329e-06,
+      "loss": 49.5578,
+      "step": 57200
+    },
+    {
+      "epoch": 0.23113563916821875,
+      "grad_norm": 1980.6900634765625,
+      "learning_rate": 6.506807301153746e-06,
+      "loss": 47.9453,
+      "step": 57210
+    },
+    {
+      "epoch": 0.2311760404335864,
+      "grad_norm": 1299.60595703125,
+      "learning_rate": 6.5054624421564204e-06,
+      "loss": 50.8445,
+      "step": 57220
+    },
+    {
+      "epoch": 0.231216441698954,
+      "grad_norm": 1376.9063720703125,
+      "learning_rate": 6.504117463376358e-06,
+      "loss": 56.114,
+      "step": 57230
+    },
+    {
+      "epoch": 0.23125684296432164,
+      "grad_norm": 463.1986999511719,
+      "learning_rate": 6.502772364920573e-06,
+      "loss": 48.9991,
+      "step": 57240
+    },
+    {
+      "epoch": 0.23129724422968928,
+      "grad_norm": 194.3710479736328,
+      "learning_rate": 6.501427146896087e-06,
+      "loss": 31.2601,
+      "step": 57250
+    },
+    {
+      "epoch": 0.2313376454950569,
+      "grad_norm": 565.7164306640625,
+      "learning_rate": 6.5000818094099345e-06,
+      "loss": 39.7205,
+      "step": 57260
+    },
+    {
+      "epoch": 0.23137804676042453,
+      "grad_norm": 511.2424011230469,
+      "learning_rate": 6.498736352569155e-06,
+      "loss": 58.0947,
+      "step": 57270
+    },
+    {
+      "epoch": 0.23141844802579217,
+      "grad_norm": 545.729248046875,
+      "learning_rate": 6.497390776480804e-06,
+      "loss": 43.7937,
+      "step": 57280
+    },
+    {
+      "epoch": 0.2314588492911598,
+      "grad_norm": 1081.4024658203125,
+      "learning_rate": 6.49604508125194e-06,
+      "loss": 57.2301,
+      "step": 57290
+    },
+    {
+      "epoch": 0.23149925055652743,
+      "grad_norm": 858.1848754882812,
+      "learning_rate": 6.4946992669896355e-06,
+      "loss": 42.6667,
+      "step": 57300
+    },
+    {
+      "epoch": 0.23153965182189506,
+      "grad_norm": 176.55555725097656,
+      "learning_rate": 6.493353333800969e-06,
+      "loss": 41.2503,
+      "step": 57310
+    },
+    {
+      "epoch": 0.2315800530872627,
+      "grad_norm": 835.7853393554688,
+      "learning_rate": 6.492007281793032e-06,
+      "loss": 69.4198,
+      "step": 57320
+    },
+    {
+      "epoch": 0.23162045435263032,
+      "grad_norm": 532.6726684570312,
+      "learning_rate": 6.490661111072923e-06,
+      "loss": 43.0533,
+      "step": 57330
+    },
+    {
+      "epoch": 0.23166085561799796,
+      "grad_norm": 395.8429260253906,
+      "learning_rate": 6.489314821747751e-06,
+      "loss": 45.0708,
+      "step": 57340
+    },
+    {
+      "epoch": 0.2317012568833656,
+      "grad_norm": 424.9057312011719,
+      "learning_rate": 6.487968413924634e-06,
+      "loss": 51.2508,
+      "step": 57350
+    },
+    {
+      "epoch": 0.2317416581487332,
+      "grad_norm": 750.9067993164062,
+      "learning_rate": 6.486621887710698e-06,
+      "loss": 43.8176,
+      "step": 57360
+    },
+    {
+      "epoch": 0.23178205941410085,
+      "grad_norm": 454.9244079589844,
+      "learning_rate": 6.485275243213081e-06,
+      "loss": 43.997,
+      "step": 57370
+    },
+    {
+      "epoch": 0.2318224606794685,
+      "grad_norm": 515.5105590820312,
+      "learning_rate": 6.4839284805389305e-06,
+      "loss": 44.7541,
+      "step": 57380
+    },
+    {
+      "epoch": 0.2318628619448361,
+      "grad_norm": 854.0843505859375,
+      "learning_rate": 6.4825815997954e-06,
+      "loss": 51.4215,
+      "step": 57390
+    },
+    {
+      "epoch": 0.23190326321020374,
+      "grad_norm": 554.3998413085938,
+      "learning_rate": 6.481234601089655e-06,
+      "loss": 50.3499,
+      "step": 57400
+    },
+    {
+      "epoch": 0.23194366447557138,
+      "grad_norm": 656.9430541992188,
+      "learning_rate": 6.4798874845288725e-06,
+      "loss": 53.9586,
+      "step": 57410
+    },
+    {
+      "epoch": 0.231984065740939,
+      "grad_norm": 429.04681396484375,
+      "learning_rate": 6.4785402502202345e-06,
+      "loss": 58.049,
+      "step": 57420
+    },
+    {
+      "epoch": 0.23202446700630663,
+      "grad_norm": 669.2875366210938,
+      "learning_rate": 6.477192898270934e-06,
+      "loss": 33.2239,
+      "step": 57430
+    },
+    {
+      "epoch": 0.23206486827167427,
+      "grad_norm": 382.91522216796875,
+      "learning_rate": 6.475845428788173e-06,
+      "loss": 50.4403,
+      "step": 57440
+    },
+    {
+      "epoch": 0.23210526953704191,
+      "grad_norm": 650.1845092773438,
+      "learning_rate": 6.474497841879166e-06,
+      "loss": 65.9487,
+      "step": 57450
+    },
+    {
+      "epoch": 0.23214567080240953,
+      "grad_norm": 991.1076049804688,
+      "learning_rate": 6.473150137651132e-06,
+      "loss": 57.2718,
+      "step": 57460
+    },
+    {
+      "epoch": 0.23218607206777717,
+      "grad_norm": 714.9050903320312,
+      "learning_rate": 6.471802316211302e-06,
+      "loss": 57.2066,
+      "step": 57470
+    },
+    {
+      "epoch": 0.2322264733331448,
+      "grad_norm": 483.4032897949219,
+      "learning_rate": 6.4704543776669174e-06,
+      "loss": 45.9662,
+      "step": 57480
+    },
+    {
+      "epoch": 0.23226687459851242,
+      "grad_norm": 353.4619445800781,
+      "learning_rate": 6.469106322125227e-06,
+      "loss": 51.166,
+      "step": 57490
+    },
+    {
+      "epoch": 0.23230727586388006,
+      "grad_norm": 700.385986328125,
+      "learning_rate": 6.467758149693486e-06,
+      "loss": 49.1312,
+      "step": 57500
+    },
+    {
+      "epoch": 0.2323476771292477,
+      "grad_norm": 429.9539794921875,
+      "learning_rate": 6.466409860478967e-06,
+      "loss": 48.9266,
+      "step": 57510
+    },
+    {
+      "epoch": 0.2323880783946153,
+      "grad_norm": 865.5701293945312,
+      "learning_rate": 6.465061454588946e-06,
+      "loss": 51.0269,
+      "step": 57520
+    },
+    {
+      "epoch": 0.23242847965998295,
+      "grad_norm": 457.370361328125,
+      "learning_rate": 6.463712932130708e-06,
+      "loss": 38.1473,
+      "step": 57530
+    },
+    {
+      "epoch": 0.2324688809253506,
+      "grad_norm": 432.3177795410156,
+      "learning_rate": 6.462364293211549e-06,
+      "loss": 33.1413,
+      "step": 57540
+    },
+    {
+      "epoch": 0.2325092821907182,
+      "grad_norm": 454.0198669433594,
+      "learning_rate": 6.4610155379387755e-06,
+      "loss": 36.9611,
+      "step": 57550
+    },
+    {
+      "epoch": 0.23254968345608584,
+      "grad_norm": 459.07940673828125,
+      "learning_rate": 6.459666666419699e-06,
+      "loss": 53.3877,
+      "step": 57560
+    },
+    {
+      "epoch": 0.23259008472145348,
+      "grad_norm": 552.9193115234375,
+      "learning_rate": 6.4583176787616466e-06,
+      "loss": 55.7331,
+      "step": 57570
+    },
+    {
+      "epoch": 0.2326304859868211,
+      "grad_norm": 187.45762634277344,
+      "learning_rate": 6.456968575071951e-06,
+      "loss": 68.7932,
+      "step": 57580
+    },
+    {
+      "epoch": 0.23267088725218874,
+      "grad_norm": 276.4294128417969,
+      "learning_rate": 6.45561935545795e-06,
+      "loss": 36.9424,
+      "step": 57590
+    },
+    {
+      "epoch": 0.23271128851755638,
+      "grad_norm": 488.11151123046875,
+      "learning_rate": 6.454270020026996e-06,
+      "loss": 59.5057,
+      "step": 57600
+    },
+    {
+      "epoch": 0.23275168978292402,
+      "grad_norm": 763.0408325195312,
+      "learning_rate": 6.452920568886452e-06,
+      "loss": 34.2426,
+      "step": 57610
+    },
+    {
+      "epoch": 0.23279209104829163,
+      "grad_norm": 325.465576171875,
+      "learning_rate": 6.451571002143687e-06,
+      "loss": 47.0651,
+      "step": 57620
+    },
+    {
+      "epoch": 0.23283249231365927,
+      "grad_norm": 1015.4959716796875,
+      "learning_rate": 6.450221319906079e-06,
+      "loss": 41.7232,
+      "step": 57630
+    },
+    {
+      "epoch": 0.2328728935790269,
+      "grad_norm": 1036.8895263671875,
+      "learning_rate": 6.448871522281016e-06,
+      "loss": 48.3217,
+      "step": 57640
+    },
+    {
+      "epoch": 0.23291329484439452,
+      "grad_norm": 424.6831359863281,
+      "learning_rate": 6.447521609375894e-06,
+      "loss": 38.218,
+      "step": 57650
+    },
+    {
+      "epoch": 0.23295369610976216,
+      "grad_norm": 742.8676147460938,
+      "learning_rate": 6.446171581298123e-06,
+      "loss": 38.9838,
+      "step": 57660
+    },
+    {
+      "epoch": 0.2329940973751298,
+      "grad_norm": 821.5770874023438,
+      "learning_rate": 6.444821438155115e-06,
+      "loss": 45.9229,
+      "step": 57670
+    },
+    {
+      "epoch": 0.2330344986404974,
+      "grad_norm": 606.6029052734375,
+      "learning_rate": 6.443471180054297e-06,
+      "loss": 53.2201,
+      "step": 57680
+    },
+    {
+      "epoch": 0.23307489990586505,
+      "grad_norm": 684.6781005859375,
+      "learning_rate": 6.442120807103102e-06,
+      "loss": 59.1518,
+      "step": 57690
+    },
+    {
+      "epoch": 0.2331153011712327,
+      "grad_norm": 678.3739624023438,
+      "learning_rate": 6.440770319408971e-06,
+      "loss": 44.6293,
+      "step": 57700
+    },
+    {
+      "epoch": 0.2331557024366003,
+      "grad_norm": 648.0844116210938,
+      "learning_rate": 6.43941971707936e-06,
+      "loss": 30.0908,
+      "step": 57710
+    },
+    {
+      "epoch": 0.23319610370196794,
+      "grad_norm": 406.64862060546875,
+      "learning_rate": 6.438069000221727e-06,
+      "loss": 50.2008,
+      "step": 57720
+    },
+    {
+      "epoch": 0.23323650496733558,
+      "grad_norm": 673.765380859375,
+      "learning_rate": 6.4367181689435434e-06,
+      "loss": 46.924,
+      "step": 57730
+    },
+    {
+      "epoch": 0.2332769062327032,
+      "grad_norm": 3202.600341796875,
+      "learning_rate": 6.435367223352289e-06,
+      "loss": 63.5285,
+      "step": 57740
+    },
+    {
+      "epoch": 0.23331730749807084,
+      "grad_norm": 1331.566162109375,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 50.6696,
+      "step": 57750
+    },
+    {
+      "epoch": 0.23335770876343848,
+      "grad_norm": 383.6949768066406,
+      "learning_rate": 6.432664989660531e-06,
+      "loss": 37.0474,
+      "step": 57760
+    },
+    {
+      "epoch": 0.2333981100288061,
+      "grad_norm": 224.4041748046875,
+      "learning_rate": 6.43131370177503e-06,
+      "loss": 46.556,
+      "step": 57770
+    },
+    {
+      "epoch": 0.23343851129417373,
+      "grad_norm": 1214.294677734375,
+      "learning_rate": 6.429962300006468e-06,
+      "loss": 72.8442,
+      "step": 57780
+    },
+    {
+      "epoch": 0.23347891255954137,
+      "grad_norm": 615.74560546875,
+      "learning_rate": 6.428610784462368e-06,
+      "loss": 46.897,
+      "step": 57790
+    },
+    {
+      "epoch": 0.233519313824909,
+      "grad_norm": 600.8942260742188,
+      "learning_rate": 6.427259155250265e-06,
+      "loss": 56.1212,
+      "step": 57800
+    },
+    {
+      "epoch": 0.23355971509027662,
+      "grad_norm": 1422.342041015625,
+      "learning_rate": 6.4259074124777e-06,
+      "loss": 53.6335,
+      "step": 57810
+    },
+    {
+      "epoch": 0.23360011635564426,
+      "grad_norm": 1549.554443359375,
+      "learning_rate": 6.4245555562522265e-06,
+      "loss": 34.9436,
+      "step": 57820
+    },
+    {
+      "epoch": 0.2336405176210119,
+      "grad_norm": 604.1146850585938,
+      "learning_rate": 6.423203586681406e-06,
+      "loss": 43.1287,
+      "step": 57830
+    },
+    {
+      "epoch": 0.2336809188863795,
+      "grad_norm": 181.16397094726562,
+      "learning_rate": 6.421851503872807e-06,
+      "loss": 42.3401,
+      "step": 57840
+    },
+    {
+      "epoch": 0.23372132015174715,
+      "grad_norm": 438.2908935546875,
+      "learning_rate": 6.42049930793401e-06,
+      "loss": 58.2065,
+      "step": 57850
+    },
+    {
+      "epoch": 0.2337617214171148,
+      "grad_norm": 865.8299560546875,
+      "learning_rate": 6.419146998972602e-06,
+      "loss": 74.7755,
+      "step": 57860
+    },
+    {
+      "epoch": 0.2338021226824824,
+      "grad_norm": 399.0161437988281,
+      "learning_rate": 6.417794577096179e-06,
+      "loss": 41.3667,
+      "step": 57870
+    },
+    {
+      "epoch": 0.23384252394785005,
+      "grad_norm": 1323.824951171875,
+      "learning_rate": 6.41644204241235e-06,
+      "loss": 48.4313,
+      "step": 57880
+    },
+    {
+      "epoch": 0.23388292521321769,
+      "grad_norm": 1061.3271484375,
+      "learning_rate": 6.4150893950287275e-06,
+      "loss": 61.6943,
+      "step": 57890
+    },
+    {
+      "epoch": 0.2339233264785853,
+      "grad_norm": 631.509521484375,
+      "learning_rate": 6.413736635052936e-06,
+      "loss": 39.0582,
+      "step": 57900
+    },
+    {
+      "epoch": 0.23396372774395294,
+      "grad_norm": 904.318359375,
+      "learning_rate": 6.41238376259261e-06,
+      "loss": 46.4853,
+      "step": 57910
+    },
+    {
+      "epoch": 0.23400412900932058,
+      "grad_norm": 380.4770812988281,
+      "learning_rate": 6.411030777755389e-06,
+      "loss": 53.1108,
+      "step": 57920
+    },
+    {
+      "epoch": 0.2340445302746882,
+      "grad_norm": 1202.594970703125,
+      "learning_rate": 6.409677680648925e-06,
+      "loss": 68.2252,
+      "step": 57930
+    },
+    {
+      "epoch": 0.23408493154005583,
+      "grad_norm": 1343.1907958984375,
+      "learning_rate": 6.4083244713808765e-06,
+      "loss": 32.3128,
+      "step": 57940
+    },
+    {
+      "epoch": 0.23412533280542347,
+      "grad_norm": 413.1898498535156,
+      "learning_rate": 6.406971150058914e-06,
+      "loss": 37.7907,
+      "step": 57950
+    },
+    {
+      "epoch": 0.2341657340707911,
+      "grad_norm": 864.8408813476562,
+      "learning_rate": 6.405617716790714e-06,
+      "loss": 37.8921,
+      "step": 57960
+    },
+    {
+      "epoch": 0.23420613533615872,
+      "grad_norm": 854.4756469726562,
+      "learning_rate": 6.404264171683965e-06,
+      "loss": 47.9397,
+      "step": 57970
+    },
+    {
+      "epoch": 0.23424653660152636,
+      "grad_norm": 743.707275390625,
+      "learning_rate": 6.402910514846358e-06,
+      "loss": 30.2608,
+      "step": 57980
+    },
+    {
+      "epoch": 0.234286937866894,
+      "grad_norm": 734.5051879882812,
+      "learning_rate": 6.4015567463856e-06,
+      "loss": 41.0997,
+      "step": 57990
+    },
+    {
+      "epoch": 0.23432733913226161,
+      "grad_norm": 744.6089477539062,
+      "learning_rate": 6.400202866409405e-06,
+      "loss": 44.8084,
+      "step": 58000
+    },
+    {
+      "epoch": 0.23436774039762925,
+      "grad_norm": 631.4942626953125,
+      "learning_rate": 6.398848875025494e-06,
+      "loss": 41.5169,
+      "step": 58010
+    },
+    {
+      "epoch": 0.2344081416629969,
+      "grad_norm": 924.2283325195312,
+      "learning_rate": 6.3974947723415985e-06,
+      "loss": 57.7236,
+      "step": 58020
+    },
+    {
+      "epoch": 0.2344485429283645,
+      "grad_norm": 748.7590942382812,
+      "learning_rate": 6.396140558465456e-06,
+      "loss": 39.0069,
+      "step": 58030
+    },
+    {
+      "epoch": 0.23448894419373215,
+      "grad_norm": 1556.5047607421875,
+      "learning_rate": 6.394786233504816e-06,
+      "loss": 67.497,
+      "step": 58040
+    },
+    {
+      "epoch": 0.2345293454590998,
+      "grad_norm": 560.0980834960938,
+      "learning_rate": 6.39343179756744e-06,
+      "loss": 36.9168,
+      "step": 58050
+    },
+    {
+      "epoch": 0.2345697467244674,
+      "grad_norm": 695.441650390625,
+      "learning_rate": 6.392077250761088e-06,
+      "loss": 43.3128,
+      "step": 58060
+    },
+    {
+      "epoch": 0.23461014798983504,
+      "grad_norm": 722.3070678710938,
+      "learning_rate": 6.390722593193538e-06,
+      "loss": 53.5355,
+      "step": 58070
+    },
+    {
+      "epoch": 0.23465054925520268,
+      "grad_norm": 390.59124755859375,
+      "learning_rate": 6.389367824972575e-06,
+      "loss": 52.2937,
+      "step": 58080
+    },
+    {
+      "epoch": 0.2346909505205703,
+      "grad_norm": 448.1783142089844,
+      "learning_rate": 6.388012946205991e-06,
+      "loss": 44.6697,
+      "step": 58090
+    },
+    {
+      "epoch": 0.23473135178593793,
+      "grad_norm": 252.83447265625,
+      "learning_rate": 6.386657957001585e-06,
+      "loss": 43.3268,
+      "step": 58100
+    },
+    {
+      "epoch": 0.23477175305130557,
+      "grad_norm": 572.3157348632812,
+      "learning_rate": 6.38530285746717e-06,
+      "loss": 32.4306,
+      "step": 58110
+    },
+    {
+      "epoch": 0.2348121543166732,
+      "grad_norm": 326.7099609375,
+      "learning_rate": 6.383947647710565e-06,
+      "loss": 30.3495,
+      "step": 58120
+    },
+    {
+      "epoch": 0.23485255558204082,
+      "grad_norm": 636.6470336914062,
+      "learning_rate": 6.382592327839596e-06,
+      "loss": 34.1573,
+      "step": 58130
+    },
+    {
+      "epoch": 0.23489295684740846,
+      "grad_norm": 648.021484375,
+      "learning_rate": 6.381236897962102e-06,
+      "loss": 45.2606,
+      "step": 58140
+    },
+    {
+      "epoch": 0.2349333581127761,
+      "grad_norm": 3382.697998046875,
+      "learning_rate": 6.379881358185926e-06,
+      "loss": 48.3623,
+      "step": 58150
+    },
+    {
+      "epoch": 0.23497375937814372,
+      "grad_norm": 376.2268981933594,
+      "learning_rate": 6.378525708618924e-06,
+      "loss": 43.6891,
+      "step": 58160
+    },
+    {
+      "epoch": 0.23501416064351136,
+      "grad_norm": 846.15625,
+      "learning_rate": 6.377169949368956e-06,
+      "loss": 55.0067,
+      "step": 58170
+    },
+    {
+      "epoch": 0.235054561908879,
+      "grad_norm": 336.41802978515625,
+      "learning_rate": 6.375814080543899e-06,
+      "loss": 33.695,
+      "step": 58180
+    },
+    {
+      "epoch": 0.2350949631742466,
+      "grad_norm": 436.46197509765625,
+      "learning_rate": 6.3744581022516285e-06,
+      "loss": 242.9059,
+      "step": 58190
+    },
+    {
+      "epoch": 0.23513536443961425,
+      "grad_norm": 740.7402954101562,
+      "learning_rate": 6.373102014600033e-06,
+      "loss": 43.7324,
+      "step": 58200
+    },
+    {
+      "epoch": 0.2351757657049819,
+      "grad_norm": 866.8933715820312,
+      "learning_rate": 6.371745817697012e-06,
+      "loss": 49.1007,
+      "step": 58210
+    },
+    {
+      "epoch": 0.2352161669703495,
+      "grad_norm": 681.402099609375,
+      "learning_rate": 6.370389511650474e-06,
+      "loss": 48.2255,
+      "step": 58220
+    },
+    {
+      "epoch": 0.23525656823571714,
+      "grad_norm": 598.37255859375,
+      "learning_rate": 6.3690330965683304e-06,
+      "loss": 43.6901,
+      "step": 58230
+    },
+    {
+      "epoch": 0.23529696950108478,
+      "grad_norm": 616.9476928710938,
+      "learning_rate": 6.367676572558506e-06,
+      "loss": 60.4343,
+      "step": 58240
+    },
+    {
+      "epoch": 0.2353373707664524,
+      "grad_norm": 881.7716674804688,
+      "learning_rate": 6.366319939728934e-06,
+      "loss": 114.1712,
+      "step": 58250
+    },
+    {
+      "epoch": 0.23537777203182003,
+      "grad_norm": 505.7012023925781,
+      "learning_rate": 6.364963198187555e-06,
+      "loss": 54.7326,
+      "step": 58260
+    },
+    {
+      "epoch": 0.23541817329718767,
+      "grad_norm": 464.7174072265625,
+      "learning_rate": 6.363606348042318e-06,
+      "loss": 38.2772,
+      "step": 58270
+    },
+    {
+      "epoch": 0.2354585745625553,
+      "grad_norm": 602.1082763671875,
+      "learning_rate": 6.362249389401183e-06,
+      "loss": 37.5626,
+      "step": 58280
+    },
+    {
+      "epoch": 0.23549897582792292,
+      "grad_norm": 578.7637939453125,
+      "learning_rate": 6.360892322372115e-06,
+      "loss": 46.3746,
+      "step": 58290
+    },
+    {
+      "epoch": 0.23553937709329056,
+      "grad_norm": 298.9007873535156,
+      "learning_rate": 6.359535147063092e-06,
+      "loss": 48.7875,
+      "step": 58300
+    },
+    {
+      "epoch": 0.2355797783586582,
+      "grad_norm": 433.61517333984375,
+      "learning_rate": 6.358177863582095e-06,
+      "loss": 41.4706,
+      "step": 58310
+    },
+    {
+      "epoch": 0.23562017962402582,
+      "grad_norm": 636.701416015625,
+      "learning_rate": 6.35682047203712e-06,
+      "loss": 37.8823,
+      "step": 58320
+    },
+    {
+      "epoch": 0.23566058088939346,
+      "grad_norm": 681.9437255859375,
+      "learning_rate": 6.355462972536166e-06,
+      "loss": 49.5436,
+      "step": 58330
+    },
+    {
+      "epoch": 0.2357009821547611,
+      "grad_norm": 658.2206420898438,
+      "learning_rate": 6.354105365187244e-06,
+      "loss": 46.6898,
+      "step": 58340
+    },
+    {
+      "epoch": 0.2357413834201287,
+      "grad_norm": 502.2849426269531,
+      "learning_rate": 6.352747650098373e-06,
+      "loss": 40.1756,
+      "step": 58350
+    },
+    {
+      "epoch": 0.23578178468549635,
+      "grad_norm": 2149.325927734375,
+      "learning_rate": 6.35138982737758e-06,
+      "loss": 61.8135,
+      "step": 58360
+    },
+    {
+      "epoch": 0.235822185950864,
+      "grad_norm": 403.00921630859375,
+      "learning_rate": 6.3500318971329e-06,
+      "loss": 40.8118,
+      "step": 58370
+    },
+    {
+      "epoch": 0.2358625872162316,
+      "grad_norm": 667.1368408203125,
+      "learning_rate": 6.348673859472378e-06,
+      "loss": 52.6098,
+      "step": 58380
+    },
+    {
+      "epoch": 0.23590298848159924,
+      "grad_norm": 488.9213562011719,
+      "learning_rate": 6.347315714504066e-06,
+      "loss": 44.4715,
+      "step": 58390
+    },
+    {
+      "epoch": 0.23594338974696688,
+      "grad_norm": 560.7942504882812,
+      "learning_rate": 6.345957462336026e-06,
+      "loss": 62.3928,
+      "step": 58400
+    },
+    {
+      "epoch": 0.2359837910123345,
+      "grad_norm": 658.2381591796875,
+      "learning_rate": 6.344599103076329e-06,
+      "loss": 47.269,
+      "step": 58410
+    },
+    {
+      "epoch": 0.23602419227770213,
+      "grad_norm": 415.65850830078125,
+      "learning_rate": 6.343240636833051e-06,
+      "loss": 47.3057,
+      "step": 58420
+    },
+    {
+      "epoch": 0.23606459354306977,
+      "grad_norm": 1434.79150390625,
+      "learning_rate": 6.341882063714282e-06,
+      "loss": 50.7882,
+      "step": 58430
+    },
+    {
+      "epoch": 0.2361049948084374,
+      "grad_norm": 574.5676879882812,
+      "learning_rate": 6.340523383828115e-06,
+      "loss": 44.911,
+      "step": 58440
+    },
+    {
+      "epoch": 0.23614539607380503,
+      "grad_norm": 289.6334533691406,
+      "learning_rate": 6.339164597282652e-06,
+      "loss": 41.7514,
+      "step": 58450
+    },
+    {
+      "epoch": 0.23618579733917267,
+      "grad_norm": 990.1171264648438,
+      "learning_rate": 6.337805704186011e-06,
+      "loss": 59.5212,
+      "step": 58460
+    },
+    {
+      "epoch": 0.2362261986045403,
+      "grad_norm": 394.655029296875,
+      "learning_rate": 6.336446704646307e-06,
+      "loss": 41.2014,
+      "step": 58470
+    },
+    {
+      "epoch": 0.23626659986990792,
+      "grad_norm": 407.64886474609375,
+      "learning_rate": 6.335087598771676e-06,
+      "loss": 34.4346,
+      "step": 58480
+    },
+    {
+      "epoch": 0.23630700113527556,
+      "grad_norm": 199.44761657714844,
+      "learning_rate": 6.333728386670249e-06,
+      "loss": 40.7228,
+      "step": 58490
+    },
+    {
+      "epoch": 0.2363474024006432,
+      "grad_norm": 615.704345703125,
+      "learning_rate": 6.332369068450175e-06,
+      "loss": 46.4799,
+      "step": 58500
+    },
+    {
+      "epoch": 0.2363878036660108,
+      "grad_norm": 356.4165954589844,
+      "learning_rate": 6.33100964421961e-06,
+      "loss": 49.1076,
+      "step": 58510
+    },
+    {
+      "epoch": 0.23642820493137845,
+      "grad_norm": 745.1015625,
+      "learning_rate": 6.329650114086717e-06,
+      "loss": 39.9381,
+      "step": 58520
+    },
+    {
+      "epoch": 0.2364686061967461,
+      "grad_norm": 222.75173950195312,
+      "learning_rate": 6.328290478159666e-06,
+      "loss": 56.0289,
+      "step": 58530
+    },
+    {
+      "epoch": 0.2365090074621137,
+      "grad_norm": 823.6661987304688,
+      "learning_rate": 6.326930736546637e-06,
+      "loss": 52.2594,
+      "step": 58540
+    },
+    {
+      "epoch": 0.23654940872748134,
+      "grad_norm": 817.8341674804688,
+      "learning_rate": 6.325570889355819e-06,
+      "loss": 38.3977,
+      "step": 58550
+    },
+    {
+      "epoch": 0.23658980999284898,
+      "grad_norm": 1076.7332763671875,
+      "learning_rate": 6.32421093669541e-06,
+      "loss": 60.3006,
+      "step": 58560
+    },
+    {
+      "epoch": 0.2366302112582166,
+      "grad_norm": 1023.4537353515625,
+      "learning_rate": 6.322850878673614e-06,
+      "loss": 55.8991,
+      "step": 58570
+    },
+    {
+      "epoch": 0.23667061252358423,
+      "grad_norm": 495.56573486328125,
+      "learning_rate": 6.321490715398644e-06,
+      "loss": 48.1315,
+      "step": 58580
+    },
+    {
+      "epoch": 0.23671101378895187,
+      "grad_norm": 806.6759033203125,
+      "learning_rate": 6.320130446978722e-06,
+      "loss": 74.4598,
+      "step": 58590
+    },
+    {
+      "epoch": 0.23675141505431951,
+      "grad_norm": 673.5408935546875,
+      "learning_rate": 6.31877007352208e-06,
+      "loss": 51.3672,
+      "step": 58600
+    },
+    {
+      "epoch": 0.23679181631968713,
+      "grad_norm": 794.4425659179688,
+      "learning_rate": 6.317409595136956e-06,
+      "loss": 34.9962,
+      "step": 58610
+    },
+    {
+      "epoch": 0.23683221758505477,
+      "grad_norm": 905.353271484375,
+      "learning_rate": 6.316049011931595e-06,
+      "loss": 53.9744,
+      "step": 58620
+    },
+    {
+      "epoch": 0.2368726188504224,
+      "grad_norm": 524.4614868164062,
+      "learning_rate": 6.314688324014255e-06,
+      "loss": 48.7819,
+      "step": 58630
+    },
+    {
+      "epoch": 0.23691302011579002,
+      "grad_norm": 865.238037109375,
+      "learning_rate": 6.3133275314931995e-06,
+      "loss": 37.5079,
+      "step": 58640
+    },
+    {
+      "epoch": 0.23695342138115766,
+      "grad_norm": 643.4755249023438,
+      "learning_rate": 6.311966634476698e-06,
+      "loss": 55.7277,
+      "step": 58650
+    },
+    {
+      "epoch": 0.2369938226465253,
+      "grad_norm": 467.0180358886719,
+      "learning_rate": 6.3106056330730335e-06,
+      "loss": 41.7435,
+      "step": 58660
+    },
+    {
+      "epoch": 0.2370342239118929,
+      "grad_norm": 504.303955078125,
+      "learning_rate": 6.309244527390493e-06,
+      "loss": 35.1575,
+      "step": 58670
+    },
+    {
+      "epoch": 0.23707462517726055,
+      "grad_norm": 573.1863403320312,
+      "learning_rate": 6.307883317537375e-06,
+      "loss": 45.4351,
+      "step": 58680
+    },
+    {
+      "epoch": 0.2371150264426282,
+      "grad_norm": 567.236328125,
+      "learning_rate": 6.306522003621983e-06,
+      "loss": 38.8636,
+      "step": 58690
+    },
+    {
+      "epoch": 0.2371554277079958,
+      "grad_norm": 846.465087890625,
+      "learning_rate": 6.305160585752632e-06,
+      "loss": 33.141,
+      "step": 58700
+    },
+    {
+      "epoch": 0.23719582897336344,
+      "grad_norm": 710.2720947265625,
+      "learning_rate": 6.303799064037643e-06,
+      "loss": 38.8927,
+      "step": 58710
+    },
+    {
+      "epoch": 0.23723623023873108,
+      "grad_norm": 320.60308837890625,
+      "learning_rate": 6.302437438585345e-06,
+      "loss": 35.5578,
+      "step": 58720
+    },
+    {
+      "epoch": 0.2372766315040987,
+      "grad_norm": 1115.669189453125,
+      "learning_rate": 6.301075709504077e-06,
+      "loss": 64.423,
+      "step": 58730
+    },
+    {
+      "epoch": 0.23731703276946634,
+      "grad_norm": 828.960205078125,
+      "learning_rate": 6.299713876902188e-06,
+      "loss": 59.0923,
+      "step": 58740
+    },
+    {
+      "epoch": 0.23735743403483398,
+      "grad_norm": 619.6551513671875,
+      "learning_rate": 6.29835194088803e-06,
+      "loss": 42.279,
+      "step": 58750
+    },
+    {
+      "epoch": 0.23739783530020162,
+      "grad_norm": 628.1817016601562,
+      "learning_rate": 6.296989901569966e-06,
+      "loss": 59.1546,
+      "step": 58760
+    },
+    {
+      "epoch": 0.23743823656556923,
+      "grad_norm": 412.5716857910156,
+      "learning_rate": 6.295627759056368e-06,
+      "loss": 42.7949,
+      "step": 58770
+    },
+    {
+      "epoch": 0.23747863783093687,
+      "grad_norm": 380.8799743652344,
+      "learning_rate": 6.294265513455616e-06,
+      "loss": 37.1053,
+      "step": 58780
+    },
+    {
+      "epoch": 0.2375190390963045,
+      "grad_norm": 796.0271606445312,
+      "learning_rate": 6.292903164876097e-06,
+      "loss": 45.2027,
+      "step": 58790
+    },
+    {
+      "epoch": 0.23755944036167212,
+      "grad_norm": 669.234619140625,
+      "learning_rate": 6.291540713426206e-06,
+      "loss": 47.7954,
+      "step": 58800
+    },
+    {
+      "epoch": 0.23759984162703976,
+      "grad_norm": 907.8953857421875,
+      "learning_rate": 6.290178159214349e-06,
+      "loss": 54.457,
+      "step": 58810
+    },
+    {
+      "epoch": 0.2376402428924074,
+      "grad_norm": 587.7103881835938,
+      "learning_rate": 6.288815502348935e-06,
+      "loss": 36.6312,
+      "step": 58820
+    },
+    {
+      "epoch": 0.237680644157775,
+      "grad_norm": 594.3585815429688,
+      "learning_rate": 6.287452742938388e-06,
+      "loss": 48.9698,
+      "step": 58830
+    },
+    {
+      "epoch": 0.23772104542314265,
+      "grad_norm": 800.2922973632812,
+      "learning_rate": 6.286089881091134e-06,
+      "loss": 48.804,
+      "step": 58840
+    },
+    {
+      "epoch": 0.2377614466885103,
+      "grad_norm": 101.07825469970703,
+      "learning_rate": 6.284726916915611e-06,
+      "loss": 37.5152,
+      "step": 58850
+    },
+    {
+      "epoch": 0.2378018479538779,
+      "grad_norm": 333.7406005859375,
+      "learning_rate": 6.2833638505202635e-06,
+      "loss": 45.3329,
+      "step": 58860
+    },
+    {
+      "epoch": 0.23784224921924554,
+      "grad_norm": 613.3673095703125,
+      "learning_rate": 6.282000682013545e-06,
+      "loss": 45.9315,
+      "step": 58870
+    },
+    {
+      "epoch": 0.23788265048461318,
+      "grad_norm": 387.0548400878906,
+      "learning_rate": 6.280637411503913e-06,
+      "loss": 40.9148,
+      "step": 58880
+    },
+    {
+      "epoch": 0.2379230517499808,
+      "grad_norm": 867.9426879882812,
+      "learning_rate": 6.279274039099842e-06,
+      "loss": 30.7292,
+      "step": 58890
+    },
+    {
+      "epoch": 0.23796345301534844,
+      "grad_norm": 942.0332641601562,
+      "learning_rate": 6.277910564909806e-06,
+      "loss": 48.0671,
+      "step": 58900
+    },
+    {
+      "epoch": 0.23800385428071608,
+      "grad_norm": 1058.768798828125,
+      "learning_rate": 6.276546989042292e-06,
+      "loss": 49.5743,
+      "step": 58910
+    },
+    {
+      "epoch": 0.23804425554608372,
+      "grad_norm": 559.1909790039062,
+      "learning_rate": 6.275183311605793e-06,
+      "loss": 33.5974,
+      "step": 58920
+    },
+    {
+      "epoch": 0.23808465681145133,
+      "grad_norm": 642.1489868164062,
+      "learning_rate": 6.273819532708807e-06,
+      "loss": 47.9694,
+      "step": 58930
+    },
+    {
+      "epoch": 0.23812505807681897,
+      "grad_norm": 478.4878234863281,
+      "learning_rate": 6.27245565245985e-06,
+      "loss": 40.388,
+      "step": 58940
+    },
+    {
+      "epoch": 0.2381654593421866,
+      "grad_norm": 506.3686218261719,
+      "learning_rate": 6.271091670967437e-06,
+      "loss": 42.4386,
+      "step": 58950
+    },
+    {
+      "epoch": 0.23820586060755422,
+      "grad_norm": 899.9596557617188,
+      "learning_rate": 6.269727588340091e-06,
+      "loss": 39.7451,
+      "step": 58960
+    },
+    {
+      "epoch": 0.23824626187292186,
+      "grad_norm": 438.31927490234375,
+      "learning_rate": 6.268363404686348e-06,
+      "loss": 44.1548,
+      "step": 58970
+    },
+    {
+      "epoch": 0.2382866631382895,
+      "grad_norm": 428.0214538574219,
+      "learning_rate": 6.26699912011475e-06,
+      "loss": 43.6406,
+      "step": 58980
+    },
+    {
+      "epoch": 0.2383270644036571,
+      "grad_norm": 491.67291259765625,
+      "learning_rate": 6.265634734733848e-06,
+      "loss": 72.9044,
+      "step": 58990
+    },
+    {
+      "epoch": 0.23836746566902475,
+      "grad_norm": 1070.6533203125,
+      "learning_rate": 6.264270248652199e-06,
+      "loss": 61.3305,
+      "step": 59000
+    },
+    {
+      "epoch": 0.2384078669343924,
+      "grad_norm": 303.5849609375,
+      "learning_rate": 6.262905661978367e-06,
+      "loss": 44.1885,
+      "step": 59010
+    },
+    {
+      "epoch": 0.23844826819976,
+      "grad_norm": 756.376708984375,
+      "learning_rate": 6.261540974820928e-06,
+      "loss": 31.1812,
+      "step": 59020
+    },
+    {
+      "epoch": 0.23848866946512765,
+      "grad_norm": 976.0094604492188,
+      "learning_rate": 6.260176187288463e-06,
+      "loss": 34.7344,
+      "step": 59030
+    },
+    {
+      "epoch": 0.23852907073049529,
+      "grad_norm": 488.3143005371094,
+      "learning_rate": 6.2588112994895636e-06,
+      "loss": 47.71,
+      "step": 59040
+    },
+    {
+      "epoch": 0.2385694719958629,
+      "grad_norm": 646.7849731445312,
+      "learning_rate": 6.257446311532824e-06,
+      "loss": 43.175,
+      "step": 59050
+    },
+    {
+      "epoch": 0.23860987326123054,
+      "grad_norm": 594.6917114257812,
+      "learning_rate": 6.256081223526854e-06,
+      "loss": 34.8907,
+      "step": 59060
+    },
+    {
+      "epoch": 0.23865027452659818,
+      "grad_norm": 275.3476867675781,
+      "learning_rate": 6.254716035580264e-06,
+      "loss": 36.1814,
+      "step": 59070
+    },
+    {
+      "epoch": 0.23869067579196582,
+      "grad_norm": 343.4564514160156,
+      "learning_rate": 6.25335074780168e-06,
+      "loss": 54.2813,
+      "step": 59080
+    },
+    {
+      "epoch": 0.23873107705733343,
+      "grad_norm": 504.92205810546875,
+      "learning_rate": 6.251985360299728e-06,
+      "loss": 39.9313,
+      "step": 59090
+    },
+    {
+      "epoch": 0.23877147832270107,
+      "grad_norm": 794.7259521484375,
+      "learning_rate": 6.250619873183046e-06,
+      "loss": 52.7031,
+      "step": 59100
+    },
+    {
+      "epoch": 0.2388118795880687,
+      "grad_norm": 610.1290283203125,
+      "learning_rate": 6.249254286560281e-06,
+      "loss": 37.6571,
+      "step": 59110
+    },
+    {
+      "epoch": 0.23885228085343632,
+      "grad_norm": 411.7903137207031,
+      "learning_rate": 6.247888600540084e-06,
+      "loss": 72.4403,
+      "step": 59120
+    },
+    {
+      "epoch": 0.23889268211880396,
+      "grad_norm": 479.38494873046875,
+      "learning_rate": 6.246522815231121e-06,
+      "loss": 60.42,
+      "step": 59130
+    },
+    {
+      "epoch": 0.2389330833841716,
+      "grad_norm": 504.35345458984375,
+      "learning_rate": 6.245156930742057e-06,
+      "loss": 42.7403,
+      "step": 59140
+    },
+    {
+      "epoch": 0.23897348464953921,
+      "grad_norm": 419.8155517578125,
+      "learning_rate": 6.24379094718157e-06,
+      "loss": 52.5672,
+      "step": 59150
+    },
+    {
+      "epoch": 0.23901388591490685,
+      "grad_norm": 340.598388671875,
+      "learning_rate": 6.2424248646583455e-06,
+      "loss": 39.9453,
+      "step": 59160
+    },
+    {
+      "epoch": 0.2390542871802745,
+      "grad_norm": 142.64559936523438,
+      "learning_rate": 6.241058683281077e-06,
+      "loss": 44.7031,
+      "step": 59170
+    },
+    {
+      "epoch": 0.2390946884456421,
+      "grad_norm": 685.390380859375,
+      "learning_rate": 6.239692403158465e-06,
+      "loss": 27.1027,
+      "step": 59180
+    },
+    {
+      "epoch": 0.23913508971100975,
+      "grad_norm": 920.209228515625,
+      "learning_rate": 6.238326024399217e-06,
+      "loss": 49.9231,
+      "step": 59190
+    },
+    {
+      "epoch": 0.2391754909763774,
+      "grad_norm": 424.9049377441406,
+      "learning_rate": 6.236959547112051e-06,
+      "loss": 45.3293,
+      "step": 59200
+    },
+    {
+      "epoch": 0.239215892241745,
+      "grad_norm": 1177.8206787109375,
+      "learning_rate": 6.235592971405691e-06,
+      "loss": 46.2852,
+      "step": 59210
+    },
+    {
+      "epoch": 0.23925629350711264,
+      "grad_norm": 496.20458984375,
+      "learning_rate": 6.234226297388869e-06,
+      "loss": 39.4153,
+      "step": 59220
+    },
+    {
+      "epoch": 0.23929669477248028,
+      "grad_norm": 635.3526000976562,
+      "learning_rate": 6.232859525170324e-06,
+      "loss": 36.4867,
+      "step": 59230
+    },
+    {
+      "epoch": 0.23933709603784792,
+      "grad_norm": 697.213134765625,
+      "learning_rate": 6.231492654858805e-06,
+      "loss": 60.3189,
+      "step": 59240
+    },
+    {
+      "epoch": 0.23937749730321553,
+      "grad_norm": 347.87237548828125,
+      "learning_rate": 6.230125686563068e-06,
+      "loss": 52.1539,
+      "step": 59250
+    },
+    {
+      "epoch": 0.23941789856858317,
+      "grad_norm": 338.2275085449219,
+      "learning_rate": 6.2287586203918745e-06,
+      "loss": 35.6523,
+      "step": 59260
+    },
+    {
+      "epoch": 0.2394582998339508,
+      "grad_norm": 843.0849609375,
+      "learning_rate": 6.227391456453997e-06,
+      "loss": 37.2032,
+      "step": 59270
+    },
+    {
+      "epoch": 0.23949870109931842,
+      "grad_norm": 558.8504638671875,
+      "learning_rate": 6.226024194858214e-06,
+      "loss": 39.231,
+      "step": 59280
+    },
+    {
+      "epoch": 0.23953910236468606,
+      "grad_norm": 1432.7646484375,
+      "learning_rate": 6.224656835713313e-06,
+      "loss": 34.3905,
+      "step": 59290
+    },
+    {
+      "epoch": 0.2395795036300537,
+      "grad_norm": 1002.6559448242188,
+      "learning_rate": 6.223289379128088e-06,
+      "loss": 46.1089,
+      "step": 59300
+    },
+    {
+      "epoch": 0.23961990489542132,
+      "grad_norm": 232.8958740234375,
+      "learning_rate": 6.221921825211342e-06,
+      "loss": 51.2115,
+      "step": 59310
+    },
+    {
+      "epoch": 0.23966030616078896,
+      "grad_norm": 811.0142822265625,
+      "learning_rate": 6.220554174071884e-06,
+      "loss": 49.028,
+      "step": 59320
+    },
+    {
+      "epoch": 0.2397007074261566,
+      "grad_norm": 1509.3494873046875,
+      "learning_rate": 6.219186425818531e-06,
+      "loss": 50.5631,
+      "step": 59330
+    },
+    {
+      "epoch": 0.2397411086915242,
+      "grad_norm": 602.2036743164062,
+      "learning_rate": 6.217818580560111e-06,
+      "loss": 56.5959,
+      "step": 59340
+    },
+    {
+      "epoch": 0.23978150995689185,
+      "grad_norm": 474.7471618652344,
+      "learning_rate": 6.216450638405454e-06,
+      "loss": 43.5263,
+      "step": 59350
+    },
+    {
+      "epoch": 0.2398219112222595,
+      "grad_norm": 537.8904418945312,
+      "learning_rate": 6.2150825994634025e-06,
+      "loss": 43.5742,
+      "step": 59360
+    },
+    {
+      "epoch": 0.2398623124876271,
+      "grad_norm": 1052.5858154296875,
+      "learning_rate": 6.2137144638428045e-06,
+      "loss": 60.8518,
+      "step": 59370
+    },
+    {
+      "epoch": 0.23990271375299474,
+      "grad_norm": 774.1162719726562,
+      "learning_rate": 6.21234623165252e-06,
+      "loss": 40.4853,
+      "step": 59380
+    },
+    {
+      "epoch": 0.23994311501836238,
+      "grad_norm": 553.7500610351562,
+      "learning_rate": 6.210977903001406e-06,
+      "loss": 47.8429,
+      "step": 59390
+    },
+    {
+      "epoch": 0.23998351628373002,
+      "grad_norm": 465.7264709472656,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 51.5527,
+      "step": 59400
+    },
+    {
+      "epoch": 0.24002391754909763,
+      "grad_norm": 535.73779296875,
+      "learning_rate": 6.2082409567521975e-06,
+      "loss": 32.3945,
+      "step": 59410
+    },
+    {
+      "epoch": 0.24006431881446527,
+      "grad_norm": 346.88336181640625,
+      "learning_rate": 6.206872339371867e-06,
+      "loss": 44.4696,
+      "step": 59420
+    },
+    {
+      "epoch": 0.2401047200798329,
+      "grad_norm": 839.6456909179688,
+      "learning_rate": 6.205503625966247e-06,
+      "loss": 43.1945,
+      "step": 59430
+    },
+    {
+      "epoch": 0.24014512134520052,
+      "grad_norm": 419.2178039550781,
+      "learning_rate": 6.204134816644233e-06,
+      "loss": 44.3108,
+      "step": 59440
+    },
+    {
+      "epoch": 0.24018552261056816,
+      "grad_norm": 528.9591064453125,
+      "learning_rate": 6.2027659115147375e-06,
+      "loss": 30.9298,
+      "step": 59450
+    },
+    {
+      "epoch": 0.2402259238759358,
+      "grad_norm": 719.5377197265625,
+      "learning_rate": 6.201396910686679e-06,
+      "loss": 44.7331,
+      "step": 59460
+    },
+    {
+      "epoch": 0.24026632514130342,
+      "grad_norm": 370.1326599121094,
+      "learning_rate": 6.200027814268984e-06,
+      "loss": 31.5706,
+      "step": 59470
+    },
+    {
+      "epoch": 0.24030672640667106,
+      "grad_norm": 719.8978881835938,
+      "learning_rate": 6.198658622370582e-06,
+      "loss": 41.7919,
+      "step": 59480
+    },
+    {
+      "epoch": 0.2403471276720387,
+      "grad_norm": 727.5550537109375,
+      "learning_rate": 6.197289335100412e-06,
+      "loss": 46.8141,
+      "step": 59490
+    },
+    {
+      "epoch": 0.2403875289374063,
+      "grad_norm": 708.8463134765625,
+      "learning_rate": 6.195919952567426e-06,
+      "loss": 50.2198,
+      "step": 59500
+    },
+    {
+      "epoch": 0.24042793020277395,
+      "grad_norm": 681.5879516601562,
+      "learning_rate": 6.194550474880579e-06,
+      "loss": 48.9477,
+      "step": 59510
+    },
+    {
+      "epoch": 0.2404683314681416,
+      "grad_norm": 456.62774658203125,
+      "learning_rate": 6.193180902148833e-06,
+      "loss": 32.21,
+      "step": 59520
+    },
+    {
+      "epoch": 0.2405087327335092,
+      "grad_norm": 308.8553466796875,
+      "learning_rate": 6.1918112344811575e-06,
+      "loss": 48.589,
+      "step": 59530
+    },
+    {
+      "epoch": 0.24054913399887684,
+      "grad_norm": 372.4393005371094,
+      "learning_rate": 6.190441471986533e-06,
+      "loss": 45.7649,
+      "step": 59540
+    },
+    {
+      "epoch": 0.24058953526424448,
+      "grad_norm": 452.86187744140625,
+      "learning_rate": 6.18907161477394e-06,
+      "loss": 51.8513,
+      "step": 59550
+    },
+    {
+      "epoch": 0.24062993652961212,
+      "grad_norm": 409.22991943359375,
+      "learning_rate": 6.187701662952381e-06,
+      "loss": 32.4357,
+      "step": 59560
+    },
+    {
+      "epoch": 0.24067033779497973,
+      "grad_norm": 589.134521484375,
+      "learning_rate": 6.18633161663085e-06,
+      "loss": 51.7794,
+      "step": 59570
+    },
+    {
+      "epoch": 0.24071073906034737,
+      "grad_norm": 250.7372283935547,
+      "learning_rate": 6.184961475918355e-06,
+      "loss": 36.2709,
+      "step": 59580
+    },
+    {
+      "epoch": 0.240751140325715,
+      "grad_norm": 337.1333312988281,
+      "learning_rate": 6.183591240923914e-06,
+      "loss": 43.1022,
+      "step": 59590
+    },
+    {
+      "epoch": 0.24079154159108263,
+      "grad_norm": 515.5701293945312,
+      "learning_rate": 6.182220911756551e-06,
+      "loss": 36.8048,
+      "step": 59600
+    },
+    {
+      "epoch": 0.24083194285645027,
+      "grad_norm": 393.62774658203125,
+      "learning_rate": 6.1808504885252955e-06,
+      "loss": 58.8733,
+      "step": 59610
+    },
+    {
+      "epoch": 0.2408723441218179,
+      "grad_norm": 486.2384033203125,
+      "learning_rate": 6.179479971339186e-06,
+      "loss": 49.6456,
+      "step": 59620
+    },
+    {
+      "epoch": 0.24091274538718552,
+      "grad_norm": 244.34669494628906,
+      "learning_rate": 6.178109360307267e-06,
+      "loss": 35.7055,
+      "step": 59630
+    },
+    {
+      "epoch": 0.24095314665255316,
+      "grad_norm": 892.8306274414062,
+      "learning_rate": 6.176738655538594e-06,
+      "loss": 45.6613,
+      "step": 59640
+    },
+    {
+      "epoch": 0.2409935479179208,
+      "grad_norm": 196.5320281982422,
+      "learning_rate": 6.175367857142227e-06,
+      "loss": 46.8177,
+      "step": 59650
+    },
+    {
+      "epoch": 0.2410339491832884,
+      "grad_norm": 404.7183532714844,
+      "learning_rate": 6.173996965227234e-06,
+      "loss": 50.3921,
+      "step": 59660
+    },
+    {
+      "epoch": 0.24107435044865605,
+      "grad_norm": 592.3640747070312,
+      "learning_rate": 6.17262597990269e-06,
+      "loss": 36.2447,
+      "step": 59670
+    },
+    {
+      "epoch": 0.2411147517140237,
+      "grad_norm": 379.3549499511719,
+      "learning_rate": 6.171254901277678e-06,
+      "loss": 36.7108,
+      "step": 59680
+    },
+    {
+      "epoch": 0.2411551529793913,
+      "grad_norm": 521.3427734375,
+      "learning_rate": 6.169883729461289e-06,
+      "loss": 52.4567,
+      "step": 59690
+    },
+    {
+      "epoch": 0.24119555424475894,
+      "grad_norm": 661.0156860351562,
+      "learning_rate": 6.16851246456262e-06,
+      "loss": 28.96,
+      "step": 59700
+    },
+    {
+      "epoch": 0.24123595551012658,
+      "grad_norm": 732.6842651367188,
+      "learning_rate": 6.167141106690778e-06,
+      "loss": 55.4517,
+      "step": 59710
+    },
+    {
+      "epoch": 0.24127635677549422,
+      "grad_norm": 612.0284423828125,
+      "learning_rate": 6.1657696559548755e-06,
+      "loss": 49.9304,
+      "step": 59720
+    },
+    {
+      "epoch": 0.24131675804086183,
+      "grad_norm": 416.08673095703125,
+      "learning_rate": 6.16439811246403e-06,
+      "loss": 50.356,
+      "step": 59730
+    },
+    {
+      "epoch": 0.24135715930622947,
+      "grad_norm": 1435.585693359375,
+      "learning_rate": 6.163026476327371e-06,
+      "loss": 48.206,
+      "step": 59740
+    },
+    {
+      "epoch": 0.24139756057159711,
+      "grad_norm": 472.382568359375,
+      "learning_rate": 6.161654747654033e-06,
+      "loss": 40.12,
+      "step": 59750
+    },
+    {
+      "epoch": 0.24143796183696473,
+      "grad_norm": 170.5372314453125,
+      "learning_rate": 6.1602829265531585e-06,
+      "loss": 30.2096,
+      "step": 59760
+    },
+    {
+      "epoch": 0.24147836310233237,
+      "grad_norm": 429.81610107421875,
+      "learning_rate": 6.158911013133896e-06,
+      "loss": 54.4487,
+      "step": 59770
+    },
+    {
+      "epoch": 0.2415187643677,
+      "grad_norm": 324.32037353515625,
+      "learning_rate": 6.157539007505402e-06,
+      "loss": 62.2513,
+      "step": 59780
+    },
+    {
+      "epoch": 0.24155916563306762,
+      "grad_norm": 808.5761108398438,
+      "learning_rate": 6.156166909776842e-06,
+      "loss": 51.42,
+      "step": 59790
+    },
+    {
+      "epoch": 0.24159956689843526,
+      "grad_norm": 782.04296875,
+      "learning_rate": 6.154794720057388e-06,
+      "loss": 40.6022,
+      "step": 59800
+    },
+    {
+      "epoch": 0.2416399681638029,
+      "grad_norm": 1242.795654296875,
+      "learning_rate": 6.153422438456218e-06,
+      "loss": 47.4253,
+      "step": 59810
+    },
+    {
+      "epoch": 0.2416803694291705,
+      "grad_norm": 647.463623046875,
+      "learning_rate": 6.1520500650825175e-06,
+      "loss": 28.0886,
+      "step": 59820
+    },
+    {
+      "epoch": 0.24172077069453815,
+      "grad_norm": 2162.615234375,
+      "learning_rate": 6.150677600045479e-06,
+      "loss": 84.5893,
+      "step": 59830
+    },
+    {
+      "epoch": 0.2417611719599058,
+      "grad_norm": 999.169677734375,
+      "learning_rate": 6.1493050434543065e-06,
+      "loss": 48.0991,
+      "step": 59840
+    },
+    {
+      "epoch": 0.2418015732252734,
+      "grad_norm": 583.0239868164062,
+      "learning_rate": 6.1479323954182055e-06,
+      "loss": 44.8052,
+      "step": 59850
+    },
+    {
+      "epoch": 0.24184197449064104,
+      "grad_norm": 596.6414184570312,
+      "learning_rate": 6.146559656046394e-06,
+      "loss": 55.0618,
+      "step": 59860
+    },
+    {
+      "epoch": 0.24188237575600868,
+      "grad_norm": 850.3500366210938,
+      "learning_rate": 6.1451868254480914e-06,
+      "loss": 51.9113,
+      "step": 59870
+    },
+    {
+      "epoch": 0.24192277702137632,
+      "grad_norm": 854.9940185546875,
+      "learning_rate": 6.143813903732527e-06,
+      "loss": 45.9024,
+      "step": 59880
+    },
+    {
+      "epoch": 0.24196317828674394,
+      "grad_norm": 819.890869140625,
+      "learning_rate": 6.142440891008941e-06,
+      "loss": 53.9581,
+      "step": 59890
+    },
+    {
+      "epoch": 0.24200357955211158,
+      "grad_norm": 587.396728515625,
+      "learning_rate": 6.141067787386579e-06,
+      "loss": 54.0361,
+      "step": 59900
+    },
+    {
+      "epoch": 0.24204398081747922,
+      "grad_norm": 547.8096923828125,
+      "learning_rate": 6.139694592974687e-06,
+      "loss": 45.5808,
+      "step": 59910
+    },
+    {
+      "epoch": 0.24208438208284683,
+      "grad_norm": 634.807373046875,
+      "learning_rate": 6.1383213078825275e-06,
+      "loss": 58.0814,
+      "step": 59920
+    },
+    {
+      "epoch": 0.24212478334821447,
+      "grad_norm": 718.9409790039062,
+      "learning_rate": 6.136947932219365e-06,
+      "loss": 43.585,
+      "step": 59930
+    },
+    {
+      "epoch": 0.2421651846135821,
+      "grad_norm": 689.8276977539062,
+      "learning_rate": 6.135574466094475e-06,
+      "loss": 53.0661,
+      "step": 59940
+    },
+    {
+      "epoch": 0.24220558587894972,
+      "grad_norm": 813.9505615234375,
+      "learning_rate": 6.134200909617135e-06,
+      "loss": 48.7048,
+      "step": 59950
+    },
+    {
+      "epoch": 0.24224598714431736,
+      "grad_norm": 703.8113403320312,
+      "learning_rate": 6.132827262896634e-06,
+      "loss": 55.3374,
+      "step": 59960
+    },
+    {
+      "epoch": 0.242286388409685,
+      "grad_norm": 574.634765625,
+      "learning_rate": 6.131453526042267e-06,
+      "loss": 47.4171,
+      "step": 59970
+    },
+    {
+      "epoch": 0.2423267896750526,
+      "grad_norm": 1104.1204833984375,
+      "learning_rate": 6.130079699163335e-06,
+      "loss": 60.4575,
+      "step": 59980
+    },
+    {
+      "epoch": 0.24236719094042025,
+      "grad_norm": 377.7409973144531,
+      "learning_rate": 6.128705782369149e-06,
+      "loss": 54.446,
+      "step": 59990
+    },
+    {
+      "epoch": 0.2424075922057879,
+      "grad_norm": 792.7305908203125,
+      "learning_rate": 6.127331775769023e-06,
+      "loss": 69.839,
+      "step": 60000
+    },
+    {
+      "epoch": 0.2424479934711555,
+      "grad_norm": 1081.14404296875,
+      "learning_rate": 6.125957679472282e-06,
+      "loss": 37.0099,
+      "step": 60010
+    },
+    {
+      "epoch": 0.24248839473652314,
+      "grad_norm": 612.3765869140625,
+      "learning_rate": 6.124583493588254e-06,
+      "loss": 41.2812,
+      "step": 60020
+    },
+    {
+      "epoch": 0.24252879600189078,
+      "grad_norm": 482.0962219238281,
+      "learning_rate": 6.123209218226282e-06,
+      "loss": 56.972,
+      "step": 60030
+    },
+    {
+      "epoch": 0.24256919726725842,
+      "grad_norm": 410.4952392578125,
+      "learning_rate": 6.121834853495704e-06,
+      "loss": 31.3736,
+      "step": 60040
+    },
+    {
+      "epoch": 0.24260959853262604,
+      "grad_norm": 522.3032836914062,
+      "learning_rate": 6.120460399505876e-06,
+      "loss": 40.1662,
+      "step": 60050
+    },
+    {
+      "epoch": 0.24264999979799368,
+      "grad_norm": 473.8763122558594,
+      "learning_rate": 6.119085856366158e-06,
+      "loss": 45.7846,
+      "step": 60060
+    },
+    {
+      "epoch": 0.24269040106336132,
+      "grad_norm": 537.3663940429688,
+      "learning_rate": 6.117711224185913e-06,
+      "loss": 58.7338,
+      "step": 60070
+    },
+    {
+      "epoch": 0.24273080232872893,
+      "grad_norm": 577.8731689453125,
+      "learning_rate": 6.116336503074516e-06,
+      "loss": 50.324,
+      "step": 60080
+    },
+    {
+      "epoch": 0.24277120359409657,
+      "grad_norm": 450.7883605957031,
+      "learning_rate": 6.114961693141346e-06,
+      "loss": 48.6675,
+      "step": 60090
+    },
+    {
+      "epoch": 0.2428116048594642,
+      "grad_norm": 739.99169921875,
+      "learning_rate": 6.113586794495792e-06,
+      "loss": 46.6423,
+      "step": 60100
+    },
+    {
+      "epoch": 0.24285200612483182,
+      "grad_norm": 830.27880859375,
+      "learning_rate": 6.112211807247246e-06,
+      "loss": 42.0815,
+      "step": 60110
+    },
+    {
+      "epoch": 0.24289240739019946,
+      "grad_norm": 787.69091796875,
+      "learning_rate": 6.110836731505112e-06,
+      "loss": 61.397,
+      "step": 60120
+    },
+    {
+      "epoch": 0.2429328086555671,
+      "grad_norm": 335.32855224609375,
+      "learning_rate": 6.109461567378796e-06,
+      "loss": 43.3532,
+      "step": 60130
+    },
+    {
+      "epoch": 0.2429732099209347,
+      "grad_norm": 610.6149291992188,
+      "learning_rate": 6.108086314977717e-06,
+      "loss": 44.857,
+      "step": 60140
+    },
+    {
+      "epoch": 0.24301361118630235,
+      "grad_norm": 545.5875244140625,
+      "learning_rate": 6.106710974411294e-06,
+      "loss": 51.1943,
+      "step": 60150
+    },
+    {
+      "epoch": 0.24305401245167,
+      "grad_norm": 548.0823974609375,
+      "learning_rate": 6.105335545788957e-06,
+      "loss": 40.8381,
+      "step": 60160
+    },
+    {
+      "epoch": 0.2430944137170376,
+      "grad_norm": 871.58251953125,
+      "learning_rate": 6.103960029220145e-06,
+      "loss": 42.5754,
+      "step": 60170
+    },
+    {
+      "epoch": 0.24313481498240525,
+      "grad_norm": 591.369140625,
+      "learning_rate": 6.102584424814299e-06,
+      "loss": 43.5635,
+      "step": 60180
+    },
+    {
+      "epoch": 0.24317521624777289,
+      "grad_norm": 768.7680053710938,
+      "learning_rate": 6.101208732680872e-06,
+      "loss": 52.6281,
+      "step": 60190
+    },
+    {
+      "epoch": 0.24321561751314053,
+      "grad_norm": 489.9885559082031,
+      "learning_rate": 6.09983295292932e-06,
+      "loss": 58.9169,
+      "step": 60200
+    },
+    {
+      "epoch": 0.24325601877850814,
+      "grad_norm": 323.9236145019531,
+      "learning_rate": 6.0984570856691046e-06,
+      "loss": 31.9827,
+      "step": 60210
+    },
+    {
+      "epoch": 0.24329642004387578,
+      "grad_norm": 422.1755676269531,
+      "learning_rate": 6.097081131009703e-06,
+      "loss": 29.6285,
+      "step": 60220
+    },
+    {
+      "epoch": 0.24333682130924342,
+      "grad_norm": 323.1979064941406,
+      "learning_rate": 6.095705089060589e-06,
+      "loss": 42.9744,
+      "step": 60230
+    },
+    {
+      "epoch": 0.24337722257461103,
+      "grad_norm": 1390.3934326171875,
+      "learning_rate": 6.094328959931252e-06,
+      "loss": 45.7392,
+      "step": 60240
+    },
+    {
+      "epoch": 0.24341762383997867,
+      "grad_norm": 841.0800170898438,
+      "learning_rate": 6.092952743731179e-06,
+      "loss": 48.7019,
+      "step": 60250
+    },
+    {
+      "epoch": 0.2434580251053463,
+      "grad_norm": 698.6349487304688,
+      "learning_rate": 6.091576440569873e-06,
+      "loss": 58.3474,
+      "step": 60260
+    },
+    {
+      "epoch": 0.24349842637071392,
+      "grad_norm": 622.3505859375,
+      "learning_rate": 6.09020005055684e-06,
+      "loss": 35.8427,
+      "step": 60270
+    },
+    {
+      "epoch": 0.24353882763608156,
+      "grad_norm": 661.6207885742188,
+      "learning_rate": 6.088823573801592e-06,
+      "loss": 56.9386,
+      "step": 60280
+    },
+    {
+      "epoch": 0.2435792289014492,
+      "grad_norm": 241.02731323242188,
+      "learning_rate": 6.087447010413651e-06,
+      "loss": 33.8124,
+      "step": 60290
+    },
+    {
+      "epoch": 0.24361963016681681,
+      "grad_norm": 544.45263671875,
+      "learning_rate": 6.08607036050254e-06,
+      "loss": 42.6043,
+      "step": 60300
+    },
+    {
+      "epoch": 0.24366003143218445,
+      "grad_norm": 353.69683837890625,
+      "learning_rate": 6.084693624177794e-06,
+      "loss": 31.6919,
+      "step": 60310
+    },
+    {
+      "epoch": 0.2437004326975521,
+      "grad_norm": 566.5994873046875,
+      "learning_rate": 6.083316801548956e-06,
+      "loss": 37.8967,
+      "step": 60320
+    },
+    {
+      "epoch": 0.2437408339629197,
+      "grad_norm": 383.8753967285156,
+      "learning_rate": 6.081939892725572e-06,
+      "loss": 46.3118,
+      "step": 60330
+    },
+    {
+      "epoch": 0.24378123522828735,
+      "grad_norm": 634.250244140625,
+      "learning_rate": 6.080562897817196e-06,
+      "loss": 37.6377,
+      "step": 60340
+    },
+    {
+      "epoch": 0.243821636493655,
+      "grad_norm": 595.0110473632812,
+      "learning_rate": 6.079185816933388e-06,
+      "loss": 39.7675,
+      "step": 60350
+    },
+    {
+      "epoch": 0.24386203775902263,
+      "grad_norm": 784.342041015625,
+      "learning_rate": 6.077808650183718e-06,
+      "loss": 46.149,
+      "step": 60360
+    },
+    {
+      "epoch": 0.24390243902439024,
+      "grad_norm": 654.6844482421875,
+      "learning_rate": 6.076431397677762e-06,
+      "loss": 46.0708,
+      "step": 60370
+    },
+    {
+      "epoch": 0.24394284028975788,
+      "grad_norm": 735.3973388671875,
+      "learning_rate": 6.0750540595250986e-06,
+      "loss": 40.0463,
+      "step": 60380
+    },
+    {
+      "epoch": 0.24398324155512552,
+      "grad_norm": 690.0650634765625,
+      "learning_rate": 6.073676635835317e-06,
+      "loss": 37.8187,
+      "step": 60390
+    },
+    {
+      "epoch": 0.24402364282049313,
+      "grad_norm": 479.63134765625,
+      "learning_rate": 6.072299126718012e-06,
+      "loss": 38.4419,
+      "step": 60400
+    },
+    {
+      "epoch": 0.24406404408586077,
+      "grad_norm": 654.6773071289062,
+      "learning_rate": 6.070921532282788e-06,
+      "loss": 34.0201,
+      "step": 60410
+    },
+    {
+      "epoch": 0.2441044453512284,
+      "grad_norm": 833.0028686523438,
+      "learning_rate": 6.0695438526392536e-06,
+      "loss": 49.9405,
+      "step": 60420
+    },
+    {
+      "epoch": 0.24414484661659602,
+      "grad_norm": 768.3114013671875,
+      "learning_rate": 6.068166087897022e-06,
+      "loss": 48.4438,
+      "step": 60430
+    },
+    {
+      "epoch": 0.24418524788196366,
+      "grad_norm": 539.3959350585938,
+      "learning_rate": 6.066788238165717e-06,
+      "loss": 45.1048,
+      "step": 60440
+    },
+    {
+      "epoch": 0.2442256491473313,
+      "grad_norm": 413.73114013671875,
+      "learning_rate": 6.0654103035549686e-06,
+      "loss": 50.7065,
+      "step": 60450
+    },
+    {
+      "epoch": 0.24426605041269892,
+      "grad_norm": 666.993408203125,
+      "learning_rate": 6.064032284174411e-06,
+      "loss": 40.3299,
+      "step": 60460
+    },
+    {
+      "epoch": 0.24430645167806656,
+      "grad_norm": 420.68548583984375,
+      "learning_rate": 6.062654180133689e-06,
+      "loss": 34.3669,
+      "step": 60470
+    },
+    {
+      "epoch": 0.2443468529434342,
+      "grad_norm": 687.9544067382812,
+      "learning_rate": 6.06127599154245e-06,
+      "loss": 38.3136,
+      "step": 60480
+    },
+    {
+      "epoch": 0.2443872542088018,
+      "grad_norm": 524.0534057617188,
+      "learning_rate": 6.059897718510351e-06,
+      "loss": 39.4131,
+      "step": 60490
+    },
+    {
+      "epoch": 0.24442765547416945,
+      "grad_norm": 639.9952392578125,
+      "learning_rate": 6.058519361147055e-06,
+      "loss": 41.8146,
+      "step": 60500
+    },
+    {
+      "epoch": 0.2444680567395371,
+      "grad_norm": 392.12811279296875,
+      "learning_rate": 6.057140919562231e-06,
+      "loss": 23.9516,
+      "step": 60510
+    },
+    {
+      "epoch": 0.24450845800490473,
+      "grad_norm": 813.45849609375,
+      "learning_rate": 6.055762393865555e-06,
+      "loss": 45.8369,
+      "step": 60520
+    },
+    {
+      "epoch": 0.24454885927027234,
+      "grad_norm": 1308.13671875,
+      "learning_rate": 6.054383784166712e-06,
+      "loss": 43.4649,
+      "step": 60530
+    },
+    {
+      "epoch": 0.24458926053563998,
+      "grad_norm": 1111.109130859375,
+      "learning_rate": 6.05300509057539e-06,
+      "loss": 61.8905,
+      "step": 60540
+    },
+    {
+      "epoch": 0.24462966180100762,
+      "grad_norm": 1149.5960693359375,
+      "learning_rate": 6.051626313201285e-06,
+      "loss": 53.8758,
+      "step": 60550
+    },
+    {
+      "epoch": 0.24467006306637523,
+      "grad_norm": 603.8845825195312,
+      "learning_rate": 6.0502474521541014e-06,
+      "loss": 53.2023,
+      "step": 60560
+    },
+    {
+      "epoch": 0.24471046433174287,
+      "grad_norm": 435.4098205566406,
+      "learning_rate": 6.048868507543547e-06,
+      "loss": 24.9156,
+      "step": 60570
+    },
+    {
+      "epoch": 0.2447508655971105,
+      "grad_norm": 702.36083984375,
+      "learning_rate": 6.047489479479339e-06,
+      "loss": 30.666,
+      "step": 60580
+    },
+    {
+      "epoch": 0.24479126686247812,
+      "grad_norm": 482.1972351074219,
+      "learning_rate": 6.046110368071201e-06,
+      "loss": 45.1941,
+      "step": 60590
+    },
+    {
+      "epoch": 0.24483166812784576,
+      "grad_norm": 290.9931945800781,
+      "learning_rate": 6.044731173428862e-06,
+      "loss": 23.8702,
+      "step": 60600
+    },
+    {
+      "epoch": 0.2448720693932134,
+      "grad_norm": 466.46673583984375,
+      "learning_rate": 6.043351895662059e-06,
+      "loss": 42.5676,
+      "step": 60610
+    },
+    {
+      "epoch": 0.24491247065858102,
+      "grad_norm": 490.4240417480469,
+      "learning_rate": 6.041972534880533e-06,
+      "loss": 21.8341,
+      "step": 60620
+    },
+    {
+      "epoch": 0.24495287192394866,
+      "grad_norm": 286.0077209472656,
+      "learning_rate": 6.040593091194035e-06,
+      "loss": 59.126,
+      "step": 60630
+    },
+    {
+      "epoch": 0.2449932731893163,
+      "grad_norm": 332.6376953125,
+      "learning_rate": 6.039213564712319e-06,
+      "loss": 50.8156,
+      "step": 60640
+    },
+    {
+      "epoch": 0.2450336744546839,
+      "grad_norm": 509.3329162597656,
+      "learning_rate": 6.03783395554515e-06,
+      "loss": 36.6754,
+      "step": 60650
+    },
+    {
+      "epoch": 0.24507407572005155,
+      "grad_norm": 362.27398681640625,
+      "learning_rate": 6.036454263802297e-06,
+      "loss": 24.3201,
+      "step": 60660
+    },
+    {
+      "epoch": 0.2451144769854192,
+      "grad_norm": 482.9433288574219,
+      "learning_rate": 6.035074489593536e-06,
+      "loss": 51.8037,
+      "step": 60670
+    },
+    {
+      "epoch": 0.24515487825078683,
+      "grad_norm": 437.08489990234375,
+      "learning_rate": 6.033694633028644e-06,
+      "loss": 43.0535,
+      "step": 60680
+    },
+    {
+      "epoch": 0.24519527951615444,
+      "grad_norm": 999.8873901367188,
+      "learning_rate": 6.032314694217416e-06,
+      "loss": 51.4657,
+      "step": 60690
+    },
+    {
+      "epoch": 0.24523568078152208,
+      "grad_norm": 1049.546142578125,
+      "learning_rate": 6.030934673269646e-06,
+      "loss": 43.613,
+      "step": 60700
+    },
+    {
+      "epoch": 0.24527608204688972,
+      "grad_norm": 786.54150390625,
+      "learning_rate": 6.029554570295135e-06,
+      "loss": 40.1556,
+      "step": 60710
+    },
+    {
+      "epoch": 0.24531648331225733,
+      "grad_norm": 773.1795043945312,
+      "learning_rate": 6.028174385403693e-06,
+      "loss": 56.3238,
+      "step": 60720
+    },
+    {
+      "epoch": 0.24535688457762497,
+      "grad_norm": 431.96539306640625,
+      "learning_rate": 6.026794118705133e-06,
+      "loss": 56.1787,
+      "step": 60730
+    },
+    {
+      "epoch": 0.2453972858429926,
+      "grad_norm": 789.9153442382812,
+      "learning_rate": 6.025413770309278e-06,
+      "loss": 53.2711,
+      "step": 60740
+    },
+    {
+      "epoch": 0.24543768710836023,
+      "grad_norm": 771.6865234375,
+      "learning_rate": 6.024033340325954e-06,
+      "loss": 45.3818,
+      "step": 60750
+    },
+    {
+      "epoch": 0.24547808837372787,
+      "grad_norm": 642.1569213867188,
+      "learning_rate": 6.022652828864999e-06,
+      "loss": 36.2313,
+      "step": 60760
+    },
+    {
+      "epoch": 0.2455184896390955,
+      "grad_norm": 1129.879638671875,
+      "learning_rate": 6.0212722360362496e-06,
+      "loss": 67.5433,
+      "step": 60770
+    },
+    {
+      "epoch": 0.24555889090446312,
+      "grad_norm": 652.3465576171875,
+      "learning_rate": 6.019891561949554e-06,
+      "loss": 44.3988,
+      "step": 60780
+    },
+    {
+      "epoch": 0.24559929216983076,
+      "grad_norm": 691.905517578125,
+      "learning_rate": 6.01851080671477e-06,
+      "loss": 43.6003,
+      "step": 60790
+    },
+    {
+      "epoch": 0.2456396934351984,
+      "grad_norm": 836.9973754882812,
+      "learning_rate": 6.017129970441756e-06,
+      "loss": 41.0295,
+      "step": 60800
+    },
+    {
+      "epoch": 0.245680094700566,
+      "grad_norm": 537.411376953125,
+      "learning_rate": 6.015749053240378e-06,
+      "loss": 35.7805,
+      "step": 60810
+    },
+    {
+      "epoch": 0.24572049596593365,
+      "grad_norm": 811.6790771484375,
+      "learning_rate": 6.0143680552205075e-06,
+      "loss": 44.9824,
+      "step": 60820
+    },
+    {
+      "epoch": 0.2457608972313013,
+      "grad_norm": 748.29541015625,
+      "learning_rate": 6.012986976492025e-06,
+      "loss": 53.0722,
+      "step": 60830
+    },
+    {
+      "epoch": 0.2458012984966689,
+      "grad_norm": 752.9380493164062,
+      "learning_rate": 6.011605817164822e-06,
+      "loss": 44.365,
+      "step": 60840
+    },
+    {
+      "epoch": 0.24584169976203654,
+      "grad_norm": 539.8973999023438,
+      "learning_rate": 6.0102245773487855e-06,
+      "loss": 38.9263,
+      "step": 60850
+    },
+    {
+      "epoch": 0.24588210102740418,
+      "grad_norm": 748.1033325195312,
+      "learning_rate": 6.008843257153815e-06,
+      "loss": 52.9087,
+      "step": 60860
+    },
+    {
+      "epoch": 0.24592250229277182,
+      "grad_norm": 397.9111328125,
+      "learning_rate": 6.007461856689815e-06,
+      "loss": 38.5012,
+      "step": 60870
+    },
+    {
+      "epoch": 0.24596290355813943,
+      "grad_norm": 532.90771484375,
+      "learning_rate": 6.0060803760667e-06,
+      "loss": 53.9542,
+      "step": 60880
+    },
+    {
+      "epoch": 0.24600330482350707,
+      "grad_norm": 331.3194885253906,
+      "learning_rate": 6.004698815394389e-06,
+      "loss": 55.5145,
+      "step": 60890
+    },
+    {
+      "epoch": 0.24604370608887471,
+      "grad_norm": 1173.3389892578125,
+      "learning_rate": 6.003317174782801e-06,
+      "loss": 59.5596,
+      "step": 60900
+    },
+    {
+      "epoch": 0.24608410735424233,
+      "grad_norm": 390.91949462890625,
+      "learning_rate": 6.001935454341872e-06,
+      "loss": 42.8467,
+      "step": 60910
+    },
+    {
+      "epoch": 0.24612450861960997,
+      "grad_norm": 771.1336669921875,
+      "learning_rate": 6.000553654181536e-06,
+      "loss": 36.5511,
+      "step": 60920
+    },
+    {
+      "epoch": 0.2461649098849776,
+      "grad_norm": 504.6431884765625,
+      "learning_rate": 5.999171774411737e-06,
+      "loss": 48.6999,
+      "step": 60930
+    },
+    {
+      "epoch": 0.24620531115034522,
+      "grad_norm": 1690.32568359375,
+      "learning_rate": 5.997789815142427e-06,
+      "loss": 46.5689,
+      "step": 60940
+    },
+    {
+      "epoch": 0.24624571241571286,
+      "grad_norm": 435.9308776855469,
+      "learning_rate": 5.99640777648356e-06,
+      "loss": 59.5126,
+      "step": 60950
+    },
+    {
+      "epoch": 0.2462861136810805,
+      "grad_norm": 508.0345458984375,
+      "learning_rate": 5.9950256585450995e-06,
+      "loss": 42.083,
+      "step": 60960
+    },
+    {
+      "epoch": 0.2463265149464481,
+      "grad_norm": 600.52392578125,
+      "learning_rate": 5.993643461437013e-06,
+      "loss": 37.8109,
+      "step": 60970
+    },
+    {
+      "epoch": 0.24636691621181575,
+      "grad_norm": 456.70550537109375,
+      "learning_rate": 5.992261185269278e-06,
+      "loss": 45.6742,
+      "step": 60980
+    },
+    {
+      "epoch": 0.2464073174771834,
+      "grad_norm": 373.98712158203125,
+      "learning_rate": 5.990878830151873e-06,
+      "loss": 38.2048,
+      "step": 60990
+    },
+    {
+      "epoch": 0.246447718742551,
+      "grad_norm": 192.475341796875,
+      "learning_rate": 5.989496396194787e-06,
+      "loss": 43.5954,
+      "step": 61000
+    },
+    {
+      "epoch": 0.24648812000791864,
+      "grad_norm": 836.5695190429688,
+      "learning_rate": 5.988113883508016e-06,
+      "loss": 79.2422,
+      "step": 61010
+    },
+    {
+      "epoch": 0.24652852127328628,
+      "grad_norm": 487.8687744140625,
+      "learning_rate": 5.986731292201555e-06,
+      "loss": 51.7228,
+      "step": 61020
+    },
+    {
+      "epoch": 0.24656892253865392,
+      "grad_norm": 342.183349609375,
+      "learning_rate": 5.985348622385415e-06,
+      "loss": 33.0495,
+      "step": 61030
+    },
+    {
+      "epoch": 0.24660932380402154,
+      "grad_norm": 1103.4874267578125,
+      "learning_rate": 5.9839658741696085e-06,
+      "loss": 64.6234,
+      "step": 61040
+    },
+    {
+      "epoch": 0.24664972506938918,
+      "grad_norm": 683.1010131835938,
+      "learning_rate": 5.982583047664151e-06,
+      "loss": 35.5353,
+      "step": 61050
+    },
+    {
+      "epoch": 0.24669012633475682,
+      "grad_norm": 318.55364990234375,
+      "learning_rate": 5.981200142979071e-06,
+      "loss": 34.0901,
+      "step": 61060
+    },
+    {
+      "epoch": 0.24673052760012443,
+      "grad_norm": 737.1567993164062,
+      "learning_rate": 5.9798171602244e-06,
+      "loss": 40.2012,
+      "step": 61070
+    },
+    {
+      "epoch": 0.24677092886549207,
+      "grad_norm": 692.0882568359375,
+      "learning_rate": 5.978434099510172e-06,
+      "loss": 47.4886,
+      "step": 61080
+    },
+    {
+      "epoch": 0.2468113301308597,
+      "grad_norm": 655.0506591796875,
+      "learning_rate": 5.977050960946433e-06,
+      "loss": 49.641,
+      "step": 61090
+    },
+    {
+      "epoch": 0.24685173139622732,
+      "grad_norm": 564.3485717773438,
+      "learning_rate": 5.975667744643235e-06,
+      "loss": 44.6221,
+      "step": 61100
+    },
+    {
+      "epoch": 0.24689213266159496,
+      "grad_norm": 896.7650146484375,
+      "learning_rate": 5.974284450710631e-06,
+      "loss": 56.5591,
+      "step": 61110
+    },
+    {
+      "epoch": 0.2469325339269626,
+      "grad_norm": 1133.868896484375,
+      "learning_rate": 5.972901079258685e-06,
+      "loss": 55.2751,
+      "step": 61120
+    },
+    {
+      "epoch": 0.2469729351923302,
+      "grad_norm": 467.185546875,
+      "learning_rate": 5.971517630397465e-06,
+      "loss": 45.7085,
+      "step": 61130
+    },
+    {
+      "epoch": 0.24701333645769785,
+      "grad_norm": 87.59776306152344,
+      "learning_rate": 5.970134104237046e-06,
+      "loss": 51.0002,
+      "step": 61140
+    },
+    {
+      "epoch": 0.2470537377230655,
+      "grad_norm": 475.14825439453125,
+      "learning_rate": 5.96875050088751e-06,
+      "loss": 50.5819,
+      "step": 61150
+    },
+    {
+      "epoch": 0.2470941389884331,
+      "grad_norm": 673.6754150390625,
+      "learning_rate": 5.9673668204589396e-06,
+      "loss": 51.5235,
+      "step": 61160
+    },
+    {
+      "epoch": 0.24713454025380074,
+      "grad_norm": 265.7752380371094,
+      "learning_rate": 5.965983063061432e-06,
+      "loss": 53.3672,
+      "step": 61170
+    },
+    {
+      "epoch": 0.24717494151916838,
+      "grad_norm": 732.6128540039062,
+      "learning_rate": 5.964599228805087e-06,
+      "loss": 49.7361,
+      "step": 61180
+    },
+    {
+      "epoch": 0.24721534278453602,
+      "grad_norm": 686.7088012695312,
+      "learning_rate": 5.963215317800008e-06,
+      "loss": 46.626,
+      "step": 61190
+    },
+    {
+      "epoch": 0.24725574404990364,
+      "grad_norm": 532.4640502929688,
+      "learning_rate": 5.961831330156306e-06,
+      "loss": 37.8897,
+      "step": 61200
+    },
+    {
+      "epoch": 0.24729614531527128,
+      "grad_norm": 495.2468566894531,
+      "learning_rate": 5.960447265984098e-06,
+      "loss": 45.9627,
+      "step": 61210
+    },
+    {
+      "epoch": 0.24733654658063892,
+      "grad_norm": 328.6434631347656,
+      "learning_rate": 5.95906312539351e-06,
+      "loss": 39.4499,
+      "step": 61220
+    },
+    {
+      "epoch": 0.24737694784600653,
+      "grad_norm": 836.5567626953125,
+      "learning_rate": 5.9576789084946705e-06,
+      "loss": 52.5922,
+      "step": 61230
+    },
+    {
+      "epoch": 0.24741734911137417,
+      "grad_norm": 379.64019775390625,
+      "learning_rate": 5.956294615397716e-06,
+      "loss": 52.1735,
+      "step": 61240
+    },
+    {
+      "epoch": 0.2474577503767418,
+      "grad_norm": 545.490966796875,
+      "learning_rate": 5.954910246212787e-06,
+      "loss": 43.6922,
+      "step": 61250
+    },
+    {
+      "epoch": 0.24749815164210942,
+      "grad_norm": 1560.308349609375,
+      "learning_rate": 5.953525801050032e-06,
+      "loss": 46.3509,
+      "step": 61260
+    },
+    {
+      "epoch": 0.24753855290747706,
+      "grad_norm": 458.22314453125,
+      "learning_rate": 5.952141280019605e-06,
+      "loss": 46.8482,
+      "step": 61270
+    },
+    {
+      "epoch": 0.2475789541728447,
+      "grad_norm": 852.26953125,
+      "learning_rate": 5.950756683231667e-06,
+      "loss": 49.4784,
+      "step": 61280
+    },
+    {
+      "epoch": 0.2476193554382123,
+      "grad_norm": 506.84912109375,
+      "learning_rate": 5.949372010796384e-06,
+      "loss": 37.7671,
+      "step": 61290
+    },
+    {
+      "epoch": 0.24765975670357995,
+      "grad_norm": 221.2855224609375,
+      "learning_rate": 5.947987262823924e-06,
+      "loss": 32.4322,
+      "step": 61300
+    },
+    {
+      "epoch": 0.2477001579689476,
+      "grad_norm": 794.8474731445312,
+      "learning_rate": 5.94660243942447e-06,
+      "loss": 44.5728,
+      "step": 61310
+    },
+    {
+      "epoch": 0.2477405592343152,
+      "grad_norm": 841.0701293945312,
+      "learning_rate": 5.945217540708206e-06,
+      "loss": 67.0786,
+      "step": 61320
+    },
+    {
+      "epoch": 0.24778096049968285,
+      "grad_norm": 450.19146728515625,
+      "learning_rate": 5.9438325667853185e-06,
+      "loss": 44.422,
+      "step": 61330
+    },
+    {
+      "epoch": 0.24782136176505049,
+      "grad_norm": 431.77667236328125,
+      "learning_rate": 5.942447517766005e-06,
+      "loss": 46.848,
+      "step": 61340
+    },
+    {
+      "epoch": 0.24786176303041813,
+      "grad_norm": 796.6408081054688,
+      "learning_rate": 5.941062393760467e-06,
+      "loss": 67.5153,
+      "step": 61350
+    },
+    {
+      "epoch": 0.24790216429578574,
+      "grad_norm": 417.2357177734375,
+      "learning_rate": 5.939677194878915e-06,
+      "loss": 36.6902,
+      "step": 61360
+    },
+    {
+      "epoch": 0.24794256556115338,
+      "grad_norm": 501.64276123046875,
+      "learning_rate": 5.93829192123156e-06,
+      "loss": 36.0204,
+      "step": 61370
+    },
+    {
+      "epoch": 0.24798296682652102,
+      "grad_norm": 1153.709228515625,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 53.7668,
+      "step": 61380
+    },
+    {
+      "epoch": 0.24802336809188863,
+      "grad_norm": 445.2611999511719,
+      "learning_rate": 5.935521150080331e-06,
+      "loss": 35.9362,
+      "step": 61390
+    },
+    {
+      "epoch": 0.24806376935725627,
+      "grad_norm": 768.938720703125,
+      "learning_rate": 5.934135652796914e-06,
+      "loss": 62.0834,
+      "step": 61400
+    },
+    {
+      "epoch": 0.2481041706226239,
+      "grad_norm": 351.4441833496094,
+      "learning_rate": 5.9327500811886095e-06,
+      "loss": 46.8324,
+      "step": 61410
+    },
+    {
+      "epoch": 0.24814457188799152,
+      "grad_norm": 872.1058959960938,
+      "learning_rate": 5.931364435365663e-06,
+      "loss": 32.937,
+      "step": 61420
+    },
+    {
+      "epoch": 0.24818497315335916,
+      "grad_norm": 1003.3562622070312,
+      "learning_rate": 5.929978715438322e-06,
+      "loss": 50.35,
+      "step": 61430
+    },
+    {
+      "epoch": 0.2482253744187268,
+      "grad_norm": 1165.629150390625,
+      "learning_rate": 5.928592921516843e-06,
+      "loss": 54.3017,
+      "step": 61440
+    },
+    {
+      "epoch": 0.24826577568409441,
+      "grad_norm": 402.1977844238281,
+      "learning_rate": 5.9272070537114855e-06,
+      "loss": 43.1218,
+      "step": 61450
+    },
+    {
+      "epoch": 0.24830617694946205,
+      "grad_norm": 525.96875,
+      "learning_rate": 5.92582111213252e-06,
+      "loss": 42.6614,
+      "step": 61460
+    },
+    {
+      "epoch": 0.2483465782148297,
+      "grad_norm": 894.915283203125,
+      "learning_rate": 5.924435096890216e-06,
+      "loss": 38.2639,
+      "step": 61470
+    },
+    {
+      "epoch": 0.2483869794801973,
+      "grad_norm": 677.1754760742188,
+      "learning_rate": 5.923049008094855e-06,
+      "loss": 34.2235,
+      "step": 61480
+    },
+    {
+      "epoch": 0.24842738074556495,
+      "grad_norm": 265.16180419921875,
+      "learning_rate": 5.921662845856719e-06,
+      "loss": 25.4316,
+      "step": 61490
+    },
+    {
+      "epoch": 0.2484677820109326,
+      "grad_norm": 627.289794921875,
+      "learning_rate": 5.920276610286102e-06,
+      "loss": 52.2437,
+      "step": 61500
+    },
+    {
+      "epoch": 0.24850818327630023,
+      "grad_norm": 433.43951416015625,
+      "learning_rate": 5.918890301493299e-06,
+      "loss": 42.8496,
+      "step": 61510
+    },
+    {
+      "epoch": 0.24854858454166784,
+      "grad_norm": 684.3920288085938,
+      "learning_rate": 5.91750391958861e-06,
+      "loss": 55.9738,
+      "step": 61520
+    },
+    {
+      "epoch": 0.24858898580703548,
+      "grad_norm": 277.8285217285156,
+      "learning_rate": 5.916117464682346e-06,
+      "loss": 57.1054,
+      "step": 61530
+    },
+    {
+      "epoch": 0.24862938707240312,
+      "grad_norm": 539.8350830078125,
+      "learning_rate": 5.914730936884819e-06,
+      "loss": 47.499,
+      "step": 61540
+    },
+    {
+      "epoch": 0.24866978833777073,
+      "grad_norm": 364.66021728515625,
+      "learning_rate": 5.91334433630635e-06,
+      "loss": 31.1749,
+      "step": 61550
+    },
+    {
+      "epoch": 0.24871018960313837,
+      "grad_norm": 655.7823486328125,
+      "learning_rate": 5.911957663057264e-06,
+      "loss": 64.6635,
+      "step": 61560
+    },
+    {
+      "epoch": 0.248750590868506,
+      "grad_norm": 418.73699951171875,
+      "learning_rate": 5.910570917247892e-06,
+      "loss": 31.336,
+      "step": 61570
+    },
+    {
+      "epoch": 0.24879099213387362,
+      "grad_norm": 576.4503173828125,
+      "learning_rate": 5.909184098988571e-06,
+      "loss": 40.6577,
+      "step": 61580
+    },
+    {
+      "epoch": 0.24883139339924126,
+      "grad_norm": 576.8017578125,
+      "learning_rate": 5.907797208389644e-06,
+      "loss": 38.9858,
+      "step": 61590
+    },
+    {
+      "epoch": 0.2488717946646089,
+      "grad_norm": 699.4044189453125,
+      "learning_rate": 5.906410245561459e-06,
+      "loss": 43.753,
+      "step": 61600
+    },
+    {
+      "epoch": 0.24891219592997652,
+      "grad_norm": 892.1261596679688,
+      "learning_rate": 5.90502321061437e-06,
+      "loss": 33.1653,
+      "step": 61610
+    },
+    {
+      "epoch": 0.24895259719534416,
+      "grad_norm": 424.92169189453125,
+      "learning_rate": 5.90363610365874e-06,
+      "loss": 60.2741,
+      "step": 61620
+    },
+    {
+      "epoch": 0.2489929984607118,
+      "grad_norm": 613.3280639648438,
+      "learning_rate": 5.9022489248049295e-06,
+      "loss": 53.3307,
+      "step": 61630
+    },
+    {
+      "epoch": 0.2490333997260794,
+      "grad_norm": 638.9053344726562,
+      "learning_rate": 5.900861674163314e-06,
+      "loss": 46.2113,
+      "step": 61640
+    },
+    {
+      "epoch": 0.24907380099144705,
+      "grad_norm": 791.4915771484375,
+      "learning_rate": 5.89947435184427e-06,
+      "loss": 38.609,
+      "step": 61650
+    },
+    {
+      "epoch": 0.2491142022568147,
+      "grad_norm": 549.6468505859375,
+      "learning_rate": 5.89808695795818e-06,
+      "loss": 36.1331,
+      "step": 61660
+    },
+    {
+      "epoch": 0.24915460352218233,
+      "grad_norm": 709.71044921875,
+      "learning_rate": 5.896699492615432e-06,
+      "loss": 43.4479,
+      "step": 61670
+    },
+    {
+      "epoch": 0.24919500478754994,
+      "grad_norm": 972.5881958007812,
+      "learning_rate": 5.895311955926419e-06,
+      "loss": 38.0126,
+      "step": 61680
+    },
+    {
+      "epoch": 0.24923540605291758,
+      "grad_norm": 423.6819152832031,
+      "learning_rate": 5.893924348001544e-06,
+      "loss": 24.6695,
+      "step": 61690
+    },
+    {
+      "epoch": 0.24927580731828522,
+      "grad_norm": 215.3943634033203,
+      "learning_rate": 5.8925366689512124e-06,
+      "loss": 69.2972,
+      "step": 61700
+    },
+    {
+      "epoch": 0.24931620858365283,
+      "grad_norm": 580.0707397460938,
+      "learning_rate": 5.891148918885834e-06,
+      "loss": 47.4895,
+      "step": 61710
+    },
+    {
+      "epoch": 0.24935660984902047,
+      "grad_norm": 405.8501892089844,
+      "learning_rate": 5.8897610979158245e-06,
+      "loss": 53.3295,
+      "step": 61720
+    },
+    {
+      "epoch": 0.2493970111143881,
+      "grad_norm": 465.32843017578125,
+      "learning_rate": 5.888373206151608e-06,
+      "loss": 46.2481,
+      "step": 61730
+    },
+    {
+      "epoch": 0.24943741237975572,
+      "grad_norm": 550.0686645507812,
+      "learning_rate": 5.886985243703612e-06,
+      "loss": 46.1988,
+      "step": 61740
+    },
+    {
+      "epoch": 0.24947781364512336,
+      "grad_norm": 691.0383911132812,
+      "learning_rate": 5.885597210682273e-06,
+      "loss": 41.3984,
+      "step": 61750
+    },
+    {
+      "epoch": 0.249518214910491,
+      "grad_norm": 709.067138671875,
+      "learning_rate": 5.884209107198027e-06,
+      "loss": 48.7272,
+      "step": 61760
+    },
+    {
+      "epoch": 0.24955861617585862,
+      "grad_norm": 583.8157348632812,
+      "learning_rate": 5.882820933361321e-06,
+      "loss": 39.1306,
+      "step": 61770
+    },
+    {
+      "epoch": 0.24959901744122626,
+      "grad_norm": 467.2850036621094,
+      "learning_rate": 5.881432689282604e-06,
+      "loss": 52.9776,
+      "step": 61780
+    },
+    {
+      "epoch": 0.2496394187065939,
+      "grad_norm": 327.1951904296875,
+      "learning_rate": 5.880044375072333e-06,
+      "loss": 53.9046,
+      "step": 61790
+    },
+    {
+      "epoch": 0.2496798199719615,
+      "grad_norm": 1364.8809814453125,
+      "learning_rate": 5.8786559908409715e-06,
+      "loss": 61.291,
+      "step": 61800
+    },
+    {
+      "epoch": 0.24972022123732915,
+      "grad_norm": 549.9939575195312,
+      "learning_rate": 5.877267536698984e-06,
+      "loss": 47.2693,
+      "step": 61810
+    },
+    {
+      "epoch": 0.2497606225026968,
+      "grad_norm": 715.7103881835938,
+      "learning_rate": 5.875879012756845e-06,
+      "loss": 89.7749,
+      "step": 61820
+    },
+    {
+      "epoch": 0.24980102376806443,
+      "grad_norm": 697.6795654296875,
+      "learning_rate": 5.8744904191250326e-06,
+      "loss": 57.4593,
+      "step": 61830
+    },
+    {
+      "epoch": 0.24984142503343204,
+      "grad_norm": 594.6947631835938,
+      "learning_rate": 5.873101755914031e-06,
+      "loss": 40.0505,
+      "step": 61840
+    },
+    {
+      "epoch": 0.24988182629879968,
+      "grad_norm": 465.7455749511719,
+      "learning_rate": 5.87171302323433e-06,
+      "loss": 26.6923,
+      "step": 61850
+    },
+    {
+      "epoch": 0.24992222756416732,
+      "grad_norm": 852.9819946289062,
+      "learning_rate": 5.870324221196424e-06,
+      "loss": 51.0079,
+      "step": 61860
+    },
+    {
+      "epoch": 0.24996262882953493,
+      "grad_norm": 619.8259887695312,
+      "learning_rate": 5.868935349910814e-06,
+      "loss": 35.0291,
+      "step": 61870
+    },
+    {
+      "epoch": 0.25000303009490255,
+      "grad_norm": 347.346923828125,
+      "learning_rate": 5.867546409488006e-06,
+      "loss": 56.9394,
+      "step": 61880
+    },
+    {
+      "epoch": 0.2500434313602702,
+      "grad_norm": 579.9848022460938,
+      "learning_rate": 5.8661574000385115e-06,
+      "loss": 34.7265,
+      "step": 61890
+    },
+    {
+      "epoch": 0.2500838326256378,
+      "grad_norm": 201.31919860839844,
+      "learning_rate": 5.864768321672848e-06,
+      "loss": 39.8269,
+      "step": 61900
+    },
+    {
+      "epoch": 0.25012423389100547,
+      "grad_norm": 767.70068359375,
+      "learning_rate": 5.863379174501538e-06,
+      "loss": 42.9614,
+      "step": 61910
+    },
+    {
+      "epoch": 0.2501646351563731,
+      "grad_norm": 682.972900390625,
+      "learning_rate": 5.861989958635109e-06,
+      "loss": 63.0275,
+      "step": 61920
+    },
+    {
+      "epoch": 0.25020503642174075,
+      "grad_norm": 606.4254150390625,
+      "learning_rate": 5.860600674184096e-06,
+      "loss": 35.5546,
+      "step": 61930
+    },
+    {
+      "epoch": 0.2502454376871084,
+      "grad_norm": 640.8799438476562,
+      "learning_rate": 5.859211321259036e-06,
+      "loss": 72.0527,
+      "step": 61940
+    },
+    {
+      "epoch": 0.25028583895247597,
+      "grad_norm": 770.442626953125,
+      "learning_rate": 5.857821899970475e-06,
+      "loss": 38.6754,
+      "step": 61950
+    },
+    {
+      "epoch": 0.2503262402178436,
+      "grad_norm": 906.4054565429688,
+      "learning_rate": 5.856432410428963e-06,
+      "loss": 38.3737,
+      "step": 61960
+    },
+    {
+      "epoch": 0.25036664148321125,
+      "grad_norm": 776.6735229492188,
+      "learning_rate": 5.8550428527450534e-06,
+      "loss": 31.9661,
+      "step": 61970
+    },
+    {
+      "epoch": 0.2504070427485789,
+      "grad_norm": 845.9974365234375,
+      "learning_rate": 5.8536532270293076e-06,
+      "loss": 44.6735,
+      "step": 61980
+    },
+    {
+      "epoch": 0.25044744401394653,
+      "grad_norm": 597.65625,
+      "learning_rate": 5.852263533392294e-06,
+      "loss": 34.4646,
+      "step": 61990
+    },
+    {
+      "epoch": 0.25048784527931417,
+      "grad_norm": 483.18707275390625,
+      "learning_rate": 5.850873771944581e-06,
+      "loss": 53.7734,
+      "step": 62000
+    },
+    {
+      "epoch": 0.25052824654468175,
+      "grad_norm": 2730.670654296875,
+      "learning_rate": 5.849483942796747e-06,
+      "loss": 51.1974,
+      "step": 62010
+    },
+    {
+      "epoch": 0.2505686478100494,
+      "grad_norm": 716.452392578125,
+      "learning_rate": 5.848094046059375e-06,
+      "loss": 41.6772,
+      "step": 62020
+    },
+    {
+      "epoch": 0.25060904907541703,
+      "grad_norm": 1254.0281982421875,
+      "learning_rate": 5.846704081843052e-06,
+      "loss": 41.3832,
+      "step": 62030
+    },
+    {
+      "epoch": 0.2506494503407847,
+      "grad_norm": 746.8511962890625,
+      "learning_rate": 5.84531405025837e-06,
+      "loss": 51.5164,
+      "step": 62040
+    },
+    {
+      "epoch": 0.2506898516061523,
+      "grad_norm": 845.3733520507812,
+      "learning_rate": 5.843923951415931e-06,
+      "loss": 56.3143,
+      "step": 62050
+    },
+    {
+      "epoch": 0.25073025287151995,
+      "grad_norm": 210.1995086669922,
+      "learning_rate": 5.842533785426334e-06,
+      "loss": 37.6338,
+      "step": 62060
+    },
+    {
+      "epoch": 0.2507706541368876,
+      "grad_norm": 706.4373779296875,
+      "learning_rate": 5.84114355240019e-06,
+      "loss": 49.6547,
+      "step": 62070
+    },
+    {
+      "epoch": 0.2508110554022552,
+      "grad_norm": 511.6235046386719,
+      "learning_rate": 5.839753252448115e-06,
+      "loss": 43.0069,
+      "step": 62080
+    },
+    {
+      "epoch": 0.2508514566676228,
+      "grad_norm": 723.2483520507812,
+      "learning_rate": 5.838362885680728e-06,
+      "loss": 49.3892,
+      "step": 62090
+    },
+    {
+      "epoch": 0.25089185793299046,
+      "grad_norm": 458.5704345703125,
+      "learning_rate": 5.8369724522086545e-06,
+      "loss": 34.8136,
+      "step": 62100
+    },
+    {
+      "epoch": 0.2509322591983581,
+      "grad_norm": 336.73876953125,
+      "learning_rate": 5.835581952142522e-06,
+      "loss": 37.1676,
+      "step": 62110
+    },
+    {
+      "epoch": 0.25097266046372574,
+      "grad_norm": 255.3216552734375,
+      "learning_rate": 5.834191385592969e-06,
+      "loss": 36.8512,
+      "step": 62120
+    },
+    {
+      "epoch": 0.2510130617290934,
+      "grad_norm": 1595.05517578125,
+      "learning_rate": 5.8328007526706354e-06,
+      "loss": 53.1324,
+      "step": 62130
+    },
+    {
+      "epoch": 0.25105346299446096,
+      "grad_norm": 484.79193115234375,
+      "learning_rate": 5.83141005348617e-06,
+      "loss": 30.9311,
+      "step": 62140
+    },
+    {
+      "epoch": 0.2510938642598286,
+      "grad_norm": 821.14990234375,
+      "learning_rate": 5.830019288150222e-06,
+      "loss": 45.7125,
+      "step": 62150
+    },
+    {
+      "epoch": 0.25113426552519624,
+      "grad_norm": 794.8558959960938,
+      "learning_rate": 5.8286284567734456e-06,
+      "loss": 53.2357,
+      "step": 62160
+    },
+    {
+      "epoch": 0.2511746667905639,
+      "grad_norm": 789.1674194335938,
+      "learning_rate": 5.827237559466508e-06,
+      "loss": 41.914,
+      "step": 62170
+    },
+    {
+      "epoch": 0.2512150680559315,
+      "grad_norm": 786.6305541992188,
+      "learning_rate": 5.825846596340075e-06,
+      "loss": 47.0211,
+      "step": 62180
+    },
+    {
+      "epoch": 0.25125546932129916,
+      "grad_norm": 574.0487060546875,
+      "learning_rate": 5.824455567504817e-06,
+      "loss": 44.7584,
+      "step": 62190
+    },
+    {
+      "epoch": 0.25129587058666675,
+      "grad_norm": 918.026123046875,
+      "learning_rate": 5.823064473071414e-06,
+      "loss": 54.0754,
+      "step": 62200
+    },
+    {
+      "epoch": 0.2513362718520344,
+      "grad_norm": 546.5556640625,
+      "learning_rate": 5.821673313150546e-06,
+      "loss": 39.635,
+      "step": 62210
+    },
+    {
+      "epoch": 0.251376673117402,
+      "grad_norm": 761.0563354492188,
+      "learning_rate": 5.820282087852906e-06,
+      "loss": 34.2317,
+      "step": 62220
+    },
+    {
+      "epoch": 0.25141707438276967,
+      "grad_norm": 604.6393432617188,
+      "learning_rate": 5.818890797289185e-06,
+      "loss": 32.208,
+      "step": 62230
+    },
+    {
+      "epoch": 0.2514574756481373,
+      "grad_norm": 803.0030517578125,
+      "learning_rate": 5.81749944157008e-06,
+      "loss": 44.4094,
+      "step": 62240
+    },
+    {
+      "epoch": 0.25149787691350495,
+      "grad_norm": 558.0752563476562,
+      "learning_rate": 5.816108020806297e-06,
+      "loss": 42.962,
+      "step": 62250
+    },
+    {
+      "epoch": 0.2515382781788726,
+      "grad_norm": 873.9310913085938,
+      "learning_rate": 5.814716535108545e-06,
+      "loss": 41.4081,
+      "step": 62260
+    },
+    {
+      "epoch": 0.25157867944424017,
+      "grad_norm": 419.5707702636719,
+      "learning_rate": 5.813324984587536e-06,
+      "loss": 58.7208,
+      "step": 62270
+    },
+    {
+      "epoch": 0.2516190807096078,
+      "grad_norm": 680.248046875,
+      "learning_rate": 5.811933369353992e-06,
+      "loss": 51.4198,
+      "step": 62280
+    },
+    {
+      "epoch": 0.25165948197497545,
+      "grad_norm": 878.7736206054688,
+      "learning_rate": 5.810541689518634e-06,
+      "loss": 53.1204,
+      "step": 62290
+    },
+    {
+      "epoch": 0.2516998832403431,
+      "grad_norm": 466.78271484375,
+      "learning_rate": 5.809149945192194e-06,
+      "loss": 42.8364,
+      "step": 62300
+    },
+    {
+      "epoch": 0.25174028450571073,
+      "grad_norm": 532.00537109375,
+      "learning_rate": 5.807758136485409e-06,
+      "loss": 48.3024,
+      "step": 62310
+    },
+    {
+      "epoch": 0.25178068577107837,
+      "grad_norm": 312.7720642089844,
+      "learning_rate": 5.8063662635090136e-06,
+      "loss": 36.9993,
+      "step": 62320
+    },
+    {
+      "epoch": 0.25182108703644596,
+      "grad_norm": 897.0571899414062,
+      "learning_rate": 5.804974326373756e-06,
+      "loss": 42.1273,
+      "step": 62330
+    },
+    {
+      "epoch": 0.2518614883018136,
+      "grad_norm": 702.832275390625,
+      "learning_rate": 5.803582325190387e-06,
+      "loss": 64.0742,
+      "step": 62340
+    },
+    {
+      "epoch": 0.25190188956718124,
+      "grad_norm": 544.4562377929688,
+      "learning_rate": 5.802190260069657e-06,
+      "loss": 41.5808,
+      "step": 62350
+    },
+    {
+      "epoch": 0.2519422908325489,
+      "grad_norm": 328.93084716796875,
+      "learning_rate": 5.800798131122332e-06,
+      "loss": 52.4388,
+      "step": 62360
+    },
+    {
+      "epoch": 0.2519826920979165,
+      "grad_norm": 516.4020385742188,
+      "learning_rate": 5.799405938459175e-06,
+      "loss": 37.1916,
+      "step": 62370
+    },
+    {
+      "epoch": 0.25202309336328416,
+      "grad_norm": 458.3615417480469,
+      "learning_rate": 5.7980136821909565e-06,
+      "loss": 34.8423,
+      "step": 62380
+    },
+    {
+      "epoch": 0.2520634946286518,
+      "grad_norm": 1137.943115234375,
+      "learning_rate": 5.79662136242845e-06,
+      "loss": 49.3595,
+      "step": 62390
+    },
+    {
+      "epoch": 0.2521038958940194,
+      "grad_norm": 421.2976379394531,
+      "learning_rate": 5.795228979282439e-06,
+      "loss": 34.1204,
+      "step": 62400
+    },
+    {
+      "epoch": 0.252144297159387,
+      "grad_norm": 634.6295776367188,
+      "learning_rate": 5.793836532863707e-06,
+      "loss": 44.5411,
+      "step": 62410
+    },
+    {
+      "epoch": 0.25218469842475466,
+      "grad_norm": 969.116943359375,
+      "learning_rate": 5.792444023283046e-06,
+      "loss": 60.2988,
+      "step": 62420
+    },
+    {
+      "epoch": 0.2522250996901223,
+      "grad_norm": 321.36590576171875,
+      "learning_rate": 5.791051450651251e-06,
+      "loss": 48.6717,
+      "step": 62430
+    },
+    {
+      "epoch": 0.25226550095548994,
+      "grad_norm": 618.8597412109375,
+      "learning_rate": 5.789658815079121e-06,
+      "loss": 47.9615,
+      "step": 62440
+    },
+    {
+      "epoch": 0.2523059022208576,
+      "grad_norm": 327.97607421875,
+      "learning_rate": 5.788266116677464e-06,
+      "loss": 32.009,
+      "step": 62450
+    },
+    {
+      "epoch": 0.25234630348622517,
+      "grad_norm": 552.7090454101562,
+      "learning_rate": 5.78687335555709e-06,
+      "loss": 38.7742,
+      "step": 62460
+    },
+    {
+      "epoch": 0.2523867047515928,
+      "grad_norm": 699.7271118164062,
+      "learning_rate": 5.785480531828815e-06,
+      "loss": 68.4585,
+      "step": 62470
+    },
+    {
+      "epoch": 0.25242710601696045,
+      "grad_norm": 448.2286376953125,
+      "learning_rate": 5.784087645603459e-06,
+      "loss": 55.5299,
+      "step": 62480
+    },
+    {
+      "epoch": 0.2524675072823281,
+      "grad_norm": 744.1871337890625,
+      "learning_rate": 5.782694696991845e-06,
+      "loss": 55.4688,
+      "step": 62490
+    },
+    {
+      "epoch": 0.2525079085476957,
+      "grad_norm": 565.6279907226562,
+      "learning_rate": 5.781301686104808e-06,
+      "loss": 43.324,
+      "step": 62500
+    },
+    {
+      "epoch": 0.25254830981306337,
+      "grad_norm": 497.8145446777344,
+      "learning_rate": 5.779908613053181e-06,
+      "loss": 37.782,
+      "step": 62510
+    },
+    {
+      "epoch": 0.25258871107843095,
+      "grad_norm": 483.60723876953125,
+      "learning_rate": 5.778515477947807e-06,
+      "loss": 44.4805,
+      "step": 62520
+    },
+    {
+      "epoch": 0.2526291123437986,
+      "grad_norm": 1020.0092163085938,
+      "learning_rate": 5.777122280899527e-06,
+      "loss": 48.6439,
+      "step": 62530
+    },
+    {
+      "epoch": 0.25266951360916623,
+      "grad_norm": 146.36549377441406,
+      "learning_rate": 5.775729022019193e-06,
+      "loss": 53.2947,
+      "step": 62540
+    },
+    {
+      "epoch": 0.25270991487453387,
+      "grad_norm": 779.2500610351562,
+      "learning_rate": 5.774335701417662e-06,
+      "loss": 58.0035,
+      "step": 62550
+    },
+    {
+      "epoch": 0.2527503161399015,
+      "grad_norm": 741.26171875,
+      "learning_rate": 5.7729423192057936e-06,
+      "loss": 54.0345,
+      "step": 62560
+    },
+    {
+      "epoch": 0.25279071740526915,
+      "grad_norm": 1187.714599609375,
+      "learning_rate": 5.771548875494453e-06,
+      "loss": 65.7193,
+      "step": 62570
+    },
+    {
+      "epoch": 0.2528311186706368,
+      "grad_norm": 312.81060791015625,
+      "learning_rate": 5.7701553703945055e-06,
+      "loss": 51.9163,
+      "step": 62580
+    },
+    {
+      "epoch": 0.2528715199360044,
+      "grad_norm": 698.7396240234375,
+      "learning_rate": 5.768761804016833e-06,
+      "loss": 39.4483,
+      "step": 62590
+    },
+    {
+      "epoch": 0.252911921201372,
+      "grad_norm": 794.9736328125,
+      "learning_rate": 5.767368176472311e-06,
+      "loss": 47.7515,
+      "step": 62600
+    },
+    {
+      "epoch": 0.25295232246673965,
+      "grad_norm": 687.7762451171875,
+      "learning_rate": 5.765974487871826e-06,
+      "loss": 65.9322,
+      "step": 62610
+    },
+    {
+      "epoch": 0.2529927237321073,
+      "grad_norm": 678.4483032226562,
+      "learning_rate": 5.764580738326265e-06,
+      "loss": 55.1801,
+      "step": 62620
+    },
+    {
+      "epoch": 0.25303312499747493,
+      "grad_norm": 329.08294677734375,
+      "learning_rate": 5.763186927946523e-06,
+      "loss": 46.4785,
+      "step": 62630
+    },
+    {
+      "epoch": 0.2530735262628426,
+      "grad_norm": 529.1029663085938,
+      "learning_rate": 5.761793056843501e-06,
+      "loss": 41.3061,
+      "step": 62640
+    },
+    {
+      "epoch": 0.25311392752821016,
+      "grad_norm": 708.1193237304688,
+      "learning_rate": 5.760399125128102e-06,
+      "loss": 47.0046,
+      "step": 62650
+    },
+    {
+      "epoch": 0.2531543287935778,
+      "grad_norm": 448.0501403808594,
+      "learning_rate": 5.759005132911233e-06,
+      "loss": 61.7254,
+      "step": 62660
+    },
+    {
+      "epoch": 0.25319473005894544,
+      "grad_norm": 955.7879028320312,
+      "learning_rate": 5.75761108030381e-06,
+      "loss": 57.2991,
+      "step": 62670
+    },
+    {
+      "epoch": 0.2532351313243131,
+      "grad_norm": 1111.9713134765625,
+      "learning_rate": 5.756216967416749e-06,
+      "loss": 58.0533,
+      "step": 62680
+    },
+    {
+      "epoch": 0.2532755325896807,
+      "grad_norm": 564.1481323242188,
+      "learning_rate": 5.754822794360976e-06,
+      "loss": 46.9675,
+      "step": 62690
+    },
+    {
+      "epoch": 0.25331593385504836,
+      "grad_norm": 683.9739990234375,
+      "learning_rate": 5.753428561247416e-06,
+      "loss": 42.0367,
+      "step": 62700
+    },
+    {
+      "epoch": 0.253356335120416,
+      "grad_norm": 409.0925598144531,
+      "learning_rate": 5.752034268187005e-06,
+      "loss": 27.2585,
+      "step": 62710
+    },
+    {
+      "epoch": 0.2533967363857836,
+      "grad_norm": 356.8011779785156,
+      "learning_rate": 5.750639915290677e-06,
+      "loss": 32.9736,
+      "step": 62720
+    },
+    {
+      "epoch": 0.2534371376511512,
+      "grad_norm": 684.747802734375,
+      "learning_rate": 5.749245502669375e-06,
+      "loss": 41.3751,
+      "step": 62730
+    },
+    {
+      "epoch": 0.25347753891651886,
+      "grad_norm": 872.2908935546875,
+      "learning_rate": 5.747851030434049e-06,
+      "loss": 63.0237,
+      "step": 62740
+    },
+    {
+      "epoch": 0.2535179401818865,
+      "grad_norm": 647.8345336914062,
+      "learning_rate": 5.746456498695648e-06,
+      "loss": 51.3102,
+      "step": 62750
+    },
+    {
+      "epoch": 0.25355834144725414,
+      "grad_norm": 264.8699035644531,
+      "learning_rate": 5.7450619075651305e-06,
+      "loss": 36.7021,
+      "step": 62760
+    },
+    {
+      "epoch": 0.2535987427126218,
+      "grad_norm": 1874.4996337890625,
+      "learning_rate": 5.743667257153454e-06,
+      "loss": 83.4588,
+      "step": 62770
+    },
+    {
+      "epoch": 0.25363914397798937,
+      "grad_norm": 350.74017333984375,
+      "learning_rate": 5.742272547571588e-06,
+      "loss": 37.704,
+      "step": 62780
+    },
+    {
+      "epoch": 0.253679545243357,
+      "grad_norm": 151.1192169189453,
+      "learning_rate": 5.740877778930503e-06,
+      "loss": 35.6705,
+      "step": 62790
+    },
+    {
+      "epoch": 0.25371994650872465,
+      "grad_norm": 1044.9840087890625,
+      "learning_rate": 5.739482951341172e-06,
+      "loss": 68.5232,
+      "step": 62800
+    },
+    {
+      "epoch": 0.2537603477740923,
+      "grad_norm": 796.45751953125,
+      "learning_rate": 5.738088064914576e-06,
+      "loss": 42.7602,
+      "step": 62810
+    },
+    {
+      "epoch": 0.2538007490394599,
+      "grad_norm": 487.5058288574219,
+      "learning_rate": 5.7366931197617e-06,
+      "loss": 34.9116,
+      "step": 62820
+    },
+    {
+      "epoch": 0.25384115030482757,
+      "grad_norm": 398.38519287109375,
+      "learning_rate": 5.735298115993535e-06,
+      "loss": 51.7369,
+      "step": 62830
+    },
+    {
+      "epoch": 0.25388155157019515,
+      "grad_norm": 546.4779663085938,
+      "learning_rate": 5.733903053721072e-06,
+      "loss": 48.8135,
+      "step": 62840
+    },
+    {
+      "epoch": 0.2539219528355628,
+      "grad_norm": 575.5430297851562,
+      "learning_rate": 5.732507933055311e-06,
+      "loss": 44.4907,
+      "step": 62850
+    },
+    {
+      "epoch": 0.25396235410093043,
+      "grad_norm": 727.0908813476562,
+      "learning_rate": 5.731112754107257e-06,
+      "loss": 57.1689,
+      "step": 62860
+    },
+    {
+      "epoch": 0.25400275536629807,
+      "grad_norm": 836.796875,
+      "learning_rate": 5.729717516987916e-06,
+      "loss": 36.4253,
+      "step": 62870
+    },
+    {
+      "epoch": 0.2540431566316657,
+      "grad_norm": 412.9502258300781,
+      "learning_rate": 5.7283222218083e-06,
+      "loss": 51.407,
+      "step": 62880
+    },
+    {
+      "epoch": 0.25408355789703335,
+      "grad_norm": 531.1953735351562,
+      "learning_rate": 5.726926868679429e-06,
+      "loss": 41.1174,
+      "step": 62890
+    },
+    {
+      "epoch": 0.254123959162401,
+      "grad_norm": 586.2398071289062,
+      "learning_rate": 5.725531457712321e-06,
+      "loss": 51.4301,
+      "step": 62900
+    },
+    {
+      "epoch": 0.2541643604277686,
+      "grad_norm": 497.56488037109375,
+      "learning_rate": 5.724135989018007e-06,
+      "loss": 61.23,
+      "step": 62910
+    },
+    {
+      "epoch": 0.2542047616931362,
+      "grad_norm": 604.0518798828125,
+      "learning_rate": 5.722740462707515e-06,
+      "loss": 55.4451,
+      "step": 62920
+    },
+    {
+      "epoch": 0.25424516295850386,
+      "grad_norm": 517.4009399414062,
+      "learning_rate": 5.72134487889188e-06,
+      "loss": 34.8546,
+      "step": 62930
+    },
+    {
+      "epoch": 0.2542855642238715,
+      "grad_norm": 219.57757568359375,
+      "learning_rate": 5.719949237682145e-06,
+      "loss": 51.7369,
+      "step": 62940
+    },
+    {
+      "epoch": 0.25432596548923914,
+      "grad_norm": 404.9817810058594,
+      "learning_rate": 5.718553539189353e-06,
+      "loss": 56.0131,
+      "step": 62950
+    },
+    {
+      "epoch": 0.2543663667546068,
+      "grad_norm": 568.6887817382812,
+      "learning_rate": 5.717157783524553e-06,
+      "loss": 38.1931,
+      "step": 62960
+    },
+    {
+      "epoch": 0.25440676801997436,
+      "grad_norm": 579.7442626953125,
+      "learning_rate": 5.7157619707988e-06,
+      "loss": 52.0563,
+      "step": 62970
+    },
+    {
+      "epoch": 0.254447169285342,
+      "grad_norm": 423.65606689453125,
+      "learning_rate": 5.714366101123152e-06,
+      "loss": 31.922,
+      "step": 62980
+    },
+    {
+      "epoch": 0.25448757055070964,
+      "grad_norm": 58.358238220214844,
+      "learning_rate": 5.712970174608671e-06,
+      "loss": 40.1323,
+      "step": 62990
+    },
+    {
+      "epoch": 0.2545279718160773,
+      "grad_norm": 1131.2806396484375,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 60.6359,
+      "step": 63000
+    },
+    {
+      "epoch": 0.2545683730814449,
+      "grad_norm": 706.688232421875,
+      "learning_rate": 5.710178151507488e-06,
+      "loss": 46.877,
+      "step": 63010
+    },
+    {
+      "epoch": 0.25460877434681256,
+      "grad_norm": 465.1105041503906,
+      "learning_rate": 5.708782055142934e-06,
+      "loss": 29.5096,
+      "step": 63020
+    },
+    {
+      "epoch": 0.2546491756121802,
+      "grad_norm": 686.3890991210938,
+      "learning_rate": 5.707385902383845e-06,
+      "loss": 38.0391,
+      "step": 63030
+    },
+    {
+      "epoch": 0.2546895768775478,
+      "grad_norm": 522.968994140625,
+      "learning_rate": 5.7059896933413076e-06,
+      "loss": 43.3018,
+      "step": 63040
+    },
+    {
+      "epoch": 0.2547299781429154,
+      "grad_norm": 521.99951171875,
+      "learning_rate": 5.7045934281264085e-06,
+      "loss": 40.3964,
+      "step": 63050
+    },
+    {
+      "epoch": 0.25477037940828307,
+      "grad_norm": 600.0154418945312,
+      "learning_rate": 5.7031971068502425e-06,
+      "loss": 39.3763,
+      "step": 63060
+    },
+    {
+      "epoch": 0.2548107806736507,
+      "grad_norm": 262.6509704589844,
+      "learning_rate": 5.701800729623911e-06,
+      "loss": 34.9332,
+      "step": 63070
+    },
+    {
+      "epoch": 0.25485118193901835,
+      "grad_norm": 471.0314025878906,
+      "learning_rate": 5.700404296558518e-06,
+      "loss": 43.9025,
+      "step": 63080
+    },
+    {
+      "epoch": 0.254891583204386,
+      "grad_norm": 612.1912231445312,
+      "learning_rate": 5.699007807765169e-06,
+      "loss": 31.0212,
+      "step": 63090
+    },
+    {
+      "epoch": 0.25493198446975357,
+      "grad_norm": 760.9694213867188,
+      "learning_rate": 5.6976112633549764e-06,
+      "loss": 56.0173,
+      "step": 63100
+    },
+    {
+      "epoch": 0.2549723857351212,
+      "grad_norm": 692.7157592773438,
+      "learning_rate": 5.696214663439055e-06,
+      "loss": 41.3706,
+      "step": 63110
+    },
+    {
+      "epoch": 0.25501278700048885,
+      "grad_norm": 385.0296630859375,
+      "learning_rate": 5.694818008128531e-06,
+      "loss": 46.966,
+      "step": 63120
+    },
+    {
+      "epoch": 0.2550531882658565,
+      "grad_norm": 423.5424499511719,
+      "learning_rate": 5.693421297534526e-06,
+      "loss": 40.4821,
+      "step": 63130
+    },
+    {
+      "epoch": 0.25509358953122413,
+      "grad_norm": 817.661376953125,
+      "learning_rate": 5.69202453176817e-06,
+      "loss": 55.8041,
+      "step": 63140
+    },
+    {
+      "epoch": 0.25513399079659177,
+      "grad_norm": 604.6948852539062,
+      "learning_rate": 5.6906277109406e-06,
+      "loss": 41.3391,
+      "step": 63150
+    },
+    {
+      "epoch": 0.25517439206195935,
+      "grad_norm": 427.2970886230469,
+      "learning_rate": 5.689230835162949e-06,
+      "loss": 36.6582,
+      "step": 63160
+    },
+    {
+      "epoch": 0.255214793327327,
+      "grad_norm": 563.9060668945312,
+      "learning_rate": 5.687833904546367e-06,
+      "loss": 53.3264,
+      "step": 63170
+    },
+    {
+      "epoch": 0.25525519459269463,
+      "grad_norm": 569.6155395507812,
+      "learning_rate": 5.686436919201996e-06,
+      "loss": 29.1016,
+      "step": 63180
+    },
+    {
+      "epoch": 0.2552955958580623,
+      "grad_norm": 760.7314453125,
+      "learning_rate": 5.68503987924099e-06,
+      "loss": 51.8447,
+      "step": 63190
+    },
+    {
+      "epoch": 0.2553359971234299,
+      "grad_norm": 929.677734375,
+      "learning_rate": 5.683642784774506e-06,
+      "loss": 50.9056,
+      "step": 63200
+    },
+    {
+      "epoch": 0.25537639838879755,
+      "grad_norm": 324.9360656738281,
+      "learning_rate": 5.682245635913701e-06,
+      "loss": 33.1183,
+      "step": 63210
+    },
+    {
+      "epoch": 0.2554167996541652,
+      "grad_norm": 309.93939208984375,
+      "learning_rate": 5.680848432769743e-06,
+      "loss": 61.7535,
+      "step": 63220
+    },
+    {
+      "epoch": 0.2554572009195328,
+      "grad_norm": 295.8665771484375,
+      "learning_rate": 5.6794511754538005e-06,
+      "loss": 44.2926,
+      "step": 63230
+    },
+    {
+      "epoch": 0.2554976021849004,
+      "grad_norm": 474.9871826171875,
+      "learning_rate": 5.6780538640770455e-06,
+      "loss": 53.1915,
+      "step": 63240
+    },
+    {
+      "epoch": 0.25553800345026806,
+      "grad_norm": 822.3848876953125,
+      "learning_rate": 5.6766564987506564e-06,
+      "loss": 35.7922,
+      "step": 63250
+    },
+    {
+      "epoch": 0.2555784047156357,
+      "grad_norm": 837.141357421875,
+      "learning_rate": 5.675259079585816e-06,
+      "loss": 44.0483,
+      "step": 63260
+    },
+    {
+      "epoch": 0.25561880598100334,
+      "grad_norm": 708.4387817382812,
+      "learning_rate": 5.673861606693708e-06,
+      "loss": 49.6165,
+      "step": 63270
+    },
+    {
+      "epoch": 0.255659207246371,
+      "grad_norm": 714.2989501953125,
+      "learning_rate": 5.672464080185526e-06,
+      "loss": 58.6901,
+      "step": 63280
+    },
+    {
+      "epoch": 0.25569960851173856,
+      "grad_norm": 832.5451049804688,
+      "learning_rate": 5.671066500172462e-06,
+      "loss": 44.7032,
+      "step": 63290
+    },
+    {
+      "epoch": 0.2557400097771062,
+      "grad_norm": 330.7470703125,
+      "learning_rate": 5.669668866765717e-06,
+      "loss": 47.6335,
+      "step": 63300
+    },
+    {
+      "epoch": 0.25578041104247384,
+      "grad_norm": 682.3972778320312,
+      "learning_rate": 5.6682711800764935e-06,
+      "loss": 38.2685,
+      "step": 63310
+    },
+    {
+      "epoch": 0.2558208123078415,
+      "grad_norm": 819.9092407226562,
+      "learning_rate": 5.6668734402159994e-06,
+      "loss": 49.0632,
+      "step": 63320
+    },
+    {
+      "epoch": 0.2558612135732091,
+      "grad_norm": 716.800048828125,
+      "learning_rate": 5.6654756472954464e-06,
+      "loss": 42.4433,
+      "step": 63330
+    },
+    {
+      "epoch": 0.25590161483857676,
+      "grad_norm": 730.6797485351562,
+      "learning_rate": 5.66407780142605e-06,
+      "loss": 39.3221,
+      "step": 63340
+    },
+    {
+      "epoch": 0.2559420161039444,
+      "grad_norm": 726.9446411132812,
+      "learning_rate": 5.66267990271903e-06,
+      "loss": 30.0828,
+      "step": 63350
+    },
+    {
+      "epoch": 0.255982417369312,
+      "grad_norm": 454.9652404785156,
+      "learning_rate": 5.661281951285613e-06,
+      "loss": 37.4903,
+      "step": 63360
+    },
+    {
+      "epoch": 0.2560228186346796,
+      "grad_norm": 798.0877075195312,
+      "learning_rate": 5.6598839472370245e-06,
+      "loss": 34.9171,
+      "step": 63370
+    },
+    {
+      "epoch": 0.25606321990004727,
+      "grad_norm": 621.2152099609375,
+      "learning_rate": 5.6584858906845e-06,
+      "loss": 48.2733,
+      "step": 63380
+    },
+    {
+      "epoch": 0.2561036211654149,
+      "grad_norm": 580.8334350585938,
+      "learning_rate": 5.657087781739274e-06,
+      "loss": 51.7214,
+      "step": 63390
+    },
+    {
+      "epoch": 0.25614402243078255,
+      "grad_norm": 728.9893188476562,
+      "learning_rate": 5.6556896205125896e-06,
+      "loss": 50.1375,
+      "step": 63400
+    },
+    {
+      "epoch": 0.2561844236961502,
+      "grad_norm": 608.4055786132812,
+      "learning_rate": 5.654291407115692e-06,
+      "loss": 37.3258,
+      "step": 63410
+    },
+    {
+      "epoch": 0.25622482496151777,
+      "grad_norm": 480.6260681152344,
+      "learning_rate": 5.652893141659829e-06,
+      "loss": 45.5747,
+      "step": 63420
+    },
+    {
+      "epoch": 0.2562652262268854,
+      "grad_norm": 906.700927734375,
+      "learning_rate": 5.651494824256256e-06,
+      "loss": 42.8529,
+      "step": 63430
+    },
+    {
+      "epoch": 0.25630562749225305,
+      "grad_norm": 660.1846923828125,
+      "learning_rate": 5.650096455016227e-06,
+      "loss": 44.0007,
+      "step": 63440
+    },
+    {
+      "epoch": 0.2563460287576207,
+      "grad_norm": 352.1722412109375,
+      "learning_rate": 5.648698034051009e-06,
+      "loss": 41.2812,
+      "step": 63450
+    },
+    {
+      "epoch": 0.25638643002298833,
+      "grad_norm": 862.4525146484375,
+      "learning_rate": 5.647299561471865e-06,
+      "loss": 94.515,
+      "step": 63460
+    },
+    {
+      "epoch": 0.25642683128835597,
+      "grad_norm": 502.7945556640625,
+      "learning_rate": 5.645901037390067e-06,
+      "loss": 53.6442,
+      "step": 63470
+    },
+    {
+      "epoch": 0.25646723255372356,
+      "grad_norm": 263.6398620605469,
+      "learning_rate": 5.644502461916886e-06,
+      "loss": 54.9007,
+      "step": 63480
+    },
+    {
+      "epoch": 0.2565076338190912,
+      "grad_norm": 570.159423828125,
+      "learning_rate": 5.643103835163602e-06,
+      "loss": 60.2439,
+      "step": 63490
+    },
+    {
+      "epoch": 0.25654803508445884,
+      "grad_norm": 954.9754638671875,
+      "learning_rate": 5.641705157241497e-06,
+      "loss": 42.2908,
+      "step": 63500
+    },
+    {
+      "epoch": 0.2565884363498265,
+      "grad_norm": 841.0116577148438,
+      "learning_rate": 5.64030642826186e-06,
+      "loss": 53.6023,
+      "step": 63510
+    },
+    {
+      "epoch": 0.2566288376151941,
+      "grad_norm": 349.643310546875,
+      "learning_rate": 5.6389076483359774e-06,
+      "loss": 45.1764,
+      "step": 63520
+    },
+    {
+      "epoch": 0.25666923888056176,
+      "grad_norm": 291.5067138671875,
+      "learning_rate": 5.637508817575145e-06,
+      "loss": 58.82,
+      "step": 63530
+    },
+    {
+      "epoch": 0.2567096401459294,
+      "grad_norm": 462.4788818359375,
+      "learning_rate": 5.636109936090661e-06,
+      "loss": 38.8552,
+      "step": 63540
+    },
+    {
+      "epoch": 0.256750041411297,
+      "grad_norm": 437.5688171386719,
+      "learning_rate": 5.634711003993832e-06,
+      "loss": 44.9762,
+      "step": 63550
+    },
+    {
+      "epoch": 0.2567904426766646,
+      "grad_norm": 559.8381958007812,
+      "learning_rate": 5.633312021395959e-06,
+      "loss": 51.4478,
+      "step": 63560
+    },
+    {
+      "epoch": 0.25683084394203226,
+      "grad_norm": 442.395263671875,
+      "learning_rate": 5.631912988408356e-06,
+      "loss": 31.9547,
+      "step": 63570
+    },
+    {
+      "epoch": 0.2568712452073999,
+      "grad_norm": 431.67193603515625,
+      "learning_rate": 5.630513905142334e-06,
+      "loss": 45.5192,
+      "step": 63580
+    },
+    {
+      "epoch": 0.25691164647276754,
+      "grad_norm": 524.9291381835938,
+      "learning_rate": 5.629114771709217e-06,
+      "loss": 42.338,
+      "step": 63590
+    },
+    {
+      "epoch": 0.2569520477381352,
+      "grad_norm": 986.5203857421875,
+      "learning_rate": 5.627715588220325e-06,
+      "loss": 46.758,
+      "step": 63600
+    },
+    {
+      "epoch": 0.25699244900350277,
+      "grad_norm": 263.0068664550781,
+      "learning_rate": 5.626316354786982e-06,
+      "loss": 54.736,
+      "step": 63610
+    },
+    {
+      "epoch": 0.2570328502688704,
+      "grad_norm": 528.8858642578125,
+      "learning_rate": 5.624917071520524e-06,
+      "loss": 46.2457,
+      "step": 63620
+    },
+    {
+      "epoch": 0.25707325153423805,
+      "grad_norm": 592.2925415039062,
+      "learning_rate": 5.62351773853228e-06,
+      "loss": 50.5073,
+      "step": 63630
+    },
+    {
+      "epoch": 0.2571136527996057,
+      "grad_norm": 473.4315490722656,
+      "learning_rate": 5.6221183559335935e-06,
+      "loss": 40.8695,
+      "step": 63640
+    },
+    {
+      "epoch": 0.2571540540649733,
+      "grad_norm": 681.96435546875,
+      "learning_rate": 5.6207189238358025e-06,
+      "loss": 40.739,
+      "step": 63650
+    },
+    {
+      "epoch": 0.25719445533034097,
+      "grad_norm": 411.7383728027344,
+      "learning_rate": 5.619319442350256e-06,
+      "loss": 39.9328,
+      "step": 63660
+    },
+    {
+      "epoch": 0.2572348565957086,
+      "grad_norm": 375.4176330566406,
+      "learning_rate": 5.617919911588304e-06,
+      "loss": 47.2908,
+      "step": 63670
+    },
+    {
+      "epoch": 0.2572752578610762,
+      "grad_norm": 451.3565979003906,
+      "learning_rate": 5.616520331661301e-06,
+      "loss": 42.4083,
+      "step": 63680
+    },
+    {
+      "epoch": 0.25731565912644383,
+      "grad_norm": 469.7231750488281,
+      "learning_rate": 5.615120702680604e-06,
+      "loss": 31.2935,
+      "step": 63690
+    },
+    {
+      "epoch": 0.25735606039181147,
+      "grad_norm": 539.3211669921875,
+      "learning_rate": 5.6137210247575754e-06,
+      "loss": 54.5643,
+      "step": 63700
+    },
+    {
+      "epoch": 0.2573964616571791,
+      "grad_norm": 791.468505859375,
+      "learning_rate": 5.6123212980035825e-06,
+      "loss": 50.1051,
+      "step": 63710
+    },
+    {
+      "epoch": 0.25743686292254675,
+      "grad_norm": 847.0855712890625,
+      "learning_rate": 5.610921522529994e-06,
+      "loss": 67.9444,
+      "step": 63720
+    },
+    {
+      "epoch": 0.2574772641879144,
+      "grad_norm": 657.5780639648438,
+      "learning_rate": 5.609521698448183e-06,
+      "loss": 53.8216,
+      "step": 63730
+    },
+    {
+      "epoch": 0.257517665453282,
+      "grad_norm": 761.1411743164062,
+      "learning_rate": 5.608121825869528e-06,
+      "loss": 53.258,
+      "step": 63740
+    },
+    {
+      "epoch": 0.2575580667186496,
+      "grad_norm": 774.0830688476562,
+      "learning_rate": 5.60672190490541e-06,
+      "loss": 38.646,
+      "step": 63750
+    },
+    {
+      "epoch": 0.25759846798401725,
+      "grad_norm": 567.2658081054688,
+      "learning_rate": 5.6053219356672155e-06,
+      "loss": 52.5933,
+      "step": 63760
+    },
+    {
+      "epoch": 0.2576388692493849,
+      "grad_norm": 410.2081298828125,
+      "learning_rate": 5.603921918266332e-06,
+      "loss": 37.3336,
+      "step": 63770
+    },
+    {
+      "epoch": 0.25767927051475253,
+      "grad_norm": 1132.9581298828125,
+      "learning_rate": 5.602521852814152e-06,
+      "loss": 65.356,
+      "step": 63780
+    },
+    {
+      "epoch": 0.2577196717801202,
+      "grad_norm": 413.339599609375,
+      "learning_rate": 5.6011217394220755e-06,
+      "loss": 62.1223,
+      "step": 63790
+    },
+    {
+      "epoch": 0.25776007304548776,
+      "grad_norm": 771.2860107421875,
+      "learning_rate": 5.599721578201499e-06,
+      "loss": 43.2339,
+      "step": 63800
+    },
+    {
+      "epoch": 0.2578004743108554,
+      "grad_norm": 581.4656372070312,
+      "learning_rate": 5.59832136926383e-06,
+      "loss": 38.4429,
+      "step": 63810
+    },
+    {
+      "epoch": 0.25784087557622304,
+      "grad_norm": 510.95404052734375,
+      "learning_rate": 5.5969211127204744e-06,
+      "loss": 29.4889,
+      "step": 63820
+    },
+    {
+      "epoch": 0.2578812768415907,
+      "grad_norm": 1371.288818359375,
+      "learning_rate": 5.595520808682848e-06,
+      "loss": 61.4885,
+      "step": 63830
+    },
+    {
+      "epoch": 0.2579216781069583,
+      "grad_norm": 395.72900390625,
+      "learning_rate": 5.594120457262361e-06,
+      "loss": 42.5042,
+      "step": 63840
+    },
+    {
+      "epoch": 0.25796207937232596,
+      "grad_norm": 435.349365234375,
+      "learning_rate": 5.592720058570438e-06,
+      "loss": 41.518,
+      "step": 63850
+    },
+    {
+      "epoch": 0.2580024806376936,
+      "grad_norm": 511.0625915527344,
+      "learning_rate": 5.591319612718498e-06,
+      "loss": 40.8742,
+      "step": 63860
+    },
+    {
+      "epoch": 0.2580428819030612,
+      "grad_norm": 502.5708312988281,
+      "learning_rate": 5.589919119817971e-06,
+      "loss": 53.2404,
+      "step": 63870
+    },
+    {
+      "epoch": 0.2580832831684288,
+      "grad_norm": 723.0562133789062,
+      "learning_rate": 5.588518579980288e-06,
+      "loss": 40.6113,
+      "step": 63880
+    },
+    {
+      "epoch": 0.25812368443379646,
+      "grad_norm": 1327.1318359375,
+      "learning_rate": 5.587117993316882e-06,
+      "loss": 37.1122,
+      "step": 63890
+    },
+    {
+      "epoch": 0.2581640856991641,
+      "grad_norm": 433.02532958984375,
+      "learning_rate": 5.585717359939192e-06,
+      "loss": 31.3232,
+      "step": 63900
+    },
+    {
+      "epoch": 0.25820448696453174,
+      "grad_norm": 538.7310791015625,
+      "learning_rate": 5.584316679958659e-06,
+      "loss": 29.2123,
+      "step": 63910
+    },
+    {
+      "epoch": 0.2582448882298994,
+      "grad_norm": 606.9814453125,
+      "learning_rate": 5.58291595348673e-06,
+      "loss": 46.0158,
+      "step": 63920
+    },
+    {
+      "epoch": 0.25828528949526697,
+      "grad_norm": 933.4443359375,
+      "learning_rate": 5.581515180634853e-06,
+      "loss": 41.5524,
+      "step": 63930
+    },
+    {
+      "epoch": 0.2583256907606346,
+      "grad_norm": 620.8282470703125,
+      "learning_rate": 5.580114361514484e-06,
+      "loss": 38.4091,
+      "step": 63940
+    },
+    {
+      "epoch": 0.25836609202600225,
+      "grad_norm": 478.485107421875,
+      "learning_rate": 5.5787134962370755e-06,
+      "loss": 35.9659,
+      "step": 63950
+    },
+    {
+      "epoch": 0.2584064932913699,
+      "grad_norm": 771.5250854492188,
+      "learning_rate": 5.57731258491409e-06,
+      "loss": 50.4317,
+      "step": 63960
+    },
+    {
+      "epoch": 0.2584468945567375,
+      "grad_norm": 213.68081665039062,
+      "learning_rate": 5.575911627656993e-06,
+      "loss": 30.2937,
+      "step": 63970
+    },
+    {
+      "epoch": 0.25848729582210517,
+      "grad_norm": 783.5086669921875,
+      "learning_rate": 5.5745106245772506e-06,
+      "loss": 54.8814,
+      "step": 63980
+    },
+    {
+      "epoch": 0.25852769708747275,
+      "grad_norm": 498.461669921875,
+      "learning_rate": 5.573109575786334e-06,
+      "loss": 32.4557,
+      "step": 63990
+    },
+    {
+      "epoch": 0.2585680983528404,
+      "grad_norm": 562.4568481445312,
+      "learning_rate": 5.571708481395719e-06,
+      "loss": 39.6354,
+      "step": 64000
+    },
+    {
+      "epoch": 0.25860849961820803,
+      "grad_norm": 724.8565673828125,
+      "learning_rate": 5.570307341516882e-06,
+      "loss": 42.1159,
+      "step": 64010
+    },
+    {
+      "epoch": 0.25864890088357567,
+      "grad_norm": 687.0962524414062,
+      "learning_rate": 5.568906156261309e-06,
+      "loss": 45.9281,
+      "step": 64020
+    },
+    {
+      "epoch": 0.2586893021489433,
+      "grad_norm": 924.2523803710938,
+      "learning_rate": 5.567504925740484e-06,
+      "loss": 36.815,
+      "step": 64030
+    },
+    {
+      "epoch": 0.25872970341431095,
+      "grad_norm": 449.19097900390625,
+      "learning_rate": 5.566103650065897e-06,
+      "loss": 45.8584,
+      "step": 64040
+    },
+    {
+      "epoch": 0.2587701046796786,
+      "grad_norm": 406.7481689453125,
+      "learning_rate": 5.564702329349041e-06,
+      "loss": 43.2192,
+      "step": 64050
+    },
+    {
+      "epoch": 0.2588105059450462,
+      "grad_norm": 845.9132080078125,
+      "learning_rate": 5.56330096370141e-06,
+      "loss": 57.1466,
+      "step": 64060
+    },
+    {
+      "epoch": 0.2588509072104138,
+      "grad_norm": 503.916748046875,
+      "learning_rate": 5.561899553234509e-06,
+      "loss": 43.5526,
+      "step": 64070
+    },
+    {
+      "epoch": 0.25889130847578146,
+      "grad_norm": 410.317138671875,
+      "learning_rate": 5.560498098059838e-06,
+      "loss": 39.6964,
+      "step": 64080
+    },
+    {
+      "epoch": 0.2589317097411491,
+      "grad_norm": 661.5330810546875,
+      "learning_rate": 5.559096598288906e-06,
+      "loss": 43.7603,
+      "step": 64090
+    },
+    {
+      "epoch": 0.25897211100651674,
+      "grad_norm": 1017.6605834960938,
+      "learning_rate": 5.557695054033223e-06,
+      "loss": 42.1348,
+      "step": 64100
+    },
+    {
+      "epoch": 0.2590125122718844,
+      "grad_norm": 738.2880859375,
+      "learning_rate": 5.556293465404304e-06,
+      "loss": 52.5627,
+      "step": 64110
+    },
+    {
+      "epoch": 0.25905291353725196,
+      "grad_norm": 281.1907043457031,
+      "learning_rate": 5.554891832513668e-06,
+      "loss": 44.3268,
+      "step": 64120
+    },
+    {
+      "epoch": 0.2590933148026196,
+      "grad_norm": 557.4099731445312,
+      "learning_rate": 5.553490155472835e-06,
+      "loss": 53.3458,
+      "step": 64130
+    },
+    {
+      "epoch": 0.25913371606798724,
+      "grad_norm": 569.7567138671875,
+      "learning_rate": 5.55208843439333e-06,
+      "loss": 46.0401,
+      "step": 64140
+    },
+    {
+      "epoch": 0.2591741173333549,
+      "grad_norm": 503.8868713378906,
+      "learning_rate": 5.550686669386683e-06,
+      "loss": 41.1913,
+      "step": 64150
+    },
+    {
+      "epoch": 0.2592145185987225,
+      "grad_norm": 928.1849975585938,
+      "learning_rate": 5.549284860564425e-06,
+      "loss": 31.0713,
+      "step": 64160
+    },
+    {
+      "epoch": 0.25925491986409016,
+      "grad_norm": 339.114013671875,
+      "learning_rate": 5.547883008038091e-06,
+      "loss": 33.9224,
+      "step": 64170
+    },
+    {
+      "epoch": 0.2592953211294578,
+      "grad_norm": 647.3558349609375,
+      "learning_rate": 5.54648111191922e-06,
+      "loss": 52.8394,
+      "step": 64180
+    },
+    {
+      "epoch": 0.2593357223948254,
+      "grad_norm": 598.630859375,
+      "learning_rate": 5.545079172319355e-06,
+      "loss": 60.7535,
+      "step": 64190
+    },
+    {
+      "epoch": 0.259376123660193,
+      "grad_norm": 746.982666015625,
+      "learning_rate": 5.543677189350043e-06,
+      "loss": 39.941,
+      "step": 64200
+    },
+    {
+      "epoch": 0.25941652492556067,
+      "grad_norm": 382.3617858886719,
+      "learning_rate": 5.542275163122831e-06,
+      "loss": 42.0451,
+      "step": 64210
+    },
+    {
+      "epoch": 0.2594569261909283,
+      "grad_norm": 486.2451171875,
+      "learning_rate": 5.540873093749274e-06,
+      "loss": 34.248,
+      "step": 64220
+    },
+    {
+      "epoch": 0.25949732745629595,
+      "grad_norm": 392.88134765625,
+      "learning_rate": 5.539470981340926e-06,
+      "loss": 56.3614,
+      "step": 64230
+    },
+    {
+      "epoch": 0.2595377287216636,
+      "grad_norm": 1441.9869384765625,
+      "learning_rate": 5.538068826009349e-06,
+      "loss": 52.6453,
+      "step": 64240
+    },
+    {
+      "epoch": 0.25957812998703117,
+      "grad_norm": 695.2682495117188,
+      "learning_rate": 5.536666627866104e-06,
+      "loss": 56.2888,
+      "step": 64250
+    },
+    {
+      "epoch": 0.2596185312523988,
+      "grad_norm": 629.8070678710938,
+      "learning_rate": 5.53526438702276e-06,
+      "loss": 58.3548,
+      "step": 64260
+    },
+    {
+      "epoch": 0.25965893251776645,
+      "grad_norm": 518.64013671875,
+      "learning_rate": 5.533862103590883e-06,
+      "loss": 53.5079,
+      "step": 64270
+    },
+    {
+      "epoch": 0.2596993337831341,
+      "grad_norm": 733.4268188476562,
+      "learning_rate": 5.532459777682051e-06,
+      "loss": 56.0399,
+      "step": 64280
+    },
+    {
+      "epoch": 0.25973973504850173,
+      "grad_norm": 455.0276184082031,
+      "learning_rate": 5.5310574094078365e-06,
+      "loss": 41.9341,
+      "step": 64290
+    },
+    {
+      "epoch": 0.25978013631386937,
+      "grad_norm": 284.0489196777344,
+      "learning_rate": 5.529654998879821e-06,
+      "loss": 34.0637,
+      "step": 64300
+    },
+    {
+      "epoch": 0.25982053757923695,
+      "grad_norm": 0.0,
+      "learning_rate": 5.528252546209588e-06,
+      "loss": 37.8221,
+      "step": 64310
+    },
+    {
+      "epoch": 0.2598609388446046,
+      "grad_norm": 401.0384521484375,
+      "learning_rate": 5.526850051508725e-06,
+      "loss": 68.1539,
+      "step": 64320
+    },
+    {
+      "epoch": 0.25990134010997223,
+      "grad_norm": 904.4043579101562,
+      "learning_rate": 5.525447514888822e-06,
+      "loss": 50.8169,
+      "step": 64330
+    },
+    {
+      "epoch": 0.2599417413753399,
+      "grad_norm": 356.97705078125,
+      "learning_rate": 5.52404493646147e-06,
+      "loss": 41.6639,
+      "step": 64340
+    },
+    {
+      "epoch": 0.2599821426407075,
+      "grad_norm": 577.2202758789062,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 44.3976,
+      "step": 64350
+    },
+    {
+      "epoch": 0.26002254390607515,
+      "grad_norm": 219.48606872558594,
+      "learning_rate": 5.521239654630816e-06,
+      "loss": 45.1079,
+      "step": 64360
+    },
+    {
+      "epoch": 0.2600629451714428,
+      "grad_norm": 567.2487182617188,
+      "learning_rate": 5.519836951450716e-06,
+      "loss": 43.7239,
+      "step": 64370
+    },
+    {
+      "epoch": 0.2601033464368104,
+      "grad_norm": 759.8638305664062,
+      "learning_rate": 5.518434206909577e-06,
+      "loss": 49.3401,
+      "step": 64380
+    },
+    {
+      "epoch": 0.260143747702178,
+      "grad_norm": 494.2590637207031,
+      "learning_rate": 5.517031421119006e-06,
+      "loss": 40.7932,
+      "step": 64390
+    },
+    {
+      "epoch": 0.26018414896754566,
+      "grad_norm": 621.8897705078125,
+      "learning_rate": 5.5156285941906175e-06,
+      "loss": 58.2495,
+      "step": 64400
+    },
+    {
+      "epoch": 0.2602245502329133,
+      "grad_norm": 336.2132263183594,
+      "learning_rate": 5.51422572623603e-06,
+      "loss": 41.9188,
+      "step": 64410
+    },
+    {
+      "epoch": 0.26026495149828094,
+      "grad_norm": 544.7830200195312,
+      "learning_rate": 5.512822817366859e-06,
+      "loss": 32.5637,
+      "step": 64420
+    },
+    {
+      "epoch": 0.2603053527636486,
+      "grad_norm": 395.4822082519531,
+      "learning_rate": 5.511419867694733e-06,
+      "loss": 38.594,
+      "step": 64430
+    },
+    {
+      "epoch": 0.26034575402901616,
+      "grad_norm": 653.3141479492188,
+      "learning_rate": 5.510016877331271e-06,
+      "loss": 62.9466,
+      "step": 64440
+    },
+    {
+      "epoch": 0.2603861552943838,
+      "grad_norm": 697.96826171875,
+      "learning_rate": 5.50861384638811e-06,
+      "loss": 40.165,
+      "step": 64450
+    },
+    {
+      "epoch": 0.26042655655975144,
+      "grad_norm": 451.45068359375,
+      "learning_rate": 5.50721077497688e-06,
+      "loss": 40.0697,
+      "step": 64460
+    },
+    {
+      "epoch": 0.2604669578251191,
+      "grad_norm": 496.3128662109375,
+      "learning_rate": 5.505807663209215e-06,
+      "loss": 42.665,
+      "step": 64470
+    },
+    {
+      "epoch": 0.2605073590904867,
+      "grad_norm": 808.9034423828125,
+      "learning_rate": 5.504404511196755e-06,
+      "loss": 33.7379,
+      "step": 64480
+    },
+    {
+      "epoch": 0.26054776035585436,
+      "grad_norm": 502.8928527832031,
+      "learning_rate": 5.503001319051142e-06,
+      "loss": 52.4628,
+      "step": 64490
+    },
+    {
+      "epoch": 0.260588161621222,
+      "grad_norm": 700.6588745117188,
+      "learning_rate": 5.5015980868840254e-06,
+      "loss": 44.8512,
+      "step": 64500
+    },
+    {
+      "epoch": 0.2606285628865896,
+      "grad_norm": 538.7782592773438,
+      "learning_rate": 5.500194814807051e-06,
+      "loss": 54.6171,
+      "step": 64510
+    },
+    {
+      "epoch": 0.2606689641519572,
+      "grad_norm": 871.9270629882812,
+      "learning_rate": 5.498791502931868e-06,
+      "loss": 57.9137,
+      "step": 64520
+    },
+    {
+      "epoch": 0.26070936541732487,
+      "grad_norm": 764.1802368164062,
+      "learning_rate": 5.497388151370136e-06,
+      "loss": 64.9328,
+      "step": 64530
+    },
+    {
+      "epoch": 0.2607497666826925,
+      "grad_norm": 487.97076416015625,
+      "learning_rate": 5.495984760233511e-06,
+      "loss": 46.9365,
+      "step": 64540
+    },
+    {
+      "epoch": 0.26079016794806015,
+      "grad_norm": 731.7493896484375,
+      "learning_rate": 5.494581329633656e-06,
+      "loss": 71.8972,
+      "step": 64550
+    },
+    {
+      "epoch": 0.2608305692134278,
+      "grad_norm": 730.07861328125,
+      "learning_rate": 5.493177859682234e-06,
+      "loss": 64.3233,
+      "step": 64560
+    },
+    {
+      "epoch": 0.26087097047879537,
+      "grad_norm": 439.54656982421875,
+      "learning_rate": 5.491774350490912e-06,
+      "loss": 39.8704,
+      "step": 64570
+    },
+    {
+      "epoch": 0.260911371744163,
+      "grad_norm": 558.0665283203125,
+      "learning_rate": 5.490370802171362e-06,
+      "loss": 45.7664,
+      "step": 64580
+    },
+    {
+      "epoch": 0.26095177300953065,
+      "grad_norm": 517.4381713867188,
+      "learning_rate": 5.488967214835259e-06,
+      "loss": 38.9769,
+      "step": 64590
+    },
+    {
+      "epoch": 0.2609921742748983,
+      "grad_norm": 306.56439208984375,
+      "learning_rate": 5.487563588594278e-06,
+      "loss": 60.6436,
+      "step": 64600
+    },
+    {
+      "epoch": 0.26103257554026593,
+      "grad_norm": 289.06494140625,
+      "learning_rate": 5.4861599235601e-06,
+      "loss": 47.6477,
+      "step": 64610
+    },
+    {
+      "epoch": 0.26107297680563357,
+      "grad_norm": 446.7611083984375,
+      "learning_rate": 5.484756219844408e-06,
+      "loss": 63.2884,
+      "step": 64620
+    },
+    {
+      "epoch": 0.26111337807100116,
+      "grad_norm": 599.0672607421875,
+      "learning_rate": 5.483352477558889e-06,
+      "loss": 36.3158,
+      "step": 64630
+    },
+    {
+      "epoch": 0.2611537793363688,
+      "grad_norm": 531.96337890625,
+      "learning_rate": 5.48194869681523e-06,
+      "loss": 67.8228,
+      "step": 64640
+    },
+    {
+      "epoch": 0.26119418060173644,
+      "grad_norm": 485.176025390625,
+      "learning_rate": 5.480544877725127e-06,
+      "loss": 41.1143,
+      "step": 64650
+    },
+    {
+      "epoch": 0.2612345818671041,
+      "grad_norm": 217.22512817382812,
+      "learning_rate": 5.479141020400271e-06,
+      "loss": 36.4919,
+      "step": 64660
+    },
+    {
+      "epoch": 0.2612749831324717,
+      "grad_norm": 906.6617431640625,
+      "learning_rate": 5.477737124952366e-06,
+      "loss": 33.1911,
+      "step": 64670
+    },
+    {
+      "epoch": 0.26131538439783936,
+      "grad_norm": 629.6244506835938,
+      "learning_rate": 5.476333191493108e-06,
+      "loss": 38.5584,
+      "step": 64680
+    },
+    {
+      "epoch": 0.261355785663207,
+      "grad_norm": 404.6357421875,
+      "learning_rate": 5.474929220134205e-06,
+      "loss": 61.765,
+      "step": 64690
+    },
+    {
+      "epoch": 0.2613961869285746,
+      "grad_norm": 490.454345703125,
+      "learning_rate": 5.473525210987363e-06,
+      "loss": 63.0138,
+      "step": 64700
+    },
+    {
+      "epoch": 0.2614365881939422,
+      "grad_norm": 570.0198974609375,
+      "learning_rate": 5.472121164164295e-06,
+      "loss": 47.7393,
+      "step": 64710
+    },
+    {
+      "epoch": 0.26147698945930986,
+      "grad_norm": 475.7129821777344,
+      "learning_rate": 5.47071707977671e-06,
+      "loss": 30.8131,
+      "step": 64720
+    },
+    {
+      "epoch": 0.2615173907246775,
+      "grad_norm": 791.043212890625,
+      "learning_rate": 5.46931295793633e-06,
+      "loss": 53.3681,
+      "step": 64730
+    },
+    {
+      "epoch": 0.26155779199004514,
+      "grad_norm": 850.3129272460938,
+      "learning_rate": 5.46790879875487e-06,
+      "loss": 41.8812,
+      "step": 64740
+    },
+    {
+      "epoch": 0.2615981932554128,
+      "grad_norm": 735.2415771484375,
+      "learning_rate": 5.466504602344055e-06,
+      "loss": 49.5888,
+      "step": 64750
+    },
+    {
+      "epoch": 0.26163859452078037,
+      "grad_norm": 838.309814453125,
+      "learning_rate": 5.465100368815609e-06,
+      "loss": 32.8048,
+      "step": 64760
+    },
+    {
+      "epoch": 0.261678995786148,
+      "grad_norm": 426.8292541503906,
+      "learning_rate": 5.463696098281262e-06,
+      "loss": 49.3965,
+      "step": 64770
+    },
+    {
+      "epoch": 0.26171939705151565,
+      "grad_norm": 447.4613952636719,
+      "learning_rate": 5.462291790852744e-06,
+      "loss": 51.6955,
+      "step": 64780
+    },
+    {
+      "epoch": 0.2617597983168833,
+      "grad_norm": 713.3033447265625,
+      "learning_rate": 5.46088744664179e-06,
+      "loss": 43.2434,
+      "step": 64790
+    },
+    {
+      "epoch": 0.2618001995822509,
+      "grad_norm": 333.4674987792969,
+      "learning_rate": 5.459483065760138e-06,
+      "loss": 37.371,
+      "step": 64800
+    },
+    {
+      "epoch": 0.26184060084761857,
+      "grad_norm": 687.7012329101562,
+      "learning_rate": 5.458078648319526e-06,
+      "loss": 52.3234,
+      "step": 64810
+    },
+    {
+      "epoch": 0.2618810021129862,
+      "grad_norm": 554.49560546875,
+      "learning_rate": 5.456674194431698e-06,
+      "loss": 42.5262,
+      "step": 64820
+    },
+    {
+      "epoch": 0.2619214033783538,
+      "grad_norm": 389.0173645019531,
+      "learning_rate": 5.455269704208401e-06,
+      "loss": 34.1464,
+      "step": 64830
+    },
+    {
+      "epoch": 0.26196180464372143,
+      "grad_norm": 333.84906005859375,
+      "learning_rate": 5.453865177761384e-06,
+      "loss": 55.9638,
+      "step": 64840
+    },
+    {
+      "epoch": 0.26200220590908907,
+      "grad_norm": 480.15155029296875,
+      "learning_rate": 5.4524606152023975e-06,
+      "loss": 49.9895,
+      "step": 64850
+    },
+    {
+      "epoch": 0.2620426071744567,
+      "grad_norm": 240.2779541015625,
+      "learning_rate": 5.4510560166431935e-06,
+      "loss": 38.5802,
+      "step": 64860
+    },
+    {
+      "epoch": 0.26208300843982435,
+      "grad_norm": 437.9780578613281,
+      "learning_rate": 5.449651382195535e-06,
+      "loss": 37.2159,
+      "step": 64870
+    },
+    {
+      "epoch": 0.262123409705192,
+      "grad_norm": 667.8101196289062,
+      "learning_rate": 5.448246711971178e-06,
+      "loss": 64.4867,
+      "step": 64880
+    },
+    {
+      "epoch": 0.2621638109705596,
+      "grad_norm": 525.559326171875,
+      "learning_rate": 5.44684200608189e-06,
+      "loss": 41.0538,
+      "step": 64890
+    },
+    {
+      "epoch": 0.2622042122359272,
+      "grad_norm": 797.489013671875,
+      "learning_rate": 5.445437264639433e-06,
+      "loss": 40.2842,
+      "step": 64900
+    },
+    {
+      "epoch": 0.26224461350129485,
+      "grad_norm": 539.8469848632812,
+      "learning_rate": 5.444032487755575e-06,
+      "loss": 39.7015,
+      "step": 64910
+    },
+    {
+      "epoch": 0.2622850147666625,
+      "grad_norm": 598.1268310546875,
+      "learning_rate": 5.442627675542092e-06,
+      "loss": 38.8527,
+      "step": 64920
+    },
+    {
+      "epoch": 0.26232541603203013,
+      "grad_norm": 509.8528137207031,
+      "learning_rate": 5.441222828110756e-06,
+      "loss": 40.2096,
+      "step": 64930
+    },
+    {
+      "epoch": 0.2623658172973978,
+      "grad_norm": 793.486083984375,
+      "learning_rate": 5.439817945573345e-06,
+      "loss": 47.9723,
+      "step": 64940
+    },
+    {
+      "epoch": 0.26240621856276536,
+      "grad_norm": 1765.65771484375,
+      "learning_rate": 5.438413028041637e-06,
+      "loss": 53.3736,
+      "step": 64950
+    },
+    {
+      "epoch": 0.262446619828133,
+      "grad_norm": 871.4049682617188,
+      "learning_rate": 5.4370080756274155e-06,
+      "loss": 74.8709,
+      "step": 64960
+    },
+    {
+      "epoch": 0.26248702109350064,
+      "grad_norm": 340.2928771972656,
+      "learning_rate": 5.435603088442471e-06,
+      "loss": 45.8343,
+      "step": 64970
+    },
+    {
+      "epoch": 0.2625274223588683,
+      "grad_norm": 465.7066345214844,
+      "learning_rate": 5.434198066598585e-06,
+      "loss": 50.4548,
+      "step": 64980
+    },
+    {
+      "epoch": 0.2625678236242359,
+      "grad_norm": 299.4384460449219,
+      "learning_rate": 5.4327930102075525e-06,
+      "loss": 36.6418,
+      "step": 64990
+    },
+    {
+      "epoch": 0.26260822488960356,
+      "grad_norm": 597.560302734375,
+      "learning_rate": 5.431387919381166e-06,
+      "loss": 51.8495,
+      "step": 65000
+    },
+    {
+      "epoch": 0.2626486261549712,
+      "grad_norm": 583.4240112304688,
+      "learning_rate": 5.429982794231221e-06,
+      "loss": 42.3373,
+      "step": 65010
+    },
+    {
+      "epoch": 0.2626890274203388,
+      "grad_norm": 256.0840148925781,
+      "learning_rate": 5.428577634869521e-06,
+      "loss": 41.6526,
+      "step": 65020
+    },
+    {
+      "epoch": 0.2627294286857064,
+      "grad_norm": 569.7841186523438,
+      "learning_rate": 5.427172441407864e-06,
+      "loss": 29.5803,
+      "step": 65030
+    },
+    {
+      "epoch": 0.26276982995107406,
+      "grad_norm": 622.056884765625,
+      "learning_rate": 5.425767213958057e-06,
+      "loss": 55.9951,
+      "step": 65040
+    },
+    {
+      "epoch": 0.2628102312164417,
+      "grad_norm": 455.84320068359375,
+      "learning_rate": 5.424361952631907e-06,
+      "loss": 61.679,
+      "step": 65050
+    },
+    {
+      "epoch": 0.26285063248180934,
+      "grad_norm": 233.52798461914062,
+      "learning_rate": 5.422956657541224e-06,
+      "loss": 43.4715,
+      "step": 65060
+    },
+    {
+      "epoch": 0.262891033747177,
+      "grad_norm": 1021.7113037109375,
+      "learning_rate": 5.421551328797821e-06,
+      "loss": 47.4148,
+      "step": 65070
+    },
+    {
+      "epoch": 0.26293143501254457,
+      "grad_norm": 489.813232421875,
+      "learning_rate": 5.420145966513513e-06,
+      "loss": 41.6413,
+      "step": 65080
+    },
+    {
+      "epoch": 0.2629718362779122,
+      "grad_norm": 449.0360412597656,
+      "learning_rate": 5.418740570800117e-06,
+      "loss": 39.4725,
+      "step": 65090
+    },
+    {
+      "epoch": 0.26301223754327985,
+      "grad_norm": 737.3682250976562,
+      "learning_rate": 5.4173351417694575e-06,
+      "loss": 41.1901,
+      "step": 65100
+    },
+    {
+      "epoch": 0.2630526388086475,
+      "grad_norm": 249.3413848876953,
+      "learning_rate": 5.415929679533356e-06,
+      "loss": 28.5086,
+      "step": 65110
+    },
+    {
+      "epoch": 0.2630930400740151,
+      "grad_norm": 436.5331726074219,
+      "learning_rate": 5.414524184203638e-06,
+      "loss": 34.6025,
+      "step": 65120
+    },
+    {
+      "epoch": 0.26313344133938277,
+      "grad_norm": 617.8345947265625,
+      "learning_rate": 5.4131186558921335e-06,
+      "loss": 44.1296,
+      "step": 65130
+    },
+    {
+      "epoch": 0.2631738426047504,
+      "grad_norm": 291.6791076660156,
+      "learning_rate": 5.411713094710673e-06,
+      "loss": 39.0024,
+      "step": 65140
+    },
+    {
+      "epoch": 0.263214243870118,
+      "grad_norm": 1020.7041625976562,
+      "learning_rate": 5.410307500771092e-06,
+      "loss": 45.0434,
+      "step": 65150
+    },
+    {
+      "epoch": 0.26325464513548563,
+      "grad_norm": 360.7496643066406,
+      "learning_rate": 5.4089018741852264e-06,
+      "loss": 32.6432,
+      "step": 65160
+    },
+    {
+      "epoch": 0.26329504640085327,
+      "grad_norm": 577.2459716796875,
+      "learning_rate": 5.407496215064915e-06,
+      "loss": 56.6949,
+      "step": 65170
+    },
+    {
+      "epoch": 0.2633354476662209,
+      "grad_norm": 660.3246459960938,
+      "learning_rate": 5.406090523521999e-06,
+      "loss": 33.376,
+      "step": 65180
+    },
+    {
+      "epoch": 0.26337584893158855,
+      "grad_norm": 493.9849548339844,
+      "learning_rate": 5.404684799668325e-06,
+      "loss": 59.4395,
+      "step": 65190
+    },
+    {
+      "epoch": 0.2634162501969562,
+      "grad_norm": 951.3013916015625,
+      "learning_rate": 5.403279043615738e-06,
+      "loss": 43.7015,
+      "step": 65200
+    },
+    {
+      "epoch": 0.2634566514623238,
+      "grad_norm": 797.259765625,
+      "learning_rate": 5.4018732554760875e-06,
+      "loss": 53.7176,
+      "step": 65210
+    },
+    {
+      "epoch": 0.2634970527276914,
+      "grad_norm": 555.95166015625,
+      "learning_rate": 5.400467435361227e-06,
+      "loss": 39.394,
+      "step": 65220
+    },
+    {
+      "epoch": 0.26353745399305906,
+      "grad_norm": 962.2315063476562,
+      "learning_rate": 5.399061583383013e-06,
+      "loss": 50.1824,
+      "step": 65230
+    },
+    {
+      "epoch": 0.2635778552584267,
+      "grad_norm": 855.8276977539062,
+      "learning_rate": 5.3976556996532965e-06,
+      "loss": 41.223,
+      "step": 65240
+    },
+    {
+      "epoch": 0.26361825652379434,
+      "grad_norm": 402.3208923339844,
+      "learning_rate": 5.396249784283943e-06,
+      "loss": 43.6482,
+      "step": 65250
+    },
+    {
+      "epoch": 0.263658657789162,
+      "grad_norm": 281.7325744628906,
+      "learning_rate": 5.394843837386812e-06,
+      "loss": 41.1225,
+      "step": 65260
+    },
+    {
+      "epoch": 0.26369905905452956,
+      "grad_norm": 998.9751586914062,
+      "learning_rate": 5.39343785907377e-06,
+      "loss": 39.7433,
+      "step": 65270
+    },
+    {
+      "epoch": 0.2637394603198972,
+      "grad_norm": 305.1476745605469,
+      "learning_rate": 5.392031849456683e-06,
+      "loss": 41.8851,
+      "step": 65280
+    },
+    {
+      "epoch": 0.26377986158526484,
+      "grad_norm": 642.9463500976562,
+      "learning_rate": 5.39062580864742e-06,
+      "loss": 33.0668,
+      "step": 65290
+    },
+    {
+      "epoch": 0.2638202628506325,
+      "grad_norm": 427.274658203125,
+      "learning_rate": 5.3892197367578535e-06,
+      "loss": 34.9511,
+      "step": 65300
+    },
+    {
+      "epoch": 0.2638606641160001,
+      "grad_norm": 281.24468994140625,
+      "learning_rate": 5.38781363389986e-06,
+      "loss": 31.0351,
+      "step": 65310
+    },
+    {
+      "epoch": 0.26390106538136776,
+      "grad_norm": 542.3818969726562,
+      "learning_rate": 5.386407500185316e-06,
+      "loss": 50.7807,
+      "step": 65320
+    },
+    {
+      "epoch": 0.2639414666467354,
+      "grad_norm": 630.6954956054688,
+      "learning_rate": 5.3850013357261e-06,
+      "loss": 37.06,
+      "step": 65330
+    },
+    {
+      "epoch": 0.263981867912103,
+      "grad_norm": 566.5939331054688,
+      "learning_rate": 5.383595140634093e-06,
+      "loss": 40.1069,
+      "step": 65340
+    },
+    {
+      "epoch": 0.2640222691774706,
+      "grad_norm": 625.3710327148438,
+      "learning_rate": 5.382188915021182e-06,
+      "loss": 46.6871,
+      "step": 65350
+    },
+    {
+      "epoch": 0.26406267044283827,
+      "grad_norm": 676.7681274414062,
+      "learning_rate": 5.380782658999256e-06,
+      "loss": 45.0811,
+      "step": 65360
+    },
+    {
+      "epoch": 0.2641030717082059,
+      "grad_norm": 297.6866149902344,
+      "learning_rate": 5.379376372680199e-06,
+      "loss": 45.0418,
+      "step": 65370
+    },
+    {
+      "epoch": 0.26414347297357355,
+      "grad_norm": 513.9822998046875,
+      "learning_rate": 5.377970056175905e-06,
+      "loss": 57.6066,
+      "step": 65380
+    },
+    {
+      "epoch": 0.2641838742389412,
+      "grad_norm": 232.52333068847656,
+      "learning_rate": 5.376563709598267e-06,
+      "loss": 30.8778,
+      "step": 65390
+    },
+    {
+      "epoch": 0.26422427550430877,
+      "grad_norm": 665.5692749023438,
+      "learning_rate": 5.3751573330591855e-06,
+      "loss": 65.3008,
+      "step": 65400
+    },
+    {
+      "epoch": 0.2642646767696764,
+      "grad_norm": 789.9412841796875,
+      "learning_rate": 5.3737509266705555e-06,
+      "loss": 49.8782,
+      "step": 65410
+    },
+    {
+      "epoch": 0.26430507803504405,
+      "grad_norm": 532.4793090820312,
+      "learning_rate": 5.37234449054428e-06,
+      "loss": 29.4968,
+      "step": 65420
+    },
+    {
+      "epoch": 0.2643454793004117,
+      "grad_norm": 121.6689224243164,
+      "learning_rate": 5.370938024792262e-06,
+      "loss": 43.7967,
+      "step": 65430
+    },
+    {
+      "epoch": 0.26438588056577933,
+      "grad_norm": 1438.095458984375,
+      "learning_rate": 5.369531529526406e-06,
+      "loss": 49.5552,
+      "step": 65440
+    },
+    {
+      "epoch": 0.26442628183114697,
+      "grad_norm": 381.73846435546875,
+      "learning_rate": 5.3681250048586246e-06,
+      "loss": 39.8286,
+      "step": 65450
+    },
+    {
+      "epoch": 0.2644666830965146,
+      "grad_norm": 1419.7532958984375,
+      "learning_rate": 5.366718450900825e-06,
+      "loss": 47.5621,
+      "step": 65460
+    },
+    {
+      "epoch": 0.2645070843618822,
+      "grad_norm": 883.9287109375,
+      "learning_rate": 5.365311867764922e-06,
+      "loss": 44.4383,
+      "step": 65470
+    },
+    {
+      "epoch": 0.26454748562724983,
+      "grad_norm": 572.197509765625,
+      "learning_rate": 5.363905255562828e-06,
+      "loss": 54.7081,
+      "step": 65480
+    },
+    {
+      "epoch": 0.2645878868926175,
+      "grad_norm": 397.60784912109375,
+      "learning_rate": 5.362498614406466e-06,
+      "loss": 34.6413,
+      "step": 65490
+    },
+    {
+      "epoch": 0.2646282881579851,
+      "grad_norm": 561.8424682617188,
+      "learning_rate": 5.361091944407751e-06,
+      "loss": 33.9671,
+      "step": 65500
+    },
+    {
+      "epoch": 0.26466868942335275,
+      "grad_norm": 459.93792724609375,
+      "learning_rate": 5.3596852456786075e-06,
+      "loss": 40.9463,
+      "step": 65510
+    },
+    {
+      "epoch": 0.2647090906887204,
+      "grad_norm": 588.5122680664062,
+      "learning_rate": 5.35827851833096e-06,
+      "loss": 37.2149,
+      "step": 65520
+    },
+    {
+      "epoch": 0.264749491954088,
+      "grad_norm": 389.89019775390625,
+      "learning_rate": 5.356871762476735e-06,
+      "loss": 36.1884,
+      "step": 65530
+    },
+    {
+      "epoch": 0.2647898932194556,
+      "grad_norm": 387.892578125,
+      "learning_rate": 5.355464978227861e-06,
+      "loss": 35.113,
+      "step": 65540
+    },
+    {
+      "epoch": 0.26483029448482326,
+      "grad_norm": 621.6873168945312,
+      "learning_rate": 5.354058165696271e-06,
+      "loss": 35.7391,
+      "step": 65550
+    },
+    {
+      "epoch": 0.2648706957501909,
+      "grad_norm": 382.91497802734375,
+      "learning_rate": 5.352651324993897e-06,
+      "loss": 52.0832,
+      "step": 65560
+    },
+    {
+      "epoch": 0.26491109701555854,
+      "grad_norm": 902.5263061523438,
+      "learning_rate": 5.351244456232676e-06,
+      "loss": 51.0694,
+      "step": 65570
+    },
+    {
+      "epoch": 0.2649514982809262,
+      "grad_norm": 356.6571960449219,
+      "learning_rate": 5.349837559524546e-06,
+      "loss": 53.7859,
+      "step": 65580
+    },
+    {
+      "epoch": 0.26499189954629376,
+      "grad_norm": 486.88458251953125,
+      "learning_rate": 5.3484306349814455e-06,
+      "loss": 43.9213,
+      "step": 65590
+    },
+    {
+      "epoch": 0.2650323008116614,
+      "grad_norm": 660.9340209960938,
+      "learning_rate": 5.34702368271532e-06,
+      "loss": 58.3023,
+      "step": 65600
+    },
+    {
+      "epoch": 0.26507270207702904,
+      "grad_norm": 559.8971557617188,
+      "learning_rate": 5.345616702838111e-06,
+      "loss": 40.1814,
+      "step": 65610
+    },
+    {
+      "epoch": 0.2651131033423967,
+      "grad_norm": 541.8464965820312,
+      "learning_rate": 5.344209695461768e-06,
+      "loss": 35.0644,
+      "step": 65620
+    },
+    {
+      "epoch": 0.2651535046077643,
+      "grad_norm": 663.918701171875,
+      "learning_rate": 5.3428026606982396e-06,
+      "loss": 60.276,
+      "step": 65630
+    },
+    {
+      "epoch": 0.26519390587313196,
+      "grad_norm": 262.1153869628906,
+      "learning_rate": 5.341395598659477e-06,
+      "loss": 47.1914,
+      "step": 65640
+    },
+    {
+      "epoch": 0.2652343071384996,
+      "grad_norm": 1048.3524169921875,
+      "learning_rate": 5.339988509457432e-06,
+      "loss": 48.2345,
+      "step": 65650
+    },
+    {
+      "epoch": 0.2652747084038672,
+      "grad_norm": 839.9951171875,
+      "learning_rate": 5.338581393204064e-06,
+      "loss": 60.1043,
+      "step": 65660
+    },
+    {
+      "epoch": 0.2653151096692348,
+      "grad_norm": 798.4373779296875,
+      "learning_rate": 5.337174250011326e-06,
+      "loss": 51.318,
+      "step": 65670
+    },
+    {
+      "epoch": 0.26535551093460247,
+      "grad_norm": 267.1536560058594,
+      "learning_rate": 5.3357670799911805e-06,
+      "loss": 37.1753,
+      "step": 65680
+    },
+    {
+      "epoch": 0.2653959121999701,
+      "grad_norm": 775.5031127929688,
+      "learning_rate": 5.334359883255591e-06,
+      "loss": 41.825,
+      "step": 65690
+    },
+    {
+      "epoch": 0.26543631346533775,
+      "grad_norm": 552.0604858398438,
+      "learning_rate": 5.33295265991652e-06,
+      "loss": 40.3956,
+      "step": 65700
+    },
+    {
+      "epoch": 0.2654767147307054,
+      "grad_norm": 814.4728393554688,
+      "learning_rate": 5.331545410085933e-06,
+      "loss": 32.473,
+      "step": 65710
+    },
+    {
+      "epoch": 0.26551711599607297,
+      "grad_norm": 706.5919799804688,
+      "learning_rate": 5.330138133875799e-06,
+      "loss": 42.804,
+      "step": 65720
+    },
+    {
+      "epoch": 0.2655575172614406,
+      "grad_norm": 1015.4716186523438,
+      "learning_rate": 5.328730831398089e-06,
+      "loss": 61.0056,
+      "step": 65730
+    },
+    {
+      "epoch": 0.26559791852680825,
+      "grad_norm": 525.5064697265625,
+      "learning_rate": 5.3273235027647764e-06,
+      "loss": 37.2049,
+      "step": 65740
+    },
+    {
+      "epoch": 0.2656383197921759,
+      "grad_norm": 556.1926879882812,
+      "learning_rate": 5.3259161480878354e-06,
+      "loss": 51.9938,
+      "step": 65750
+    },
+    {
+      "epoch": 0.26567872105754353,
+      "grad_norm": 1342.614990234375,
+      "learning_rate": 5.324508767479239e-06,
+      "loss": 34.7253,
+      "step": 65760
+    },
+    {
+      "epoch": 0.26571912232291117,
+      "grad_norm": 541.5657958984375,
+      "learning_rate": 5.323101361050972e-06,
+      "loss": 47.4547,
+      "step": 65770
+    },
+    {
+      "epoch": 0.2657595235882788,
+      "grad_norm": 827.6627197265625,
+      "learning_rate": 5.321693928915012e-06,
+      "loss": 72.2926,
+      "step": 65780
+    },
+    {
+      "epoch": 0.2657999248536464,
+      "grad_norm": 1144.3697509765625,
+      "learning_rate": 5.320286471183343e-06,
+      "loss": 42.7462,
+      "step": 65790
+    },
+    {
+      "epoch": 0.26584032611901404,
+      "grad_norm": 361.05267333984375,
+      "learning_rate": 5.3188789879679496e-06,
+      "loss": 56.6374,
+      "step": 65800
+    },
+    {
+      "epoch": 0.2658807273843817,
+      "grad_norm": 448.1735534667969,
+      "learning_rate": 5.317471479380816e-06,
+      "loss": 55.7517,
+      "step": 65810
+    },
+    {
+      "epoch": 0.2659211286497493,
+      "grad_norm": 377.99957275390625,
+      "learning_rate": 5.3160639455339355e-06,
+      "loss": 48.2162,
+      "step": 65820
+    },
+    {
+      "epoch": 0.26596152991511696,
+      "grad_norm": 236.86538696289062,
+      "learning_rate": 5.314656386539298e-06,
+      "loss": 38.0658,
+      "step": 65830
+    },
+    {
+      "epoch": 0.2660019311804846,
+      "grad_norm": 929.2343139648438,
+      "learning_rate": 5.313248802508896e-06,
+      "loss": 41.3343,
+      "step": 65840
+    },
+    {
+      "epoch": 0.2660423324458522,
+      "grad_norm": 566.799072265625,
+      "learning_rate": 5.311841193554723e-06,
+      "loss": 36.5541,
+      "step": 65850
+    },
+    {
+      "epoch": 0.2660827337112198,
+      "grad_norm": 549.5016479492188,
+      "learning_rate": 5.310433559788778e-06,
+      "loss": 44.7664,
+      "step": 65860
+    },
+    {
+      "epoch": 0.26612313497658746,
+      "grad_norm": 476.6116638183594,
+      "learning_rate": 5.309025901323059e-06,
+      "loss": 32.1816,
+      "step": 65870
+    },
+    {
+      "epoch": 0.2661635362419551,
+      "grad_norm": 607.8992919921875,
+      "learning_rate": 5.307618218269569e-06,
+      "loss": 43.7821,
+      "step": 65880
+    },
+    {
+      "epoch": 0.26620393750732274,
+      "grad_norm": 1422.3590087890625,
+      "learning_rate": 5.306210510740307e-06,
+      "loss": 44.0493,
+      "step": 65890
+    },
+    {
+      "epoch": 0.2662443387726904,
+      "grad_norm": 651.806396484375,
+      "learning_rate": 5.304802778847281e-06,
+      "loss": 46.3512,
+      "step": 65900
+    },
+    {
+      "epoch": 0.26628474003805797,
+      "grad_norm": 414.97845458984375,
+      "learning_rate": 5.303395022702495e-06,
+      "loss": 59.5663,
+      "step": 65910
+    },
+    {
+      "epoch": 0.2663251413034256,
+      "grad_norm": 1217.07958984375,
+      "learning_rate": 5.301987242417963e-06,
+      "loss": 45.3674,
+      "step": 65920
+    },
+    {
+      "epoch": 0.26636554256879325,
+      "grad_norm": 584.6605224609375,
+      "learning_rate": 5.300579438105689e-06,
+      "loss": 47.7693,
+      "step": 65930
+    },
+    {
+      "epoch": 0.2664059438341609,
+      "grad_norm": 273.66412353515625,
+      "learning_rate": 5.29917160987769e-06,
+      "loss": 48.6378,
+      "step": 65940
+    },
+    {
+      "epoch": 0.2664463450995285,
+      "grad_norm": 497.5111999511719,
+      "learning_rate": 5.297763757845979e-06,
+      "loss": 30.083,
+      "step": 65950
+    },
+    {
+      "epoch": 0.26648674636489617,
+      "grad_norm": 574.8904418945312,
+      "learning_rate": 5.296355882122572e-06,
+      "loss": 36.8044,
+      "step": 65960
+    },
+    {
+      "epoch": 0.2665271476302638,
+      "grad_norm": 411.0905456542969,
+      "learning_rate": 5.294947982819488e-06,
+      "loss": 48.731,
+      "step": 65970
+    },
+    {
+      "epoch": 0.2665675488956314,
+      "grad_norm": 511.0521545410156,
+      "learning_rate": 5.293540060048746e-06,
+      "loss": 41.3318,
+      "step": 65980
+    },
+    {
+      "epoch": 0.26660795016099903,
+      "grad_norm": 740.4434204101562,
+      "learning_rate": 5.292132113922369e-06,
+      "loss": 47.0271,
+      "step": 65990
+    },
+    {
+      "epoch": 0.26664835142636667,
+      "grad_norm": 319.0716247558594,
+      "learning_rate": 5.290724144552379e-06,
+      "loss": 30.8917,
+      "step": 66000
+    },
+    {
+      "epoch": 0.2666887526917343,
+      "grad_norm": 979.0654907226562,
+      "learning_rate": 5.2893161520508055e-06,
+      "loss": 44.61,
+      "step": 66010
+    },
+    {
+      "epoch": 0.26672915395710195,
+      "grad_norm": 736.4896240234375,
+      "learning_rate": 5.287908136529671e-06,
+      "loss": 61.8157,
+      "step": 66020
+    },
+    {
+      "epoch": 0.2667695552224696,
+      "grad_norm": 523.2166137695312,
+      "learning_rate": 5.28650009810101e-06,
+      "loss": 49.0017,
+      "step": 66030
+    },
+    {
+      "epoch": 0.2668099564878372,
+      "grad_norm": 593.52685546875,
+      "learning_rate": 5.28509203687685e-06,
+      "loss": 42.2598,
+      "step": 66040
+    },
+    {
+      "epoch": 0.2668503577532048,
+      "grad_norm": 732.2353515625,
+      "learning_rate": 5.283683952969224e-06,
+      "loss": 37.1897,
+      "step": 66050
+    },
+    {
+      "epoch": 0.26689075901857245,
+      "grad_norm": 608.0689086914062,
+      "learning_rate": 5.282275846490169e-06,
+      "loss": 42.6663,
+      "step": 66060
+    },
+    {
+      "epoch": 0.2669311602839401,
+      "grad_norm": 514.1216430664062,
+      "learning_rate": 5.280867717551719e-06,
+      "loss": 41.3538,
+      "step": 66070
+    },
+    {
+      "epoch": 0.26697156154930773,
+      "grad_norm": 489.35546875,
+      "learning_rate": 5.279459566265915e-06,
+      "loss": 42.4444,
+      "step": 66080
+    },
+    {
+      "epoch": 0.2670119628146754,
+      "grad_norm": 850.6774291992188,
+      "learning_rate": 5.278051392744796e-06,
+      "loss": 40.5687,
+      "step": 66090
+    },
+    {
+      "epoch": 0.267052364080043,
+      "grad_norm": 546.3876953125,
+      "learning_rate": 5.2766431971004025e-06,
+      "loss": 43.6863,
+      "step": 66100
+    },
+    {
+      "epoch": 0.2670927653454106,
+      "grad_norm": 303.21832275390625,
+      "learning_rate": 5.275234979444781e-06,
+      "loss": 58.7996,
+      "step": 66110
+    },
+    {
+      "epoch": 0.26713316661077824,
+      "grad_norm": 436.77923583984375,
+      "learning_rate": 5.273826739889975e-06,
+      "loss": 37.3343,
+      "step": 66120
+    },
+    {
+      "epoch": 0.2671735678761459,
+      "grad_norm": 225.6940460205078,
+      "learning_rate": 5.272418478548031e-06,
+      "loss": 51.1029,
+      "step": 66130
+    },
+    {
+      "epoch": 0.2672139691415135,
+      "grad_norm": 1763.029296875,
+      "learning_rate": 5.271010195530999e-06,
+      "loss": 47.1112,
+      "step": 66140
+    },
+    {
+      "epoch": 0.26725437040688116,
+      "grad_norm": 422.47314453125,
+      "learning_rate": 5.26960189095093e-06,
+      "loss": 31.8788,
+      "step": 66150
+    },
+    {
+      "epoch": 0.2672947716722488,
+      "grad_norm": 424.02606201171875,
+      "learning_rate": 5.268193564919876e-06,
+      "loss": 37.1531,
+      "step": 66160
+    },
+    {
+      "epoch": 0.2673351729376164,
+      "grad_norm": 672.9403686523438,
+      "learning_rate": 5.26678521754989e-06,
+      "loss": 34.8339,
+      "step": 66170
+    },
+    {
+      "epoch": 0.267375574202984,
+      "grad_norm": 486.558349609375,
+      "learning_rate": 5.265376848953031e-06,
+      "loss": 35.5,
+      "step": 66180
+    },
+    {
+      "epoch": 0.26741597546835166,
+      "grad_norm": 749.306640625,
+      "learning_rate": 5.263968459241351e-06,
+      "loss": 52.8159,
+      "step": 66190
+    },
+    {
+      "epoch": 0.2674563767337193,
+      "grad_norm": 346.21722412109375,
+      "learning_rate": 5.262560048526913e-06,
+      "loss": 40.2258,
+      "step": 66200
+    },
+    {
+      "epoch": 0.26749677799908694,
+      "grad_norm": 826.6575317382812,
+      "learning_rate": 5.261151616921778e-06,
+      "loss": 36.5579,
+      "step": 66210
+    },
+    {
+      "epoch": 0.2675371792644546,
+      "grad_norm": 593.7576293945312,
+      "learning_rate": 5.259743164538008e-06,
+      "loss": 55.5614,
+      "step": 66220
+    },
+    {
+      "epoch": 0.26757758052982217,
+      "grad_norm": 535.8445434570312,
+      "learning_rate": 5.2583346914876655e-06,
+      "loss": 45.7979,
+      "step": 66230
+    },
+    {
+      "epoch": 0.2676179817951898,
+      "grad_norm": 521.997314453125,
+      "learning_rate": 5.2569261978828155e-06,
+      "loss": 48.649,
+      "step": 66240
+    },
+    {
+      "epoch": 0.26765838306055745,
+      "grad_norm": 961.8884887695312,
+      "learning_rate": 5.255517683835528e-06,
+      "loss": 52.0874,
+      "step": 66250
+    },
+    {
+      "epoch": 0.2676987843259251,
+      "grad_norm": 573.283935546875,
+      "learning_rate": 5.254109149457873e-06,
+      "loss": 40.6045,
+      "step": 66260
+    },
+    {
+      "epoch": 0.2677391855912927,
+      "grad_norm": 510.8491516113281,
+      "learning_rate": 5.252700594861918e-06,
+      "loss": 48.0254,
+      "step": 66270
+    },
+    {
+      "epoch": 0.26777958685666037,
+      "grad_norm": 4054.195556640625,
+      "learning_rate": 5.251292020159736e-06,
+      "loss": 66.1508,
+      "step": 66280
+    },
+    {
+      "epoch": 0.267819988122028,
+      "grad_norm": 983.9483642578125,
+      "learning_rate": 5.2498834254634005e-06,
+      "loss": 59.9476,
+      "step": 66290
+    },
+    {
+      "epoch": 0.2678603893873956,
+      "grad_norm": 531.3615112304688,
+      "learning_rate": 5.248474810884988e-06,
+      "loss": 62.1785,
+      "step": 66300
+    },
+    {
+      "epoch": 0.26790079065276323,
+      "grad_norm": 438.48516845703125,
+      "learning_rate": 5.247066176536577e-06,
+      "loss": 33.2004,
+      "step": 66310
+    },
+    {
+      "epoch": 0.26794119191813087,
+      "grad_norm": 0.0,
+      "learning_rate": 5.245657522530243e-06,
+      "loss": 44.228,
+      "step": 66320
+    },
+    {
+      "epoch": 0.2679815931834985,
+      "grad_norm": 299.0640563964844,
+      "learning_rate": 5.244248848978067e-06,
+      "loss": 40.0723,
+      "step": 66330
+    },
+    {
+      "epoch": 0.26802199444886615,
+      "grad_norm": 297.197509765625,
+      "learning_rate": 5.242840155992131e-06,
+      "loss": 53.5966,
+      "step": 66340
+    },
+    {
+      "epoch": 0.2680623957142338,
+      "grad_norm": 772.6343994140625,
+      "learning_rate": 5.24143144368452e-06,
+      "loss": 50.8811,
+      "step": 66350
+    },
+    {
+      "epoch": 0.2681027969796014,
+      "grad_norm": 720.2068481445312,
+      "learning_rate": 5.240022712167315e-06,
+      "loss": 69.1438,
+      "step": 66360
+    },
+    {
+      "epoch": 0.268143198244969,
+      "grad_norm": 718.2180786132812,
+      "learning_rate": 5.2386139615526046e-06,
+      "loss": 37.676,
+      "step": 66370
+    },
+    {
+      "epoch": 0.26818359951033666,
+      "grad_norm": 858.3838500976562,
+      "learning_rate": 5.237205191952477e-06,
+      "loss": 31.7104,
+      "step": 66380
+    },
+    {
+      "epoch": 0.2682240007757043,
+      "grad_norm": 808.7693481445312,
+      "learning_rate": 5.235796403479021e-06,
+      "loss": 43.2298,
+      "step": 66390
+    },
+    {
+      "epoch": 0.26826440204107194,
+      "grad_norm": 3157.16015625,
+      "learning_rate": 5.2343875962443255e-06,
+      "loss": 61.2995,
+      "step": 66400
+    },
+    {
+      "epoch": 0.2683048033064396,
+      "grad_norm": 784.2205810546875,
+      "learning_rate": 5.2329787703604875e-06,
+      "loss": 50.4512,
+      "step": 66410
+    },
+    {
+      "epoch": 0.2683452045718072,
+      "grad_norm": 592.7188720703125,
+      "learning_rate": 5.231569925939596e-06,
+      "loss": 37.0912,
+      "step": 66420
+    },
+    {
+      "epoch": 0.2683856058371748,
+      "grad_norm": 523.109619140625,
+      "learning_rate": 5.230161063093749e-06,
+      "loss": 52.2018,
+      "step": 66430
+    },
+    {
+      "epoch": 0.26842600710254244,
+      "grad_norm": 605.025146484375,
+      "learning_rate": 5.228752181935042e-06,
+      "loss": 52.8809,
+      "step": 66440
+    },
+    {
+      "epoch": 0.2684664083679101,
+      "grad_norm": 201.62841796875,
+      "learning_rate": 5.227343282575574e-06,
+      "loss": 39.6615,
+      "step": 66450
+    },
+    {
+      "epoch": 0.2685068096332777,
+      "grad_norm": 899.1422729492188,
+      "learning_rate": 5.225934365127445e-06,
+      "loss": 44.044,
+      "step": 66460
+    },
+    {
+      "epoch": 0.26854721089864536,
+      "grad_norm": 980.2147216796875,
+      "learning_rate": 5.224525429702755e-06,
+      "loss": 35.6335,
+      "step": 66470
+    },
+    {
+      "epoch": 0.268587612164013,
+      "grad_norm": 800.3404541015625,
+      "learning_rate": 5.223116476413606e-06,
+      "loss": 34.9568,
+      "step": 66480
+    },
+    {
+      "epoch": 0.2686280134293806,
+      "grad_norm": 373.1787109375,
+      "learning_rate": 5.221707505372105e-06,
+      "loss": 50.0092,
+      "step": 66490
+    },
+    {
+      "epoch": 0.2686684146947482,
+      "grad_norm": 548.1878051757812,
+      "learning_rate": 5.220298516690353e-06,
+      "loss": 51.9183,
+      "step": 66500
+    },
+    {
+      "epoch": 0.26870881596011587,
+      "grad_norm": 546.2302856445312,
+      "learning_rate": 5.21888951048046e-06,
+      "loss": 46.6924,
+      "step": 66510
+    },
+    {
+      "epoch": 0.2687492172254835,
+      "grad_norm": 404.8308410644531,
+      "learning_rate": 5.217480486854534e-06,
+      "loss": 42.3545,
+      "step": 66520
+    },
+    {
+      "epoch": 0.26878961849085115,
+      "grad_norm": 787.6769409179688,
+      "learning_rate": 5.216071445924683e-06,
+      "loss": 50.6945,
+      "step": 66530
+    },
+    {
+      "epoch": 0.2688300197562188,
+      "grad_norm": 637.3775634765625,
+      "learning_rate": 5.214662387803019e-06,
+      "loss": 34.9455,
+      "step": 66540
+    },
+    {
+      "epoch": 0.26887042102158637,
+      "grad_norm": 840.2587280273438,
+      "learning_rate": 5.213253312601654e-06,
+      "loss": 48.897,
+      "step": 66550
+    },
+    {
+      "epoch": 0.268910822286954,
+      "grad_norm": 668.3641967773438,
+      "learning_rate": 5.211844220432702e-06,
+      "loss": 60.593,
+      "step": 66560
+    },
+    {
+      "epoch": 0.26895122355232165,
+      "grad_norm": 520.17626953125,
+      "learning_rate": 5.210435111408276e-06,
+      "loss": 47.4215,
+      "step": 66570
+    },
+    {
+      "epoch": 0.2689916248176893,
+      "grad_norm": 718.3225708007812,
+      "learning_rate": 5.209025985640496e-06,
+      "loss": 43.8603,
+      "step": 66580
+    },
+    {
+      "epoch": 0.26903202608305693,
+      "grad_norm": 857.1797485351562,
+      "learning_rate": 5.207616843241476e-06,
+      "loss": 59.8737,
+      "step": 66590
+    },
+    {
+      "epoch": 0.26907242734842457,
+      "grad_norm": 0.0,
+      "learning_rate": 5.206207684323337e-06,
+      "loss": 47.3329,
+      "step": 66600
+    },
+    {
+      "epoch": 0.2691128286137922,
+      "grad_norm": 804.732666015625,
+      "learning_rate": 5.2047985089982e-06,
+      "loss": 50.3234,
+      "step": 66610
+    },
+    {
+      "epoch": 0.2691532298791598,
+      "grad_norm": 600.4869995117188,
+      "learning_rate": 5.203389317378183e-06,
+      "loss": 35.6028,
+      "step": 66620
+    },
+    {
+      "epoch": 0.26919363114452743,
+      "grad_norm": 507.7207336425781,
+      "learning_rate": 5.201980109575414e-06,
+      "loss": 37.8279,
+      "step": 66630
+    },
+    {
+      "epoch": 0.2692340324098951,
+      "grad_norm": 1408.29248046875,
+      "learning_rate": 5.200570885702013e-06,
+      "loss": 49.3994,
+      "step": 66640
+    },
+    {
+      "epoch": 0.2692744336752627,
+      "grad_norm": 756.8723754882812,
+      "learning_rate": 5.19916164587011e-06,
+      "loss": 50.934,
+      "step": 66650
+    },
+    {
+      "epoch": 0.26931483494063035,
+      "grad_norm": 715.3894653320312,
+      "learning_rate": 5.197752390191827e-06,
+      "loss": 53.1523,
+      "step": 66660
+    },
+    {
+      "epoch": 0.269355236205998,
+      "grad_norm": 523.8912353515625,
+      "learning_rate": 5.196343118779292e-06,
+      "loss": 35.7084,
+      "step": 66670
+    },
+    {
+      "epoch": 0.2693956374713656,
+      "grad_norm": 721.2276000976562,
+      "learning_rate": 5.194933831744637e-06,
+      "loss": 47.4896,
+      "step": 66680
+    },
+    {
+      "epoch": 0.2694360387367332,
+      "grad_norm": 588.867919921875,
+      "learning_rate": 5.1935245291999945e-06,
+      "loss": 40.2639,
+      "step": 66690
+    },
+    {
+      "epoch": 0.26947644000210086,
+      "grad_norm": 624.7872924804688,
+      "learning_rate": 5.192115211257491e-06,
+      "loss": 44.807,
+      "step": 66700
+    },
+    {
+      "epoch": 0.2695168412674685,
+      "grad_norm": 458.06085205078125,
+      "learning_rate": 5.19070587802926e-06,
+      "loss": 36.0,
+      "step": 66710
+    },
+    {
+      "epoch": 0.26955724253283614,
+      "grad_norm": 344.6346435546875,
+      "learning_rate": 5.189296529627441e-06,
+      "loss": 40.4112,
+      "step": 66720
+    },
+    {
+      "epoch": 0.2695976437982038,
+      "grad_norm": 421.5151062011719,
+      "learning_rate": 5.187887166164165e-06,
+      "loss": 45.2732,
+      "step": 66730
+    },
+    {
+      "epoch": 0.2696380450635714,
+      "grad_norm": 1014.9851684570312,
+      "learning_rate": 5.186477787751569e-06,
+      "loss": 68.0941,
+      "step": 66740
+    },
+    {
+      "epoch": 0.269678446328939,
+      "grad_norm": 277.7436828613281,
+      "learning_rate": 5.185068394501791e-06,
+      "loss": 33.1461,
+      "step": 66750
+    },
+    {
+      "epoch": 0.26971884759430664,
+      "grad_norm": 788.6626586914062,
+      "learning_rate": 5.183658986526969e-06,
+      "loss": 59.109,
+      "step": 66760
+    },
+    {
+      "epoch": 0.2697592488596743,
+      "grad_norm": 823.033935546875,
+      "learning_rate": 5.1822495639392465e-06,
+      "loss": 43.8207,
+      "step": 66770
+    },
+    {
+      "epoch": 0.2697996501250419,
+      "grad_norm": 388.10076904296875,
+      "learning_rate": 5.180840126850764e-06,
+      "loss": 26.7396,
+      "step": 66780
+    },
+    {
+      "epoch": 0.26984005139040956,
+      "grad_norm": 616.7316284179688,
+      "learning_rate": 5.179430675373659e-06,
+      "loss": 41.6738,
+      "step": 66790
+    },
+    {
+      "epoch": 0.2698804526557772,
+      "grad_norm": 395.72637939453125,
+      "learning_rate": 5.17802120962008e-06,
+      "loss": 36.2188,
+      "step": 66800
+    },
+    {
+      "epoch": 0.2699208539211448,
+      "grad_norm": 534.4721069335938,
+      "learning_rate": 5.17661172970217e-06,
+      "loss": 55.1734,
+      "step": 66810
+    },
+    {
+      "epoch": 0.2699612551865124,
+      "grad_norm": 718.8760986328125,
+      "learning_rate": 5.175202235732077e-06,
+      "loss": 47.2039,
+      "step": 66820
+    },
+    {
+      "epoch": 0.27000165645188007,
+      "grad_norm": 181.80419921875,
+      "learning_rate": 5.1737927278219446e-06,
+      "loss": 54.5886,
+      "step": 66830
+    },
+    {
+      "epoch": 0.2700420577172477,
+      "grad_norm": 657.7708740234375,
+      "learning_rate": 5.1723832060839216e-06,
+      "loss": 43.6877,
+      "step": 66840
+    },
+    {
+      "epoch": 0.27008245898261535,
+      "grad_norm": 582.9547119140625,
+      "learning_rate": 5.170973670630159e-06,
+      "loss": 32.2856,
+      "step": 66850
+    },
+    {
+      "epoch": 0.270122860247983,
+      "grad_norm": 908.3861083984375,
+      "learning_rate": 5.169564121572806e-06,
+      "loss": 51.184,
+      "step": 66860
+    },
+    {
+      "epoch": 0.27016326151335057,
+      "grad_norm": 933.0931396484375,
+      "learning_rate": 5.168154559024014e-06,
+      "loss": 37.9435,
+      "step": 66870
+    },
+    {
+      "epoch": 0.2702036627787182,
+      "grad_norm": 288.22430419921875,
+      "learning_rate": 5.166744983095937e-06,
+      "loss": 56.0103,
+      "step": 66880
+    },
+    {
+      "epoch": 0.27024406404408585,
+      "grad_norm": 686.4309692382812,
+      "learning_rate": 5.165335393900726e-06,
+      "loss": 34.439,
+      "step": 66890
+    },
+    {
+      "epoch": 0.2702844653094535,
+      "grad_norm": 558.5487060546875,
+      "learning_rate": 5.163925791550536e-06,
+      "loss": 33.7994,
+      "step": 66900
+    },
+    {
+      "epoch": 0.27032486657482113,
+      "grad_norm": 720.1372680664062,
+      "learning_rate": 5.162516176157523e-06,
+      "loss": 53.6696,
+      "step": 66910
+    },
+    {
+      "epoch": 0.27036526784018877,
+      "grad_norm": 512.8276977539062,
+      "learning_rate": 5.161106547833843e-06,
+      "loss": 38.8246,
+      "step": 66920
+    },
+    {
+      "epoch": 0.2704056691055564,
+      "grad_norm": 397.11785888671875,
+      "learning_rate": 5.159696906691656e-06,
+      "loss": 35.0052,
+      "step": 66930
+    },
+    {
+      "epoch": 0.270446070370924,
+      "grad_norm": 753.915771484375,
+      "learning_rate": 5.158287252843118e-06,
+      "loss": 48.151,
+      "step": 66940
+    },
+    {
+      "epoch": 0.27048647163629164,
+      "grad_norm": 870.271240234375,
+      "learning_rate": 5.1568775864003894e-06,
+      "loss": 64.0858,
+      "step": 66950
+    },
+    {
+      "epoch": 0.2705268729016593,
+      "grad_norm": 715.2255859375,
+      "learning_rate": 5.155467907475632e-06,
+      "loss": 42.3206,
+      "step": 66960
+    },
+    {
+      "epoch": 0.2705672741670269,
+      "grad_norm": 336.386962890625,
+      "learning_rate": 5.154058216181007e-06,
+      "loss": 32.85,
+      "step": 66970
+    },
+    {
+      "epoch": 0.27060767543239456,
+      "grad_norm": 1711.7962646484375,
+      "learning_rate": 5.1526485126286766e-06,
+      "loss": 45.585,
+      "step": 66980
+    },
+    {
+      "epoch": 0.2706480766977622,
+      "grad_norm": 689.0267944335938,
+      "learning_rate": 5.151238796930804e-06,
+      "loss": 42.3475,
+      "step": 66990
+    },
+    {
+      "epoch": 0.2706884779631298,
+      "grad_norm": 0.0,
+      "learning_rate": 5.149829069199555e-06,
+      "loss": 36.7057,
+      "step": 67000
+    },
+    {
+      "epoch": 0.2707288792284974,
+      "grad_norm": 492.4189758300781,
+      "learning_rate": 5.148419329547094e-06,
+      "loss": 41.9212,
+      "step": 67010
+    },
+    {
+      "epoch": 0.27076928049386506,
+      "grad_norm": 746.7906494140625,
+      "learning_rate": 5.147009578085589e-06,
+      "loss": 37.3403,
+      "step": 67020
+    },
+    {
+      "epoch": 0.2708096817592327,
+      "grad_norm": 696.8953247070312,
+      "learning_rate": 5.145599814927205e-06,
+      "loss": 48.6399,
+      "step": 67030
+    },
+    {
+      "epoch": 0.27085008302460034,
+      "grad_norm": 1013.6634521484375,
+      "learning_rate": 5.144190040184114e-06,
+      "loss": 36.1067,
+      "step": 67040
+    },
+    {
+      "epoch": 0.270890484289968,
+      "grad_norm": 962.2171630859375,
+      "learning_rate": 5.142780253968481e-06,
+      "loss": 29.0633,
+      "step": 67050
+    },
+    {
+      "epoch": 0.27093088555533557,
+      "grad_norm": 683.1814575195312,
+      "learning_rate": 5.14137045639248e-06,
+      "loss": 36.5833,
+      "step": 67060
+    },
+    {
+      "epoch": 0.2709712868207032,
+      "grad_norm": 419.5286560058594,
+      "learning_rate": 5.13996064756828e-06,
+      "loss": 36.7421,
+      "step": 67070
+    },
+    {
+      "epoch": 0.27101168808607085,
+      "grad_norm": 428.304931640625,
+      "learning_rate": 5.138550827608055e-06,
+      "loss": 28.9419,
+      "step": 67080
+    },
+    {
+      "epoch": 0.2710520893514385,
+      "grad_norm": 347.9562072753906,
+      "learning_rate": 5.137140996623975e-06,
+      "loss": 49.9478,
+      "step": 67090
+    },
+    {
+      "epoch": 0.2710924906168061,
+      "grad_norm": 354.6157531738281,
+      "learning_rate": 5.135731154728215e-06,
+      "loss": 46.6038,
+      "step": 67100
+    },
+    {
+      "epoch": 0.27113289188217377,
+      "grad_norm": 589.3129272460938,
+      "learning_rate": 5.134321302032951e-06,
+      "loss": 42.932,
+      "step": 67110
+    },
+    {
+      "epoch": 0.2711732931475414,
+      "grad_norm": 361.30029296875,
+      "learning_rate": 5.1329114386503585e-06,
+      "loss": 42.2918,
+      "step": 67120
+    },
+    {
+      "epoch": 0.271213694412909,
+      "grad_norm": 420.7300720214844,
+      "learning_rate": 5.131501564692611e-06,
+      "loss": 51.0995,
+      "step": 67130
+    },
+    {
+      "epoch": 0.27125409567827663,
+      "grad_norm": 478.6790466308594,
+      "learning_rate": 5.130091680271887e-06,
+      "loss": 64.9722,
+      "step": 67140
+    },
+    {
+      "epoch": 0.27129449694364427,
+      "grad_norm": 850.8605346679688,
+      "learning_rate": 5.128681785500365e-06,
+      "loss": 60.1893,
+      "step": 67150
+    },
+    {
+      "epoch": 0.2713348982090119,
+      "grad_norm": 451.3896484375,
+      "learning_rate": 5.127271880490227e-06,
+      "loss": 31.8813,
+      "step": 67160
+    },
+    {
+      "epoch": 0.27137529947437955,
+      "grad_norm": 980.57177734375,
+      "learning_rate": 5.125861965353647e-06,
+      "loss": 42.3975,
+      "step": 67170
+    },
+    {
+      "epoch": 0.2714157007397472,
+      "grad_norm": 517.5213623046875,
+      "learning_rate": 5.124452040202809e-06,
+      "loss": 52.2581,
+      "step": 67180
+    },
+    {
+      "epoch": 0.2714561020051148,
+      "grad_norm": 550.5774536132812,
+      "learning_rate": 5.1230421051498914e-06,
+      "loss": 36.1163,
+      "step": 67190
+    },
+    {
+      "epoch": 0.2714965032704824,
+      "grad_norm": 427.63421630859375,
+      "learning_rate": 5.121632160307078e-06,
+      "loss": 48.8638,
+      "step": 67200
+    },
+    {
+      "epoch": 0.27153690453585005,
+      "grad_norm": 471.3889465332031,
+      "learning_rate": 5.120222205786556e-06,
+      "loss": 44.1868,
+      "step": 67210
+    },
+    {
+      "epoch": 0.2715773058012177,
+      "grad_norm": 610.775390625,
+      "learning_rate": 5.118812241700501e-06,
+      "loss": 45.4912,
+      "step": 67220
+    },
+    {
+      "epoch": 0.27161770706658533,
+      "grad_norm": 654.1630859375,
+      "learning_rate": 5.117402268161103e-06,
+      "loss": 39.7674,
+      "step": 67230
+    },
+    {
+      "epoch": 0.271658108331953,
+      "grad_norm": 311.9112243652344,
+      "learning_rate": 5.115992285280543e-06,
+      "loss": 46.7622,
+      "step": 67240
+    },
+    {
+      "epoch": 0.2716985095973206,
+      "grad_norm": 916.1666870117188,
+      "learning_rate": 5.114582293171012e-06,
+      "loss": 45.3463,
+      "step": 67250
+    },
+    {
+      "epoch": 0.2717389108626882,
+      "grad_norm": 538.2660522460938,
+      "learning_rate": 5.113172291944693e-06,
+      "loss": 37.8112,
+      "step": 67260
+    },
+    {
+      "epoch": 0.27177931212805584,
+      "grad_norm": 652.853515625,
+      "learning_rate": 5.111762281713773e-06,
+      "loss": 50.5335,
+      "step": 67270
+    },
+    {
+      "epoch": 0.2718197133934235,
+      "grad_norm": 459.0245056152344,
+      "learning_rate": 5.110352262590442e-06,
+      "loss": 53.5647,
+      "step": 67280
+    },
+    {
+      "epoch": 0.2718601146587911,
+      "grad_norm": 322.7192687988281,
+      "learning_rate": 5.108942234686889e-06,
+      "loss": 28.7373,
+      "step": 67290
+    },
+    {
+      "epoch": 0.27190051592415876,
+      "grad_norm": 364.63555908203125,
+      "learning_rate": 5.1075321981153014e-06,
+      "loss": 39.3614,
+      "step": 67300
+    },
+    {
+      "epoch": 0.2719409171895264,
+      "grad_norm": 842.6339111328125,
+      "learning_rate": 5.106122152987869e-06,
+      "loss": 45.2849,
+      "step": 67310
+    },
+    {
+      "epoch": 0.271981318454894,
+      "grad_norm": 477.67950439453125,
+      "learning_rate": 5.1047120994167855e-06,
+      "loss": 51.1875,
+      "step": 67320
+    },
+    {
+      "epoch": 0.2720217197202616,
+      "grad_norm": 544.8759155273438,
+      "learning_rate": 5.103302037514241e-06,
+      "loss": 44.9741,
+      "step": 67330
+    },
+    {
+      "epoch": 0.27206212098562926,
+      "grad_norm": 393.1825256347656,
+      "learning_rate": 5.101891967392426e-06,
+      "loss": 41.374,
+      "step": 67340
+    },
+    {
+      "epoch": 0.2721025222509969,
+      "grad_norm": 307.2347106933594,
+      "learning_rate": 5.100481889163535e-06,
+      "loss": 43.7367,
+      "step": 67350
+    },
+    {
+      "epoch": 0.27214292351636454,
+      "grad_norm": 328.61029052734375,
+      "learning_rate": 5.099071802939763e-06,
+      "loss": 38.4899,
+      "step": 67360
+    },
+    {
+      "epoch": 0.2721833247817322,
+      "grad_norm": 253.48068237304688,
+      "learning_rate": 5.097661708833302e-06,
+      "loss": 42.7777,
+      "step": 67370
+    },
+    {
+      "epoch": 0.27222372604709977,
+      "grad_norm": 648.9194946289062,
+      "learning_rate": 5.096251606956345e-06,
+      "loss": 49.0186,
+      "step": 67380
+    },
+    {
+      "epoch": 0.2722641273124674,
+      "grad_norm": 743.0326538085938,
+      "learning_rate": 5.0948414974210906e-06,
+      "loss": 52.1385,
+      "step": 67390
+    },
+    {
+      "epoch": 0.27230452857783505,
+      "grad_norm": 369.1799621582031,
+      "learning_rate": 5.093431380339734e-06,
+      "loss": 47.9512,
+      "step": 67400
+    },
+    {
+      "epoch": 0.2723449298432027,
+      "grad_norm": 240.17050170898438,
+      "learning_rate": 5.092021255824471e-06,
+      "loss": 73.6545,
+      "step": 67410
+    },
+    {
+      "epoch": 0.2723853311085703,
+      "grad_norm": 577.8941040039062,
+      "learning_rate": 5.090611123987498e-06,
+      "loss": 46.5082,
+      "step": 67420
+    },
+    {
+      "epoch": 0.27242573237393797,
+      "grad_norm": 952.72216796875,
+      "learning_rate": 5.089200984941014e-06,
+      "loss": 37.1385,
+      "step": 67430
+    },
+    {
+      "epoch": 0.2724661336393056,
+      "grad_norm": 367.08306884765625,
+      "learning_rate": 5.087790838797217e-06,
+      "loss": 44.5161,
+      "step": 67440
+    },
+    {
+      "epoch": 0.2725065349046732,
+      "grad_norm": 593.40478515625,
+      "learning_rate": 5.0863806856683076e-06,
+      "loss": 40.6398,
+      "step": 67450
+    },
+    {
+      "epoch": 0.27254693617004083,
+      "grad_norm": 474.2576599121094,
+      "learning_rate": 5.084970525666481e-06,
+      "loss": 46.1414,
+      "step": 67460
+    },
+    {
+      "epoch": 0.27258733743540847,
+      "grad_norm": 354.111572265625,
+      "learning_rate": 5.083560358903942e-06,
+      "loss": 28.794,
+      "step": 67470
+    },
+    {
+      "epoch": 0.2726277387007761,
+      "grad_norm": 985.1436157226562,
+      "learning_rate": 5.082150185492887e-06,
+      "loss": 41.4091,
+      "step": 67480
+    },
+    {
+      "epoch": 0.27266813996614375,
+      "grad_norm": 744.9746704101562,
+      "learning_rate": 5.080740005545519e-06,
+      "loss": 46.0218,
+      "step": 67490
+    },
+    {
+      "epoch": 0.2727085412315114,
+      "grad_norm": 549.0433959960938,
+      "learning_rate": 5.07932981917404e-06,
+      "loss": 41.1397,
+      "step": 67500
+    },
+    {
+      "epoch": 0.272748942496879,
+      "grad_norm": 403.9040222167969,
+      "learning_rate": 5.077919626490651e-06,
+      "loss": 38.3747,
+      "step": 67510
+    },
+    {
+      "epoch": 0.2727893437622466,
+      "grad_norm": 690.0769653320312,
+      "learning_rate": 5.076509427607555e-06,
+      "loss": 33.2111,
+      "step": 67520
+    },
+    {
+      "epoch": 0.27282974502761426,
+      "grad_norm": 209.73768615722656,
+      "learning_rate": 5.075099222636954e-06,
+      "loss": 41.2934,
+      "step": 67530
+    },
+    {
+      "epoch": 0.2728701462929819,
+      "grad_norm": 397.9496154785156,
+      "learning_rate": 5.073689011691054e-06,
+      "loss": 35.0581,
+      "step": 67540
+    },
+    {
+      "epoch": 0.27291054755834954,
+      "grad_norm": 731.2933959960938,
+      "learning_rate": 5.072278794882058e-06,
+      "loss": 47.9663,
+      "step": 67550
+    },
+    {
+      "epoch": 0.2729509488237172,
+      "grad_norm": 845.2896118164062,
+      "learning_rate": 5.07086857232217e-06,
+      "loss": 33.0361,
+      "step": 67560
+    },
+    {
+      "epoch": 0.2729913500890848,
+      "grad_norm": 670.6997680664062,
+      "learning_rate": 5.069458344123592e-06,
+      "loss": 43.4707,
+      "step": 67570
+    },
+    {
+      "epoch": 0.2730317513544524,
+      "grad_norm": 782.651123046875,
+      "learning_rate": 5.068048110398535e-06,
+      "loss": 48.6419,
+      "step": 67580
+    },
+    {
+      "epoch": 0.27307215261982004,
+      "grad_norm": 327.63372802734375,
+      "learning_rate": 5.066637871259201e-06,
+      "loss": 37.3878,
+      "step": 67590
+    },
+    {
+      "epoch": 0.2731125538851877,
+      "grad_norm": 942.0525512695312,
+      "learning_rate": 5.065227626817798e-06,
+      "loss": 49.2512,
+      "step": 67600
+    },
+    {
+      "epoch": 0.2731529551505553,
+      "grad_norm": 563.7671508789062,
+      "learning_rate": 5.063817377186531e-06,
+      "loss": 39.5605,
+      "step": 67610
+    },
+    {
+      "epoch": 0.27319335641592296,
+      "grad_norm": 550.6803588867188,
+      "learning_rate": 5.062407122477609e-06,
+      "loss": 43.2036,
+      "step": 67620
+    },
+    {
+      "epoch": 0.2732337576812906,
+      "grad_norm": 556.4356689453125,
+      "learning_rate": 5.060996862803239e-06,
+      "loss": 36.3034,
+      "step": 67630
+    },
+    {
+      "epoch": 0.2732741589466582,
+      "grad_norm": 498.069091796875,
+      "learning_rate": 5.0595865982756284e-06,
+      "loss": 44.1085,
+      "step": 67640
+    },
+    {
+      "epoch": 0.2733145602120258,
+      "grad_norm": 525.10693359375,
+      "learning_rate": 5.0581763290069865e-06,
+      "loss": 48.0721,
+      "step": 67650
+    },
+    {
+      "epoch": 0.27335496147739347,
+      "grad_norm": 1076.3779296875,
+      "learning_rate": 5.05676605510952e-06,
+      "loss": 55.7806,
+      "step": 67660
+    },
+    {
+      "epoch": 0.2733953627427611,
+      "grad_norm": 956.6840209960938,
+      "learning_rate": 5.055355776695437e-06,
+      "loss": 50.8037,
+      "step": 67670
+    },
+    {
+      "epoch": 0.27343576400812875,
+      "grad_norm": 934.687255859375,
+      "learning_rate": 5.0539454938769525e-06,
+      "loss": 53.6633,
+      "step": 67680
+    },
+    {
+      "epoch": 0.2734761652734964,
+      "grad_norm": 789.1433715820312,
+      "learning_rate": 5.052535206766271e-06,
+      "loss": 67.9196,
+      "step": 67690
+    },
+    {
+      "epoch": 0.27351656653886397,
+      "grad_norm": 509.7837219238281,
+      "learning_rate": 5.051124915475604e-06,
+      "loss": 47.9745,
+      "step": 67700
+    },
+    {
+      "epoch": 0.2735569678042316,
+      "grad_norm": 806.1221313476562,
+      "learning_rate": 5.049714620117162e-06,
+      "loss": 54.7842,
+      "step": 67710
+    },
+    {
+      "epoch": 0.27359736906959925,
+      "grad_norm": 789.8609008789062,
+      "learning_rate": 5.0483043208031575e-06,
+      "loss": 42.8648,
+      "step": 67720
+    },
+    {
+      "epoch": 0.2736377703349669,
+      "grad_norm": 378.3641662597656,
+      "learning_rate": 5.0468940176458e-06,
+      "loss": 25.4453,
+      "step": 67730
+    },
+    {
+      "epoch": 0.27367817160033453,
+      "grad_norm": 377.1431579589844,
+      "learning_rate": 5.045483710757298e-06,
+      "loss": 44.5361,
+      "step": 67740
+    },
+    {
+      "epoch": 0.27371857286570217,
+      "grad_norm": 717.8798217773438,
+      "learning_rate": 5.044073400249867e-06,
+      "loss": 60.3492,
+      "step": 67750
+    },
+    {
+      "epoch": 0.2737589741310698,
+      "grad_norm": 612.9976806640625,
+      "learning_rate": 5.0426630862357176e-06,
+      "loss": 34.6233,
+      "step": 67760
+    },
+    {
+      "epoch": 0.2737993753964374,
+      "grad_norm": 242.31356811523438,
+      "learning_rate": 5.041252768827064e-06,
+      "loss": 32.6884,
+      "step": 67770
+    },
+    {
+      "epoch": 0.27383977666180503,
+      "grad_norm": 875.4049072265625,
+      "learning_rate": 5.039842448136115e-06,
+      "loss": 58.8626,
+      "step": 67780
+    },
+    {
+      "epoch": 0.2738801779271727,
+      "grad_norm": 347.9784240722656,
+      "learning_rate": 5.038432124275087e-06,
+      "loss": 38.7157,
+      "step": 67790
+    },
+    {
+      "epoch": 0.2739205791925403,
+      "grad_norm": 528.5045166015625,
+      "learning_rate": 5.03702179735619e-06,
+      "loss": 62.6629,
+      "step": 67800
+    },
+    {
+      "epoch": 0.27396098045790795,
+      "grad_norm": 613.0690307617188,
+      "learning_rate": 5.035611467491638e-06,
+      "loss": 30.0438,
+      "step": 67810
+    },
+    {
+      "epoch": 0.2740013817232756,
+      "grad_norm": 510.3870544433594,
+      "learning_rate": 5.034201134793646e-06,
+      "loss": 35.476,
+      "step": 67820
+    },
+    {
+      "epoch": 0.2740417829886432,
+      "grad_norm": 491.538330078125,
+      "learning_rate": 5.032790799374426e-06,
+      "loss": 31.3243,
+      "step": 67830
+    },
+    {
+      "epoch": 0.2740821842540108,
+      "grad_norm": 360.6121520996094,
+      "learning_rate": 5.0313804613461925e-06,
+      "loss": 50.7477,
+      "step": 67840
+    },
+    {
+      "epoch": 0.27412258551937846,
+      "grad_norm": 502.55706787109375,
+      "learning_rate": 5.0299701208211605e-06,
+      "loss": 32.5469,
+      "step": 67850
+    },
+    {
+      "epoch": 0.2741629867847461,
+      "grad_norm": 524.4901733398438,
+      "learning_rate": 5.028559777911543e-06,
+      "loss": 37.7005,
+      "step": 67860
+    },
+    {
+      "epoch": 0.27420338805011374,
+      "grad_norm": 647.3909912109375,
+      "learning_rate": 5.027149432729555e-06,
+      "loss": 69.5209,
+      "step": 67870
+    },
+    {
+      "epoch": 0.2742437893154814,
+      "grad_norm": 1294.49365234375,
+      "learning_rate": 5.025739085387411e-06,
+      "loss": 41.7499,
+      "step": 67880
+    },
+    {
+      "epoch": 0.274284190580849,
+      "grad_norm": 801.7745361328125,
+      "learning_rate": 5.024328735997327e-06,
+      "loss": 59.8358,
+      "step": 67890
+    },
+    {
+      "epoch": 0.2743245918462166,
+      "grad_norm": 262.2741394042969,
+      "learning_rate": 5.0229183846715154e-06,
+      "loss": 26.463,
+      "step": 67900
+    },
+    {
+      "epoch": 0.27436499311158424,
+      "grad_norm": 978.8836059570312,
+      "learning_rate": 5.021508031522195e-06,
+      "loss": 66.1999,
+      "step": 67910
+    },
+    {
+      "epoch": 0.2744053943769519,
+      "grad_norm": 960.217529296875,
+      "learning_rate": 5.0200976766615785e-06,
+      "loss": 47.8755,
+      "step": 67920
+    },
+    {
+      "epoch": 0.2744457956423195,
+      "grad_norm": 899.3736572265625,
+      "learning_rate": 5.018687320201882e-06,
+      "loss": 57.0194,
+      "step": 67930
+    },
+    {
+      "epoch": 0.27448619690768716,
+      "grad_norm": 572.0177612304688,
+      "learning_rate": 5.017276962255323e-06,
+      "loss": 38.2684,
+      "step": 67940
+    },
+    {
+      "epoch": 0.2745265981730548,
+      "grad_norm": 650.45361328125,
+      "learning_rate": 5.015866602934112e-06,
+      "loss": 38.218,
+      "step": 67950
+    },
+    {
+      "epoch": 0.2745669994384224,
+      "grad_norm": 491.1336669921875,
+      "learning_rate": 5.01445624235047e-06,
+      "loss": 46.4445,
+      "step": 67960
+    },
+    {
+      "epoch": 0.27460740070379,
+      "grad_norm": 622.7539672851562,
+      "learning_rate": 5.013045880616612e-06,
+      "loss": 60.7711,
+      "step": 67970
+    },
+    {
+      "epoch": 0.27464780196915767,
+      "grad_norm": 467.92230224609375,
+      "learning_rate": 5.011635517844753e-06,
+      "loss": 49.4,
+      "step": 67980
+    },
+    {
+      "epoch": 0.2746882032345253,
+      "grad_norm": 635.2073364257812,
+      "learning_rate": 5.010225154147107e-06,
+      "loss": 59.7034,
+      "step": 67990
+    },
+    {
+      "epoch": 0.27472860449989295,
+      "grad_norm": 533.5791015625,
+      "learning_rate": 5.008814789635894e-06,
+      "loss": 39.1137,
+      "step": 68000
+    },
+    {
+      "epoch": 0.2747690057652606,
+      "grad_norm": 494.2832336425781,
+      "learning_rate": 5.007404424423329e-06,
+      "loss": 35.455,
+      "step": 68010
+    },
+    {
+      "epoch": 0.27480940703062817,
+      "grad_norm": 331.5168151855469,
+      "learning_rate": 5.0059940586216284e-06,
+      "loss": 33.3954,
+      "step": 68020
+    },
+    {
+      "epoch": 0.2748498082959958,
+      "grad_norm": 687.5106811523438,
+      "learning_rate": 5.004583692343007e-06,
+      "loss": 37.4024,
+      "step": 68030
+    },
+    {
+      "epoch": 0.27489020956136345,
+      "grad_norm": 685.0592651367188,
+      "learning_rate": 5.003173325699682e-06,
+      "loss": 50.4846,
+      "step": 68040
+    },
+    {
+      "epoch": 0.2749306108267311,
+      "grad_norm": 316.0040588378906,
+      "learning_rate": 5.00176295880387e-06,
+      "loss": 43.7119,
+      "step": 68050
+    },
+    {
+      "epoch": 0.27497101209209873,
+      "grad_norm": 368.1940612792969,
+      "learning_rate": 5.000352591767787e-06,
+      "loss": 40.2011,
+      "step": 68060
+    },
+    {
+      "epoch": 0.27501141335746637,
+      "grad_norm": 483.9440002441406,
+      "learning_rate": 4.998942224703651e-06,
+      "loss": 40.0359,
+      "step": 68070
+    },
+    {
+      "epoch": 0.275051814622834,
+      "grad_norm": 1012.7457275390625,
+      "learning_rate": 4.997531857723678e-06,
+      "loss": 39.9802,
+      "step": 68080
+    },
+    {
+      "epoch": 0.2750922158882016,
+      "grad_norm": 481.0828857421875,
+      "learning_rate": 4.996121490940084e-06,
+      "loss": 41.3946,
+      "step": 68090
+    },
+    {
+      "epoch": 0.27513261715356924,
+      "grad_norm": 619.2692260742188,
+      "learning_rate": 4.994711124465084e-06,
+      "loss": 44.6533,
+      "step": 68100
+    },
+    {
+      "epoch": 0.2751730184189369,
+      "grad_norm": 598.6572875976562,
+      "learning_rate": 4.993300758410895e-06,
+      "loss": 45.2937,
+      "step": 68110
+    },
+    {
+      "epoch": 0.2752134196843045,
+      "grad_norm": 475.28607177734375,
+      "learning_rate": 4.991890392889735e-06,
+      "loss": 26.7339,
+      "step": 68120
+    },
+    {
+      "epoch": 0.27525382094967216,
+      "grad_norm": 296.6506042480469,
+      "learning_rate": 4.990480028013818e-06,
+      "loss": 49.4723,
+      "step": 68130
+    },
+    {
+      "epoch": 0.2752942222150398,
+      "grad_norm": 319.9155578613281,
+      "learning_rate": 4.989069663895361e-06,
+      "loss": 45.966,
+      "step": 68140
+    },
+    {
+      "epoch": 0.2753346234804074,
+      "grad_norm": 629.8634033203125,
+      "learning_rate": 4.9876593006465825e-06,
+      "loss": 48.5483,
+      "step": 68150
+    },
+    {
+      "epoch": 0.275375024745775,
+      "grad_norm": 704.5947265625,
+      "learning_rate": 4.986248938379696e-06,
+      "loss": 46.9476,
+      "step": 68160
+    },
+    {
+      "epoch": 0.27541542601114266,
+      "grad_norm": 696.213134765625,
+      "learning_rate": 4.984838577206921e-06,
+      "loss": 58.3571,
+      "step": 68170
+    },
+    {
+      "epoch": 0.2754558272765103,
+      "grad_norm": 624.2259521484375,
+      "learning_rate": 4.9834282172404665e-06,
+      "loss": 49.6634,
+      "step": 68180
+    },
+    {
+      "epoch": 0.27549622854187794,
+      "grad_norm": 592.13623046875,
+      "learning_rate": 4.982017858592555e-06,
+      "loss": 35.5057,
+      "step": 68190
+    },
+    {
+      "epoch": 0.2755366298072456,
+      "grad_norm": 311.4937438964844,
+      "learning_rate": 4.980607501375399e-06,
+      "loss": 48.7469,
+      "step": 68200
+    },
+    {
+      "epoch": 0.2755770310726132,
+      "grad_norm": 721.6624755859375,
+      "learning_rate": 4.979197145701216e-06,
+      "loss": 47.0013,
+      "step": 68210
+    },
+    {
+      "epoch": 0.2756174323379808,
+      "grad_norm": 306.4320068359375,
+      "learning_rate": 4.977786791682221e-06,
+      "loss": 36.7569,
+      "step": 68220
+    },
+    {
+      "epoch": 0.27565783360334845,
+      "grad_norm": 378.13616943359375,
+      "learning_rate": 4.976376439430627e-06,
+      "loss": 50.454,
+      "step": 68230
+    },
+    {
+      "epoch": 0.2756982348687161,
+      "grad_norm": 651.3588256835938,
+      "learning_rate": 4.974966089058652e-06,
+      "loss": 36.0825,
+      "step": 68240
+    },
+    {
+      "epoch": 0.2757386361340837,
+      "grad_norm": 538.3560180664062,
+      "learning_rate": 4.973555740678512e-06,
+      "loss": 33.1485,
+      "step": 68250
+    },
+    {
+      "epoch": 0.27577903739945137,
+      "grad_norm": 349.70184326171875,
+      "learning_rate": 4.972145394402421e-06,
+      "loss": 73.0145,
+      "step": 68260
+    },
+    {
+      "epoch": 0.275819438664819,
+      "grad_norm": 448.4791259765625,
+      "learning_rate": 4.9707350503425905e-06,
+      "loss": 41.0152,
+      "step": 68270
+    },
+    {
+      "epoch": 0.2758598399301866,
+      "grad_norm": 353.9982604980469,
+      "learning_rate": 4.969324708611239e-06,
+      "loss": 40.9351,
+      "step": 68280
+    },
+    {
+      "epoch": 0.27590024119555423,
+      "grad_norm": 779.5863647460938,
+      "learning_rate": 4.9679143693205785e-06,
+      "loss": 52.2675,
+      "step": 68290
+    },
+    {
+      "epoch": 0.27594064246092187,
+      "grad_norm": 714.5828857421875,
+      "learning_rate": 4.966504032582826e-06,
+      "loss": 36.8268,
+      "step": 68300
+    },
+    {
+      "epoch": 0.2759810437262895,
+      "grad_norm": 930.7247314453125,
+      "learning_rate": 4.965093698510192e-06,
+      "loss": 62.9139,
+      "step": 68310
+    },
+    {
+      "epoch": 0.27602144499165715,
+      "grad_norm": 710.6253051757812,
+      "learning_rate": 4.963683367214895e-06,
+      "loss": 59.0211,
+      "step": 68320
+    },
+    {
+      "epoch": 0.2760618462570248,
+      "grad_norm": 716.7061157226562,
+      "learning_rate": 4.962273038809143e-06,
+      "loss": 75.9322,
+      "step": 68330
+    },
+    {
+      "epoch": 0.2761022475223924,
+      "grad_norm": 957.3465576171875,
+      "learning_rate": 4.960862713405153e-06,
+      "loss": 46.2151,
+      "step": 68340
+    },
+    {
+      "epoch": 0.27614264878776,
+      "grad_norm": 339.1300048828125,
+      "learning_rate": 4.95945239111514e-06,
+      "loss": 36.6535,
+      "step": 68350
+    },
+    {
+      "epoch": 0.27618305005312765,
+      "grad_norm": 740.7291259765625,
+      "learning_rate": 4.9580420720513115e-06,
+      "loss": 55.0259,
+      "step": 68360
+    },
+    {
+      "epoch": 0.2762234513184953,
+      "grad_norm": 521.9520263671875,
+      "learning_rate": 4.956631756325882e-06,
+      "loss": 39.7061,
+      "step": 68370
+    },
+    {
+      "epoch": 0.27626385258386293,
+      "grad_norm": 475.54473876953125,
+      "learning_rate": 4.955221444051066e-06,
+      "loss": 37.5069,
+      "step": 68380
+    },
+    {
+      "epoch": 0.2763042538492306,
+      "grad_norm": 665.6690063476562,
+      "learning_rate": 4.953811135339073e-06,
+      "loss": 51.0758,
+      "step": 68390
+    },
+    {
+      "epoch": 0.2763446551145982,
+      "grad_norm": 360.49273681640625,
+      "learning_rate": 4.952400830302117e-06,
+      "loss": 47.2758,
+      "step": 68400
+    },
+    {
+      "epoch": 0.2763850563799658,
+      "grad_norm": 306.6684265136719,
+      "learning_rate": 4.950990529052409e-06,
+      "loss": 38.9826,
+      "step": 68410
+    },
+    {
+      "epoch": 0.27642545764533344,
+      "grad_norm": 776.9755249023438,
+      "learning_rate": 4.949580231702158e-06,
+      "loss": 45.6411,
+      "step": 68420
+    },
+    {
+      "epoch": 0.2764658589107011,
+      "grad_norm": 275.6937561035156,
+      "learning_rate": 4.94816993836358e-06,
+      "loss": 25.9277,
+      "step": 68430
+    },
+    {
+      "epoch": 0.2765062601760687,
+      "grad_norm": 131.82835388183594,
+      "learning_rate": 4.946759649148879e-06,
+      "loss": 30.0371,
+      "step": 68440
+    },
+    {
+      "epoch": 0.27654666144143636,
+      "grad_norm": 733.06103515625,
+      "learning_rate": 4.945349364170269e-06,
+      "loss": 45.6553,
+      "step": 68450
+    },
+    {
+      "epoch": 0.276587062706804,
+      "grad_norm": 650.412353515625,
+      "learning_rate": 4.94393908353996e-06,
+      "loss": 37.3156,
+      "step": 68460
+    },
+    {
+      "epoch": 0.2766274639721716,
+      "grad_norm": 412.7141418457031,
+      "learning_rate": 4.942528807370158e-06,
+      "loss": 38.6044,
+      "step": 68470
+    },
+    {
+      "epoch": 0.2766678652375392,
+      "grad_norm": 331.10858154296875,
+      "learning_rate": 4.941118535773078e-06,
+      "loss": 39.9055,
+      "step": 68480
+    },
+    {
+      "epoch": 0.27670826650290686,
+      "grad_norm": 708.8507690429688,
+      "learning_rate": 4.9397082688609245e-06,
+      "loss": 52.6693,
+      "step": 68490
+    },
+    {
+      "epoch": 0.2767486677682745,
+      "grad_norm": 411.26239013671875,
+      "learning_rate": 4.938298006745909e-06,
+      "loss": 36.8732,
+      "step": 68500
+    },
+    {
+      "epoch": 0.27678906903364214,
+      "grad_norm": 518.560791015625,
+      "learning_rate": 4.936887749540236e-06,
+      "loss": 31.6104,
+      "step": 68510
+    },
+    {
+      "epoch": 0.2768294702990098,
+      "grad_norm": 1007.0764770507812,
+      "learning_rate": 4.935477497356118e-06,
+      "loss": 45.9277,
+      "step": 68520
+    },
+    {
+      "epoch": 0.2768698715643774,
+      "grad_norm": 5712.80810546875,
+      "learning_rate": 4.934067250305757e-06,
+      "loss": 59.7893,
+      "step": 68530
+    },
+    {
+      "epoch": 0.276910272829745,
+      "grad_norm": 718.469970703125,
+      "learning_rate": 4.932657008501362e-06,
+      "loss": 63.6629,
+      "step": 68540
+    },
+    {
+      "epoch": 0.27695067409511265,
+      "grad_norm": 361.53131103515625,
+      "learning_rate": 4.931246772055141e-06,
+      "loss": 47.2617,
+      "step": 68550
+    },
+    {
+      "epoch": 0.2769910753604803,
+      "grad_norm": 730.575439453125,
+      "learning_rate": 4.9298365410792985e-06,
+      "loss": 39.6719,
+      "step": 68560
+    },
+    {
+      "epoch": 0.27703147662584793,
+      "grad_norm": 630.3861694335938,
+      "learning_rate": 4.928426315686039e-06,
+      "loss": 52.2734,
+      "step": 68570
+    },
+    {
+      "epoch": 0.27707187789121557,
+      "grad_norm": 597.8245849609375,
+      "learning_rate": 4.92701609598757e-06,
+      "loss": 42.7517,
+      "step": 68580
+    },
+    {
+      "epoch": 0.2771122791565832,
+      "grad_norm": 628.2814331054688,
+      "learning_rate": 4.925605882096096e-06,
+      "loss": 40.1165,
+      "step": 68590
+    },
+    {
+      "epoch": 0.2771526804219508,
+      "grad_norm": 862.098876953125,
+      "learning_rate": 4.924195674123821e-06,
+      "loss": 45.4163,
+      "step": 68600
+    },
+    {
+      "epoch": 0.27719308168731843,
+      "grad_norm": 431.5798034667969,
+      "learning_rate": 4.922785472182948e-06,
+      "loss": 32.5109,
+      "step": 68610
+    },
+    {
+      "epoch": 0.27723348295268607,
+      "grad_norm": 568.9783935546875,
+      "learning_rate": 4.92137527638568e-06,
+      "loss": 38.1525,
+      "step": 68620
+    },
+    {
+      "epoch": 0.2772738842180537,
+      "grad_norm": 349.13421630859375,
+      "learning_rate": 4.919965086844221e-06,
+      "loss": 49.8859,
+      "step": 68630
+    },
+    {
+      "epoch": 0.27731428548342135,
+      "grad_norm": 524.2000122070312,
+      "learning_rate": 4.9185549036707715e-06,
+      "loss": 40.2317,
+      "step": 68640
+    },
+    {
+      "epoch": 0.277354686748789,
+      "grad_norm": 556.9549560546875,
+      "learning_rate": 4.917144726977535e-06,
+      "loss": 40.5639,
+      "step": 68650
+    },
+    {
+      "epoch": 0.2773950880141566,
+      "grad_norm": 464.0063781738281,
+      "learning_rate": 4.915734556876713e-06,
+      "loss": 41.1225,
+      "step": 68660
+    },
+    {
+      "epoch": 0.2774354892795242,
+      "grad_norm": 646.3726196289062,
+      "learning_rate": 4.914324393480504e-06,
+      "loss": 42.3612,
+      "step": 68670
+    },
+    {
+      "epoch": 0.27747589054489186,
+      "grad_norm": 475.73748779296875,
+      "learning_rate": 4.9129142369011105e-06,
+      "loss": 39.3261,
+      "step": 68680
+    },
+    {
+      "epoch": 0.2775162918102595,
+      "grad_norm": 429.1013488769531,
+      "learning_rate": 4.911504087250735e-06,
+      "loss": 44.6936,
+      "step": 68690
+    },
+    {
+      "epoch": 0.27755669307562714,
+      "grad_norm": 395.2467041015625,
+      "learning_rate": 4.910093944641569e-06,
+      "loss": 49.2282,
+      "step": 68700
+    },
+    {
+      "epoch": 0.2775970943409948,
+      "grad_norm": 1080.4151611328125,
+      "learning_rate": 4.9086838091858155e-06,
+      "loss": 50.2219,
+      "step": 68710
+    },
+    {
+      "epoch": 0.2776374956063624,
+      "grad_norm": 725.0263671875,
+      "learning_rate": 4.9072736809956735e-06,
+      "loss": 41.7178,
+      "step": 68720
+    },
+    {
+      "epoch": 0.27767789687173,
+      "grad_norm": 240.93812561035156,
+      "learning_rate": 4.9058635601833384e-06,
+      "loss": 34.3757,
+      "step": 68730
+    },
+    {
+      "epoch": 0.27771829813709764,
+      "grad_norm": 422.4692687988281,
+      "learning_rate": 4.904453446861008e-06,
+      "loss": 47.2706,
+      "step": 68740
+    },
+    {
+      "epoch": 0.2777586994024653,
+      "grad_norm": 992.8607177734375,
+      "learning_rate": 4.903043341140879e-06,
+      "loss": 40.3693,
+      "step": 68750
+    },
+    {
+      "epoch": 0.2777991006678329,
+      "grad_norm": 671.8703002929688,
+      "learning_rate": 4.901633243135144e-06,
+      "loss": 41.3787,
+      "step": 68760
+    },
+    {
+      "epoch": 0.27783950193320056,
+      "grad_norm": 748.8638916015625,
+      "learning_rate": 4.900223152956003e-06,
+      "loss": 47.3397,
+      "step": 68770
+    },
+    {
+      "epoch": 0.2778799031985682,
+      "grad_norm": 290.8716735839844,
+      "learning_rate": 4.898813070715649e-06,
+      "loss": 37.6664,
+      "step": 68780
+    },
+    {
+      "epoch": 0.2779203044639358,
+      "grad_norm": 565.4688720703125,
+      "learning_rate": 4.897402996526273e-06,
+      "loss": 51.1212,
+      "step": 68790
+    },
+    {
+      "epoch": 0.2779607057293034,
+      "grad_norm": 898.2797241210938,
+      "learning_rate": 4.895992930500068e-06,
+      "loss": 42.9865,
+      "step": 68800
+    },
+    {
+      "epoch": 0.27800110699467107,
+      "grad_norm": 316.641845703125,
+      "learning_rate": 4.894582872749229e-06,
+      "loss": 37.7981,
+      "step": 68810
+    },
+    {
+      "epoch": 0.2780415082600387,
+      "grad_norm": 701.3886108398438,
+      "learning_rate": 4.893172823385947e-06,
+      "loss": 55.1056,
+      "step": 68820
+    },
+    {
+      "epoch": 0.27808190952540635,
+      "grad_norm": 780.6195068359375,
+      "learning_rate": 4.8917627825224146e-06,
+      "loss": 45.9489,
+      "step": 68830
+    },
+    {
+      "epoch": 0.278122310790774,
+      "grad_norm": 767.5986328125,
+      "learning_rate": 4.89035275027082e-06,
+      "loss": 42.4477,
+      "step": 68840
+    },
+    {
+      "epoch": 0.2781627120561416,
+      "grad_norm": 665.1085205078125,
+      "learning_rate": 4.888942726743353e-06,
+      "loss": 42.3077,
+      "step": 68850
+    },
+    {
+      "epoch": 0.2782031133215092,
+      "grad_norm": 823.7915649414062,
+      "learning_rate": 4.887532712052206e-06,
+      "loss": 48.6418,
+      "step": 68860
+    },
+    {
+      "epoch": 0.27824351458687685,
+      "grad_norm": 601.9495239257812,
+      "learning_rate": 4.886122706309563e-06,
+      "loss": 36.7191,
+      "step": 68870
+    },
+    {
+      "epoch": 0.2782839158522445,
+      "grad_norm": 538.6893310546875,
+      "learning_rate": 4.884712709627614e-06,
+      "loss": 42.2536,
+      "step": 68880
+    },
+    {
+      "epoch": 0.27832431711761213,
+      "grad_norm": 619.4857177734375,
+      "learning_rate": 4.8833027221185455e-06,
+      "loss": 37.4783,
+      "step": 68890
+    },
+    {
+      "epoch": 0.27836471838297977,
+      "grad_norm": 803.7899169921875,
+      "learning_rate": 4.881892743894543e-06,
+      "loss": 39.6659,
+      "step": 68900
+    },
+    {
+      "epoch": 0.2784051196483474,
+      "grad_norm": 377.7758483886719,
+      "learning_rate": 4.880482775067794e-06,
+      "loss": 35.0229,
+      "step": 68910
+    },
+    {
+      "epoch": 0.278445520913715,
+      "grad_norm": 546.9606323242188,
+      "learning_rate": 4.879072815750481e-06,
+      "loss": 38.2003,
+      "step": 68920
+    },
+    {
+      "epoch": 0.27848592217908263,
+      "grad_norm": 751.6867065429688,
+      "learning_rate": 4.87766286605479e-06,
+      "loss": 45.7661,
+      "step": 68930
+    },
+    {
+      "epoch": 0.2785263234444503,
+      "grad_norm": 623.1182250976562,
+      "learning_rate": 4.876252926092903e-06,
+      "loss": 45.6871,
+      "step": 68940
+    },
+    {
+      "epoch": 0.2785667247098179,
+      "grad_norm": 698.1951904296875,
+      "learning_rate": 4.874842995977004e-06,
+      "loss": 38.3063,
+      "step": 68950
+    },
+    {
+      "epoch": 0.27860712597518555,
+      "grad_norm": 1337.2913818359375,
+      "learning_rate": 4.873433075819272e-06,
+      "loss": 44.2702,
+      "step": 68960
+    },
+    {
+      "epoch": 0.2786475272405532,
+      "grad_norm": 386.43536376953125,
+      "learning_rate": 4.87202316573189e-06,
+      "loss": 31.441,
+      "step": 68970
+    },
+    {
+      "epoch": 0.2786879285059208,
+      "grad_norm": 242.0117645263672,
+      "learning_rate": 4.870613265827037e-06,
+      "loss": 38.5598,
+      "step": 68980
+    },
+    {
+      "epoch": 0.2787283297712884,
+      "grad_norm": 598.38818359375,
+      "learning_rate": 4.869203376216891e-06,
+      "loss": 33.6361,
+      "step": 68990
+    },
+    {
+      "epoch": 0.27876873103665606,
+      "grad_norm": 527.7412109375,
+      "learning_rate": 4.867793497013634e-06,
+      "loss": 46.4232,
+      "step": 69000
+    },
+    {
+      "epoch": 0.2788091323020237,
+      "grad_norm": 782.029541015625,
+      "learning_rate": 4.866383628329442e-06,
+      "loss": 49.8267,
+      "step": 69010
+    },
+    {
+      "epoch": 0.27884953356739134,
+      "grad_norm": 409.10711669921875,
+      "learning_rate": 4.86497377027649e-06,
+      "loss": 43.6214,
+      "step": 69020
+    },
+    {
+      "epoch": 0.278889934832759,
+      "grad_norm": 471.22906494140625,
+      "learning_rate": 4.863563922966957e-06,
+      "loss": 67.8977,
+      "step": 69030
+    },
+    {
+      "epoch": 0.2789303360981266,
+      "grad_norm": 400.7550354003906,
+      "learning_rate": 4.862154086513016e-06,
+      "loss": 59.4513,
+      "step": 69040
+    },
+    {
+      "epoch": 0.2789707373634942,
+      "grad_norm": 381.7715148925781,
+      "learning_rate": 4.860744261026841e-06,
+      "loss": 38.826,
+      "step": 69050
+    },
+    {
+      "epoch": 0.27901113862886184,
+      "grad_norm": 245.989501953125,
+      "learning_rate": 4.8593344466206075e-06,
+      "loss": 38.2717,
+      "step": 69060
+    },
+    {
+      "epoch": 0.2790515398942295,
+      "grad_norm": 419.75469970703125,
+      "learning_rate": 4.857924643406485e-06,
+      "loss": 27.1463,
+      "step": 69070
+    },
+    {
+      "epoch": 0.2790919411595971,
+      "grad_norm": 1163.119873046875,
+      "learning_rate": 4.856514851496647e-06,
+      "loss": 55.6151,
+      "step": 69080
+    },
+    {
+      "epoch": 0.27913234242496476,
+      "grad_norm": 999.8130493164062,
+      "learning_rate": 4.8551050710032625e-06,
+      "loss": 36.9399,
+      "step": 69090
+    },
+    {
+      "epoch": 0.2791727436903324,
+      "grad_norm": 463.41046142578125,
+      "learning_rate": 4.853695302038504e-06,
+      "loss": 47.6074,
+      "step": 69100
+    },
+    {
+      "epoch": 0.2792131449557,
+      "grad_norm": 614.2119140625,
+      "learning_rate": 4.8522855447145385e-06,
+      "loss": 54.542,
+      "step": 69110
+    },
+    {
+      "epoch": 0.27925354622106763,
+      "grad_norm": 488.19732666015625,
+      "learning_rate": 4.850875799143537e-06,
+      "loss": 38.1682,
+      "step": 69120
+    },
+    {
+      "epoch": 0.27929394748643527,
+      "grad_norm": 302.4384765625,
+      "learning_rate": 4.84946606543766e-06,
+      "loss": 38.7543,
+      "step": 69130
+    },
+    {
+      "epoch": 0.2793343487518029,
+      "grad_norm": 514.2261352539062,
+      "learning_rate": 4.848056343709079e-06,
+      "loss": 41.4556,
+      "step": 69140
+    },
+    {
+      "epoch": 0.27937475001717055,
+      "grad_norm": 966.3445434570312,
+      "learning_rate": 4.846646634069957e-06,
+      "loss": 53.6878,
+      "step": 69150
+    },
+    {
+      "epoch": 0.2794151512825382,
+      "grad_norm": 1981.818603515625,
+      "learning_rate": 4.845236936632458e-06,
+      "loss": 62.9226,
+      "step": 69160
+    },
+    {
+      "epoch": 0.27945555254790583,
+      "grad_norm": 507.5809631347656,
+      "learning_rate": 4.843827251508747e-06,
+      "loss": 63.0771,
+      "step": 69170
+    },
+    {
+      "epoch": 0.2794959538132734,
+      "grad_norm": 562.9615478515625,
+      "learning_rate": 4.842417578810984e-06,
+      "loss": 59.4227,
+      "step": 69180
+    },
+    {
+      "epoch": 0.27953635507864105,
+      "grad_norm": 459.4322814941406,
+      "learning_rate": 4.841007918651329e-06,
+      "loss": 47.2068,
+      "step": 69190
+    },
+    {
+      "epoch": 0.2795767563440087,
+      "grad_norm": 816.6414184570312,
+      "learning_rate": 4.839598271141947e-06,
+      "loss": 45.1818,
+      "step": 69200
+    },
+    {
+      "epoch": 0.27961715760937633,
+      "grad_norm": 435.2837829589844,
+      "learning_rate": 4.8381886363949956e-06,
+      "loss": 50.4104,
+      "step": 69210
+    },
+    {
+      "epoch": 0.27965755887474397,
+      "grad_norm": 3312.6845703125,
+      "learning_rate": 4.83677901452263e-06,
+      "loss": 59.2026,
+      "step": 69220
+    },
+    {
+      "epoch": 0.2796979601401116,
+      "grad_norm": 560.1244506835938,
+      "learning_rate": 4.835369405637009e-06,
+      "loss": 42.2866,
+      "step": 69230
+    },
+    {
+      "epoch": 0.2797383614054792,
+      "grad_norm": 674.087646484375,
+      "learning_rate": 4.833959809850288e-06,
+      "loss": 34.2167,
+      "step": 69240
+    },
+    {
+      "epoch": 0.27977876267084684,
+      "grad_norm": 978.073974609375,
+      "learning_rate": 4.832550227274624e-06,
+      "loss": 44.3173,
+      "step": 69250
+    },
+    {
+      "epoch": 0.2798191639362145,
+      "grad_norm": 613.9957275390625,
+      "learning_rate": 4.83114065802217e-06,
+      "loss": 41.8678,
+      "step": 69260
+    },
+    {
+      "epoch": 0.2798595652015821,
+      "grad_norm": 899.7896118164062,
+      "learning_rate": 4.829731102205079e-06,
+      "loss": 61.8224,
+      "step": 69270
+    },
+    {
+      "epoch": 0.27989996646694976,
+      "grad_norm": 672.7186279296875,
+      "learning_rate": 4.828321559935502e-06,
+      "loss": 52.7221,
+      "step": 69280
+    },
+    {
+      "epoch": 0.2799403677323174,
+      "grad_norm": 515.7698974609375,
+      "learning_rate": 4.826912031325592e-06,
+      "loss": 57.5608,
+      "step": 69290
+    },
+    {
+      "epoch": 0.279980768997685,
+      "grad_norm": 702.9324951171875,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 36.5247,
+      "step": 69300
+    },
+    {
+      "epoch": 0.2800211702630526,
+      "grad_norm": 620.7726440429688,
+      "learning_rate": 4.824093015533365e-06,
+      "loss": 47.5829,
+      "step": 69310
+    },
+    {
+      "epoch": 0.28006157152842026,
+      "grad_norm": 100.11849212646484,
+      "learning_rate": 4.822683528575344e-06,
+      "loss": 39.2016,
+      "step": 69320
+    },
+    {
+      "epoch": 0.2801019727937879,
+      "grad_norm": 275.9478759765625,
+      "learning_rate": 4.8212740557255815e-06,
+      "loss": 47.3615,
+      "step": 69330
+    },
+    {
+      "epoch": 0.28014237405915554,
+      "grad_norm": 548.790771484375,
+      "learning_rate": 4.819864597096222e-06,
+      "loss": 40.4179,
+      "step": 69340
+    },
+    {
+      "epoch": 0.2801827753245232,
+      "grad_norm": 593.6949462890625,
+      "learning_rate": 4.81845515279941e-06,
+      "loss": 55.0802,
+      "step": 69350
+    },
+    {
+      "epoch": 0.2802231765898908,
+      "grad_norm": 691.48193359375,
+      "learning_rate": 4.817045722947288e-06,
+      "loss": 47.7153,
+      "step": 69360
+    },
+    {
+      "epoch": 0.2802635778552584,
+      "grad_norm": 613.411376953125,
+      "learning_rate": 4.815636307651998e-06,
+      "loss": 40.1139,
+      "step": 69370
+    },
+    {
+      "epoch": 0.28030397912062605,
+      "grad_norm": 517.376220703125,
+      "learning_rate": 4.814226907025683e-06,
+      "loss": 35.4418,
+      "step": 69380
+    },
+    {
+      "epoch": 0.2803443803859937,
+      "grad_norm": 319.6499328613281,
+      "learning_rate": 4.812817521180479e-06,
+      "loss": 45.9838,
+      "step": 69390
+    },
+    {
+      "epoch": 0.2803847816513613,
+      "grad_norm": 973.4129028320312,
+      "learning_rate": 4.811408150228526e-06,
+      "loss": 54.5951,
+      "step": 69400
+    },
+    {
+      "epoch": 0.28042518291672897,
+      "grad_norm": 366.27972412109375,
+      "learning_rate": 4.80999879428196e-06,
+      "loss": 39.5605,
+      "step": 69410
+    },
+    {
+      "epoch": 0.2804655841820966,
+      "grad_norm": 646.7029418945312,
+      "learning_rate": 4.808589453452918e-06,
+      "loss": 42.2973,
+      "step": 69420
+    },
+    {
+      "epoch": 0.2805059854474642,
+      "grad_norm": 615.7694702148438,
+      "learning_rate": 4.807180127853535e-06,
+      "loss": 53.6376,
+      "step": 69430
+    },
+    {
+      "epoch": 0.28054638671283183,
+      "grad_norm": 758.0111083984375,
+      "learning_rate": 4.8057708175959446e-06,
+      "loss": 36.8911,
+      "step": 69440
+    },
+    {
+      "epoch": 0.28058678797819947,
+      "grad_norm": 847.0225830078125,
+      "learning_rate": 4.804361522792278e-06,
+      "loss": 43.2223,
+      "step": 69450
+    },
+    {
+      "epoch": 0.2806271892435671,
+      "grad_norm": 1029.685791015625,
+      "learning_rate": 4.8029522435546695e-06,
+      "loss": 40.3139,
+      "step": 69460
+    },
+    {
+      "epoch": 0.28066759050893475,
+      "grad_norm": 920.887451171875,
+      "learning_rate": 4.801542979995245e-06,
+      "loss": 40.4103,
+      "step": 69470
+    },
+    {
+      "epoch": 0.2807079917743024,
+      "grad_norm": 520.3526000976562,
+      "learning_rate": 4.800133732226135e-06,
+      "loss": 46.0557,
+      "step": 69480
+    },
+    {
+      "epoch": 0.28074839303967003,
+      "grad_norm": 504.0417175292969,
+      "learning_rate": 4.798724500359467e-06,
+      "loss": 32.6267,
+      "step": 69490
+    },
+    {
+      "epoch": 0.2807887943050376,
+      "grad_norm": 643.9005126953125,
+      "learning_rate": 4.7973152845073666e-06,
+      "loss": 88.861,
+      "step": 69500
+    },
+    {
+      "epoch": 0.28082919557040525,
+      "grad_norm": 539.6871948242188,
+      "learning_rate": 4.795906084781958e-06,
+      "loss": 43.0512,
+      "step": 69510
+    },
+    {
+      "epoch": 0.2808695968357729,
+      "grad_norm": 347.9564208984375,
+      "learning_rate": 4.7944969012953656e-06,
+      "loss": 53.9477,
+      "step": 69520
+    },
+    {
+      "epoch": 0.28090999810114053,
+      "grad_norm": 591.0164794921875,
+      "learning_rate": 4.793087734159711e-06,
+      "loss": 49.4503,
+      "step": 69530
+    },
+    {
+      "epoch": 0.2809503993665082,
+      "grad_norm": 681.713623046875,
+      "learning_rate": 4.791678583487118e-06,
+      "loss": 48.3232,
+      "step": 69540
+    },
+    {
+      "epoch": 0.2809908006318758,
+      "grad_norm": 392.1138610839844,
+      "learning_rate": 4.790269449389703e-06,
+      "loss": 43.5036,
+      "step": 69550
+    },
+    {
+      "epoch": 0.2810312018972434,
+      "grad_norm": 160.6968231201172,
+      "learning_rate": 4.788860331979586e-06,
+      "loss": 59.5492,
+      "step": 69560
+    },
+    {
+      "epoch": 0.28107160316261104,
+      "grad_norm": 382.0798034667969,
+      "learning_rate": 4.787451231368883e-06,
+      "loss": 35.5043,
+      "step": 69570
+    },
+    {
+      "epoch": 0.2811120044279787,
+      "grad_norm": 1062.69970703125,
+      "learning_rate": 4.786042147669709e-06,
+      "loss": 46.8691,
+      "step": 69580
+    },
+    {
+      "epoch": 0.2811524056933463,
+      "grad_norm": 397.5672607421875,
+      "learning_rate": 4.784633080994181e-06,
+      "loss": 29.8181,
+      "step": 69590
+    },
+    {
+      "epoch": 0.28119280695871396,
+      "grad_norm": 1378.0069580078125,
+      "learning_rate": 4.783224031454409e-06,
+      "loss": 55.9483,
+      "step": 69600
+    },
+    {
+      "epoch": 0.2812332082240816,
+      "grad_norm": 929.8328857421875,
+      "learning_rate": 4.781814999162507e-06,
+      "loss": 50.3427,
+      "step": 69610
+    },
+    {
+      "epoch": 0.2812736094894492,
+      "grad_norm": 401.11334228515625,
+      "learning_rate": 4.780405984230582e-06,
+      "loss": 37.6489,
+      "step": 69620
+    },
+    {
+      "epoch": 0.2813140107548168,
+      "grad_norm": 762.3184814453125,
+      "learning_rate": 4.778996986770747e-06,
+      "loss": 45.4038,
+      "step": 69630
+    },
+    {
+      "epoch": 0.28135441202018446,
+      "grad_norm": 762.9951171875,
+      "learning_rate": 4.777588006895109e-06,
+      "loss": 53.0929,
+      "step": 69640
+    },
+    {
+      "epoch": 0.2813948132855521,
+      "grad_norm": 726.9749755859375,
+      "learning_rate": 4.77617904471577e-06,
+      "loss": 40.7251,
+      "step": 69650
+    },
+    {
+      "epoch": 0.28143521455091974,
+      "grad_norm": 578.7279663085938,
+      "learning_rate": 4.774770100344838e-06,
+      "loss": 48.0613,
+      "step": 69660
+    },
+    {
+      "epoch": 0.2814756158162874,
+      "grad_norm": 322.8981628417969,
+      "learning_rate": 4.7733611738944155e-06,
+      "loss": 55.718,
+      "step": 69670
+    },
+    {
+      "epoch": 0.281516017081655,
+      "grad_norm": 658.8078002929688,
+      "learning_rate": 4.7719522654766044e-06,
+      "loss": 26.6201,
+      "step": 69680
+    },
+    {
+      "epoch": 0.2815564183470226,
+      "grad_norm": 553.6875,
+      "learning_rate": 4.7705433752035045e-06,
+      "loss": 59.374,
+      "step": 69690
+    },
+    {
+      "epoch": 0.28159681961239025,
+      "grad_norm": 843.7166748046875,
+      "learning_rate": 4.7691345031872156e-06,
+      "loss": 41.141,
+      "step": 69700
+    },
+    {
+      "epoch": 0.2816372208777579,
+      "grad_norm": 402.2727355957031,
+      "learning_rate": 4.767725649539833e-06,
+      "loss": 37.5811,
+      "step": 69710
+    },
+    {
+      "epoch": 0.28167762214312553,
+      "grad_norm": 288.26202392578125,
+      "learning_rate": 4.766316814373458e-06,
+      "loss": 55.0614,
+      "step": 69720
+    },
+    {
+      "epoch": 0.28171802340849317,
+      "grad_norm": 320.1476745605469,
+      "learning_rate": 4.76490799780018e-06,
+      "loss": 39.0787,
+      "step": 69730
+    },
+    {
+      "epoch": 0.2817584246738608,
+      "grad_norm": 546.8268432617188,
+      "learning_rate": 4.763499199932093e-06,
+      "loss": 64.9588,
+      "step": 69740
+    },
+    {
+      "epoch": 0.2817988259392284,
+      "grad_norm": 594.4404907226562,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 40.4867,
+      "step": 69750
+    },
+    {
+      "epoch": 0.28183922720459603,
+      "grad_norm": 742.597412109375,
+      "learning_rate": 4.760681660759859e-06,
+      "loss": 46.6788,
+      "step": 69760
+    },
+    {
+      "epoch": 0.2818796284699637,
+      "grad_norm": 452.48114013671875,
+      "learning_rate": 4.7592729196798905e-06,
+      "loss": 44.5193,
+      "step": 69770
+    },
+    {
+      "epoch": 0.2819200297353313,
+      "grad_norm": 762.345703125,
+      "learning_rate": 4.757864197753472e-06,
+      "loss": 45.2634,
+      "step": 69780
+    },
+    {
+      "epoch": 0.28196043100069895,
+      "grad_norm": 723.4268188476562,
+      "learning_rate": 4.7564554950926876e-06,
+      "loss": 47.0695,
+      "step": 69790
+    },
+    {
+      "epoch": 0.2820008322660666,
+      "grad_norm": 645.5535888671875,
+      "learning_rate": 4.755046811809621e-06,
+      "loss": 41.8543,
+      "step": 69800
+    },
+    {
+      "epoch": 0.28204123353143423,
+      "grad_norm": 886.3253784179688,
+      "learning_rate": 4.7536381480163575e-06,
+      "loss": 63.1216,
+      "step": 69810
+    },
+    {
+      "epoch": 0.2820816347968018,
+      "grad_norm": 304.3733825683594,
+      "learning_rate": 4.752229503824974e-06,
+      "loss": 27.179,
+      "step": 69820
+    },
+    {
+      "epoch": 0.28212203606216946,
+      "grad_norm": 712.4049072265625,
+      "learning_rate": 4.7508208793475515e-06,
+      "loss": 50.8489,
+      "step": 69830
+    },
+    {
+      "epoch": 0.2821624373275371,
+      "grad_norm": 627.2713012695312,
+      "learning_rate": 4.749412274696169e-06,
+      "loss": 65.211,
+      "step": 69840
+    },
+    {
+      "epoch": 0.28220283859290474,
+      "grad_norm": 517.7521362304688,
+      "learning_rate": 4.748003689982901e-06,
+      "loss": 38.2601,
+      "step": 69850
+    },
+    {
+      "epoch": 0.2822432398582724,
+      "grad_norm": 901.2315063476562,
+      "learning_rate": 4.746595125319823e-06,
+      "loss": 38.5679,
+      "step": 69860
+    },
+    {
+      "epoch": 0.28228364112364,
+      "grad_norm": 483.2677917480469,
+      "learning_rate": 4.745186580819008e-06,
+      "loss": 51.5532,
+      "step": 69870
+    },
+    {
+      "epoch": 0.2823240423890076,
+      "grad_norm": 455.71343994140625,
+      "learning_rate": 4.743778056592528e-06,
+      "loss": 50.9613,
+      "step": 69880
+    },
+    {
+      "epoch": 0.28236444365437524,
+      "grad_norm": 576.9340209960938,
+      "learning_rate": 4.742369552752453e-06,
+      "loss": 61.8175,
+      "step": 69890
+    },
+    {
+      "epoch": 0.2824048449197429,
+      "grad_norm": 586.7548217773438,
+      "learning_rate": 4.740961069410848e-06,
+      "loss": 38.9002,
+      "step": 69900
+    },
+    {
+      "epoch": 0.2824452461851105,
+      "grad_norm": 273.7797546386719,
+      "learning_rate": 4.7395526066797835e-06,
+      "loss": 39.9881,
+      "step": 69910
+    },
+    {
+      "epoch": 0.28248564745047816,
+      "grad_norm": 414.7781677246094,
+      "learning_rate": 4.738144164671322e-06,
+      "loss": 32.6235,
+      "step": 69920
+    },
+    {
+      "epoch": 0.2825260487158458,
+      "grad_norm": 523.3553466796875,
+      "learning_rate": 4.736735743497528e-06,
+      "loss": 35.1946,
+      "step": 69930
+    },
+    {
+      "epoch": 0.2825664499812134,
+      "grad_norm": 704.5969848632812,
+      "learning_rate": 4.735327343270461e-06,
+      "loss": 43.2801,
+      "step": 69940
+    },
+    {
+      "epoch": 0.282606851246581,
+      "grad_norm": 715.0891723632812,
+      "learning_rate": 4.733918964102185e-06,
+      "loss": 47.1886,
+      "step": 69950
+    },
+    {
+      "epoch": 0.28264725251194867,
+      "grad_norm": 824.6829833984375,
+      "learning_rate": 4.732510606104754e-06,
+      "loss": 37.4036,
+      "step": 69960
+    },
+    {
+      "epoch": 0.2826876537773163,
+      "grad_norm": 535.701904296875,
+      "learning_rate": 4.731102269390227e-06,
+      "loss": 66.5,
+      "step": 69970
+    },
+    {
+      "epoch": 0.28272805504268395,
+      "grad_norm": 602.7445678710938,
+      "learning_rate": 4.729693954070661e-06,
+      "loss": 34.4767,
+      "step": 69980
+    },
+    {
+      "epoch": 0.2827684563080516,
+      "grad_norm": 102.27546691894531,
+      "learning_rate": 4.728285660258104e-06,
+      "loss": 25.0424,
+      "step": 69990
+    },
+    {
+      "epoch": 0.2828088575734192,
+      "grad_norm": 678.1873168945312,
+      "learning_rate": 4.726877388064609e-06,
+      "loss": 50.8831,
+      "step": 70000
+    },
+    {
+      "epoch": 0.2828492588387868,
+      "grad_norm": 401.9143981933594,
+      "learning_rate": 4.725469137602229e-06,
+      "loss": 49.9991,
+      "step": 70010
+    },
+    {
+      "epoch": 0.28288966010415445,
+      "grad_norm": 745.3547973632812,
+      "learning_rate": 4.724060908983008e-06,
+      "loss": 39.7754,
+      "step": 70020
+    },
+    {
+      "epoch": 0.2829300613695221,
+      "grad_norm": 663.1378173828125,
+      "learning_rate": 4.7226527023189954e-06,
+      "loss": 42.3307,
+      "step": 70030
+    },
+    {
+      "epoch": 0.28297046263488973,
+      "grad_norm": 273.6854248046875,
+      "learning_rate": 4.721244517722233e-06,
+      "loss": 42.688,
+      "step": 70040
+    },
+    {
+      "epoch": 0.28301086390025737,
+      "grad_norm": 568.3843994140625,
+      "learning_rate": 4.719836355304766e-06,
+      "loss": 56.1069,
+      "step": 70050
+    },
+    {
+      "epoch": 0.283051265165625,
+      "grad_norm": 619.1213989257812,
+      "learning_rate": 4.718428215178634e-06,
+      "loss": 40.7905,
+      "step": 70060
+    },
+    {
+      "epoch": 0.2830916664309926,
+      "grad_norm": 857.2811889648438,
+      "learning_rate": 4.717020097455879e-06,
+      "loss": 54.3016,
+      "step": 70070
+    },
+    {
+      "epoch": 0.28313206769636023,
+      "grad_norm": 300.54571533203125,
+      "learning_rate": 4.715612002248533e-06,
+      "loss": 39.4976,
+      "step": 70080
+    },
+    {
+      "epoch": 0.2831724689617279,
+      "grad_norm": 1189.8184814453125,
+      "learning_rate": 4.714203929668637e-06,
+      "loss": 43.9997,
+      "step": 70090
+    },
+    {
+      "epoch": 0.2832128702270955,
+      "grad_norm": 450.262939453125,
+      "learning_rate": 4.712795879828221e-06,
+      "loss": 42.1777,
+      "step": 70100
+    },
+    {
+      "epoch": 0.28325327149246315,
+      "grad_norm": 565.9345092773438,
+      "learning_rate": 4.71138785283932e-06,
+      "loss": 44.1722,
+      "step": 70110
+    },
+    {
+      "epoch": 0.2832936727578308,
+      "grad_norm": 372.7854309082031,
+      "learning_rate": 4.709979848813963e-06,
+      "loss": 34.5973,
+      "step": 70120
+    },
+    {
+      "epoch": 0.2833340740231984,
+      "grad_norm": 623.8758544921875,
+      "learning_rate": 4.7085718678641776e-06,
+      "loss": 43.4469,
+      "step": 70130
+    },
+    {
+      "epoch": 0.283374475288566,
+      "grad_norm": 970.5692138671875,
+      "learning_rate": 4.70716391010199e-06,
+      "loss": 50.1333,
+      "step": 70140
+    },
+    {
+      "epoch": 0.28341487655393366,
+      "grad_norm": 793.2423095703125,
+      "learning_rate": 4.70575597563943e-06,
+      "loss": 39.2793,
+      "step": 70150
+    },
+    {
+      "epoch": 0.2834552778193013,
+      "grad_norm": 482.6946716308594,
+      "learning_rate": 4.704348064588514e-06,
+      "loss": 41.613,
+      "step": 70160
+    },
+    {
+      "epoch": 0.28349567908466894,
+      "grad_norm": 557.0717163085938,
+      "learning_rate": 4.702940177061266e-06,
+      "loss": 43.9438,
+      "step": 70170
+    },
+    {
+      "epoch": 0.2835360803500366,
+      "grad_norm": 970.6007690429688,
+      "learning_rate": 4.7015323131697035e-06,
+      "loss": 44.938,
+      "step": 70180
+    },
+    {
+      "epoch": 0.2835764816154042,
+      "grad_norm": 463.7577209472656,
+      "learning_rate": 4.700124473025846e-06,
+      "loss": 33.1951,
+      "step": 70190
+    },
+    {
+      "epoch": 0.2836168828807718,
+      "grad_norm": 236.90660095214844,
+      "learning_rate": 4.6987166567417085e-06,
+      "loss": 24.4781,
+      "step": 70200
+    },
+    {
+      "epoch": 0.28365728414613944,
+      "grad_norm": 600.7906494140625,
+      "learning_rate": 4.697308864429303e-06,
+      "loss": 35.3102,
+      "step": 70210
+    },
+    {
+      "epoch": 0.2836976854115071,
+      "grad_norm": 244.70358276367188,
+      "learning_rate": 4.695901096200643e-06,
+      "loss": 35.8941,
+      "step": 70220
+    },
+    {
+      "epoch": 0.2837380866768747,
+      "grad_norm": 543.962646484375,
+      "learning_rate": 4.694493352167736e-06,
+      "loss": 57.4858,
+      "step": 70230
+    },
+    {
+      "epoch": 0.28377848794224236,
+      "grad_norm": 584.345947265625,
+      "learning_rate": 4.693085632442593e-06,
+      "loss": 33.6563,
+      "step": 70240
+    },
+    {
+      "epoch": 0.28381888920761,
+      "grad_norm": 627.2140502929688,
+      "learning_rate": 4.691677937137217e-06,
+      "loss": 47.8382,
+      "step": 70250
+    },
+    {
+      "epoch": 0.2838592904729776,
+      "grad_norm": 156.89602661132812,
+      "learning_rate": 4.690270266363612e-06,
+      "loss": 48.3714,
+      "step": 70260
+    },
+    {
+      "epoch": 0.28389969173834523,
+      "grad_norm": 591.1685791015625,
+      "learning_rate": 4.688862620233779e-06,
+      "loss": 31.6053,
+      "step": 70270
+    },
+    {
+      "epoch": 0.28394009300371287,
+      "grad_norm": 174.83114624023438,
+      "learning_rate": 4.687454998859721e-06,
+      "loss": 27.5564,
+      "step": 70280
+    },
+    {
+      "epoch": 0.2839804942690805,
+      "grad_norm": 489.0082092285156,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 51.9983,
+      "step": 70290
+    },
+    {
+      "epoch": 0.28402089553444815,
+      "grad_norm": 3722.665771484375,
+      "learning_rate": 4.684639830826913e-06,
+      "loss": 57.5925,
+      "step": 70300
+    },
+    {
+      "epoch": 0.2840612967998158,
+      "grad_norm": 668.0916137695312,
+      "learning_rate": 4.683232284392155e-06,
+      "loss": 53.0558,
+      "step": 70310
+    },
+    {
+      "epoch": 0.28410169806518343,
+      "grad_norm": 560.221923828125,
+      "learning_rate": 4.681824763161151e-06,
+      "loss": 46.5698,
+      "step": 70320
+    },
+    {
+      "epoch": 0.284142099330551,
+      "grad_norm": 267.6634826660156,
+      "learning_rate": 4.6804172672458905e-06,
+      "loss": 37.9598,
+      "step": 70330
+    },
+    {
+      "epoch": 0.28418250059591865,
+      "grad_norm": 385.9459533691406,
+      "learning_rate": 4.67900979675836e-06,
+      "loss": 44.058,
+      "step": 70340
+    },
+    {
+      "epoch": 0.2842229018612863,
+      "grad_norm": 889.5348510742188,
+      "learning_rate": 4.677602351810547e-06,
+      "loss": 49.9251,
+      "step": 70350
+    },
+    {
+      "epoch": 0.28426330312665393,
+      "grad_norm": 780.3435668945312,
+      "learning_rate": 4.676194932514435e-06,
+      "loss": 35.594,
+      "step": 70360
+    },
+    {
+      "epoch": 0.2843037043920216,
+      "grad_norm": 920.3884887695312,
+      "learning_rate": 4.674787538982006e-06,
+      "loss": 66.1759,
+      "step": 70370
+    },
+    {
+      "epoch": 0.2843441056573892,
+      "grad_norm": 562.5645751953125,
+      "learning_rate": 4.6733801713252405e-06,
+      "loss": 33.8848,
+      "step": 70380
+    },
+    {
+      "epoch": 0.2843845069227568,
+      "grad_norm": 711.589599609375,
+      "learning_rate": 4.671972829656116e-06,
+      "loss": 55.0332,
+      "step": 70390
+    },
+    {
+      "epoch": 0.28442490818812444,
+      "grad_norm": 682.6295776367188,
+      "learning_rate": 4.670565514086607e-06,
+      "loss": 47.3515,
+      "step": 70400
+    },
+    {
+      "epoch": 0.2844653094534921,
+      "grad_norm": 227.3060302734375,
+      "learning_rate": 4.669158224728691e-06,
+      "loss": 36.4143,
+      "step": 70410
+    },
+    {
+      "epoch": 0.2845057107188597,
+      "grad_norm": 2018.75048828125,
+      "learning_rate": 4.667750961694334e-06,
+      "loss": 30.8619,
+      "step": 70420
+    },
+    {
+      "epoch": 0.28454611198422736,
+      "grad_norm": 277.3977355957031,
+      "learning_rate": 4.666343725095509e-06,
+      "loss": 41.897,
+      "step": 70430
+    },
+    {
+      "epoch": 0.284586513249595,
+      "grad_norm": 90.26355743408203,
+      "learning_rate": 4.6649365150441825e-06,
+      "loss": 40.3659,
+      "step": 70440
+    },
+    {
+      "epoch": 0.2846269145149626,
+      "grad_norm": 364.5423583984375,
+      "learning_rate": 4.66352933165232e-06,
+      "loss": 37.3343,
+      "step": 70450
+    },
+    {
+      "epoch": 0.2846673157803302,
+      "grad_norm": 640.2406005859375,
+      "learning_rate": 4.6621221750318835e-06,
+      "loss": 51.8274,
+      "step": 70460
+    },
+    {
+      "epoch": 0.28470771704569786,
+      "grad_norm": 935.7151489257812,
+      "learning_rate": 4.660715045294834e-06,
+      "loss": 33.1354,
+      "step": 70470
+    },
+    {
+      "epoch": 0.2847481183110655,
+      "grad_norm": 367.4470520019531,
+      "learning_rate": 4.659307942553133e-06,
+      "loss": 32.8019,
+      "step": 70480
+    },
+    {
+      "epoch": 0.28478851957643314,
+      "grad_norm": 305.1088562011719,
+      "learning_rate": 4.657900866918735e-06,
+      "loss": 50.5888,
+      "step": 70490
+    },
+    {
+      "epoch": 0.2848289208418008,
+      "grad_norm": 319.4598388671875,
+      "learning_rate": 4.6564938185035954e-06,
+      "loss": 42.0434,
+      "step": 70500
+    },
+    {
+      "epoch": 0.2848693221071684,
+      "grad_norm": 491.0206604003906,
+      "learning_rate": 4.655086797419666e-06,
+      "loss": 50.2816,
+      "step": 70510
+    },
+    {
+      "epoch": 0.284909723372536,
+      "grad_norm": 647.287353515625,
+      "learning_rate": 4.653679803778897e-06,
+      "loss": 58.8979,
+      "step": 70520
+    },
+    {
+      "epoch": 0.28495012463790365,
+      "grad_norm": 276.81549072265625,
+      "learning_rate": 4.652272837693237e-06,
+      "loss": 54.2775,
+      "step": 70530
+    },
+    {
+      "epoch": 0.2849905259032713,
+      "grad_norm": 768.6326904296875,
+      "learning_rate": 4.650865899274632e-06,
+      "loss": 61.2162,
+      "step": 70540
+    },
+    {
+      "epoch": 0.2850309271686389,
+      "grad_norm": 990.27294921875,
+      "learning_rate": 4.649458988635023e-06,
+      "loss": 60.081,
+      "step": 70550
+    },
+    {
+      "epoch": 0.28507132843400657,
+      "grad_norm": 754.9923706054688,
+      "learning_rate": 4.6480521058863546e-06,
+      "loss": 43.4132,
+      "step": 70560
+    },
+    {
+      "epoch": 0.2851117296993742,
+      "grad_norm": 635.0836791992188,
+      "learning_rate": 4.646645251140564e-06,
+      "loss": 26.996,
+      "step": 70570
+    },
+    {
+      "epoch": 0.2851521309647418,
+      "grad_norm": 1053.806640625,
+      "learning_rate": 4.6452384245095924e-06,
+      "loss": 30.2651,
+      "step": 70580
+    },
+    {
+      "epoch": 0.28519253223010943,
+      "grad_norm": 500.2285461425781,
+      "learning_rate": 4.643831626105369e-06,
+      "loss": 45.7198,
+      "step": 70590
+    },
+    {
+      "epoch": 0.28523293349547707,
+      "grad_norm": 801.8435668945312,
+      "learning_rate": 4.642424856039827e-06,
+      "loss": 43.3361,
+      "step": 70600
+    },
+    {
+      "epoch": 0.2852733347608447,
+      "grad_norm": 685.59033203125,
+      "learning_rate": 4.6410181144249e-06,
+      "loss": 54.9417,
+      "step": 70610
+    },
+    {
+      "epoch": 0.28531373602621235,
+      "grad_norm": 471.1318054199219,
+      "learning_rate": 4.639611401372514e-06,
+      "loss": 44.3713,
+      "step": 70620
+    },
+    {
+      "epoch": 0.28535413729158,
+      "grad_norm": 1329.538818359375,
+      "learning_rate": 4.638204716994594e-06,
+      "loss": 52.105,
+      "step": 70630
+    },
+    {
+      "epoch": 0.28539453855694763,
+      "grad_norm": 437.1035461425781,
+      "learning_rate": 4.636798061403065e-06,
+      "loss": 38.8913,
+      "step": 70640
+    },
+    {
+      "epoch": 0.2854349398223152,
+      "grad_norm": 517.3322143554688,
+      "learning_rate": 4.635391434709847e-06,
+      "loss": 69.1764,
+      "step": 70650
+    },
+    {
+      "epoch": 0.28547534108768285,
+      "grad_norm": 502.039306640625,
+      "learning_rate": 4.6339848370268585e-06,
+      "loss": 37.4621,
+      "step": 70660
+    },
+    {
+      "epoch": 0.2855157423530505,
+      "grad_norm": 329.94049072265625,
+      "learning_rate": 4.63257826846602e-06,
+      "loss": 41.4155,
+      "step": 70670
+    },
+    {
+      "epoch": 0.28555614361841813,
+      "grad_norm": 1370.1334228515625,
+      "learning_rate": 4.6311717291392396e-06,
+      "loss": 33.5599,
+      "step": 70680
+    },
+    {
+      "epoch": 0.2855965448837858,
+      "grad_norm": 773.607177734375,
+      "learning_rate": 4.629765219158433e-06,
+      "loss": 48.0365,
+      "step": 70690
+    },
+    {
+      "epoch": 0.2856369461491534,
+      "grad_norm": 497.14300537109375,
+      "learning_rate": 4.628358738635507e-06,
+      "loss": 41.8579,
+      "step": 70700
+    },
+    {
+      "epoch": 0.285677347414521,
+      "grad_norm": 1049.352294921875,
+      "learning_rate": 4.626952287682372e-06,
+      "loss": 44.9301,
+      "step": 70710
+    },
+    {
+      "epoch": 0.28571774867988864,
+      "grad_norm": 446.8216552734375,
+      "learning_rate": 4.6255458664109306e-06,
+      "loss": 38.319,
+      "step": 70720
+    },
+    {
+      "epoch": 0.2857581499452563,
+      "grad_norm": 256.68109130859375,
+      "learning_rate": 4.624139474933087e-06,
+      "loss": 41.6502,
+      "step": 70730
+    },
+    {
+      "epoch": 0.2857985512106239,
+      "grad_norm": 548.9017944335938,
+      "learning_rate": 4.62273311336074e-06,
+      "loss": 37.4628,
+      "step": 70740
+    },
+    {
+      "epoch": 0.28583895247599156,
+      "grad_norm": 468.376708984375,
+      "learning_rate": 4.62132678180579e-06,
+      "loss": 51.4482,
+      "step": 70750
+    },
+    {
+      "epoch": 0.2858793537413592,
+      "grad_norm": 866.9331665039062,
+      "learning_rate": 4.619920480380127e-06,
+      "loss": 52.7439,
+      "step": 70760
+    },
+    {
+      "epoch": 0.2859197550067268,
+      "grad_norm": 532.3231201171875,
+      "learning_rate": 4.618514209195648e-06,
+      "loss": 54.0081,
+      "step": 70770
+    },
+    {
+      "epoch": 0.2859601562720944,
+      "grad_norm": 326.88629150390625,
+      "learning_rate": 4.617107968364243e-06,
+      "loss": 78.2161,
+      "step": 70780
+    },
+    {
+      "epoch": 0.28600055753746206,
+      "grad_norm": 436.6565856933594,
+      "learning_rate": 4.615701757997799e-06,
+      "loss": 53.2956,
+      "step": 70790
+    },
+    {
+      "epoch": 0.2860409588028297,
+      "grad_norm": 329.3333435058594,
+      "learning_rate": 4.614295578208202e-06,
+      "loss": 43.7113,
+      "step": 70800
+    },
+    {
+      "epoch": 0.28608136006819734,
+      "grad_norm": 401.3062438964844,
+      "learning_rate": 4.612889429107337e-06,
+      "loss": 61.199,
+      "step": 70810
+    },
+    {
+      "epoch": 0.286121761333565,
+      "grad_norm": 983.342529296875,
+      "learning_rate": 4.611483310807082e-06,
+      "loss": 51.3303,
+      "step": 70820
+    },
+    {
+      "epoch": 0.2861621625989326,
+      "grad_norm": 571.834716796875,
+      "learning_rate": 4.610077223419319e-06,
+      "loss": 37.479,
+      "step": 70830
+    },
+    {
+      "epoch": 0.2862025638643002,
+      "grad_norm": 622.2101440429688,
+      "learning_rate": 4.608671167055922e-06,
+      "loss": 41.2861,
+      "step": 70840
+    },
+    {
+      "epoch": 0.28624296512966785,
+      "grad_norm": 771.7720336914062,
+      "learning_rate": 4.607265141828762e-06,
+      "loss": 46.7046,
+      "step": 70850
+    },
+    {
+      "epoch": 0.2862833663950355,
+      "grad_norm": 733.5690307617188,
+      "learning_rate": 4.605859147849713e-06,
+      "loss": 50.6617,
+      "step": 70860
+    },
+    {
+      "epoch": 0.28632376766040313,
+      "grad_norm": 564.01513671875,
+      "learning_rate": 4.604453185230643e-06,
+      "loss": 35.724,
+      "step": 70870
+    },
+    {
+      "epoch": 0.28636416892577077,
+      "grad_norm": 383.891845703125,
+      "learning_rate": 4.603047254083418e-06,
+      "loss": 26.681,
+      "step": 70880
+    },
+    {
+      "epoch": 0.2864045701911384,
+      "grad_norm": 464.1899719238281,
+      "learning_rate": 4.601641354519901e-06,
+      "loss": 39.4786,
+      "step": 70890
+    },
+    {
+      "epoch": 0.286444971456506,
+      "grad_norm": 455.7287292480469,
+      "learning_rate": 4.6002354866519526e-06,
+      "loss": 53.9625,
+      "step": 70900
+    },
+    {
+      "epoch": 0.28648537272187363,
+      "grad_norm": 873.5855102539062,
+      "learning_rate": 4.598829650591432e-06,
+      "loss": 36.5101,
+      "step": 70910
+    },
+    {
+      "epoch": 0.2865257739872413,
+      "grad_norm": 599.9312133789062,
+      "learning_rate": 4.597423846450196e-06,
+      "loss": 40.5595,
+      "step": 70920
+    },
+    {
+      "epoch": 0.2865661752526089,
+      "grad_norm": 925.1723022460938,
+      "learning_rate": 4.596018074340097e-06,
+      "loss": 55.3253,
+      "step": 70930
+    },
+    {
+      "epoch": 0.28660657651797655,
+      "grad_norm": 484.8789367675781,
+      "learning_rate": 4.594612334372985e-06,
+      "loss": 52.8668,
+      "step": 70940
+    },
+    {
+      "epoch": 0.2866469777833442,
+      "grad_norm": 645.4853515625,
+      "learning_rate": 4.59320662666071e-06,
+      "loss": 42.3059,
+      "step": 70950
+    },
+    {
+      "epoch": 0.28668737904871183,
+      "grad_norm": 333.7181701660156,
+      "learning_rate": 4.591800951315116e-06,
+      "loss": 77.4099,
+      "step": 70960
+    },
+    {
+      "epoch": 0.2867277803140794,
+      "grad_norm": 438.87396240234375,
+      "learning_rate": 4.590395308448046e-06,
+      "loss": 53.1856,
+      "step": 70970
+    },
+    {
+      "epoch": 0.28676818157944706,
+      "grad_norm": 698.4015502929688,
+      "learning_rate": 4.588989698171343e-06,
+      "loss": 54.2699,
+      "step": 70980
+    },
+    {
+      "epoch": 0.2868085828448147,
+      "grad_norm": 607.1814575195312,
+      "learning_rate": 4.587584120596842e-06,
+      "loss": 39.6917,
+      "step": 70990
+    },
+    {
+      "epoch": 0.28684898411018234,
+      "grad_norm": 1104.1712646484375,
+      "learning_rate": 4.58617857583638e-06,
+      "loss": 41.4022,
+      "step": 71000
+    },
+    {
+      "epoch": 0.28688938537555,
+      "grad_norm": 380.30560302734375,
+      "learning_rate": 4.5847730640017926e-06,
+      "loss": 43.8242,
+      "step": 71010
+    },
+    {
+      "epoch": 0.2869297866409176,
+      "grad_norm": 425.9400329589844,
+      "learning_rate": 4.5833675852049045e-06,
+      "loss": 46.4129,
+      "step": 71020
+    },
+    {
+      "epoch": 0.2869701879062852,
+      "grad_norm": 443.4739074707031,
+      "learning_rate": 4.5819621395575445e-06,
+      "loss": 37.8977,
+      "step": 71030
+    },
+    {
+      "epoch": 0.28701058917165284,
+      "grad_norm": 384.0914001464844,
+      "learning_rate": 4.5805567271715395e-06,
+      "loss": 28.2375,
+      "step": 71040
+    },
+    {
+      "epoch": 0.2870509904370205,
+      "grad_norm": 606.654541015625,
+      "learning_rate": 4.5791513481587105e-06,
+      "loss": 40.5883,
+      "step": 71050
+    },
+    {
+      "epoch": 0.2870913917023881,
+      "grad_norm": 816.3211669921875,
+      "learning_rate": 4.577746002630878e-06,
+      "loss": 38.1152,
+      "step": 71060
+    },
+    {
+      "epoch": 0.28713179296775576,
+      "grad_norm": 1249.8865966796875,
+      "learning_rate": 4.576340690699857e-06,
+      "loss": 26.9667,
+      "step": 71070
+    },
+    {
+      "epoch": 0.2871721942331234,
+      "grad_norm": 846.2115478515625,
+      "learning_rate": 4.574935412477464e-06,
+      "loss": 34.3009,
+      "step": 71080
+    },
+    {
+      "epoch": 0.287212595498491,
+      "grad_norm": 278.2926940917969,
+      "learning_rate": 4.573530168075508e-06,
+      "loss": 38.9183,
+      "step": 71090
+    },
+    {
+      "epoch": 0.2872529967638586,
+      "grad_norm": 823.379150390625,
+      "learning_rate": 4.572124957605803e-06,
+      "loss": 47.3542,
+      "step": 71100
+    },
+    {
+      "epoch": 0.28729339802922627,
+      "grad_norm": 279.0650939941406,
+      "learning_rate": 4.5707197811801484e-06,
+      "loss": 36.309,
+      "step": 71110
+    },
+    {
+      "epoch": 0.2873337992945939,
+      "grad_norm": 255.07122802734375,
+      "learning_rate": 4.569314638910352e-06,
+      "loss": 22.5709,
+      "step": 71120
+    },
+    {
+      "epoch": 0.28737420055996155,
+      "grad_norm": 752.4155883789062,
+      "learning_rate": 4.56790953090821e-06,
+      "loss": 65.827,
+      "step": 71130
+    },
+    {
+      "epoch": 0.2874146018253292,
+      "grad_norm": 731.970458984375,
+      "learning_rate": 4.566504457285527e-06,
+      "loss": 46.3308,
+      "step": 71140
+    },
+    {
+      "epoch": 0.2874550030906968,
+      "grad_norm": 5122.3427734375,
+      "learning_rate": 4.565099418154093e-06,
+      "loss": 75.2653,
+      "step": 71150
+    },
+    {
+      "epoch": 0.2874954043560644,
+      "grad_norm": 815.7695922851562,
+      "learning_rate": 4.563694413625703e-06,
+      "loss": 47.1574,
+      "step": 71160
+    },
+    {
+      "epoch": 0.28753580562143205,
+      "grad_norm": 724.8348999023438,
+      "learning_rate": 4.5622894438121465e-06,
+      "loss": 30.6963,
+      "step": 71170
+    },
+    {
+      "epoch": 0.2875762068867997,
+      "grad_norm": 459.6385498046875,
+      "learning_rate": 4.560884508825212e-06,
+      "loss": 49.0495,
+      "step": 71180
+    },
+    {
+      "epoch": 0.28761660815216733,
+      "grad_norm": 1621.1512451171875,
+      "learning_rate": 4.559479608776679e-06,
+      "loss": 74.2594,
+      "step": 71190
+    },
+    {
+      "epoch": 0.28765700941753497,
+      "grad_norm": 684.2808227539062,
+      "learning_rate": 4.558074743778333e-06,
+      "loss": 46.0973,
+      "step": 71200
+    },
+    {
+      "epoch": 0.2876974106829026,
+      "grad_norm": 600.880859375,
+      "learning_rate": 4.556669913941951e-06,
+      "loss": 40.5554,
+      "step": 71210
+    },
+    {
+      "epoch": 0.2877378119482702,
+      "grad_norm": 1257.7164306640625,
+      "learning_rate": 4.555265119379308e-06,
+      "loss": 42.7592,
+      "step": 71220
+    },
+    {
+      "epoch": 0.28777821321363783,
+      "grad_norm": 493.02850341796875,
+      "learning_rate": 4.55386036020218e-06,
+      "loss": 32.3741,
+      "step": 71230
+    },
+    {
+      "epoch": 0.2878186144790055,
+      "grad_norm": 273.1938171386719,
+      "learning_rate": 4.552455636522335e-06,
+      "loss": 54.3847,
+      "step": 71240
+    },
+    {
+      "epoch": 0.2878590157443731,
+      "grad_norm": 698.703369140625,
+      "learning_rate": 4.551050948451542e-06,
+      "loss": 54.6396,
+      "step": 71250
+    },
+    {
+      "epoch": 0.28789941700974075,
+      "grad_norm": 141.8015594482422,
+      "learning_rate": 4.549646296101564e-06,
+      "loss": 53.9065,
+      "step": 71260
+    },
+    {
+      "epoch": 0.2879398182751084,
+      "grad_norm": 335.04571533203125,
+      "learning_rate": 4.548241679584165e-06,
+      "loss": 40.8237,
+      "step": 71270
+    },
+    {
+      "epoch": 0.28798021954047603,
+      "grad_norm": 483.5780029296875,
+      "learning_rate": 4.546837099011101e-06,
+      "loss": 63.2328,
+      "step": 71280
+    },
+    {
+      "epoch": 0.2880206208058436,
+      "grad_norm": 878.0452270507812,
+      "learning_rate": 4.545432554494128e-06,
+      "loss": 39.5612,
+      "step": 71290
+    },
+    {
+      "epoch": 0.28806102207121126,
+      "grad_norm": 244.78079223632812,
+      "learning_rate": 4.544028046145002e-06,
+      "loss": 41.3371,
+      "step": 71300
+    },
+    {
+      "epoch": 0.2881014233365789,
+      "grad_norm": 905.211669921875,
+      "learning_rate": 4.542623574075471e-06,
+      "loss": 42.8943,
+      "step": 71310
+    },
+    {
+      "epoch": 0.28814182460194654,
+      "grad_norm": 414.203125,
+      "learning_rate": 4.541219138397283e-06,
+      "loss": 40.4702,
+      "step": 71320
+    },
+    {
+      "epoch": 0.2881822258673142,
+      "grad_norm": 591.1458129882812,
+      "learning_rate": 4.539814739222182e-06,
+      "loss": 38.5632,
+      "step": 71330
+    },
+    {
+      "epoch": 0.2882226271326818,
+      "grad_norm": 525.4293212890625,
+      "learning_rate": 4.538410376661912e-06,
+      "loss": 35.0831,
+      "step": 71340
+    },
+    {
+      "epoch": 0.2882630283980494,
+      "grad_norm": 491.466796875,
+      "learning_rate": 4.537006050828209e-06,
+      "loss": 49.4348,
+      "step": 71350
+    },
+    {
+      "epoch": 0.28830342966341704,
+      "grad_norm": 768.8724975585938,
+      "learning_rate": 4.535601761832811e-06,
+      "loss": 46.8842,
+      "step": 71360
+    },
+    {
+      "epoch": 0.2883438309287847,
+      "grad_norm": 522.435791015625,
+      "learning_rate": 4.534197509787448e-06,
+      "loss": 40.2206,
+      "step": 71370
+    },
+    {
+      "epoch": 0.2883842321941523,
+      "grad_norm": 519.8639526367188,
+      "learning_rate": 4.5327932948038525e-06,
+      "loss": 34.5767,
+      "step": 71380
+    },
+    {
+      "epoch": 0.28842463345951996,
+      "grad_norm": 626.932861328125,
+      "learning_rate": 4.5313891169937495e-06,
+      "loss": 42.644,
+      "step": 71390
+    },
+    {
+      "epoch": 0.2884650347248876,
+      "grad_norm": 629.30517578125,
+      "learning_rate": 4.529984976468864e-06,
+      "loss": 46.1373,
+      "step": 71400
+    },
+    {
+      "epoch": 0.2885054359902552,
+      "grad_norm": 1649.0540771484375,
+      "learning_rate": 4.528580873340916e-06,
+      "loss": 65.9521,
+      "step": 71410
+    },
+    {
+      "epoch": 0.28854583725562283,
+      "grad_norm": 472.5841369628906,
+      "learning_rate": 4.5271768077216245e-06,
+      "loss": 45.2228,
+      "step": 71420
+    },
+    {
+      "epoch": 0.28858623852099047,
+      "grad_norm": 546.1096801757812,
+      "learning_rate": 4.525772779722705e-06,
+      "loss": 43.5789,
+      "step": 71430
+    },
+    {
+      "epoch": 0.2886266397863581,
+      "grad_norm": 660.35302734375,
+      "learning_rate": 4.524368789455872e-06,
+      "loss": 36.7619,
+      "step": 71440
+    },
+    {
+      "epoch": 0.28866704105172575,
+      "grad_norm": 575.0551147460938,
+      "learning_rate": 4.5229648370328276e-06,
+      "loss": 62.3495,
+      "step": 71450
+    },
+    {
+      "epoch": 0.2887074423170934,
+      "grad_norm": 673.6522827148438,
+      "learning_rate": 4.521560922565282e-06,
+      "loss": 35.792,
+      "step": 71460
+    },
+    {
+      "epoch": 0.28874784358246103,
+      "grad_norm": 661.4155883789062,
+      "learning_rate": 4.52015704616494e-06,
+      "loss": 36.4599,
+      "step": 71470
+    },
+    {
+      "epoch": 0.2887882448478286,
+      "grad_norm": 745.6897583007812,
+      "learning_rate": 4.518753207943498e-06,
+      "loss": 47.3433,
+      "step": 71480
+    },
+    {
+      "epoch": 0.28882864611319625,
+      "grad_norm": 543.9266967773438,
+      "learning_rate": 4.517349408012656e-06,
+      "loss": 42.4632,
+      "step": 71490
+    },
+    {
+      "epoch": 0.2888690473785639,
+      "grad_norm": 1051.2540283203125,
+      "learning_rate": 4.515945646484105e-06,
+      "loss": 37.7069,
+      "step": 71500
+    },
+    {
+      "epoch": 0.28890944864393153,
+      "grad_norm": 561.8253173828125,
+      "learning_rate": 4.514541923469538e-06,
+      "loss": 43.3329,
+      "step": 71510
+    },
+    {
+      "epoch": 0.2889498499092992,
+      "grad_norm": 700.4878540039062,
+      "learning_rate": 4.513138239080641e-06,
+      "loss": 37.6151,
+      "step": 71520
+    },
+    {
+      "epoch": 0.2889902511746668,
+      "grad_norm": 837.8364868164062,
+      "learning_rate": 4.511734593429104e-06,
+      "loss": 57.4711,
+      "step": 71530
+    },
+    {
+      "epoch": 0.2890306524400344,
+      "grad_norm": 277.2738342285156,
+      "learning_rate": 4.510330986626602e-06,
+      "loss": 30.9063,
+      "step": 71540
+    },
+    {
+      "epoch": 0.28907105370540204,
+      "grad_norm": 717.367431640625,
+      "learning_rate": 4.5089274187848144e-06,
+      "loss": 29.4986,
+      "step": 71550
+    },
+    {
+      "epoch": 0.2891114549707697,
+      "grad_norm": 545.1079711914062,
+      "learning_rate": 4.507523890015421e-06,
+      "loss": 42.2378,
+      "step": 71560
+    },
+    {
+      "epoch": 0.2891518562361373,
+      "grad_norm": 694.6339721679688,
+      "learning_rate": 4.5061204004300905e-06,
+      "loss": 57.6911,
+      "step": 71570
+    },
+    {
+      "epoch": 0.28919225750150496,
+      "grad_norm": 1314.3990478515625,
+      "learning_rate": 4.504716950140492e-06,
+      "loss": 41.1658,
+      "step": 71580
+    },
+    {
+      "epoch": 0.2892326587668726,
+      "grad_norm": 1044.653076171875,
+      "learning_rate": 4.503313539258294e-06,
+      "loss": 50.9684,
+      "step": 71590
+    },
+    {
+      "epoch": 0.28927306003224024,
+      "grad_norm": 604.3131713867188,
+      "learning_rate": 4.501910167895158e-06,
+      "loss": 42.9722,
+      "step": 71600
+    },
+    {
+      "epoch": 0.2893134612976078,
+      "grad_norm": 585.9352416992188,
+      "learning_rate": 4.500506836162746e-06,
+      "loss": 27.7333,
+      "step": 71610
+    },
+    {
+      "epoch": 0.28935386256297546,
+      "grad_norm": 377.50897216796875,
+      "learning_rate": 4.499103544172711e-06,
+      "loss": 41.019,
+      "step": 71620
+    },
+    {
+      "epoch": 0.2893942638283431,
+      "grad_norm": 694.3080444335938,
+      "learning_rate": 4.497700292036708e-06,
+      "loss": 52.2126,
+      "step": 71630
+    },
+    {
+      "epoch": 0.28943466509371074,
+      "grad_norm": 3354.697265625,
+      "learning_rate": 4.4962970798663865e-06,
+      "loss": 63.4105,
+      "step": 71640
+    },
+    {
+      "epoch": 0.2894750663590784,
+      "grad_norm": 435.9419250488281,
+      "learning_rate": 4.494893907773394e-06,
+      "loss": 53.5595,
+      "step": 71650
+    },
+    {
+      "epoch": 0.289515467624446,
+      "grad_norm": 662.7257080078125,
+      "learning_rate": 4.493490775869377e-06,
+      "loss": 32.0002,
+      "step": 71660
+    },
+    {
+      "epoch": 0.2895558688898136,
+      "grad_norm": 323.3299560546875,
+      "learning_rate": 4.492087684265975e-06,
+      "loss": 38.7141,
+      "step": 71670
+    },
+    {
+      "epoch": 0.28959627015518125,
+      "grad_norm": 1028.1826171875,
+      "learning_rate": 4.490684633074824e-06,
+      "loss": 41.6504,
+      "step": 71680
+    },
+    {
+      "epoch": 0.2896366714205489,
+      "grad_norm": 346.1958312988281,
+      "learning_rate": 4.489281622407559e-06,
+      "loss": 58.8113,
+      "step": 71690
+    },
+    {
+      "epoch": 0.2896770726859165,
+      "grad_norm": 248.37896728515625,
+      "learning_rate": 4.487878652375813e-06,
+      "loss": 36.2968,
+      "step": 71700
+    },
+    {
+      "epoch": 0.28971747395128417,
+      "grad_norm": 625.3243408203125,
+      "learning_rate": 4.486475723091211e-06,
+      "loss": 57.1991,
+      "step": 71710
+    },
+    {
+      "epoch": 0.2897578752166518,
+      "grad_norm": 796.6832885742188,
+      "learning_rate": 4.485072834665379e-06,
+      "loss": 52.9969,
+      "step": 71720
+    },
+    {
+      "epoch": 0.2897982764820194,
+      "grad_norm": 711.8778076171875,
+      "learning_rate": 4.483669987209938e-06,
+      "loss": 52.1264,
+      "step": 71730
+    },
+    {
+      "epoch": 0.28983867774738703,
+      "grad_norm": 399.5564880371094,
+      "learning_rate": 4.482267180836508e-06,
+      "loss": 49.1718,
+      "step": 71740
+    },
+    {
+      "epoch": 0.28987907901275467,
+      "grad_norm": 529.950439453125,
+      "learning_rate": 4.4808644156567e-06,
+      "loss": 40.0581,
+      "step": 71750
+    },
+    {
+      "epoch": 0.2899194802781223,
+      "grad_norm": 269.11956787109375,
+      "learning_rate": 4.479461691782129e-06,
+      "loss": 25.8915,
+      "step": 71760
+    },
+    {
+      "epoch": 0.28995988154348995,
+      "grad_norm": 417.9612121582031,
+      "learning_rate": 4.478059009324403e-06,
+      "loss": 37.0698,
+      "step": 71770
+    },
+    {
+      "epoch": 0.2900002828088576,
+      "grad_norm": 345.50360107421875,
+      "learning_rate": 4.476656368395126e-06,
+      "loss": 44.7259,
+      "step": 71780
+    },
+    {
+      "epoch": 0.29004068407422523,
+      "grad_norm": 623.320068359375,
+      "learning_rate": 4.4752537691059e-06,
+      "loss": 40.9761,
+      "step": 71790
+    },
+    {
+      "epoch": 0.2900810853395928,
+      "grad_norm": 718.0443115234375,
+      "learning_rate": 4.473851211568323e-06,
+      "loss": 47.5845,
+      "step": 71800
+    },
+    {
+      "epoch": 0.29012148660496045,
+      "grad_norm": 710.3790893554688,
+      "learning_rate": 4.472448695893991e-06,
+      "loss": 37.0356,
+      "step": 71810
+    },
+    {
+      "epoch": 0.2901618878703281,
+      "grad_norm": 1138.6190185546875,
+      "learning_rate": 4.471046222194494e-06,
+      "loss": 44.9854,
+      "step": 71820
+    },
+    {
+      "epoch": 0.29020228913569573,
+      "grad_norm": 415.5585021972656,
+      "learning_rate": 4.469643790581422e-06,
+      "loss": 34.5813,
+      "step": 71830
+    },
+    {
+      "epoch": 0.2902426904010634,
+      "grad_norm": 749.3125,
+      "learning_rate": 4.468241401166359e-06,
+      "loss": 44.3407,
+      "step": 71840
+    },
+    {
+      "epoch": 0.290283091666431,
+      "grad_norm": 1295.281494140625,
+      "learning_rate": 4.466839054060888e-06,
+      "loss": 62.4487,
+      "step": 71850
+    },
+    {
+      "epoch": 0.2903234929317986,
+      "grad_norm": 762.2303466796875,
+      "learning_rate": 4.465436749376586e-06,
+      "loss": 35.7702,
+      "step": 71860
+    },
+    {
+      "epoch": 0.29036389419716624,
+      "grad_norm": 366.7358703613281,
+      "learning_rate": 4.464034487225031e-06,
+      "loss": 63.1885,
+      "step": 71870
+    },
+    {
+      "epoch": 0.2904042954625339,
+      "grad_norm": 332.02301025390625,
+      "learning_rate": 4.462632267717789e-06,
+      "loss": 49.8979,
+      "step": 71880
+    },
+    {
+      "epoch": 0.2904446967279015,
+      "grad_norm": 388.0854187011719,
+      "learning_rate": 4.461230090966433e-06,
+      "loss": 35.094,
+      "step": 71890
+    },
+    {
+      "epoch": 0.29048509799326916,
+      "grad_norm": 580.3858642578125,
+      "learning_rate": 4.4598279570825244e-06,
+      "loss": 41.0283,
+      "step": 71900
+    },
+    {
+      "epoch": 0.2905254992586368,
+      "grad_norm": 396.2524108886719,
+      "learning_rate": 4.458425866177628e-06,
+      "loss": 31.7255,
+      "step": 71910
+    },
+    {
+      "epoch": 0.29056590052400444,
+      "grad_norm": 331.2504577636719,
+      "learning_rate": 4.457023818363299e-06,
+      "loss": 53.3679,
+      "step": 71920
+    },
+    {
+      "epoch": 0.290606301789372,
+      "grad_norm": 744.08544921875,
+      "learning_rate": 4.455621813751093e-06,
+      "loss": 41.8191,
+      "step": 71930
+    },
+    {
+      "epoch": 0.29064670305473966,
+      "grad_norm": 645.9574584960938,
+      "learning_rate": 4.45421985245256e-06,
+      "loss": 42.5509,
+      "step": 71940
+    },
+    {
+      "epoch": 0.2906871043201073,
+      "grad_norm": 233.99517822265625,
+      "learning_rate": 4.452817934579249e-06,
+      "loss": 59.6218,
+      "step": 71950
+    },
+    {
+      "epoch": 0.29072750558547494,
+      "grad_norm": 557.802490234375,
+      "learning_rate": 4.451416060242707e-06,
+      "loss": 42.233,
+      "step": 71960
+    },
+    {
+      "epoch": 0.2907679068508426,
+      "grad_norm": 1119.917236328125,
+      "learning_rate": 4.450014229554468e-06,
+      "loss": 60.5082,
+      "step": 71970
+    },
+    {
+      "epoch": 0.2908083081162102,
+      "grad_norm": 366.4770812988281,
+      "learning_rate": 4.448612442626073e-06,
+      "loss": 34.5282,
+      "step": 71980
+    },
+    {
+      "epoch": 0.2908487093815778,
+      "grad_norm": 1179.2496337890625,
+      "learning_rate": 4.447210699569055e-06,
+      "loss": 61.8394,
+      "step": 71990
+    },
+    {
+      "epoch": 0.29088911064694545,
+      "grad_norm": 833.4341430664062,
+      "learning_rate": 4.445809000494945e-06,
+      "loss": 56.3729,
+      "step": 72000
+    },
+    {
+      "epoch": 0.2909295119123131,
+      "grad_norm": 417.32440185546875,
+      "learning_rate": 4.4444073455152705e-06,
+      "loss": 36.397,
+      "step": 72010
+    },
+    {
+      "epoch": 0.29096991317768073,
+      "grad_norm": 324.6310119628906,
+      "learning_rate": 4.443005734741553e-06,
+      "loss": 46.3118,
+      "step": 72020
+    },
+    {
+      "epoch": 0.29101031444304837,
+      "grad_norm": 386.0984191894531,
+      "learning_rate": 4.441604168285313e-06,
+      "loss": 27.7411,
+      "step": 72030
+    },
+    {
+      "epoch": 0.291050715708416,
+      "grad_norm": 444.1304626464844,
+      "learning_rate": 4.440202646258067e-06,
+      "loss": 47.9088,
+      "step": 72040
+    },
+    {
+      "epoch": 0.2910911169737836,
+      "grad_norm": 783.587890625,
+      "learning_rate": 4.4388011687713274e-06,
+      "loss": 57.362,
+      "step": 72050
+    },
+    {
+      "epoch": 0.29113151823915123,
+      "grad_norm": 1120.068115234375,
+      "learning_rate": 4.437399735936603e-06,
+      "loss": 68.9344,
+      "step": 72060
+    },
+    {
+      "epoch": 0.2911719195045189,
+      "grad_norm": 496.18292236328125,
+      "learning_rate": 4.435998347865399e-06,
+      "loss": 31.082,
+      "step": 72070
+    },
+    {
+      "epoch": 0.2912123207698865,
+      "grad_norm": 427.5905456542969,
+      "learning_rate": 4.4345970046692174e-06,
+      "loss": 39.1392,
+      "step": 72080
+    },
+    {
+      "epoch": 0.29125272203525415,
+      "grad_norm": 828.7031860351562,
+      "learning_rate": 4.433195706459558e-06,
+      "loss": 54.4549,
+      "step": 72090
+    },
+    {
+      "epoch": 0.2912931233006218,
+      "grad_norm": 974.0640869140625,
+      "learning_rate": 4.431794453347915e-06,
+      "loss": 45.0464,
+      "step": 72100
+    },
+    {
+      "epoch": 0.29133352456598943,
+      "grad_norm": 227.1929931640625,
+      "learning_rate": 4.430393245445781e-06,
+      "loss": 38.7168,
+      "step": 72110
+    },
+    {
+      "epoch": 0.291373925831357,
+      "grad_norm": 416.6019287109375,
+      "learning_rate": 4.42899208286464e-06,
+      "loss": 44.9181,
+      "step": 72120
+    },
+    {
+      "epoch": 0.29141432709672466,
+      "grad_norm": 688.7039184570312,
+      "learning_rate": 4.427590965715981e-06,
+      "loss": 43.0894,
+      "step": 72130
+    },
+    {
+      "epoch": 0.2914547283620923,
+      "grad_norm": 250.0460205078125,
+      "learning_rate": 4.426189894111281e-06,
+      "loss": 29.5613,
+      "step": 72140
+    },
+    {
+      "epoch": 0.29149512962745994,
+      "grad_norm": 1028.0450439453125,
+      "learning_rate": 4.4247888681620165e-06,
+      "loss": 59.4425,
+      "step": 72150
+    },
+    {
+      "epoch": 0.2915355308928276,
+      "grad_norm": 1062.1776123046875,
+      "learning_rate": 4.423387887979663e-06,
+      "loss": 45.5027,
+      "step": 72160
+    },
+    {
+      "epoch": 0.2915759321581952,
+      "grad_norm": 967.5091552734375,
+      "learning_rate": 4.421986953675687e-06,
+      "loss": 57.5422,
+      "step": 72170
+    },
+    {
+      "epoch": 0.2916163334235628,
+      "grad_norm": 281.9029541015625,
+      "learning_rate": 4.420586065361558e-06,
+      "loss": 33.8364,
+      "step": 72180
+    },
+    {
+      "epoch": 0.29165673468893044,
+      "grad_norm": 475.996826171875,
+      "learning_rate": 4.419185223148737e-06,
+      "loss": 56.7115,
+      "step": 72190
+    },
+    {
+      "epoch": 0.2916971359542981,
+      "grad_norm": 900.3701782226562,
+      "learning_rate": 4.417784427148681e-06,
+      "loss": 35.606,
+      "step": 72200
+    },
+    {
+      "epoch": 0.2917375372196657,
+      "grad_norm": 495.081787109375,
+      "learning_rate": 4.4163836774728466e-06,
+      "loss": 52.596,
+      "step": 72210
+    },
+    {
+      "epoch": 0.29177793848503336,
+      "grad_norm": 385.602294921875,
+      "learning_rate": 4.414982974232686e-06,
+      "loss": 33.3124,
+      "step": 72220
+    },
+    {
+      "epoch": 0.291818339750401,
+      "grad_norm": 741.7224731445312,
+      "learning_rate": 4.413582317539644e-06,
+      "loss": 35.2805,
+      "step": 72230
+    },
+    {
+      "epoch": 0.29185874101576864,
+      "grad_norm": 842.2901611328125,
+      "learning_rate": 4.412181707505167e-06,
+      "loss": 56.8002,
+      "step": 72240
+    },
+    {
+      "epoch": 0.2918991422811362,
+      "grad_norm": 537.3816528320312,
+      "learning_rate": 4.410781144240692e-06,
+      "loss": 48.8103,
+      "step": 72250
+    },
+    {
+      "epoch": 0.29193954354650387,
+      "grad_norm": 677.678466796875,
+      "learning_rate": 4.409380627857658e-06,
+      "loss": 70.5054,
+      "step": 72260
+    },
+    {
+      "epoch": 0.2919799448118715,
+      "grad_norm": 445.90057373046875,
+      "learning_rate": 4.4079801584674955e-06,
+      "loss": 39.0066,
+      "step": 72270
+    },
+    {
+      "epoch": 0.29202034607723915,
+      "grad_norm": 579.4309692382812,
+      "learning_rate": 4.406579736181636e-06,
+      "loss": 47.5335,
+      "step": 72280
+    },
+    {
+      "epoch": 0.2920607473426068,
+      "grad_norm": 373.58270263671875,
+      "learning_rate": 4.405179361111503e-06,
+      "loss": 46.5012,
+      "step": 72290
+    },
+    {
+      "epoch": 0.2921011486079744,
+      "grad_norm": 292.4507141113281,
+      "learning_rate": 4.403779033368521e-06,
+      "loss": 39.3859,
+      "step": 72300
+    },
+    {
+      "epoch": 0.292141549873342,
+      "grad_norm": 446.4667663574219,
+      "learning_rate": 4.402378753064102e-06,
+      "loss": 59.9139,
+      "step": 72310
+    },
+    {
+      "epoch": 0.29218195113870965,
+      "grad_norm": 756.2067260742188,
+      "learning_rate": 4.400978520309663e-06,
+      "loss": 40.9019,
+      "step": 72320
+    },
+    {
+      "epoch": 0.2922223524040773,
+      "grad_norm": 707.1049194335938,
+      "learning_rate": 4.399578335216615e-06,
+      "loss": 59.0028,
+      "step": 72330
+    },
+    {
+      "epoch": 0.29226275366944493,
+      "grad_norm": 583.9354248046875,
+      "learning_rate": 4.3981781978963625e-06,
+      "loss": 35.8286,
+      "step": 72340
+    },
+    {
+      "epoch": 0.29230315493481257,
+      "grad_norm": 637.59130859375,
+      "learning_rate": 4.39677810846031e-06,
+      "loss": 63.2968,
+      "step": 72350
+    },
+    {
+      "epoch": 0.2923435562001802,
+      "grad_norm": 811.93408203125,
+      "learning_rate": 4.395378067019854e-06,
+      "loss": 47.6136,
+      "step": 72360
+    },
+    {
+      "epoch": 0.2923839574655478,
+      "grad_norm": 1073.439697265625,
+      "learning_rate": 4.39397807368639e-06,
+      "loss": 57.2526,
+      "step": 72370
+    },
+    {
+      "epoch": 0.29242435873091543,
+      "grad_norm": 374.0229797363281,
+      "learning_rate": 4.39257812857131e-06,
+      "loss": 41.0583,
+      "step": 72380
+    },
+    {
+      "epoch": 0.2924647599962831,
+      "grad_norm": 757.6024780273438,
+      "learning_rate": 4.391178231786003e-06,
+      "loss": 38.2484,
+      "step": 72390
+    },
+    {
+      "epoch": 0.2925051612616507,
+      "grad_norm": 797.3472290039062,
+      "learning_rate": 4.389778383441847e-06,
+      "loss": 38.5529,
+      "step": 72400
+    },
+    {
+      "epoch": 0.29254556252701835,
+      "grad_norm": 927.91796875,
+      "learning_rate": 4.388378583650225e-06,
+      "loss": 38.3963,
+      "step": 72410
+    },
+    {
+      "epoch": 0.292585963792386,
+      "grad_norm": 467.8027648925781,
+      "learning_rate": 4.386978832522512e-06,
+      "loss": 37.6053,
+      "step": 72420
+    },
+    {
+      "epoch": 0.29262636505775363,
+      "grad_norm": 491.3323059082031,
+      "learning_rate": 4.38557913017008e-06,
+      "loss": 55.3208,
+      "step": 72430
+    },
+    {
+      "epoch": 0.2926667663231212,
+      "grad_norm": 931.24609375,
+      "learning_rate": 4.384179476704297e-06,
+      "loss": 49.4638,
+      "step": 72440
+    },
+    {
+      "epoch": 0.29270716758848886,
+      "grad_norm": 854.8884887695312,
+      "learning_rate": 4.382779872236527e-06,
+      "loss": 44.3615,
+      "step": 72450
+    },
+    {
+      "epoch": 0.2927475688538565,
+      "grad_norm": 498.6365966796875,
+      "learning_rate": 4.3813803168781295e-06,
+      "loss": 39.2645,
+      "step": 72460
+    },
+    {
+      "epoch": 0.29278797011922414,
+      "grad_norm": 658.3452758789062,
+      "learning_rate": 4.379980810740463e-06,
+      "loss": 41.7759,
+      "step": 72470
+    },
+    {
+      "epoch": 0.2928283713845918,
+      "grad_norm": 596.740478515625,
+      "learning_rate": 4.378581353934876e-06,
+      "loss": 49.1443,
+      "step": 72480
+    },
+    {
+      "epoch": 0.2928687726499594,
+      "grad_norm": 496.052734375,
+      "learning_rate": 4.3771819465727185e-06,
+      "loss": 40.2331,
+      "step": 72490
+    },
+    {
+      "epoch": 0.292909173915327,
+      "grad_norm": 340.43646240234375,
+      "learning_rate": 4.3757825887653345e-06,
+      "loss": 38.848,
+      "step": 72500
+    },
+    {
+      "epoch": 0.29294957518069464,
+      "grad_norm": 613.5763549804688,
+      "learning_rate": 4.374383280624066e-06,
+      "loss": 36.6635,
+      "step": 72510
+    },
+    {
+      "epoch": 0.2929899764460623,
+      "grad_norm": 450.40399169921875,
+      "learning_rate": 4.372984022260249e-06,
+      "loss": 37.3646,
+      "step": 72520
+    },
+    {
+      "epoch": 0.2930303777114299,
+      "grad_norm": 751.0297241210938,
+      "learning_rate": 4.371584813785216e-06,
+      "loss": 30.7279,
+      "step": 72530
+    },
+    {
+      "epoch": 0.29307077897679756,
+      "grad_norm": 347.6400146484375,
+      "learning_rate": 4.370185655310295e-06,
+      "loss": 41.535,
+      "step": 72540
+    },
+    {
+      "epoch": 0.2931111802421652,
+      "grad_norm": 928.6194458007812,
+      "learning_rate": 4.368786546946811e-06,
+      "loss": 41.9351,
+      "step": 72550
+    },
+    {
+      "epoch": 0.29315158150753284,
+      "grad_norm": 542.6160278320312,
+      "learning_rate": 4.367387488806086e-06,
+      "loss": 49.6833,
+      "step": 72560
+    },
+    {
+      "epoch": 0.29319198277290043,
+      "grad_norm": 351.0897216796875,
+      "learning_rate": 4.365988480999434e-06,
+      "loss": 32.4432,
+      "step": 72570
+    },
+    {
+      "epoch": 0.29323238403826807,
+      "grad_norm": 381.7917785644531,
+      "learning_rate": 4.364589523638168e-06,
+      "loss": 34.2274,
+      "step": 72580
+    },
+    {
+      "epoch": 0.2932727853036357,
+      "grad_norm": 1160.76611328125,
+      "learning_rate": 4.363190616833598e-06,
+      "loss": 41.1181,
+      "step": 72590
+    },
+    {
+      "epoch": 0.29331318656900335,
+      "grad_norm": 766.3687744140625,
+      "learning_rate": 4.361791760697027e-06,
+      "loss": 42.0144,
+      "step": 72600
+    },
+    {
+      "epoch": 0.293353587834371,
+      "grad_norm": 619.919677734375,
+      "learning_rate": 4.360392955339758e-06,
+      "loss": 31.6507,
+      "step": 72610
+    },
+    {
+      "epoch": 0.29339398909973863,
+      "grad_norm": 236.89402770996094,
+      "learning_rate": 4.358994200873085e-06,
+      "loss": 32.084,
+      "step": 72620
+    },
+    {
+      "epoch": 0.2934343903651062,
+      "grad_norm": 587.992919921875,
+      "learning_rate": 4.357595497408303e-06,
+      "loss": 40.2523,
+      "step": 72630
+    },
+    {
+      "epoch": 0.29347479163047385,
+      "grad_norm": 640.2550048828125,
+      "learning_rate": 4.356196845056699e-06,
+      "loss": 61.5953,
+      "step": 72640
+    },
+    {
+      "epoch": 0.2935151928958415,
+      "grad_norm": 653.5460205078125,
+      "learning_rate": 4.3547982439295576e-06,
+      "loss": 60.019,
+      "step": 72650
+    },
+    {
+      "epoch": 0.29355559416120913,
+      "grad_norm": 462.1523132324219,
+      "learning_rate": 4.353399694138158e-06,
+      "loss": 41.3274,
+      "step": 72660
+    },
+    {
+      "epoch": 0.2935959954265768,
+      "grad_norm": 541.9874877929688,
+      "learning_rate": 4.352001195793778e-06,
+      "loss": 48.6624,
+      "step": 72670
+    },
+    {
+      "epoch": 0.2936363966919444,
+      "grad_norm": 190.7097625732422,
+      "learning_rate": 4.350602749007688e-06,
+      "loss": 34.3381,
+      "step": 72680
+    },
+    {
+      "epoch": 0.293676797957312,
+      "grad_norm": 630.536376953125,
+      "learning_rate": 4.349204353891158e-06,
+      "loss": 39.0707,
+      "step": 72690
+    },
+    {
+      "epoch": 0.29371719922267964,
+      "grad_norm": 421.9498596191406,
+      "learning_rate": 4.347806010555448e-06,
+      "loss": 34.1856,
+      "step": 72700
+    },
+    {
+      "epoch": 0.2937576004880473,
+      "grad_norm": 679.4033203125,
+      "learning_rate": 4.346407719111823e-06,
+      "loss": 49.3176,
+      "step": 72710
+    },
+    {
+      "epoch": 0.2937980017534149,
+      "grad_norm": 749.3023681640625,
+      "learning_rate": 4.3450094796715354e-06,
+      "loss": 66.0755,
+      "step": 72720
+    },
+    {
+      "epoch": 0.29383840301878256,
+      "grad_norm": 614.3016357421875,
+      "learning_rate": 4.343611292345839e-06,
+      "loss": 49.1222,
+      "step": 72730
+    },
+    {
+      "epoch": 0.2938788042841502,
+      "grad_norm": 644.041748046875,
+      "learning_rate": 4.342213157245979e-06,
+      "loss": 40.4218,
+      "step": 72740
+    },
+    {
+      "epoch": 0.29391920554951784,
+      "grad_norm": 681.5894775390625,
+      "learning_rate": 4.340815074483199e-06,
+      "loss": 56.7761,
+      "step": 72750
+    },
+    {
+      "epoch": 0.2939596068148854,
+      "grad_norm": 474.3656311035156,
+      "learning_rate": 4.339417044168738e-06,
+      "loss": 49.5246,
+      "step": 72760
+    },
+    {
+      "epoch": 0.29400000808025306,
+      "grad_norm": 701.384033203125,
+      "learning_rate": 4.338019066413832e-06,
+      "loss": 47.6203,
+      "step": 72770
+    },
+    {
+      "epoch": 0.2940404093456207,
+      "grad_norm": 322.1119384765625,
+      "learning_rate": 4.33662114132971e-06,
+      "loss": 37.8673,
+      "step": 72780
+    },
+    {
+      "epoch": 0.29408081061098834,
+      "grad_norm": 406.5722351074219,
+      "learning_rate": 4.335223269027599e-06,
+      "loss": 44.9244,
+      "step": 72790
+    },
+    {
+      "epoch": 0.294121211876356,
+      "grad_norm": 520.870361328125,
+      "learning_rate": 4.333825449618721e-06,
+      "loss": 48.5541,
+      "step": 72800
+    },
+    {
+      "epoch": 0.2941616131417236,
+      "grad_norm": 768.7017822265625,
+      "learning_rate": 4.332427683214295e-06,
+      "loss": 41.3215,
+      "step": 72810
+    },
+    {
+      "epoch": 0.2942020144070912,
+      "grad_norm": 882.6221313476562,
+      "learning_rate": 4.331029969925538e-06,
+      "loss": 39.3176,
+      "step": 72820
+    },
+    {
+      "epoch": 0.29424241567245885,
+      "grad_norm": 4053.163818359375,
+      "learning_rate": 4.329632309863652e-06,
+      "loss": 61.8202,
+      "step": 72830
+    },
+    {
+      "epoch": 0.2942828169378265,
+      "grad_norm": 517.92724609375,
+      "learning_rate": 4.328234703139847e-06,
+      "loss": 43.2393,
+      "step": 72840
+    },
+    {
+      "epoch": 0.2943232182031941,
+      "grad_norm": 284.5147399902344,
+      "learning_rate": 4.326837149865325e-06,
+      "loss": 27.8085,
+      "step": 72850
+    },
+    {
+      "epoch": 0.29436361946856177,
+      "grad_norm": 856.0078125,
+      "learning_rate": 4.325439650151281e-06,
+      "loss": 57.4546,
+      "step": 72860
+    },
+    {
+      "epoch": 0.2944040207339294,
+      "grad_norm": 613.6627807617188,
+      "learning_rate": 4.324042204108908e-06,
+      "loss": 44.772,
+      "step": 72870
+    },
+    {
+      "epoch": 0.29444442199929705,
+      "grad_norm": 905.0805053710938,
+      "learning_rate": 4.322644811849395e-06,
+      "loss": 47.959,
+      "step": 72880
+    },
+    {
+      "epoch": 0.29448482326466463,
+      "grad_norm": 596.2250366210938,
+      "learning_rate": 4.321247473483924e-06,
+      "loss": 29.8382,
+      "step": 72890
+    },
+    {
+      "epoch": 0.29452522453003227,
+      "grad_norm": 327.65338134765625,
+      "learning_rate": 4.319850189123681e-06,
+      "loss": 47.0369,
+      "step": 72900
+    },
+    {
+      "epoch": 0.2945656257953999,
+      "grad_norm": 716.1499633789062,
+      "learning_rate": 4.3184529588798335e-06,
+      "loss": 46.2548,
+      "step": 72910
+    },
+    {
+      "epoch": 0.29460602706076755,
+      "grad_norm": 270.2619934082031,
+      "learning_rate": 4.3170557828635565e-06,
+      "loss": 53.3165,
+      "step": 72920
+    },
+    {
+      "epoch": 0.2946464283261352,
+      "grad_norm": 517.9827880859375,
+      "learning_rate": 4.315658661186016e-06,
+      "loss": 37.7342,
+      "step": 72930
+    },
+    {
+      "epoch": 0.29468682959150283,
+      "grad_norm": 508.7879638671875,
+      "learning_rate": 4.314261593958376e-06,
+      "loss": 30.2865,
+      "step": 72940
+    },
+    {
+      "epoch": 0.2947272308568704,
+      "grad_norm": 840.9309692382812,
+      "learning_rate": 4.3128645812917935e-06,
+      "loss": 43.7234,
+      "step": 72950
+    },
+    {
+      "epoch": 0.29476763212223805,
+      "grad_norm": 649.1232299804688,
+      "learning_rate": 4.311467623297423e-06,
+      "loss": 43.8122,
+      "step": 72960
+    },
+    {
+      "epoch": 0.2948080333876057,
+      "grad_norm": 886.0907592773438,
+      "learning_rate": 4.310070720086414e-06,
+      "loss": 51.4335,
+      "step": 72970
+    },
+    {
+      "epoch": 0.29484843465297333,
+      "grad_norm": 1010.0867919921875,
+      "learning_rate": 4.30867387176991e-06,
+      "loss": 54.0462,
+      "step": 72980
+    },
+    {
+      "epoch": 0.294888835918341,
+      "grad_norm": 563.8475341796875,
+      "learning_rate": 4.307277078459057e-06,
+      "loss": 33.6128,
+      "step": 72990
+    },
+    {
+      "epoch": 0.2949292371837086,
+      "grad_norm": 507.4144287109375,
+      "learning_rate": 4.305880340264985e-06,
+      "loss": 41.6365,
+      "step": 73000
+    },
+    {
+      "epoch": 0.2949696384490762,
+      "grad_norm": 901.0919189453125,
+      "learning_rate": 4.3044836572988285e-06,
+      "loss": 53.9174,
+      "step": 73010
+    },
+    {
+      "epoch": 0.29501003971444384,
+      "grad_norm": 384.7436828613281,
+      "learning_rate": 4.3030870296717155e-06,
+      "loss": 52.3031,
+      "step": 73020
+    },
+    {
+      "epoch": 0.2950504409798115,
+      "grad_norm": 578.6005249023438,
+      "learning_rate": 4.301690457494769e-06,
+      "loss": 40.9771,
+      "step": 73030
+    },
+    {
+      "epoch": 0.2950908422451791,
+      "grad_norm": 544.58642578125,
+      "learning_rate": 4.300293940879108e-06,
+      "loss": 42.7261,
+      "step": 73040
+    },
+    {
+      "epoch": 0.29513124351054676,
+      "grad_norm": 279.6160583496094,
+      "learning_rate": 4.298897479935847e-06,
+      "loss": 66.8447,
+      "step": 73050
+    },
+    {
+      "epoch": 0.2951716447759144,
+      "grad_norm": 217.38037109375,
+      "learning_rate": 4.297501074776097e-06,
+      "loss": 33.5354,
+      "step": 73060
+    },
+    {
+      "epoch": 0.29521204604128204,
+      "grad_norm": 0.0,
+      "learning_rate": 4.296104725510961e-06,
+      "loss": 49.2657,
+      "step": 73070
+    },
+    {
+      "epoch": 0.2952524473066496,
+      "grad_norm": 860.3460693359375,
+      "learning_rate": 4.294708432251544e-06,
+      "loss": 50.6563,
+      "step": 73080
+    },
+    {
+      "epoch": 0.29529284857201726,
+      "grad_norm": 500.3200988769531,
+      "learning_rate": 4.293312195108938e-06,
+      "loss": 35.9054,
+      "step": 73090
+    },
+    {
+      "epoch": 0.2953332498373849,
+      "grad_norm": 642.476318359375,
+      "learning_rate": 4.291916014194238e-06,
+      "loss": 42.9805,
+      "step": 73100
+    },
+    {
+      "epoch": 0.29537365110275254,
+      "grad_norm": 453.5120544433594,
+      "learning_rate": 4.290519889618531e-06,
+      "loss": 44.4742,
+      "step": 73110
+    },
+    {
+      "epoch": 0.2954140523681202,
+      "grad_norm": 604.7741088867188,
+      "learning_rate": 4.2891238214928995e-06,
+      "loss": 49.2943,
+      "step": 73120
+    },
+    {
+      "epoch": 0.2954544536334878,
+      "grad_norm": 632.0167846679688,
+      "learning_rate": 4.287727809928423e-06,
+      "loss": 48.4525,
+      "step": 73130
+    },
+    {
+      "epoch": 0.2954948548988554,
+      "grad_norm": 1167.5364990234375,
+      "learning_rate": 4.286331855036177e-06,
+      "loss": 40.5812,
+      "step": 73140
+    },
+    {
+      "epoch": 0.29553525616422305,
+      "grad_norm": 569.1380004882812,
+      "learning_rate": 4.284935956927229e-06,
+      "loss": 38.5044,
+      "step": 73150
+    },
+    {
+      "epoch": 0.2955756574295907,
+      "grad_norm": 628.6519775390625,
+      "learning_rate": 4.283540115712647e-06,
+      "loss": 46.5134,
+      "step": 73160
+    },
+    {
+      "epoch": 0.29561605869495833,
+      "grad_norm": 384.6105041503906,
+      "learning_rate": 4.282144331503488e-06,
+      "loss": 59.5178,
+      "step": 73170
+    },
+    {
+      "epoch": 0.29565645996032597,
+      "grad_norm": 526.2188720703125,
+      "learning_rate": 4.280748604410811e-06,
+      "loss": 81.0941,
+      "step": 73180
+    },
+    {
+      "epoch": 0.2956968612256936,
+      "grad_norm": 550.6795654296875,
+      "learning_rate": 4.279352934545666e-06,
+      "loss": 34.5867,
+      "step": 73190
+    },
+    {
+      "epoch": 0.2957372624910612,
+      "grad_norm": 1194.502197265625,
+      "learning_rate": 4.277957322019101e-06,
+      "loss": 61.5651,
+      "step": 73200
+    },
+    {
+      "epoch": 0.29577766375642883,
+      "grad_norm": 566.8056640625,
+      "learning_rate": 4.276561766942158e-06,
+      "loss": 40.1345,
+      "step": 73210
+    },
+    {
+      "epoch": 0.2958180650217965,
+      "grad_norm": 623.376953125,
+      "learning_rate": 4.275166269425874e-06,
+      "loss": 34.6147,
+      "step": 73220
+    },
+    {
+      "epoch": 0.2958584662871641,
+      "grad_norm": 596.8927001953125,
+      "learning_rate": 4.273770829581285e-06,
+      "loss": 62.5187,
+      "step": 73230
+    },
+    {
+      "epoch": 0.29589886755253175,
+      "grad_norm": 770.3164672851562,
+      "learning_rate": 4.272375447519418e-06,
+      "loss": 54.9159,
+      "step": 73240
+    },
+    {
+      "epoch": 0.2959392688178994,
+      "grad_norm": 425.0506286621094,
+      "learning_rate": 4.270980123351299e-06,
+      "loss": 54.7439,
+      "step": 73250
+    },
+    {
+      "epoch": 0.29597967008326703,
+      "grad_norm": 690.937255859375,
+      "learning_rate": 4.269584857187942e-06,
+      "loss": 38.0473,
+      "step": 73260
+    },
+    {
+      "epoch": 0.2960200713486346,
+      "grad_norm": 606.042236328125,
+      "learning_rate": 4.268189649140369e-06,
+      "loss": 40.5556,
+      "step": 73270
+    },
+    {
+      "epoch": 0.29606047261400226,
+      "grad_norm": 1045.22119140625,
+      "learning_rate": 4.266794499319585e-06,
+      "loss": 54.4675,
+      "step": 73280
+    },
+    {
+      "epoch": 0.2961008738793699,
+      "grad_norm": 514.9558715820312,
+      "learning_rate": 4.265399407836598e-06,
+      "loss": 47.2978,
+      "step": 73290
+    },
+    {
+      "epoch": 0.29614127514473754,
+      "grad_norm": 976.2498779296875,
+      "learning_rate": 4.26400437480241e-06,
+      "loss": 41.2603,
+      "step": 73300
+    },
+    {
+      "epoch": 0.2961816764101052,
+      "grad_norm": 423.5660705566406,
+      "learning_rate": 4.262609400328015e-06,
+      "loss": 43.1175,
+      "step": 73310
+    },
+    {
+      "epoch": 0.2962220776754728,
+      "grad_norm": 505.2185974121094,
+      "learning_rate": 4.2612144845244044e-06,
+      "loss": 31.8775,
+      "step": 73320
+    },
+    {
+      "epoch": 0.2962624789408404,
+      "grad_norm": 577.039794921875,
+      "learning_rate": 4.259819627502571e-06,
+      "loss": 40.0951,
+      "step": 73330
+    },
+    {
+      "epoch": 0.29630288020620804,
+      "grad_norm": 580.8576049804688,
+      "learning_rate": 4.258424829373491e-06,
+      "loss": 63.3412,
+      "step": 73340
+    },
+    {
+      "epoch": 0.2963432814715757,
+      "grad_norm": 413.0155944824219,
+      "learning_rate": 4.257030090248142e-06,
+      "loss": 31.0133,
+      "step": 73350
+    },
+    {
+      "epoch": 0.2963836827369433,
+      "grad_norm": 473.4555358886719,
+      "learning_rate": 4.2556354102374994e-06,
+      "loss": 45.6765,
+      "step": 73360
+    },
+    {
+      "epoch": 0.29642408400231096,
+      "grad_norm": 777.5567626953125,
+      "learning_rate": 4.254240789452532e-06,
+      "loss": 42.2331,
+      "step": 73370
+    },
+    {
+      "epoch": 0.2964644852676786,
+      "grad_norm": 639.9664306640625,
+      "learning_rate": 4.252846228004203e-06,
+      "loss": 77.8432,
+      "step": 73380
+    },
+    {
+      "epoch": 0.29650488653304624,
+      "grad_norm": 884.723388671875,
+      "learning_rate": 4.25145172600347e-06,
+      "loss": 46.5431,
+      "step": 73390
+    },
+    {
+      "epoch": 0.2965452877984138,
+      "grad_norm": 92.27436828613281,
+      "learning_rate": 4.2500572835612876e-06,
+      "loss": 31.1916,
+      "step": 73400
+    },
+    {
+      "epoch": 0.29658568906378147,
+      "grad_norm": 469.61456298828125,
+      "learning_rate": 4.248662900788605e-06,
+      "loss": 36.7089,
+      "step": 73410
+    },
+    {
+      "epoch": 0.2966260903291491,
+      "grad_norm": 654.217041015625,
+      "learning_rate": 4.247268577796368e-06,
+      "loss": 41.0459,
+      "step": 73420
+    },
+    {
+      "epoch": 0.29666649159451675,
+      "grad_norm": 433.7368469238281,
+      "learning_rate": 4.245874314695516e-06,
+      "loss": 35.8187,
+      "step": 73430
+    },
+    {
+      "epoch": 0.2967068928598844,
+      "grad_norm": 413.3462219238281,
+      "learning_rate": 4.244480111596984e-06,
+      "loss": 37.6828,
+      "step": 73440
+    },
+    {
+      "epoch": 0.296747294125252,
+      "grad_norm": 837.417236328125,
+      "learning_rate": 4.2430859686117e-06,
+      "loss": 44.1146,
+      "step": 73450
+    },
+    {
+      "epoch": 0.2967876953906196,
+      "grad_norm": 599.1912841796875,
+      "learning_rate": 4.241691885850593e-06,
+      "loss": 42.8803,
+      "step": 73460
+    },
+    {
+      "epoch": 0.29682809665598725,
+      "grad_norm": 382.2321472167969,
+      "learning_rate": 4.240297863424582e-06,
+      "loss": 36.6295,
+      "step": 73470
+    },
+    {
+      "epoch": 0.2968684979213549,
+      "grad_norm": 680.3961791992188,
+      "learning_rate": 4.2389039014445846e-06,
+      "loss": 52.8627,
+      "step": 73480
+    },
+    {
+      "epoch": 0.29690889918672253,
+      "grad_norm": 654.2417602539062,
+      "learning_rate": 4.23751000002151e-06,
+      "loss": 37.8739,
+      "step": 73490
+    },
+    {
+      "epoch": 0.29694930045209017,
+      "grad_norm": 235.81236267089844,
+      "learning_rate": 4.2361161592662655e-06,
+      "loss": 28.3701,
+      "step": 73500
+    },
+    {
+      "epoch": 0.2969897017174578,
+      "grad_norm": 1637.2437744140625,
+      "learning_rate": 4.234722379289753e-06,
+      "loss": 49.2392,
+      "step": 73510
+    },
+    {
+      "epoch": 0.2970301029828254,
+      "grad_norm": 315.6921081542969,
+      "learning_rate": 4.233328660202869e-06,
+      "loss": 34.3896,
+      "step": 73520
+    },
+    {
+      "epoch": 0.29707050424819303,
+      "grad_norm": 500.3575134277344,
+      "learning_rate": 4.231935002116504e-06,
+      "loss": 49.8805,
+      "step": 73530
+    },
+    {
+      "epoch": 0.2971109055135607,
+      "grad_norm": 137.28250122070312,
+      "learning_rate": 4.230541405141546e-06,
+      "loss": 36.5432,
+      "step": 73540
+    },
+    {
+      "epoch": 0.2971513067789283,
+      "grad_norm": 558.1651000976562,
+      "learning_rate": 4.229147869388875e-06,
+      "loss": 44.4456,
+      "step": 73550
+    },
+    {
+      "epoch": 0.29719170804429595,
+      "grad_norm": 243.798828125,
+      "learning_rate": 4.227754394969373e-06,
+      "loss": 45.8601,
+      "step": 73560
+    },
+    {
+      "epoch": 0.2972321093096636,
+      "grad_norm": 1039.82177734375,
+      "learning_rate": 4.226360981993909e-06,
+      "loss": 57.3914,
+      "step": 73570
+    },
+    {
+      "epoch": 0.29727251057503123,
+      "grad_norm": 504.4435729980469,
+      "learning_rate": 4.224967630573351e-06,
+      "loss": 51.3584,
+      "step": 73580
+    },
+    {
+      "epoch": 0.2973129118403988,
+      "grad_norm": 357.03570556640625,
+      "learning_rate": 4.2235743408185635e-06,
+      "loss": 55.9675,
+      "step": 73590
+    },
+    {
+      "epoch": 0.29735331310576646,
+      "grad_norm": 398.0602722167969,
+      "learning_rate": 4.222181112840401e-06,
+      "loss": 47.9505,
+      "step": 73600
+    },
+    {
+      "epoch": 0.2973937143711341,
+      "grad_norm": 774.7249145507812,
+      "learning_rate": 4.220787946749717e-06,
+      "loss": 45.4625,
+      "step": 73610
+    },
+    {
+      "epoch": 0.29743411563650174,
+      "grad_norm": 754.9330444335938,
+      "learning_rate": 4.219394842657361e-06,
+      "loss": 42.2232,
+      "step": 73620
+    },
+    {
+      "epoch": 0.2974745169018694,
+      "grad_norm": 583.9215698242188,
+      "learning_rate": 4.218001800674174e-06,
+      "loss": 34.6186,
+      "step": 73630
+    },
+    {
+      "epoch": 0.297514918167237,
+      "grad_norm": 481.25177001953125,
+      "learning_rate": 4.216608820910995e-06,
+      "loss": 42.5184,
+      "step": 73640
+    },
+    {
+      "epoch": 0.2975553194326046,
+      "grad_norm": 1010.7393188476562,
+      "learning_rate": 4.2152159034786554e-06,
+      "loss": 49.1567,
+      "step": 73650
+    },
+    {
+      "epoch": 0.29759572069797224,
+      "grad_norm": 429.19256591796875,
+      "learning_rate": 4.213823048487987e-06,
+      "loss": 45.6626,
+      "step": 73660
+    },
+    {
+      "epoch": 0.2976361219633399,
+      "grad_norm": 602.9405517578125,
+      "learning_rate": 4.212430256049809e-06,
+      "loss": 40.8663,
+      "step": 73670
+    },
+    {
+      "epoch": 0.2976765232287075,
+      "grad_norm": 341.2919921875,
+      "learning_rate": 4.2110375262749435e-06,
+      "loss": 42.1535,
+      "step": 73680
+    },
+    {
+      "epoch": 0.29771692449407516,
+      "grad_norm": 795.6687622070312,
+      "learning_rate": 4.209644859274199e-06,
+      "loss": 60.3106,
+      "step": 73690
+    },
+    {
+      "epoch": 0.2977573257594428,
+      "grad_norm": 534.8817138671875,
+      "learning_rate": 4.208252255158387e-06,
+      "loss": 31.561,
+      "step": 73700
+    },
+    {
+      "epoch": 0.29779772702481044,
+      "grad_norm": 672.9677734375,
+      "learning_rate": 4.2068597140383084e-06,
+      "loss": 38.9706,
+      "step": 73710
+    },
+    {
+      "epoch": 0.29783812829017803,
+      "grad_norm": 590.738037109375,
+      "learning_rate": 4.205467236024763e-06,
+      "loss": 44.2501,
+      "step": 73720
+    },
+    {
+      "epoch": 0.29787852955554567,
+      "grad_norm": 580.1815185546875,
+      "learning_rate": 4.204074821228542e-06,
+      "loss": 50.2648,
+      "step": 73730
+    },
+    {
+      "epoch": 0.2979189308209133,
+      "grad_norm": 895.4475708007812,
+      "learning_rate": 4.202682469760436e-06,
+      "loss": 57.8472,
+      "step": 73740
+    },
+    {
+      "epoch": 0.29795933208628095,
+      "grad_norm": 537.7899780273438,
+      "learning_rate": 4.2012901817312255e-06,
+      "loss": 48.3372,
+      "step": 73750
+    },
+    {
+      "epoch": 0.2979997333516486,
+      "grad_norm": 437.3421936035156,
+      "learning_rate": 4.199897957251693e-06,
+      "loss": 36.2092,
+      "step": 73760
+    },
+    {
+      "epoch": 0.29804013461701623,
+      "grad_norm": 567.8798828125,
+      "learning_rate": 4.198505796432605e-06,
+      "loss": 35.1477,
+      "step": 73770
+    },
+    {
+      "epoch": 0.2980805358823838,
+      "grad_norm": 838.5132446289062,
+      "learning_rate": 4.197113699384732e-06,
+      "loss": 50.7975,
+      "step": 73780
+    },
+    {
+      "epoch": 0.29812093714775145,
+      "grad_norm": 676.006103515625,
+      "learning_rate": 4.1957216662188385e-06,
+      "loss": 47.005,
+      "step": 73790
+    },
+    {
+      "epoch": 0.2981613384131191,
+      "grad_norm": 473.2902526855469,
+      "learning_rate": 4.194329697045681e-06,
+      "loss": 57.9121,
+      "step": 73800
+    },
+    {
+      "epoch": 0.29820173967848673,
+      "grad_norm": 315.8070983886719,
+      "learning_rate": 4.19293779197601e-06,
+      "loss": 37.5221,
+      "step": 73810
+    },
+    {
+      "epoch": 0.2982421409438544,
+      "grad_norm": 545.0514526367188,
+      "learning_rate": 4.191545951120577e-06,
+      "loss": 44.9015,
+      "step": 73820
+    },
+    {
+      "epoch": 0.298282542209222,
+      "grad_norm": 588.2539672851562,
+      "learning_rate": 4.190154174590122e-06,
+      "loss": 35.0877,
+      "step": 73830
+    },
+    {
+      "epoch": 0.2983229434745896,
+      "grad_norm": 513.2553100585938,
+      "learning_rate": 4.188762462495381e-06,
+      "loss": 34.4793,
+      "step": 73840
+    },
+    {
+      "epoch": 0.29836334473995724,
+      "grad_norm": 1437.9708251953125,
+      "learning_rate": 4.187370814947091e-06,
+      "loss": 47.6676,
+      "step": 73850
+    },
+    {
+      "epoch": 0.2984037460053249,
+      "grad_norm": 2023.705810546875,
+      "learning_rate": 4.185979232055975e-06,
+      "loss": 77.7692,
+      "step": 73860
+    },
+    {
+      "epoch": 0.2984441472706925,
+      "grad_norm": 292.62713623046875,
+      "learning_rate": 4.184587713932755e-06,
+      "loss": 38.4317,
+      "step": 73870
+    },
+    {
+      "epoch": 0.29848454853606016,
+      "grad_norm": 679.334716796875,
+      "learning_rate": 4.183196260688147e-06,
+      "loss": 38.1238,
+      "step": 73880
+    },
+    {
+      "epoch": 0.2985249498014278,
+      "grad_norm": 554.282958984375,
+      "learning_rate": 4.1818048724328646e-06,
+      "loss": 41.1552,
+      "step": 73890
+    },
+    {
+      "epoch": 0.29856535106679544,
+      "grad_norm": 1359.9815673828125,
+      "learning_rate": 4.180413549277614e-06,
+      "loss": 42.1656,
+      "step": 73900
+    },
+    {
+      "epoch": 0.298605752332163,
+      "grad_norm": 544.4028930664062,
+      "learning_rate": 4.1790222913330955e-06,
+      "loss": 31.6292,
+      "step": 73910
+    },
+    {
+      "epoch": 0.29864615359753066,
+      "grad_norm": 779.8333740234375,
+      "learning_rate": 4.1776310987100054e-06,
+      "loss": 59.4306,
+      "step": 73920
+    },
+    {
+      "epoch": 0.2986865548628983,
+      "grad_norm": 299.58441162109375,
+      "learning_rate": 4.1762399715190366e-06,
+      "loss": 34.8977,
+      "step": 73930
+    },
+    {
+      "epoch": 0.29872695612826594,
+      "grad_norm": 1023.2490234375,
+      "learning_rate": 4.1748489098708715e-06,
+      "loss": 61.7636,
+      "step": 73940
+    },
+    {
+      "epoch": 0.2987673573936336,
+      "grad_norm": 496.14874267578125,
+      "learning_rate": 4.173457913876191e-06,
+      "loss": 37.2103,
+      "step": 73950
+    },
+    {
+      "epoch": 0.2988077586590012,
+      "grad_norm": 419.87164306640625,
+      "learning_rate": 4.172066983645671e-06,
+      "loss": 26.5351,
+      "step": 73960
+    },
+    {
+      "epoch": 0.2988481599243688,
+      "grad_norm": 309.26873779296875,
+      "learning_rate": 4.170676119289982e-06,
+      "loss": 45.8094,
+      "step": 73970
+    },
+    {
+      "epoch": 0.29888856118973645,
+      "grad_norm": 278.8150329589844,
+      "learning_rate": 4.1692853209197865e-06,
+      "loss": 49.6217,
+      "step": 73980
+    },
+    {
+      "epoch": 0.2989289624551041,
+      "grad_norm": 314.2029724121094,
+      "learning_rate": 4.167894588645746e-06,
+      "loss": 28.5693,
+      "step": 73990
+    },
+    {
+      "epoch": 0.2989693637204717,
+      "grad_norm": 554.4902954101562,
+      "learning_rate": 4.166503922578516e-06,
+      "loss": 39.707,
+      "step": 74000
+    },
+    {
+      "epoch": 0.29900976498583937,
+      "grad_norm": 635.1714477539062,
+      "learning_rate": 4.165113322828743e-06,
+      "loss": 43.1086,
+      "step": 74010
+    },
+    {
+      "epoch": 0.299050166251207,
+      "grad_norm": 404.27142333984375,
+      "learning_rate": 4.163722789507071e-06,
+      "loss": 33.8543,
+      "step": 74020
+    },
+    {
+      "epoch": 0.29909056751657465,
+      "grad_norm": 376.3638916015625,
+      "learning_rate": 4.162332322724139e-06,
+      "loss": 37.309,
+      "step": 74030
+    },
+    {
+      "epoch": 0.29913096878194223,
+      "grad_norm": 647.9591064453125,
+      "learning_rate": 4.16094192259058e-06,
+      "loss": 37.8924,
+      "step": 74040
+    },
+    {
+      "epoch": 0.29917137004730987,
+      "grad_norm": 824.0821533203125,
+      "learning_rate": 4.15955158921702e-06,
+      "loss": 48.8199,
+      "step": 74050
+    },
+    {
+      "epoch": 0.2992117713126775,
+      "grad_norm": 1072.1903076171875,
+      "learning_rate": 4.158161322714085e-06,
+      "loss": 50.3598,
+      "step": 74060
+    },
+    {
+      "epoch": 0.29925217257804515,
+      "grad_norm": 660.3749389648438,
+      "learning_rate": 4.1567711231923876e-06,
+      "loss": 36.4483,
+      "step": 74070
+    },
+    {
+      "epoch": 0.2992925738434128,
+      "grad_norm": 907.34423828125,
+      "learning_rate": 4.155380990762542e-06,
+      "loss": 56.8118,
+      "step": 74080
+    },
+    {
+      "epoch": 0.29933297510878043,
+      "grad_norm": 1100.1082763671875,
+      "learning_rate": 4.153990925535157e-06,
+      "loss": 34.7482,
+      "step": 74090
+    },
+    {
+      "epoch": 0.299373376374148,
+      "grad_norm": 784.7665405273438,
+      "learning_rate": 4.15260092762083e-06,
+      "loss": 64.8724,
+      "step": 74100
+    },
+    {
+      "epoch": 0.29941377763951565,
+      "grad_norm": 332.3941955566406,
+      "learning_rate": 4.151210997130159e-06,
+      "loss": 44.2105,
+      "step": 74110
+    },
+    {
+      "epoch": 0.2994541789048833,
+      "grad_norm": 1009.3142700195312,
+      "learning_rate": 4.1498211341737335e-06,
+      "loss": 68.1092,
+      "step": 74120
+    },
+    {
+      "epoch": 0.29949458017025093,
+      "grad_norm": 584.8860473632812,
+      "learning_rate": 4.148431338862138e-06,
+      "loss": 54.0953,
+      "step": 74130
+    },
+    {
+      "epoch": 0.2995349814356186,
+      "grad_norm": 696.730224609375,
+      "learning_rate": 4.147041611305952e-06,
+      "loss": 40.4448,
+      "step": 74140
+    },
+    {
+      "epoch": 0.2995753827009862,
+      "grad_norm": 566.9742431640625,
+      "learning_rate": 4.145651951615752e-06,
+      "loss": 40.5708,
+      "step": 74150
+    },
+    {
+      "epoch": 0.2996157839663538,
+      "grad_norm": 766.1707153320312,
+      "learning_rate": 4.144262359902104e-06,
+      "loss": 43.4644,
+      "step": 74160
+    },
+    {
+      "epoch": 0.29965618523172144,
+      "grad_norm": 870.1135864257812,
+      "learning_rate": 4.142872836275572e-06,
+      "loss": 50.9456,
+      "step": 74170
+    },
+    {
+      "epoch": 0.2996965864970891,
+      "grad_norm": 333.26947021484375,
+      "learning_rate": 4.141483380846716e-06,
+      "loss": 50.0214,
+      "step": 74180
+    },
+    {
+      "epoch": 0.2997369877624567,
+      "grad_norm": 222.6185302734375,
+      "learning_rate": 4.1400939937260894e-06,
+      "loss": 25.4425,
+      "step": 74190
+    },
+    {
+      "epoch": 0.29977738902782436,
+      "grad_norm": 474.7249755859375,
+      "learning_rate": 4.138704675024235e-06,
+      "loss": 39.1855,
+      "step": 74200
+    },
+    {
+      "epoch": 0.299817790293192,
+      "grad_norm": 416.831298828125,
+      "learning_rate": 4.137315424851696e-06,
+      "loss": 29.6196,
+      "step": 74210
+    },
+    {
+      "epoch": 0.29985819155855964,
+      "grad_norm": 377.79144287109375,
+      "learning_rate": 4.1359262433190105e-06,
+      "loss": 46.4644,
+      "step": 74220
+    },
+    {
+      "epoch": 0.2998985928239272,
+      "grad_norm": 1291.645263671875,
+      "learning_rate": 4.134537130536708e-06,
+      "loss": 46.8967,
+      "step": 74230
+    },
+    {
+      "epoch": 0.29993899408929486,
+      "grad_norm": 511.9327392578125,
+      "learning_rate": 4.133148086615314e-06,
+      "loss": 37.661,
+      "step": 74240
+    },
+    {
+      "epoch": 0.2999793953546625,
+      "grad_norm": 490.469970703125,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 37.2974,
+      "step": 74250
+    },
+    {
+      "epoch": 0.30001979662003014,
+      "grad_norm": 643.108154296875,
+      "learning_rate": 4.130370205797326e-06,
+      "loss": 36.1692,
+      "step": 74260
+    },
+    {
+      "epoch": 0.3000601978853978,
+      "grad_norm": 365.9134521484375,
+      "learning_rate": 4.128981369121754e-06,
+      "loss": 40.0982,
+      "step": 74270
+    },
+    {
+      "epoch": 0.3001005991507654,
+      "grad_norm": 735.3226318359375,
+      "learning_rate": 4.127592601749141e-06,
+      "loss": 48.7191,
+      "step": 74280
+    },
+    {
+      "epoch": 0.300141000416133,
+      "grad_norm": 362.46533203125,
+      "learning_rate": 4.1262039037899775e-06,
+      "loss": 38.9229,
+      "step": 74290
+    },
+    {
+      "epoch": 0.30018140168150065,
+      "grad_norm": 391.4541931152344,
+      "learning_rate": 4.12481527535476e-06,
+      "loss": 44.5621,
+      "step": 74300
+    },
+    {
+      "epoch": 0.3002218029468683,
+      "grad_norm": 1557.863525390625,
+      "learning_rate": 4.123426716553972e-06,
+      "loss": 61.9819,
+      "step": 74310
+    },
+    {
+      "epoch": 0.30026220421223593,
+      "grad_norm": 845.4327392578125,
+      "learning_rate": 4.122038227498101e-06,
+      "loss": 58.8395,
+      "step": 74320
+    },
+    {
+      "epoch": 0.30030260547760357,
+      "grad_norm": 483.0872802734375,
+      "learning_rate": 4.120649808297616e-06,
+      "loss": 84.1862,
+      "step": 74330
+    },
+    {
+      "epoch": 0.3003430067429712,
+      "grad_norm": 752.4678344726562,
+      "learning_rate": 4.119261459062992e-06,
+      "loss": 49.809,
+      "step": 74340
+    },
+    {
+      "epoch": 0.30038340800833885,
+      "grad_norm": 527.9071655273438,
+      "learning_rate": 4.1178731799046915e-06,
+      "loss": 48.4955,
+      "step": 74350
+    },
+    {
+      "epoch": 0.30042380927370643,
+      "grad_norm": 1314.4130859375,
+      "learning_rate": 4.116484970933174e-06,
+      "loss": 41.3964,
+      "step": 74360
+    },
+    {
+      "epoch": 0.3004642105390741,
+      "grad_norm": 586.785400390625,
+      "learning_rate": 4.1150968322588915e-06,
+      "loss": 56.8744,
+      "step": 74370
+    },
+    {
+      "epoch": 0.3005046118044417,
+      "grad_norm": 906.0205078125,
+      "learning_rate": 4.113708763992294e-06,
+      "loss": 45.647,
+      "step": 74380
+    },
+    {
+      "epoch": 0.30054501306980935,
+      "grad_norm": 271.83453369140625,
+      "learning_rate": 4.1123207662438216e-06,
+      "loss": 53.6612,
+      "step": 74390
+    },
+    {
+      "epoch": 0.300585414335177,
+      "grad_norm": 772.8023681640625,
+      "learning_rate": 4.110932839123911e-06,
+      "loss": 45.7652,
+      "step": 74400
+    },
+    {
+      "epoch": 0.30062581560054463,
+      "grad_norm": 1087.2738037109375,
+      "learning_rate": 4.109544982742995e-06,
+      "loss": 47.2202,
+      "step": 74410
+    },
+    {
+      "epoch": 0.3006662168659122,
+      "grad_norm": 195.87841796875,
+      "learning_rate": 4.108157197211499e-06,
+      "loss": 37.906,
+      "step": 74420
+    },
+    {
+      "epoch": 0.30070661813127986,
+      "grad_norm": 625.649169921875,
+      "learning_rate": 4.1067694826398405e-06,
+      "loss": 57.2702,
+      "step": 74430
+    },
+    {
+      "epoch": 0.3007470193966475,
+      "grad_norm": 1060.251220703125,
+      "learning_rate": 4.105381839138436e-06,
+      "loss": 54.2827,
+      "step": 74440
+    },
+    {
+      "epoch": 0.30078742066201514,
+      "grad_norm": 236.252197265625,
+      "learning_rate": 4.103994266817694e-06,
+      "loss": 29.2483,
+      "step": 74450
+    },
+    {
+      "epoch": 0.3008278219273828,
+      "grad_norm": 607.3275756835938,
+      "learning_rate": 4.102606765788014e-06,
+      "loss": 41.7655,
+      "step": 74460
+    },
+    {
+      "epoch": 0.3008682231927504,
+      "grad_norm": 469.4609069824219,
+      "learning_rate": 4.101219336159795e-06,
+      "loss": 39.0194,
+      "step": 74470
+    },
+    {
+      "epoch": 0.300908624458118,
+      "grad_norm": 527.9530639648438,
+      "learning_rate": 4.09983197804343e-06,
+      "loss": 49.2814,
+      "step": 74480
+    },
+    {
+      "epoch": 0.30094902572348564,
+      "grad_norm": 707.5580444335938,
+      "learning_rate": 4.098444691549302e-06,
+      "loss": 48.481,
+      "step": 74490
+    },
+    {
+      "epoch": 0.3009894269888533,
+      "grad_norm": 266.0844421386719,
+      "learning_rate": 4.097057476787792e-06,
+      "loss": 37.1013,
+      "step": 74500
+    },
+    {
+      "epoch": 0.3010298282542209,
+      "grad_norm": 565.5172119140625,
+      "learning_rate": 4.0956703338692755e-06,
+      "loss": 50.1374,
+      "step": 74510
+    },
+    {
+      "epoch": 0.30107022951958856,
+      "grad_norm": 756.9215087890625,
+      "learning_rate": 4.09428326290412e-06,
+      "loss": 47.4215,
+      "step": 74520
+    },
+    {
+      "epoch": 0.3011106307849562,
+      "grad_norm": 368.2695617675781,
+      "learning_rate": 4.092896264002689e-06,
+      "loss": 66.5132,
+      "step": 74530
+    },
+    {
+      "epoch": 0.30115103205032384,
+      "grad_norm": 397.1333923339844,
+      "learning_rate": 4.09150933727534e-06,
+      "loss": 43.6294,
+      "step": 74540
+    },
+    {
+      "epoch": 0.3011914333156914,
+      "grad_norm": 443.4328918457031,
+      "learning_rate": 4.0901224828324225e-06,
+      "loss": 55.6375,
+      "step": 74550
+    },
+    {
+      "epoch": 0.30123183458105907,
+      "grad_norm": 1249.0457763671875,
+      "learning_rate": 4.088735700784283e-06,
+      "loss": 51.6747,
+      "step": 74560
+    },
+    {
+      "epoch": 0.3012722358464267,
+      "grad_norm": 779.8118896484375,
+      "learning_rate": 4.087348991241262e-06,
+      "loss": 32.0234,
+      "step": 74570
+    },
+    {
+      "epoch": 0.30131263711179435,
+      "grad_norm": 902.4772338867188,
+      "learning_rate": 4.0859623543136935e-06,
+      "loss": 39.6898,
+      "step": 74580
+    },
+    {
+      "epoch": 0.301353038377162,
+      "grad_norm": 556.6895751953125,
+      "learning_rate": 4.084575790111905e-06,
+      "loss": 38.2554,
+      "step": 74590
+    },
+    {
+      "epoch": 0.3013934396425296,
+      "grad_norm": 1023.3602294921875,
+      "learning_rate": 4.08318929874622e-06,
+      "loss": 46.229,
+      "step": 74600
+    },
+    {
+      "epoch": 0.3014338409078972,
+      "grad_norm": 625.0224609375,
+      "learning_rate": 4.081802880326955e-06,
+      "loss": 52.6646,
+      "step": 74610
+    },
+    {
+      "epoch": 0.30147424217326485,
+      "grad_norm": 291.3082580566406,
+      "learning_rate": 4.080416534964422e-06,
+      "loss": 42.6597,
+      "step": 74620
+    },
+    {
+      "epoch": 0.3015146434386325,
+      "grad_norm": 2209.510009765625,
+      "learning_rate": 4.079030262768924e-06,
+      "loss": 50.5197,
+      "step": 74630
+    },
+    {
+      "epoch": 0.30155504470400013,
+      "grad_norm": 542.4468994140625,
+      "learning_rate": 4.077644063850761e-06,
+      "loss": 39.2088,
+      "step": 74640
+    },
+    {
+      "epoch": 0.30159544596936777,
+      "grad_norm": 584.303466796875,
+      "learning_rate": 4.076257938320226e-06,
+      "loss": 46.4614,
+      "step": 74650
+    },
+    {
+      "epoch": 0.3016358472347354,
+      "grad_norm": 516.731689453125,
+      "learning_rate": 4.074871886287609e-06,
+      "loss": 31.4396,
+      "step": 74660
+    },
+    {
+      "epoch": 0.30167624850010305,
+      "grad_norm": 464.3906555175781,
+      "learning_rate": 4.073485907863189e-06,
+      "loss": 59.6245,
+      "step": 74670
+    },
+    {
+      "epoch": 0.30171664976547063,
+      "grad_norm": 535.5070190429688,
+      "learning_rate": 4.0721000031572445e-06,
+      "loss": 42.0167,
+      "step": 74680
+    },
+    {
+      "epoch": 0.3017570510308383,
+      "grad_norm": 697.2291870117188,
+      "learning_rate": 4.070714172280043e-06,
+      "loss": 51.2677,
+      "step": 74690
+    },
+    {
+      "epoch": 0.3017974522962059,
+      "grad_norm": 596.2420654296875,
+      "learning_rate": 4.06932841534185e-06,
+      "loss": 36.7729,
+      "step": 74700
+    },
+    {
+      "epoch": 0.30183785356157355,
+      "grad_norm": 797.6749267578125,
+      "learning_rate": 4.067942732452926e-06,
+      "loss": 30.443,
+      "step": 74710
+    },
+    {
+      "epoch": 0.3018782548269412,
+      "grad_norm": 984.3765869140625,
+      "learning_rate": 4.06655712372352e-06,
+      "loss": 42.8845,
+      "step": 74720
+    },
+    {
+      "epoch": 0.30191865609230883,
+      "grad_norm": 435.9882507324219,
+      "learning_rate": 4.065171589263878e-06,
+      "loss": 41.4888,
+      "step": 74730
+    },
+    {
+      "epoch": 0.3019590573576764,
+      "grad_norm": 613.9766845703125,
+      "learning_rate": 4.063786129184243e-06,
+      "loss": 47.646,
+      "step": 74740
+    },
+    {
+      "epoch": 0.30199945862304406,
+      "grad_norm": 795.3075561523438,
+      "learning_rate": 4.06240074359485e-06,
+      "loss": 64.4297,
+      "step": 74750
+    },
+    {
+      "epoch": 0.3020398598884117,
+      "grad_norm": 880.7652587890625,
+      "learning_rate": 4.061015432605927e-06,
+      "loss": 35.325,
+      "step": 74760
+    },
+    {
+      "epoch": 0.30208026115377934,
+      "grad_norm": 693.7550048828125,
+      "learning_rate": 4.059630196327696e-06,
+      "loss": 57.8272,
+      "step": 74770
+    },
+    {
+      "epoch": 0.302120662419147,
+      "grad_norm": 653.6392211914062,
+      "learning_rate": 4.058245034870375e-06,
+      "loss": 42.8275,
+      "step": 74780
+    },
+    {
+      "epoch": 0.3021610636845146,
+      "grad_norm": 270.9452819824219,
+      "learning_rate": 4.056859948344175e-06,
+      "loss": 31.6941,
+      "step": 74790
+    },
+    {
+      "epoch": 0.3022014649498822,
+      "grad_norm": 675.5482788085938,
+      "learning_rate": 4.0554749368593e-06,
+      "loss": 36.3624,
+      "step": 74800
+    },
+    {
+      "epoch": 0.30224186621524984,
+      "grad_norm": 372.732177734375,
+      "learning_rate": 4.054090000525949e-06,
+      "loss": 39.691,
+      "step": 74810
+    },
+    {
+      "epoch": 0.3022822674806175,
+      "grad_norm": 543.2750244140625,
+      "learning_rate": 4.052705139454316e-06,
+      "loss": 27.9947,
+      "step": 74820
+    },
+    {
+      "epoch": 0.3023226687459851,
+      "grad_norm": 817.23779296875,
+      "learning_rate": 4.051320353754586e-06,
+      "loss": 52.1108,
+      "step": 74830
+    },
+    {
+      "epoch": 0.30236307001135276,
+      "grad_norm": 704.6561279296875,
+      "learning_rate": 4.049935643536943e-06,
+      "loss": 53.9041,
+      "step": 74840
+    },
+    {
+      "epoch": 0.3024034712767204,
+      "grad_norm": 456.3458251953125,
+      "learning_rate": 4.048551008911561e-06,
+      "loss": 33.8443,
+      "step": 74850
+    },
+    {
+      "epoch": 0.30244387254208804,
+      "grad_norm": 758.6323852539062,
+      "learning_rate": 4.0471664499886074e-06,
+      "loss": 58.1973,
+      "step": 74860
+    },
+    {
+      "epoch": 0.30248427380745563,
+      "grad_norm": 336.8901062011719,
+      "learning_rate": 4.045781966878247e-06,
+      "loss": 28.6064,
+      "step": 74870
+    },
+    {
+      "epoch": 0.30252467507282327,
+      "grad_norm": 528.4429931640625,
+      "learning_rate": 4.044397559690638e-06,
+      "loss": 33.9238,
+      "step": 74880
+    },
+    {
+      "epoch": 0.3025650763381909,
+      "grad_norm": 1521.8575439453125,
+      "learning_rate": 4.043013228535928e-06,
+      "loss": 54.3209,
+      "step": 74890
+    },
+    {
+      "epoch": 0.30260547760355855,
+      "grad_norm": 787.5988159179688,
+      "learning_rate": 4.041628973524264e-06,
+      "loss": 36.775,
+      "step": 74900
+    },
+    {
+      "epoch": 0.3026458788689262,
+      "grad_norm": 830.396240234375,
+      "learning_rate": 4.040244794765783e-06,
+      "loss": 47.7584,
+      "step": 74910
+    },
+    {
+      "epoch": 0.30268628013429383,
+      "grad_norm": 459.3642272949219,
+      "learning_rate": 4.03886069237062e-06,
+      "loss": 38.857,
+      "step": 74920
+    },
+    {
+      "epoch": 0.3027266813996614,
+      "grad_norm": 396.26116943359375,
+      "learning_rate": 4.037476666448899e-06,
+      "loss": 40.7984,
+      "step": 74930
+    },
+    {
+      "epoch": 0.30276708266502905,
+      "grad_norm": 744.2449340820312,
+      "learning_rate": 4.0360927171107436e-06,
+      "loss": 69.9115,
+      "step": 74940
+    },
+    {
+      "epoch": 0.3028074839303967,
+      "grad_norm": 788.3013916015625,
+      "learning_rate": 4.034708844466267e-06,
+      "loss": 42.2636,
+      "step": 74950
+    },
+    {
+      "epoch": 0.30284788519576433,
+      "grad_norm": 576.2045288085938,
+      "learning_rate": 4.033325048625578e-06,
+      "loss": 48.013,
+      "step": 74960
+    },
+    {
+      "epoch": 0.302888286461132,
+      "grad_norm": 430.5680847167969,
+      "learning_rate": 4.031941329698778e-06,
+      "loss": 40.4983,
+      "step": 74970
+    },
+    {
+      "epoch": 0.3029286877264996,
+      "grad_norm": 563.1224365234375,
+      "learning_rate": 4.030557687795965e-06,
+      "loss": 33.5722,
+      "step": 74980
+    },
+    {
+      "epoch": 0.30296908899186725,
+      "grad_norm": 424.4299011230469,
+      "learning_rate": 4.029174123027226e-06,
+      "loss": 47.6854,
+      "step": 74990
+    },
+    {
+      "epoch": 0.30300949025723484,
+      "grad_norm": 1378.790283203125,
+      "learning_rate": 4.027790635502646e-06,
+      "loss": 37.4708,
+      "step": 75000
+    },
+    {
+      "epoch": 0.3030498915226025,
+      "grad_norm": 1003.11865234375,
+      "learning_rate": 4.026407225332305e-06,
+      "loss": 71.0551,
+      "step": 75010
+    },
+    {
+      "epoch": 0.3030902927879701,
+      "grad_norm": 544.5147705078125,
+      "learning_rate": 4.025023892626272e-06,
+      "loss": 33.1692,
+      "step": 75020
+    },
+    {
+      "epoch": 0.30313069405333776,
+      "grad_norm": 786.7648315429688,
+      "learning_rate": 4.023640637494612e-06,
+      "loss": 44.8312,
+      "step": 75030
+    },
+    {
+      "epoch": 0.3031710953187054,
+      "grad_norm": 748.3566284179688,
+      "learning_rate": 4.022257460047387e-06,
+      "loss": 35.4565,
+      "step": 75040
+    },
+    {
+      "epoch": 0.30321149658407304,
+      "grad_norm": 452.1760559082031,
+      "learning_rate": 4.0208743603946505e-06,
+      "loss": 42.6222,
+      "step": 75050
+    },
+    {
+      "epoch": 0.3032518978494406,
+      "grad_norm": 388.5080871582031,
+      "learning_rate": 4.0194913386464445e-06,
+      "loss": 59.1521,
+      "step": 75060
+    },
+    {
+      "epoch": 0.30329229911480826,
+      "grad_norm": 1736.18408203125,
+      "learning_rate": 4.018108394912814e-06,
+      "loss": 73.9259,
+      "step": 75070
+    },
+    {
+      "epoch": 0.3033327003801759,
+      "grad_norm": 324.61883544921875,
+      "learning_rate": 4.016725529303792e-06,
+      "loss": 37.6378,
+      "step": 75080
+    },
+    {
+      "epoch": 0.30337310164554354,
+      "grad_norm": 844.913818359375,
+      "learning_rate": 4.015342741929407e-06,
+      "loss": 29.1962,
+      "step": 75090
+    },
+    {
+      "epoch": 0.3034135029109112,
+      "grad_norm": 572.6766357421875,
+      "learning_rate": 4.013960032899681e-06,
+      "loss": 39.9053,
+      "step": 75100
+    },
+    {
+      "epoch": 0.3034539041762788,
+      "grad_norm": 394.77142333984375,
+      "learning_rate": 4.012577402324631e-06,
+      "loss": 47.6606,
+      "step": 75110
+    },
+    {
+      "epoch": 0.3034943054416464,
+      "grad_norm": 731.8491821289062,
+      "learning_rate": 4.011194850314263e-06,
+      "loss": 47.0361,
+      "step": 75120
+    },
+    {
+      "epoch": 0.30353470670701405,
+      "grad_norm": 450.2945251464844,
+      "learning_rate": 4.009812376978585e-06,
+      "loss": 45.0056,
+      "step": 75130
+    },
+    {
+      "epoch": 0.3035751079723817,
+      "grad_norm": 512.0255126953125,
+      "learning_rate": 4.0084299824275926e-06,
+      "loss": 29.1424,
+      "step": 75140
+    },
+    {
+      "epoch": 0.3036155092377493,
+      "grad_norm": 599.2891845703125,
+      "learning_rate": 4.007047666771274e-06,
+      "loss": 47.6622,
+      "step": 75150
+    },
+    {
+      "epoch": 0.30365591050311697,
+      "grad_norm": 533.6417236328125,
+      "learning_rate": 4.005665430119615e-06,
+      "loss": 30.358,
+      "step": 75160
+    },
+    {
+      "epoch": 0.3036963117684846,
+      "grad_norm": 729.7269897460938,
+      "learning_rate": 4.0042832725825954e-06,
+      "loss": 37.5167,
+      "step": 75170
+    },
+    {
+      "epoch": 0.30373671303385225,
+      "grad_norm": 487.214599609375,
+      "learning_rate": 4.002901194270186e-06,
+      "loss": 50.324,
+      "step": 75180
+    },
+    {
+      "epoch": 0.30377711429921983,
+      "grad_norm": 566.6151123046875,
+      "learning_rate": 4.001519195292352e-06,
+      "loss": 41.4339,
+      "step": 75190
+    },
+    {
+      "epoch": 0.30381751556458747,
+      "grad_norm": 470.7705078125,
+      "learning_rate": 4.000137275759053e-06,
+      "loss": 50.9527,
+      "step": 75200
+    },
+    {
+      "epoch": 0.3038579168299551,
+      "grad_norm": 518.1776733398438,
+      "learning_rate": 3.9987554357802435e-06,
+      "loss": 51.7486,
+      "step": 75210
+    },
+    {
+      "epoch": 0.30389831809532275,
+      "grad_norm": 374.93310546875,
+      "learning_rate": 3.997373675465869e-06,
+      "loss": 42.1258,
+      "step": 75220
+    },
+    {
+      "epoch": 0.3039387193606904,
+      "grad_norm": 643.6387329101562,
+      "learning_rate": 3.995991994925869e-06,
+      "loss": 41.6441,
+      "step": 75230
+    },
+    {
+      "epoch": 0.30397912062605803,
+      "grad_norm": 419.3672790527344,
+      "learning_rate": 3.994610394270178e-06,
+      "loss": 52.551,
+      "step": 75240
+    },
+    {
+      "epoch": 0.3040195218914256,
+      "grad_norm": 828.9144897460938,
+      "learning_rate": 3.993228873608724e-06,
+      "loss": 32.2149,
+      "step": 75250
+    },
+    {
+      "epoch": 0.30405992315679325,
+      "grad_norm": 451.1707458496094,
+      "learning_rate": 3.991847433051427e-06,
+      "loss": 43.9339,
+      "step": 75260
+    },
+    {
+      "epoch": 0.3041003244221609,
+      "grad_norm": 533.591552734375,
+      "learning_rate": 3.990466072708204e-06,
+      "loss": 39.6879,
+      "step": 75270
+    },
+    {
+      "epoch": 0.30414072568752853,
+      "grad_norm": 701.5623779296875,
+      "learning_rate": 3.989084792688962e-06,
+      "loss": 39.0662,
+      "step": 75280
+    },
+    {
+      "epoch": 0.3041811269528962,
+      "grad_norm": 263.5152893066406,
+      "learning_rate": 3.987703593103604e-06,
+      "loss": 40.9297,
+      "step": 75290
+    },
+    {
+      "epoch": 0.3042215282182638,
+      "grad_norm": 351.6845703125,
+      "learning_rate": 3.986322474062025e-06,
+      "loss": 26.069,
+      "step": 75300
+    },
+    {
+      "epoch": 0.30426192948363145,
+      "grad_norm": 593.352783203125,
+      "learning_rate": 3.9849414356741165e-06,
+      "loss": 44.6515,
+      "step": 75310
+    },
+    {
+      "epoch": 0.30430233074899904,
+      "grad_norm": 724.2471313476562,
+      "learning_rate": 3.9835604780497575e-06,
+      "loss": 37.1545,
+      "step": 75320
+    },
+    {
+      "epoch": 0.3043427320143667,
+      "grad_norm": 689.8421020507812,
+      "learning_rate": 3.982179601298827e-06,
+      "loss": 41.7073,
+      "step": 75330
+    },
+    {
+      "epoch": 0.3043831332797343,
+      "grad_norm": 335.9830627441406,
+      "learning_rate": 3.9807988055311946e-06,
+      "loss": 44.3201,
+      "step": 75340
+    },
+    {
+      "epoch": 0.30442353454510196,
+      "grad_norm": 1735.399658203125,
+      "learning_rate": 3.979418090856723e-06,
+      "loss": 52.7617,
+      "step": 75350
+    },
+    {
+      "epoch": 0.3044639358104696,
+      "grad_norm": 764.639892578125,
+      "learning_rate": 3.978037457385268e-06,
+      "loss": 44.3283,
+      "step": 75360
+    },
+    {
+      "epoch": 0.30450433707583724,
+      "grad_norm": 583.4554443359375,
+      "learning_rate": 3.976656905226686e-06,
+      "loss": 49.1295,
+      "step": 75370
+    },
+    {
+      "epoch": 0.3045447383412048,
+      "grad_norm": 756.2467041015625,
+      "learning_rate": 3.9752764344908155e-06,
+      "loss": 52.4948,
+      "step": 75380
+    },
+    {
+      "epoch": 0.30458513960657246,
+      "grad_norm": 580.8175048828125,
+      "learning_rate": 3.9738960452874975e-06,
+      "loss": 57.1519,
+      "step": 75390
+    },
+    {
+      "epoch": 0.3046255408719401,
+      "grad_norm": 538.5950317382812,
+      "learning_rate": 3.972515737726563e-06,
+      "loss": 43.3488,
+      "step": 75400
+    },
+    {
+      "epoch": 0.30466594213730774,
+      "grad_norm": 716.4103393554688,
+      "learning_rate": 3.9711355119178345e-06,
+      "loss": 47.843,
+      "step": 75410
+    },
+    {
+      "epoch": 0.3047063434026754,
+      "grad_norm": 1004.1705932617188,
+      "learning_rate": 3.969755367971131e-06,
+      "loss": 38.9337,
+      "step": 75420
+    },
+    {
+      "epoch": 0.304746744668043,
+      "grad_norm": 560.037109375,
+      "learning_rate": 3.9683753059962646e-06,
+      "loss": 33.0987,
+      "step": 75430
+    },
+    {
+      "epoch": 0.3047871459334106,
+      "grad_norm": 945.2614135742188,
+      "learning_rate": 3.966995326103041e-06,
+      "loss": 54.9676,
+      "step": 75440
+    },
+    {
+      "epoch": 0.30482754719877825,
+      "grad_norm": 543.81787109375,
+      "learning_rate": 3.965615428401257e-06,
+      "loss": 47.2744,
+      "step": 75450
+    },
+    {
+      "epoch": 0.3048679484641459,
+      "grad_norm": 773.4525756835938,
+      "learning_rate": 3.964235613000708e-06,
+      "loss": 57.7128,
+      "step": 75460
+    },
+    {
+      "epoch": 0.30490834972951353,
+      "grad_norm": 652.48046875,
+      "learning_rate": 3.962855880011177e-06,
+      "loss": 23.7307,
+      "step": 75470
+    },
+    {
+      "epoch": 0.30494875099488117,
+      "grad_norm": 485.0013122558594,
+      "learning_rate": 3.961476229542446e-06,
+      "loss": 38.741,
+      "step": 75480
+    },
+    {
+      "epoch": 0.3049891522602488,
+      "grad_norm": 599.990234375,
+      "learning_rate": 3.9600966617042825e-06,
+      "loss": 36.2357,
+      "step": 75490
+    },
+    {
+      "epoch": 0.30502955352561645,
+      "grad_norm": 1090.4490966796875,
+      "learning_rate": 3.958717176606456e-06,
+      "loss": 69.718,
+      "step": 75500
+    },
+    {
+      "epoch": 0.30506995479098403,
+      "grad_norm": 520.7333374023438,
+      "learning_rate": 3.957337774358725e-06,
+      "loss": 53.6938,
+      "step": 75510
+    },
+    {
+      "epoch": 0.3051103560563517,
+      "grad_norm": 568.2504272460938,
+      "learning_rate": 3.955958455070842e-06,
+      "loss": 38.5345,
+      "step": 75520
+    },
+    {
+      "epoch": 0.3051507573217193,
+      "grad_norm": 397.0588073730469,
+      "learning_rate": 3.954579218852553e-06,
+      "loss": 44.8067,
+      "step": 75530
+    },
+    {
+      "epoch": 0.30519115858708695,
+      "grad_norm": 592.46875,
+      "learning_rate": 3.953200065813597e-06,
+      "loss": 45.6429,
+      "step": 75540
+    },
+    {
+      "epoch": 0.3052315598524546,
+      "grad_norm": 597.1698608398438,
+      "learning_rate": 3.951820996063708e-06,
+      "loss": 50.6538,
+      "step": 75550
+    },
+    {
+      "epoch": 0.30527196111782223,
+      "grad_norm": 148.9978790283203,
+      "learning_rate": 3.950442009712612e-06,
+      "loss": 54.7872,
+      "step": 75560
+    },
+    {
+      "epoch": 0.3053123623831898,
+      "grad_norm": 614.67333984375,
+      "learning_rate": 3.949063106870031e-06,
+      "loss": 39.9307,
+      "step": 75570
+    },
+    {
+      "epoch": 0.30535276364855746,
+      "grad_norm": 268.4046325683594,
+      "learning_rate": 3.9476842876456735e-06,
+      "loss": 24.0442,
+      "step": 75580
+    },
+    {
+      "epoch": 0.3053931649139251,
+      "grad_norm": 314.32305908203125,
+      "learning_rate": 3.946305552149247e-06,
+      "loss": 34.3205,
+      "step": 75590
+    },
+    {
+      "epoch": 0.30543356617929274,
+      "grad_norm": 629.6476440429688,
+      "learning_rate": 3.944926900490452e-06,
+      "loss": 35.2416,
+      "step": 75600
+    },
+    {
+      "epoch": 0.3054739674446604,
+      "grad_norm": 873.33984375,
+      "learning_rate": 3.943548332778982e-06,
+      "loss": 44.9323,
+      "step": 75610
+    },
+    {
+      "epoch": 0.305514368710028,
+      "grad_norm": 588.8949584960938,
+      "learning_rate": 3.942169849124523e-06,
+      "loss": 38.1177,
+      "step": 75620
+    },
+    {
+      "epoch": 0.30555476997539566,
+      "grad_norm": 594.586669921875,
+      "learning_rate": 3.940791449636753e-06,
+      "loss": 26.6151,
+      "step": 75630
+    },
+    {
+      "epoch": 0.30559517124076324,
+      "grad_norm": 1274.4742431640625,
+      "learning_rate": 3.939413134425347e-06,
+      "loss": 62.9132,
+      "step": 75640
+    },
+    {
+      "epoch": 0.3056355725061309,
+      "grad_norm": 434.52105712890625,
+      "learning_rate": 3.938034903599972e-06,
+      "loss": 44.5439,
+      "step": 75650
+    },
+    {
+      "epoch": 0.3056759737714985,
+      "grad_norm": 701.6295166015625,
+      "learning_rate": 3.9366567572702845e-06,
+      "loss": 45.023,
+      "step": 75660
+    },
+    {
+      "epoch": 0.30571637503686616,
+      "grad_norm": 408.67327880859375,
+      "learning_rate": 3.935278695545939e-06,
+      "loss": 50.6655,
+      "step": 75670
+    },
+    {
+      "epoch": 0.3057567763022338,
+      "grad_norm": 536.2960815429688,
+      "learning_rate": 3.933900718536579e-06,
+      "loss": 38.9414,
+      "step": 75680
+    },
+    {
+      "epoch": 0.30579717756760144,
+      "grad_norm": 608.7130126953125,
+      "learning_rate": 3.932522826351849e-06,
+      "loss": 44.9229,
+      "step": 75690
+    },
+    {
+      "epoch": 0.305837578832969,
+      "grad_norm": 874.279052734375,
+      "learning_rate": 3.9311450191013774e-06,
+      "loss": 32.2318,
+      "step": 75700
+    },
+    {
+      "epoch": 0.30587798009833667,
+      "grad_norm": 882.767578125,
+      "learning_rate": 3.929767296894792e-06,
+      "loss": 35.0614,
+      "step": 75710
+    },
+    {
+      "epoch": 0.3059183813637043,
+      "grad_norm": 480.5823974609375,
+      "learning_rate": 3.9283896598417104e-06,
+      "loss": 31.6882,
+      "step": 75720
+    },
+    {
+      "epoch": 0.30595878262907195,
+      "grad_norm": 481.7046203613281,
+      "learning_rate": 3.927012108051746e-06,
+      "loss": 39.5504,
+      "step": 75730
+    },
+    {
+      "epoch": 0.3059991838944396,
+      "grad_norm": 164.4939727783203,
+      "learning_rate": 3.925634641634505e-06,
+      "loss": 31.9604,
+      "step": 75740
+    },
+    {
+      "epoch": 0.3060395851598072,
+      "grad_norm": 249.40625,
+      "learning_rate": 3.924257260699583e-06,
+      "loss": 36.5495,
+      "step": 75750
+    },
+    {
+      "epoch": 0.3060799864251748,
+      "grad_norm": 398.69757080078125,
+      "learning_rate": 3.922879965356574e-06,
+      "loss": 49.0819,
+      "step": 75760
+    },
+    {
+      "epoch": 0.30612038769054245,
+      "grad_norm": 1438.6021728515625,
+      "learning_rate": 3.921502755715064e-06,
+      "loss": 57.2429,
+      "step": 75770
+    },
+    {
+      "epoch": 0.3061607889559101,
+      "grad_norm": 600.7191162109375,
+      "learning_rate": 3.920125631884628e-06,
+      "loss": 38.778,
+      "step": 75780
+    },
+    {
+      "epoch": 0.30620119022127773,
+      "grad_norm": 894.872802734375,
+      "learning_rate": 3.918748593974841e-06,
+      "loss": 55.2803,
+      "step": 75790
+    },
+    {
+      "epoch": 0.30624159148664537,
+      "grad_norm": 968.234130859375,
+      "learning_rate": 3.917371642095265e-06,
+      "loss": 66.3114,
+      "step": 75800
+    },
+    {
+      "epoch": 0.306281992752013,
+      "grad_norm": 483.552734375,
+      "learning_rate": 3.91599477635546e-06,
+      "loss": 59.5434,
+      "step": 75810
+    },
+    {
+      "epoch": 0.30632239401738065,
+      "grad_norm": 442.4198303222656,
+      "learning_rate": 3.914617996864976e-06,
+      "loss": 49.4996,
+      "step": 75820
+    },
+    {
+      "epoch": 0.30636279528274823,
+      "grad_norm": 1012.351318359375,
+      "learning_rate": 3.9132413037333565e-06,
+      "loss": 59.5431,
+      "step": 75830
+    },
+    {
+      "epoch": 0.3064031965481159,
+      "grad_norm": 269.81292724609375,
+      "learning_rate": 3.911864697070139e-06,
+      "loss": 30.562,
+      "step": 75840
+    },
+    {
+      "epoch": 0.3064435978134835,
+      "grad_norm": 939.7927856445312,
+      "learning_rate": 3.910488176984853e-06,
+      "loss": 39.42,
+      "step": 75850
+    },
+    {
+      "epoch": 0.30648399907885115,
+      "grad_norm": 710.1830444335938,
+      "learning_rate": 3.909111743587023e-06,
+      "loss": 55.374,
+      "step": 75860
+    },
+    {
+      "epoch": 0.3065244003442188,
+      "grad_norm": 811.8646850585938,
+      "learning_rate": 3.907735396986166e-06,
+      "loss": 36.7296,
+      "step": 75870
+    },
+    {
+      "epoch": 0.30656480160958643,
+      "grad_norm": 520.3335571289062,
+      "learning_rate": 3.9063591372917875e-06,
+      "loss": 32.1305,
+      "step": 75880
+    },
+    {
+      "epoch": 0.306605202874954,
+      "grad_norm": 369.95404052734375,
+      "learning_rate": 3.904982964613395e-06,
+      "loss": 50.2928,
+      "step": 75890
+    },
+    {
+      "epoch": 0.30664560414032166,
+      "grad_norm": 723.0009155273438,
+      "learning_rate": 3.903606879060483e-06,
+      "loss": 31.4758,
+      "step": 75900
+    },
+    {
+      "epoch": 0.3066860054056893,
+      "grad_norm": 626.061767578125,
+      "learning_rate": 3.902230880742541e-06,
+      "loss": 26.1713,
+      "step": 75910
+    },
+    {
+      "epoch": 0.30672640667105694,
+      "grad_norm": 463.85137939453125,
+      "learning_rate": 3.900854969769049e-06,
+      "loss": 41.5159,
+      "step": 75920
+    },
+    {
+      "epoch": 0.3067668079364246,
+      "grad_norm": 1027.1707763671875,
+      "learning_rate": 3.899479146249482e-06,
+      "loss": 45.4365,
+      "step": 75930
+    },
+    {
+      "epoch": 0.3068072092017922,
+      "grad_norm": 424.14703369140625,
+      "learning_rate": 3.898103410293309e-06,
+      "loss": 49.6982,
+      "step": 75940
+    },
+    {
+      "epoch": 0.3068476104671598,
+      "grad_norm": 546.7954711914062,
+      "learning_rate": 3.89672776200999e-06,
+      "loss": 46.8047,
+      "step": 75950
+    },
+    {
+      "epoch": 0.30688801173252744,
+      "grad_norm": 440.770751953125,
+      "learning_rate": 3.895352201508981e-06,
+      "loss": 56.3813,
+      "step": 75960
+    },
+    {
+      "epoch": 0.3069284129978951,
+      "grad_norm": 568.0527954101562,
+      "learning_rate": 3.893976728899726e-06,
+      "loss": 38.7457,
+      "step": 75970
+    },
+    {
+      "epoch": 0.3069688142632627,
+      "grad_norm": 691.5803833007812,
+      "learning_rate": 3.892601344291667e-06,
+      "loss": 44.3474,
+      "step": 75980
+    },
+    {
+      "epoch": 0.30700921552863036,
+      "grad_norm": 467.8775939941406,
+      "learning_rate": 3.891226047794237e-06,
+      "loss": 54.3494,
+      "step": 75990
+    },
+    {
+      "epoch": 0.307049616793998,
+      "grad_norm": 608.9412231445312,
+      "learning_rate": 3.8898508395168645e-06,
+      "loss": 52.8284,
+      "step": 76000
+    },
+    {
+      "epoch": 0.30709001805936564,
+      "grad_norm": 727.375,
+      "learning_rate": 3.888475719568961e-06,
+      "loss": 45.566,
+      "step": 76010
+    },
+    {
+      "epoch": 0.30713041932473323,
+      "grad_norm": 687.283935546875,
+      "learning_rate": 3.887100688059947e-06,
+      "loss": 32.538,
+      "step": 76020
+    },
+    {
+      "epoch": 0.30717082059010087,
+      "grad_norm": 316.9288024902344,
+      "learning_rate": 3.885725745099222e-06,
+      "loss": 36.429,
+      "step": 76030
+    },
+    {
+      "epoch": 0.3072112218554685,
+      "grad_norm": 504.8291015625,
+      "learning_rate": 3.8843508907961855e-06,
+      "loss": 48.7635,
+      "step": 76040
+    },
+    {
+      "epoch": 0.30725162312083615,
+      "grad_norm": 709.7803344726562,
+      "learning_rate": 3.882976125260229e-06,
+      "loss": 49.7421,
+      "step": 76050
+    },
+    {
+      "epoch": 0.3072920243862038,
+      "grad_norm": 788.4102172851562,
+      "learning_rate": 3.881601448600736e-06,
+      "loss": 40.394,
+      "step": 76060
+    },
+    {
+      "epoch": 0.30733242565157143,
+      "grad_norm": 542.3693237304688,
+      "learning_rate": 3.880226860927082e-06,
+      "loss": 49.177,
+      "step": 76070
+    },
+    {
+      "epoch": 0.307372826916939,
+      "grad_norm": 520.00048828125,
+      "learning_rate": 3.8788523623486405e-06,
+      "loss": 43.6388,
+      "step": 76080
+    },
+    {
+      "epoch": 0.30741322818230665,
+      "grad_norm": 415.659912109375,
+      "learning_rate": 3.877477952974768e-06,
+      "loss": 27.0779,
+      "step": 76090
+    },
+    {
+      "epoch": 0.3074536294476743,
+      "grad_norm": 1292.7166748046875,
+      "learning_rate": 3.876103632914825e-06,
+      "loss": 44.3588,
+      "step": 76100
+    },
+    {
+      "epoch": 0.30749403071304193,
+      "grad_norm": 520.9183959960938,
+      "learning_rate": 3.8747294022781555e-06,
+      "loss": 33.352,
+      "step": 76110
+    },
+    {
+      "epoch": 0.3075344319784096,
+      "grad_norm": 587.1658325195312,
+      "learning_rate": 3.873355261174105e-06,
+      "loss": 54.5753,
+      "step": 76120
+    },
+    {
+      "epoch": 0.3075748332437772,
+      "grad_norm": 754.8411254882812,
+      "learning_rate": 3.871981209712006e-06,
+      "loss": 50.3107,
+      "step": 76130
+    },
+    {
+      "epoch": 0.30761523450914485,
+      "grad_norm": 493.7840881347656,
+      "learning_rate": 3.870607248001184e-06,
+      "loss": 37.3539,
+      "step": 76140
+    },
+    {
+      "epoch": 0.30765563577451244,
+      "grad_norm": 714.5708618164062,
+      "learning_rate": 3.869233376150961e-06,
+      "loss": 41.1049,
+      "step": 76150
+    },
+    {
+      "epoch": 0.3076960370398801,
+      "grad_norm": 833.8482055664062,
+      "learning_rate": 3.867859594270649e-06,
+      "loss": 58.7063,
+      "step": 76160
+    },
+    {
+      "epoch": 0.3077364383052477,
+      "grad_norm": 620.5109252929688,
+      "learning_rate": 3.866485902469554e-06,
+      "loss": 36.8431,
+      "step": 76170
+    },
+    {
+      "epoch": 0.30777683957061536,
+      "grad_norm": 558.1962280273438,
+      "learning_rate": 3.865112300856972e-06,
+      "loss": 42.8776,
+      "step": 76180
+    },
+    {
+      "epoch": 0.307817240835983,
+      "grad_norm": 631.0726318359375,
+      "learning_rate": 3.863738789542196e-06,
+      "loss": 34.5687,
+      "step": 76190
+    },
+    {
+      "epoch": 0.30785764210135064,
+      "grad_norm": 667.056640625,
+      "learning_rate": 3.86236536863451e-06,
+      "loss": 40.6587,
+      "step": 76200
+    },
+    {
+      "epoch": 0.3078980433667182,
+      "grad_norm": 284.8694152832031,
+      "learning_rate": 3.860992038243189e-06,
+      "loss": 70.5536,
+      "step": 76210
+    },
+    {
+      "epoch": 0.30793844463208586,
+      "grad_norm": 476.61114501953125,
+      "learning_rate": 3.8596187984775064e-06,
+      "loss": 35.2596,
+      "step": 76220
+    },
+    {
+      "epoch": 0.3079788458974535,
+      "grad_norm": 729.2440185546875,
+      "learning_rate": 3.8582456494467214e-06,
+      "loss": 50.2296,
+      "step": 76230
+    },
+    {
+      "epoch": 0.30801924716282114,
+      "grad_norm": 963.9013061523438,
+      "learning_rate": 3.8568725912600904e-06,
+      "loss": 48.3303,
+      "step": 76240
+    },
+    {
+      "epoch": 0.3080596484281888,
+      "grad_norm": 659.7503051757812,
+      "learning_rate": 3.855499624026861e-06,
+      "loss": 34.8169,
+      "step": 76250
+    },
+    {
+      "epoch": 0.3081000496935564,
+      "grad_norm": 397.644775390625,
+      "learning_rate": 3.854126747856275e-06,
+      "loss": 59.7892,
+      "step": 76260
+    },
+    {
+      "epoch": 0.308140450958924,
+      "grad_norm": 847.7427978515625,
+      "learning_rate": 3.8527539628575635e-06,
+      "loss": 56.8665,
+      "step": 76270
+    },
+    {
+      "epoch": 0.30818085222429165,
+      "grad_norm": 489.5332336425781,
+      "learning_rate": 3.851381269139955e-06,
+      "loss": 37.4831,
+      "step": 76280
+    },
+    {
+      "epoch": 0.3082212534896593,
+      "grad_norm": 817.7260131835938,
+      "learning_rate": 3.8500086668126666e-06,
+      "loss": 48.8452,
+      "step": 76290
+    },
+    {
+      "epoch": 0.3082616547550269,
+      "grad_norm": 610.6978759765625,
+      "learning_rate": 3.848636155984912e-06,
+      "loss": 51.255,
+      "step": 76300
+    },
+    {
+      "epoch": 0.30830205602039457,
+      "grad_norm": 567.9066162109375,
+      "learning_rate": 3.847263736765892e-06,
+      "loss": 57.638,
+      "step": 76310
+    },
+    {
+      "epoch": 0.3083424572857622,
+      "grad_norm": 912.1642456054688,
+      "learning_rate": 3.8458914092648074e-06,
+      "loss": 40.1567,
+      "step": 76320
+    },
+    {
+      "epoch": 0.30838285855112985,
+      "grad_norm": 742.2052612304688,
+      "learning_rate": 3.844519173590847e-06,
+      "loss": 29.8157,
+      "step": 76330
+    },
+    {
+      "epoch": 0.30842325981649743,
+      "grad_norm": 748.3175048828125,
+      "learning_rate": 3.843147029853194e-06,
+      "loss": 50.3242,
+      "step": 76340
+    },
+    {
+      "epoch": 0.30846366108186507,
+      "grad_norm": 515.7076416015625,
+      "learning_rate": 3.841774978161022e-06,
+      "loss": 38.1882,
+      "step": 76350
+    },
+    {
+      "epoch": 0.3085040623472327,
+      "grad_norm": 294.39068603515625,
+      "learning_rate": 3.840403018623499e-06,
+      "loss": 36.2087,
+      "step": 76360
+    },
+    {
+      "epoch": 0.30854446361260035,
+      "grad_norm": 644.3060913085938,
+      "learning_rate": 3.839031151349786e-06,
+      "loss": 54.5945,
+      "step": 76370
+    },
+    {
+      "epoch": 0.308584864877968,
+      "grad_norm": 278.3051452636719,
+      "learning_rate": 3.837659376449036e-06,
+      "loss": 40.8898,
+      "step": 76380
+    },
+    {
+      "epoch": 0.30862526614333563,
+      "grad_norm": 542.3574829101562,
+      "learning_rate": 3.836287694030395e-06,
+      "loss": 40.5055,
+      "step": 76390
+    },
+    {
+      "epoch": 0.3086656674087032,
+      "grad_norm": 800.6304321289062,
+      "learning_rate": 3.834916104203e-06,
+      "loss": 58.5879,
+      "step": 76400
+    },
+    {
+      "epoch": 0.30870606867407085,
+      "grad_norm": 350.9243469238281,
+      "learning_rate": 3.833544607075986e-06,
+      "loss": 32.7555,
+      "step": 76410
+    },
+    {
+      "epoch": 0.3087464699394385,
+      "grad_norm": 389.3909606933594,
+      "learning_rate": 3.8321732027584734e-06,
+      "loss": 52.5082,
+      "step": 76420
+    },
+    {
+      "epoch": 0.30878687120480613,
+      "grad_norm": 806.8661499023438,
+      "learning_rate": 3.830801891359582e-06,
+      "loss": 37.8104,
+      "step": 76430
+    },
+    {
+      "epoch": 0.3088272724701738,
+      "grad_norm": 169.90428161621094,
+      "learning_rate": 3.829430672988414e-06,
+      "loss": 42.1398,
+      "step": 76440
+    },
+    {
+      "epoch": 0.3088676737355414,
+      "grad_norm": 662.7548828125,
+      "learning_rate": 3.828059547754078e-06,
+      "loss": 34.0822,
+      "step": 76450
+    },
+    {
+      "epoch": 0.30890807500090905,
+      "grad_norm": 853.8652954101562,
+      "learning_rate": 3.826688515765664e-06,
+      "loss": 49.2767,
+      "step": 76460
+    },
+    {
+      "epoch": 0.30894847626627664,
+      "grad_norm": 509.5336608886719,
+      "learning_rate": 3.82531757713226e-06,
+      "loss": 50.4154,
+      "step": 76470
+    },
+    {
+      "epoch": 0.3089888775316443,
+      "grad_norm": 701.5994262695312,
+      "learning_rate": 3.823946731962945e-06,
+      "loss": 35.8648,
+      "step": 76480
+    },
+    {
+      "epoch": 0.3090292787970119,
+      "grad_norm": 315.60101318359375,
+      "learning_rate": 3.8225759803667925e-06,
+      "loss": 31.0514,
+      "step": 76490
+    },
+    {
+      "epoch": 0.30906968006237956,
+      "grad_norm": 365.6060791015625,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 44.3825,
+      "step": 76500
+    },
+    {
+      "epoch": 0.3091100813277472,
+      "grad_norm": 290.24053955078125,
+      "learning_rate": 3.81983475833022e-06,
+      "loss": 38.933,
+      "step": 76510
+    },
+    {
+      "epoch": 0.30915048259311484,
+      "grad_norm": 501.88043212890625,
+      "learning_rate": 3.818464288107908e-06,
+      "loss": 52.8634,
+      "step": 76520
+    },
+    {
+      "epoch": 0.3091908838584824,
+      "grad_norm": 809.6956176757812,
+      "learning_rate": 3.817093911894968e-06,
+      "loss": 40.3679,
+      "step": 76530
+    },
+    {
+      "epoch": 0.30923128512385006,
+      "grad_norm": 764.794677734375,
+      "learning_rate": 3.8157236298004375e-06,
+      "loss": 48.3359,
+      "step": 76540
+    },
+    {
+      "epoch": 0.3092716863892177,
+      "grad_norm": 888.4735717773438,
+      "learning_rate": 3.814353441933343e-06,
+      "loss": 52.2985,
+      "step": 76550
+    },
+    {
+      "epoch": 0.30931208765458534,
+      "grad_norm": 623.149169921875,
+      "learning_rate": 3.812983348402703e-06,
+      "loss": 66.8884,
+      "step": 76560
+    },
+    {
+      "epoch": 0.309352488919953,
+      "grad_norm": 762.5501098632812,
+      "learning_rate": 3.811613349317531e-06,
+      "loss": 43.1388,
+      "step": 76570
+    },
+    {
+      "epoch": 0.3093928901853206,
+      "grad_norm": 920.5236206054688,
+      "learning_rate": 3.810243444786831e-06,
+      "loss": 49.3094,
+      "step": 76580
+    },
+    {
+      "epoch": 0.3094332914506882,
+      "grad_norm": 735.0086059570312,
+      "learning_rate": 3.8088736349195995e-06,
+      "loss": 36.29,
+      "step": 76590
+    },
+    {
+      "epoch": 0.30947369271605585,
+      "grad_norm": 2006.9073486328125,
+      "learning_rate": 3.8075039198248274e-06,
+      "loss": 64.1557,
+      "step": 76600
+    },
+    {
+      "epoch": 0.3095140939814235,
+      "grad_norm": 397.25018310546875,
+      "learning_rate": 3.8061342996114946e-06,
+      "loss": 48.698,
+      "step": 76610
+    },
+    {
+      "epoch": 0.30955449524679113,
+      "grad_norm": 694.335205078125,
+      "learning_rate": 3.8047647743885762e-06,
+      "loss": 53.1335,
+      "step": 76620
+    },
+    {
+      "epoch": 0.30959489651215877,
+      "grad_norm": 1197.7288818359375,
+      "learning_rate": 3.8033953442650382e-06,
+      "loss": 60.2836,
+      "step": 76630
+    },
+    {
+      "epoch": 0.3096352977775264,
+      "grad_norm": 780.9642944335938,
+      "learning_rate": 3.802026009349843e-06,
+      "loss": 51.8936,
+      "step": 76640
+    },
+    {
+      "epoch": 0.30967569904289405,
+      "grad_norm": 882.420166015625,
+      "learning_rate": 3.800656769751939e-06,
+      "loss": 46.3737,
+      "step": 76650
+    },
+    {
+      "epoch": 0.30971610030826163,
+      "grad_norm": 623.1157836914062,
+      "learning_rate": 3.799287625580273e-06,
+      "loss": 53.2067,
+      "step": 76660
+    },
+    {
+      "epoch": 0.3097565015736293,
+      "grad_norm": 851.390869140625,
+      "learning_rate": 3.7979185769437795e-06,
+      "loss": 33.3779,
+      "step": 76670
+    },
+    {
+      "epoch": 0.3097969028389969,
+      "grad_norm": 323.7216796875,
+      "learning_rate": 3.7965496239513875e-06,
+      "loss": 35.5668,
+      "step": 76680
+    },
+    {
+      "epoch": 0.30983730410436455,
+      "grad_norm": 810.2097778320312,
+      "learning_rate": 3.79518076671202e-06,
+      "loss": 42.1789,
+      "step": 76690
+    },
+    {
+      "epoch": 0.3098777053697322,
+      "grad_norm": 715.9926147460938,
+      "learning_rate": 3.793812005334589e-06,
+      "loss": 54.3969,
+      "step": 76700
+    },
+    {
+      "epoch": 0.30991810663509983,
+      "grad_norm": 341.7637939453125,
+      "learning_rate": 3.792443339928001e-06,
+      "loss": 45.9676,
+      "step": 76710
+    },
+    {
+      "epoch": 0.3099585079004674,
+      "grad_norm": 652.6634521484375,
+      "learning_rate": 3.7910747706011543e-06,
+      "loss": 43.1486,
+      "step": 76720
+    },
+    {
+      "epoch": 0.30999890916583506,
+      "grad_norm": 288.5016174316406,
+      "learning_rate": 3.7897062974629384e-06,
+      "loss": 27.6082,
+      "step": 76730
+    },
+    {
+      "epoch": 0.3100393104312027,
+      "grad_norm": 672.7640991210938,
+      "learning_rate": 3.78833792062224e-06,
+      "loss": 40.3693,
+      "step": 76740
+    },
+    {
+      "epoch": 0.31007971169657034,
+      "grad_norm": 559.7825317382812,
+      "learning_rate": 3.786969640187932e-06,
+      "loss": 34.8804,
+      "step": 76750
+    },
+    {
+      "epoch": 0.310120112961938,
+      "grad_norm": 1037.9384765625,
+      "learning_rate": 3.785601456268882e-06,
+      "loss": 49.6401,
+      "step": 76760
+    },
+    {
+      "epoch": 0.3101605142273056,
+      "grad_norm": 614.68115234375,
+      "learning_rate": 3.7842333689739524e-06,
+      "loss": 43.7798,
+      "step": 76770
+    },
+    {
+      "epoch": 0.31020091549267326,
+      "grad_norm": 1368.322021484375,
+      "learning_rate": 3.782865378411993e-06,
+      "loss": 49.2492,
+      "step": 76780
+    },
+    {
+      "epoch": 0.31024131675804084,
+      "grad_norm": 436.1568298339844,
+      "learning_rate": 3.7814974846918496e-06,
+      "loss": 55.0184,
+      "step": 76790
+    },
+    {
+      "epoch": 0.3102817180234085,
+      "grad_norm": 635.700927734375,
+      "learning_rate": 3.7801296879223594e-06,
+      "loss": 46.0046,
+      "step": 76800
+    },
+    {
+      "epoch": 0.3103221192887761,
+      "grad_norm": 366.9886779785156,
+      "learning_rate": 3.7787619882123506e-06,
+      "loss": 29.4515,
+      "step": 76810
+    },
+    {
+      "epoch": 0.31036252055414376,
+      "grad_norm": 640.870849609375,
+      "learning_rate": 3.7773943856706463e-06,
+      "loss": 50.9594,
+      "step": 76820
+    },
+    {
+      "epoch": 0.3104029218195114,
+      "grad_norm": 816.541015625,
+      "learning_rate": 3.7760268804060583e-06,
+      "loss": 39.7423,
+      "step": 76830
+    },
+    {
+      "epoch": 0.31044332308487904,
+      "grad_norm": 704.5332641601562,
+      "learning_rate": 3.774659472527396e-06,
+      "loss": 49.7774,
+      "step": 76840
+    },
+    {
+      "epoch": 0.3104837243502466,
+      "grad_norm": 621.05419921875,
+      "learning_rate": 3.7732921621434553e-06,
+      "loss": 36.368,
+      "step": 76850
+    },
+    {
+      "epoch": 0.31052412561561427,
+      "grad_norm": 326.81329345703125,
+      "learning_rate": 3.77192494936303e-06,
+      "loss": 65.5857,
+      "step": 76860
+    },
+    {
+      "epoch": 0.3105645268809819,
+      "grad_norm": 564.2059326171875,
+      "learning_rate": 3.7705578342948967e-06,
+      "loss": 37.76,
+      "step": 76870
+    },
+    {
+      "epoch": 0.31060492814634955,
+      "grad_norm": 49.22088623046875,
+      "learning_rate": 3.7691908170478352e-06,
+      "loss": 53.3579,
+      "step": 76880
+    },
+    {
+      "epoch": 0.3106453294117172,
+      "grad_norm": 522.652587890625,
+      "learning_rate": 3.767823897730612e-06,
+      "loss": 35.3693,
+      "step": 76890
+    },
+    {
+      "epoch": 0.3106857306770848,
+      "grad_norm": 507.7791442871094,
+      "learning_rate": 3.7664570764519865e-06,
+      "loss": 33.2377,
+      "step": 76900
+    },
+    {
+      "epoch": 0.3107261319424524,
+      "grad_norm": 318.0624694824219,
+      "learning_rate": 3.76509035332071e-06,
+      "loss": 49.797,
+      "step": 76910
+    },
+    {
+      "epoch": 0.31076653320782005,
+      "grad_norm": 701.2640991210938,
+      "learning_rate": 3.7637237284455264e-06,
+      "loss": 37.4456,
+      "step": 76920
+    },
+    {
+      "epoch": 0.3108069344731877,
+      "grad_norm": 441.8192443847656,
+      "learning_rate": 3.762357201935171e-06,
+      "loss": 37.457,
+      "step": 76930
+    },
+    {
+      "epoch": 0.31084733573855533,
+      "grad_norm": 556.957763671875,
+      "learning_rate": 3.7609907738983762e-06,
+      "loss": 27.0198,
+      "step": 76940
+    },
+    {
+      "epoch": 0.31088773700392297,
+      "grad_norm": 758.1605834960938,
+      "learning_rate": 3.7596244444438577e-06,
+      "loss": 68.1611,
+      "step": 76950
+    },
+    {
+      "epoch": 0.3109281382692906,
+      "grad_norm": 503.20672607421875,
+      "learning_rate": 3.758258213680328e-06,
+      "loss": 40.469,
+      "step": 76960
+    },
+    {
+      "epoch": 0.31096853953465825,
+      "grad_norm": 579.7887573242188,
+      "learning_rate": 3.7568920817164945e-06,
+      "loss": 45.6774,
+      "step": 76970
+    },
+    {
+      "epoch": 0.31100894080002583,
+      "grad_norm": 799.3428344726562,
+      "learning_rate": 3.755526048661053e-06,
+      "loss": 40.041,
+      "step": 76980
+    },
+    {
+      "epoch": 0.3110493420653935,
+      "grad_norm": 619.9786987304688,
+      "learning_rate": 3.7541601146226924e-06,
+      "loss": 38.2264,
+      "step": 76990
+    },
+    {
+      "epoch": 0.3110897433307611,
+      "grad_norm": 1017.5667724609375,
+      "learning_rate": 3.752794279710094e-06,
+      "loss": 65.4237,
+      "step": 77000
+    },
+    {
+      "epoch": 0.31113014459612875,
+      "grad_norm": 744.3829956054688,
+      "learning_rate": 3.751428544031931e-06,
+      "loss": 53.0718,
+      "step": 77010
+    },
+    {
+      "epoch": 0.3111705458614964,
+      "grad_norm": 1114.8953857421875,
+      "learning_rate": 3.750062907696868e-06,
+      "loss": 46.9796,
+      "step": 77020
+    },
+    {
+      "epoch": 0.31121094712686403,
+      "grad_norm": 789.780517578125,
+      "learning_rate": 3.7486973708135643e-06,
+      "loss": 73.1727,
+      "step": 77030
+    },
+    {
+      "epoch": 0.3112513483922316,
+      "grad_norm": 281.07928466796875,
+      "learning_rate": 3.7473319334906678e-06,
+      "loss": 25.536,
+      "step": 77040
+    },
+    {
+      "epoch": 0.31129174965759926,
+      "grad_norm": 378.5385437011719,
+      "learning_rate": 3.7459665958368197e-06,
+      "loss": 52.1635,
+      "step": 77050
+    },
+    {
+      "epoch": 0.3113321509229669,
+      "grad_norm": 972.1957397460938,
+      "learning_rate": 3.7446013579606534e-06,
+      "loss": 55.4378,
+      "step": 77060
+    },
+    {
+      "epoch": 0.31137255218833454,
+      "grad_norm": 585.01171875,
+      "learning_rate": 3.743236219970796e-06,
+      "loss": 44.6457,
+      "step": 77070
+    },
+    {
+      "epoch": 0.3114129534537022,
+      "grad_norm": 391.7244873046875,
+      "learning_rate": 3.741871181975866e-06,
+      "loss": 40.0719,
+      "step": 77080
+    },
+    {
+      "epoch": 0.3114533547190698,
+      "grad_norm": 1258.5592041015625,
+      "learning_rate": 3.740506244084471e-06,
+      "loss": 44.3012,
+      "step": 77090
+    },
+    {
+      "epoch": 0.31149375598443746,
+      "grad_norm": 786.6157836914062,
+      "learning_rate": 3.7391414064052138e-06,
+      "loss": 56.0882,
+      "step": 77100
+    },
+    {
+      "epoch": 0.31153415724980504,
+      "grad_norm": 362.84814453125,
+      "learning_rate": 3.737776669046689e-06,
+      "loss": 40.2787,
+      "step": 77110
+    },
+    {
+      "epoch": 0.3115745585151727,
+      "grad_norm": 436.129150390625,
+      "learning_rate": 3.7364120321174826e-06,
+      "loss": 40.8267,
+      "step": 77120
+    },
+    {
+      "epoch": 0.3116149597805403,
+      "grad_norm": 357.44561767578125,
+      "learning_rate": 3.7350474957261705e-06,
+      "loss": 39.113,
+      "step": 77130
+    },
+    {
+      "epoch": 0.31165536104590796,
+      "grad_norm": 732.1150512695312,
+      "learning_rate": 3.7336830599813245e-06,
+      "loss": 39.6991,
+      "step": 77140
+    },
+    {
+      "epoch": 0.3116957623112756,
+      "grad_norm": 336.4695129394531,
+      "learning_rate": 3.732318724991505e-06,
+      "loss": 26.2738,
+      "step": 77150
+    },
+    {
+      "epoch": 0.31173616357664324,
+      "grad_norm": 959.9412841796875,
+      "learning_rate": 3.730954490865266e-06,
+      "loss": 58.7199,
+      "step": 77160
+    },
+    {
+      "epoch": 0.31177656484201083,
+      "grad_norm": 521.5714721679688,
+      "learning_rate": 3.7295903577111548e-06,
+      "loss": 37.6181,
+      "step": 77170
+    },
+    {
+      "epoch": 0.31181696610737847,
+      "grad_norm": 231.63853454589844,
+      "learning_rate": 3.728226325637709e-06,
+      "loss": 64.1254,
+      "step": 77180
+    },
+    {
+      "epoch": 0.3118573673727461,
+      "grad_norm": 1082.9320068359375,
+      "learning_rate": 3.726862394753457e-06,
+      "loss": 43.6313,
+      "step": 77190
+    },
+    {
+      "epoch": 0.31189776863811375,
+      "grad_norm": 525.0565185546875,
+      "learning_rate": 3.725498565166923e-06,
+      "loss": 40.8923,
+      "step": 77200
+    },
+    {
+      "epoch": 0.3119381699034814,
+      "grad_norm": 248.5260467529297,
+      "learning_rate": 3.7241348369866183e-06,
+      "loss": 34.3744,
+      "step": 77210
+    },
+    {
+      "epoch": 0.31197857116884903,
+      "grad_norm": 994.7620239257812,
+      "learning_rate": 3.7227712103210485e-06,
+      "loss": 52.4718,
+      "step": 77220
+    },
+    {
+      "epoch": 0.3120189724342166,
+      "grad_norm": 979.7284545898438,
+      "learning_rate": 3.721407685278712e-06,
+      "loss": 45.3776,
+      "step": 77230
+    },
+    {
+      "epoch": 0.31205937369958425,
+      "grad_norm": 410.4310302734375,
+      "learning_rate": 3.7200442619680976e-06,
+      "loss": 51.1721,
+      "step": 77240
+    },
+    {
+      "epoch": 0.3120997749649519,
+      "grad_norm": 821.8783569335938,
+      "learning_rate": 3.7186809404976877e-06,
+      "loss": 37.6959,
+      "step": 77250
+    },
+    {
+      "epoch": 0.31214017623031953,
+      "grad_norm": 598.2932739257812,
+      "learning_rate": 3.7173177209759538e-06,
+      "loss": 78.5719,
+      "step": 77260
+    },
+    {
+      "epoch": 0.3121805774956872,
+      "grad_norm": 599.0471801757812,
+      "learning_rate": 3.715954603511363e-06,
+      "loss": 40.2954,
+      "step": 77270
+    },
+    {
+      "epoch": 0.3122209787610548,
+      "grad_norm": 778.6463623046875,
+      "learning_rate": 3.714591588212372e-06,
+      "loss": 28.0776,
+      "step": 77280
+    },
+    {
+      "epoch": 0.31226138002642245,
+      "grad_norm": 922.8678588867188,
+      "learning_rate": 3.713228675187429e-06,
+      "loss": 37.1462,
+      "step": 77290
+    },
+    {
+      "epoch": 0.31230178129179004,
+      "grad_norm": 521.8157958984375,
+      "learning_rate": 3.7118658645449745e-06,
+      "loss": 47.8419,
+      "step": 77300
+    },
+    {
+      "epoch": 0.3123421825571577,
+      "grad_norm": 1121.2880859375,
+      "learning_rate": 3.710503156393441e-06,
+      "loss": 52.1181,
+      "step": 77310
+    },
+    {
+      "epoch": 0.3123825838225253,
+      "grad_norm": 557.4747314453125,
+      "learning_rate": 3.7091405508412538e-06,
+      "loss": 47.8416,
+      "step": 77320
+    },
+    {
+      "epoch": 0.31242298508789296,
+      "grad_norm": 654.5795288085938,
+      "learning_rate": 3.7077780479968286e-06,
+      "loss": 44.9925,
+      "step": 77330
+    },
+    {
+      "epoch": 0.3124633863532606,
+      "grad_norm": 550.8849487304688,
+      "learning_rate": 3.7064156479685736e-06,
+      "loss": 46.5616,
+      "step": 77340
+    },
+    {
+      "epoch": 0.31250378761862824,
+      "grad_norm": 441.02020263671875,
+      "learning_rate": 3.705053350864888e-06,
+      "loss": 58.3744,
+      "step": 77350
+    },
+    {
+      "epoch": 0.3125441888839958,
+      "grad_norm": 981.9597778320312,
+      "learning_rate": 3.703691156794165e-06,
+      "loss": 38.7906,
+      "step": 77360
+    },
+    {
+      "epoch": 0.31258459014936346,
+      "grad_norm": 517.7142333984375,
+      "learning_rate": 3.7023290658647893e-06,
+      "loss": 43.4523,
+      "step": 77370
+    },
+    {
+      "epoch": 0.3126249914147311,
+      "grad_norm": 432.1444091796875,
+      "learning_rate": 3.7009670781851326e-06,
+      "loss": 34.6007,
+      "step": 77380
+    },
+    {
+      "epoch": 0.31266539268009874,
+      "grad_norm": 5524.80029296875,
+      "learning_rate": 3.6996051938635626e-06,
+      "loss": 57.5598,
+      "step": 77390
+    },
+    {
+      "epoch": 0.3127057939454664,
+      "grad_norm": 469.6958312988281,
+      "learning_rate": 3.69824341300844e-06,
+      "loss": 54.4597,
+      "step": 77400
+    },
+    {
+      "epoch": 0.312746195210834,
+      "grad_norm": 568.2667236328125,
+      "learning_rate": 3.696881735728115e-06,
+      "loss": 72.8601,
+      "step": 77410
+    },
+    {
+      "epoch": 0.31278659647620166,
+      "grad_norm": 342.8549499511719,
+      "learning_rate": 3.6955201621309302e-06,
+      "loss": 33.34,
+      "step": 77420
+    },
+    {
+      "epoch": 0.31282699774156925,
+      "grad_norm": 553.44921875,
+      "learning_rate": 3.6941586923252194e-06,
+      "loss": 39.574,
+      "step": 77430
+    },
+    {
+      "epoch": 0.3128673990069369,
+      "grad_norm": 322.31494140625,
+      "learning_rate": 3.6927973264193074e-06,
+      "loss": 42.0066,
+      "step": 77440
+    },
+    {
+      "epoch": 0.3129078002723045,
+      "grad_norm": 746.9494018554688,
+      "learning_rate": 3.691436064521513e-06,
+      "loss": 50.2764,
+      "step": 77450
+    },
+    {
+      "epoch": 0.31294820153767217,
+      "grad_norm": 535.7817993164062,
+      "learning_rate": 3.6900749067401474e-06,
+      "loss": 45.719,
+      "step": 77460
+    },
+    {
+      "epoch": 0.3129886028030398,
+      "grad_norm": 494.1651306152344,
+      "learning_rate": 3.6887138531835085e-06,
+      "loss": 47.2059,
+      "step": 77470
+    },
+    {
+      "epoch": 0.31302900406840745,
+      "grad_norm": 458.833740234375,
+      "learning_rate": 3.6873529039598903e-06,
+      "loss": 83.885,
+      "step": 77480
+    },
+    {
+      "epoch": 0.31306940533377503,
+      "grad_norm": 452.3143615722656,
+      "learning_rate": 3.6859920591775763e-06,
+      "loss": 28.6129,
+      "step": 77490
+    },
+    {
+      "epoch": 0.31310980659914267,
+      "grad_norm": 1721.304443359375,
+      "learning_rate": 3.6846313189448447e-06,
+      "loss": 69.1576,
+      "step": 77500
+    },
+    {
+      "epoch": 0.3131502078645103,
+      "grad_norm": 1018.3944702148438,
+      "learning_rate": 3.6832706833699616e-06,
+      "loss": 58.151,
+      "step": 77510
+    },
+    {
+      "epoch": 0.31319060912987795,
+      "grad_norm": 710.81494140625,
+      "learning_rate": 3.681910152561188e-06,
+      "loss": 41.7799,
+      "step": 77520
+    },
+    {
+      "epoch": 0.3132310103952456,
+      "grad_norm": 244.9594268798828,
+      "learning_rate": 3.6805497266267742e-06,
+      "loss": 33.3409,
+      "step": 77530
+    },
+    {
+      "epoch": 0.31327141166061323,
+      "grad_norm": 736.140625,
+      "learning_rate": 3.679189405674963e-06,
+      "loss": 40.06,
+      "step": 77540
+    },
+    {
+      "epoch": 0.3133118129259808,
+      "grad_norm": 528.4320678710938,
+      "learning_rate": 3.6778291898139907e-06,
+      "loss": 46.3387,
+      "step": 77550
+    },
+    {
+      "epoch": 0.31335221419134845,
+      "grad_norm": 563.1051635742188,
+      "learning_rate": 3.6764690791520797e-06,
+      "loss": 38.3095,
+      "step": 77560
+    },
+    {
+      "epoch": 0.3133926154567161,
+      "grad_norm": 330.9205627441406,
+      "learning_rate": 3.6751090737974506e-06,
+      "loss": 36.1311,
+      "step": 77570
+    },
+    {
+      "epoch": 0.31343301672208373,
+      "grad_norm": 987.652099609375,
+      "learning_rate": 3.673749173858312e-06,
+      "loss": 62.4743,
+      "step": 77580
+    },
+    {
+      "epoch": 0.3134734179874514,
+      "grad_norm": 787.0823364257812,
+      "learning_rate": 3.672389379442864e-06,
+      "loss": 41.8553,
+      "step": 77590
+    },
+    {
+      "epoch": 0.313513819252819,
+      "grad_norm": 606.4526977539062,
+      "learning_rate": 3.6710296906593012e-06,
+      "loss": 53.3784,
+      "step": 77600
+    },
+    {
+      "epoch": 0.31355422051818665,
+      "grad_norm": 313.4430847167969,
+      "learning_rate": 3.6696701076158064e-06,
+      "loss": 37.6288,
+      "step": 77610
+    },
+    {
+      "epoch": 0.31359462178355424,
+      "grad_norm": 472.91925048828125,
+      "learning_rate": 3.6683106304205564e-06,
+      "loss": 35.023,
+      "step": 77620
+    },
+    {
+      "epoch": 0.3136350230489219,
+      "grad_norm": 933.0067138671875,
+      "learning_rate": 3.666951259181718e-06,
+      "loss": 47.8511,
+      "step": 77630
+    },
+    {
+      "epoch": 0.3136754243142895,
+      "grad_norm": 593.52001953125,
+      "learning_rate": 3.6655919940074497e-06,
+      "loss": 46.1541,
+      "step": 77640
+    },
+    {
+      "epoch": 0.31371582557965716,
+      "grad_norm": 1112.1434326171875,
+      "learning_rate": 3.664232835005902e-06,
+      "loss": 43.5285,
+      "step": 77650
+    },
+    {
+      "epoch": 0.3137562268450248,
+      "grad_norm": 630.2261352539062,
+      "learning_rate": 3.6628737822852177e-06,
+      "loss": 46.3687,
+      "step": 77660
+    },
+    {
+      "epoch": 0.31379662811039244,
+      "grad_norm": 676.2584838867188,
+      "learning_rate": 3.66151483595353e-06,
+      "loss": 36.21,
+      "step": 77670
+    },
+    {
+      "epoch": 0.31383702937576,
+      "grad_norm": 206.3310089111328,
+      "learning_rate": 3.6601559961189626e-06,
+      "loss": 52.2429,
+      "step": 77680
+    },
+    {
+      "epoch": 0.31387743064112766,
+      "grad_norm": 439.7651672363281,
+      "learning_rate": 3.6587972628896345e-06,
+      "loss": 44.6034,
+      "step": 77690
+    },
+    {
+      "epoch": 0.3139178319064953,
+      "grad_norm": 848.1891479492188,
+      "learning_rate": 3.6574386363736532e-06,
+      "loss": 31.7829,
+      "step": 77700
+    },
+    {
+      "epoch": 0.31395823317186294,
+      "grad_norm": 673.7623901367188,
+      "learning_rate": 3.6560801166791183e-06,
+      "loss": 50.483,
+      "step": 77710
+    },
+    {
+      "epoch": 0.3139986344372306,
+      "grad_norm": 611.6978149414062,
+      "learning_rate": 3.654721703914121e-06,
+      "loss": 62.5249,
+      "step": 77720
+    },
+    {
+      "epoch": 0.3140390357025982,
+      "grad_norm": 628.8622436523438,
+      "learning_rate": 3.6533633981867433e-06,
+      "loss": 56.999,
+      "step": 77730
+    },
+    {
+      "epoch": 0.31407943696796586,
+      "grad_norm": 655.521728515625,
+      "learning_rate": 3.652005199605059e-06,
+      "loss": 49.1772,
+      "step": 77740
+    },
+    {
+      "epoch": 0.31411983823333345,
+      "grad_norm": 607.6434326171875,
+      "learning_rate": 3.6506471082771357e-06,
+      "loss": 42.7941,
+      "step": 77750
+    },
+    {
+      "epoch": 0.3141602394987011,
+      "grad_norm": 562.9298095703125,
+      "learning_rate": 3.6492891243110283e-06,
+      "loss": 41.0694,
+      "step": 77760
+    },
+    {
+      "epoch": 0.31420064076406873,
+      "grad_norm": 867.0905151367188,
+      "learning_rate": 3.6479312478147866e-06,
+      "loss": 42.9423,
+      "step": 77770
+    },
+    {
+      "epoch": 0.31424104202943637,
+      "grad_norm": 363.48370361328125,
+      "learning_rate": 3.64657347889645e-06,
+      "loss": 32.8319,
+      "step": 77780
+    },
+    {
+      "epoch": 0.314281443294804,
+      "grad_norm": 822.6988525390625,
+      "learning_rate": 3.6452158176640505e-06,
+      "loss": 58.5034,
+      "step": 77790
+    },
+    {
+      "epoch": 0.31432184456017165,
+      "grad_norm": 778.3346557617188,
+      "learning_rate": 3.6438582642256138e-06,
+      "loss": 46.2177,
+      "step": 77800
+    },
+    {
+      "epoch": 0.31436224582553923,
+      "grad_norm": 823.2440185546875,
+      "learning_rate": 3.642500818689148e-06,
+      "loss": 30.1332,
+      "step": 77810
+    },
+    {
+      "epoch": 0.3144026470909069,
+      "grad_norm": 582.5382690429688,
+      "learning_rate": 3.641143481162661e-06,
+      "loss": 35.8509,
+      "step": 77820
+    },
+    {
+      "epoch": 0.3144430483562745,
+      "grad_norm": 512.2465209960938,
+      "learning_rate": 3.639786251754153e-06,
+      "loss": 45.7103,
+      "step": 77830
+    },
+    {
+      "epoch": 0.31448344962164215,
+      "grad_norm": 445.96392822265625,
+      "learning_rate": 3.638429130571609e-06,
+      "loss": 26.201,
+      "step": 77840
+    },
+    {
+      "epoch": 0.3145238508870098,
+      "grad_norm": 388.62579345703125,
+      "learning_rate": 3.637072117723012e-06,
+      "loss": 50.9161,
+      "step": 77850
+    },
+    {
+      "epoch": 0.31456425215237743,
+      "grad_norm": 1324.3643798828125,
+      "learning_rate": 3.6357152133163297e-06,
+      "loss": 60.3968,
+      "step": 77860
+    },
+    {
+      "epoch": 0.314604653417745,
+      "grad_norm": 485.1225891113281,
+      "learning_rate": 3.634358417459528e-06,
+      "loss": 38.8524,
+      "step": 77870
+    },
+    {
+      "epoch": 0.31464505468311266,
+      "grad_norm": 712.991943359375,
+      "learning_rate": 3.633001730260558e-06,
+      "loss": 40.6697,
+      "step": 77880
+    },
+    {
+      "epoch": 0.3146854559484803,
+      "grad_norm": 565.8276977539062,
+      "learning_rate": 3.63164515182737e-06,
+      "loss": 54.9966,
+      "step": 77890
+    },
+    {
+      "epoch": 0.31472585721384794,
+      "grad_norm": 826.9208984375,
+      "learning_rate": 3.630288682267895e-06,
+      "loss": 43.2319,
+      "step": 77900
+    },
+    {
+      "epoch": 0.3147662584792156,
+      "grad_norm": 603.3365478515625,
+      "learning_rate": 3.628932321690063e-06,
+      "loss": 42.2038,
+      "step": 77910
+    },
+    {
+      "epoch": 0.3148066597445832,
+      "grad_norm": 751.0957641601562,
+      "learning_rate": 3.6275760702017938e-06,
+      "loss": 54.4473,
+      "step": 77920
+    },
+    {
+      "epoch": 0.31484706100995086,
+      "grad_norm": 419.89166259765625,
+      "learning_rate": 3.626219927910999e-06,
+      "loss": 38.9901,
+      "step": 77930
+    },
+    {
+      "epoch": 0.31488746227531844,
+      "grad_norm": 467.87078857421875,
+      "learning_rate": 3.6248638949255795e-06,
+      "loss": 49.0393,
+      "step": 77940
+    },
+    {
+      "epoch": 0.3149278635406861,
+      "grad_norm": 711.9083251953125,
+      "learning_rate": 3.6235079713534287e-06,
+      "loss": 44.0929,
+      "step": 77950
+    },
+    {
+      "epoch": 0.3149682648060537,
+      "grad_norm": 811.3135986328125,
+      "learning_rate": 3.6221521573024316e-06,
+      "loss": 40.8963,
+      "step": 77960
+    },
+    {
+      "epoch": 0.31500866607142136,
+      "grad_norm": 466.0732421875,
+      "learning_rate": 3.620796452880464e-06,
+      "loss": 30.8488,
+      "step": 77970
+    },
+    {
+      "epoch": 0.315049067336789,
+      "grad_norm": 291.6542663574219,
+      "learning_rate": 3.6194408581953934e-06,
+      "loss": 22.6698,
+      "step": 77980
+    },
+    {
+      "epoch": 0.31508946860215664,
+      "grad_norm": 639.0294189453125,
+      "learning_rate": 3.618085373355077e-06,
+      "loss": 45.7208,
+      "step": 77990
+    },
+    {
+      "epoch": 0.3151298698675242,
+      "grad_norm": 1079.23828125,
+      "learning_rate": 3.6167299984673655e-06,
+      "loss": 55.0113,
+      "step": 78000
+    },
+    {
+      "epoch": 0.31517027113289187,
+      "grad_norm": 269.6738586425781,
+      "learning_rate": 3.615374733640099e-06,
+      "loss": 35.7515,
+      "step": 78010
+    },
+    {
+      "epoch": 0.3152106723982595,
+      "grad_norm": 657.4027709960938,
+      "learning_rate": 3.6140195789811108e-06,
+      "loss": 35.9039,
+      "step": 78020
+    },
+    {
+      "epoch": 0.31525107366362715,
+      "grad_norm": 291.2002258300781,
+      "learning_rate": 3.6126645345982243e-06,
+      "loss": 57.9338,
+      "step": 78030
+    },
+    {
+      "epoch": 0.3152914749289948,
+      "grad_norm": 812.0730590820312,
+      "learning_rate": 3.611309600599253e-06,
+      "loss": 43.1736,
+      "step": 78040
+    },
+    {
+      "epoch": 0.3153318761943624,
+      "grad_norm": 754.0714111328125,
+      "learning_rate": 3.6099547770920046e-06,
+      "loss": 48.0824,
+      "step": 78050
+    },
+    {
+      "epoch": 0.31537227745973007,
+      "grad_norm": 477.45733642578125,
+      "learning_rate": 3.6086000641842757e-06,
+      "loss": 27.9684,
+      "step": 78060
+    },
+    {
+      "epoch": 0.31541267872509765,
+      "grad_norm": 324.0244140625,
+      "learning_rate": 3.6072454619838525e-06,
+      "loss": 34.7192,
+      "step": 78070
+    },
+    {
+      "epoch": 0.3154530799904653,
+      "grad_norm": 680.8942260742188,
+      "learning_rate": 3.6058909705985166e-06,
+      "loss": 38.3838,
+      "step": 78080
+    },
+    {
+      "epoch": 0.31549348125583293,
+      "grad_norm": 566.3120727539062,
+      "learning_rate": 3.6045365901360385e-06,
+      "loss": 36.4708,
+      "step": 78090
+    },
+    {
+      "epoch": 0.31553388252120057,
+      "grad_norm": 453.73834228515625,
+      "learning_rate": 3.603182320704179e-06,
+      "loss": 47.0383,
+      "step": 78100
+    },
+    {
+      "epoch": 0.3155742837865682,
+      "grad_norm": 1445.5745849609375,
+      "learning_rate": 3.601828162410691e-06,
+      "loss": 51.0274,
+      "step": 78110
+    },
+    {
+      "epoch": 0.31561468505193585,
+      "grad_norm": 367.10394287109375,
+      "learning_rate": 3.6004741153633194e-06,
+      "loss": 30.2823,
+      "step": 78120
+    },
+    {
+      "epoch": 0.31565508631730343,
+      "grad_norm": 385.9931640625,
+      "learning_rate": 3.5991201796698006e-06,
+      "loss": 31.6421,
+      "step": 78130
+    },
+    {
+      "epoch": 0.3156954875826711,
+      "grad_norm": 457.8427429199219,
+      "learning_rate": 3.5977663554378594e-06,
+      "loss": 55.4395,
+      "step": 78140
+    },
+    {
+      "epoch": 0.3157358888480387,
+      "grad_norm": 644.6337280273438,
+      "learning_rate": 3.5964126427752155e-06,
+      "loss": 36.9936,
+      "step": 78150
+    },
+    {
+      "epoch": 0.31577629011340635,
+      "grad_norm": 459.3667907714844,
+      "learning_rate": 3.595059041789575e-06,
+      "loss": 39.2877,
+      "step": 78160
+    },
+    {
+      "epoch": 0.315816691378774,
+      "grad_norm": 527.1011352539062,
+      "learning_rate": 3.5937055525886377e-06,
+      "loss": 46.0837,
+      "step": 78170
+    },
+    {
+      "epoch": 0.31585709264414163,
+      "grad_norm": 789.7222900390625,
+      "learning_rate": 3.592352175280096e-06,
+      "loss": 51.5472,
+      "step": 78180
+    },
+    {
+      "epoch": 0.3158974939095092,
+      "grad_norm": 703.373046875,
+      "learning_rate": 3.5909989099716325e-06,
+      "loss": 51.672,
+      "step": 78190
+    },
+    {
+      "epoch": 0.31593789517487686,
+      "grad_norm": 1366.6173095703125,
+      "learning_rate": 3.589645756770918e-06,
+      "loss": 54.4186,
+      "step": 78200
+    },
+    {
+      "epoch": 0.3159782964402445,
+      "grad_norm": 651.6315307617188,
+      "learning_rate": 3.5882927157856175e-06,
+      "loss": 37.2753,
+      "step": 78210
+    },
+    {
+      "epoch": 0.31601869770561214,
+      "grad_norm": 457.5976867675781,
+      "learning_rate": 3.586939787123388e-06,
+      "loss": 58.4296,
+      "step": 78220
+    },
+    {
+      "epoch": 0.3160590989709798,
+      "grad_norm": 463.1532287597656,
+      "learning_rate": 3.585586970891876e-06,
+      "loss": 42.9541,
+      "step": 78230
+    },
+    {
+      "epoch": 0.3160995002363474,
+      "grad_norm": 802.164794921875,
+      "learning_rate": 3.584234267198715e-06,
+      "loss": 37.8752,
+      "step": 78240
+    },
+    {
+      "epoch": 0.31613990150171506,
+      "grad_norm": 438.13385009765625,
+      "learning_rate": 3.582881676151536e-06,
+      "loss": 35.8725,
+      "step": 78250
+    },
+    {
+      "epoch": 0.31618030276708264,
+      "grad_norm": 605.7740478515625,
+      "learning_rate": 3.581529197857959e-06,
+      "loss": 34.4586,
+      "step": 78260
+    },
+    {
+      "epoch": 0.3162207040324503,
+      "grad_norm": 503.4420471191406,
+      "learning_rate": 3.580176832425594e-06,
+      "loss": 46.2822,
+      "step": 78270
+    },
+    {
+      "epoch": 0.3162611052978179,
+      "grad_norm": 762.1167602539062,
+      "learning_rate": 3.5788245799620425e-06,
+      "loss": 51.9271,
+      "step": 78280
+    },
+    {
+      "epoch": 0.31630150656318556,
+      "grad_norm": 671.9507446289062,
+      "learning_rate": 3.577472440574896e-06,
+      "loss": 41.3695,
+      "step": 78290
+    },
+    {
+      "epoch": 0.3163419078285532,
+      "grad_norm": 408.4527282714844,
+      "learning_rate": 3.5761204143717387e-06,
+      "loss": 56.6144,
+      "step": 78300
+    },
+    {
+      "epoch": 0.31638230909392084,
+      "grad_norm": 357.1184387207031,
+      "learning_rate": 3.5747685014601456e-06,
+      "loss": 33.9895,
+      "step": 78310
+    },
+    {
+      "epoch": 0.31642271035928843,
+      "grad_norm": 260.8604431152344,
+      "learning_rate": 3.5734167019476845e-06,
+      "loss": 32.9133,
+      "step": 78320
+    },
+    {
+      "epoch": 0.31646311162465607,
+      "grad_norm": 756.7543334960938,
+      "learning_rate": 3.572065015941907e-06,
+      "loss": 46.5736,
+      "step": 78330
+    },
+    {
+      "epoch": 0.3165035128900237,
+      "grad_norm": 694.8875732421875,
+      "learning_rate": 3.570713443550362e-06,
+      "loss": 68.6546,
+      "step": 78340
+    },
+    {
+      "epoch": 0.31654391415539135,
+      "grad_norm": 347.2257385253906,
+      "learning_rate": 3.5693619848805892e-06,
+      "loss": 32.6081,
+      "step": 78350
+    },
+    {
+      "epoch": 0.316584315420759,
+      "grad_norm": 580.4989624023438,
+      "learning_rate": 3.568010640040118e-06,
+      "loss": 55.1243,
+      "step": 78360
+    },
+    {
+      "epoch": 0.31662471668612663,
+      "grad_norm": 487.68292236328125,
+      "learning_rate": 3.566659409136468e-06,
+      "loss": 56.2001,
+      "step": 78370
+    },
+    {
+      "epoch": 0.31666511795149427,
+      "grad_norm": 441.1833190917969,
+      "learning_rate": 3.565308292277151e-06,
+      "loss": 49.3039,
+      "step": 78380
+    },
+    {
+      "epoch": 0.31670551921686185,
+      "grad_norm": 497.8100891113281,
+      "learning_rate": 3.563957289569669e-06,
+      "loss": 38.5456,
+      "step": 78390
+    },
+    {
+      "epoch": 0.3167459204822295,
+      "grad_norm": 685.7149047851562,
+      "learning_rate": 3.5626064011215135e-06,
+      "loss": 47.0403,
+      "step": 78400
+    },
+    {
+      "epoch": 0.31678632174759713,
+      "grad_norm": 540.7225341796875,
+      "learning_rate": 3.5612556270401733e-06,
+      "loss": 44.1251,
+      "step": 78410
+    },
+    {
+      "epoch": 0.3168267230129648,
+      "grad_norm": 810.5453491210938,
+      "learning_rate": 3.5599049674331175e-06,
+      "loss": 44.7801,
+      "step": 78420
+    },
+    {
+      "epoch": 0.3168671242783324,
+      "grad_norm": 172.29164123535156,
+      "learning_rate": 3.5585544224078143e-06,
+      "loss": 38.7469,
+      "step": 78430
+    },
+    {
+      "epoch": 0.31690752554370005,
+      "grad_norm": 433.4833984375,
+      "learning_rate": 3.5572039920717192e-06,
+      "loss": 45.1474,
+      "step": 78440
+    },
+    {
+      "epoch": 0.31694792680906764,
+      "grad_norm": 456.20703125,
+      "learning_rate": 3.5558536765322825e-06,
+      "loss": 56.6234,
+      "step": 78450
+    },
+    {
+      "epoch": 0.3169883280744353,
+      "grad_norm": 547.5340576171875,
+      "learning_rate": 3.554503475896941e-06,
+      "loss": 49.693,
+      "step": 78460
+    },
+    {
+      "epoch": 0.3170287293398029,
+      "grad_norm": 616.480712890625,
+      "learning_rate": 3.553153390273124e-06,
+      "loss": 46.7905,
+      "step": 78470
+    },
+    {
+      "epoch": 0.31706913060517056,
+      "grad_norm": 618.6358642578125,
+      "learning_rate": 3.551803419768251e-06,
+      "loss": 41.6367,
+      "step": 78480
+    },
+    {
+      "epoch": 0.3171095318705382,
+      "grad_norm": 359.9627685546875,
+      "learning_rate": 3.5504535644897352e-06,
+      "loss": 34.8955,
+      "step": 78490
+    },
+    {
+      "epoch": 0.31714993313590584,
+      "grad_norm": 928.6984252929688,
+      "learning_rate": 3.549103824544975e-06,
+      "loss": 39.994,
+      "step": 78500
+    },
+    {
+      "epoch": 0.3171903344012734,
+      "grad_norm": 594.8177490234375,
+      "learning_rate": 3.5477542000413657e-06,
+      "loss": 64.4823,
+      "step": 78510
+    },
+    {
+      "epoch": 0.31723073566664106,
+      "grad_norm": 427.8562927246094,
+      "learning_rate": 3.546404691086289e-06,
+      "loss": 34.0224,
+      "step": 78520
+    },
+    {
+      "epoch": 0.3172711369320087,
+      "grad_norm": 298.39031982421875,
+      "learning_rate": 3.5450552977871207e-06,
+      "loss": 46.3302,
+      "step": 78530
+    },
+    {
+      "epoch": 0.31731153819737634,
+      "grad_norm": 604.6382446289062,
+      "learning_rate": 3.543706020251223e-06,
+      "loss": 32.3792,
+      "step": 78540
+    },
+    {
+      "epoch": 0.317351939462744,
+      "grad_norm": 1150.4559326171875,
+      "learning_rate": 3.542356858585956e-06,
+      "loss": 36.3491,
+      "step": 78550
+    },
+    {
+      "epoch": 0.3173923407281116,
+      "grad_norm": 1116.10888671875,
+      "learning_rate": 3.541007812898663e-06,
+      "loss": 37.8419,
+      "step": 78560
+    },
+    {
+      "epoch": 0.31743274199347926,
+      "grad_norm": 477.97076416015625,
+      "learning_rate": 3.539658883296683e-06,
+      "loss": 31.0409,
+      "step": 78570
+    },
+    {
+      "epoch": 0.31747314325884685,
+      "grad_norm": 743.943603515625,
+      "learning_rate": 3.5383100698873446e-06,
+      "loss": 49.6261,
+      "step": 78580
+    },
+    {
+      "epoch": 0.3175135445242145,
+      "grad_norm": 767.6546020507812,
+      "learning_rate": 3.536961372777965e-06,
+      "loss": 41.5514,
+      "step": 78590
+    },
+    {
+      "epoch": 0.3175539457895821,
+      "grad_norm": 705.5023193359375,
+      "learning_rate": 3.535612792075854e-06,
+      "loss": 32.2963,
+      "step": 78600
+    },
+    {
+      "epoch": 0.31759434705494977,
+      "grad_norm": 384.8016662597656,
+      "learning_rate": 3.5342643278883127e-06,
+      "loss": 29.7352,
+      "step": 78610
+    },
+    {
+      "epoch": 0.3176347483203174,
+      "grad_norm": 411.0057678222656,
+      "learning_rate": 3.532915980322632e-06,
+      "loss": 46.1627,
+      "step": 78620
+    },
+    {
+      "epoch": 0.31767514958568505,
+      "grad_norm": 466.7144775390625,
+      "learning_rate": 3.5315677494860923e-06,
+      "loss": 44.4728,
+      "step": 78630
+    },
+    {
+      "epoch": 0.31771555085105263,
+      "grad_norm": 560.2733154296875,
+      "learning_rate": 3.5302196354859693e-06,
+      "loss": 38.7301,
+      "step": 78640
+    },
+    {
+      "epoch": 0.31775595211642027,
+      "grad_norm": 541.011474609375,
+      "learning_rate": 3.528871638429524e-06,
+      "loss": 61.4878,
+      "step": 78650
+    },
+    {
+      "epoch": 0.3177963533817879,
+      "grad_norm": 497.6824645996094,
+      "learning_rate": 3.527523758424013e-06,
+      "loss": 32.1103,
+      "step": 78660
+    },
+    {
+      "epoch": 0.31783675464715555,
+      "grad_norm": 418.5374755859375,
+      "learning_rate": 3.526175995576676e-06,
+      "loss": 42.6755,
+      "step": 78670
+    },
+    {
+      "epoch": 0.3178771559125232,
+      "grad_norm": 686.6862182617188,
+      "learning_rate": 3.524828349994752e-06,
+      "loss": 63.9243,
+      "step": 78680
+    },
+    {
+      "epoch": 0.31791755717789083,
+      "grad_norm": 908.7049560546875,
+      "learning_rate": 3.523480821785466e-06,
+      "loss": 66.8907,
+      "step": 78690
+    },
+    {
+      "epoch": 0.31795795844325847,
+      "grad_norm": 904.7369384765625,
+      "learning_rate": 3.5221334110560345e-06,
+      "loss": 37.5363,
+      "step": 78700
+    },
+    {
+      "epoch": 0.31799835970862605,
+      "grad_norm": 797.6988525390625,
+      "learning_rate": 3.5207861179136654e-06,
+      "loss": 35.7445,
+      "step": 78710
+    },
+    {
+      "epoch": 0.3180387609739937,
+      "grad_norm": 441.76275634765625,
+      "learning_rate": 3.519438942465556e-06,
+      "loss": 47.8045,
+      "step": 78720
+    },
+    {
+      "epoch": 0.31807916223936133,
+      "grad_norm": 709.9418334960938,
+      "learning_rate": 3.5180918848188937e-06,
+      "loss": 62.3529,
+      "step": 78730
+    },
+    {
+      "epoch": 0.318119563504729,
+      "grad_norm": 540.959228515625,
+      "learning_rate": 3.516744945080861e-06,
+      "loss": 40.3376,
+      "step": 78740
+    },
+    {
+      "epoch": 0.3181599647700966,
+      "grad_norm": 539.8733520507812,
+      "learning_rate": 3.5153981233586277e-06,
+      "loss": 57.6363,
+      "step": 78750
+    },
+    {
+      "epoch": 0.31820036603546425,
+      "grad_norm": 602.8059692382812,
+      "learning_rate": 3.5140514197593494e-06,
+      "loss": 41.2759,
+      "step": 78760
+    },
+    {
+      "epoch": 0.31824076730083184,
+      "grad_norm": 464.77081298828125,
+      "learning_rate": 3.512704834390179e-06,
+      "loss": 44.2819,
+      "step": 78770
+    },
+    {
+      "epoch": 0.3182811685661995,
+      "grad_norm": 691.4537963867188,
+      "learning_rate": 3.5113583673582613e-06,
+      "loss": 49.4957,
+      "step": 78780
+    },
+    {
+      "epoch": 0.3183215698315671,
+      "grad_norm": 294.8431701660156,
+      "learning_rate": 3.510012018770726e-06,
+      "loss": 62.4848,
+      "step": 78790
+    },
+    {
+      "epoch": 0.31836197109693476,
+      "grad_norm": 987.3035888671875,
+      "learning_rate": 3.508665788734696e-06,
+      "loss": 49.3034,
+      "step": 78800
+    },
+    {
+      "epoch": 0.3184023723623024,
+      "grad_norm": 410.7585144042969,
+      "learning_rate": 3.507319677357285e-06,
+      "loss": 34.8022,
+      "step": 78810
+    },
+    {
+      "epoch": 0.31844277362767004,
+      "grad_norm": 519.374755859375,
+      "learning_rate": 3.5059736847455967e-06,
+      "loss": 37.6209,
+      "step": 78820
+    },
+    {
+      "epoch": 0.3184831748930376,
+      "grad_norm": 925.8695068359375,
+      "learning_rate": 3.5046278110067242e-06,
+      "loss": 33.3094,
+      "step": 78830
+    },
+    {
+      "epoch": 0.31852357615840526,
+      "grad_norm": 1023.0116577148438,
+      "learning_rate": 3.5032820562477577e-06,
+      "loss": 59.342,
+      "step": 78840
+    },
+    {
+      "epoch": 0.3185639774237729,
+      "grad_norm": 725.557861328125,
+      "learning_rate": 3.5019364205757667e-06,
+      "loss": 38.1884,
+      "step": 78850
+    },
+    {
+      "epoch": 0.31860437868914054,
+      "grad_norm": 432.7353820800781,
+      "learning_rate": 3.5005909040978188e-06,
+      "loss": 47.6009,
+      "step": 78860
+    },
+    {
+      "epoch": 0.3186447799545082,
+      "grad_norm": 505.9002380371094,
+      "learning_rate": 3.4992455069209717e-06,
+      "loss": 52.5041,
+      "step": 78870
+    },
+    {
+      "epoch": 0.3186851812198758,
+      "grad_norm": 526.93701171875,
+      "learning_rate": 3.4979002291522723e-06,
+      "loss": 54.7521,
+      "step": 78880
+    },
+    {
+      "epoch": 0.31872558248524346,
+      "grad_norm": 586.6494750976562,
+      "learning_rate": 3.4965550708987583e-06,
+      "loss": 43.5702,
+      "step": 78890
+    },
+    {
+      "epoch": 0.31876598375061105,
+      "grad_norm": 226.22586059570312,
+      "learning_rate": 3.4952100322674574e-06,
+      "loss": 41.4644,
+      "step": 78900
+    },
+    {
+      "epoch": 0.3188063850159787,
+      "grad_norm": 1272.3206787109375,
+      "learning_rate": 3.4938651133653877e-06,
+      "loss": 45.8692,
+      "step": 78910
+    },
+    {
+      "epoch": 0.31884678628134633,
+      "grad_norm": 577.4758911132812,
+      "learning_rate": 3.49252031429956e-06,
+      "loss": 41.5907,
+      "step": 78920
+    },
+    {
+      "epoch": 0.31888718754671397,
+      "grad_norm": 622.424560546875,
+      "learning_rate": 3.4911756351769722e-06,
+      "loss": 39.5042,
+      "step": 78930
+    },
+    {
+      "epoch": 0.3189275888120816,
+      "grad_norm": 486.1512451171875,
+      "learning_rate": 3.4898310761046133e-06,
+      "loss": 43.4679,
+      "step": 78940
+    },
+    {
+      "epoch": 0.31896799007744925,
+      "grad_norm": 465.1907958984375,
+      "learning_rate": 3.4884866371894654e-06,
+      "loss": 34.0905,
+      "step": 78950
+    },
+    {
+      "epoch": 0.31900839134281683,
+      "grad_norm": 488.0005798339844,
+      "learning_rate": 3.487142318538498e-06,
+      "loss": 42.7252,
+      "step": 78960
+    },
+    {
+      "epoch": 0.3190487926081845,
+      "grad_norm": 430.1434020996094,
+      "learning_rate": 3.4857981202586742e-06,
+      "loss": 35.6475,
+      "step": 78970
+    },
+    {
+      "epoch": 0.3190891938735521,
+      "grad_norm": 789.178955078125,
+      "learning_rate": 3.4844540424569453e-06,
+      "loss": 56.79,
+      "step": 78980
+    },
+    {
+      "epoch": 0.31912959513891975,
+      "grad_norm": 1318.2086181640625,
+      "learning_rate": 3.483110085240252e-06,
+      "loss": 51.1702,
+      "step": 78990
+    },
+    {
+      "epoch": 0.3191699964042874,
+      "grad_norm": 588.171142578125,
+      "learning_rate": 3.481766248715528e-06,
+      "loss": 37.9113,
+      "step": 79000
+    },
+    {
+      "epoch": 0.31921039766965503,
+      "grad_norm": 609.7747802734375,
+      "learning_rate": 3.4804225329896963e-06,
+      "loss": 25.8852,
+      "step": 79010
+    },
+    {
+      "epoch": 0.3192507989350226,
+      "grad_norm": 442.40960693359375,
+      "learning_rate": 3.4790789381696686e-06,
+      "loss": 54.1903,
+      "step": 79020
+    },
+    {
+      "epoch": 0.31929120020039026,
+      "grad_norm": 548.3629150390625,
+      "learning_rate": 3.4777354643623506e-06,
+      "loss": 39.2755,
+      "step": 79030
+    },
+    {
+      "epoch": 0.3193316014657579,
+      "grad_norm": 675.93603515625,
+      "learning_rate": 3.4763921116746352e-06,
+      "loss": 60.0762,
+      "step": 79040
+    },
+    {
+      "epoch": 0.31937200273112554,
+      "grad_norm": 426.993408203125,
+      "learning_rate": 3.475048880213407e-06,
+      "loss": 53.1464,
+      "step": 79050
+    },
+    {
+      "epoch": 0.3194124039964932,
+      "grad_norm": 585.4061279296875,
+      "learning_rate": 3.473705770085539e-06,
+      "loss": 29.5674,
+      "step": 79060
+    },
+    {
+      "epoch": 0.3194528052618608,
+      "grad_norm": 322.57574462890625,
+      "learning_rate": 3.4723627813979005e-06,
+      "loss": 41.8893,
+      "step": 79070
+    },
+    {
+      "epoch": 0.31949320652722846,
+      "grad_norm": 962.873046875,
+      "learning_rate": 3.471019914257344e-06,
+      "loss": 50.1003,
+      "step": 79080
+    },
+    {
+      "epoch": 0.31953360779259604,
+      "grad_norm": 496.5466613769531,
+      "learning_rate": 3.4696771687707176e-06,
+      "loss": 34.1614,
+      "step": 79090
+    },
+    {
+      "epoch": 0.3195740090579637,
+      "grad_norm": 351.3705749511719,
+      "learning_rate": 3.468334545044853e-06,
+      "loss": 36.1759,
+      "step": 79100
+    },
+    {
+      "epoch": 0.3196144103233313,
+      "grad_norm": 515.8349609375,
+      "learning_rate": 3.46699204318658e-06,
+      "loss": 57.3094,
+      "step": 79110
+    },
+    {
+      "epoch": 0.31965481158869896,
+      "grad_norm": 215.4729766845703,
+      "learning_rate": 3.465649663302715e-06,
+      "loss": 20.6461,
+      "step": 79120
+    },
+    {
+      "epoch": 0.3196952128540666,
+      "grad_norm": 395.2944641113281,
+      "learning_rate": 3.464307405500064e-06,
+      "loss": 45.1714,
+      "step": 79130
+    },
+    {
+      "epoch": 0.31973561411943424,
+      "grad_norm": 424.1663818359375,
+      "learning_rate": 3.4629652698854254e-06,
+      "loss": 43.4122,
+      "step": 79140
+    },
+    {
+      "epoch": 0.3197760153848018,
+      "grad_norm": 641.1630859375,
+      "learning_rate": 3.461623256565586e-06,
+      "loss": 45.5314,
+      "step": 79150
+    },
+    {
+      "epoch": 0.31981641665016947,
+      "grad_norm": 721.894775390625,
+      "learning_rate": 3.4602813656473223e-06,
+      "loss": 48.3435,
+      "step": 79160
+    },
+    {
+      "epoch": 0.3198568179155371,
+      "grad_norm": 1735.0408935546875,
+      "learning_rate": 3.4589395972374055e-06,
+      "loss": 65.0877,
+      "step": 79170
+    },
+    {
+      "epoch": 0.31989721918090475,
+      "grad_norm": 844.60205078125,
+      "learning_rate": 3.457597951442595e-06,
+      "loss": 28.3744,
+      "step": 79180
+    },
+    {
+      "epoch": 0.3199376204462724,
+      "grad_norm": 402.255859375,
+      "learning_rate": 3.456256428369633e-06,
+      "loss": 39.7704,
+      "step": 79190
+    },
+    {
+      "epoch": 0.31997802171164,
+      "grad_norm": 719.4677124023438,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 60.1323,
+      "step": 79200
+    },
+    {
+      "epoch": 0.32001842297700767,
+      "grad_norm": 797.6649780273438,
+      "learning_rate": 3.453573750816214e-06,
+      "loss": 50.052,
+      "step": 79210
+    },
+    {
+      "epoch": 0.32005882424237525,
+      "grad_norm": 1009.09716796875,
+      "learning_rate": 3.452232596549204e-06,
+      "loss": 51.3804,
+      "step": 79220
+    },
+    {
+      "epoch": 0.3200992255077429,
+      "grad_norm": 684.1012573242188,
+      "learning_rate": 3.4508915654309438e-06,
+      "loss": 42.5982,
+      "step": 79230
+    },
+    {
+      "epoch": 0.32013962677311053,
+      "grad_norm": 571.1311645507812,
+      "learning_rate": 3.4495506575681313e-06,
+      "loss": 39.391,
+      "step": 79240
+    },
+    {
+      "epoch": 0.32018002803847817,
+      "grad_norm": 363.0473937988281,
+      "learning_rate": 3.4482098730674577e-06,
+      "loss": 39.8878,
+      "step": 79250
+    },
+    {
+      "epoch": 0.3202204293038458,
+      "grad_norm": 476.1780700683594,
+      "learning_rate": 3.4468692120356017e-06,
+      "loss": 42.8986,
+      "step": 79260
+    },
+    {
+      "epoch": 0.32026083056921345,
+      "grad_norm": 883.4768676757812,
+      "learning_rate": 3.4455286745792383e-06,
+      "loss": 53.311,
+      "step": 79270
+    },
+    {
+      "epoch": 0.32030123183458103,
+      "grad_norm": 594.4937133789062,
+      "learning_rate": 3.4441882608050216e-06,
+      "loss": 35.4785,
+      "step": 79280
+    },
+    {
+      "epoch": 0.3203416330999487,
+      "grad_norm": 534.9506225585938,
+      "learning_rate": 3.442847970819604e-06,
+      "loss": 42.1486,
+      "step": 79290
+    },
+    {
+      "epoch": 0.3203820343653163,
+      "grad_norm": 3948.400634765625,
+      "learning_rate": 3.441507804729627e-06,
+      "loss": 73.9457,
+      "step": 79300
+    },
+    {
+      "epoch": 0.32042243563068395,
+      "grad_norm": 287.04266357421875,
+      "learning_rate": 3.440167762641722e-06,
+      "loss": 64.0647,
+      "step": 79310
+    },
+    {
+      "epoch": 0.3204628368960516,
+      "grad_norm": 1426.3807373046875,
+      "learning_rate": 3.43882784466251e-06,
+      "loss": 37.0213,
+      "step": 79320
+    },
+    {
+      "epoch": 0.32050323816141923,
+      "grad_norm": 613.2059936523438,
+      "learning_rate": 3.4374880508986013e-06,
+      "loss": 38.8521,
+      "step": 79330
+    },
+    {
+      "epoch": 0.3205436394267868,
+      "grad_norm": 494.06988525390625,
+      "learning_rate": 3.436148381456598e-06,
+      "loss": 44.2171,
+      "step": 79340
+    },
+    {
+      "epoch": 0.32058404069215446,
+      "grad_norm": 764.4878540039062,
+      "learning_rate": 3.434808836443091e-06,
+      "loss": 58.2755,
+      "step": 79350
+    },
+    {
+      "epoch": 0.3206244419575221,
+      "grad_norm": 1218.543212890625,
+      "learning_rate": 3.4334694159646608e-06,
+      "loss": 40.1468,
+      "step": 79360
+    },
+    {
+      "epoch": 0.32066484322288974,
+      "grad_norm": 611.8609008789062,
+      "learning_rate": 3.43213012012788e-06,
+      "loss": 35.2115,
+      "step": 79370
+    },
+    {
+      "epoch": 0.3207052444882574,
+      "grad_norm": 702.8851928710938,
+      "learning_rate": 3.43079094903931e-06,
+      "loss": 41.0661,
+      "step": 79380
+    },
+    {
+      "epoch": 0.320745645753625,
+      "grad_norm": 959.5906982421875,
+      "learning_rate": 3.4294519028055014e-06,
+      "loss": 51.1332,
+      "step": 79390
+    },
+    {
+      "epoch": 0.32078604701899266,
+      "grad_norm": 552.6378784179688,
+      "learning_rate": 3.428112981532998e-06,
+      "loss": 63.2162,
+      "step": 79400
+    },
+    {
+      "epoch": 0.32082644828436024,
+      "grad_norm": 819.8400268554688,
+      "learning_rate": 3.4267741853283305e-06,
+      "loss": 41.3017,
+      "step": 79410
+    },
+    {
+      "epoch": 0.3208668495497279,
+      "grad_norm": 797.3480834960938,
+      "learning_rate": 3.425435514298021e-06,
+      "loss": 70.3864,
+      "step": 79420
+    },
+    {
+      "epoch": 0.3209072508150955,
+      "grad_norm": 281.98406982421875,
+      "learning_rate": 3.4240969685485813e-06,
+      "loss": 45.3279,
+      "step": 79430
+    },
+    {
+      "epoch": 0.32094765208046316,
+      "grad_norm": 497.9682922363281,
+      "learning_rate": 3.422758548186515e-06,
+      "loss": 49.0692,
+      "step": 79440
+    },
+    {
+      "epoch": 0.3209880533458308,
+      "grad_norm": 532.9519653320312,
+      "learning_rate": 3.4214202533183104e-06,
+      "loss": 47.7864,
+      "step": 79450
+    },
+    {
+      "epoch": 0.32102845461119844,
+      "grad_norm": 327.18487548828125,
+      "learning_rate": 3.420082084050453e-06,
+      "loss": 62.1244,
+      "step": 79460
+    },
+    {
+      "epoch": 0.32106885587656603,
+      "grad_norm": 522.5250244140625,
+      "learning_rate": 3.4187440404894123e-06,
+      "loss": 33.9887,
+      "step": 79470
+    },
+    {
+      "epoch": 0.32110925714193367,
+      "grad_norm": 465.98486328125,
+      "learning_rate": 3.417406122741651e-06,
+      "loss": 50.0454,
+      "step": 79480
+    },
+    {
+      "epoch": 0.3211496584073013,
+      "grad_norm": 641.7701416015625,
+      "learning_rate": 3.416068330913621e-06,
+      "loss": 45.9428,
+      "step": 79490
+    },
+    {
+      "epoch": 0.32119005967266895,
+      "grad_norm": 220.279296875,
+      "learning_rate": 3.4147306651117663e-06,
+      "loss": 37.3395,
+      "step": 79500
+    },
+    {
+      "epoch": 0.3212304609380366,
+      "grad_norm": 474.749267578125,
+      "learning_rate": 3.4133931254425156e-06,
+      "loss": 45.0709,
+      "step": 79510
+    },
+    {
+      "epoch": 0.32127086220340423,
+      "grad_norm": 253.58042907714844,
+      "learning_rate": 3.4120557120122944e-06,
+      "loss": 44.7656,
+      "step": 79520
+    },
+    {
+      "epoch": 0.32131126346877187,
+      "grad_norm": 597.0999145507812,
+      "learning_rate": 3.4107184249275114e-06,
+      "loss": 52.3735,
+      "step": 79530
+    },
+    {
+      "epoch": 0.32135166473413945,
+      "grad_norm": 484.10125732421875,
+      "learning_rate": 3.4093812642945694e-06,
+      "loss": 39.8811,
+      "step": 79540
+    },
+    {
+      "epoch": 0.3213920659995071,
+      "grad_norm": 523.0706176757812,
+      "learning_rate": 3.40804423021986e-06,
+      "loss": 48.5399,
+      "step": 79550
+    },
+    {
+      "epoch": 0.32143246726487473,
+      "grad_norm": 797.7967529296875,
+      "learning_rate": 3.4067073228097656e-06,
+      "loss": 36.3043,
+      "step": 79560
+    },
+    {
+      "epoch": 0.3214728685302424,
+      "grad_norm": 534.5451049804688,
+      "learning_rate": 3.4053705421706574e-06,
+      "loss": 36.7537,
+      "step": 79570
+    },
+    {
+      "epoch": 0.32151326979561,
+      "grad_norm": 664.1915893554688,
+      "learning_rate": 3.4040338884088955e-06,
+      "loss": 44.3147,
+      "step": 79580
+    },
+    {
+      "epoch": 0.32155367106097765,
+      "grad_norm": 295.6415710449219,
+      "learning_rate": 3.4026973616308334e-06,
+      "loss": 40.6299,
+      "step": 79590
+    },
+    {
+      "epoch": 0.32159407232634524,
+      "grad_norm": 624.4505004882812,
+      "learning_rate": 3.401360961942812e-06,
+      "loss": 53.7231,
+      "step": 79600
+    },
+    {
+      "epoch": 0.3216344735917129,
+      "grad_norm": 735.50244140625,
+      "learning_rate": 3.4000246894511634e-06,
+      "loss": 32.1379,
+      "step": 79610
+    },
+    {
+      "epoch": 0.3216748748570805,
+      "grad_norm": 268.3006896972656,
+      "learning_rate": 3.398688544262205e-06,
+      "loss": 30.7915,
+      "step": 79620
+    },
+    {
+      "epoch": 0.32171527612244816,
+      "grad_norm": 868.1389770507812,
+      "learning_rate": 3.397352526482251e-06,
+      "loss": 39.305,
+      "step": 79630
+    },
+    {
+      "epoch": 0.3217556773878158,
+      "grad_norm": 487.7118225097656,
+      "learning_rate": 3.396016636217601e-06,
+      "loss": 40.3482,
+      "step": 79640
+    },
+    {
+      "epoch": 0.32179607865318344,
+      "grad_norm": 551.2245483398438,
+      "learning_rate": 3.394680873574546e-06,
+      "loss": 46.2264,
+      "step": 79650
+    },
+    {
+      "epoch": 0.321836479918551,
+      "grad_norm": 634.7960205078125,
+      "learning_rate": 3.3933452386593666e-06,
+      "loss": 42.2122,
+      "step": 79660
+    },
+    {
+      "epoch": 0.32187688118391866,
+      "grad_norm": 272.61962890625,
+      "learning_rate": 3.392009731578334e-06,
+      "loss": 28.0738,
+      "step": 79670
+    },
+    {
+      "epoch": 0.3219172824492863,
+      "grad_norm": 273.2590026855469,
+      "learning_rate": 3.3906743524377053e-06,
+      "loss": 33.9681,
+      "step": 79680
+    },
+    {
+      "epoch": 0.32195768371465394,
+      "grad_norm": 400.4376220703125,
+      "learning_rate": 3.3893391013437338e-06,
+      "loss": 53.93,
+      "step": 79690
+    },
+    {
+      "epoch": 0.3219980849800216,
+      "grad_norm": 381.15850830078125,
+      "learning_rate": 3.38800397840266e-06,
+      "loss": 38.3037,
+      "step": 79700
+    },
+    {
+      "epoch": 0.3220384862453892,
+      "grad_norm": 668.4859008789062,
+      "learning_rate": 3.3866689837207094e-06,
+      "loss": 43.7831,
+      "step": 79710
+    },
+    {
+      "epoch": 0.32207888751075686,
+      "grad_norm": 599.7156372070312,
+      "learning_rate": 3.3853341174041025e-06,
+      "loss": 36.1802,
+      "step": 79720
+    },
+    {
+      "epoch": 0.32211928877612445,
+      "grad_norm": 673.56005859375,
+      "learning_rate": 3.3839993795590507e-06,
+      "loss": 33.4147,
+      "step": 79730
+    },
+    {
+      "epoch": 0.3221596900414921,
+      "grad_norm": 613.9022216796875,
+      "learning_rate": 3.3826647702917526e-06,
+      "loss": 51.1556,
+      "step": 79740
+    },
+    {
+      "epoch": 0.3222000913068597,
+      "grad_norm": 726.7110595703125,
+      "learning_rate": 3.3813302897083955e-06,
+      "loss": 31.6607,
+      "step": 79750
+    },
+    {
+      "epoch": 0.32224049257222737,
+      "grad_norm": 848.6845092773438,
+      "learning_rate": 3.379995937915158e-06,
+      "loss": 45.0145,
+      "step": 79760
+    },
+    {
+      "epoch": 0.322280893837595,
+      "grad_norm": 804.3870849609375,
+      "learning_rate": 3.37866171501821e-06,
+      "loss": 35.8172,
+      "step": 79770
+    },
+    {
+      "epoch": 0.32232129510296265,
+      "grad_norm": 282.2019958496094,
+      "learning_rate": 3.3773276211237087e-06,
+      "loss": 38.8783,
+      "step": 79780
+    },
+    {
+      "epoch": 0.32236169636833023,
+      "grad_norm": 501.5666198730469,
+      "learning_rate": 3.3759936563378004e-06,
+      "loss": 37.0371,
+      "step": 79790
+    },
+    {
+      "epoch": 0.32240209763369787,
+      "grad_norm": 299.2339782714844,
+      "learning_rate": 3.374659820766625e-06,
+      "loss": 35.9554,
+      "step": 79800
+    },
+    {
+      "epoch": 0.3224424988990655,
+      "grad_norm": 376.84515380859375,
+      "learning_rate": 3.3733261145163064e-06,
+      "loss": 68.8255,
+      "step": 79810
+    },
+    {
+      "epoch": 0.32248290016443315,
+      "grad_norm": 342.44073486328125,
+      "learning_rate": 3.371992537692964e-06,
+      "loss": 30.2029,
+      "step": 79820
+    },
+    {
+      "epoch": 0.3225233014298008,
+      "grad_norm": 251.28659057617188,
+      "learning_rate": 3.370659090402704e-06,
+      "loss": 31.2192,
+      "step": 79830
+    },
+    {
+      "epoch": 0.32256370269516843,
+      "grad_norm": 563.5650024414062,
+      "learning_rate": 3.3693257727516227e-06,
+      "loss": 40.0149,
+      "step": 79840
+    },
+    {
+      "epoch": 0.32260410396053607,
+      "grad_norm": 611.80419921875,
+      "learning_rate": 3.367992584845806e-06,
+      "loss": 31.1052,
+      "step": 79850
+    },
+    {
+      "epoch": 0.32264450522590365,
+      "grad_norm": 707.4664916992188,
+      "learning_rate": 3.3666595267913293e-06,
+      "loss": 50.2898,
+      "step": 79860
+    },
+    {
+      "epoch": 0.3226849064912713,
+      "grad_norm": 855.7467041015625,
+      "learning_rate": 3.365326598694259e-06,
+      "loss": 38.313,
+      "step": 79870
+    },
+    {
+      "epoch": 0.32272530775663893,
+      "grad_norm": 1066.11865234375,
+      "learning_rate": 3.3639938006606483e-06,
+      "loss": 57.3661,
+      "step": 79880
+    },
+    {
+      "epoch": 0.3227657090220066,
+      "grad_norm": 662.3603515625,
+      "learning_rate": 3.3626611327965418e-06,
+      "loss": 55.8432,
+      "step": 79890
+    },
+    {
+      "epoch": 0.3228061102873742,
+      "grad_norm": 557.5881958007812,
+      "learning_rate": 3.3613285952079754e-06,
+      "loss": 31.4979,
+      "step": 79900
+    },
+    {
+      "epoch": 0.32284651155274185,
+      "grad_norm": 336.6396789550781,
+      "learning_rate": 3.3599961880009713e-06,
+      "loss": 39.1249,
+      "step": 79910
+    },
+    {
+      "epoch": 0.32288691281810944,
+      "grad_norm": 611.0641479492188,
+      "learning_rate": 3.3586639112815446e-06,
+      "loss": 47.8904,
+      "step": 79920
+    },
+    {
+      "epoch": 0.3229273140834771,
+      "grad_norm": 436.7526550292969,
+      "learning_rate": 3.357331765155698e-06,
+      "loss": 38.4208,
+      "step": 79930
+    },
+    {
+      "epoch": 0.3229677153488447,
+      "grad_norm": 791.1549072265625,
+      "learning_rate": 3.355999749729424e-06,
+      "loss": 54.1118,
+      "step": 79940
+    },
+    {
+      "epoch": 0.32300811661421236,
+      "grad_norm": 308.76812744140625,
+      "learning_rate": 3.354667865108706e-06,
+      "loss": 45.5083,
+      "step": 79950
+    },
+    {
+      "epoch": 0.32304851787958,
+      "grad_norm": 392.00079345703125,
+      "learning_rate": 3.353336111399513e-06,
+      "loss": 41.9508,
+      "step": 79960
+    },
+    {
+      "epoch": 0.32308891914494764,
+      "grad_norm": 682.669921875,
+      "learning_rate": 3.3520044887078096e-06,
+      "loss": 56.1205,
+      "step": 79970
+    },
+    {
+      "epoch": 0.3231293204103152,
+      "grad_norm": 477.31451416015625,
+      "learning_rate": 3.350672997139546e-06,
+      "loss": 37.7137,
+      "step": 79980
+    },
+    {
+      "epoch": 0.32316972167568286,
+      "grad_norm": 604.2123413085938,
+      "learning_rate": 3.3493416368006614e-06,
+      "loss": 45.0942,
+      "step": 79990
+    },
+    {
+      "epoch": 0.3232101229410505,
+      "grad_norm": 340.8969421386719,
+      "learning_rate": 3.348010407797088e-06,
+      "loss": 27.6048,
+      "step": 80000
+    },
+    {
+      "epoch": 0.32325052420641814,
+      "grad_norm": 520.0179443359375,
+      "learning_rate": 3.346679310234744e-06,
+      "loss": 55.725,
+      "step": 80010
+    },
+    {
+      "epoch": 0.3232909254717858,
+      "grad_norm": 773.4434204101562,
+      "learning_rate": 3.34534834421954e-06,
+      "loss": 28.6225,
+      "step": 80020
+    },
+    {
+      "epoch": 0.3233313267371534,
+      "grad_norm": 632.611328125,
+      "learning_rate": 3.3440175098573748e-06,
+      "loss": 39.8326,
+      "step": 80030
+    },
+    {
+      "epoch": 0.32337172800252106,
+      "grad_norm": 1086.8714599609375,
+      "learning_rate": 3.3426868072541386e-06,
+      "loss": 44.8706,
+      "step": 80040
+    },
+    {
+      "epoch": 0.32341212926788865,
+      "grad_norm": 742.6467895507812,
+      "learning_rate": 3.3413562365157037e-06,
+      "loss": 51.3557,
+      "step": 80050
+    },
+    {
+      "epoch": 0.3234525305332563,
+      "grad_norm": 493.33428955078125,
+      "learning_rate": 3.340025797747942e-06,
+      "loss": 29.4505,
+      "step": 80060
+    },
+    {
+      "epoch": 0.32349293179862393,
+      "grad_norm": 574.6460571289062,
+      "learning_rate": 3.3386954910567094e-06,
+      "loss": 48.0383,
+      "step": 80070
+    },
+    {
+      "epoch": 0.32353333306399157,
+      "grad_norm": 314.26397705078125,
+      "learning_rate": 3.337365316547852e-06,
+      "loss": 35.6812,
+      "step": 80080
+    },
+    {
+      "epoch": 0.3235737343293592,
+      "grad_norm": 521.7124633789062,
+      "learning_rate": 3.336035274327206e-06,
+      "loss": 27.6825,
+      "step": 80090
+    },
+    {
+      "epoch": 0.32361413559472685,
+      "grad_norm": 745.5689697265625,
+      "learning_rate": 3.3347053645005965e-06,
+      "loss": 48.5578,
+      "step": 80100
+    },
+    {
+      "epoch": 0.32365453686009443,
+      "grad_norm": 226.4112091064453,
+      "learning_rate": 3.333375587173838e-06,
+      "loss": 41.1653,
+      "step": 80110
+    },
+    {
+      "epoch": 0.3236949381254621,
+      "grad_norm": 737.1196899414062,
+      "learning_rate": 3.332045942452738e-06,
+      "loss": 65.6083,
+      "step": 80120
+    },
+    {
+      "epoch": 0.3237353393908297,
+      "grad_norm": 573.3949584960938,
+      "learning_rate": 3.330716430443085e-06,
+      "loss": 44.0588,
+      "step": 80130
+    },
+    {
+      "epoch": 0.32377574065619735,
+      "grad_norm": 943.7135009765625,
+      "learning_rate": 3.329387051250664e-06,
+      "loss": 62.2634,
+      "step": 80140
+    },
+    {
+      "epoch": 0.323816141921565,
+      "grad_norm": 686.1159057617188,
+      "learning_rate": 3.3280578049812493e-06,
+      "loss": 34.7635,
+      "step": 80150
+    },
+    {
+      "epoch": 0.32385654318693263,
+      "grad_norm": 514.3086547851562,
+      "learning_rate": 3.3267286917406027e-06,
+      "loss": 27.8198,
+      "step": 80160
+    },
+    {
+      "epoch": 0.3238969444523003,
+      "grad_norm": 286.08856201171875,
+      "learning_rate": 3.3253997116344737e-06,
+      "loss": 52.2875,
+      "step": 80170
+    },
+    {
+      "epoch": 0.32393734571766786,
+      "grad_norm": 1013.5391845703125,
+      "learning_rate": 3.3240708647686047e-06,
+      "loss": 43.7235,
+      "step": 80180
+    },
+    {
+      "epoch": 0.3239777469830355,
+      "grad_norm": 311.2264404296875,
+      "learning_rate": 3.322742151248726e-06,
+      "loss": 48.4024,
+      "step": 80190
+    },
+    {
+      "epoch": 0.32401814824840314,
+      "grad_norm": 620.4436645507812,
+      "learning_rate": 3.3214135711805555e-06,
+      "loss": 48.492,
+      "step": 80200
+    },
+    {
+      "epoch": 0.3240585495137708,
+      "grad_norm": 514.2919311523438,
+      "learning_rate": 3.3200851246698053e-06,
+      "loss": 35.2087,
+      "step": 80210
+    },
+    {
+      "epoch": 0.3240989507791384,
+      "grad_norm": 673.8460083007812,
+      "learning_rate": 3.318756811822171e-06,
+      "loss": 53.043,
+      "step": 80220
+    },
+    {
+      "epoch": 0.32413935204450606,
+      "grad_norm": 529.0668334960938,
+      "learning_rate": 3.3174286327433408e-06,
+      "loss": 48.762,
+      "step": 80230
+    },
+    {
+      "epoch": 0.32417975330987364,
+      "grad_norm": 274.8592834472656,
+      "learning_rate": 3.3161005875389916e-06,
+      "loss": 44.827,
+      "step": 80240
+    },
+    {
+      "epoch": 0.3242201545752413,
+      "grad_norm": 572.7136840820312,
+      "learning_rate": 3.3147726763147913e-06,
+      "loss": 39.5248,
+      "step": 80250
+    },
+    {
+      "epoch": 0.3242605558406089,
+      "grad_norm": 772.9658813476562,
+      "learning_rate": 3.3134448991763957e-06,
+      "loss": 39.5004,
+      "step": 80260
+    },
+    {
+      "epoch": 0.32430095710597656,
+      "grad_norm": 361.0927734375,
+      "learning_rate": 3.312117256229449e-06,
+      "loss": 38.6941,
+      "step": 80270
+    },
+    {
+      "epoch": 0.3243413583713442,
+      "grad_norm": 563.4647216796875,
+      "learning_rate": 3.310789747579586e-06,
+      "loss": 73.1953,
+      "step": 80280
+    },
+    {
+      "epoch": 0.32438175963671184,
+      "grad_norm": 383.66326904296875,
+      "learning_rate": 3.30946237333243e-06,
+      "loss": 34.4704,
+      "step": 80290
+    },
+    {
+      "epoch": 0.3244221609020794,
+      "grad_norm": 652.5866088867188,
+      "learning_rate": 3.308135133593595e-06,
+      "loss": 41.3068,
+      "step": 80300
+    },
+    {
+      "epoch": 0.32446256216744707,
+      "grad_norm": 180.12716674804688,
+      "learning_rate": 3.3068080284686825e-06,
+      "loss": 51.6751,
+      "step": 80310
+    },
+    {
+      "epoch": 0.3245029634328147,
+      "grad_norm": 239.9126739501953,
+      "learning_rate": 3.3054810580632844e-06,
+      "loss": 33.6634,
+      "step": 80320
+    },
+    {
+      "epoch": 0.32454336469818235,
+      "grad_norm": 615.7425537109375,
+      "learning_rate": 3.304154222482982e-06,
+      "loss": 41.3266,
+      "step": 80330
+    },
+    {
+      "epoch": 0.32458376596355,
+      "grad_norm": 391.79119873046875,
+      "learning_rate": 3.3028275218333438e-06,
+      "loss": 34.0825,
+      "step": 80340
+    },
+    {
+      "epoch": 0.3246241672289176,
+      "grad_norm": 380.4940490722656,
+      "learning_rate": 3.301500956219932e-06,
+      "loss": 64.4449,
+      "step": 80350
+    },
+    {
+      "epoch": 0.32466456849428527,
+      "grad_norm": 569.7992553710938,
+      "learning_rate": 3.3001745257482935e-06,
+      "loss": 45.7908,
+      "step": 80360
+    },
+    {
+      "epoch": 0.32470496975965285,
+      "grad_norm": 666.3257446289062,
+      "learning_rate": 3.2988482305239673e-06,
+      "loss": 45.82,
+      "step": 80370
+    },
+    {
+      "epoch": 0.3247453710250205,
+      "grad_norm": 590.6626586914062,
+      "learning_rate": 3.2975220706524813e-06,
+      "loss": 38.1958,
+      "step": 80380
+    },
+    {
+      "epoch": 0.32478577229038813,
+      "grad_norm": 621.2801513671875,
+      "learning_rate": 3.2961960462393492e-06,
+      "loss": 37.6412,
+      "step": 80390
+    },
+    {
+      "epoch": 0.32482617355575577,
+      "grad_norm": 708.3571166992188,
+      "learning_rate": 3.2948701573900786e-06,
+      "loss": 43.4666,
+      "step": 80400
+    },
+    {
+      "epoch": 0.3248665748211234,
+      "grad_norm": 481.2866516113281,
+      "learning_rate": 3.2935444042101646e-06,
+      "loss": 26.2208,
+      "step": 80410
+    },
+    {
+      "epoch": 0.32490697608649105,
+      "grad_norm": 521.428955078125,
+      "learning_rate": 3.29221878680509e-06,
+      "loss": 52.5316,
+      "step": 80420
+    },
+    {
+      "epoch": 0.32494737735185864,
+      "grad_norm": 793.7952270507812,
+      "learning_rate": 3.2908933052803292e-06,
+      "loss": 64.951,
+      "step": 80430
+    },
+    {
+      "epoch": 0.3249877786172263,
+      "grad_norm": 466.8500671386719,
+      "learning_rate": 3.2895679597413433e-06,
+      "loss": 28.5023,
+      "step": 80440
+    },
+    {
+      "epoch": 0.3250281798825939,
+      "grad_norm": 404.64739990234375,
+      "learning_rate": 3.2882427502935867e-06,
+      "loss": 37.0899,
+      "step": 80450
+    },
+    {
+      "epoch": 0.32506858114796156,
+      "grad_norm": 475.4322509765625,
+      "learning_rate": 3.2869176770424976e-06,
+      "loss": 41.2995,
+      "step": 80460
+    },
+    {
+      "epoch": 0.3251089824133292,
+      "grad_norm": 679.5100708007812,
+      "learning_rate": 3.2855927400935085e-06,
+      "loss": 36.223,
+      "step": 80470
+    },
+    {
+      "epoch": 0.32514938367869683,
+      "grad_norm": 426.2971496582031,
+      "learning_rate": 3.2842679395520363e-06,
+      "loss": 44.444,
+      "step": 80480
+    },
+    {
+      "epoch": 0.3251897849440645,
+      "grad_norm": 727.0317993164062,
+      "learning_rate": 3.282943275523489e-06,
+      "loss": 46.9828,
+      "step": 80490
+    },
+    {
+      "epoch": 0.32523018620943206,
+      "grad_norm": 522.9331665039062,
+      "learning_rate": 3.2816187481132655e-06,
+      "loss": 47.7081,
+      "step": 80500
+    },
+    {
+      "epoch": 0.3252705874747997,
+      "grad_norm": 675.2789916992188,
+      "learning_rate": 3.280294357426752e-06,
+      "loss": 50.2561,
+      "step": 80510
+    },
+    {
+      "epoch": 0.32531098874016734,
+      "grad_norm": 735.783447265625,
+      "learning_rate": 3.2789701035693242e-06,
+      "loss": 58.5167,
+      "step": 80520
+    },
+    {
+      "epoch": 0.325351390005535,
+      "grad_norm": 601.2733154296875,
+      "learning_rate": 3.277645986646346e-06,
+      "loss": 52.778,
+      "step": 80530
+    },
+    {
+      "epoch": 0.3253917912709026,
+      "grad_norm": 649.8107299804688,
+      "learning_rate": 3.276322006763172e-06,
+      "loss": 44.8131,
+      "step": 80540
+    },
+    {
+      "epoch": 0.32543219253627026,
+      "grad_norm": 731.3230590820312,
+      "learning_rate": 3.274998164025148e-06,
+      "loss": 46.8925,
+      "step": 80550
+    },
+    {
+      "epoch": 0.32547259380163784,
+      "grad_norm": 824.4846801757812,
+      "learning_rate": 3.2736744585376016e-06,
+      "loss": 41.6556,
+      "step": 80560
+    },
+    {
+      "epoch": 0.3255129950670055,
+      "grad_norm": 402.4475402832031,
+      "learning_rate": 3.2723508904058547e-06,
+      "loss": 41.3893,
+      "step": 80570
+    },
+    {
+      "epoch": 0.3255533963323731,
+      "grad_norm": 762.2625732421875,
+      "learning_rate": 3.27102745973522e-06,
+      "loss": 45.3046,
+      "step": 80580
+    },
+    {
+      "epoch": 0.32559379759774076,
+      "grad_norm": 526.9542236328125,
+      "learning_rate": 3.269704166630995e-06,
+      "loss": 30.6245,
+      "step": 80590
+    },
+    {
+      "epoch": 0.3256341988631084,
+      "grad_norm": 1798.8516845703125,
+      "learning_rate": 3.268381011198468e-06,
+      "loss": 55.7041,
+      "step": 80600
+    },
+    {
+      "epoch": 0.32567460012847604,
+      "grad_norm": 701.319091796875,
+      "learning_rate": 3.2670579935429176e-06,
+      "loss": 34.5113,
+      "step": 80610
+    },
+    {
+      "epoch": 0.32571500139384363,
+      "grad_norm": 708.3885498046875,
+      "learning_rate": 3.265735113769609e-06,
+      "loss": 41.173,
+      "step": 80620
+    },
+    {
+      "epoch": 0.32575540265921127,
+      "grad_norm": 384.1140441894531,
+      "learning_rate": 3.264412371983797e-06,
+      "loss": 49.1882,
+      "step": 80630
+    },
+    {
+      "epoch": 0.3257958039245789,
+      "grad_norm": 618.54931640625,
+      "learning_rate": 3.2630897682907312e-06,
+      "loss": 44.0357,
+      "step": 80640
+    },
+    {
+      "epoch": 0.32583620518994655,
+      "grad_norm": 781.7505493164062,
+      "learning_rate": 3.261767302795639e-06,
+      "loss": 45.6742,
+      "step": 80650
+    },
+    {
+      "epoch": 0.3258766064553142,
+      "grad_norm": 602.9747924804688,
+      "learning_rate": 3.2604449756037447e-06,
+      "loss": 33.2267,
+      "step": 80660
+    },
+    {
+      "epoch": 0.32591700772068183,
+      "grad_norm": 307.5460205078125,
+      "learning_rate": 3.2591227868202592e-06,
+      "loss": 40.4157,
+      "step": 80670
+    },
+    {
+      "epoch": 0.32595740898604947,
+      "grad_norm": 867.035400390625,
+      "learning_rate": 3.257800736550385e-06,
+      "loss": 47.2259,
+      "step": 80680
+    },
+    {
+      "epoch": 0.32599781025141705,
+      "grad_norm": 432.8591613769531,
+      "learning_rate": 3.2564788248993105e-06,
+      "loss": 48.1717,
+      "step": 80690
+    },
+    {
+      "epoch": 0.3260382115167847,
+      "grad_norm": 551.8834838867188,
+      "learning_rate": 3.2551570519722155e-06,
+      "loss": 37.013,
+      "step": 80700
+    },
+    {
+      "epoch": 0.32607861278215233,
+      "grad_norm": 527.9805908203125,
+      "learning_rate": 3.2538354178742648e-06,
+      "loss": 53.8853,
+      "step": 80710
+    },
+    {
+      "epoch": 0.32611901404752,
+      "grad_norm": 1692.8126220703125,
+      "learning_rate": 3.2525139227106163e-06,
+      "loss": 59.593,
+      "step": 80720
+    },
+    {
+      "epoch": 0.3261594153128876,
+      "grad_norm": 61.837989807128906,
+      "learning_rate": 3.2511925665864164e-06,
+      "loss": 42.8212,
+      "step": 80730
+    },
+    {
+      "epoch": 0.32619981657825525,
+      "grad_norm": 617.0887451171875,
+      "learning_rate": 3.2498713496067963e-06,
+      "loss": 40.8015,
+      "step": 80740
+    },
+    {
+      "epoch": 0.32624021784362284,
+      "grad_norm": 558.3300170898438,
+      "learning_rate": 3.2485502718768814e-06,
+      "loss": 38.1439,
+      "step": 80750
+    },
+    {
+      "epoch": 0.3262806191089905,
+      "grad_norm": 410.2859802246094,
+      "learning_rate": 3.2472293335017836e-06,
+      "loss": 42.2704,
+      "step": 80760
+    },
+    {
+      "epoch": 0.3263210203743581,
+      "grad_norm": 666.5332641601562,
+      "learning_rate": 3.245908534586602e-06,
+      "loss": 40.3663,
+      "step": 80770
+    },
+    {
+      "epoch": 0.32636142163972576,
+      "grad_norm": 737.5230712890625,
+      "learning_rate": 3.2445878752364298e-06,
+      "loss": 33.5636,
+      "step": 80780
+    },
+    {
+      "epoch": 0.3264018229050934,
+      "grad_norm": 496.13604736328125,
+      "learning_rate": 3.2432673555563433e-06,
+      "loss": 38.1529,
+      "step": 80790
+    },
+    {
+      "epoch": 0.32644222417046104,
+      "grad_norm": 966.6318359375,
+      "learning_rate": 3.2419469756514116e-06,
+      "loss": 49.5031,
+      "step": 80800
+    },
+    {
+      "epoch": 0.3264826254358287,
+      "grad_norm": 447.5719299316406,
+      "learning_rate": 3.2406267356266918e-06,
+      "loss": 29.6554,
+      "step": 80810
+    },
+    {
+      "epoch": 0.32652302670119626,
+      "grad_norm": 769.01025390625,
+      "learning_rate": 3.2393066355872264e-06,
+      "loss": 57.3569,
+      "step": 80820
+    },
+    {
+      "epoch": 0.3265634279665639,
+      "grad_norm": 454.7648010253906,
+      "learning_rate": 3.237986675638052e-06,
+      "loss": 46.1115,
+      "step": 80830
+    },
+    {
+      "epoch": 0.32660382923193154,
+      "grad_norm": 398.7479248046875,
+      "learning_rate": 3.236666855884192e-06,
+      "loss": 27.5807,
+      "step": 80840
+    },
+    {
+      "epoch": 0.3266442304972992,
+      "grad_norm": 687.7910766601562,
+      "learning_rate": 3.2353471764306567e-06,
+      "loss": 44.8174,
+      "step": 80850
+    },
+    {
+      "epoch": 0.3266846317626668,
+      "grad_norm": 510.6954345703125,
+      "learning_rate": 3.234027637382447e-06,
+      "loss": 42.4717,
+      "step": 80860
+    },
+    {
+      "epoch": 0.32672503302803446,
+      "grad_norm": 430.68328857421875,
+      "learning_rate": 3.2327082388445545e-06,
+      "loss": 41.8108,
+      "step": 80870
+    },
+    {
+      "epoch": 0.32676543429340205,
+      "grad_norm": 56.9016227722168,
+      "learning_rate": 3.2313889809219568e-06,
+      "loss": 30.2116,
+      "step": 80880
+    },
+    {
+      "epoch": 0.3268058355587697,
+      "grad_norm": 467.46502685546875,
+      "learning_rate": 3.2300698637196217e-06,
+      "loss": 31.7899,
+      "step": 80890
+    },
+    {
+      "epoch": 0.3268462368241373,
+      "grad_norm": 572.5570068359375,
+      "learning_rate": 3.2287508873425043e-06,
+      "loss": 43.237,
+      "step": 80900
+    },
+    {
+      "epoch": 0.32688663808950497,
+      "grad_norm": 874.2039794921875,
+      "learning_rate": 3.22743205189555e-06,
+      "loss": 55.5983,
+      "step": 80910
+    },
+    {
+      "epoch": 0.3269270393548726,
+      "grad_norm": 823.0765380859375,
+      "learning_rate": 3.2261133574836918e-06,
+      "loss": 37.0331,
+      "step": 80920
+    },
+    {
+      "epoch": 0.32696744062024025,
+      "grad_norm": 701.4224243164062,
+      "learning_rate": 3.2247948042118525e-06,
+      "loss": 50.2638,
+      "step": 80930
+    },
+    {
+      "epoch": 0.32700784188560783,
+      "grad_norm": 280.49932861328125,
+      "learning_rate": 3.223476392184944e-06,
+      "loss": 37.2814,
+      "step": 80940
+    },
+    {
+      "epoch": 0.32704824315097547,
+      "grad_norm": 562.7162475585938,
+      "learning_rate": 3.2221581215078656e-06,
+      "loss": 36.172,
+      "step": 80950
+    },
+    {
+      "epoch": 0.3270886444163431,
+      "grad_norm": 1017.1797485351562,
+      "learning_rate": 3.2208399922855055e-06,
+      "loss": 40.6515,
+      "step": 80960
+    },
+    {
+      "epoch": 0.32712904568171075,
+      "grad_norm": 565.8228149414062,
+      "learning_rate": 3.2195220046227425e-06,
+      "loss": 35.703,
+      "step": 80970
+    },
+    {
+      "epoch": 0.3271694469470784,
+      "grad_norm": 852.8961181640625,
+      "learning_rate": 3.218204158624445e-06,
+      "loss": 60.7625,
+      "step": 80980
+    },
+    {
+      "epoch": 0.32720984821244603,
+      "grad_norm": 808.6995239257812,
+      "learning_rate": 3.216886454395463e-06,
+      "loss": 53.9434,
+      "step": 80990
+    },
+    {
+      "epoch": 0.32725024947781367,
+      "grad_norm": 386.1480407714844,
+      "learning_rate": 3.2155688920406415e-06,
+      "loss": 28.4667,
+      "step": 81000
+    },
+    {
+      "epoch": 0.32729065074318126,
+      "grad_norm": 798.99755859375,
+      "learning_rate": 3.2142514716648143e-06,
+      "loss": 39.1096,
+      "step": 81010
+    },
+    {
+      "epoch": 0.3273310520085489,
+      "grad_norm": 242.5078582763672,
+      "learning_rate": 3.212934193372803e-06,
+      "loss": 37.2829,
+      "step": 81020
+    },
+    {
+      "epoch": 0.32737145327391654,
+      "grad_norm": 237.69207763671875,
+      "learning_rate": 3.2116170572694156e-06,
+      "loss": 41.0571,
+      "step": 81030
+    },
+    {
+      "epoch": 0.3274118545392842,
+      "grad_norm": 570.440673828125,
+      "learning_rate": 3.2103000634594518e-06,
+      "loss": 51.5456,
+      "step": 81040
+    },
+    {
+      "epoch": 0.3274522558046518,
+      "grad_norm": 616.5958862304688,
+      "learning_rate": 3.2089832120476983e-06,
+      "loss": 30.9374,
+      "step": 81050
+    },
+    {
+      "epoch": 0.32749265707001946,
+      "grad_norm": 810.25927734375,
+      "learning_rate": 3.2076665031389294e-06,
+      "loss": 30.6905,
+      "step": 81060
+    },
+    {
+      "epoch": 0.32753305833538704,
+      "grad_norm": 584.2343139648438,
+      "learning_rate": 3.2063499368379146e-06,
+      "loss": 38.2822,
+      "step": 81070
+    },
+    {
+      "epoch": 0.3275734596007547,
+      "grad_norm": 442.2859802246094,
+      "learning_rate": 3.2050335132494014e-06,
+      "loss": 54.2211,
+      "step": 81080
+    },
+    {
+      "epoch": 0.3276138608661223,
+      "grad_norm": 422.188232421875,
+      "learning_rate": 3.203717232478133e-06,
+      "loss": 26.3646,
+      "step": 81090
+    },
+    {
+      "epoch": 0.32765426213148996,
+      "grad_norm": 462.74261474609375,
+      "learning_rate": 3.2024010946288415e-06,
+      "loss": 42.2259,
+      "step": 81100
+    },
+    {
+      "epoch": 0.3276946633968576,
+      "grad_norm": 511.82464599609375,
+      "learning_rate": 3.201085099806245e-06,
+      "loss": 47.9144,
+      "step": 81110
+    },
+    {
+      "epoch": 0.32773506466222524,
+      "grad_norm": 1044.2135009765625,
+      "learning_rate": 3.199769248115051e-06,
+      "loss": 36.4315,
+      "step": 81120
+    },
+    {
+      "epoch": 0.3277754659275929,
+      "grad_norm": 2148.4111328125,
+      "learning_rate": 3.1984535396599565e-06,
+      "loss": 51.7218,
+      "step": 81130
+    },
+    {
+      "epoch": 0.32781586719296046,
+      "grad_norm": 278.0746154785156,
+      "learning_rate": 3.1971379745456452e-06,
+      "loss": 54.2099,
+      "step": 81140
+    },
+    {
+      "epoch": 0.3278562684583281,
+      "grad_norm": 1001.82568359375,
+      "learning_rate": 3.1958225528767918e-06,
+      "loss": 44.9904,
+      "step": 81150
+    },
+    {
+      "epoch": 0.32789666972369574,
+      "grad_norm": 594.7664794921875,
+      "learning_rate": 3.1945072747580585e-06,
+      "loss": 46.4409,
+      "step": 81160
+    },
+    {
+      "epoch": 0.3279370709890634,
+      "grad_norm": 623.8257446289062,
+      "learning_rate": 3.1931921402940946e-06,
+      "loss": 42.915,
+      "step": 81170
+    },
+    {
+      "epoch": 0.327977472254431,
+      "grad_norm": 392.2928466796875,
+      "learning_rate": 3.1918771495895395e-06,
+      "loss": 40.8185,
+      "step": 81180
+    },
+    {
+      "epoch": 0.32801787351979866,
+      "grad_norm": 710.558349609375,
+      "learning_rate": 3.1905623027490205e-06,
+      "loss": 42.129,
+      "step": 81190
+    },
+    {
+      "epoch": 0.32805827478516625,
+      "grad_norm": 529.6282958984375,
+      "learning_rate": 3.1892475998771567e-06,
+      "loss": 39.6081,
+      "step": 81200
+    },
+    {
+      "epoch": 0.3280986760505339,
+      "grad_norm": 546.3908081054688,
+      "learning_rate": 3.1879330410785503e-06,
+      "loss": 42.8713,
+      "step": 81210
+    },
+    {
+      "epoch": 0.32813907731590153,
+      "grad_norm": 811.4768676757812,
+      "learning_rate": 3.186618626457796e-06,
+      "loss": 26.8995,
+      "step": 81220
+    },
+    {
+      "epoch": 0.32817947858126917,
+      "grad_norm": 190.24868774414062,
+      "learning_rate": 3.1853043561194748e-06,
+      "loss": 35.0625,
+      "step": 81230
+    },
+    {
+      "epoch": 0.3282198798466368,
+      "grad_norm": 684.8766479492188,
+      "learning_rate": 3.183990230168159e-06,
+      "loss": 42.4473,
+      "step": 81240
+    },
+    {
+      "epoch": 0.32826028111200445,
+      "grad_norm": 831.6922607421875,
+      "learning_rate": 3.1826762487084053e-06,
+      "loss": 57.3726,
+      "step": 81250
+    },
+    {
+      "epoch": 0.32830068237737203,
+      "grad_norm": 587.2714233398438,
+      "learning_rate": 3.1813624118447615e-06,
+      "loss": 29.0623,
+      "step": 81260
+    },
+    {
+      "epoch": 0.3283410836427397,
+      "grad_norm": 541.7936401367188,
+      "learning_rate": 3.180048719681765e-06,
+      "loss": 41.3275,
+      "step": 81270
+    },
+    {
+      "epoch": 0.3283814849081073,
+      "grad_norm": 532.4156494140625,
+      "learning_rate": 3.178735172323939e-06,
+      "loss": 37.0309,
+      "step": 81280
+    },
+    {
+      "epoch": 0.32842188617347495,
+      "grad_norm": 144.05352783203125,
+      "learning_rate": 3.177421769875796e-06,
+      "loss": 29.7735,
+      "step": 81290
+    },
+    {
+      "epoch": 0.3284622874388426,
+      "grad_norm": 662.11376953125,
+      "learning_rate": 3.176108512441839e-06,
+      "loss": 49.1473,
+      "step": 81300
+    },
+    {
+      "epoch": 0.32850268870421023,
+      "grad_norm": 401.6919860839844,
+      "learning_rate": 3.174795400126557e-06,
+      "loss": 47.5695,
+      "step": 81310
+    },
+    {
+      "epoch": 0.3285430899695779,
+      "grad_norm": 732.7622680664062,
+      "learning_rate": 3.173482433034429e-06,
+      "loss": 41.8201,
+      "step": 81320
+    },
+    {
+      "epoch": 0.32858349123494546,
+      "grad_norm": 596.14794921875,
+      "learning_rate": 3.1721696112699217e-06,
+      "loss": 39.5877,
+      "step": 81330
+    },
+    {
+      "epoch": 0.3286238925003131,
+      "grad_norm": 818.538818359375,
+      "learning_rate": 3.1708569349374896e-06,
+      "loss": 43.1352,
+      "step": 81340
+    },
+    {
+      "epoch": 0.32866429376568074,
+      "grad_norm": 1098.4739990234375,
+      "learning_rate": 3.1695444041415757e-06,
+      "loss": 32.2592,
+      "step": 81350
+    },
+    {
+      "epoch": 0.3287046950310484,
+      "grad_norm": 402.6197509765625,
+      "learning_rate": 3.1682320189866133e-06,
+      "loss": 51.2497,
+      "step": 81360
+    },
+    {
+      "epoch": 0.328745096296416,
+      "grad_norm": 793.1287841796875,
+      "learning_rate": 3.1669197795770225e-06,
+      "loss": 59.4796,
+      "step": 81370
+    },
+    {
+      "epoch": 0.32878549756178366,
+      "grad_norm": 379.2230224609375,
+      "learning_rate": 3.165607686017212e-06,
+      "loss": 36.5757,
+      "step": 81380
+    },
+    {
+      "epoch": 0.32882589882715124,
+      "grad_norm": 1003.5530395507812,
+      "learning_rate": 3.164295738411578e-06,
+      "loss": 40.0238,
+      "step": 81390
+    },
+    {
+      "epoch": 0.3288663000925189,
+      "grad_norm": 501.1110534667969,
+      "learning_rate": 3.1629839368645087e-06,
+      "loss": 41.732,
+      "step": 81400
+    },
+    {
+      "epoch": 0.3289067013578865,
+      "grad_norm": 8355.318359375,
+      "learning_rate": 3.161672281480379e-06,
+      "loss": 61.845,
+      "step": 81410
+    },
+    {
+      "epoch": 0.32894710262325416,
+      "grad_norm": 376.4793701171875,
+      "learning_rate": 3.1603607723635455e-06,
+      "loss": 44.7023,
+      "step": 81420
+    },
+    {
+      "epoch": 0.3289875038886218,
+      "grad_norm": 326.89520263671875,
+      "learning_rate": 3.1590494096183643e-06,
+      "loss": 41.9557,
+      "step": 81430
+    },
+    {
+      "epoch": 0.32902790515398944,
+      "grad_norm": 659.9878540039062,
+      "learning_rate": 3.1577381933491718e-06,
+      "loss": 45.3585,
+      "step": 81440
+    },
+    {
+      "epoch": 0.3290683064193571,
+      "grad_norm": 671.0325927734375,
+      "learning_rate": 3.156427123660297e-06,
+      "loss": 47.1818,
+      "step": 81450
+    },
+    {
+      "epoch": 0.32910870768472467,
+      "grad_norm": 991.5573120117188,
+      "learning_rate": 3.1551162006560554e-06,
+      "loss": 55.3784,
+      "step": 81460
+    },
+    {
+      "epoch": 0.3291491089500923,
+      "grad_norm": 628.5300903320312,
+      "learning_rate": 3.15380542444075e-06,
+      "loss": 53.9291,
+      "step": 81470
+    },
+    {
+      "epoch": 0.32918951021545995,
+      "grad_norm": 505.6431884765625,
+      "learning_rate": 3.1524947951186746e-06,
+      "loss": 31.4605,
+      "step": 81480
+    },
+    {
+      "epoch": 0.3292299114808276,
+      "grad_norm": 439.1938171386719,
+      "learning_rate": 3.1511843127941085e-06,
+      "loss": 53.568,
+      "step": 81490
+    },
+    {
+      "epoch": 0.3292703127461952,
+      "grad_norm": 600.3242797851562,
+      "learning_rate": 3.149873977571324e-06,
+      "loss": 45.2601,
+      "step": 81500
+    },
+    {
+      "epoch": 0.32931071401156287,
+      "grad_norm": 869.1049194335938,
+      "learning_rate": 3.148563789554575e-06,
+      "loss": 42.587,
+      "step": 81510
+    },
+    {
+      "epoch": 0.32935111527693045,
+      "grad_norm": 645.8557739257812,
+      "learning_rate": 3.147253748848107e-06,
+      "loss": 64.908,
+      "step": 81520
+    },
+    {
+      "epoch": 0.3293915165422981,
+      "grad_norm": 309.96282958984375,
+      "learning_rate": 3.1459438555561565e-06,
+      "loss": 40.8375,
+      "step": 81530
+    },
+    {
+      "epoch": 0.32943191780766573,
+      "grad_norm": 725.4754028320312,
+      "learning_rate": 3.1446341097829446e-06,
+      "loss": 37.4378,
+      "step": 81540
+    },
+    {
+      "epoch": 0.32947231907303337,
+      "grad_norm": 490.93060302734375,
+      "learning_rate": 3.1433245116326812e-06,
+      "loss": 46.1421,
+      "step": 81550
+    },
+    {
+      "epoch": 0.329512720338401,
+      "grad_norm": 952.149658203125,
+      "learning_rate": 3.1420150612095653e-06,
+      "loss": 44.979,
+      "step": 81560
+    },
+    {
+      "epoch": 0.32955312160376865,
+      "grad_norm": 567.7958984375,
+      "learning_rate": 3.140705758617784e-06,
+      "loss": 32.8766,
+      "step": 81570
+    },
+    {
+      "epoch": 0.32959352286913624,
+      "grad_norm": 887.6821899414062,
+      "learning_rate": 3.139396603961512e-06,
+      "loss": 41.5468,
+      "step": 81580
+    },
+    {
+      "epoch": 0.3296339241345039,
+      "grad_norm": 384.929931640625,
+      "learning_rate": 3.1380875973449155e-06,
+      "loss": 50.3088,
+      "step": 81590
+    },
+    {
+      "epoch": 0.3296743253998715,
+      "grad_norm": 1463.9766845703125,
+      "learning_rate": 3.1367787388721427e-06,
+      "loss": 42.5035,
+      "step": 81600
+    },
+    {
+      "epoch": 0.32971472666523916,
+      "grad_norm": 222.37277221679688,
+      "learning_rate": 3.135470028647334e-06,
+      "loss": 47.8513,
+      "step": 81610
+    },
+    {
+      "epoch": 0.3297551279306068,
+      "grad_norm": 515.9407348632812,
+      "learning_rate": 3.134161466774617e-06,
+      "loss": 29.2294,
+      "step": 81620
+    },
+    {
+      "epoch": 0.32979552919597444,
+      "grad_norm": 489.002197265625,
+      "learning_rate": 3.1328530533581102e-06,
+      "loss": 43.5699,
+      "step": 81630
+    },
+    {
+      "epoch": 0.3298359304613421,
+      "grad_norm": 396.6682434082031,
+      "learning_rate": 3.131544788501917e-06,
+      "loss": 42.7945,
+      "step": 81640
+    },
+    {
+      "epoch": 0.32987633172670966,
+      "grad_norm": 813.0165405273438,
+      "learning_rate": 3.1302366723101294e-06,
+      "loss": 43.1957,
+      "step": 81650
+    },
+    {
+      "epoch": 0.3299167329920773,
+      "grad_norm": 106.5440673828125,
+      "learning_rate": 3.12892870488683e-06,
+      "loss": 38.8655,
+      "step": 81660
+    },
+    {
+      "epoch": 0.32995713425744494,
+      "grad_norm": 814.0963134765625,
+      "learning_rate": 3.1276208863360862e-06,
+      "loss": 43.5237,
+      "step": 81670
+    },
+    {
+      "epoch": 0.3299975355228126,
+      "grad_norm": 731.88330078125,
+      "learning_rate": 3.126313216761955e-06,
+      "loss": 57.3485,
+      "step": 81680
+    },
+    {
+      "epoch": 0.3300379367881802,
+      "grad_norm": 706.98779296875,
+      "learning_rate": 3.125005696268482e-06,
+      "loss": 59.5241,
+      "step": 81690
+    },
+    {
+      "epoch": 0.33007833805354786,
+      "grad_norm": 991.2210693359375,
+      "learning_rate": 3.1236983249597007e-06,
+      "loss": 54.4068,
+      "step": 81700
+    },
+    {
+      "epoch": 0.33011873931891544,
+      "grad_norm": 610.1210327148438,
+      "learning_rate": 3.1223911029396324e-06,
+      "loss": 53.5363,
+      "step": 81710
+    },
+    {
+      "epoch": 0.3301591405842831,
+      "grad_norm": 763.7863159179688,
+      "learning_rate": 3.121084030312286e-06,
+      "loss": 39.8277,
+      "step": 81720
+    },
+    {
+      "epoch": 0.3301995418496507,
+      "grad_norm": 429.45367431640625,
+      "learning_rate": 3.1197771071816617e-06,
+      "loss": 46.1526,
+      "step": 81730
+    },
+    {
+      "epoch": 0.33023994311501836,
+      "grad_norm": 561.5693969726562,
+      "learning_rate": 3.118470333651744e-06,
+      "loss": 54.1054,
+      "step": 81740
+    },
+    {
+      "epoch": 0.330280344380386,
+      "grad_norm": 761.9989624023438,
+      "learning_rate": 3.1171637098265063e-06,
+      "loss": 50.2312,
+      "step": 81750
+    },
+    {
+      "epoch": 0.33032074564575364,
+      "grad_norm": 901.3028564453125,
+      "learning_rate": 3.1158572358099127e-06,
+      "loss": 49.2848,
+      "step": 81760
+    },
+    {
+      "epoch": 0.3303611469111213,
+      "grad_norm": 768.6314086914062,
+      "learning_rate": 3.11455091170591e-06,
+      "loss": 44.6113,
+      "step": 81770
+    },
+    {
+      "epoch": 0.33040154817648887,
+      "grad_norm": 368.1278076171875,
+      "learning_rate": 3.1132447376184383e-06,
+      "loss": 57.5698,
+      "step": 81780
+    },
+    {
+      "epoch": 0.3304419494418565,
+      "grad_norm": 644.8842163085938,
+      "learning_rate": 3.1119387136514246e-06,
+      "loss": 38.2045,
+      "step": 81790
+    },
+    {
+      "epoch": 0.33048235070722415,
+      "grad_norm": 889.154296875,
+      "learning_rate": 3.1106328399087814e-06,
+      "loss": 47.1308,
+      "step": 81800
+    },
+    {
+      "epoch": 0.3305227519725918,
+      "grad_norm": 339.5101013183594,
+      "learning_rate": 3.1093271164944116e-06,
+      "loss": 48.5115,
+      "step": 81810
+    },
+    {
+      "epoch": 0.33056315323795943,
+      "grad_norm": 555.5851440429688,
+      "learning_rate": 3.1080215435122072e-06,
+      "loss": 45.9561,
+      "step": 81820
+    },
+    {
+      "epoch": 0.33060355450332707,
+      "grad_norm": 558.2139282226562,
+      "learning_rate": 3.106716121066046e-06,
+      "loss": 48.2627,
+      "step": 81830
+    },
+    {
+      "epoch": 0.33064395576869465,
+      "grad_norm": 889.7554931640625,
+      "learning_rate": 3.105410849259796e-06,
+      "loss": 57.0113,
+      "step": 81840
+    },
+    {
+      "epoch": 0.3306843570340623,
+      "grad_norm": 928.018798828125,
+      "learning_rate": 3.104105728197306e-06,
+      "loss": 39.6977,
+      "step": 81850
+    },
+    {
+      "epoch": 0.33072475829942993,
+      "grad_norm": 594.8426513671875,
+      "learning_rate": 3.1028007579824234e-06,
+      "loss": 43.1991,
+      "step": 81860
+    },
+    {
+      "epoch": 0.3307651595647976,
+      "grad_norm": 367.9489440917969,
+      "learning_rate": 3.1014959387189774e-06,
+      "loss": 51.1528,
+      "step": 81870
+    },
+    {
+      "epoch": 0.3308055608301652,
+      "grad_norm": 727.341552734375,
+      "learning_rate": 3.1001912705107874e-06,
+      "loss": 49.7708,
+      "step": 81880
+    },
+    {
+      "epoch": 0.33084596209553285,
+      "grad_norm": 572.1885375976562,
+      "learning_rate": 3.0988867534616586e-06,
+      "loss": 23.4917,
+      "step": 81890
+    },
+    {
+      "epoch": 0.33088636336090044,
+      "grad_norm": 244.5078125,
+      "learning_rate": 3.097582387675385e-06,
+      "loss": 45.9483,
+      "step": 81900
+    },
+    {
+      "epoch": 0.3309267646262681,
+      "grad_norm": 250.58753967285156,
+      "learning_rate": 3.09627817325575e-06,
+      "loss": 48.8653,
+      "step": 81910
+    },
+    {
+      "epoch": 0.3309671658916357,
+      "grad_norm": 494.48101806640625,
+      "learning_rate": 3.0949741103065246e-06,
+      "loss": 48.1861,
+      "step": 81920
+    },
+    {
+      "epoch": 0.33100756715700336,
+      "grad_norm": 641.30908203125,
+      "learning_rate": 3.093670198931469e-06,
+      "loss": 35.6989,
+      "step": 81930
+    },
+    {
+      "epoch": 0.331047968422371,
+      "grad_norm": 886.3203125,
+      "learning_rate": 3.0923664392343233e-06,
+      "loss": 59.5055,
+      "step": 81940
+    },
+    {
+      "epoch": 0.33108836968773864,
+      "grad_norm": 666.1203002929688,
+      "learning_rate": 3.091062831318825e-06,
+      "loss": 37.953,
+      "step": 81950
+    },
+    {
+      "epoch": 0.3311287709531063,
+      "grad_norm": 341.7474060058594,
+      "learning_rate": 3.089759375288698e-06,
+      "loss": 40.19,
+      "step": 81960
+    },
+    {
+      "epoch": 0.33116917221847386,
+      "grad_norm": 473.49053955078125,
+      "learning_rate": 3.0884560712476497e-06,
+      "loss": 48.5036,
+      "step": 81970
+    },
+    {
+      "epoch": 0.3312095734838415,
+      "grad_norm": 964.3297119140625,
+      "learning_rate": 3.0871529192993794e-06,
+      "loss": 56.6274,
+      "step": 81980
+    },
+    {
+      "epoch": 0.33124997474920914,
+      "grad_norm": 469.9378662109375,
+      "learning_rate": 3.085849919547572e-06,
+      "loss": 36.335,
+      "step": 81990
+    },
+    {
+      "epoch": 0.3312903760145768,
+      "grad_norm": 692.3630981445312,
+      "learning_rate": 3.0845470720959027e-06,
+      "loss": 45.9318,
+      "step": 82000
+    },
+    {
+      "epoch": 0.3313307772799444,
+      "grad_norm": 696.519775390625,
+      "learning_rate": 3.08324437704803e-06,
+      "loss": 48.5088,
+      "step": 82010
+    },
+    {
+      "epoch": 0.33137117854531206,
+      "grad_norm": 556.7713012695312,
+      "learning_rate": 3.0819418345076095e-06,
+      "loss": 20.585,
+      "step": 82020
+    },
+    {
+      "epoch": 0.33141157981067965,
+      "grad_norm": 387.6054992675781,
+      "learning_rate": 3.080639444578272e-06,
+      "loss": 42.5861,
+      "step": 82030
+    },
+    {
+      "epoch": 0.3314519810760473,
+      "grad_norm": 733.8251953125,
+      "learning_rate": 3.0793372073636455e-06,
+      "loss": 45.5413,
+      "step": 82040
+    },
+    {
+      "epoch": 0.3314923823414149,
+      "grad_norm": 1118.459228515625,
+      "learning_rate": 3.0780351229673423e-06,
+      "loss": 47.3603,
+      "step": 82050
+    },
+    {
+      "epoch": 0.33153278360678257,
+      "grad_norm": 620.4292602539062,
+      "learning_rate": 3.0767331914929638e-06,
+      "loss": 39.6754,
+      "step": 82060
+    },
+    {
+      "epoch": 0.3315731848721502,
+      "grad_norm": 351.4848327636719,
+      "learning_rate": 3.075431413044099e-06,
+      "loss": 43.1012,
+      "step": 82070
+    },
+    {
+      "epoch": 0.33161358613751785,
+      "grad_norm": 887.3251342773438,
+      "learning_rate": 3.074129787724324e-06,
+      "loss": 44.2201,
+      "step": 82080
+    },
+    {
+      "epoch": 0.33165398740288543,
+      "grad_norm": 575.6647338867188,
+      "learning_rate": 3.072828315637203e-06,
+      "loss": 41.9672,
+      "step": 82090
+    },
+    {
+      "epoch": 0.33169438866825307,
+      "grad_norm": 275.0605163574219,
+      "learning_rate": 3.0715269968862898e-06,
+      "loss": 42.3283,
+      "step": 82100
+    },
+    {
+      "epoch": 0.3317347899336207,
+      "grad_norm": 374.13812255859375,
+      "learning_rate": 3.0702258315751223e-06,
+      "loss": 72.7533,
+      "step": 82110
+    },
+    {
+      "epoch": 0.33177519119898835,
+      "grad_norm": 677.5405883789062,
+      "learning_rate": 3.0689248198072282e-06,
+      "loss": 30.5168,
+      "step": 82120
+    },
+    {
+      "epoch": 0.331815592464356,
+      "grad_norm": 456.5289306640625,
+      "learning_rate": 3.0676239616861234e-06,
+      "loss": 34.6827,
+      "step": 82130
+    },
+    {
+      "epoch": 0.33185599372972363,
+      "grad_norm": 652.3780517578125,
+      "learning_rate": 3.066323257315311e-06,
+      "loss": 44.3979,
+      "step": 82140
+    },
+    {
+      "epoch": 0.33189639499509127,
+      "grad_norm": 304.3308410644531,
+      "learning_rate": 3.065022706798284e-06,
+      "loss": 47.26,
+      "step": 82150
+    },
+    {
+      "epoch": 0.33193679626045886,
+      "grad_norm": 446.0997314453125,
+      "learning_rate": 3.06372231023852e-06,
+      "loss": 36.9697,
+      "step": 82160
+    },
+    {
+      "epoch": 0.3319771975258265,
+      "grad_norm": 525.1025390625,
+      "learning_rate": 3.0624220677394854e-06,
+      "loss": 34.6405,
+      "step": 82170
+    },
+    {
+      "epoch": 0.33201759879119414,
+      "grad_norm": 621.7910766601562,
+      "learning_rate": 3.0611219794046344e-06,
+      "loss": 39.364,
+      "step": 82180
+    },
+    {
+      "epoch": 0.3320580000565618,
+      "grad_norm": 484.9457702636719,
+      "learning_rate": 3.05982204533741e-06,
+      "loss": 39.2439,
+      "step": 82190
+    },
+    {
+      "epoch": 0.3320984013219294,
+      "grad_norm": 609.784912109375,
+      "learning_rate": 3.0585222656412406e-06,
+      "loss": 43.1885,
+      "step": 82200
+    },
+    {
+      "epoch": 0.33213880258729706,
+      "grad_norm": 746.94580078125,
+      "learning_rate": 3.0572226404195436e-06,
+      "loss": 28.1384,
+      "step": 82210
+    },
+    {
+      "epoch": 0.33217920385266464,
+      "grad_norm": 416.6310119628906,
+      "learning_rate": 3.055923169775726e-06,
+      "loss": 71.9208,
+      "step": 82220
+    },
+    {
+      "epoch": 0.3322196051180323,
+      "grad_norm": 378.2218017578125,
+      "learning_rate": 3.054623853813179e-06,
+      "loss": 37.3235,
+      "step": 82230
+    },
+    {
+      "epoch": 0.3322600063833999,
+      "grad_norm": 425.2323913574219,
+      "learning_rate": 3.0533246926352834e-06,
+      "loss": 46.6814,
+      "step": 82240
+    },
+    {
+      "epoch": 0.33230040764876756,
+      "grad_norm": 664.06982421875,
+      "learning_rate": 3.0520256863454077e-06,
+      "loss": 46.2079,
+      "step": 82250
+    },
+    {
+      "epoch": 0.3323408089141352,
+      "grad_norm": 527.8890380859375,
+      "learning_rate": 3.05072683504691e-06,
+      "loss": 33.8584,
+      "step": 82260
+    },
+    {
+      "epoch": 0.33238121017950284,
+      "grad_norm": 854.1585693359375,
+      "learning_rate": 3.049428138843133e-06,
+      "loss": 35.6598,
+      "step": 82270
+    },
+    {
+      "epoch": 0.3324216114448705,
+      "grad_norm": 596.1838989257812,
+      "learning_rate": 3.0481295978374037e-06,
+      "loss": 66.4695,
+      "step": 82280
+    },
+    {
+      "epoch": 0.33246201271023806,
+      "grad_norm": 691.150146484375,
+      "learning_rate": 3.0468312121330464e-06,
+      "loss": 37.7867,
+      "step": 82290
+    },
+    {
+      "epoch": 0.3325024139756057,
+      "grad_norm": 436.70391845703125,
+      "learning_rate": 3.0455329818333652e-06,
+      "loss": 31.3879,
+      "step": 82300
+    },
+    {
+      "epoch": 0.33254281524097334,
+      "grad_norm": 331.2305603027344,
+      "learning_rate": 3.044234907041655e-06,
+      "loss": 39.9435,
+      "step": 82310
+    },
+    {
+      "epoch": 0.332583216506341,
+      "grad_norm": 519.8748779296875,
+      "learning_rate": 3.0429369878611968e-06,
+      "loss": 34.4187,
+      "step": 82320
+    },
+    {
+      "epoch": 0.3326236177717086,
+      "grad_norm": 582.2870483398438,
+      "learning_rate": 3.041639224395262e-06,
+      "loss": 47.4416,
+      "step": 82330
+    },
+    {
+      "epoch": 0.33266401903707626,
+      "grad_norm": 534.1697998046875,
+      "learning_rate": 3.0403416167471044e-06,
+      "loss": 45.1939,
+      "step": 82340
+    },
+    {
+      "epoch": 0.33270442030244385,
+      "grad_norm": 538.1617431640625,
+      "learning_rate": 3.0390441650199727e-06,
+      "loss": 38.0616,
+      "step": 82350
+    },
+    {
+      "epoch": 0.3327448215678115,
+      "grad_norm": 930.120849609375,
+      "learning_rate": 3.0377468693170985e-06,
+      "loss": 48.7219,
+      "step": 82360
+    },
+    {
+      "epoch": 0.33278522283317913,
+      "grad_norm": 571.488037109375,
+      "learning_rate": 3.0364497297416973e-06,
+      "loss": 43.8716,
+      "step": 82370
+    },
+    {
+      "epoch": 0.33282562409854677,
+      "grad_norm": 642.69970703125,
+      "learning_rate": 3.035152746396981e-06,
+      "loss": 50.2554,
+      "step": 82380
+    },
+    {
+      "epoch": 0.3328660253639144,
+      "grad_norm": 424.82440185546875,
+      "learning_rate": 3.0338559193861434e-06,
+      "loss": 40.0815,
+      "step": 82390
+    },
+    {
+      "epoch": 0.33290642662928205,
+      "grad_norm": 641.3731079101562,
+      "learning_rate": 3.032559248812367e-06,
+      "loss": 49.77,
+      "step": 82400
+    },
+    {
+      "epoch": 0.33294682789464963,
+      "grad_norm": 903.556640625,
+      "learning_rate": 3.0312627347788208e-06,
+      "loss": 67.3948,
+      "step": 82410
+    },
+    {
+      "epoch": 0.3329872291600173,
+      "grad_norm": 711.0341186523438,
+      "learning_rate": 3.0299663773886646e-06,
+      "loss": 37.6624,
+      "step": 82420
+    },
+    {
+      "epoch": 0.3330276304253849,
+      "grad_norm": 830.165283203125,
+      "learning_rate": 3.0286701767450423e-06,
+      "loss": 49.855,
+      "step": 82430
+    },
+    {
+      "epoch": 0.33306803169075255,
+      "grad_norm": 729.2254638671875,
+      "learning_rate": 3.0273741329510852e-06,
+      "loss": 45.3208,
+      "step": 82440
+    },
+    {
+      "epoch": 0.3331084329561202,
+      "grad_norm": 316.6321716308594,
+      "learning_rate": 3.0260782461099192e-06,
+      "loss": 43.257,
+      "step": 82450
+    },
+    {
+      "epoch": 0.33314883422148783,
+      "grad_norm": 592.7379760742188,
+      "learning_rate": 3.024782516324645e-06,
+      "loss": 44.4966,
+      "step": 82460
+    },
+    {
+      "epoch": 0.3331892354868555,
+      "grad_norm": 571.3594360351562,
+      "learning_rate": 3.0234869436983606e-06,
+      "loss": 56.4878,
+      "step": 82470
+    },
+    {
+      "epoch": 0.33322963675222306,
+      "grad_norm": 375.4861145019531,
+      "learning_rate": 3.02219152833415e-06,
+      "loss": 62.9724,
+      "step": 82480
+    },
+    {
+      "epoch": 0.3332700380175907,
+      "grad_norm": 466.45416259765625,
+      "learning_rate": 3.0208962703350832e-06,
+      "loss": 59.3834,
+      "step": 82490
+    },
+    {
+      "epoch": 0.33331043928295834,
+      "grad_norm": 544.7888793945312,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 35.7769,
+      "step": 82500
+    },
+    {
+      "epoch": 0.333350840548326,
+      "grad_norm": 743.954833984375,
+      "learning_rate": 3.0183062268445964e-06,
+      "loss": 44.2142,
+      "step": 82510
+    },
+    {
+      "epoch": 0.3333912418136936,
+      "grad_norm": 448.221923828125,
+      "learning_rate": 3.0170114415592543e-06,
+      "loss": 39.1324,
+      "step": 82520
+    },
+    {
+      "epoch": 0.33343164307906126,
+      "grad_norm": 653.385986328125,
+      "learning_rate": 3.015716814051213e-06,
+      "loss": 40.8635,
+      "step": 82530
+    },
+    {
+      "epoch": 0.33347204434442884,
+      "grad_norm": 776.0372924804688,
+      "learning_rate": 3.0144223444234767e-06,
+      "loss": 34.3393,
+      "step": 82540
+    },
+    {
+      "epoch": 0.3335124456097965,
+      "grad_norm": 789.0279541015625,
+      "learning_rate": 3.0131280327790412e-06,
+      "loss": 59.5271,
+      "step": 82550
+    },
+    {
+      "epoch": 0.3335528468751641,
+      "grad_norm": 619.6022338867188,
+      "learning_rate": 3.0118338792208912e-06,
+      "loss": 29.2193,
+      "step": 82560
+    },
+    {
+      "epoch": 0.33359324814053176,
+      "grad_norm": 437.3066101074219,
+      "learning_rate": 3.010539883851993e-06,
+      "loss": 42.7655,
+      "step": 82570
+    },
+    {
+      "epoch": 0.3336336494058994,
+      "grad_norm": 492.7397766113281,
+      "learning_rate": 3.009246046775307e-06,
+      "loss": 30.9015,
+      "step": 82580
+    },
+    {
+      "epoch": 0.33367405067126704,
+      "grad_norm": 699.6427001953125,
+      "learning_rate": 3.0079523680937766e-06,
+      "loss": 60.2017,
+      "step": 82590
+    },
+    {
+      "epoch": 0.3337144519366347,
+      "grad_norm": 835.25244140625,
+      "learning_rate": 3.006658847910334e-06,
+      "loss": 43.552,
+      "step": 82600
+    },
+    {
+      "epoch": 0.33375485320200227,
+      "grad_norm": 489.4479064941406,
+      "learning_rate": 3.005365486327899e-06,
+      "loss": 29.7535,
+      "step": 82610
+    },
+    {
+      "epoch": 0.3337952544673699,
+      "grad_norm": 289.3746032714844,
+      "learning_rate": 3.004072283449379e-06,
+      "loss": 34.3409,
+      "step": 82620
+    },
+    {
+      "epoch": 0.33383565573273755,
+      "grad_norm": 358.6887512207031,
+      "learning_rate": 3.0027792393776666e-06,
+      "loss": 42.0769,
+      "step": 82630
+    },
+    {
+      "epoch": 0.3338760569981052,
+      "grad_norm": 478.79754638671875,
+      "learning_rate": 3.001486354215644e-06,
+      "loss": 41.234,
+      "step": 82640
+    },
+    {
+      "epoch": 0.3339164582634728,
+      "grad_norm": 336.5907897949219,
+      "learning_rate": 3.0001936280661794e-06,
+      "loss": 49.105,
+      "step": 82650
+    },
+    {
+      "epoch": 0.33395685952884047,
+      "grad_norm": 326.53662109375,
+      "learning_rate": 2.998901061032131e-06,
+      "loss": 46.3112,
+      "step": 82660
+    },
+    {
+      "epoch": 0.33399726079420805,
+      "grad_norm": 877.6404418945312,
+      "learning_rate": 2.9976086532163397e-06,
+      "loss": 46.8901,
+      "step": 82670
+    },
+    {
+      "epoch": 0.3340376620595757,
+      "grad_norm": 372.0155334472656,
+      "learning_rate": 2.9963164047216397e-06,
+      "loss": 39.4705,
+      "step": 82680
+    },
+    {
+      "epoch": 0.33407806332494333,
+      "grad_norm": 940.6842651367188,
+      "learning_rate": 2.9950243156508473e-06,
+      "loss": 39.6171,
+      "step": 82690
+    },
+    {
+      "epoch": 0.33411846459031097,
+      "grad_norm": 433.3446350097656,
+      "learning_rate": 2.9937323861067695e-06,
+      "loss": 25.4052,
+      "step": 82700
+    },
+    {
+      "epoch": 0.3341588658556786,
+      "grad_norm": 868.9320678710938,
+      "learning_rate": 2.992440616192197e-06,
+      "loss": 36.8388,
+      "step": 82710
+    },
+    {
+      "epoch": 0.33419926712104625,
+      "grad_norm": 717.5938720703125,
+      "learning_rate": 2.9911490060099117e-06,
+      "loss": 53.1356,
+      "step": 82720
+    },
+    {
+      "epoch": 0.33423966838641384,
+      "grad_norm": 458.17218017578125,
+      "learning_rate": 2.9898575556626807e-06,
+      "loss": 60.6068,
+      "step": 82730
+    },
+    {
+      "epoch": 0.3342800696517815,
+      "grad_norm": 542.5181274414062,
+      "learning_rate": 2.9885662652532586e-06,
+      "loss": 37.2215,
+      "step": 82740
+    },
+    {
+      "epoch": 0.3343204709171491,
+      "grad_norm": 611.09228515625,
+      "learning_rate": 2.9872751348843875e-06,
+      "loss": 38.3881,
+      "step": 82750
+    },
+    {
+      "epoch": 0.33436087218251676,
+      "grad_norm": 958.848876953125,
+      "learning_rate": 2.985984164658796e-06,
+      "loss": 43.9771,
+      "step": 82760
+    },
+    {
+      "epoch": 0.3344012734478844,
+      "grad_norm": 751.4885864257812,
+      "learning_rate": 2.9846933546792012e-06,
+      "loss": 45.087,
+      "step": 82770
+    },
+    {
+      "epoch": 0.33444167471325204,
+      "grad_norm": 444.13800048828125,
+      "learning_rate": 2.9834027050483085e-06,
+      "loss": 33.4151,
+      "step": 82780
+    },
+    {
+      "epoch": 0.3344820759786197,
+      "grad_norm": 975.753662109375,
+      "learning_rate": 2.9821122158688086e-06,
+      "loss": 37.9123,
+      "step": 82790
+    },
+    {
+      "epoch": 0.33452247724398726,
+      "grad_norm": 350.3032531738281,
+      "learning_rate": 2.980821887243377e-06,
+      "loss": 41.8596,
+      "step": 82800
+    },
+    {
+      "epoch": 0.3345628785093549,
+      "grad_norm": 494.2857360839844,
+      "learning_rate": 2.979531719274681e-06,
+      "loss": 42.1246,
+      "step": 82810
+    },
+    {
+      "epoch": 0.33460327977472254,
+      "grad_norm": 386.30780029296875,
+      "learning_rate": 2.978241712065374e-06,
+      "loss": 44.5044,
+      "step": 82820
+    },
+    {
+      "epoch": 0.3346436810400902,
+      "grad_norm": 1870.94287109375,
+      "learning_rate": 2.9769518657180953e-06,
+      "loss": 47.9243,
+      "step": 82830
+    },
+    {
+      "epoch": 0.3346840823054578,
+      "grad_norm": 412.831298828125,
+      "learning_rate": 2.9756621803354722e-06,
+      "loss": 44.2663,
+      "step": 82840
+    },
+    {
+      "epoch": 0.33472448357082546,
+      "grad_norm": 748.1416015625,
+      "learning_rate": 2.9743726560201185e-06,
+      "loss": 47.1023,
+      "step": 82850
+    },
+    {
+      "epoch": 0.33476488483619304,
+      "grad_norm": 672.1915893554688,
+      "learning_rate": 2.9730832928746355e-06,
+      "loss": 40.4036,
+      "step": 82860
+    },
+    {
+      "epoch": 0.3348052861015607,
+      "grad_norm": 468.0141296386719,
+      "learning_rate": 2.9717940910016135e-06,
+      "loss": 45.7574,
+      "step": 82870
+    },
+    {
+      "epoch": 0.3348456873669283,
+      "grad_norm": 888.3388671875,
+      "learning_rate": 2.9705050505036294e-06,
+      "loss": 38.8895,
+      "step": 82880
+    },
+    {
+      "epoch": 0.33488608863229596,
+      "grad_norm": 371.0845947265625,
+      "learning_rate": 2.9692161714832422e-06,
+      "loss": 51.6445,
+      "step": 82890
+    },
+    {
+      "epoch": 0.3349264898976636,
+      "grad_norm": 931.8453979492188,
+      "learning_rate": 2.9679274540430037e-06,
+      "loss": 38.1988,
+      "step": 82900
+    },
+    {
+      "epoch": 0.33496689116303124,
+      "grad_norm": 438.6957702636719,
+      "learning_rate": 2.966638898285452e-06,
+      "loss": 40.7775,
+      "step": 82910
+    },
+    {
+      "epoch": 0.3350072924283989,
+      "grad_norm": 355.041015625,
+      "learning_rate": 2.9653505043131125e-06,
+      "loss": 36.4231,
+      "step": 82920
+    },
+    {
+      "epoch": 0.33504769369376647,
+      "grad_norm": 220.08999633789062,
+      "learning_rate": 2.9640622722284944e-06,
+      "loss": 44.5257,
+      "step": 82930
+    },
+    {
+      "epoch": 0.3350880949591341,
+      "grad_norm": 300.3365783691406,
+      "learning_rate": 2.962774202134098e-06,
+      "loss": 44.7493,
+      "step": 82940
+    },
+    {
+      "epoch": 0.33512849622450175,
+      "grad_norm": 364.51422119140625,
+      "learning_rate": 2.961486294132409e-06,
+      "loss": 47.445,
+      "step": 82950
+    },
+    {
+      "epoch": 0.3351688974898694,
+      "grad_norm": 754.2266235351562,
+      "learning_rate": 2.960198548325901e-06,
+      "loss": 40.6013,
+      "step": 82960
+    },
+    {
+      "epoch": 0.33520929875523703,
+      "grad_norm": 781.6005859375,
+      "learning_rate": 2.958910964817032e-06,
+      "loss": 39.7547,
+      "step": 82970
+    },
+    {
+      "epoch": 0.33524970002060467,
+      "grad_norm": 538.5778198242188,
+      "learning_rate": 2.9576235437082502e-06,
+      "loss": 37.5946,
+      "step": 82980
+    },
+    {
+      "epoch": 0.33529010128597225,
+      "grad_norm": 514.644287109375,
+      "learning_rate": 2.9563362851019893e-06,
+      "loss": 68.5898,
+      "step": 82990
+    },
+    {
+      "epoch": 0.3353305025513399,
+      "grad_norm": 700.5049438476562,
+      "learning_rate": 2.9550491891006704e-06,
+      "loss": 45.753,
+      "step": 83000
+    },
+    {
+      "epoch": 0.33537090381670753,
+      "grad_norm": 443.4197998046875,
+      "learning_rate": 2.9537622558067036e-06,
+      "loss": 42.1891,
+      "step": 83010
+    },
+    {
+      "epoch": 0.3354113050820752,
+      "grad_norm": 298.11846923828125,
+      "learning_rate": 2.9524754853224837e-06,
+      "loss": 46.5696,
+      "step": 83020
+    },
+    {
+      "epoch": 0.3354517063474428,
+      "grad_norm": 699.7203369140625,
+      "learning_rate": 2.9511888777503916e-06,
+      "loss": 41.8185,
+      "step": 83030
+    },
+    {
+      "epoch": 0.33549210761281045,
+      "grad_norm": 1089.260009765625,
+      "learning_rate": 2.949902433192798e-06,
+      "loss": 38.7898,
+      "step": 83040
+    },
+    {
+      "epoch": 0.33553250887817804,
+      "grad_norm": 848.50244140625,
+      "learning_rate": 2.94861615175206e-06,
+      "loss": 36.6572,
+      "step": 83050
+    },
+    {
+      "epoch": 0.3355729101435457,
+      "grad_norm": 503.3612976074219,
+      "learning_rate": 2.9473300335305193e-06,
+      "loss": 41.0212,
+      "step": 83060
+    },
+    {
+      "epoch": 0.3356133114089133,
+      "grad_norm": 485.5074768066406,
+      "learning_rate": 2.946044078630508e-06,
+      "loss": 59.2754,
+      "step": 83070
+    },
+    {
+      "epoch": 0.33565371267428096,
+      "grad_norm": 322.2576599121094,
+      "learning_rate": 2.9447582871543423e-06,
+      "loss": 35.568,
+      "step": 83080
+    },
+    {
+      "epoch": 0.3356941139396486,
+      "grad_norm": 496.3887939453125,
+      "learning_rate": 2.9434726592043263e-06,
+      "loss": 45.2664,
+      "step": 83090
+    },
+    {
+      "epoch": 0.33573451520501624,
+      "grad_norm": 578.519287109375,
+      "learning_rate": 2.942187194882754e-06,
+      "loss": 39.4166,
+      "step": 83100
+    },
+    {
+      "epoch": 0.3357749164703839,
+      "grad_norm": 889.9583740234375,
+      "learning_rate": 2.940901894291902e-06,
+      "loss": 43.7717,
+      "step": 83110
+    },
+    {
+      "epoch": 0.33581531773575146,
+      "grad_norm": 479.1361083984375,
+      "learning_rate": 2.939616757534037e-06,
+      "loss": 44.8988,
+      "step": 83120
+    },
+    {
+      "epoch": 0.3358557190011191,
+      "grad_norm": 570.2633666992188,
+      "learning_rate": 2.938331784711411e-06,
+      "loss": 39.3279,
+      "step": 83130
+    },
+    {
+      "epoch": 0.33589612026648674,
+      "grad_norm": 812.4302368164062,
+      "learning_rate": 2.937046975926262e-06,
+      "loss": 54.3701,
+      "step": 83140
+    },
+    {
+      "epoch": 0.3359365215318544,
+      "grad_norm": 727.7770385742188,
+      "learning_rate": 2.9357623312808183e-06,
+      "loss": 44.7289,
+      "step": 83150
+    },
+    {
+      "epoch": 0.335976922797222,
+      "grad_norm": 206.3565216064453,
+      "learning_rate": 2.934477850877292e-06,
+      "loss": 41.506,
+      "step": 83160
+    },
+    {
+      "epoch": 0.33601732406258966,
+      "grad_norm": 480.6734924316406,
+      "learning_rate": 2.9331935348178838e-06,
+      "loss": 33.4442,
+      "step": 83170
+    },
+    {
+      "epoch": 0.33605772532795725,
+      "grad_norm": 1374.9521484375,
+      "learning_rate": 2.931909383204781e-06,
+      "loss": 71.191,
+      "step": 83180
+    },
+    {
+      "epoch": 0.3360981265933249,
+      "grad_norm": 594.8909301757812,
+      "learning_rate": 2.9306253961401553e-06,
+      "loss": 38.9371,
+      "step": 83190
+    },
+    {
+      "epoch": 0.3361385278586925,
+      "grad_norm": 477.9680480957031,
+      "learning_rate": 2.929341573726171e-06,
+      "loss": 39.7385,
+      "step": 83200
+    },
+    {
+      "epoch": 0.33617892912406017,
+      "grad_norm": 692.1814575195312,
+      "learning_rate": 2.928057916064975e-06,
+      "loss": 45.9146,
+      "step": 83210
+    },
+    {
+      "epoch": 0.3362193303894278,
+      "grad_norm": 755.2357788085938,
+      "learning_rate": 2.9267744232587035e-06,
+      "loss": 51.6069,
+      "step": 83220
+    },
+    {
+      "epoch": 0.33625973165479545,
+      "grad_norm": 175.2359619140625,
+      "learning_rate": 2.925491095409473e-06,
+      "loss": 33.7543,
+      "step": 83230
+    },
+    {
+      "epoch": 0.3363001329201631,
+      "grad_norm": 1155.9080810546875,
+      "learning_rate": 2.924207932619397e-06,
+      "loss": 40.8623,
+      "step": 83240
+    },
+    {
+      "epoch": 0.33634053418553067,
+      "grad_norm": 795.6212768554688,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 38.2946,
+      "step": 83250
+    },
+    {
+      "epoch": 0.3363809354508983,
+      "grad_norm": 507.70013427734375,
+      "learning_rate": 2.9216421026250707e-06,
+      "loss": 58.5413,
+      "step": 83260
+    },
+    {
+      "epoch": 0.33642133671626595,
+      "grad_norm": 418.37408447265625,
+      "learning_rate": 2.9203594356249726e-06,
+      "loss": 40.5312,
+      "step": 83270
+    },
+    {
+      "epoch": 0.3364617379816336,
+      "grad_norm": 339.6582946777344,
+      "learning_rate": 2.919076934092329e-06,
+      "loss": 43.9843,
+      "step": 83280
+    },
+    {
+      "epoch": 0.33650213924700123,
+      "grad_norm": 544.9318237304688,
+      "learning_rate": 2.9177945981291843e-06,
+      "loss": 31.948,
+      "step": 83290
+    },
+    {
+      "epoch": 0.33654254051236887,
+      "grad_norm": 1640.63818359375,
+      "learning_rate": 2.916512427837568e-06,
+      "loss": 47.5451,
+      "step": 83300
+    },
+    {
+      "epoch": 0.33658294177773646,
+      "grad_norm": 394.36474609375,
+      "learning_rate": 2.9152304233194974e-06,
+      "loss": 36.8862,
+      "step": 83310
+    },
+    {
+      "epoch": 0.3366233430431041,
+      "grad_norm": 240.1521759033203,
+      "learning_rate": 2.9139485846769723e-06,
+      "loss": 33.977,
+      "step": 83320
+    },
+    {
+      "epoch": 0.33666374430847174,
+      "grad_norm": 683.187255859375,
+      "learning_rate": 2.9126669120119846e-06,
+      "loss": 45.2668,
+      "step": 83330
+    },
+    {
+      "epoch": 0.3367041455738394,
+      "grad_norm": 715.0475463867188,
+      "learning_rate": 2.9113854054265112e-06,
+      "loss": 46.4243,
+      "step": 83340
+    },
+    {
+      "epoch": 0.336744546839207,
+      "grad_norm": 333.6781921386719,
+      "learning_rate": 2.9101040650225155e-06,
+      "loss": 43.0255,
+      "step": 83350
+    },
+    {
+      "epoch": 0.33678494810457466,
+      "grad_norm": 460.63641357421875,
+      "learning_rate": 2.9088228909019455e-06,
+      "loss": 44.8078,
+      "step": 83360
+    },
+    {
+      "epoch": 0.33682534936994224,
+      "grad_norm": 1160.0489501953125,
+      "learning_rate": 2.9075418831667436e-06,
+      "loss": 56.1693,
+      "step": 83370
+    },
+    {
+      "epoch": 0.3368657506353099,
+      "grad_norm": 558.1758422851562,
+      "learning_rate": 2.906261041918831e-06,
+      "loss": 43.3211,
+      "step": 83380
+    },
+    {
+      "epoch": 0.3369061519006775,
+      "grad_norm": 640.573974609375,
+      "learning_rate": 2.90498036726012e-06,
+      "loss": 39.1272,
+      "step": 83390
+    },
+    {
+      "epoch": 0.33694655316604516,
+      "grad_norm": 498.0477294921875,
+      "learning_rate": 2.903699859292505e-06,
+      "loss": 35.0058,
+      "step": 83400
+    },
+    {
+      "epoch": 0.3369869544314128,
+      "grad_norm": 343.3585510253906,
+      "learning_rate": 2.9024195181178704e-06,
+      "loss": 54.491,
+      "step": 83410
+    },
+    {
+      "epoch": 0.33702735569678044,
+      "grad_norm": 537.1847534179688,
+      "learning_rate": 2.9011393438380884e-06,
+      "loss": 28.7002,
+      "step": 83420
+    },
+    {
+      "epoch": 0.3370677569621481,
+      "grad_norm": 502.0111999511719,
+      "learning_rate": 2.8998593365550178e-06,
+      "loss": 29.3126,
+      "step": 83430
+    },
+    {
+      "epoch": 0.33710815822751566,
+      "grad_norm": 477.94482421875,
+      "learning_rate": 2.8985794963704992e-06,
+      "loss": 44.7282,
+      "step": 83440
+    },
+    {
+      "epoch": 0.3371485594928833,
+      "grad_norm": 167.1127166748047,
+      "learning_rate": 2.8972998233863657e-06,
+      "loss": 31.71,
+      "step": 83450
+    },
+    {
+      "epoch": 0.33718896075825094,
+      "grad_norm": 982.8727416992188,
+      "learning_rate": 2.8960203177044364e-06,
+      "loss": 52.744,
+      "step": 83460
+    },
+    {
+      "epoch": 0.3372293620236186,
+      "grad_norm": 587.1280517578125,
+      "learning_rate": 2.8947409794265146e-06,
+      "loss": 45.7435,
+      "step": 83470
+    },
+    {
+      "epoch": 0.3372697632889862,
+      "grad_norm": 859.0236206054688,
+      "learning_rate": 2.893461808654393e-06,
+      "loss": 36.1571,
+      "step": 83480
+    },
+    {
+      "epoch": 0.33731016455435386,
+      "grad_norm": 489.2738037109375,
+      "learning_rate": 2.892182805489846e-06,
+      "loss": 51.3487,
+      "step": 83490
+    },
+    {
+      "epoch": 0.33735056581972145,
+      "grad_norm": 716.7078857421875,
+      "learning_rate": 2.8909039700346385e-06,
+      "loss": 33.1731,
+      "step": 83500
+    },
+    {
+      "epoch": 0.3373909670850891,
+      "grad_norm": 479.42559814453125,
+      "learning_rate": 2.889625302390524e-06,
+      "loss": 32.8824,
+      "step": 83510
+    },
+    {
+      "epoch": 0.33743136835045673,
+      "grad_norm": 645.82275390625,
+      "learning_rate": 2.8883468026592382e-06,
+      "loss": 48.3588,
+      "step": 83520
+    },
+    {
+      "epoch": 0.33747176961582437,
+      "grad_norm": 285.9710388183594,
+      "learning_rate": 2.8870684709425063e-06,
+      "loss": 46.8221,
+      "step": 83530
+    },
+    {
+      "epoch": 0.337512170881192,
+      "grad_norm": 698.7156982421875,
+      "learning_rate": 2.885790307342039e-06,
+      "loss": 39.8056,
+      "step": 83540
+    },
+    {
+      "epoch": 0.33755257214655965,
+      "grad_norm": 644.73974609375,
+      "learning_rate": 2.884512311959532e-06,
+      "loss": 55.1326,
+      "step": 83550
+    },
+    {
+      "epoch": 0.3375929734119273,
+      "grad_norm": 595.5823974609375,
+      "learning_rate": 2.8832344848966758e-06,
+      "loss": 48.1498,
+      "step": 83560
+    },
+    {
+      "epoch": 0.3376333746772949,
+      "grad_norm": 777.5430908203125,
+      "learning_rate": 2.8819568262551344e-06,
+      "loss": 47.8501,
+      "step": 83570
+    },
+    {
+      "epoch": 0.3376737759426625,
+      "grad_norm": 695.3916015625,
+      "learning_rate": 2.8806793361365686e-06,
+      "loss": 63.2564,
+      "step": 83580
+    },
+    {
+      "epoch": 0.33771417720803015,
+      "grad_norm": 802.4063720703125,
+      "learning_rate": 2.8794020146426217e-06,
+      "loss": 48.957,
+      "step": 83590
+    },
+    {
+      "epoch": 0.3377545784733978,
+      "grad_norm": 430.29315185546875,
+      "learning_rate": 2.8781248618749235e-06,
+      "loss": 40.1431,
+      "step": 83600
+    },
+    {
+      "epoch": 0.33779497973876543,
+      "grad_norm": 359.31207275390625,
+      "learning_rate": 2.8768478779350927e-06,
+      "loss": 33.8353,
+      "step": 83610
+    },
+    {
+      "epoch": 0.3378353810041331,
+      "grad_norm": 204.70376586914062,
+      "learning_rate": 2.875571062924732e-06,
+      "loss": 46.5473,
+      "step": 83620
+    },
+    {
+      "epoch": 0.33787578226950066,
+      "grad_norm": 780.0933227539062,
+      "learning_rate": 2.874294416945432e-06,
+      "loss": 40.4211,
+      "step": 83630
+    },
+    {
+      "epoch": 0.3379161835348683,
+      "grad_norm": 535.6966552734375,
+      "learning_rate": 2.8730179400987697e-06,
+      "loss": 67.5744,
+      "step": 83640
+    },
+    {
+      "epoch": 0.33795658480023594,
+      "grad_norm": 576.902587890625,
+      "learning_rate": 2.871741632486308e-06,
+      "loss": 34.0992,
+      "step": 83650
+    },
+    {
+      "epoch": 0.3379969860656036,
+      "grad_norm": 611.9027709960938,
+      "learning_rate": 2.8704654942095977e-06,
+      "loss": 26.9851,
+      "step": 83660
+    },
+    {
+      "epoch": 0.3380373873309712,
+      "grad_norm": 414.39654541015625,
+      "learning_rate": 2.869189525370174e-06,
+      "loss": 43.7683,
+      "step": 83670
+    },
+    {
+      "epoch": 0.33807778859633886,
+      "grad_norm": 864.5162963867188,
+      "learning_rate": 2.8679137260695614e-06,
+      "loss": 53.398,
+      "step": 83680
+    },
+    {
+      "epoch": 0.33811818986170644,
+      "grad_norm": 492.2377624511719,
+      "learning_rate": 2.866638096409269e-06,
+      "loss": 34.3411,
+      "step": 83690
+    },
+    {
+      "epoch": 0.3381585911270741,
+      "grad_norm": 759.2435302734375,
+      "learning_rate": 2.8653626364907918e-06,
+      "loss": 41.1706,
+      "step": 83700
+    },
+    {
+      "epoch": 0.3381989923924417,
+      "grad_norm": 688.8447265625,
+      "learning_rate": 2.8640873464156127e-06,
+      "loss": 49.1247,
+      "step": 83710
+    },
+    {
+      "epoch": 0.33823939365780936,
+      "grad_norm": 655.9478759765625,
+      "learning_rate": 2.8628122262852015e-06,
+      "loss": 43.5474,
+      "step": 83720
+    },
+    {
+      "epoch": 0.338279794923177,
+      "grad_norm": 255.67202758789062,
+      "learning_rate": 2.861537276201013e-06,
+      "loss": 28.4624,
+      "step": 83730
+    },
+    {
+      "epoch": 0.33832019618854464,
+      "grad_norm": 987.0891723632812,
+      "learning_rate": 2.860262496264489e-06,
+      "loss": 47.4445,
+      "step": 83740
+    },
+    {
+      "epoch": 0.3383605974539123,
+      "grad_norm": 204.87025451660156,
+      "learning_rate": 2.858987886577058e-06,
+      "loss": 69.179,
+      "step": 83750
+    },
+    {
+      "epoch": 0.33840099871927987,
+      "grad_norm": 296.1997375488281,
+      "learning_rate": 2.857713447240135e-06,
+      "loss": 64.5847,
+      "step": 83760
+    },
+    {
+      "epoch": 0.3384413999846475,
+      "grad_norm": 541.6371459960938,
+      "learning_rate": 2.8564391783551214e-06,
+      "loss": 43.0074,
+      "step": 83770
+    },
+    {
+      "epoch": 0.33848180125001515,
+      "grad_norm": 551.6318359375,
+      "learning_rate": 2.855165080023405e-06,
+      "loss": 34.2838,
+      "step": 83780
+    },
+    {
+      "epoch": 0.3385222025153828,
+      "grad_norm": 357.3740234375,
+      "learning_rate": 2.85389115234636e-06,
+      "loss": 33.5051,
+      "step": 83790
+    },
+    {
+      "epoch": 0.3385626037807504,
+      "grad_norm": 280.4405822753906,
+      "learning_rate": 2.8526173954253458e-06,
+      "loss": 35.4893,
+      "step": 83800
+    },
+    {
+      "epoch": 0.33860300504611807,
+      "grad_norm": 968.9623413085938,
+      "learning_rate": 2.8513438093617107e-06,
+      "loss": 53.0172,
+      "step": 83810
+    },
+    {
+      "epoch": 0.33864340631148565,
+      "grad_norm": 247.39732360839844,
+      "learning_rate": 2.8500703942567874e-06,
+      "loss": 36.6919,
+      "step": 83820
+    },
+    {
+      "epoch": 0.3386838075768533,
+      "grad_norm": 554.3946533203125,
+      "learning_rate": 2.848797150211896e-06,
+      "loss": 39.9293,
+      "step": 83830
+    },
+    {
+      "epoch": 0.33872420884222093,
+      "grad_norm": 329.2684020996094,
+      "learning_rate": 2.847524077328343e-06,
+      "loss": 44.9997,
+      "step": 83840
+    },
+    {
+      "epoch": 0.33876461010758857,
+      "grad_norm": 585.6620483398438,
+      "learning_rate": 2.8462511757074205e-06,
+      "loss": 42.0342,
+      "step": 83850
+    },
+    {
+      "epoch": 0.3388050113729562,
+      "grad_norm": 481.63720703125,
+      "learning_rate": 2.844978445450408e-06,
+      "loss": 39.1469,
+      "step": 83860
+    },
+    {
+      "epoch": 0.33884541263832385,
+      "grad_norm": 857.4867553710938,
+      "learning_rate": 2.8437058866585698e-06,
+      "loss": 46.5103,
+      "step": 83870
+    },
+    {
+      "epoch": 0.3388858139036915,
+      "grad_norm": 676.7787475585938,
+      "learning_rate": 2.842433499433158e-06,
+      "loss": 52.469,
+      "step": 83880
+    },
+    {
+      "epoch": 0.3389262151690591,
+      "grad_norm": 746.2361450195312,
+      "learning_rate": 2.841161283875411e-06,
+      "loss": 30.7549,
+      "step": 83890
+    },
+    {
+      "epoch": 0.3389666164344267,
+      "grad_norm": 416.3562316894531,
+      "learning_rate": 2.8398892400865537e-06,
+      "loss": 41.4338,
+      "step": 83900
+    },
+    {
+      "epoch": 0.33900701769979436,
+      "grad_norm": 398.10546875,
+      "learning_rate": 2.838617368167797e-06,
+      "loss": 56.6096,
+      "step": 83910
+    },
+    {
+      "epoch": 0.339047418965162,
+      "grad_norm": 443.0265808105469,
+      "learning_rate": 2.837345668220333e-06,
+      "loss": 57.8073,
+      "step": 83920
+    },
+    {
+      "epoch": 0.33908782023052964,
+      "grad_norm": 297.83135986328125,
+      "learning_rate": 2.836074140345352e-06,
+      "loss": 22.5456,
+      "step": 83930
+    },
+    {
+      "epoch": 0.3391282214958973,
+      "grad_norm": 1189.527587890625,
+      "learning_rate": 2.834802784644019e-06,
+      "loss": 78.1559,
+      "step": 83940
+    },
+    {
+      "epoch": 0.33916862276126486,
+      "grad_norm": 482.82147216796875,
+      "learning_rate": 2.8335316012174925e-06,
+      "loss": 21.9731,
+      "step": 83950
+    },
+    {
+      "epoch": 0.3392090240266325,
+      "grad_norm": 192.12002563476562,
+      "learning_rate": 2.8322605901669133e-06,
+      "loss": 42.8519,
+      "step": 83960
+    },
+    {
+      "epoch": 0.33924942529200014,
+      "grad_norm": 912.255126953125,
+      "learning_rate": 2.8309897515934104e-06,
+      "loss": 74.6317,
+      "step": 83970
+    },
+    {
+      "epoch": 0.3392898265573678,
+      "grad_norm": 405.94451904296875,
+      "learning_rate": 2.8297190855980987e-06,
+      "loss": 48.1218,
+      "step": 83980
+    },
+    {
+      "epoch": 0.3393302278227354,
+      "grad_norm": 685.3220825195312,
+      "learning_rate": 2.8284485922820814e-06,
+      "loss": 42.0125,
+      "step": 83990
+    },
+    {
+      "epoch": 0.33937062908810306,
+      "grad_norm": 376.51837158203125,
+      "learning_rate": 2.8271782717464413e-06,
+      "loss": 41.4176,
+      "step": 84000
+    },
+    {
+      "epoch": 0.33941103035347064,
+      "grad_norm": 305.7981872558594,
+      "learning_rate": 2.8259081240922522e-06,
+      "loss": 30.9832,
+      "step": 84010
+    },
+    {
+      "epoch": 0.3394514316188383,
+      "grad_norm": 537.4862670898438,
+      "learning_rate": 2.8246381494205775e-06,
+      "loss": 56.0932,
+      "step": 84020
+    },
+    {
+      "epoch": 0.3394918328842059,
+      "grad_norm": 712.7220458984375,
+      "learning_rate": 2.8233683478324627e-06,
+      "loss": 52.1589,
+      "step": 84030
+    },
+    {
+      "epoch": 0.33953223414957356,
+      "grad_norm": 647.290283203125,
+      "learning_rate": 2.822098719428938e-06,
+      "loss": 42.8148,
+      "step": 84040
+    },
+    {
+      "epoch": 0.3395726354149412,
+      "grad_norm": 1051.1763916015625,
+      "learning_rate": 2.8208292643110237e-06,
+      "loss": 49.4385,
+      "step": 84050
+    },
+    {
+      "epoch": 0.33961303668030884,
+      "grad_norm": 846.8150024414062,
+      "learning_rate": 2.8195599825797233e-06,
+      "loss": 46.9994,
+      "step": 84060
+    },
+    {
+      "epoch": 0.3396534379456765,
+      "grad_norm": 455.5355224609375,
+      "learning_rate": 2.818290874336028e-06,
+      "loss": 37.593,
+      "step": 84070
+    },
+    {
+      "epoch": 0.33969383921104407,
+      "grad_norm": 1034.6007080078125,
+      "learning_rate": 2.817021939680918e-06,
+      "loss": 38.3971,
+      "step": 84080
+    },
+    {
+      "epoch": 0.3397342404764117,
+      "grad_norm": 585.2464599609375,
+      "learning_rate": 2.8157531787153515e-06,
+      "loss": 46.1575,
+      "step": 84090
+    },
+    {
+      "epoch": 0.33977464174177935,
+      "grad_norm": 437.1051940917969,
+      "learning_rate": 2.8144845915402796e-06,
+      "loss": 51.5693,
+      "step": 84100
+    },
+    {
+      "epoch": 0.339815043007147,
+      "grad_norm": 694.668701171875,
+      "learning_rate": 2.813216178256637e-06,
+      "loss": 40.9471,
+      "step": 84110
+    },
+    {
+      "epoch": 0.33985544427251463,
+      "grad_norm": 698.83544921875,
+      "learning_rate": 2.8119479389653492e-06,
+      "loss": 44.3773,
+      "step": 84120
+    },
+    {
+      "epoch": 0.33989584553788227,
+      "grad_norm": 734.264892578125,
+      "learning_rate": 2.8106798737673223e-06,
+      "loss": 35.5071,
+      "step": 84130
+    },
+    {
+      "epoch": 0.33993624680324985,
+      "grad_norm": 631.1513671875,
+      "learning_rate": 2.8094119827634496e-06,
+      "loss": 38.4813,
+      "step": 84140
+    },
+    {
+      "epoch": 0.3399766480686175,
+      "grad_norm": 497.7986145019531,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 32.6974,
+      "step": 84150
+    },
+    {
+      "epoch": 0.34001704933398513,
+      "grad_norm": 440.0523376464844,
+      "learning_rate": 2.806876723741677e-06,
+      "loss": 38.9369,
+      "step": 84160
+    },
+    {
+      "epoch": 0.3400574505993528,
+      "grad_norm": 932.92578125,
+      "learning_rate": 2.805609355925497e-06,
+      "loss": 35.6964,
+      "step": 84170
+    },
+    {
+      "epoch": 0.3400978518647204,
+      "grad_norm": 119.84386444091797,
+      "learning_rate": 2.8043421627069077e-06,
+      "loss": 43.8795,
+      "step": 84180
+    },
+    {
+      "epoch": 0.34013825313008805,
+      "grad_norm": 854.2145385742188,
+      "learning_rate": 2.8030751441867364e-06,
+      "loss": 44.9602,
+      "step": 84190
+    },
+    {
+      "epoch": 0.3401786543954557,
+      "grad_norm": 480.2217712402344,
+      "learning_rate": 2.8018083004657924e-06,
+      "loss": 43.9618,
+      "step": 84200
+    },
+    {
+      "epoch": 0.3402190556608233,
+      "grad_norm": 560.4413452148438,
+      "learning_rate": 2.800541631644873e-06,
+      "loss": 48.4142,
+      "step": 84210
+    },
+    {
+      "epoch": 0.3402594569261909,
+      "grad_norm": 828.934326171875,
+      "learning_rate": 2.7992751378247627e-06,
+      "loss": 37.967,
+      "step": 84220
+    },
+    {
+      "epoch": 0.34029985819155856,
+      "grad_norm": 965.2238159179688,
+      "learning_rate": 2.79800881910623e-06,
+      "loss": 46.0749,
+      "step": 84230
+    },
+    {
+      "epoch": 0.3403402594569262,
+      "grad_norm": 784.2520141601562,
+      "learning_rate": 2.7967426755900293e-06,
+      "loss": 32.5352,
+      "step": 84240
+    },
+    {
+      "epoch": 0.34038066072229384,
+      "grad_norm": 514.36328125,
+      "learning_rate": 2.795476707376905e-06,
+      "loss": 38.8823,
+      "step": 84250
+    },
+    {
+      "epoch": 0.3404210619876615,
+      "grad_norm": 858.2100219726562,
+      "learning_rate": 2.79421091456758e-06,
+      "loss": 61.8575,
+      "step": 84260
+    },
+    {
+      "epoch": 0.34046146325302906,
+      "grad_norm": 581.9533081054688,
+      "learning_rate": 2.7929452972627685e-06,
+      "loss": 35.4017,
+      "step": 84270
+    },
+    {
+      "epoch": 0.3405018645183967,
+      "grad_norm": 777.8386840820312,
+      "learning_rate": 2.791679855563171e-06,
+      "loss": 31.8655,
+      "step": 84280
+    },
+    {
+      "epoch": 0.34054226578376434,
+      "grad_norm": 607.4931640625,
+      "learning_rate": 2.790414589569473e-06,
+      "loss": 46.2957,
+      "step": 84290
+    },
+    {
+      "epoch": 0.340582667049132,
+      "grad_norm": 752.4151611328125,
+      "learning_rate": 2.789149499382345e-06,
+      "loss": 50.8046,
+      "step": 84300
+    },
+    {
+      "epoch": 0.3406230683144996,
+      "grad_norm": 243.44146728515625,
+      "learning_rate": 2.7878845851024426e-06,
+      "loss": 56.0194,
+      "step": 84310
+    },
+    {
+      "epoch": 0.34066346957986726,
+      "grad_norm": 934.1202392578125,
+      "learning_rate": 2.786619846830414e-06,
+      "loss": 60.4464,
+      "step": 84320
+    },
+    {
+      "epoch": 0.34070387084523485,
+      "grad_norm": 461.3244934082031,
+      "learning_rate": 2.7853552846668865e-06,
+      "loss": 50.0204,
+      "step": 84330
+    },
+    {
+      "epoch": 0.3407442721106025,
+      "grad_norm": 704.9860229492188,
+      "learning_rate": 2.784090898712476e-06,
+      "loss": 39.5185,
+      "step": 84340
+    },
+    {
+      "epoch": 0.3407846733759701,
+      "grad_norm": 841.1322021484375,
+      "learning_rate": 2.7828266890677825e-06,
+      "loss": 48.853,
+      "step": 84350
+    },
+    {
+      "epoch": 0.34082507464133777,
+      "grad_norm": 707.6873779296875,
+      "learning_rate": 2.781562655833393e-06,
+      "loss": 41.3904,
+      "step": 84360
+    },
+    {
+      "epoch": 0.3408654759067054,
+      "grad_norm": 622.2953491210938,
+      "learning_rate": 2.7802987991098816e-06,
+      "loss": 61.1507,
+      "step": 84370
+    },
+    {
+      "epoch": 0.34090587717207305,
+      "grad_norm": 278.2420959472656,
+      "learning_rate": 2.7790351189978083e-06,
+      "loss": 43.5655,
+      "step": 84380
+    },
+    {
+      "epoch": 0.3409462784374407,
+      "grad_norm": 440.2988586425781,
+      "learning_rate": 2.777771615597717e-06,
+      "loss": 34.7467,
+      "step": 84390
+    },
+    {
+      "epoch": 0.34098667970280827,
+      "grad_norm": 794.9293212890625,
+      "learning_rate": 2.776508289010138e-06,
+      "loss": 36.8944,
+      "step": 84400
+    },
+    {
+      "epoch": 0.3410270809681759,
+      "grad_norm": 393.08367919921875,
+      "learning_rate": 2.7752451393355916e-06,
+      "loss": 44.2507,
+      "step": 84410
+    },
+    {
+      "epoch": 0.34106748223354355,
+      "grad_norm": 574.3208618164062,
+      "learning_rate": 2.773982166674582e-06,
+      "loss": 28.7962,
+      "step": 84420
+    },
+    {
+      "epoch": 0.3411078834989112,
+      "grad_norm": 557.8585205078125,
+      "learning_rate": 2.772719371127593e-06,
+      "loss": 36.2364,
+      "step": 84430
+    },
+    {
+      "epoch": 0.34114828476427883,
+      "grad_norm": 377.4037170410156,
+      "learning_rate": 2.771456752795102e-06,
+      "loss": 36.0346,
+      "step": 84440
+    },
+    {
+      "epoch": 0.34118868602964647,
+      "grad_norm": 555.8394775390625,
+      "learning_rate": 2.7701943117775686e-06,
+      "loss": 40.5826,
+      "step": 84450
+    },
+    {
+      "epoch": 0.34122908729501406,
+      "grad_norm": 604.6018676757812,
+      "learning_rate": 2.7689320481754414e-06,
+      "loss": 38.2319,
+      "step": 84460
+    },
+    {
+      "epoch": 0.3412694885603817,
+      "grad_norm": 1093.6312255859375,
+      "learning_rate": 2.7676699620891514e-06,
+      "loss": 46.0036,
+      "step": 84470
+    },
+    {
+      "epoch": 0.34130988982574934,
+      "grad_norm": 574.8224487304688,
+      "learning_rate": 2.7664080536191178e-06,
+      "loss": 44.9199,
+      "step": 84480
+    },
+    {
+      "epoch": 0.341350291091117,
+      "grad_norm": 312.3489074707031,
+      "learning_rate": 2.7651463228657444e-06,
+      "loss": 43.5578,
+      "step": 84490
+    },
+    {
+      "epoch": 0.3413906923564846,
+      "grad_norm": 869.42138671875,
+      "learning_rate": 2.7638847699294196e-06,
+      "loss": 47.0098,
+      "step": 84500
+    },
+    {
+      "epoch": 0.34143109362185226,
+      "grad_norm": 417.8177795410156,
+      "learning_rate": 2.7626233949105252e-06,
+      "loss": 33.5986,
+      "step": 84510
+    },
+    {
+      "epoch": 0.3414714948872199,
+      "grad_norm": 552.8901977539062,
+      "learning_rate": 2.7613621979094173e-06,
+      "loss": 44.5389,
+      "step": 84520
+    },
+    {
+      "epoch": 0.3415118961525875,
+      "grad_norm": 702.9528198242188,
+      "learning_rate": 2.7601011790264454e-06,
+      "loss": 45.9088,
+      "step": 84530
+    },
+    {
+      "epoch": 0.3415522974179551,
+      "grad_norm": 388.2279357910156,
+      "learning_rate": 2.758840338361942e-06,
+      "loss": 52.9619,
+      "step": 84540
+    },
+    {
+      "epoch": 0.34159269868332276,
+      "grad_norm": 322.6661071777344,
+      "learning_rate": 2.7575796760162288e-06,
+      "loss": 33.669,
+      "step": 84550
+    },
+    {
+      "epoch": 0.3416330999486904,
+      "grad_norm": 429.1786193847656,
+      "learning_rate": 2.7563191920896084e-06,
+      "loss": 49.6612,
+      "step": 84560
+    },
+    {
+      "epoch": 0.34167350121405804,
+      "grad_norm": 371.429931640625,
+      "learning_rate": 2.755058886682373e-06,
+      "loss": 32.2629,
+      "step": 84570
+    },
+    {
+      "epoch": 0.3417139024794257,
+      "grad_norm": 440.96319580078125,
+      "learning_rate": 2.753798759894799e-06,
+      "loss": 49.125,
+      "step": 84580
+    },
+    {
+      "epoch": 0.34175430374479326,
+      "grad_norm": 227.27528381347656,
+      "learning_rate": 2.7525388118271495e-06,
+      "loss": 32.9526,
+      "step": 84590
+    },
+    {
+      "epoch": 0.3417947050101609,
+      "grad_norm": 755.2015380859375,
+      "learning_rate": 2.751279042579672e-06,
+      "loss": 39.707,
+      "step": 84600
+    },
+    {
+      "epoch": 0.34183510627552854,
+      "grad_norm": 572.055419921875,
+      "learning_rate": 2.7500194522526007e-06,
+      "loss": 60.7821,
+      "step": 84610
+    },
+    {
+      "epoch": 0.3418755075408962,
+      "grad_norm": 568.6116943359375,
+      "learning_rate": 2.748760040946156e-06,
+      "loss": 32.712,
+      "step": 84620
+    },
+    {
+      "epoch": 0.3419159088062638,
+      "grad_norm": 421.6330261230469,
+      "learning_rate": 2.7475008087605428e-06,
+      "loss": 42.8288,
+      "step": 84630
+    },
+    {
+      "epoch": 0.34195631007163146,
+      "grad_norm": 512.63427734375,
+      "learning_rate": 2.746241755795952e-06,
+      "loss": 52.5931,
+      "step": 84640
+    },
+    {
+      "epoch": 0.34199671133699905,
+      "grad_norm": 439.4118957519531,
+      "learning_rate": 2.7449828821525624e-06,
+      "loss": 38.7648,
+      "step": 84650
+    },
+    {
+      "epoch": 0.3420371126023667,
+      "grad_norm": 661.09765625,
+      "learning_rate": 2.7437241879305354e-06,
+      "loss": 51.7134,
+      "step": 84660
+    },
+    {
+      "epoch": 0.34207751386773433,
+      "grad_norm": 643.4342651367188,
+      "learning_rate": 2.7424656732300193e-06,
+      "loss": 49.4388,
+      "step": 84670
+    },
+    {
+      "epoch": 0.34211791513310197,
+      "grad_norm": 667.8513793945312,
+      "learning_rate": 2.7412073381511495e-06,
+      "loss": 34.0474,
+      "step": 84680
+    },
+    {
+      "epoch": 0.3421583163984696,
+      "grad_norm": 191.4943084716797,
+      "learning_rate": 2.739949182794045e-06,
+      "loss": 41.1864,
+      "step": 84690
+    },
+    {
+      "epoch": 0.34219871766383725,
+      "grad_norm": 528.9896850585938,
+      "learning_rate": 2.7386912072588123e-06,
+      "loss": 42.0747,
+      "step": 84700
+    },
+    {
+      "epoch": 0.3422391189292049,
+      "grad_norm": 625.2055053710938,
+      "learning_rate": 2.737433411645542e-06,
+      "loss": 38.8828,
+      "step": 84710
+    },
+    {
+      "epoch": 0.3422795201945725,
+      "grad_norm": 581.0269775390625,
+      "learning_rate": 2.7361757960543114e-06,
+      "loss": 44.3446,
+      "step": 84720
+    },
+    {
+      "epoch": 0.3423199214599401,
+      "grad_norm": 819.0857543945312,
+      "learning_rate": 2.7349183605851824e-06,
+      "loss": 35.713,
+      "step": 84730
+    },
+    {
+      "epoch": 0.34236032272530775,
+      "grad_norm": 745.9638671875,
+      "learning_rate": 2.733661105338205e-06,
+      "loss": 37.4942,
+      "step": 84740
+    },
+    {
+      "epoch": 0.3424007239906754,
+      "grad_norm": 733.6754150390625,
+      "learning_rate": 2.7324040304134125e-06,
+      "loss": 51.0984,
+      "step": 84750
+    },
+    {
+      "epoch": 0.34244112525604303,
+      "grad_norm": 247.33990478515625,
+      "learning_rate": 2.731147135910824e-06,
+      "loss": 33.1103,
+      "step": 84760
+    },
+    {
+      "epoch": 0.3424815265214107,
+      "grad_norm": 876.4989013671875,
+      "learning_rate": 2.729890421930445e-06,
+      "loss": 43.8137,
+      "step": 84770
+    },
+    {
+      "epoch": 0.34252192778677826,
+      "grad_norm": 448.0574951171875,
+      "learning_rate": 2.7286338885722674e-06,
+      "loss": 34.5072,
+      "step": 84780
+    },
+    {
+      "epoch": 0.3425623290521459,
+      "grad_norm": 567.6174926757812,
+      "learning_rate": 2.7273775359362665e-06,
+      "loss": 29.5398,
+      "step": 84790
+    },
+    {
+      "epoch": 0.34260273031751354,
+      "grad_norm": 390.6498107910156,
+      "learning_rate": 2.7261213641224056e-06,
+      "loss": 45.2981,
+      "step": 84800
+    },
+    {
+      "epoch": 0.3426431315828812,
+      "grad_norm": 614.93359375,
+      "learning_rate": 2.724865373230632e-06,
+      "loss": 38.1664,
+      "step": 84810
+    },
+    {
+      "epoch": 0.3426835328482488,
+      "grad_norm": 1281.5938720703125,
+      "learning_rate": 2.723609563360879e-06,
+      "loss": 45.2071,
+      "step": 84820
+    },
+    {
+      "epoch": 0.34272393411361646,
+      "grad_norm": 871.1010131835938,
+      "learning_rate": 2.7223539346130655e-06,
+      "loss": 46.9967,
+      "step": 84830
+    },
+    {
+      "epoch": 0.3427643353789841,
+      "grad_norm": 876.0147094726562,
+      "learning_rate": 2.7210984870870972e-06,
+      "loss": 41.4255,
+      "step": 84840
+    },
+    {
+      "epoch": 0.3428047366443517,
+      "grad_norm": 284.36773681640625,
+      "learning_rate": 2.7198432208828653e-06,
+      "loss": 33.469,
+      "step": 84850
+    },
+    {
+      "epoch": 0.3428451379097193,
+      "grad_norm": 416.6922607421875,
+      "learning_rate": 2.7185881361002415e-06,
+      "loss": 49.1144,
+      "step": 84860
+    },
+    {
+      "epoch": 0.34288553917508696,
+      "grad_norm": 864.996337890625,
+      "learning_rate": 2.717333232839088e-06,
+      "loss": 44.824,
+      "step": 84870
+    },
+    {
+      "epoch": 0.3429259404404546,
+      "grad_norm": 680.2763671875,
+      "learning_rate": 2.7160785111992546e-06,
+      "loss": 46.4814,
+      "step": 84880
+    },
+    {
+      "epoch": 0.34296634170582224,
+      "grad_norm": 371.52783203125,
+      "learning_rate": 2.7148239712805725e-06,
+      "loss": 42.2676,
+      "step": 84890
+    },
+    {
+      "epoch": 0.3430067429711899,
+      "grad_norm": 485.0791931152344,
+      "learning_rate": 2.713569613182859e-06,
+      "loss": 51.0596,
+      "step": 84900
+    },
+    {
+      "epoch": 0.34304714423655747,
+      "grad_norm": 598.8778076171875,
+      "learning_rate": 2.7123154370059185e-06,
+      "loss": 39.7807,
+      "step": 84910
+    },
+    {
+      "epoch": 0.3430875455019251,
+      "grad_norm": 786.4949340820312,
+      "learning_rate": 2.7110614428495396e-06,
+      "loss": 49.6095,
+      "step": 84920
+    },
+    {
+      "epoch": 0.34312794676729275,
+      "grad_norm": 443.7628479003906,
+      "learning_rate": 2.709807630813497e-06,
+      "loss": 44.824,
+      "step": 84930
+    },
+    {
+      "epoch": 0.3431683480326604,
+      "grad_norm": 480.0895080566406,
+      "learning_rate": 2.7085540009975526e-06,
+      "loss": 42.2488,
+      "step": 84940
+    },
+    {
+      "epoch": 0.343208749298028,
+      "grad_norm": 1161.1788330078125,
+      "learning_rate": 2.707300553501448e-06,
+      "loss": 65.0683,
+      "step": 84950
+    },
+    {
+      "epoch": 0.34324915056339567,
+      "grad_norm": 620.4024658203125,
+      "learning_rate": 2.7060472884249145e-06,
+      "loss": 53.8362,
+      "step": 84960
+    },
+    {
+      "epoch": 0.34328955182876325,
+      "grad_norm": 409.4412841796875,
+      "learning_rate": 2.7047942058676717e-06,
+      "loss": 32.1152,
+      "step": 84970
+    },
+    {
+      "epoch": 0.3433299530941309,
+      "grad_norm": 478.91217041015625,
+      "learning_rate": 2.703541305929421e-06,
+      "loss": 33.8495,
+      "step": 84980
+    },
+    {
+      "epoch": 0.34337035435949853,
+      "grad_norm": 581.8789672851562,
+      "learning_rate": 2.7022885887098492e-06,
+      "loss": 28.2309,
+      "step": 84990
+    },
+    {
+      "epoch": 0.34341075562486617,
+      "grad_norm": 454.3522033691406,
+      "learning_rate": 2.701036054308629e-06,
+      "loss": 41.7049,
+      "step": 85000
+    },
+    {
+      "epoch": 0.3434511568902338,
+      "grad_norm": 500.46746826171875,
+      "learning_rate": 2.699783702825419e-06,
+      "loss": 45.1027,
+      "step": 85010
+    },
+    {
+      "epoch": 0.34349155815560145,
+      "grad_norm": 626.4382934570312,
+      "learning_rate": 2.698531534359864e-06,
+      "loss": 41.5983,
+      "step": 85020
+    },
+    {
+      "epoch": 0.3435319594209691,
+      "grad_norm": 800.3005981445312,
+      "learning_rate": 2.6972795490115944e-06,
+      "loss": 46.0404,
+      "step": 85030
+    },
+    {
+      "epoch": 0.3435723606863367,
+      "grad_norm": 644.60986328125,
+      "learning_rate": 2.6960277468802203e-06,
+      "loss": 52.5239,
+      "step": 85040
+    },
+    {
+      "epoch": 0.3436127619517043,
+      "grad_norm": 430.5956115722656,
+      "learning_rate": 2.694776128065345e-06,
+      "loss": 43.1113,
+      "step": 85050
+    },
+    {
+      "epoch": 0.34365316321707196,
+      "grad_norm": 516.9263305664062,
+      "learning_rate": 2.6935246926665513e-06,
+      "loss": 47.5073,
+      "step": 85060
+    },
+    {
+      "epoch": 0.3436935644824396,
+      "grad_norm": 911.3888549804688,
+      "learning_rate": 2.692273440783415e-06,
+      "loss": 48.9684,
+      "step": 85070
+    },
+    {
+      "epoch": 0.34373396574780724,
+      "grad_norm": 1070.81201171875,
+      "learning_rate": 2.6910223725154903e-06,
+      "loss": 51.5814,
+      "step": 85080
+    },
+    {
+      "epoch": 0.3437743670131749,
+      "grad_norm": 786.39013671875,
+      "learning_rate": 2.6897714879623184e-06,
+      "loss": 49.3049,
+      "step": 85090
+    },
+    {
+      "epoch": 0.34381476827854246,
+      "grad_norm": 882.5459594726562,
+      "learning_rate": 2.688520787223426e-06,
+      "loss": 36.6178,
+      "step": 85100
+    },
+    {
+      "epoch": 0.3438551695439101,
+      "grad_norm": 631.80517578125,
+      "learning_rate": 2.6872702703983287e-06,
+      "loss": 41.4636,
+      "step": 85110
+    },
+    {
+      "epoch": 0.34389557080927774,
+      "grad_norm": 607.47119140625,
+      "learning_rate": 2.6860199375865203e-06,
+      "loss": 32.1181,
+      "step": 85120
+    },
+    {
+      "epoch": 0.3439359720746454,
+      "grad_norm": 692.5158081054688,
+      "learning_rate": 2.6847697888874853e-06,
+      "loss": 43.0668,
+      "step": 85130
+    },
+    {
+      "epoch": 0.343976373340013,
+      "grad_norm": 360.70697021484375,
+      "learning_rate": 2.683519824400693e-06,
+      "loss": 51.3872,
+      "step": 85140
+    },
+    {
+      "epoch": 0.34401677460538066,
+      "grad_norm": 551.6885375976562,
+      "learning_rate": 2.6822700442255965e-06,
+      "loss": 30.0492,
+      "step": 85150
+    },
+    {
+      "epoch": 0.34405717587074824,
+      "grad_norm": 700.1077880859375,
+      "learning_rate": 2.681020448461634e-06,
+      "loss": 43.414,
+      "step": 85160
+    },
+    {
+      "epoch": 0.3440975771361159,
+      "grad_norm": 409.4812316894531,
+      "learning_rate": 2.679771037208234e-06,
+      "loss": 49.9183,
+      "step": 85170
+    },
+    {
+      "epoch": 0.3441379784014835,
+      "grad_norm": 522.0422973632812,
+      "learning_rate": 2.678521810564804e-06,
+      "loss": 33.3543,
+      "step": 85180
+    },
+    {
+      "epoch": 0.34417837966685116,
+      "grad_norm": 918.2327270507812,
+      "learning_rate": 2.6772727686307398e-06,
+      "loss": 55.8203,
+      "step": 85190
+    },
+    {
+      "epoch": 0.3442187809322188,
+      "grad_norm": 532.8182373046875,
+      "learning_rate": 2.676023911505423e-06,
+      "loss": 43.2552,
+      "step": 85200
+    },
+    {
+      "epoch": 0.34425918219758644,
+      "grad_norm": 448.65093994140625,
+      "learning_rate": 2.674775239288216e-06,
+      "loss": 90.8124,
+      "step": 85210
+    },
+    {
+      "epoch": 0.3442995834629541,
+      "grad_norm": 640.6787719726562,
+      "learning_rate": 2.673526752078472e-06,
+      "loss": 29.4502,
+      "step": 85220
+    },
+    {
+      "epoch": 0.34433998472832167,
+      "grad_norm": 580.9517211914062,
+      "learning_rate": 2.6722784499755273e-06,
+      "loss": 44.5188,
+      "step": 85230
+    },
+    {
+      "epoch": 0.3443803859936893,
+      "grad_norm": 927.4952392578125,
+      "learning_rate": 2.6710303330787035e-06,
+      "loss": 42.2006,
+      "step": 85240
+    },
+    {
+      "epoch": 0.34442078725905695,
+      "grad_norm": 873.9500732421875,
+      "learning_rate": 2.6697824014873076e-06,
+      "loss": 45.5853,
+      "step": 85250
+    },
+    {
+      "epoch": 0.3444611885244246,
+      "grad_norm": 587.55126953125,
+      "learning_rate": 2.6685346553006293e-06,
+      "loss": 51.6585,
+      "step": 85260
+    },
+    {
+      "epoch": 0.34450158978979223,
+      "grad_norm": 814.5745239257812,
+      "learning_rate": 2.6672870946179506e-06,
+      "loss": 34.1723,
+      "step": 85270
+    },
+    {
+      "epoch": 0.34454199105515987,
+      "grad_norm": 784.3179931640625,
+      "learning_rate": 2.6660397195385344e-06,
+      "loss": 46.9616,
+      "step": 85280
+    },
+    {
+      "epoch": 0.34458239232052745,
+      "grad_norm": 558.2401733398438,
+      "learning_rate": 2.664792530161624e-06,
+      "loss": 35.8878,
+      "step": 85290
+    },
+    {
+      "epoch": 0.3446227935858951,
+      "grad_norm": 741.8731079101562,
+      "learning_rate": 2.6635455265864553e-06,
+      "loss": 39.7892,
+      "step": 85300
+    },
+    {
+      "epoch": 0.34466319485126273,
+      "grad_norm": 650.9552001953125,
+      "learning_rate": 2.662298708912246e-06,
+      "loss": 54.3428,
+      "step": 85310
+    },
+    {
+      "epoch": 0.3447035961166304,
+      "grad_norm": 0.0,
+      "learning_rate": 2.6610520772382e-06,
+      "loss": 31.7625,
+      "step": 85320
+    },
+    {
+      "epoch": 0.344743997381998,
+      "grad_norm": 1837.4625244140625,
+      "learning_rate": 2.659805631663505e-06,
+      "loss": 54.597,
+      "step": 85330
+    },
+    {
+      "epoch": 0.34478439864736565,
+      "grad_norm": 561.35302734375,
+      "learning_rate": 2.658559372287337e-06,
+      "loss": 39.6425,
+      "step": 85340
+    },
+    {
+      "epoch": 0.3448247999127333,
+      "grad_norm": 373.4984130859375,
+      "learning_rate": 2.6573132992088534e-06,
+      "loss": 25.0325,
+      "step": 85350
+    },
+    {
+      "epoch": 0.3448652011781009,
+      "grad_norm": 1178.658935546875,
+      "learning_rate": 2.656067412527197e-06,
+      "loss": 59.2496,
+      "step": 85360
+    },
+    {
+      "epoch": 0.3449056024434685,
+      "grad_norm": 391.9385681152344,
+      "learning_rate": 2.6548217123415033e-06,
+      "loss": 51.5498,
+      "step": 85370
+    },
+    {
+      "epoch": 0.34494600370883616,
+      "grad_norm": 933.973876953125,
+      "learning_rate": 2.6535761987508813e-06,
+      "loss": 43.9957,
+      "step": 85380
+    },
+    {
+      "epoch": 0.3449864049742038,
+      "grad_norm": 526.8036499023438,
+      "learning_rate": 2.652330871854433e-06,
+      "loss": 48.7626,
+      "step": 85390
+    },
+    {
+      "epoch": 0.34502680623957144,
+      "grad_norm": 312.12994384765625,
+      "learning_rate": 2.651085731751242e-06,
+      "loss": 47.8764,
+      "step": 85400
+    },
+    {
+      "epoch": 0.3450672075049391,
+      "grad_norm": 357.3439025878906,
+      "learning_rate": 2.6498407785403794e-06,
+      "loss": 35.1316,
+      "step": 85410
+    },
+    {
+      "epoch": 0.34510760877030666,
+      "grad_norm": 447.86126708984375,
+      "learning_rate": 2.648596012320901e-06,
+      "loss": 44.3176,
+      "step": 85420
+    },
+    {
+      "epoch": 0.3451480100356743,
+      "grad_norm": 734.8046264648438,
+      "learning_rate": 2.647351433191846e-06,
+      "loss": 32.598,
+      "step": 85430
+    },
+    {
+      "epoch": 0.34518841130104194,
+      "grad_norm": 710.0639038085938,
+      "learning_rate": 2.64610704125224e-06,
+      "loss": 41.4129,
+      "step": 85440
+    },
+    {
+      "epoch": 0.3452288125664096,
+      "grad_norm": 1368.0191650390625,
+      "learning_rate": 2.644862836601092e-06,
+      "loss": 62.2333,
+      "step": 85450
+    },
+    {
+      "epoch": 0.3452692138317772,
+      "grad_norm": 411.8841857910156,
+      "learning_rate": 2.6436188193374035e-06,
+      "loss": 32.3257,
+      "step": 85460
+    },
+    {
+      "epoch": 0.34530961509714486,
+      "grad_norm": 720.62890625,
+      "learning_rate": 2.6423749895601494e-06,
+      "loss": 38.4964,
+      "step": 85470
+    },
+    {
+      "epoch": 0.34535001636251245,
+      "grad_norm": 444.2539367675781,
+      "learning_rate": 2.6411313473682966e-06,
+      "loss": 52.2975,
+      "step": 85480
+    },
+    {
+      "epoch": 0.3453904176278801,
+      "grad_norm": 674.4296264648438,
+      "learning_rate": 2.6398878928607973e-06,
+      "loss": 46.51,
+      "step": 85490
+    },
+    {
+      "epoch": 0.3454308188932477,
+      "grad_norm": 707.7427978515625,
+      "learning_rate": 2.6386446261365874e-06,
+      "loss": 42.681,
+      "step": 85500
+    },
+    {
+      "epoch": 0.34547122015861537,
+      "grad_norm": 959.9139404296875,
+      "learning_rate": 2.6374015472945868e-06,
+      "loss": 42.9944,
+      "step": 85510
+    },
+    {
+      "epoch": 0.345511621423983,
+      "grad_norm": 620.0403442382812,
+      "learning_rate": 2.6361586564337023e-06,
+      "loss": 43.4066,
+      "step": 85520
+    },
+    {
+      "epoch": 0.34555202268935065,
+      "grad_norm": 894.035400390625,
+      "learning_rate": 2.6349159536528245e-06,
+      "loss": 44.0186,
+      "step": 85530
+    },
+    {
+      "epoch": 0.3455924239547183,
+      "grad_norm": 703.9459228515625,
+      "learning_rate": 2.633673439050831e-06,
+      "loss": 42.7953,
+      "step": 85540
+    },
+    {
+      "epoch": 0.34563282522008587,
+      "grad_norm": 1000.5932006835938,
+      "learning_rate": 2.6324311127265812e-06,
+      "loss": 54.5857,
+      "step": 85550
+    },
+    {
+      "epoch": 0.3456732264854535,
+      "grad_norm": 4672.994140625,
+      "learning_rate": 2.6311889747789225e-06,
+      "loss": 52.1665,
+      "step": 85560
+    },
+    {
+      "epoch": 0.34571362775082115,
+      "grad_norm": 723.791748046875,
+      "learning_rate": 2.6299470253066863e-06,
+      "loss": 44.0532,
+      "step": 85570
+    },
+    {
+      "epoch": 0.3457540290161888,
+      "grad_norm": 531.83740234375,
+      "learning_rate": 2.628705264408687e-06,
+      "loss": 41.6017,
+      "step": 85580
+    },
+    {
+      "epoch": 0.34579443028155643,
+      "grad_norm": 622.5613403320312,
+      "learning_rate": 2.6274636921837272e-06,
+      "loss": 42.075,
+      "step": 85590
+    },
+    {
+      "epoch": 0.34583483154692407,
+      "grad_norm": 635.7496948242188,
+      "learning_rate": 2.626222308730594e-06,
+      "loss": 44.8181,
+      "step": 85600
+    },
+    {
+      "epoch": 0.34587523281229166,
+      "grad_norm": 520.9039306640625,
+      "learning_rate": 2.6249811141480564e-06,
+      "loss": 37.7106,
+      "step": 85610
+    },
+    {
+      "epoch": 0.3459156340776593,
+      "grad_norm": 809.7235717773438,
+      "learning_rate": 2.6237401085348723e-06,
+      "loss": 55.4172,
+      "step": 85620
+    },
+    {
+      "epoch": 0.34595603534302694,
+      "grad_norm": 748.8391723632812,
+      "learning_rate": 2.6224992919897817e-06,
+      "loss": 60.3569,
+      "step": 85630
+    },
+    {
+      "epoch": 0.3459964366083946,
+      "grad_norm": 422.0188903808594,
+      "learning_rate": 2.6212586646115114e-06,
+      "loss": 32.4093,
+      "step": 85640
+    },
+    {
+      "epoch": 0.3460368378737622,
+      "grad_norm": 386.18365478515625,
+      "learning_rate": 2.620018226498772e-06,
+      "loss": 56.6944,
+      "step": 85650
+    },
+    {
+      "epoch": 0.34607723913912986,
+      "grad_norm": 704.84765625,
+      "learning_rate": 2.61877797775026e-06,
+      "loss": 39.3209,
+      "step": 85660
+    },
+    {
+      "epoch": 0.3461176404044975,
+      "grad_norm": 497.8809509277344,
+      "learning_rate": 2.6175379184646565e-06,
+      "loss": 33.6977,
+      "step": 85670
+    },
+    {
+      "epoch": 0.3461580416698651,
+      "grad_norm": 733.8531494140625,
+      "learning_rate": 2.616298048740626e-06,
+      "loss": 41.8165,
+      "step": 85680
+    },
+    {
+      "epoch": 0.3461984429352327,
+      "grad_norm": 733.565185546875,
+      "learning_rate": 2.6150583686768203e-06,
+      "loss": 33.9747,
+      "step": 85690
+    },
+    {
+      "epoch": 0.34623884420060036,
+      "grad_norm": 598.9244384765625,
+      "learning_rate": 2.6138188783718745e-06,
+      "loss": 57.9881,
+      "step": 85700
+    },
+    {
+      "epoch": 0.346279245465968,
+      "grad_norm": 471.6096496582031,
+      "learning_rate": 2.6125795779244125e-06,
+      "loss": 37.6456,
+      "step": 85710
+    },
+    {
+      "epoch": 0.34631964673133564,
+      "grad_norm": 816.8072509765625,
+      "learning_rate": 2.611340467433031e-06,
+      "loss": 48.4208,
+      "step": 85720
+    },
+    {
+      "epoch": 0.3463600479967033,
+      "grad_norm": 368.94610595703125,
+      "learning_rate": 2.61010154699633e-06,
+      "loss": 37.9098,
+      "step": 85730
+    },
+    {
+      "epoch": 0.34640044926207086,
+      "grad_norm": 1260.583984375,
+      "learning_rate": 2.6088628167128794e-06,
+      "loss": 55.5421,
+      "step": 85740
+    },
+    {
+      "epoch": 0.3464408505274385,
+      "grad_norm": 934.0879516601562,
+      "learning_rate": 2.607624276681241e-06,
+      "loss": 43.342,
+      "step": 85750
+    },
+    {
+      "epoch": 0.34648125179280614,
+      "grad_norm": 547.6620483398438,
+      "learning_rate": 2.6063859269999594e-06,
+      "loss": 42.7008,
+      "step": 85760
+    },
+    {
+      "epoch": 0.3465216530581738,
+      "grad_norm": 475.74371337890625,
+      "learning_rate": 2.605147767767564e-06,
+      "loss": 34.0014,
+      "step": 85770
+    },
+    {
+      "epoch": 0.3465620543235414,
+      "grad_norm": 366.348388671875,
+      "learning_rate": 2.6039097990825703e-06,
+      "loss": 31.9253,
+      "step": 85780
+    },
+    {
+      "epoch": 0.34660245558890906,
+      "grad_norm": 999.9945678710938,
+      "learning_rate": 2.602672021043477e-06,
+      "loss": 84.3831,
+      "step": 85790
+    },
+    {
+      "epoch": 0.34664285685427665,
+      "grad_norm": 1005.42578125,
+      "learning_rate": 2.601434433748771e-06,
+      "loss": 37.5773,
+      "step": 85800
+    },
+    {
+      "epoch": 0.3466832581196443,
+      "grad_norm": 458.24029541015625,
+      "learning_rate": 2.600197037296917e-06,
+      "loss": 50.0763,
+      "step": 85810
+    },
+    {
+      "epoch": 0.34672365938501193,
+      "grad_norm": 732.8392944335938,
+      "learning_rate": 2.5989598317863694e-06,
+      "loss": 46.9295,
+      "step": 85820
+    },
+    {
+      "epoch": 0.34676406065037957,
+      "grad_norm": 282.8758544921875,
+      "learning_rate": 2.59772281731557e-06,
+      "loss": 38.0548,
+      "step": 85830
+    },
+    {
+      "epoch": 0.3468044619157472,
+      "grad_norm": 1127.44384765625,
+      "learning_rate": 2.5964859939829423e-06,
+      "loss": 52.8228,
+      "step": 85840
+    },
+    {
+      "epoch": 0.34684486318111485,
+      "grad_norm": 1677.3736572265625,
+      "learning_rate": 2.595249361886892e-06,
+      "loss": 53.9349,
+      "step": 85850
+    },
+    {
+      "epoch": 0.3468852644464825,
+      "grad_norm": 543.578369140625,
+      "learning_rate": 2.5940129211258147e-06,
+      "loss": 42.8509,
+      "step": 85860
+    },
+    {
+      "epoch": 0.3469256657118501,
+      "grad_norm": 700.8523559570312,
+      "learning_rate": 2.5927766717980873e-06,
+      "loss": 51.0615,
+      "step": 85870
+    },
+    {
+      "epoch": 0.3469660669772177,
+      "grad_norm": 297.7542419433594,
+      "learning_rate": 2.5915406140020738e-06,
+      "loss": 29.6853,
+      "step": 85880
+    },
+    {
+      "epoch": 0.34700646824258535,
+      "grad_norm": 316.5313415527344,
+      "learning_rate": 2.590304747836119e-06,
+      "loss": 34.6187,
+      "step": 85890
+    },
+    {
+      "epoch": 0.347046869507953,
+      "grad_norm": 995.6156005859375,
+      "learning_rate": 2.5890690733985555e-06,
+      "loss": 48.3624,
+      "step": 85900
+    },
+    {
+      "epoch": 0.34708727077332063,
+      "grad_norm": 438.5686950683594,
+      "learning_rate": 2.5878335907876997e-06,
+      "loss": 51.1746,
+      "step": 85910
+    },
+    {
+      "epoch": 0.3471276720386883,
+      "grad_norm": 1318.769775390625,
+      "learning_rate": 2.5865983001018567e-06,
+      "loss": 44.332,
+      "step": 85920
+    },
+    {
+      "epoch": 0.34716807330405586,
+      "grad_norm": 188.23451232910156,
+      "learning_rate": 2.5853632014393108e-06,
+      "loss": 28.3907,
+      "step": 85930
+    },
+    {
+      "epoch": 0.3472084745694235,
+      "grad_norm": 539.7623291015625,
+      "learning_rate": 2.584128294898334e-06,
+      "loss": 43.8977,
+      "step": 85940
+    },
+    {
+      "epoch": 0.34724887583479114,
+      "grad_norm": 445.7605285644531,
+      "learning_rate": 2.5828935805771804e-06,
+      "loss": 46.062,
+      "step": 85950
+    },
+    {
+      "epoch": 0.3472892771001588,
+      "grad_norm": 504.9610290527344,
+      "learning_rate": 2.581659058574092e-06,
+      "loss": 77.293,
+      "step": 85960
+    },
+    {
+      "epoch": 0.3473296783655264,
+      "grad_norm": 569.9614868164062,
+      "learning_rate": 2.580424728987296e-06,
+      "loss": 42.8085,
+      "step": 85970
+    },
+    {
+      "epoch": 0.34737007963089406,
+      "grad_norm": 786.86767578125,
+      "learning_rate": 2.5791905919149973e-06,
+      "loss": 35.5293,
+      "step": 85980
+    },
+    {
+      "epoch": 0.3474104808962617,
+      "grad_norm": 899.1405639648438,
+      "learning_rate": 2.5779566474553934e-06,
+      "loss": 54.769,
+      "step": 85990
+    },
+    {
+      "epoch": 0.3474508821616293,
+      "grad_norm": 331.9537353515625,
+      "learning_rate": 2.5767228957066635e-06,
+      "loss": 31.8642,
+      "step": 86000
+    },
+    {
+      "epoch": 0.3474912834269969,
+      "grad_norm": 952.2335205078125,
+      "learning_rate": 2.5754893367669697e-06,
+      "loss": 36.8524,
+      "step": 86010
+    },
+    {
+      "epoch": 0.34753168469236456,
+      "grad_norm": 415.8359680175781,
+      "learning_rate": 2.5742559707344638e-06,
+      "loss": 35.8399,
+      "step": 86020
+    },
+    {
+      "epoch": 0.3475720859577322,
+      "grad_norm": 397.8514709472656,
+      "learning_rate": 2.573022797707278e-06,
+      "loss": 42.1008,
+      "step": 86030
+    },
+    {
+      "epoch": 0.34761248722309984,
+      "grad_norm": 588.5228271484375,
+      "learning_rate": 2.57178981778353e-06,
+      "loss": 44.473,
+      "step": 86040
+    },
+    {
+      "epoch": 0.3476528884884675,
+      "grad_norm": 790.072998046875,
+      "learning_rate": 2.5705570310613215e-06,
+      "loss": 39.8272,
+      "step": 86050
+    },
+    {
+      "epoch": 0.34769328975383507,
+      "grad_norm": 400.7402038574219,
+      "learning_rate": 2.5693244376387435e-06,
+      "loss": 30.9018,
+      "step": 86060
+    },
+    {
+      "epoch": 0.3477336910192027,
+      "grad_norm": 293.9530944824219,
+      "learning_rate": 2.568092037613862e-06,
+      "loss": 25.6735,
+      "step": 86070
+    },
+    {
+      "epoch": 0.34777409228457035,
+      "grad_norm": 873.0491333007812,
+      "learning_rate": 2.566859831084736e-06,
+      "loss": 60.489,
+      "step": 86080
+    },
+    {
+      "epoch": 0.347814493549938,
+      "grad_norm": 487.86639404296875,
+      "learning_rate": 2.5656278181494072e-06,
+      "loss": 34.5589,
+      "step": 86090
+    },
+    {
+      "epoch": 0.3478548948153056,
+      "grad_norm": 350.3863830566406,
+      "learning_rate": 2.5643959989058997e-06,
+      "loss": 37.4708,
+      "step": 86100
+    },
+    {
+      "epoch": 0.34789529608067327,
+      "grad_norm": 1139.77294921875,
+      "learning_rate": 2.563164373452224e-06,
+      "loss": 34.991,
+      "step": 86110
+    },
+    {
+      "epoch": 0.34793569734604085,
+      "grad_norm": 422.9922180175781,
+      "learning_rate": 2.561932941886377e-06,
+      "loss": 63.9281,
+      "step": 86120
+    },
+    {
+      "epoch": 0.3479760986114085,
+      "grad_norm": 423.64495849609375,
+      "learning_rate": 2.560701704306336e-06,
+      "loss": 39.6468,
+      "step": 86130
+    },
+    {
+      "epoch": 0.34801649987677613,
+      "grad_norm": 496.6314392089844,
+      "learning_rate": 2.5594706608100677e-06,
+      "loss": 35.5593,
+      "step": 86140
+    },
+    {
+      "epoch": 0.34805690114214377,
+      "grad_norm": 201.16738891601562,
+      "learning_rate": 2.5582398114955164e-06,
+      "loss": 39.8651,
+      "step": 86150
+    },
+    {
+      "epoch": 0.3480973024075114,
+      "grad_norm": 464.44390869140625,
+      "learning_rate": 2.5570091564606182e-06,
+      "loss": 41.5413,
+      "step": 86160
+    },
+    {
+      "epoch": 0.34813770367287905,
+      "grad_norm": 518.0660400390625,
+      "learning_rate": 2.555778695803288e-06,
+      "loss": 46.4022,
+      "step": 86170
+    },
+    {
+      "epoch": 0.3481781049382467,
+      "grad_norm": 617.8944091796875,
+      "learning_rate": 2.554548429621431e-06,
+      "loss": 53.7246,
+      "step": 86180
+    },
+    {
+      "epoch": 0.3482185062036143,
+      "grad_norm": 409.9566345214844,
+      "learning_rate": 2.5533183580129317e-06,
+      "loss": 43.618,
+      "step": 86190
+    },
+    {
+      "epoch": 0.3482589074689819,
+      "grad_norm": 608.5611572265625,
+      "learning_rate": 2.5520884810756614e-06,
+      "loss": 39.9297,
+      "step": 86200
+    },
+    {
+      "epoch": 0.34829930873434956,
+      "grad_norm": 672.5448608398438,
+      "learning_rate": 2.550858798907475e-06,
+      "loss": 39.3397,
+      "step": 86210
+    },
+    {
+      "epoch": 0.3483397099997172,
+      "grad_norm": 183.29920959472656,
+      "learning_rate": 2.5496293116062154e-06,
+      "loss": 41.2196,
+      "step": 86220
+    },
+    {
+      "epoch": 0.34838011126508484,
+      "grad_norm": 903.1591796875,
+      "learning_rate": 2.5484000192697078e-06,
+      "loss": 56.9694,
+      "step": 86230
+    },
+    {
+      "epoch": 0.3484205125304525,
+      "grad_norm": 2445.053955078125,
+      "learning_rate": 2.5471709219957573e-06,
+      "loss": 68.2111,
+      "step": 86240
+    },
+    {
+      "epoch": 0.34846091379582006,
+      "grad_norm": 504.24951171875,
+      "learning_rate": 2.5459420198821604e-06,
+      "loss": 57.0127,
+      "step": 86250
+    },
+    {
+      "epoch": 0.3485013150611877,
+      "grad_norm": 447.6020812988281,
+      "learning_rate": 2.5447133130266937e-06,
+      "loss": 38.3485,
+      "step": 86260
+    },
+    {
+      "epoch": 0.34854171632655534,
+      "grad_norm": 619.9303588867188,
+      "learning_rate": 2.5434848015271206e-06,
+      "loss": 39.8455,
+      "step": 86270
+    },
+    {
+      "epoch": 0.348582117591923,
+      "grad_norm": 1106.0968017578125,
+      "learning_rate": 2.542256485481188e-06,
+      "loss": 51.2592,
+      "step": 86280
+    },
+    {
+      "epoch": 0.3486225188572906,
+      "grad_norm": 550.4239501953125,
+      "learning_rate": 2.5410283649866272e-06,
+      "loss": 56.5999,
+      "step": 86290
+    },
+    {
+      "epoch": 0.34866292012265826,
+      "grad_norm": 353.9264221191406,
+      "learning_rate": 2.539800440141154e-06,
+      "loss": 38.7674,
+      "step": 86300
+    },
+    {
+      "epoch": 0.3487033213880259,
+      "grad_norm": 819.6434326171875,
+      "learning_rate": 2.5385727110424697e-06,
+      "loss": 38.6414,
+      "step": 86310
+    },
+    {
+      "epoch": 0.3487437226533935,
+      "grad_norm": 751.7317504882812,
+      "learning_rate": 2.5373451777882575e-06,
+      "loss": 50.2327,
+      "step": 86320
+    },
+    {
+      "epoch": 0.3487841239187611,
+      "grad_norm": 264.96881103515625,
+      "learning_rate": 2.5361178404761876e-06,
+      "loss": 34.7408,
+      "step": 86330
+    },
+    {
+      "epoch": 0.34882452518412876,
+      "grad_norm": 508.8009948730469,
+      "learning_rate": 2.534890699203914e-06,
+      "loss": 37.9676,
+      "step": 86340
+    },
+    {
+      "epoch": 0.3488649264494964,
+      "grad_norm": 450.41363525390625,
+      "learning_rate": 2.533663754069074e-06,
+      "loss": 36.202,
+      "step": 86350
+    },
+    {
+      "epoch": 0.34890532771486404,
+      "grad_norm": 1836.413330078125,
+      "learning_rate": 2.5324370051692905e-06,
+      "loss": 51.1927,
+      "step": 86360
+    },
+    {
+      "epoch": 0.3489457289802317,
+      "grad_norm": 431.05322265625,
+      "learning_rate": 2.5312104526021687e-06,
+      "loss": 37.8777,
+      "step": 86370
+    },
+    {
+      "epoch": 0.34898613024559927,
+      "grad_norm": 606.6944580078125,
+      "learning_rate": 2.529984096465302e-06,
+      "loss": 46.8281,
+      "step": 86380
+    },
+    {
+      "epoch": 0.3490265315109669,
+      "grad_norm": 449.3538513183594,
+      "learning_rate": 2.528757936856264e-06,
+      "loss": 47.6959,
+      "step": 86390
+    },
+    {
+      "epoch": 0.34906693277633455,
+      "grad_norm": 682.8927001953125,
+      "learning_rate": 2.527531973872617e-06,
+      "loss": 44.2995,
+      "step": 86400
+    },
+    {
+      "epoch": 0.3491073340417022,
+      "grad_norm": 621.0684814453125,
+      "learning_rate": 2.5263062076119026e-06,
+      "loss": 55.0341,
+      "step": 86410
+    },
+    {
+      "epoch": 0.34914773530706983,
+      "grad_norm": 959.4788818359375,
+      "learning_rate": 2.525080638171651e-06,
+      "loss": 38.6664,
+      "step": 86420
+    },
+    {
+      "epoch": 0.34918813657243747,
+      "grad_norm": 329.4792175292969,
+      "learning_rate": 2.5238552656493743e-06,
+      "loss": 54.4498,
+      "step": 86430
+    },
+    {
+      "epoch": 0.34922853783780505,
+      "grad_norm": 547.2073974609375,
+      "learning_rate": 2.52263009014257e-06,
+      "loss": 50.6669,
+      "step": 86440
+    },
+    {
+      "epoch": 0.3492689391031727,
+      "grad_norm": 348.137939453125,
+      "learning_rate": 2.5214051117487205e-06,
+      "loss": 53.2841,
+      "step": 86450
+    },
+    {
+      "epoch": 0.34930934036854033,
+      "grad_norm": 838.4050903320312,
+      "learning_rate": 2.52018033056529e-06,
+      "loss": 46.9813,
+      "step": 86460
+    },
+    {
+      "epoch": 0.349349741633908,
+      "grad_norm": 662.2925415039062,
+      "learning_rate": 2.5189557466897306e-06,
+      "loss": 41.6777,
+      "step": 86470
+    },
+    {
+      "epoch": 0.3493901428992756,
+      "grad_norm": 541.9384155273438,
+      "learning_rate": 2.517731360219476e-06,
+      "loss": 41.1058,
+      "step": 86480
+    },
+    {
+      "epoch": 0.34943054416464325,
+      "grad_norm": 446.6821594238281,
+      "learning_rate": 2.5165071712519447e-06,
+      "loss": 46.4978,
+      "step": 86490
+    },
+    {
+      "epoch": 0.3494709454300109,
+      "grad_norm": 493.4598083496094,
+      "learning_rate": 2.51528317988454e-06,
+      "loss": 34.3595,
+      "step": 86500
+    },
+    {
+      "epoch": 0.3495113466953785,
+      "grad_norm": 368.7260437011719,
+      "learning_rate": 2.5140593862146496e-06,
+      "loss": 33.893,
+      "step": 86510
+    },
+    {
+      "epoch": 0.3495517479607461,
+      "grad_norm": 360.5600891113281,
+      "learning_rate": 2.512835790339645e-06,
+      "loss": 91.0386,
+      "step": 86520
+    },
+    {
+      "epoch": 0.34959214922611376,
+      "grad_norm": 759.7279052734375,
+      "learning_rate": 2.5116123923568815e-06,
+      "loss": 61.5651,
+      "step": 86530
+    },
+    {
+      "epoch": 0.3496325504914814,
+      "grad_norm": 482.3156433105469,
+      "learning_rate": 2.5103891923637e-06,
+      "loss": 35.2281,
+      "step": 86540
+    },
+    {
+      "epoch": 0.34967295175684904,
+      "grad_norm": 1186.7445068359375,
+      "learning_rate": 2.509166190457425e-06,
+      "loss": 40.6471,
+      "step": 86550
+    },
+    {
+      "epoch": 0.3497133530222167,
+      "grad_norm": 926.4749145507812,
+      "learning_rate": 2.5079433867353646e-06,
+      "loss": 52.9848,
+      "step": 86560
+    },
+    {
+      "epoch": 0.34975375428758426,
+      "grad_norm": 282.3522644042969,
+      "learning_rate": 2.5067207812948123e-06,
+      "loss": 35.276,
+      "step": 86570
+    },
+    {
+      "epoch": 0.3497941555529519,
+      "grad_norm": 592.4163818359375,
+      "learning_rate": 2.505498374233044e-06,
+      "loss": 50.2927,
+      "step": 86580
+    },
+    {
+      "epoch": 0.34983455681831954,
+      "grad_norm": 749.9614868164062,
+      "learning_rate": 2.5042761656473226e-06,
+      "loss": 58.0474,
+      "step": 86590
+    },
+    {
+      "epoch": 0.3498749580836872,
+      "grad_norm": 476.7281799316406,
+      "learning_rate": 2.503054155634893e-06,
+      "loss": 46.2431,
+      "step": 86600
+    },
+    {
+      "epoch": 0.3499153593490548,
+      "grad_norm": 544.9255981445312,
+      "learning_rate": 2.5018323442929844e-06,
+      "loss": 36.0948,
+      "step": 86610
+    },
+    {
+      "epoch": 0.34995576061442246,
+      "grad_norm": 615.7764892578125,
+      "learning_rate": 2.500610731718811e-06,
+      "loss": 44.5007,
+      "step": 86620
+    },
+    {
+      "epoch": 0.3499961618797901,
+      "grad_norm": 615.6270141601562,
+      "learning_rate": 2.499389318009571e-06,
+      "loss": 32.2692,
+      "step": 86630
+    },
+    {
+      "epoch": 0.3500365631451577,
+      "grad_norm": 724.1041259765625,
+      "learning_rate": 2.4981681032624473e-06,
+      "loss": 47.3459,
+      "step": 86640
+    },
+    {
+      "epoch": 0.3500769644105253,
+      "grad_norm": 1264.620849609375,
+      "learning_rate": 2.4969470875746055e-06,
+      "loss": 41.2133,
+      "step": 86650
+    },
+    {
+      "epoch": 0.35011736567589297,
+      "grad_norm": 616.659423828125,
+      "learning_rate": 2.495726271043198e-06,
+      "loss": 41.7627,
+      "step": 86660
+    },
+    {
+      "epoch": 0.3501577669412606,
+      "grad_norm": 519.1875610351562,
+      "learning_rate": 2.4945056537653545e-06,
+      "loss": 54.5645,
+      "step": 86670
+    },
+    {
+      "epoch": 0.35019816820662825,
+      "grad_norm": 255.1110382080078,
+      "learning_rate": 2.493285235838199e-06,
+      "loss": 35.6709,
+      "step": 86680
+    },
+    {
+      "epoch": 0.3502385694719959,
+      "grad_norm": 702.441650390625,
+      "learning_rate": 2.492065017358834e-06,
+      "loss": 46.2376,
+      "step": 86690
+    },
+    {
+      "epoch": 0.35027897073736347,
+      "grad_norm": 773.6206665039062,
+      "learning_rate": 2.4908449984243448e-06,
+      "loss": 47.9963,
+      "step": 86700
+    },
+    {
+      "epoch": 0.3503193720027311,
+      "grad_norm": 1610.5216064453125,
+      "learning_rate": 2.4896251791318036e-06,
+      "loss": 73.3314,
+      "step": 86710
+    },
+    {
+      "epoch": 0.35035977326809875,
+      "grad_norm": 515.2232666015625,
+      "learning_rate": 2.4884055595782666e-06,
+      "loss": 62.2861,
+      "step": 86720
+    },
+    {
+      "epoch": 0.3504001745334664,
+      "grad_norm": 603.9746704101562,
+      "learning_rate": 2.487186139860772e-06,
+      "loss": 69.002,
+      "step": 86730
+    },
+    {
+      "epoch": 0.35044057579883403,
+      "grad_norm": 410.34613037109375,
+      "learning_rate": 2.485966920076346e-06,
+      "loss": 29.3686,
+      "step": 86740
+    },
+    {
+      "epoch": 0.35048097706420167,
+      "grad_norm": 396.1544189453125,
+      "learning_rate": 2.4847479003219926e-06,
+      "loss": 29.2824,
+      "step": 86750
+    },
+    {
+      "epoch": 0.35052137832956926,
+      "grad_norm": 501.6451110839844,
+      "learning_rate": 2.4835290806947047e-06,
+      "loss": 51.5449,
+      "step": 86760
+    },
+    {
+      "epoch": 0.3505617795949369,
+      "grad_norm": 617.4530639648438,
+      "learning_rate": 2.4823104612914578e-06,
+      "loss": 41.4633,
+      "step": 86770
+    },
+    {
+      "epoch": 0.35060218086030454,
+      "grad_norm": 1717.3043212890625,
+      "learning_rate": 2.4810920422092137e-06,
+      "loss": 58.29,
+      "step": 86780
+    },
+    {
+      "epoch": 0.3506425821256722,
+      "grad_norm": 481.8612365722656,
+      "learning_rate": 2.4798738235449164e-06,
+      "loss": 52.251,
+      "step": 86790
+    },
+    {
+      "epoch": 0.3506829833910398,
+      "grad_norm": 303.1776123046875,
+      "learning_rate": 2.478655805395493e-06,
+      "loss": 29.4278,
+      "step": 86800
+    },
+    {
+      "epoch": 0.35072338465640746,
+      "grad_norm": 697.6649169921875,
+      "learning_rate": 2.477437987857856e-06,
+      "loss": 28.0373,
+      "step": 86810
+    },
+    {
+      "epoch": 0.3507637859217751,
+      "grad_norm": 433.3942565917969,
+      "learning_rate": 2.4762203710289008e-06,
+      "loss": 43.1259,
+      "step": 86820
+    },
+    {
+      "epoch": 0.3508041871871427,
+      "grad_norm": 377.6124267578125,
+      "learning_rate": 2.4750029550055098e-06,
+      "loss": 41.783,
+      "step": 86830
+    },
+    {
+      "epoch": 0.3508445884525103,
+      "grad_norm": 617.4815063476562,
+      "learning_rate": 2.473785739884544e-06,
+      "loss": 31.3333,
+      "step": 86840
+    },
+    {
+      "epoch": 0.35088498971787796,
+      "grad_norm": 614.5011596679688,
+      "learning_rate": 2.4725687257628533e-06,
+      "loss": 49.7944,
+      "step": 86850
+    },
+    {
+      "epoch": 0.3509253909832456,
+      "grad_norm": 322.9035949707031,
+      "learning_rate": 2.47135191273727e-06,
+      "loss": 47.4259,
+      "step": 86860
+    },
+    {
+      "epoch": 0.35096579224861324,
+      "grad_norm": 1108.1063232421875,
+      "learning_rate": 2.4701353009046075e-06,
+      "loss": 54.4417,
+      "step": 86870
+    },
+    {
+      "epoch": 0.3510061935139809,
+      "grad_norm": 719.0805053710938,
+      "learning_rate": 2.4689188903616707e-06,
+      "loss": 39.9168,
+      "step": 86880
+    },
+    {
+      "epoch": 0.35104659477934846,
+      "grad_norm": 545.470947265625,
+      "learning_rate": 2.467702681205241e-06,
+      "loss": 51.4684,
+      "step": 86890
+    },
+    {
+      "epoch": 0.3510869960447161,
+      "grad_norm": 158.2139129638672,
+      "learning_rate": 2.4664866735320886e-06,
+      "loss": 31.18,
+      "step": 86900
+    },
+    {
+      "epoch": 0.35112739731008374,
+      "grad_norm": 437.4032287597656,
+      "learning_rate": 2.4652708674389636e-06,
+      "loss": 29.5146,
+      "step": 86910
+    },
+    {
+      "epoch": 0.3511677985754514,
+      "grad_norm": 606.1836547851562,
+      "learning_rate": 2.464055263022605e-06,
+      "loss": 39.0574,
+      "step": 86920
+    },
+    {
+      "epoch": 0.351208199840819,
+      "grad_norm": 485.1193542480469,
+      "learning_rate": 2.462839860379729e-06,
+      "loss": 38.1961,
+      "step": 86930
+    },
+    {
+      "epoch": 0.35124860110618666,
+      "grad_norm": 377.4209899902344,
+      "learning_rate": 2.46162465960704e-06,
+      "loss": 34.7745,
+      "step": 86940
+    },
+    {
+      "epoch": 0.3512890023715543,
+      "grad_norm": 717.3018188476562,
+      "learning_rate": 2.460409660801229e-06,
+      "loss": 44.4682,
+      "step": 86950
+    },
+    {
+      "epoch": 0.3513294036369219,
+      "grad_norm": 621.1063842773438,
+      "learning_rate": 2.459194864058963e-06,
+      "loss": 64.5044,
+      "step": 86960
+    },
+    {
+      "epoch": 0.35136980490228953,
+      "grad_norm": 244.81198120117188,
+      "learning_rate": 2.457980269476903e-06,
+      "loss": 46.7034,
+      "step": 86970
+    },
+    {
+      "epoch": 0.35141020616765717,
+      "grad_norm": 1095.035888671875,
+      "learning_rate": 2.4567658771516876e-06,
+      "loss": 57.7799,
+      "step": 86980
+    },
+    {
+      "epoch": 0.3514506074330248,
+      "grad_norm": 163.78842163085938,
+      "learning_rate": 2.455551687179939e-06,
+      "loss": 27.2865,
+      "step": 86990
+    },
+    {
+      "epoch": 0.35149100869839245,
+      "grad_norm": 539.2749633789062,
+      "learning_rate": 2.454337699658267e-06,
+      "loss": 38.7721,
+      "step": 87000
+    },
+    {
+      "epoch": 0.3515314099637601,
+      "grad_norm": 544.2691040039062,
+      "learning_rate": 2.453123914683259e-06,
+      "loss": 43.5339,
+      "step": 87010
+    },
+    {
+      "epoch": 0.3515718112291277,
+      "grad_norm": 597.60546875,
+      "learning_rate": 2.4519103323514932e-06,
+      "loss": 45.6523,
+      "step": 87020
+    },
+    {
+      "epoch": 0.3516122124944953,
+      "grad_norm": 546.766845703125,
+      "learning_rate": 2.4506969527595277e-06,
+      "loss": 27.7563,
+      "step": 87030
+    },
+    {
+      "epoch": 0.35165261375986295,
+      "grad_norm": 974.383544921875,
+      "learning_rate": 2.4494837760039057e-06,
+      "loss": 39.1787,
+      "step": 87040
+    },
+    {
+      "epoch": 0.3516930150252306,
+      "grad_norm": 832.134033203125,
+      "learning_rate": 2.4482708021811546e-06,
+      "loss": 46.4646,
+      "step": 87050
+    },
+    {
+      "epoch": 0.35173341629059823,
+      "grad_norm": 668.964111328125,
+      "learning_rate": 2.4470580313877833e-06,
+      "loss": 48.1042,
+      "step": 87060
+    },
+    {
+      "epoch": 0.3517738175559659,
+      "grad_norm": 441.5453186035156,
+      "learning_rate": 2.44584546372029e-06,
+      "loss": 58.8005,
+      "step": 87070
+    },
+    {
+      "epoch": 0.35181421882133346,
+      "grad_norm": 1058.5311279296875,
+      "learning_rate": 2.4446330992751504e-06,
+      "loss": 34.4747,
+      "step": 87080
+    },
+    {
+      "epoch": 0.3518546200867011,
+      "grad_norm": 242.58157348632812,
+      "learning_rate": 2.44342093814883e-06,
+      "loss": 42.5668,
+      "step": 87090
+    },
+    {
+      "epoch": 0.35189502135206874,
+      "grad_norm": 530.7557373046875,
+      "learning_rate": 2.442208980437771e-06,
+      "loss": 38.0475,
+      "step": 87100
+    },
+    {
+      "epoch": 0.3519354226174364,
+      "grad_norm": 694.8820190429688,
+      "learning_rate": 2.4409972262384037e-06,
+      "loss": 52.2897,
+      "step": 87110
+    },
+    {
+      "epoch": 0.351975823882804,
+      "grad_norm": 1054.0111083984375,
+      "learning_rate": 2.4397856756471435e-06,
+      "loss": 56.8561,
+      "step": 87120
+    },
+    {
+      "epoch": 0.35201622514817166,
+      "grad_norm": 287.3025817871094,
+      "learning_rate": 2.438574328760387e-06,
+      "loss": 35.5717,
+      "step": 87130
+    },
+    {
+      "epoch": 0.3520566264135393,
+      "grad_norm": 538.0453491210938,
+      "learning_rate": 2.437363185674516e-06,
+      "loss": 47.153,
+      "step": 87140
+    },
+    {
+      "epoch": 0.3520970276789069,
+      "grad_norm": 286.7646179199219,
+      "learning_rate": 2.4361522464858956e-06,
+      "loss": 34.1812,
+      "step": 87150
+    },
+    {
+      "epoch": 0.3521374289442745,
+      "grad_norm": 717.658935546875,
+      "learning_rate": 2.434941511290872e-06,
+      "loss": 44.6221,
+      "step": 87160
+    },
+    {
+      "epoch": 0.35217783020964216,
+      "grad_norm": 574.4979858398438,
+      "learning_rate": 2.4337309801857846e-06,
+      "loss": 70.7453,
+      "step": 87170
+    },
+    {
+      "epoch": 0.3522182314750098,
+      "grad_norm": 518.8090209960938,
+      "learning_rate": 2.432520653266943e-06,
+      "loss": 35.0774,
+      "step": 87180
+    },
+    {
+      "epoch": 0.35225863274037744,
+      "grad_norm": 457.4261779785156,
+      "learning_rate": 2.4313105306306505e-06,
+      "loss": 44.7714,
+      "step": 87190
+    },
+    {
+      "epoch": 0.3522990340057451,
+      "grad_norm": 740.4395141601562,
+      "learning_rate": 2.4301006123731908e-06,
+      "loss": 42.6572,
+      "step": 87200
+    },
+    {
+      "epoch": 0.35233943527111267,
+      "grad_norm": 760.9652709960938,
+      "learning_rate": 2.4288908985908304e-06,
+      "loss": 60.802,
+      "step": 87210
+    },
+    {
+      "epoch": 0.3523798365364803,
+      "grad_norm": 451.7291564941406,
+      "learning_rate": 2.4276813893798212e-06,
+      "loss": 35.5246,
+      "step": 87220
+    },
+    {
+      "epoch": 0.35242023780184795,
+      "grad_norm": 364.14007568359375,
+      "learning_rate": 2.4264720848363992e-06,
+      "loss": 34.915,
+      "step": 87230
+    },
+    {
+      "epoch": 0.3524606390672156,
+      "grad_norm": 738.9100341796875,
+      "learning_rate": 2.4252629850567823e-06,
+      "loss": 42.3692,
+      "step": 87240
+    },
+    {
+      "epoch": 0.3525010403325832,
+      "grad_norm": 501.7611083984375,
+      "learning_rate": 2.4240540901371727e-06,
+      "loss": 40.0392,
+      "step": 87250
+    },
+    {
+      "epoch": 0.35254144159795087,
+      "grad_norm": 598.90771484375,
+      "learning_rate": 2.4228454001737576e-06,
+      "loss": 38.7727,
+      "step": 87260
+    },
+    {
+      "epoch": 0.3525818428633185,
+      "grad_norm": 807.11181640625,
+      "learning_rate": 2.421636915262707e-06,
+      "loss": 35.2477,
+      "step": 87270
+    },
+    {
+      "epoch": 0.3526222441286861,
+      "grad_norm": 421.2793273925781,
+      "learning_rate": 2.420428635500173e-06,
+      "loss": 28.5157,
+      "step": 87280
+    },
+    {
+      "epoch": 0.35266264539405373,
+      "grad_norm": 662.828369140625,
+      "learning_rate": 2.419220560982294e-06,
+      "loss": 55.7317,
+      "step": 87290
+    },
+    {
+      "epoch": 0.35270304665942137,
+      "grad_norm": 341.786865234375,
+      "learning_rate": 2.418012691805191e-06,
+      "loss": 36.8736,
+      "step": 87300
+    },
+    {
+      "epoch": 0.352743447924789,
+      "grad_norm": 423.5752258300781,
+      "learning_rate": 2.4168050280649686e-06,
+      "loss": 43.2021,
+      "step": 87310
+    },
+    {
+      "epoch": 0.35278384919015665,
+      "grad_norm": 643.5564575195312,
+      "learning_rate": 2.4155975698577146e-06,
+      "loss": 43.2181,
+      "step": 87320
+    },
+    {
+      "epoch": 0.3528242504555243,
+      "grad_norm": 885.1715698242188,
+      "learning_rate": 2.4143903172795014e-06,
+      "loss": 39.284,
+      "step": 87330
+    },
+    {
+      "epoch": 0.3528646517208919,
+      "grad_norm": 827.6829223632812,
+      "learning_rate": 2.4131832704263842e-06,
+      "loss": 57.596,
+      "step": 87340
+    },
+    {
+      "epoch": 0.3529050529862595,
+      "grad_norm": 947.4823608398438,
+      "learning_rate": 2.411976429394402e-06,
+      "loss": 29.5421,
+      "step": 87350
+    },
+    {
+      "epoch": 0.35294545425162716,
+      "grad_norm": 432.5012512207031,
+      "learning_rate": 2.4107697942795782e-06,
+      "loss": 95.7835,
+      "step": 87360
+    },
+    {
+      "epoch": 0.3529858555169948,
+      "grad_norm": 816.228515625,
+      "learning_rate": 2.4095633651779186e-06,
+      "loss": 42.3632,
+      "step": 87370
+    },
+    {
+      "epoch": 0.35302625678236244,
+      "grad_norm": 245.00999450683594,
+      "learning_rate": 2.4083571421854137e-06,
+      "loss": 32.5642,
+      "step": 87380
+    },
+    {
+      "epoch": 0.3530666580477301,
+      "grad_norm": 515.5824584960938,
+      "learning_rate": 2.407151125398037e-06,
+      "loss": 42.5482,
+      "step": 87390
+    },
+    {
+      "epoch": 0.35310705931309766,
+      "grad_norm": 478.2258605957031,
+      "learning_rate": 2.405945314911746e-06,
+      "loss": 39.545,
+      "step": 87400
+    },
+    {
+      "epoch": 0.3531474605784653,
+      "grad_norm": 330.6973876953125,
+      "learning_rate": 2.4047397108224807e-06,
+      "loss": 38.6373,
+      "step": 87410
+    },
+    {
+      "epoch": 0.35318786184383294,
+      "grad_norm": 654.366943359375,
+      "learning_rate": 2.403534313226166e-06,
+      "loss": 51.1304,
+      "step": 87420
+    },
+    {
+      "epoch": 0.3532282631092006,
+      "grad_norm": 235.28494262695312,
+      "learning_rate": 2.40232912221871e-06,
+      "loss": 38.8305,
+      "step": 87430
+    },
+    {
+      "epoch": 0.3532686643745682,
+      "grad_norm": 553.5276489257812,
+      "learning_rate": 2.4011241378960037e-06,
+      "loss": 65.192,
+      "step": 87440
+    },
+    {
+      "epoch": 0.35330906563993586,
+      "grad_norm": 500.3632507324219,
+      "learning_rate": 2.3999193603539234e-06,
+      "loss": 23.022,
+      "step": 87450
+    },
+    {
+      "epoch": 0.3533494669053035,
+      "grad_norm": 544.8861083984375,
+      "learning_rate": 2.3987147896883263e-06,
+      "loss": 34.551,
+      "step": 87460
+    },
+    {
+      "epoch": 0.3533898681706711,
+      "grad_norm": 642.550048828125,
+      "learning_rate": 2.397510425995055e-06,
+      "loss": 40.3579,
+      "step": 87470
+    },
+    {
+      "epoch": 0.3534302694360387,
+      "grad_norm": 371.6269836425781,
+      "learning_rate": 2.3963062693699353e-06,
+      "loss": 20.749,
+      "step": 87480
+    },
+    {
+      "epoch": 0.35347067070140636,
+      "grad_norm": 568.1721801757812,
+      "learning_rate": 2.3951023199087763e-06,
+      "loss": 50.5051,
+      "step": 87490
+    },
+    {
+      "epoch": 0.353511071966774,
+      "grad_norm": 823.19921875,
+      "learning_rate": 2.393898577707371e-06,
+      "loss": 25.9185,
+      "step": 87500
+    },
+    {
+      "epoch": 0.35355147323214164,
+      "grad_norm": 589.0402221679688,
+      "learning_rate": 2.392695042861495e-06,
+      "loss": 42.5533,
+      "step": 87510
+    },
+    {
+      "epoch": 0.3535918744975093,
+      "grad_norm": 417.7690124511719,
+      "learning_rate": 2.391491715466909e-06,
+      "loss": 35.2292,
+      "step": 87520
+    },
+    {
+      "epoch": 0.35363227576287687,
+      "grad_norm": 408.7934875488281,
+      "learning_rate": 2.390288595619356e-06,
+      "loss": 56.9616,
+      "step": 87530
+    },
+    {
+      "epoch": 0.3536726770282445,
+      "grad_norm": 623.8540649414062,
+      "learning_rate": 2.3890856834145625e-06,
+      "loss": 33.9323,
+      "step": 87540
+    },
+    {
+      "epoch": 0.35371307829361215,
+      "grad_norm": 927.5078125,
+      "learning_rate": 2.3878829789482385e-06,
+      "loss": 45.9919,
+      "step": 87550
+    },
+    {
+      "epoch": 0.3537534795589798,
+      "grad_norm": 467.7714538574219,
+      "learning_rate": 2.3866804823160776e-06,
+      "loss": 48.9872,
+      "step": 87560
+    },
+    {
+      "epoch": 0.35379388082434743,
+      "grad_norm": 214.3078155517578,
+      "learning_rate": 2.385478193613758e-06,
+      "loss": 44.1506,
+      "step": 87570
+    },
+    {
+      "epoch": 0.35383428208971507,
+      "grad_norm": 691.3468017578125,
+      "learning_rate": 2.3842761129369387e-06,
+      "loss": 37.2927,
+      "step": 87580
+    },
+    {
+      "epoch": 0.3538746833550827,
+      "grad_norm": 428.5848388671875,
+      "learning_rate": 2.3830742403812646e-06,
+      "loss": 29.9571,
+      "step": 87590
+    },
+    {
+      "epoch": 0.3539150846204503,
+      "grad_norm": 209.60064697265625,
+      "learning_rate": 2.381872576042365e-06,
+      "loss": 38.5981,
+      "step": 87600
+    },
+    {
+      "epoch": 0.35395548588581793,
+      "grad_norm": 737.27587890625,
+      "learning_rate": 2.3806711200158473e-06,
+      "loss": 44.0908,
+      "step": 87610
+    },
+    {
+      "epoch": 0.3539958871511856,
+      "grad_norm": 645.7940063476562,
+      "learning_rate": 2.3794698723973057e-06,
+      "loss": 36.5283,
+      "step": 87620
+    },
+    {
+      "epoch": 0.3540362884165532,
+      "grad_norm": 312.0558776855469,
+      "learning_rate": 2.3782688332823212e-06,
+      "loss": 23.5262,
+      "step": 87630
+    },
+    {
+      "epoch": 0.35407668968192085,
+      "grad_norm": 971.0711059570312,
+      "learning_rate": 2.3770680027664537e-06,
+      "loss": 44.3331,
+      "step": 87640
+    },
+    {
+      "epoch": 0.3541170909472885,
+      "grad_norm": 438.9453125,
+      "learning_rate": 2.3758673809452484e-06,
+      "loss": 32.176,
+      "step": 87650
+    },
+    {
+      "epoch": 0.3541574922126561,
+      "grad_norm": 674.4141845703125,
+      "learning_rate": 2.3746669679142315e-06,
+      "loss": 32.0949,
+      "step": 87660
+    },
+    {
+      "epoch": 0.3541978934780237,
+      "grad_norm": 443.1610412597656,
+      "learning_rate": 2.373466763768915e-06,
+      "loss": 47.1898,
+      "step": 87670
+    },
+    {
+      "epoch": 0.35423829474339136,
+      "grad_norm": 1148.020263671875,
+      "learning_rate": 2.3722667686047945e-06,
+      "loss": 40.4775,
+      "step": 87680
+    },
+    {
+      "epoch": 0.354278696008759,
+      "grad_norm": 792.3489379882812,
+      "learning_rate": 2.37106698251735e-06,
+      "loss": 54.9188,
+      "step": 87690
+    },
+    {
+      "epoch": 0.35431909727412664,
+      "grad_norm": 716.711669921875,
+      "learning_rate": 2.3698674056020378e-06,
+      "loss": 57.037,
+      "step": 87700
+    },
+    {
+      "epoch": 0.3543594985394943,
+      "grad_norm": 751.2996215820312,
+      "learning_rate": 2.3686680379543057e-06,
+      "loss": 44.8936,
+      "step": 87710
+    },
+    {
+      "epoch": 0.35439989980486186,
+      "grad_norm": 607.5596313476562,
+      "learning_rate": 2.36746887966958e-06,
+      "loss": 36.4576,
+      "step": 87720
+    },
+    {
+      "epoch": 0.3544403010702295,
+      "grad_norm": 530.6735229492188,
+      "learning_rate": 2.366269930843275e-06,
+      "loss": 47.7748,
+      "step": 87730
+    },
+    {
+      "epoch": 0.35448070233559714,
+      "grad_norm": 324.5372314453125,
+      "learning_rate": 2.3650711915707852e-06,
+      "loss": 29.4486,
+      "step": 87740
+    },
+    {
+      "epoch": 0.3545211036009648,
+      "grad_norm": 1344.857666015625,
+      "learning_rate": 2.363872661947488e-06,
+      "loss": 47.2667,
+      "step": 87750
+    },
+    {
+      "epoch": 0.3545615048663324,
+      "grad_norm": 412.0323791503906,
+      "learning_rate": 2.362674342068744e-06,
+      "loss": 31.9723,
+      "step": 87760
+    },
+    {
+      "epoch": 0.35460190613170006,
+      "grad_norm": 391.2769470214844,
+      "learning_rate": 2.3614762320299e-06,
+      "loss": 38.9474,
+      "step": 87770
+    },
+    {
+      "epoch": 0.3546423073970677,
+      "grad_norm": 716.4356079101562,
+      "learning_rate": 2.3602783319262847e-06,
+      "loss": 42.3854,
+      "step": 87780
+    },
+    {
+      "epoch": 0.3546827086624353,
+      "grad_norm": 445.8362121582031,
+      "learning_rate": 2.3590806418532052e-06,
+      "loss": 43.7234,
+      "step": 87790
+    },
+    {
+      "epoch": 0.3547231099278029,
+      "grad_norm": 353.9080505371094,
+      "learning_rate": 2.3578831619059595e-06,
+      "loss": 36.2803,
+      "step": 87800
+    },
+    {
+      "epoch": 0.35476351119317057,
+      "grad_norm": 510.5903625488281,
+      "learning_rate": 2.3566858921798246e-06,
+      "loss": 40.1896,
+      "step": 87810
+    },
+    {
+      "epoch": 0.3548039124585382,
+      "grad_norm": 433.15924072265625,
+      "learning_rate": 2.3554888327700604e-06,
+      "loss": 46.1273,
+      "step": 87820
+    },
+    {
+      "epoch": 0.35484431372390585,
+      "grad_norm": 345.576171875,
+      "learning_rate": 2.3542919837719154e-06,
+      "loss": 38.3756,
+      "step": 87830
+    },
+    {
+      "epoch": 0.3548847149892735,
+      "grad_norm": 631.8955688476562,
+      "learning_rate": 2.3530953452806143e-06,
+      "loss": 45.1479,
+      "step": 87840
+    },
+    {
+      "epoch": 0.35492511625464107,
+      "grad_norm": 516.5052490234375,
+      "learning_rate": 2.351898917391369e-06,
+      "loss": 39.7578,
+      "step": 87850
+    },
+    {
+      "epoch": 0.3549655175200087,
+      "grad_norm": 617.366943359375,
+      "learning_rate": 2.350702700199376e-06,
+      "loss": 36.3532,
+      "step": 87860
+    },
+    {
+      "epoch": 0.35500591878537635,
+      "grad_norm": 514.8583984375,
+      "learning_rate": 2.3495066937998085e-06,
+      "loss": 47.3744,
+      "step": 87870
+    },
+    {
+      "epoch": 0.355046320050744,
+      "grad_norm": 856.524658203125,
+      "learning_rate": 2.3483108982878294e-06,
+      "loss": 38.0377,
+      "step": 87880
+    },
+    {
+      "epoch": 0.35508672131611163,
+      "grad_norm": 236.63670349121094,
+      "learning_rate": 2.3471153137585823e-06,
+      "loss": 25.1089,
+      "step": 87890
+    },
+    {
+      "epoch": 0.35512712258147927,
+      "grad_norm": 530.3377075195312,
+      "learning_rate": 2.345919940307195e-06,
+      "loss": 51.7881,
+      "step": 87900
+    },
+    {
+      "epoch": 0.3551675238468469,
+      "grad_norm": 340.4045104980469,
+      "learning_rate": 2.3447247780287746e-06,
+      "loss": 47.1669,
+      "step": 87910
+    },
+    {
+      "epoch": 0.3552079251122145,
+      "grad_norm": 433.123291015625,
+      "learning_rate": 2.3435298270184204e-06,
+      "loss": 24.2755,
+      "step": 87920
+    },
+    {
+      "epoch": 0.35524832637758214,
+      "grad_norm": 393.7866516113281,
+      "learning_rate": 2.3423350873712057e-06,
+      "loss": 50.8094,
+      "step": 87930
+    },
+    {
+      "epoch": 0.3552887276429498,
+      "grad_norm": 695.3616333007812,
+      "learning_rate": 2.341140559182192e-06,
+      "loss": 38.8656,
+      "step": 87940
+    },
+    {
+      "epoch": 0.3553291289083174,
+      "grad_norm": 655.1812133789062,
+      "learning_rate": 2.339946242546422e-06,
+      "loss": 58.7881,
+      "step": 87950
+    },
+    {
+      "epoch": 0.35536953017368506,
+      "grad_norm": 789.6603393554688,
+      "learning_rate": 2.3387521375589205e-06,
+      "loss": 45.0847,
+      "step": 87960
+    },
+    {
+      "epoch": 0.3554099314390527,
+      "grad_norm": 634.9356079101562,
+      "learning_rate": 2.3375582443146977e-06,
+      "loss": 35.3675,
+      "step": 87970
+    },
+    {
+      "epoch": 0.3554503327044203,
+      "grad_norm": 287.3161926269531,
+      "learning_rate": 2.3363645629087467e-06,
+      "loss": 30.8269,
+      "step": 87980
+    },
+    {
+      "epoch": 0.3554907339697879,
+      "grad_norm": 454.5109558105469,
+      "learning_rate": 2.3351710934360426e-06,
+      "loss": 37.7506,
+      "step": 87990
+    },
+    {
+      "epoch": 0.35553113523515556,
+      "grad_norm": 556.6475219726562,
+      "learning_rate": 2.333977835991545e-06,
+      "loss": 41.6105,
+      "step": 88000
+    },
+    {
+      "epoch": 0.3555715365005232,
+      "grad_norm": 367.1153564453125,
+      "learning_rate": 2.3327847906701932e-06,
+      "loss": 31.0596,
+      "step": 88010
+    },
+    {
+      "epoch": 0.35561193776589084,
+      "grad_norm": 910.18798828125,
+      "learning_rate": 2.3315919575669172e-06,
+      "loss": 40.8386,
+      "step": 88020
+    },
+    {
+      "epoch": 0.3556523390312585,
+      "grad_norm": 909.4761352539062,
+      "learning_rate": 2.330399336776625e-06,
+      "loss": 28.7883,
+      "step": 88030
+    },
+    {
+      "epoch": 0.35569274029662606,
+      "grad_norm": 1147.876953125,
+      "learning_rate": 2.329206928394203e-06,
+      "loss": 50.8721,
+      "step": 88040
+    },
+    {
+      "epoch": 0.3557331415619937,
+      "grad_norm": 455.9250793457031,
+      "learning_rate": 2.3280147325145285e-06,
+      "loss": 39.5445,
+      "step": 88050
+    },
+    {
+      "epoch": 0.35577354282736134,
+      "grad_norm": 406.4186706542969,
+      "learning_rate": 2.3268227492324594e-06,
+      "loss": 43.1811,
+      "step": 88060
+    },
+    {
+      "epoch": 0.355813944092729,
+      "grad_norm": 719.1936645507812,
+      "learning_rate": 2.325630978642836e-06,
+      "loss": 45.485,
+      "step": 88070
+    },
+    {
+      "epoch": 0.3558543453580966,
+      "grad_norm": 972.1707153320312,
+      "learning_rate": 2.3244394208404816e-06,
+      "loss": 46.7824,
+      "step": 88080
+    },
+    {
+      "epoch": 0.35589474662346426,
+      "grad_norm": 1423.513427734375,
+      "learning_rate": 2.3232480759202035e-06,
+      "loss": 43.499,
+      "step": 88090
+    },
+    {
+      "epoch": 0.3559351478888319,
+      "grad_norm": 448.8571472167969,
+      "learning_rate": 2.3220569439767907e-06,
+      "loss": 58.1903,
+      "step": 88100
+    },
+    {
+      "epoch": 0.3559755491541995,
+      "grad_norm": 458.1613464355469,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 45.6451,
+      "step": 88110
+    },
+    {
+      "epoch": 0.35601595041956713,
+      "grad_norm": 691.4515991210938,
+      "learning_rate": 2.319675319399639e-06,
+      "loss": 41.7642,
+      "step": 88120
+    },
+    {
+      "epoch": 0.35605635168493477,
+      "grad_norm": 776.6409301757812,
+      "learning_rate": 2.3184848269553944e-06,
+      "loss": 46.5186,
+      "step": 88130
+    },
+    {
+      "epoch": 0.3560967529503024,
+      "grad_norm": 742.6858520507812,
+      "learning_rate": 2.3172945478670056e-06,
+      "loss": 30.7103,
+      "step": 88140
+    },
+    {
+      "epoch": 0.35613715421567005,
+      "grad_norm": 545.2284545898438,
+      "learning_rate": 2.316104482229178e-06,
+      "loss": 50.1192,
+      "step": 88150
+    },
+    {
+      "epoch": 0.3561775554810377,
+      "grad_norm": 771.6258544921875,
+      "learning_rate": 2.314914630136599e-06,
+      "loss": 53.8035,
+      "step": 88160
+    },
+    {
+      "epoch": 0.3562179567464053,
+      "grad_norm": 383.92913818359375,
+      "learning_rate": 2.3137249916839394e-06,
+      "loss": 38.666,
+      "step": 88170
+    },
+    {
+      "epoch": 0.3562583580117729,
+      "grad_norm": 750.7330322265625,
+      "learning_rate": 2.3125355669658547e-06,
+      "loss": 34.8775,
+      "step": 88180
+    },
+    {
+      "epoch": 0.35629875927714055,
+      "grad_norm": 659.4488525390625,
+      "learning_rate": 2.3113463560769807e-06,
+      "loss": 43.7715,
+      "step": 88190
+    },
+    {
+      "epoch": 0.3563391605425082,
+      "grad_norm": 899.4495239257812,
+      "learning_rate": 2.310157359111938e-06,
+      "loss": 44.3542,
+      "step": 88200
+    },
+    {
+      "epoch": 0.35637956180787583,
+      "grad_norm": 589.4261474609375,
+      "learning_rate": 2.3089685761653296e-06,
+      "loss": 40.3078,
+      "step": 88210
+    },
+    {
+      "epoch": 0.3564199630732435,
+      "grad_norm": 3914.93994140625,
+      "learning_rate": 2.3077800073317415e-06,
+      "loss": 73.6616,
+      "step": 88220
+    },
+    {
+      "epoch": 0.35646036433861106,
+      "grad_norm": 402.8006591796875,
+      "learning_rate": 2.3065916527057426e-06,
+      "loss": 45.9411,
+      "step": 88230
+    },
+    {
+      "epoch": 0.3565007656039787,
+      "grad_norm": 539.3480834960938,
+      "learning_rate": 2.305403512381884e-06,
+      "loss": 71.7619,
+      "step": 88240
+    },
+    {
+      "epoch": 0.35654116686934634,
+      "grad_norm": 945.126708984375,
+      "learning_rate": 2.3042155864547024e-06,
+      "loss": 43.7147,
+      "step": 88250
+    },
+    {
+      "epoch": 0.356581568134714,
+      "grad_norm": 355.734375,
+      "learning_rate": 2.303027875018714e-06,
+      "loss": 27.9518,
+      "step": 88260
+    },
+    {
+      "epoch": 0.3566219694000816,
+      "grad_norm": 208.47483825683594,
+      "learning_rate": 2.3018403781684205e-06,
+      "loss": 33.6595,
+      "step": 88270
+    },
+    {
+      "epoch": 0.35666237066544926,
+      "grad_norm": 313.4412536621094,
+      "learning_rate": 2.3006530959983055e-06,
+      "loss": 34.5666,
+      "step": 88280
+    },
+    {
+      "epoch": 0.3567027719308169,
+      "grad_norm": 224.41334533691406,
+      "learning_rate": 2.299466028602835e-06,
+      "loss": 44.1448,
+      "step": 88290
+    },
+    {
+      "epoch": 0.3567431731961845,
+      "grad_norm": 644.6503295898438,
+      "learning_rate": 2.298279176076459e-06,
+      "loss": 39.4249,
+      "step": 88300
+    },
+    {
+      "epoch": 0.3567835744615521,
+      "grad_norm": 1147.67919921875,
+      "learning_rate": 2.2970925385136093e-06,
+      "loss": 65.6626,
+      "step": 88310
+    },
+    {
+      "epoch": 0.35682397572691976,
+      "grad_norm": 724.1583862304688,
+      "learning_rate": 2.295906116008702e-06,
+      "loss": 47.0453,
+      "step": 88320
+    },
+    {
+      "epoch": 0.3568643769922874,
+      "grad_norm": 757.2796630859375,
+      "learning_rate": 2.2947199086561346e-06,
+      "loss": 36.3413,
+      "step": 88330
+    },
+    {
+      "epoch": 0.35690477825765504,
+      "grad_norm": 926.6180419921875,
+      "learning_rate": 2.293533916550289e-06,
+      "loss": 52.0724,
+      "step": 88340
+    },
+    {
+      "epoch": 0.3569451795230227,
+      "grad_norm": 459.0198059082031,
+      "learning_rate": 2.292348139785528e-06,
+      "loss": 44.2652,
+      "step": 88350
+    },
+    {
+      "epoch": 0.35698558078839027,
+      "grad_norm": 405.31219482421875,
+      "learning_rate": 2.2911625784562e-06,
+      "loss": 35.9227,
+      "step": 88360
+    },
+    {
+      "epoch": 0.3570259820537579,
+      "grad_norm": 568.8250122070312,
+      "learning_rate": 2.2899772326566327e-06,
+      "loss": 35.6614,
+      "step": 88370
+    },
+    {
+      "epoch": 0.35706638331912555,
+      "grad_norm": 277.4350891113281,
+      "learning_rate": 2.2887921024811405e-06,
+      "loss": 39.3197,
+      "step": 88380
+    },
+    {
+      "epoch": 0.3571067845844932,
+      "grad_norm": 483.451171875,
+      "learning_rate": 2.2876071880240174e-06,
+      "loss": 41.9486,
+      "step": 88390
+    },
+    {
+      "epoch": 0.3571471858498608,
+      "grad_norm": 765.7012939453125,
+      "learning_rate": 2.2864224893795423e-06,
+      "loss": 42.5878,
+      "step": 88400
+    },
+    {
+      "epoch": 0.35718758711522847,
+      "grad_norm": 363.0635681152344,
+      "learning_rate": 2.285238006641976e-06,
+      "loss": 42.8957,
+      "step": 88410
+    },
+    {
+      "epoch": 0.3572279883805961,
+      "grad_norm": 417.99847412109375,
+      "learning_rate": 2.284053739905563e-06,
+      "loss": 37.0467,
+      "step": 88420
+    },
+    {
+      "epoch": 0.3572683896459637,
+      "grad_norm": 748.5416870117188,
+      "learning_rate": 2.282869689264529e-06,
+      "loss": 42.7998,
+      "step": 88430
+    },
+    {
+      "epoch": 0.35730879091133133,
+      "grad_norm": 738.539794921875,
+      "learning_rate": 2.2816858548130837e-06,
+      "loss": 29.4397,
+      "step": 88440
+    },
+    {
+      "epoch": 0.35734919217669897,
+      "grad_norm": 297.82244873046875,
+      "learning_rate": 2.28050223664542e-06,
+      "loss": 33.6989,
+      "step": 88450
+    },
+    {
+      "epoch": 0.3573895934420666,
+      "grad_norm": 619.92822265625,
+      "learning_rate": 2.2793188348557136e-06,
+      "loss": 57.3341,
+      "step": 88460
+    },
+    {
+      "epoch": 0.35742999470743425,
+      "grad_norm": 862.1347045898438,
+      "learning_rate": 2.2781356495381186e-06,
+      "loss": 43.8701,
+      "step": 88470
+    },
+    {
+      "epoch": 0.3574703959728019,
+      "grad_norm": 345.45556640625,
+      "learning_rate": 2.276952680786779e-06,
+      "loss": 33.5217,
+      "step": 88480
+    },
+    {
+      "epoch": 0.3575107972381695,
+      "grad_norm": 1218.187744140625,
+      "learning_rate": 2.2757699286958186e-06,
+      "loss": 47.879,
+      "step": 88490
+    },
+    {
+      "epoch": 0.3575511985035371,
+      "grad_norm": 938.0349731445312,
+      "learning_rate": 2.274587393359342e-06,
+      "loss": 37.6885,
+      "step": 88500
+    },
+    {
+      "epoch": 0.35759159976890476,
+      "grad_norm": 448.3917541503906,
+      "learning_rate": 2.273405074871438e-06,
+      "loss": 51.9774,
+      "step": 88510
+    },
+    {
+      "epoch": 0.3576320010342724,
+      "grad_norm": 254.905029296875,
+      "learning_rate": 2.2722229733261795e-06,
+      "loss": 35.7465,
+      "step": 88520
+    },
+    {
+      "epoch": 0.35767240229964004,
+      "grad_norm": 524.2755126953125,
+      "learning_rate": 2.2710410888176205e-06,
+      "loss": 51.7829,
+      "step": 88530
+    },
+    {
+      "epoch": 0.3577128035650077,
+      "grad_norm": 1088.100830078125,
+      "learning_rate": 2.2698594214397966e-06,
+      "loss": 48.4789,
+      "step": 88540
+    },
+    {
+      "epoch": 0.35775320483037526,
+      "grad_norm": 358.4494323730469,
+      "learning_rate": 2.268677971286732e-06,
+      "loss": 32.0552,
+      "step": 88550
+    },
+    {
+      "epoch": 0.3577936060957429,
+      "grad_norm": 286.228515625,
+      "learning_rate": 2.2674967384524237e-06,
+      "loss": 49.5649,
+      "step": 88560
+    },
+    {
+      "epoch": 0.35783400736111054,
+      "grad_norm": 322.6353454589844,
+      "learning_rate": 2.2663157230308576e-06,
+      "loss": 38.7459,
+      "step": 88570
+    },
+    {
+      "epoch": 0.3578744086264782,
+      "grad_norm": 520.438232421875,
+      "learning_rate": 2.2651349251160055e-06,
+      "loss": 50.6289,
+      "step": 88580
+    },
+    {
+      "epoch": 0.3579148098918458,
+      "grad_norm": 638.7794799804688,
+      "learning_rate": 2.263954344801816e-06,
+      "loss": 38.1684,
+      "step": 88590
+    },
+    {
+      "epoch": 0.35795521115721346,
+      "grad_norm": 1405.0643310546875,
+      "learning_rate": 2.2627739821822226e-06,
+      "loss": 46.8707,
+      "step": 88600
+    },
+    {
+      "epoch": 0.3579956124225811,
+      "grad_norm": 1341.0897216796875,
+      "learning_rate": 2.261593837351141e-06,
+      "loss": 52.1858,
+      "step": 88610
+    },
+    {
+      "epoch": 0.3580360136879487,
+      "grad_norm": 425.5408020019531,
+      "learning_rate": 2.26041391040247e-06,
+      "loss": 34.8545,
+      "step": 88620
+    },
+    {
+      "epoch": 0.3580764149533163,
+      "grad_norm": 587.6309814453125,
+      "learning_rate": 2.259234201430092e-06,
+      "loss": 28.1153,
+      "step": 88630
+    },
+    {
+      "epoch": 0.35811681621868396,
+      "grad_norm": 962.7505493164062,
+      "learning_rate": 2.2580547105278716e-06,
+      "loss": 48.4188,
+      "step": 88640
+    },
+    {
+      "epoch": 0.3581572174840516,
+      "grad_norm": 438.6845703125,
+      "learning_rate": 2.2568754377896516e-06,
+      "loss": 35.5792,
+      "step": 88650
+    },
+    {
+      "epoch": 0.35819761874941924,
+      "grad_norm": 464.7343444824219,
+      "learning_rate": 2.255696383309265e-06,
+      "loss": 46.6391,
+      "step": 88660
+    },
+    {
+      "epoch": 0.3582380200147869,
+      "grad_norm": 867.2921752929688,
+      "learning_rate": 2.2545175471805197e-06,
+      "loss": 61.4531,
+      "step": 88670
+    },
+    {
+      "epoch": 0.35827842128015447,
+      "grad_norm": 221.92137145996094,
+      "learning_rate": 2.2533389294972153e-06,
+      "loss": 36.2097,
+      "step": 88680
+    },
+    {
+      "epoch": 0.3583188225455221,
+      "grad_norm": 346.0577087402344,
+      "learning_rate": 2.2521605303531267e-06,
+      "loss": 42.6694,
+      "step": 88690
+    },
+    {
+      "epoch": 0.35835922381088975,
+      "grad_norm": 953.293701171875,
+      "learning_rate": 2.2509823498420142e-06,
+      "loss": 41.7837,
+      "step": 88700
+    },
+    {
+      "epoch": 0.3583996250762574,
+      "grad_norm": 459.3766174316406,
+      "learning_rate": 2.2498043880576193e-06,
+      "loss": 35.375,
+      "step": 88710
+    },
+    {
+      "epoch": 0.35844002634162503,
+      "grad_norm": 735.3365478515625,
+      "learning_rate": 2.2486266450936695e-06,
+      "loss": 49.8161,
+      "step": 88720
+    },
+    {
+      "epoch": 0.35848042760699267,
+      "grad_norm": 488.2784118652344,
+      "learning_rate": 2.2474491210438687e-06,
+      "loss": 45.6757,
+      "step": 88730
+    },
+    {
+      "epoch": 0.3585208288723603,
+      "grad_norm": 918.28515625,
+      "learning_rate": 2.2462718160019086e-06,
+      "loss": 69.573,
+      "step": 88740
+    },
+    {
+      "epoch": 0.3585612301377279,
+      "grad_norm": 472.2066345214844,
+      "learning_rate": 2.245094730061463e-06,
+      "loss": 39.619,
+      "step": 88750
+    },
+    {
+      "epoch": 0.35860163140309553,
+      "grad_norm": 425.4399719238281,
+      "learning_rate": 2.2439178633161855e-06,
+      "loss": 46.1708,
+      "step": 88760
+    },
+    {
+      "epoch": 0.3586420326684632,
+      "grad_norm": 776.6825561523438,
+      "learning_rate": 2.2427412158597133e-06,
+      "loss": 49.0356,
+      "step": 88770
+    },
+    {
+      "epoch": 0.3586824339338308,
+      "grad_norm": 911.3529052734375,
+      "learning_rate": 2.2415647877856706e-06,
+      "loss": 59.9118,
+      "step": 88780
+    },
+    {
+      "epoch": 0.35872283519919845,
+      "grad_norm": 230.06333923339844,
+      "learning_rate": 2.240388579187658e-06,
+      "loss": 48.1842,
+      "step": 88790
+    },
+    {
+      "epoch": 0.3587632364645661,
+      "grad_norm": 549.273193359375,
+      "learning_rate": 2.2392125901592615e-06,
+      "loss": 37.3137,
+      "step": 88800
+    },
+    {
+      "epoch": 0.3588036377299337,
+      "grad_norm": 1127.8165283203125,
+      "learning_rate": 2.23803682079405e-06,
+      "loss": 48.1943,
+      "step": 88810
+    },
+    {
+      "epoch": 0.3588440389953013,
+      "grad_norm": 4227.5986328125,
+      "learning_rate": 2.236861271185572e-06,
+      "loss": 75.6474,
+      "step": 88820
+    },
+    {
+      "epoch": 0.35888444026066896,
+      "grad_norm": 596.3049926757812,
+      "learning_rate": 2.2356859414273613e-06,
+      "loss": 34.7695,
+      "step": 88830
+    },
+    {
+      "epoch": 0.3589248415260366,
+      "grad_norm": 529.9871826171875,
+      "learning_rate": 2.2345108316129333e-06,
+      "loss": 53.1124,
+      "step": 88840
+    },
+    {
+      "epoch": 0.35896524279140424,
+      "grad_norm": 577.8896484375,
+      "learning_rate": 2.233335941835787e-06,
+      "loss": 46.6854,
+      "step": 88850
+    },
+    {
+      "epoch": 0.3590056440567719,
+      "grad_norm": 656.1891479492188,
+      "learning_rate": 2.232161272189401e-06,
+      "loss": 48.7999,
+      "step": 88860
+    },
+    {
+      "epoch": 0.35904604532213946,
+      "grad_norm": 328.6973876953125,
+      "learning_rate": 2.230986822767241e-06,
+      "loss": 36.8326,
+      "step": 88870
+    },
+    {
+      "epoch": 0.3590864465875071,
+      "grad_norm": 777.5918579101562,
+      "learning_rate": 2.2298125936627517e-06,
+      "loss": 44.1089,
+      "step": 88880
+    },
+    {
+      "epoch": 0.35912684785287474,
+      "grad_norm": 395.1194152832031,
+      "learning_rate": 2.228638584969363e-06,
+      "loss": 28.7259,
+      "step": 88890
+    },
+    {
+      "epoch": 0.3591672491182424,
+      "grad_norm": 637.8306274414062,
+      "learning_rate": 2.227464796780481e-06,
+      "loss": 40.9933,
+      "step": 88900
+    },
+    {
+      "epoch": 0.35920765038361,
+      "grad_norm": 685.5401000976562,
+      "learning_rate": 2.226291229189501e-06,
+      "loss": 53.4521,
+      "step": 88910
+    },
+    {
+      "epoch": 0.35924805164897766,
+      "grad_norm": 592.7680053710938,
+      "learning_rate": 2.225117882289799e-06,
+      "loss": 35.8013,
+      "step": 88920
+    },
+    {
+      "epoch": 0.3592884529143453,
+      "grad_norm": 433.8973388671875,
+      "learning_rate": 2.223944756174731e-06,
+      "loss": 44.9199,
+      "step": 88930
+    },
+    {
+      "epoch": 0.3593288541797129,
+      "grad_norm": 706.0247802734375,
+      "learning_rate": 2.2227718509376395e-06,
+      "loss": 44.5,
+      "step": 88940
+    },
+    {
+      "epoch": 0.3593692554450805,
+      "grad_norm": 437.9292907714844,
+      "learning_rate": 2.221599166671845e-06,
+      "loss": 35.4113,
+      "step": 88950
+    },
+    {
+      "epoch": 0.35940965671044817,
+      "grad_norm": 680.7326049804688,
+      "learning_rate": 2.220426703470653e-06,
+      "loss": 33.6723,
+      "step": 88960
+    },
+    {
+      "epoch": 0.3594500579758158,
+      "grad_norm": 340.7998046875,
+      "learning_rate": 2.2192544614273526e-06,
+      "loss": 59.2249,
+      "step": 88970
+    },
+    {
+      "epoch": 0.35949045924118345,
+      "grad_norm": 569.3303833007812,
+      "learning_rate": 2.218082440635215e-06,
+      "loss": 51.8298,
+      "step": 88980
+    },
+    {
+      "epoch": 0.3595308605065511,
+      "grad_norm": 586.7805786132812,
+      "learning_rate": 2.216910641187488e-06,
+      "loss": 33.3345,
+      "step": 88990
+    },
+    {
+      "epoch": 0.35957126177191867,
+      "grad_norm": 660.5562133789062,
+      "learning_rate": 2.215739063177409e-06,
+      "loss": 39.8882,
+      "step": 89000
+    },
+    {
+      "epoch": 0.3596116630372863,
+      "grad_norm": 328.0276794433594,
+      "learning_rate": 2.2145677066981948e-06,
+      "loss": 48.4375,
+      "step": 89010
+    },
+    {
+      "epoch": 0.35965206430265395,
+      "grad_norm": 740.8335571289062,
+      "learning_rate": 2.213396571843045e-06,
+      "loss": 41.5751,
+      "step": 89020
+    },
+    {
+      "epoch": 0.3596924655680216,
+      "grad_norm": 663.4422607421875,
+      "learning_rate": 2.2122256587051404e-06,
+      "loss": 44.6787,
+      "step": 89030
+    },
+    {
+      "epoch": 0.35973286683338923,
+      "grad_norm": 693.177001953125,
+      "learning_rate": 2.211054967377647e-06,
+      "loss": 39.8156,
+      "step": 89040
+    },
+    {
+      "epoch": 0.35977326809875687,
+      "grad_norm": 424.6520690917969,
+      "learning_rate": 2.2098844979537093e-06,
+      "loss": 43.0032,
+      "step": 89050
+    },
+    {
+      "epoch": 0.3598136693641245,
+      "grad_norm": 398.33245849609375,
+      "learning_rate": 2.208714250526456e-06,
+      "loss": 51.8167,
+      "step": 89060
+    },
+    {
+      "epoch": 0.3598540706294921,
+      "grad_norm": 235.34942626953125,
+      "learning_rate": 2.207544225189003e-06,
+      "loss": 31.962,
+      "step": 89070
+    },
+    {
+      "epoch": 0.35989447189485974,
+      "grad_norm": 524.5186767578125,
+      "learning_rate": 2.2063744220344386e-06,
+      "loss": 31.0559,
+      "step": 89080
+    },
+    {
+      "epoch": 0.3599348731602274,
+      "grad_norm": 447.48516845703125,
+      "learning_rate": 2.2052048411558403e-06,
+      "loss": 26.2319,
+      "step": 89090
+    },
+    {
+      "epoch": 0.359975274425595,
+      "grad_norm": 971.1300659179688,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 51.9521,
+      "step": 89100
+    },
+    {
+      "epoch": 0.36001567569096266,
+      "grad_norm": 617.1143798828125,
+      "learning_rate": 2.2028663465987576e-06,
+      "loss": 36.2012,
+      "step": 89110
+    },
+    {
+      "epoch": 0.3600560769563303,
+      "grad_norm": 752.3304443359375,
+      "learning_rate": 2.201697433106336e-06,
+      "loss": 37.9972,
+      "step": 89120
+    },
+    {
+      "epoch": 0.3600964782216979,
+      "grad_norm": 575.4898681640625,
+      "learning_rate": 2.2005287422620083e-06,
+      "loss": 47.5751,
+      "step": 89130
+    },
+    {
+      "epoch": 0.3601368794870655,
+      "grad_norm": 440.1522521972656,
+      "learning_rate": 2.19936027415876e-06,
+      "loss": 29.9773,
+      "step": 89140
+    },
+    {
+      "epoch": 0.36017728075243316,
+      "grad_norm": 423.9078369140625,
+      "learning_rate": 2.1981920288895615e-06,
+      "loss": 36.8926,
+      "step": 89150
+    },
+    {
+      "epoch": 0.3602176820178008,
+      "grad_norm": 772.270751953125,
+      "learning_rate": 2.197024006547364e-06,
+      "loss": 32.4956,
+      "step": 89160
+    },
+    {
+      "epoch": 0.36025808328316844,
+      "grad_norm": 669.4854736328125,
+      "learning_rate": 2.1958562072251033e-06,
+      "loss": 46.3644,
+      "step": 89170
+    },
+    {
+      "epoch": 0.3602984845485361,
+      "grad_norm": 997.831298828125,
+      "learning_rate": 2.1946886310156945e-06,
+      "loss": 45.0814,
+      "step": 89180
+    },
+    {
+      "epoch": 0.36033888581390366,
+      "grad_norm": 661.4156494140625,
+      "learning_rate": 2.193521278012037e-06,
+      "loss": 59.7783,
+      "step": 89190
+    },
+    {
+      "epoch": 0.3603792870792713,
+      "grad_norm": 539.9761352539062,
+      "learning_rate": 2.1923541483070114e-06,
+      "loss": 34.3439,
+      "step": 89200
+    },
+    {
+      "epoch": 0.36041968834463894,
+      "grad_norm": 385.9214172363281,
+      "learning_rate": 2.1911872419934804e-06,
+      "loss": 55.5859,
+      "step": 89210
+    },
+    {
+      "epoch": 0.3604600896100066,
+      "grad_norm": 266.3016052246094,
+      "learning_rate": 2.1900205591642904e-06,
+      "loss": 33.0835,
+      "step": 89220
+    },
+    {
+      "epoch": 0.3605004908753742,
+      "grad_norm": 682.5214233398438,
+      "learning_rate": 2.188854099912268e-06,
+      "loss": 40.3408,
+      "step": 89230
+    },
+    {
+      "epoch": 0.36054089214074186,
+      "grad_norm": 460.3638916015625,
+      "learning_rate": 2.187687864330224e-06,
+      "loss": 30.4106,
+      "step": 89240
+    },
+    {
+      "epoch": 0.3605812934061095,
+      "grad_norm": 724.8146362304688,
+      "learning_rate": 2.1865218525109496e-06,
+      "loss": 37.4181,
+      "step": 89250
+    },
+    {
+      "epoch": 0.3606216946714771,
+      "grad_norm": 436.9694519042969,
+      "learning_rate": 2.185356064547219e-06,
+      "loss": 46.0752,
+      "step": 89260
+    },
+    {
+      "epoch": 0.36066209593684473,
+      "grad_norm": 517.351806640625,
+      "learning_rate": 2.1841905005317893e-06,
+      "loss": 51.404,
+      "step": 89270
+    },
+    {
+      "epoch": 0.36070249720221237,
+      "grad_norm": 626.5015258789062,
+      "learning_rate": 2.1830251605573984e-06,
+      "loss": 68.4701,
+      "step": 89280
+    },
+    {
+      "epoch": 0.36074289846758,
+      "grad_norm": 686.57275390625,
+      "learning_rate": 2.181860044716767e-06,
+      "loss": 53.143,
+      "step": 89290
+    },
+    {
+      "epoch": 0.36078329973294765,
+      "grad_norm": 6181.7060546875,
+      "learning_rate": 2.180695153102599e-06,
+      "loss": 52.3037,
+      "step": 89300
+    },
+    {
+      "epoch": 0.3608237009983153,
+      "grad_norm": 739.2041625976562,
+      "learning_rate": 2.179530485807578e-06,
+      "loss": 46.834,
+      "step": 89310
+    },
+    {
+      "epoch": 0.3608641022636829,
+      "grad_norm": 340.6275329589844,
+      "learning_rate": 2.1783660429243747e-06,
+      "loss": 42.4703,
+      "step": 89320
+    },
+    {
+      "epoch": 0.3609045035290505,
+      "grad_norm": 594.989501953125,
+      "learning_rate": 2.177201824545632e-06,
+      "loss": 35.3637,
+      "step": 89330
+    },
+    {
+      "epoch": 0.36094490479441815,
+      "grad_norm": 560.0608520507812,
+      "learning_rate": 2.1760378307639867e-06,
+      "loss": 35.9234,
+      "step": 89340
+    },
+    {
+      "epoch": 0.3609853060597858,
+      "grad_norm": 583.403076171875,
+      "learning_rate": 2.174874061672051e-06,
+      "loss": 59.6438,
+      "step": 89350
+    },
+    {
+      "epoch": 0.36102570732515343,
+      "grad_norm": 567.5617065429688,
+      "learning_rate": 2.173710517362421e-06,
+      "loss": 37.6246,
+      "step": 89360
+    },
+    {
+      "epoch": 0.3610661085905211,
+      "grad_norm": 424.9427795410156,
+      "learning_rate": 2.1725471979276734e-06,
+      "loss": 24.9531,
+      "step": 89370
+    },
+    {
+      "epoch": 0.3611065098558887,
+      "grad_norm": 458.8323974609375,
+      "learning_rate": 2.17138410346037e-06,
+      "loss": 50.1594,
+      "step": 89380
+    },
+    {
+      "epoch": 0.3611469111212563,
+      "grad_norm": 488.1316223144531,
+      "learning_rate": 2.1702212340530515e-06,
+      "loss": 48.1364,
+      "step": 89390
+    },
+    {
+      "epoch": 0.36118731238662394,
+      "grad_norm": 711.227294921875,
+      "learning_rate": 2.1690585897982423e-06,
+      "loss": 32.577,
+      "step": 89400
+    },
+    {
+      "epoch": 0.3612277136519916,
+      "grad_norm": 309.82220458984375,
+      "learning_rate": 2.167896170788451e-06,
+      "loss": 20.7099,
+      "step": 89410
+    },
+    {
+      "epoch": 0.3612681149173592,
+      "grad_norm": 744.2542114257812,
+      "learning_rate": 2.16673397711616e-06,
+      "loss": 51.5083,
+      "step": 89420
+    },
+    {
+      "epoch": 0.36130851618272686,
+      "grad_norm": 960.5636596679688,
+      "learning_rate": 2.165572008873845e-06,
+      "loss": 36.2448,
+      "step": 89430
+    },
+    {
+      "epoch": 0.3613489174480945,
+      "grad_norm": 556.6470947265625,
+      "learning_rate": 2.1644102661539573e-06,
+      "loss": 50.8981,
+      "step": 89440
+    },
+    {
+      "epoch": 0.3613893187134621,
+      "grad_norm": 344.86383056640625,
+      "learning_rate": 2.1632487490489314e-06,
+      "loss": 35.3776,
+      "step": 89450
+    },
+    {
+      "epoch": 0.3614297199788297,
+      "grad_norm": 1143.32861328125,
+      "learning_rate": 2.162087457651183e-06,
+      "loss": 55.0619,
+      "step": 89460
+    },
+    {
+      "epoch": 0.36147012124419736,
+      "grad_norm": 555.110107421875,
+      "learning_rate": 2.1609263920531115e-06,
+      "loss": 49.6263,
+      "step": 89470
+    },
+    {
+      "epoch": 0.361510522509565,
+      "grad_norm": 599.9452514648438,
+      "learning_rate": 2.159765552347098e-06,
+      "loss": 32.7204,
+      "step": 89480
+    },
+    {
+      "epoch": 0.36155092377493264,
+      "grad_norm": 696.069091796875,
+      "learning_rate": 2.1586049386255036e-06,
+      "loss": 31.2028,
+      "step": 89490
+    },
+    {
+      "epoch": 0.3615913250403003,
+      "grad_norm": 623.7810668945312,
+      "learning_rate": 2.1574445509806764e-06,
+      "loss": 51.1999,
+      "step": 89500
+    },
+    {
+      "epoch": 0.36163172630566787,
+      "grad_norm": 922.6550903320312,
+      "learning_rate": 2.156284389504939e-06,
+      "loss": 50.31,
+      "step": 89510
+    },
+    {
+      "epoch": 0.3616721275710355,
+      "grad_norm": 480.38385009765625,
+      "learning_rate": 2.1551244542905995e-06,
+      "loss": 53.4307,
+      "step": 89520
+    },
+    {
+      "epoch": 0.36171252883640315,
+      "grad_norm": 381.1392517089844,
+      "learning_rate": 2.1539647454299535e-06,
+      "loss": 37.8111,
+      "step": 89530
+    },
+    {
+      "epoch": 0.3617529301017708,
+      "grad_norm": 807.1481323242188,
+      "learning_rate": 2.152805263015271e-06,
+      "loss": 35.2265,
+      "step": 89540
+    },
+    {
+      "epoch": 0.3617933313671384,
+      "grad_norm": 414.92742919921875,
+      "learning_rate": 2.1516460071388062e-06,
+      "loss": 34.6921,
+      "step": 89550
+    },
+    {
+      "epoch": 0.36183373263250607,
+      "grad_norm": 878.7418823242188,
+      "learning_rate": 2.1504869778927965e-06,
+      "loss": 56.5126,
+      "step": 89560
+    },
+    {
+      "epoch": 0.3618741338978737,
+      "grad_norm": 207.6223602294922,
+      "learning_rate": 2.149328175369461e-06,
+      "loss": 23.0313,
+      "step": 89570
+    },
+    {
+      "epoch": 0.3619145351632413,
+      "grad_norm": 780.982177734375,
+      "learning_rate": 2.148169599661001e-06,
+      "loss": 39.0533,
+      "step": 89580
+    },
+    {
+      "epoch": 0.36195493642860893,
+      "grad_norm": 403.15582275390625,
+      "learning_rate": 2.147011250859597e-06,
+      "loss": 32.8454,
+      "step": 89590
+    },
+    {
+      "epoch": 0.36199533769397657,
+      "grad_norm": 231.27716064453125,
+      "learning_rate": 2.1458531290574138e-06,
+      "loss": 28.3538,
+      "step": 89600
+    },
+    {
+      "epoch": 0.3620357389593442,
+      "grad_norm": 410.24169921875,
+      "learning_rate": 2.144695234346598e-06,
+      "loss": 49.4636,
+      "step": 89610
+    },
+    {
+      "epoch": 0.36207614022471185,
+      "grad_norm": 581.3450317382812,
+      "learning_rate": 2.143537566819277e-06,
+      "loss": 52.0878,
+      "step": 89620
+    },
+    {
+      "epoch": 0.3621165414900795,
+      "grad_norm": 1102.1473388671875,
+      "learning_rate": 2.1423801265675643e-06,
+      "loss": 44.8626,
+      "step": 89630
+    },
+    {
+      "epoch": 0.3621569427554471,
+      "grad_norm": 450.6357727050781,
+      "learning_rate": 2.14122291368355e-06,
+      "loss": 65.0009,
+      "step": 89640
+    },
+    {
+      "epoch": 0.3621973440208147,
+      "grad_norm": 576.7796630859375,
+      "learning_rate": 2.1400659282593083e-06,
+      "loss": 37.2811,
+      "step": 89650
+    },
+    {
+      "epoch": 0.36223774528618236,
+      "grad_norm": 653.2671508789062,
+      "learning_rate": 2.1389091703868954e-06,
+      "loss": 46.5598,
+      "step": 89660
+    },
+    {
+      "epoch": 0.36227814655155,
+      "grad_norm": 701.052001953125,
+      "learning_rate": 2.137752640158351e-06,
+      "loss": 47.4967,
+      "step": 89670
+    },
+    {
+      "epoch": 0.36231854781691764,
+      "grad_norm": 1010.1072998046875,
+      "learning_rate": 2.136596337665691e-06,
+      "loss": 38.5774,
+      "step": 89680
+    },
+    {
+      "epoch": 0.3623589490822853,
+      "grad_norm": 623.6953735351562,
+      "learning_rate": 2.13544026300092e-06,
+      "loss": 49.8961,
+      "step": 89690
+    },
+    {
+      "epoch": 0.3623993503476529,
+      "grad_norm": 1026.1448974609375,
+      "learning_rate": 2.13428441625602e-06,
+      "loss": 31.1046,
+      "step": 89700
+    },
+    {
+      "epoch": 0.3624397516130205,
+      "grad_norm": 621.78076171875,
+      "learning_rate": 2.1331287975229574e-06,
+      "loss": 54.1089,
+      "step": 89710
+    },
+    {
+      "epoch": 0.36248015287838814,
+      "grad_norm": 418.7895812988281,
+      "learning_rate": 2.131973406893677e-06,
+      "loss": 37.3024,
+      "step": 89720
+    },
+    {
+      "epoch": 0.3625205541437558,
+      "grad_norm": 580.6715698242188,
+      "learning_rate": 2.1308182444601126e-06,
+      "loss": 33.1834,
+      "step": 89730
+    },
+    {
+      "epoch": 0.3625609554091234,
+      "grad_norm": 653.8624267578125,
+      "learning_rate": 2.1296633103141724e-06,
+      "loss": 48.9852,
+      "step": 89740
+    },
+    {
+      "epoch": 0.36260135667449106,
+      "grad_norm": 646.8182373046875,
+      "learning_rate": 2.1285086045477515e-06,
+      "loss": 41.7328,
+      "step": 89750
+    },
+    {
+      "epoch": 0.3626417579398587,
+      "grad_norm": 696.7265014648438,
+      "learning_rate": 2.12735412725272e-06,
+      "loss": 49.5961,
+      "step": 89760
+    },
+    {
+      "epoch": 0.3626821592052263,
+      "grad_norm": 735.8131103515625,
+      "learning_rate": 2.1261998785209382e-06,
+      "loss": 36.8307,
+      "step": 89770
+    },
+    {
+      "epoch": 0.3627225604705939,
+      "grad_norm": 528.7770385742188,
+      "learning_rate": 2.125045858444242e-06,
+      "loss": 41.6708,
+      "step": 89780
+    },
+    {
+      "epoch": 0.36276296173596156,
+      "grad_norm": 535.4943237304688,
+      "learning_rate": 2.1238920671144534e-06,
+      "loss": 38.8069,
+      "step": 89790
+    },
+    {
+      "epoch": 0.3628033630013292,
+      "grad_norm": 1034.5419921875,
+      "learning_rate": 2.122738504623373e-06,
+      "loss": 37.8137,
+      "step": 89800
+    },
+    {
+      "epoch": 0.36284376426669684,
+      "grad_norm": 762.473388671875,
+      "learning_rate": 2.121585171062785e-06,
+      "loss": 53.8039,
+      "step": 89810
+    },
+    {
+      "epoch": 0.3628841655320645,
+      "grad_norm": 613.7921142578125,
+      "learning_rate": 2.1204320665244533e-06,
+      "loss": 38.9286,
+      "step": 89820
+    },
+    {
+      "epoch": 0.36292456679743207,
+      "grad_norm": 513.7335205078125,
+      "learning_rate": 2.119279191100128e-06,
+      "loss": 33.4057,
+      "step": 89830
+    },
+    {
+      "epoch": 0.3629649680627997,
+      "grad_norm": 864.5242919921875,
+      "learning_rate": 2.1181265448815388e-06,
+      "loss": 42.9364,
+      "step": 89840
+    },
+    {
+      "epoch": 0.36300536932816735,
+      "grad_norm": 422.1235656738281,
+      "learning_rate": 2.1169741279603927e-06,
+      "loss": 40.5869,
+      "step": 89850
+    },
+    {
+      "epoch": 0.363045770593535,
+      "grad_norm": 693.3779296875,
+      "learning_rate": 2.1158219404283836e-06,
+      "loss": 50.8548,
+      "step": 89860
+    },
+    {
+      "epoch": 0.36308617185890263,
+      "grad_norm": 466.7857971191406,
+      "learning_rate": 2.1146699823771867e-06,
+      "loss": 34.1342,
+      "step": 89870
+    },
+    {
+      "epoch": 0.36312657312427027,
+      "grad_norm": 703.0391235351562,
+      "learning_rate": 2.1135182538984565e-06,
+      "loss": 54.1938,
+      "step": 89880
+    },
+    {
+      "epoch": 0.3631669743896379,
+      "grad_norm": 404.7064208984375,
+      "learning_rate": 2.1123667550838322e-06,
+      "loss": 29.0139,
+      "step": 89890
+    },
+    {
+      "epoch": 0.3632073756550055,
+      "grad_norm": 502.29351806640625,
+      "learning_rate": 2.1112154860249327e-06,
+      "loss": 54.9764,
+      "step": 89900
+    },
+    {
+      "epoch": 0.36324777692037313,
+      "grad_norm": 510.11895751953125,
+      "learning_rate": 2.1100644468133574e-06,
+      "loss": 51.0422,
+      "step": 89910
+    },
+    {
+      "epoch": 0.3632881781857408,
+      "grad_norm": 299.7879333496094,
+      "learning_rate": 2.1089136375406934e-06,
+      "loss": 48.2378,
+      "step": 89920
+    },
+    {
+      "epoch": 0.3633285794511084,
+      "grad_norm": 240.06448364257812,
+      "learning_rate": 2.107763058298504e-06,
+      "loss": 31.8403,
+      "step": 89930
+    },
+    {
+      "epoch": 0.36336898071647605,
+      "grad_norm": 1164.992431640625,
+      "learning_rate": 2.106612709178333e-06,
+      "loss": 34.8477,
+      "step": 89940
+    },
+    {
+      "epoch": 0.3634093819818437,
+      "grad_norm": 565.4199829101562,
+      "learning_rate": 2.10546259027171e-06,
+      "loss": 53.9241,
+      "step": 89950
+    },
+    {
+      "epoch": 0.3634497832472113,
+      "grad_norm": 584.436767578125,
+      "learning_rate": 2.1043127016701442e-06,
+      "loss": 45.8077,
+      "step": 89960
+    },
+    {
+      "epoch": 0.3634901845125789,
+      "grad_norm": 475.01226806640625,
+      "learning_rate": 2.1031630434651277e-06,
+      "loss": 44.3417,
+      "step": 89970
+    },
+    {
+      "epoch": 0.36353058577794656,
+      "grad_norm": 569.577880859375,
+      "learning_rate": 2.102013615748133e-06,
+      "loss": 49.0946,
+      "step": 89980
+    },
+    {
+      "epoch": 0.3635709870433142,
+      "grad_norm": 443.4644775390625,
+      "learning_rate": 2.1008644186106146e-06,
+      "loss": 38.6309,
+      "step": 89990
+    },
+    {
+      "epoch": 0.36361138830868184,
+      "grad_norm": 2156.372314453125,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": 51.6774,
+      "step": 90000
+    },
+    {
+      "epoch": 0.3636517895740495,
+      "grad_norm": 564.2222290039062,
+      "learning_rate": 2.0985667164397355e-06,
+      "loss": 50.4205,
+      "step": 90010
+    },
+    {
+      "epoch": 0.3636921908394171,
+      "grad_norm": 803.0197143554688,
+      "learning_rate": 2.0974182115891924e-06,
+      "loss": 48.016,
+      "step": 90020
+    },
+    {
+      "epoch": 0.3637325921047847,
+      "grad_norm": 715.5997314453125,
+      "learning_rate": 2.0962699376837604e-06,
+      "loss": 51.9438,
+      "step": 90030
+    },
+    {
+      "epoch": 0.36377299337015234,
+      "grad_norm": 553.6121215820312,
+      "learning_rate": 2.0951218948148034e-06,
+      "loss": 46.5583,
+      "step": 90040
+    },
+    {
+      "epoch": 0.36381339463552,
+      "grad_norm": 923.4013061523438,
+      "learning_rate": 2.093974083073666e-06,
+      "loss": 51.7148,
+      "step": 90050
+    },
+    {
+      "epoch": 0.3638537959008876,
+      "grad_norm": 919.7776489257812,
+      "learning_rate": 2.0928265025516737e-06,
+      "loss": 55.002,
+      "step": 90060
+    },
+    {
+      "epoch": 0.36389419716625526,
+      "grad_norm": 522.1885375976562,
+      "learning_rate": 2.0916791533401344e-06,
+      "loss": 41.7103,
+      "step": 90070
+    },
+    {
+      "epoch": 0.3639345984316229,
+      "grad_norm": 721.49609375,
+      "learning_rate": 2.090532035530337e-06,
+      "loss": 40.6271,
+      "step": 90080
+    },
+    {
+      "epoch": 0.3639749996969905,
+      "grad_norm": 361.6112060546875,
+      "learning_rate": 2.0893851492135536e-06,
+      "loss": 39.6089,
+      "step": 90090
+    },
+    {
+      "epoch": 0.3640154009623581,
+      "grad_norm": 666.88720703125,
+      "learning_rate": 2.0882384944810358e-06,
+      "loss": 36.2407,
+      "step": 90100
+    },
+    {
+      "epoch": 0.36405580222772577,
+      "grad_norm": 370.4792175292969,
+      "learning_rate": 2.087092071424017e-06,
+      "loss": 39.4503,
+      "step": 90110
+    },
+    {
+      "epoch": 0.3640962034930934,
+      "grad_norm": 317.5982971191406,
+      "learning_rate": 2.085945880133715e-06,
+      "loss": 56.8149,
+      "step": 90120
+    },
+    {
+      "epoch": 0.36413660475846105,
+      "grad_norm": 743.6160888671875,
+      "learning_rate": 2.0847999207013247e-06,
+      "loss": 50.2978,
+      "step": 90130
+    },
+    {
+      "epoch": 0.3641770060238287,
+      "grad_norm": 482.4396057128906,
+      "learning_rate": 2.083654193218026e-06,
+      "loss": 37.2876,
+      "step": 90140
+    },
+    {
+      "epoch": 0.36421740728919627,
+      "grad_norm": 737.9346923828125,
+      "learning_rate": 2.0825086977749793e-06,
+      "loss": 47.1252,
+      "step": 90150
+    },
+    {
+      "epoch": 0.3642578085545639,
+      "grad_norm": 375.45465087890625,
+      "learning_rate": 2.0813634344633256e-06,
+      "loss": 39.6987,
+      "step": 90160
+    },
+    {
+      "epoch": 0.36429820981993155,
+      "grad_norm": 493.0916748046875,
+      "learning_rate": 2.0802184033741886e-06,
+      "loss": 38.7509,
+      "step": 90170
+    },
+    {
+      "epoch": 0.3643386110852992,
+      "grad_norm": 551.7267456054688,
+      "learning_rate": 2.0790736045986737e-06,
+      "loss": 41.7807,
+      "step": 90180
+    },
+    {
+      "epoch": 0.36437901235066683,
+      "grad_norm": 742.0836181640625,
+      "learning_rate": 2.077929038227867e-06,
+      "loss": 36.7443,
+      "step": 90190
+    },
+    {
+      "epoch": 0.36441941361603447,
+      "grad_norm": 635.392578125,
+      "learning_rate": 2.076784704352835e-06,
+      "loss": 37.1757,
+      "step": 90200
+    },
+    {
+      "epoch": 0.3644598148814021,
+      "grad_norm": 1009.4746704101562,
+      "learning_rate": 2.075640603064629e-06,
+      "loss": 32.1593,
+      "step": 90210
+    },
+    {
+      "epoch": 0.3645002161467697,
+      "grad_norm": 679.4805297851562,
+      "learning_rate": 2.07449673445428e-06,
+      "loss": 57.6325,
+      "step": 90220
+    },
+    {
+      "epoch": 0.36454061741213734,
+      "grad_norm": 404.5597839355469,
+      "learning_rate": 2.0733530986127985e-06,
+      "loss": 25.9294,
+      "step": 90230
+    },
+    {
+      "epoch": 0.364581018677505,
+      "grad_norm": 305.4102478027344,
+      "learning_rate": 2.07220969563118e-06,
+      "loss": 36.3092,
+      "step": 90240
+    },
+    {
+      "epoch": 0.3646214199428726,
+      "grad_norm": 915.2492065429688,
+      "learning_rate": 2.0710665256003994e-06,
+      "loss": 54.6068,
+      "step": 90250
+    },
+    {
+      "epoch": 0.36466182120824026,
+      "grad_norm": 867.6870727539062,
+      "learning_rate": 2.069923588611413e-06,
+      "loss": 47.5033,
+      "step": 90260
+    },
+    {
+      "epoch": 0.3647022224736079,
+      "grad_norm": 616.1192016601562,
+      "learning_rate": 2.068780884755161e-06,
+      "loss": 43.4188,
+      "step": 90270
+    },
+    {
+      "epoch": 0.3647426237389755,
+      "grad_norm": 644.8280029296875,
+      "learning_rate": 2.0676384141225586e-06,
+      "loss": 55.7568,
+      "step": 90280
+    },
+    {
+      "epoch": 0.3647830250043431,
+      "grad_norm": 492.7054443359375,
+      "learning_rate": 2.066496176804511e-06,
+      "loss": 47.7336,
+      "step": 90290
+    },
+    {
+      "epoch": 0.36482342626971076,
+      "grad_norm": 1280.5413818359375,
+      "learning_rate": 2.0653541728919002e-06,
+      "loss": 46.9477,
+      "step": 90300
+    },
+    {
+      "epoch": 0.3648638275350784,
+      "grad_norm": 603.3501586914062,
+      "learning_rate": 2.0642124024755895e-06,
+      "loss": 31.9769,
+      "step": 90310
+    },
+    {
+      "epoch": 0.36490422880044604,
+      "grad_norm": 976.8206176757812,
+      "learning_rate": 2.0630708656464245e-06,
+      "loss": 50.2211,
+      "step": 90320
+    },
+    {
+      "epoch": 0.3649446300658137,
+      "grad_norm": 698.71240234375,
+      "learning_rate": 2.0619295624952318e-06,
+      "loss": 55.8766,
+      "step": 90330
+    },
+    {
+      "epoch": 0.3649850313311813,
+      "grad_norm": 1087.606689453125,
+      "learning_rate": 2.0607884931128205e-06,
+      "loss": 50.9868,
+      "step": 90340
+    },
+    {
+      "epoch": 0.3650254325965489,
+      "grad_norm": 817.5459594726562,
+      "learning_rate": 2.059647657589979e-06,
+      "loss": 47.0761,
+      "step": 90350
+    },
+    {
+      "epoch": 0.36506583386191654,
+      "grad_norm": 929.0218505859375,
+      "learning_rate": 2.0585070560174807e-06,
+      "loss": 44.2793,
+      "step": 90360
+    },
+    {
+      "epoch": 0.3651062351272842,
+      "grad_norm": 316.37432861328125,
+      "learning_rate": 2.057366688486073e-06,
+      "loss": 24.7103,
+      "step": 90370
+    },
+    {
+      "epoch": 0.3651466363926518,
+      "grad_norm": 1726.5743408203125,
+      "learning_rate": 2.056226555086495e-06,
+      "loss": 37.4747,
+      "step": 90380
+    },
+    {
+      "epoch": 0.36518703765801946,
+      "grad_norm": 703.32275390625,
+      "learning_rate": 2.0550866559094597e-06,
+      "loss": 49.8901,
+      "step": 90390
+    },
+    {
+      "epoch": 0.3652274389233871,
+      "grad_norm": 1690.470703125,
+      "learning_rate": 2.053946991045664e-06,
+      "loss": 53.2018,
+      "step": 90400
+    },
+    {
+      "epoch": 0.3652678401887547,
+      "grad_norm": 431.23388671875,
+      "learning_rate": 2.0528075605857855e-06,
+      "loss": 33.2038,
+      "step": 90410
+    },
+    {
+      "epoch": 0.36530824145412233,
+      "grad_norm": 323.9116516113281,
+      "learning_rate": 2.0516683646204836e-06,
+      "loss": 55.9418,
+      "step": 90420
+    },
+    {
+      "epoch": 0.36534864271948997,
+      "grad_norm": 671.396240234375,
+      "learning_rate": 2.0505294032403987e-06,
+      "loss": 30.016,
+      "step": 90430
+    },
+    {
+      "epoch": 0.3653890439848576,
+      "grad_norm": 641.1298217773438,
+      "learning_rate": 2.0493906765361556e-06,
+      "loss": 36.2649,
+      "step": 90440
+    },
+    {
+      "epoch": 0.36542944525022525,
+      "grad_norm": 364.4647521972656,
+      "learning_rate": 2.0482521845983522e-06,
+      "loss": 47.5725,
+      "step": 90450
+    },
+    {
+      "epoch": 0.3654698465155929,
+      "grad_norm": 323.13031005859375,
+      "learning_rate": 2.047113927517576e-06,
+      "loss": 40.9744,
+      "step": 90460
+    },
+    {
+      "epoch": 0.3655102477809605,
+      "grad_norm": 541.819580078125,
+      "learning_rate": 2.0459759053843913e-06,
+      "loss": 46.2442,
+      "step": 90470
+    },
+    {
+      "epoch": 0.3655506490463281,
+      "grad_norm": 424.2214050292969,
+      "learning_rate": 2.0448381182893485e-06,
+      "loss": 47.5028,
+      "step": 90480
+    },
+    {
+      "epoch": 0.36559105031169575,
+      "grad_norm": 330.5768737792969,
+      "learning_rate": 2.043700566322974e-06,
+      "loss": 49.1583,
+      "step": 90490
+    },
+    {
+      "epoch": 0.3656314515770634,
+      "grad_norm": 751.4622192382812,
+      "learning_rate": 2.0425632495757776e-06,
+      "loss": 37.6808,
+      "step": 90500
+    },
+    {
+      "epoch": 0.36567185284243103,
+      "grad_norm": 431.5377502441406,
+      "learning_rate": 2.0414261681382507e-06,
+      "loss": 47.902,
+      "step": 90510
+    },
+    {
+      "epoch": 0.3657122541077987,
+      "grad_norm": 363.0134582519531,
+      "learning_rate": 2.0402893221008657e-06,
+      "loss": 30.096,
+      "step": 90520
+    },
+    {
+      "epoch": 0.3657526553731663,
+      "grad_norm": 270.1390075683594,
+      "learning_rate": 2.0391527115540777e-06,
+      "loss": 52.4842,
+      "step": 90530
+    },
+    {
+      "epoch": 0.3657930566385339,
+      "grad_norm": 316.0973815917969,
+      "learning_rate": 2.0380163365883188e-06,
+      "loss": 35.3534,
+      "step": 90540
+    },
+    {
+      "epoch": 0.36583345790390154,
+      "grad_norm": 871.845947265625,
+      "learning_rate": 2.0368801972940055e-06,
+      "loss": 43.3384,
+      "step": 90550
+    },
+    {
+      "epoch": 0.3658738591692692,
+      "grad_norm": 948.296142578125,
+      "learning_rate": 2.0357442937615367e-06,
+      "loss": 37.9077,
+      "step": 90560
+    },
+    {
+      "epoch": 0.3659142604346368,
+      "grad_norm": 310.0594177246094,
+      "learning_rate": 2.034608626081288e-06,
+      "loss": 44.4959,
+      "step": 90570
+    },
+    {
+      "epoch": 0.36595466170000446,
+      "grad_norm": 783.2125244140625,
+      "learning_rate": 2.0334731943436235e-06,
+      "loss": 40.5028,
+      "step": 90580
+    },
+    {
+      "epoch": 0.3659950629653721,
+      "grad_norm": 963.5101318359375,
+      "learning_rate": 2.032337998638883e-06,
+      "loss": 45.5643,
+      "step": 90590
+    },
+    {
+      "epoch": 0.3660354642307397,
+      "grad_norm": 530.5311889648438,
+      "learning_rate": 2.031203039057388e-06,
+      "loss": 26.7716,
+      "step": 90600
+    },
+    {
+      "epoch": 0.3660758654961073,
+      "grad_norm": 477.712890625,
+      "learning_rate": 2.0300683156894435e-06,
+      "loss": 45.5529,
+      "step": 90610
+    },
+    {
+      "epoch": 0.36611626676147496,
+      "grad_norm": 734.0820922851562,
+      "learning_rate": 2.028933828625332e-06,
+      "loss": 42.5537,
+      "step": 90620
+    },
+    {
+      "epoch": 0.3661566680268426,
+      "grad_norm": 878.6141967773438,
+      "learning_rate": 2.0277995779553193e-06,
+      "loss": 40.6559,
+      "step": 90630
+    },
+    {
+      "epoch": 0.36619706929221024,
+      "grad_norm": 740.7584228515625,
+      "learning_rate": 2.026665563769655e-06,
+      "loss": 59.3021,
+      "step": 90640
+    },
+    {
+      "epoch": 0.3662374705575779,
+      "grad_norm": 406.0100402832031,
+      "learning_rate": 2.025531786158565e-06,
+      "loss": 48.8978,
+      "step": 90650
+    },
+    {
+      "epoch": 0.3662778718229455,
+      "grad_norm": 351.8523254394531,
+      "learning_rate": 2.02439824521226e-06,
+      "loss": 28.7422,
+      "step": 90660
+    },
+    {
+      "epoch": 0.3663182730883131,
+      "grad_norm": 373.048095703125,
+      "learning_rate": 2.023264941020929e-06,
+      "loss": 36.5842,
+      "step": 90670
+    },
+    {
+      "epoch": 0.36635867435368075,
+      "grad_norm": 251.12399291992188,
+      "learning_rate": 2.022131873674747e-06,
+      "loss": 47.2558,
+      "step": 90680
+    },
+    {
+      "epoch": 0.3663990756190484,
+      "grad_norm": 412.8178405761719,
+      "learning_rate": 2.020999043263865e-06,
+      "loss": 39.3716,
+      "step": 90690
+    },
+    {
+      "epoch": 0.366439476884416,
+      "grad_norm": 483.9591369628906,
+      "learning_rate": 2.0198664498784194e-06,
+      "loss": 42.5158,
+      "step": 90700
+    },
+    {
+      "epoch": 0.36647987814978367,
+      "grad_norm": 479.1996765136719,
+      "learning_rate": 2.018734093608521e-06,
+      "loss": 42.1002,
+      "step": 90710
+    },
+    {
+      "epoch": 0.3665202794151513,
+      "grad_norm": 617.8897094726562,
+      "learning_rate": 2.017601974544269e-06,
+      "loss": 45.9508,
+      "step": 90720
+    },
+    {
+      "epoch": 0.3665606806805189,
+      "grad_norm": 807.5328369140625,
+      "learning_rate": 2.0164700927757407e-06,
+      "loss": 61.1654,
+      "step": 90730
+    },
+    {
+      "epoch": 0.36660108194588653,
+      "grad_norm": 610.870361328125,
+      "learning_rate": 2.0153384483929946e-06,
+      "loss": 34.2054,
+      "step": 90740
+    },
+    {
+      "epoch": 0.36664148321125417,
+      "grad_norm": 337.25994873046875,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 47.6262,
+      "step": 90750
+    },
+    {
+      "epoch": 0.3666818844766218,
+      "grad_norm": 1225.8677978515625,
+      "learning_rate": 2.0130758721449887e-06,
+      "loss": 54.6125,
+      "step": 90760
+    },
+    {
+      "epoch": 0.36672228574198945,
+      "grad_norm": 583.123291015625,
+      "learning_rate": 2.01194494045975e-06,
+      "loss": 41.9564,
+      "step": 90770
+    },
+    {
+      "epoch": 0.3667626870073571,
+      "grad_norm": 532.25,
+      "learning_rate": 2.0108142465203413e-06,
+      "loss": 37.121,
+      "step": 90780
+    },
+    {
+      "epoch": 0.3668030882727247,
+      "grad_norm": 1506.677490234375,
+      "learning_rate": 2.0096837904167252e-06,
+      "loss": 35.0181,
+      "step": 90790
+    },
+    {
+      "epoch": 0.3668434895380923,
+      "grad_norm": 492.0948486328125,
+      "learning_rate": 2.0085535722388454e-06,
+      "loss": 35.3306,
+      "step": 90800
+    },
+    {
+      "epoch": 0.36688389080345996,
+      "grad_norm": 672.6559448242188,
+      "learning_rate": 2.007423592076629e-06,
+      "loss": 34.055,
+      "step": 90810
+    },
+    {
+      "epoch": 0.3669242920688276,
+      "grad_norm": 543.7750854492188,
+      "learning_rate": 2.006293850019983e-06,
+      "loss": 35.9712,
+      "step": 90820
+    },
+    {
+      "epoch": 0.36696469333419524,
+      "grad_norm": 712.0992431640625,
+      "learning_rate": 2.005164346158796e-06,
+      "loss": 40.947,
+      "step": 90830
+    },
+    {
+      "epoch": 0.3670050945995629,
+      "grad_norm": 722.9002075195312,
+      "learning_rate": 2.004035080582938e-06,
+      "loss": 48.6245,
+      "step": 90840
+    },
+    {
+      "epoch": 0.3670454958649305,
+      "grad_norm": 952.40234375,
+      "learning_rate": 2.002906053382258e-06,
+      "loss": 46.5803,
+      "step": 90850
+    },
+    {
+      "epoch": 0.3670858971302981,
+      "grad_norm": 416.2144470214844,
+      "learning_rate": 2.001777264646588e-06,
+      "loss": 40.4567,
+      "step": 90860
+    },
+    {
+      "epoch": 0.36712629839566574,
+      "grad_norm": 403.0811767578125,
+      "learning_rate": 2.000648714465744e-06,
+      "loss": 43.3017,
+      "step": 90870
+    },
+    {
+      "epoch": 0.3671666996610334,
+      "grad_norm": 580.00048828125,
+      "learning_rate": 1.9995204029295147e-06,
+      "loss": 38.2412,
+      "step": 90880
+    },
+    {
+      "epoch": 0.367207100926401,
+      "grad_norm": 378.8351745605469,
+      "learning_rate": 1.9983923301276764e-06,
+      "loss": 36.4264,
+      "step": 90890
+    },
+    {
+      "epoch": 0.36724750219176866,
+      "grad_norm": 707.3033447265625,
+      "learning_rate": 1.9972644961499853e-06,
+      "loss": 19.8426,
+      "step": 90900
+    },
+    {
+      "epoch": 0.3672879034571363,
+      "grad_norm": 932.6064453125,
+      "learning_rate": 1.9961369010861777e-06,
+      "loss": 52.4729,
+      "step": 90910
+    },
+    {
+      "epoch": 0.3673283047225039,
+      "grad_norm": 568.361572265625,
+      "learning_rate": 1.995009545025971e-06,
+      "loss": 32.0684,
+      "step": 90920
+    },
+    {
+      "epoch": 0.3673687059878715,
+      "grad_norm": 768.6260986328125,
+      "learning_rate": 1.9938824280590635e-06,
+      "loss": 53.6914,
+      "step": 90930
+    },
+    {
+      "epoch": 0.36740910725323916,
+      "grad_norm": 680.199951171875,
+      "learning_rate": 1.992755550275135e-06,
+      "loss": 29.9278,
+      "step": 90940
+    },
+    {
+      "epoch": 0.3674495085186068,
+      "grad_norm": 772.3173217773438,
+      "learning_rate": 1.991628911763846e-06,
+      "loss": 62.1594,
+      "step": 90950
+    },
+    {
+      "epoch": 0.36748990978397444,
+      "grad_norm": 571.9541015625,
+      "learning_rate": 1.990502512614838e-06,
+      "loss": 32.7492,
+      "step": 90960
+    },
+    {
+      "epoch": 0.3675303110493421,
+      "grad_norm": 404.0989685058594,
+      "learning_rate": 1.989376352917733e-06,
+      "loss": 37.1862,
+      "step": 90970
+    },
+    {
+      "epoch": 0.3675707123147097,
+      "grad_norm": 659.4763793945312,
+      "learning_rate": 1.988250432762135e-06,
+      "loss": 38.4783,
+      "step": 90980
+    },
+    {
+      "epoch": 0.3676111135800773,
+      "grad_norm": 665.2921142578125,
+      "learning_rate": 1.987124752237628e-06,
+      "loss": 58.7645,
+      "step": 90990
+    },
+    {
+      "epoch": 0.36765151484544495,
+      "grad_norm": 378.5953369140625,
+      "learning_rate": 1.9859993114337773e-06,
+      "loss": 35.1157,
+      "step": 91000
+    },
+    {
+      "epoch": 0.3676919161108126,
+      "grad_norm": 826.9314575195312,
+      "learning_rate": 1.984874110440129e-06,
+      "loss": 48.5521,
+      "step": 91010
+    },
+    {
+      "epoch": 0.36773231737618023,
+      "grad_norm": 584.8626708984375,
+      "learning_rate": 1.9837491493462104e-06,
+      "loss": 50.472,
+      "step": 91020
+    },
+    {
+      "epoch": 0.36777271864154787,
+      "grad_norm": 381.18341064453125,
+      "learning_rate": 1.9826244282415285e-06,
+      "loss": 43.9919,
+      "step": 91030
+    },
+    {
+      "epoch": 0.3678131199069155,
+      "grad_norm": 365.3213195800781,
+      "learning_rate": 1.9814999472155736e-06,
+      "loss": 60.1717,
+      "step": 91040
+    },
+    {
+      "epoch": 0.3678535211722831,
+      "grad_norm": 525.1259155273438,
+      "learning_rate": 1.9803757063578146e-06,
+      "loss": 44.7315,
+      "step": 91050
+    },
+    {
+      "epoch": 0.36789392243765073,
+      "grad_norm": 526.0704956054688,
+      "learning_rate": 1.9792517057577026e-06,
+      "loss": 50.6954,
+      "step": 91060
+    },
+    {
+      "epoch": 0.3679343237030184,
+      "grad_norm": 279.87548828125,
+      "learning_rate": 1.978127945504669e-06,
+      "loss": 38.3538,
+      "step": 91070
+    },
+    {
+      "epoch": 0.367974724968386,
+      "grad_norm": 474.6632995605469,
+      "learning_rate": 1.977004425688126e-06,
+      "loss": 37.0405,
+      "step": 91080
+    },
+    {
+      "epoch": 0.36801512623375365,
+      "grad_norm": 480.8251037597656,
+      "learning_rate": 1.9758811463974677e-06,
+      "loss": 51.2929,
+      "step": 91090
+    },
+    {
+      "epoch": 0.3680555274991213,
+      "grad_norm": 547.632568359375,
+      "learning_rate": 1.9747581077220675e-06,
+      "loss": 28.9646,
+      "step": 91100
+    },
+    {
+      "epoch": 0.3680959287644889,
+      "grad_norm": 635.3892211914062,
+      "learning_rate": 1.9736353097512802e-06,
+      "loss": 45.2826,
+      "step": 91110
+    },
+    {
+      "epoch": 0.3681363300298565,
+      "grad_norm": 260.98651123046875,
+      "learning_rate": 1.9725127525744423e-06,
+      "loss": 77.2748,
+      "step": 91120
+    },
+    {
+      "epoch": 0.36817673129522416,
+      "grad_norm": 728.8169555664062,
+      "learning_rate": 1.971390436280871e-06,
+      "loss": 40.5998,
+      "step": 91130
+    },
+    {
+      "epoch": 0.3682171325605918,
+      "grad_norm": 411.4289245605469,
+      "learning_rate": 1.970268360959863e-06,
+      "loss": 45.5529,
+      "step": 91140
+    },
+    {
+      "epoch": 0.36825753382595944,
+      "grad_norm": 753.0279541015625,
+      "learning_rate": 1.9691465267006965e-06,
+      "loss": 35.7461,
+      "step": 91150
+    },
+    {
+      "epoch": 0.3682979350913271,
+      "grad_norm": 488.8835144042969,
+      "learning_rate": 1.9680249335926314e-06,
+      "loss": 37.8499,
+      "step": 91160
+    },
+    {
+      "epoch": 0.3683383363566947,
+      "grad_norm": 243.46627807617188,
+      "learning_rate": 1.9669035817249077e-06,
+      "loss": 45.215,
+      "step": 91170
+    },
+    {
+      "epoch": 0.3683787376220623,
+      "grad_norm": 599.044189453125,
+      "learning_rate": 1.9657824711867457e-06,
+      "loss": 56.9096,
+      "step": 91180
+    },
+    {
+      "epoch": 0.36841913888742994,
+      "grad_norm": 932.3475952148438,
+      "learning_rate": 1.9646616020673474e-06,
+      "loss": 34.8462,
+      "step": 91190
+    },
+    {
+      "epoch": 0.3684595401527976,
+      "grad_norm": 707.1528930664062,
+      "learning_rate": 1.9635409744558953e-06,
+      "loss": 42.6267,
+      "step": 91200
+    },
+    {
+      "epoch": 0.3684999414181652,
+      "grad_norm": 377.12896728515625,
+      "learning_rate": 1.962420588441552e-06,
+      "loss": 34.0437,
+      "step": 91210
+    },
+    {
+      "epoch": 0.36854034268353286,
+      "grad_norm": 756.0499877929688,
+      "learning_rate": 1.9613004441134635e-06,
+      "loss": 46.7615,
+      "step": 91220
+    },
+    {
+      "epoch": 0.3685807439489005,
+      "grad_norm": 1075.0089111328125,
+      "learning_rate": 1.96018054156075e-06,
+      "loss": 56.312,
+      "step": 91230
+    },
+    {
+      "epoch": 0.3686211452142681,
+      "grad_norm": 680.0381469726562,
+      "learning_rate": 1.9590608808725214e-06,
+      "loss": 44.0678,
+      "step": 91240
+    },
+    {
+      "epoch": 0.3686615464796357,
+      "grad_norm": 464.0368347167969,
+      "learning_rate": 1.9579414621378624e-06,
+      "loss": 45.7525,
+      "step": 91250
+    },
+    {
+      "epoch": 0.36870194774500337,
+      "grad_norm": 205.71853637695312,
+      "learning_rate": 1.9568222854458403e-06,
+      "loss": 49.5888,
+      "step": 91260
+    },
+    {
+      "epoch": 0.368742349010371,
+      "grad_norm": 599.9993286132812,
+      "learning_rate": 1.955703350885502e-06,
+      "loss": 46.9081,
+      "step": 91270
+    },
+    {
+      "epoch": 0.36878275027573865,
+      "grad_norm": 766.0186767578125,
+      "learning_rate": 1.954584658545877e-06,
+      "loss": 43.123,
+      "step": 91280
+    },
+    {
+      "epoch": 0.3688231515411063,
+      "grad_norm": 579.0977172851562,
+      "learning_rate": 1.9534662085159746e-06,
+      "loss": 47.9196,
+      "step": 91290
+    },
+    {
+      "epoch": 0.36886355280647387,
+      "grad_norm": 710.0382690429688,
+      "learning_rate": 1.9523480008847856e-06,
+      "loss": 39.5002,
+      "step": 91300
+    },
+    {
+      "epoch": 0.3689039540718415,
+      "grad_norm": 449.13397216796875,
+      "learning_rate": 1.9512300357412778e-06,
+      "loss": 42.9828,
+      "step": 91310
+    },
+    {
+      "epoch": 0.36894435533720915,
+      "grad_norm": 657.600830078125,
+      "learning_rate": 1.950112313174404e-06,
+      "loss": 36.0095,
+      "step": 91320
+    },
+    {
+      "epoch": 0.3689847566025768,
+      "grad_norm": 842.6561279296875,
+      "learning_rate": 1.9489948332730945e-06,
+      "loss": 44.1807,
+      "step": 91330
+    },
+    {
+      "epoch": 0.36902515786794443,
+      "grad_norm": 284.8597717285156,
+      "learning_rate": 1.947877596126266e-06,
+      "loss": 30.7171,
+      "step": 91340
+    },
+    {
+      "epoch": 0.36906555913331207,
+      "grad_norm": 921.1884765625,
+      "learning_rate": 1.946760601822809e-06,
+      "loss": 47.5682,
+      "step": 91350
+    },
+    {
+      "epoch": 0.3691059603986797,
+      "grad_norm": 486.9660339355469,
+      "learning_rate": 1.945643850451599e-06,
+      "loss": 45.4097,
+      "step": 91360
+    },
+    {
+      "epoch": 0.3691463616640473,
+      "grad_norm": 481.0872497558594,
+      "learning_rate": 1.9445273421014903e-06,
+      "loss": 34.2548,
+      "step": 91370
+    },
+    {
+      "epoch": 0.36918676292941494,
+      "grad_norm": 297.3924560546875,
+      "learning_rate": 1.9434110768613184e-06,
+      "loss": 53.0149,
+      "step": 91380
+    },
+    {
+      "epoch": 0.3692271641947826,
+      "grad_norm": 721.890869140625,
+      "learning_rate": 1.9422950548199004e-06,
+      "loss": 66.6985,
+      "step": 91390
+    },
+    {
+      "epoch": 0.3692675654601502,
+      "grad_norm": 407.1198425292969,
+      "learning_rate": 1.941179276066031e-06,
+      "loss": 42.478,
+      "step": 91400
+    },
+    {
+      "epoch": 0.36930796672551786,
+      "grad_norm": 982.4834594726562,
+      "learning_rate": 1.9400637406884875e-06,
+      "loss": 42.1372,
+      "step": 91410
+    },
+    {
+      "epoch": 0.3693483679908855,
+      "grad_norm": 547.2681884765625,
+      "learning_rate": 1.938948448776028e-06,
+      "loss": 22.6065,
+      "step": 91420
+    },
+    {
+      "epoch": 0.3693887692562531,
+      "grad_norm": 751.7438354492188,
+      "learning_rate": 1.9378334004173936e-06,
+      "loss": 43.4024,
+      "step": 91430
+    },
+    {
+      "epoch": 0.3694291705216207,
+      "grad_norm": 515.72802734375,
+      "learning_rate": 1.9367185957013024e-06,
+      "loss": 38.0327,
+      "step": 91440
+    },
+    {
+      "epoch": 0.36946957178698836,
+      "grad_norm": 481.98919677734375,
+      "learning_rate": 1.9356040347164533e-06,
+      "loss": 71.2502,
+      "step": 91450
+    },
+    {
+      "epoch": 0.369509973052356,
+      "grad_norm": 274.7797546386719,
+      "learning_rate": 1.9344897175515283e-06,
+      "loss": 39.7428,
+      "step": 91460
+    },
+    {
+      "epoch": 0.36955037431772364,
+      "grad_norm": 390.0133056640625,
+      "learning_rate": 1.9333756442951886e-06,
+      "loss": 47.6374,
+      "step": 91470
+    },
+    {
+      "epoch": 0.3695907755830913,
+      "grad_norm": 556.7567749023438,
+      "learning_rate": 1.9322618150360732e-06,
+      "loss": 49.5216,
+      "step": 91480
+    },
+    {
+      "epoch": 0.3696311768484589,
+      "grad_norm": 442.883056640625,
+      "learning_rate": 1.931148229862807e-06,
+      "loss": 36.484,
+      "step": 91490
+    },
+    {
+      "epoch": 0.3696715781138265,
+      "grad_norm": 499.36090087890625,
+      "learning_rate": 1.9300348888639915e-06,
+      "loss": 51.1593,
+      "step": 91500
+    },
+    {
+      "epoch": 0.36971197937919414,
+      "grad_norm": 750.4673461914062,
+      "learning_rate": 1.9289217921282104e-06,
+      "loss": 48.1591,
+      "step": 91510
+    },
+    {
+      "epoch": 0.3697523806445618,
+      "grad_norm": 307.9422607421875,
+      "learning_rate": 1.927808939744027e-06,
+      "loss": 46.4992,
+      "step": 91520
+    },
+    {
+      "epoch": 0.3697927819099294,
+      "grad_norm": 907.98046875,
+      "learning_rate": 1.9266963317999884e-06,
+      "loss": 39.0636,
+      "step": 91530
+    },
+    {
+      "epoch": 0.36983318317529706,
+      "grad_norm": 797.4759521484375,
+      "learning_rate": 1.9255839683846174e-06,
+      "loss": 52.2098,
+      "step": 91540
+    },
+    {
+      "epoch": 0.3698735844406647,
+      "grad_norm": 372.3993225097656,
+      "learning_rate": 1.9244718495864206e-06,
+      "loss": 37.4535,
+      "step": 91550
+    },
+    {
+      "epoch": 0.3699139857060323,
+      "grad_norm": 488.9723815917969,
+      "learning_rate": 1.9233599754938857e-06,
+      "loss": 32.6998,
+      "step": 91560
+    },
+    {
+      "epoch": 0.36995438697139993,
+      "grad_norm": 843.7347412109375,
+      "learning_rate": 1.922248346195477e-06,
+      "loss": 47.1682,
+      "step": 91570
+    },
+    {
+      "epoch": 0.36999478823676757,
+      "grad_norm": 472.1070251464844,
+      "learning_rate": 1.921136961779641e-06,
+      "loss": 38.4525,
+      "step": 91580
+    },
+    {
+      "epoch": 0.3700351895021352,
+      "grad_norm": 768.330810546875,
+      "learning_rate": 1.9200258223348072e-06,
+      "loss": 35.5316,
+      "step": 91590
+    },
+    {
+      "epoch": 0.37007559076750285,
+      "grad_norm": 917.4964599609375,
+      "learning_rate": 1.918914927949384e-06,
+      "loss": 32.2703,
+      "step": 91600
+    },
+    {
+      "epoch": 0.3701159920328705,
+      "grad_norm": 1343.5494384765625,
+      "learning_rate": 1.9178042787117594e-06,
+      "loss": 35.542,
+      "step": 91610
+    },
+    {
+      "epoch": 0.3701563932982381,
+      "grad_norm": 463.442626953125,
+      "learning_rate": 1.9166938747103013e-06,
+      "loss": 33.1795,
+      "step": 91620
+    },
+    {
+      "epoch": 0.3701967945636057,
+      "grad_norm": 389.9674377441406,
+      "learning_rate": 1.915583716033363e-06,
+      "loss": 42.821,
+      "step": 91630
+    },
+    {
+      "epoch": 0.37023719582897335,
+      "grad_norm": 404.5822448730469,
+      "learning_rate": 1.9144738027692746e-06,
+      "loss": 43.7155,
+      "step": 91640
+    },
+    {
+      "epoch": 0.370277597094341,
+      "grad_norm": 587.7801513671875,
+      "learning_rate": 1.913364135006343e-06,
+      "loss": 29.5249,
+      "step": 91650
+    },
+    {
+      "epoch": 0.37031799835970863,
+      "grad_norm": 835.2510986328125,
+      "learning_rate": 1.9122547128328616e-06,
+      "loss": 41.1438,
+      "step": 91660
+    },
+    {
+      "epoch": 0.3703583996250763,
+      "grad_norm": 342.636962890625,
+      "learning_rate": 1.9111455363371016e-06,
+      "loss": 34.5593,
+      "step": 91670
+    },
+    {
+      "epoch": 0.3703988008904439,
+      "grad_norm": 424.19354248046875,
+      "learning_rate": 1.910036605607316e-06,
+      "loss": 32.6238,
+      "step": 91680
+    },
+    {
+      "epoch": 0.3704392021558115,
+      "grad_norm": 920.641845703125,
+      "learning_rate": 1.908927920731736e-06,
+      "loss": 37.6558,
+      "step": 91690
+    },
+    {
+      "epoch": 0.37047960342117914,
+      "grad_norm": 857.9271240234375,
+      "learning_rate": 1.9078194817985755e-06,
+      "loss": 41.7298,
+      "step": 91700
+    },
+    {
+      "epoch": 0.3705200046865468,
+      "grad_norm": 368.0783996582031,
+      "learning_rate": 1.9067112888960283e-06,
+      "loss": 48.4999,
+      "step": 91710
+    },
+    {
+      "epoch": 0.3705604059519144,
+      "grad_norm": 644.633056640625,
+      "learning_rate": 1.905603342112265e-06,
+      "loss": 45.2764,
+      "step": 91720
+    },
+    {
+      "epoch": 0.37060080721728206,
+      "grad_norm": 310.33648681640625,
+      "learning_rate": 1.904495641535446e-06,
+      "loss": 49.4517,
+      "step": 91730
+    },
+    {
+      "epoch": 0.3706412084826497,
+      "grad_norm": 407.6061096191406,
+      "learning_rate": 1.9033881872537009e-06,
+      "loss": 39.223,
+      "step": 91740
+    },
+    {
+      "epoch": 0.3706816097480173,
+      "grad_norm": 665.1161499023438,
+      "learning_rate": 1.902280979355146e-06,
+      "loss": 38.5211,
+      "step": 91750
+    },
+    {
+      "epoch": 0.3707220110133849,
+      "grad_norm": 392.2351989746094,
+      "learning_rate": 1.901174017927877e-06,
+      "loss": 41.5336,
+      "step": 91760
+    },
+    {
+      "epoch": 0.37076241227875256,
+      "grad_norm": 520.5103759765625,
+      "learning_rate": 1.9000673030599698e-06,
+      "loss": 35.4438,
+      "step": 91770
+    },
+    {
+      "epoch": 0.3708028135441202,
+      "grad_norm": 579.9551391601562,
+      "learning_rate": 1.89896083483948e-06,
+      "loss": 39.8775,
+      "step": 91780
+    },
+    {
+      "epoch": 0.37084321480948784,
+      "grad_norm": 1177.031494140625,
+      "learning_rate": 1.897854613354445e-06,
+      "loss": 45.311,
+      "step": 91790
+    },
+    {
+      "epoch": 0.3708836160748555,
+      "grad_norm": 481.6818542480469,
+      "learning_rate": 1.8967486386928819e-06,
+      "loss": 40.572,
+      "step": 91800
+    },
+    {
+      "epoch": 0.3709240173402231,
+      "grad_norm": 631.0228881835938,
+      "learning_rate": 1.8956429109427855e-06,
+      "loss": 59.4919,
+      "step": 91810
+    },
+    {
+      "epoch": 0.3709644186055907,
+      "grad_norm": 459.4544372558594,
+      "learning_rate": 1.8945374301921393e-06,
+      "loss": 32.1201,
+      "step": 91820
+    },
+    {
+      "epoch": 0.37100481987095835,
+      "grad_norm": 340.3385925292969,
+      "learning_rate": 1.893432196528896e-06,
+      "loss": 60.1047,
+      "step": 91830
+    },
+    {
+      "epoch": 0.371045221136326,
+      "grad_norm": 647.6593017578125,
+      "learning_rate": 1.892327210040995e-06,
+      "loss": 34.1518,
+      "step": 91840
+    },
+    {
+      "epoch": 0.3710856224016936,
+      "grad_norm": 718.9778442382812,
+      "learning_rate": 1.8912224708163561e-06,
+      "loss": 34.7443,
+      "step": 91850
+    },
+    {
+      "epoch": 0.37112602366706127,
+      "grad_norm": 784.8222045898438,
+      "learning_rate": 1.890117978942878e-06,
+      "loss": 28.6377,
+      "step": 91860
+    },
+    {
+      "epoch": 0.3711664249324289,
+      "grad_norm": 482.3733825683594,
+      "learning_rate": 1.8890137345084392e-06,
+      "loss": 43.0016,
+      "step": 91870
+    },
+    {
+      "epoch": 0.3712068261977965,
+      "grad_norm": 1036.407470703125,
+      "learning_rate": 1.8879097376009009e-06,
+      "loss": 44.0674,
+      "step": 91880
+    },
+    {
+      "epoch": 0.37124722746316413,
+      "grad_norm": 554.4319458007812,
+      "learning_rate": 1.8868059883081015e-06,
+      "loss": 41.1695,
+      "step": 91890
+    },
+    {
+      "epoch": 0.37128762872853177,
+      "grad_norm": 642.7158203125,
+      "learning_rate": 1.8857024867178625e-06,
+      "loss": 29.2449,
+      "step": 91900
+    },
+    {
+      "epoch": 0.3713280299938994,
+      "grad_norm": 451.4582214355469,
+      "learning_rate": 1.8845992329179835e-06,
+      "loss": 41.4946,
+      "step": 91910
+    },
+    {
+      "epoch": 0.37136843125926705,
+      "grad_norm": 1148.1690673828125,
+      "learning_rate": 1.883496226996246e-06,
+      "loss": 31.4822,
+      "step": 91920
+    },
+    {
+      "epoch": 0.3714088325246347,
+      "grad_norm": 1108.7608642578125,
+      "learning_rate": 1.8823934690404106e-06,
+      "loss": 44.5192,
+      "step": 91930
+    },
+    {
+      "epoch": 0.3714492337900023,
+      "grad_norm": 407.66729736328125,
+      "learning_rate": 1.8812909591382195e-06,
+      "loss": 44.9309,
+      "step": 91940
+    },
+    {
+      "epoch": 0.3714896350553699,
+      "grad_norm": 833.0994262695312,
+      "learning_rate": 1.8801886973773936e-06,
+      "loss": 41.494,
+      "step": 91950
+    },
+    {
+      "epoch": 0.37153003632073756,
+      "grad_norm": 356.8611755371094,
+      "learning_rate": 1.8790866838456351e-06,
+      "loss": 32.0275,
+      "step": 91960
+    },
+    {
+      "epoch": 0.3715704375861052,
+      "grad_norm": 552.33837890625,
+      "learning_rate": 1.877984918630626e-06,
+      "loss": 43.8171,
+      "step": 91970
+    },
+    {
+      "epoch": 0.37161083885147284,
+      "grad_norm": 882.7359619140625,
+      "learning_rate": 1.876883401820029e-06,
+      "loss": 68.7707,
+      "step": 91980
+    },
+    {
+      "epoch": 0.3716512401168405,
+      "grad_norm": 222.77023315429688,
+      "learning_rate": 1.8757821335014858e-06,
+      "loss": 28.0252,
+      "step": 91990
+    },
+    {
+      "epoch": 0.3716916413822081,
+      "grad_norm": 569.3855590820312,
+      "learning_rate": 1.8746811137626208e-06,
+      "loss": 41.8199,
+      "step": 92000
+    },
+    {
+      "epoch": 0.3717320426475757,
+      "grad_norm": 337.5414123535156,
+      "learning_rate": 1.8735803426910366e-06,
+      "loss": 44.0779,
+      "step": 92010
+    },
+    {
+      "epoch": 0.37177244391294334,
+      "grad_norm": 307.907470703125,
+      "learning_rate": 1.8724798203743154e-06,
+      "loss": 34.9715,
+      "step": 92020
+    },
+    {
+      "epoch": 0.371812845178311,
+      "grad_norm": 410.8703308105469,
+      "learning_rate": 1.8713795469000218e-06,
+      "loss": 41.9369,
+      "step": 92030
+    },
+    {
+      "epoch": 0.3718532464436786,
+      "grad_norm": 526.17138671875,
+      "learning_rate": 1.8702795223556992e-06,
+      "loss": 41.1116,
+      "step": 92040
+    },
+    {
+      "epoch": 0.37189364770904626,
+      "grad_norm": 546.6736450195312,
+      "learning_rate": 1.8691797468288713e-06,
+      "loss": 40.5777,
+      "step": 92050
+    },
+    {
+      "epoch": 0.3719340489744139,
+      "grad_norm": 665.9561157226562,
+      "learning_rate": 1.8680802204070432e-06,
+      "loss": 52.4794,
+      "step": 92060
+    },
+    {
+      "epoch": 0.3719744502397815,
+      "grad_norm": 656.8494262695312,
+      "learning_rate": 1.8669809431776991e-06,
+      "loss": 43.0441,
+      "step": 92070
+    },
+    {
+      "epoch": 0.3720148515051491,
+      "grad_norm": 835.1004638671875,
+      "learning_rate": 1.8658819152283003e-06,
+      "loss": 54.4179,
+      "step": 92080
+    },
+    {
+      "epoch": 0.37205525277051676,
+      "grad_norm": 738.4459838867188,
+      "learning_rate": 1.8647831366462948e-06,
+      "loss": 39.4756,
+      "step": 92090
+    },
+    {
+      "epoch": 0.3720956540358844,
+      "grad_norm": 497.9049377441406,
+      "learning_rate": 1.8636846075191067e-06,
+      "loss": 34.005,
+      "step": 92100
+    },
+    {
+      "epoch": 0.37213605530125204,
+      "grad_norm": 334.0683898925781,
+      "learning_rate": 1.8625863279341406e-06,
+      "loss": 36.7819,
+      "step": 92110
+    },
+    {
+      "epoch": 0.3721764565666197,
+      "grad_norm": 432.1595764160156,
+      "learning_rate": 1.8614882979787818e-06,
+      "loss": 42.6485,
+      "step": 92120
+    },
+    {
+      "epoch": 0.3722168578319873,
+      "grad_norm": 563.0101318359375,
+      "learning_rate": 1.8603905177403953e-06,
+      "loss": 40.2115,
+      "step": 92130
+    },
+    {
+      "epoch": 0.3722572590973549,
+      "grad_norm": 603.9617919921875,
+      "learning_rate": 1.8592929873063259e-06,
+      "loss": 47.0243,
+      "step": 92140
+    },
+    {
+      "epoch": 0.37229766036272255,
+      "grad_norm": 250.31321716308594,
+      "learning_rate": 1.8581957067639e-06,
+      "loss": 45.3303,
+      "step": 92150
+    },
+    {
+      "epoch": 0.3723380616280902,
+      "grad_norm": 486.78363037109375,
+      "learning_rate": 1.8570986762004246e-06,
+      "loss": 43.6162,
+      "step": 92160
+    },
+    {
+      "epoch": 0.37237846289345783,
+      "grad_norm": 798.7283935546875,
+      "learning_rate": 1.8560018957031816e-06,
+      "loss": 35.2461,
+      "step": 92170
+    },
+    {
+      "epoch": 0.37241886415882547,
+      "grad_norm": 576.589599609375,
+      "learning_rate": 1.8549053653594373e-06,
+      "loss": 41.0202,
+      "step": 92180
+    },
+    {
+      "epoch": 0.3724592654241931,
+      "grad_norm": 590.5379028320312,
+      "learning_rate": 1.8538090852564405e-06,
+      "loss": 46.1836,
+      "step": 92190
+    },
+    {
+      "epoch": 0.3724996666895607,
+      "grad_norm": 313.347900390625,
+      "learning_rate": 1.852713055481416e-06,
+      "loss": 41.0,
+      "step": 92200
+    },
+    {
+      "epoch": 0.37254006795492833,
+      "grad_norm": 903.6192626953125,
+      "learning_rate": 1.8516172761215695e-06,
+      "loss": 46.3808,
+      "step": 92210
+    },
+    {
+      "epoch": 0.372580469220296,
+      "grad_norm": 626.2720947265625,
+      "learning_rate": 1.8505217472640868e-06,
+      "loss": 53.388,
+      "step": 92220
+    },
+    {
+      "epoch": 0.3726208704856636,
+      "grad_norm": 644.8720703125,
+      "learning_rate": 1.849426468996135e-06,
+      "loss": 38.492,
+      "step": 92230
+    },
+    {
+      "epoch": 0.37266127175103125,
+      "grad_norm": 750.325439453125,
+      "learning_rate": 1.8483314414048597e-06,
+      "loss": 37.5022,
+      "step": 92240
+    },
+    {
+      "epoch": 0.3727016730163989,
+      "grad_norm": 726.2249145507812,
+      "learning_rate": 1.8472366645773892e-06,
+      "loss": 46.9339,
+      "step": 92250
+    },
+    {
+      "epoch": 0.3727420742817665,
+      "grad_norm": 622.7699584960938,
+      "learning_rate": 1.846142138600826e-06,
+      "loss": 37.4274,
+      "step": 92260
+    },
+    {
+      "epoch": 0.3727824755471341,
+      "grad_norm": 666.1739501953125,
+      "learning_rate": 1.8450478635622592e-06,
+      "loss": 43.8214,
+      "step": 92270
+    },
+    {
+      "epoch": 0.37282287681250176,
+      "grad_norm": 818.9193115234375,
+      "learning_rate": 1.8439538395487528e-06,
+      "loss": 57.6135,
+      "step": 92280
+    },
+    {
+      "epoch": 0.3728632780778694,
+      "grad_norm": 784.80859375,
+      "learning_rate": 1.842860066647356e-06,
+      "loss": 40.4329,
+      "step": 92290
+    },
+    {
+      "epoch": 0.37290367934323704,
+      "grad_norm": 768.7633056640625,
+      "learning_rate": 1.841766544945095e-06,
+      "loss": 42.8496,
+      "step": 92300
+    },
+    {
+      "epoch": 0.3729440806086047,
+      "grad_norm": 563.5238037109375,
+      "learning_rate": 1.8406732745289757e-06,
+      "loss": 46.8466,
+      "step": 92310
+    },
+    {
+      "epoch": 0.3729844818739723,
+      "grad_norm": 678.9468383789062,
+      "learning_rate": 1.839580255485985e-06,
+      "loss": 52.8284,
+      "step": 92320
+    },
+    {
+      "epoch": 0.3730248831393399,
+      "grad_norm": 314.1755676269531,
+      "learning_rate": 1.83848748790309e-06,
+      "loss": 56.7375,
+      "step": 92330
+    },
+    {
+      "epoch": 0.37306528440470754,
+      "grad_norm": 756.5408325195312,
+      "learning_rate": 1.8373949718672345e-06,
+      "loss": 51.5074,
+      "step": 92340
+    },
+    {
+      "epoch": 0.3731056856700752,
+      "grad_norm": 402.6850891113281,
+      "learning_rate": 1.8363027074653473e-06,
+      "loss": 32.0473,
+      "step": 92350
+    },
+    {
+      "epoch": 0.3731460869354428,
+      "grad_norm": 610.68017578125,
+      "learning_rate": 1.835210694784334e-06,
+      "loss": 40.5902,
+      "step": 92360
+    },
+    {
+      "epoch": 0.37318648820081046,
+      "grad_norm": 776.7313842773438,
+      "learning_rate": 1.8341189339110793e-06,
+      "loss": 47.7953,
+      "step": 92370
+    },
+    {
+      "epoch": 0.3732268894661781,
+      "grad_norm": 171.4274444580078,
+      "learning_rate": 1.8330274249324537e-06,
+      "loss": 54.6801,
+      "step": 92380
+    },
+    {
+      "epoch": 0.3732672907315457,
+      "grad_norm": 527.5751953125,
+      "learning_rate": 1.831936167935301e-06,
+      "loss": 44.6827,
+      "step": 92390
+    },
+    {
+      "epoch": 0.3733076919969133,
+      "grad_norm": 542.8246459960938,
+      "learning_rate": 1.8308451630064484e-06,
+      "loss": 44.6548,
+      "step": 92400
+    },
+    {
+      "epoch": 0.37334809326228097,
+      "grad_norm": 558.8591918945312,
+      "learning_rate": 1.8297544102327014e-06,
+      "loss": 40.691,
+      "step": 92410
+    },
+    {
+      "epoch": 0.3733884945276486,
+      "grad_norm": 635.3528442382812,
+      "learning_rate": 1.8286639097008484e-06,
+      "loss": 46.9977,
+      "step": 92420
+    },
+    {
+      "epoch": 0.37342889579301625,
+      "grad_norm": 540.139892578125,
+      "learning_rate": 1.827573661497652e-06,
+      "loss": 44.0475,
+      "step": 92430
+    },
+    {
+      "epoch": 0.3734692970583839,
+      "grad_norm": 400.1092224121094,
+      "learning_rate": 1.8264836657098595e-06,
+      "loss": 35.4605,
+      "step": 92440
+    },
+    {
+      "epoch": 0.3735096983237515,
+      "grad_norm": 504.9743347167969,
+      "learning_rate": 1.8253939224241974e-06,
+      "loss": 41.2205,
+      "step": 92450
+    },
+    {
+      "epoch": 0.3735500995891191,
+      "grad_norm": 403.0384521484375,
+      "learning_rate": 1.8243044317273717e-06,
+      "loss": 49.5701,
+      "step": 92460
+    },
+    {
+      "epoch": 0.37359050085448675,
+      "grad_norm": 214.4297637939453,
+      "learning_rate": 1.823215193706066e-06,
+      "loss": 33.6296,
+      "step": 92470
+    },
+    {
+      "epoch": 0.3736309021198544,
+      "grad_norm": 435.58782958984375,
+      "learning_rate": 1.82212620844695e-06,
+      "loss": 60.6322,
+      "step": 92480
+    },
+    {
+      "epoch": 0.37367130338522203,
+      "grad_norm": 820.32373046875,
+      "learning_rate": 1.8210374760366662e-06,
+      "loss": 42.7006,
+      "step": 92490
+    },
+    {
+      "epoch": 0.37371170465058967,
+      "grad_norm": 488.87689208984375,
+      "learning_rate": 1.8199489965618433e-06,
+      "loss": 40.5734,
+      "step": 92500
+    },
+    {
+      "epoch": 0.3737521059159573,
+      "grad_norm": 1221.661865234375,
+      "learning_rate": 1.8188607701090827e-06,
+      "loss": 72.2834,
+      "step": 92510
+    },
+    {
+      "epoch": 0.3737925071813249,
+      "grad_norm": 712.658447265625,
+      "learning_rate": 1.8177727967649705e-06,
+      "loss": 45.4339,
+      "step": 92520
+    },
+    {
+      "epoch": 0.37383290844669254,
+      "grad_norm": 703.703369140625,
+      "learning_rate": 1.816685076616073e-06,
+      "loss": 59.1592,
+      "step": 92530
+    },
+    {
+      "epoch": 0.3738733097120602,
+      "grad_norm": 384.11309814453125,
+      "learning_rate": 1.8155976097489342e-06,
+      "loss": 46.3461,
+      "step": 92540
+    },
+    {
+      "epoch": 0.3739137109774278,
+      "grad_norm": 322.06011962890625,
+      "learning_rate": 1.8145103962500792e-06,
+      "loss": 36.9918,
+      "step": 92550
+    },
+    {
+      "epoch": 0.37395411224279546,
+      "grad_norm": 349.7514953613281,
+      "learning_rate": 1.8134234362060128e-06,
+      "loss": 42.063,
+      "step": 92560
+    },
+    {
+      "epoch": 0.3739945135081631,
+      "grad_norm": 679.798828125,
+      "learning_rate": 1.8123367297032175e-06,
+      "loss": 40.9792,
+      "step": 92570
+    },
+    {
+      "epoch": 0.3740349147735307,
+      "grad_norm": 891.4302978515625,
+      "learning_rate": 1.8112502768281608e-06,
+      "loss": 49.7337,
+      "step": 92580
+    },
+    {
+      "epoch": 0.3740753160388983,
+      "grad_norm": 359.6378479003906,
+      "learning_rate": 1.810164077667287e-06,
+      "loss": 32.4731,
+      "step": 92590
+    },
+    {
+      "epoch": 0.37411571730426596,
+      "grad_norm": 348.2173767089844,
+      "learning_rate": 1.809078132307016e-06,
+      "loss": 29.6351,
+      "step": 92600
+    },
+    {
+      "epoch": 0.3741561185696336,
+      "grad_norm": 579.4752197265625,
+      "learning_rate": 1.807992440833754e-06,
+      "loss": 38.418,
+      "step": 92610
+    },
+    {
+      "epoch": 0.37419651983500124,
+      "grad_norm": 738.5474853515625,
+      "learning_rate": 1.8069070033338842e-06,
+      "loss": 23.2401,
+      "step": 92620
+    },
+    {
+      "epoch": 0.3742369211003689,
+      "grad_norm": 619.52978515625,
+      "learning_rate": 1.8058218198937695e-06,
+      "loss": 51.5335,
+      "step": 92630
+    },
+    {
+      "epoch": 0.3742773223657365,
+      "grad_norm": 578.2020874023438,
+      "learning_rate": 1.8047368905997536e-06,
+      "loss": 30.5904,
+      "step": 92640
+    },
+    {
+      "epoch": 0.3743177236311041,
+      "grad_norm": 55.084449768066406,
+      "learning_rate": 1.8036522155381592e-06,
+      "loss": 44.467,
+      "step": 92650
+    },
+    {
+      "epoch": 0.37435812489647174,
+      "grad_norm": 323.4886474609375,
+      "learning_rate": 1.8025677947952879e-06,
+      "loss": 45.4456,
+      "step": 92660
+    },
+    {
+      "epoch": 0.3743985261618394,
+      "grad_norm": 235.45973205566406,
+      "learning_rate": 1.8014836284574223e-06,
+      "loss": 44.9233,
+      "step": 92670
+    },
+    {
+      "epoch": 0.374438927427207,
+      "grad_norm": 758.6615600585938,
+      "learning_rate": 1.8003997166108278e-06,
+      "loss": 31.2891,
+      "step": 92680
+    },
+    {
+      "epoch": 0.37447932869257466,
+      "grad_norm": 692.9884033203125,
+      "learning_rate": 1.7993160593417424e-06,
+      "loss": 28.9124,
+      "step": 92690
+    },
+    {
+      "epoch": 0.3745197299579423,
+      "grad_norm": 354.16278076171875,
+      "learning_rate": 1.798232656736389e-06,
+      "loss": 29.815,
+      "step": 92700
+    },
+    {
+      "epoch": 0.3745601312233099,
+      "grad_norm": 709.9598388671875,
+      "learning_rate": 1.7971495088809688e-06,
+      "loss": 42.7331,
+      "step": 92710
+    },
+    {
+      "epoch": 0.37460053248867753,
+      "grad_norm": 788.5820922851562,
+      "learning_rate": 1.796066615861663e-06,
+      "loss": 34.2348,
+      "step": 92720
+    },
+    {
+      "epoch": 0.37464093375404517,
+      "grad_norm": 347.8127136230469,
+      "learning_rate": 1.7949839777646327e-06,
+      "loss": 33.6161,
+      "step": 92730
+    },
+    {
+      "epoch": 0.3746813350194128,
+      "grad_norm": 483.2420349121094,
+      "learning_rate": 1.7939015946760186e-06,
+      "loss": 41.9075,
+      "step": 92740
+    },
+    {
+      "epoch": 0.37472173628478045,
+      "grad_norm": 424.3015441894531,
+      "learning_rate": 1.7928194666819398e-06,
+      "loss": 46.9531,
+      "step": 92750
+    },
+    {
+      "epoch": 0.3747621375501481,
+      "grad_norm": 582.9439697265625,
+      "learning_rate": 1.7917375938684979e-06,
+      "loss": 28.903,
+      "step": 92760
+    },
+    {
+      "epoch": 0.37480253881551573,
+      "grad_norm": 697.3135375976562,
+      "learning_rate": 1.7906559763217713e-06,
+      "loss": 43.9791,
+      "step": 92770
+    },
+    {
+      "epoch": 0.3748429400808833,
+      "grad_norm": 733.6925659179688,
+      "learning_rate": 1.7895746141278198e-06,
+      "loss": 45.0568,
+      "step": 92780
+    },
+    {
+      "epoch": 0.37488334134625095,
+      "grad_norm": 426.135498046875,
+      "learning_rate": 1.7884935073726822e-06,
+      "loss": 46.5901,
+      "step": 92790
+    },
+    {
+      "epoch": 0.3749237426116186,
+      "grad_norm": 500.1527099609375,
+      "learning_rate": 1.7874126561423771e-06,
+      "loss": 47.9253,
+      "step": 92800
+    },
+    {
+      "epoch": 0.37496414387698623,
+      "grad_norm": 129.91612243652344,
+      "learning_rate": 1.786332060522904e-06,
+      "loss": 45.9043,
+      "step": 92810
+    },
+    {
+      "epoch": 0.3750045451423539,
+      "grad_norm": 514.854736328125,
+      "learning_rate": 1.7852517206002396e-06,
+      "loss": 32.7294,
+      "step": 92820
+    },
+    {
+      "epoch": 0.3750449464077215,
+      "grad_norm": 657.1265258789062,
+      "learning_rate": 1.7841716364603423e-06,
+      "loss": 37.4753,
+      "step": 92830
+    },
+    {
+      "epoch": 0.3750853476730891,
+      "grad_norm": 806.763916015625,
+      "learning_rate": 1.783091808189149e-06,
+      "loss": 43.7356,
+      "step": 92840
+    },
+    {
+      "epoch": 0.37512574893845674,
+      "grad_norm": 445.2367858886719,
+      "learning_rate": 1.7820122358725772e-06,
+      "loss": 46.6244,
+      "step": 92850
+    },
+    {
+      "epoch": 0.3751661502038244,
+      "grad_norm": 181.48643493652344,
+      "learning_rate": 1.780932919596523e-06,
+      "loss": 27.6955,
+      "step": 92860
+    },
+    {
+      "epoch": 0.375206551469192,
+      "grad_norm": 853.9912719726562,
+      "learning_rate": 1.779853859446863e-06,
+      "loss": 52.9919,
+      "step": 92870
+    },
+    {
+      "epoch": 0.37524695273455966,
+      "grad_norm": 575.7948608398438,
+      "learning_rate": 1.778775055509453e-06,
+      "loss": 35.4131,
+      "step": 92880
+    },
+    {
+      "epoch": 0.3752873539999273,
+      "grad_norm": 772.3094482421875,
+      "learning_rate": 1.777696507870128e-06,
+      "loss": 57.1095,
+      "step": 92890
+    },
+    {
+      "epoch": 0.3753277552652949,
+      "grad_norm": 674.185546875,
+      "learning_rate": 1.776618216614704e-06,
+      "loss": 37.1195,
+      "step": 92900
+    },
+    {
+      "epoch": 0.3753681565306625,
+      "grad_norm": 735.7659301757812,
+      "learning_rate": 1.7755401818289748e-06,
+      "loss": 48.0352,
+      "step": 92910
+    },
+    {
+      "epoch": 0.37540855779603016,
+      "grad_norm": 256.3131408691406,
+      "learning_rate": 1.774462403598715e-06,
+      "loss": 37.8858,
+      "step": 92920
+    },
+    {
+      "epoch": 0.3754489590613978,
+      "grad_norm": 484.47283935546875,
+      "learning_rate": 1.7733848820096789e-06,
+      "loss": 41.405,
+      "step": 92930
+    },
+    {
+      "epoch": 0.37548936032676544,
+      "grad_norm": 271.2965393066406,
+      "learning_rate": 1.7723076171475995e-06,
+      "loss": 32.3131,
+      "step": 92940
+    },
+    {
+      "epoch": 0.3755297615921331,
+      "grad_norm": 523.7734375,
+      "learning_rate": 1.7712306090981896e-06,
+      "loss": 34.7863,
+      "step": 92950
+    },
+    {
+      "epoch": 0.3755701628575007,
+      "grad_norm": 724.2642822265625,
+      "learning_rate": 1.7701538579471423e-06,
+      "loss": 45.1211,
+      "step": 92960
+    },
+    {
+      "epoch": 0.3756105641228683,
+      "grad_norm": 291.0926513671875,
+      "learning_rate": 1.7690773637801295e-06,
+      "loss": 39.4207,
+      "step": 92970
+    },
+    {
+      "epoch": 0.37565096538823595,
+      "grad_norm": 673.997314453125,
+      "learning_rate": 1.768001126682803e-06,
+      "loss": 59.2063,
+      "step": 92980
+    },
+    {
+      "epoch": 0.3756913666536036,
+      "grad_norm": 1037.6884765625,
+      "learning_rate": 1.7669251467407938e-06,
+      "loss": 26.6047,
+      "step": 92990
+    },
+    {
+      "epoch": 0.3757317679189712,
+      "grad_norm": 931.728759765625,
+      "learning_rate": 1.7658494240397127e-06,
+      "loss": 50.3487,
+      "step": 93000
+    },
+    {
+      "epoch": 0.37577216918433887,
+      "grad_norm": 617.3193969726562,
+      "learning_rate": 1.7647739586651508e-06,
+      "loss": 38.8255,
+      "step": 93010
+    },
+    {
+      "epoch": 0.3758125704497065,
+      "grad_norm": 591.8453979492188,
+      "learning_rate": 1.7636987507026787e-06,
+      "loss": 51.4436,
+      "step": 93020
+    },
+    {
+      "epoch": 0.3758529717150741,
+      "grad_norm": 428.7930603027344,
+      "learning_rate": 1.762623800237841e-06,
+      "loss": 44.514,
+      "step": 93030
+    },
+    {
+      "epoch": 0.37589337298044173,
+      "grad_norm": 644.0867309570312,
+      "learning_rate": 1.7615491073561714e-06,
+      "loss": 52.178,
+      "step": 93040
+    },
+    {
+      "epoch": 0.37593377424580937,
+      "grad_norm": 565.926025390625,
+      "learning_rate": 1.760474672143177e-06,
+      "loss": 30.3734,
+      "step": 93050
+    },
+    {
+      "epoch": 0.375974175511177,
+      "grad_norm": 1001.1051635742188,
+      "learning_rate": 1.7594004946843458e-06,
+      "loss": 62.9556,
+      "step": 93060
+    },
+    {
+      "epoch": 0.37601457677654465,
+      "grad_norm": 1308.181884765625,
+      "learning_rate": 1.7583265750651446e-06,
+      "loss": 55.2899,
+      "step": 93070
+    },
+    {
+      "epoch": 0.3760549780419123,
+      "grad_norm": 611.5211181640625,
+      "learning_rate": 1.7572529133710204e-06,
+      "loss": 45.4431,
+      "step": 93080
+    },
+    {
+      "epoch": 0.37609537930727993,
+      "grad_norm": 416.8019104003906,
+      "learning_rate": 1.7561795096874002e-06,
+      "loss": 57.3495,
+      "step": 93090
+    },
+    {
+      "epoch": 0.3761357805726475,
+      "grad_norm": 222.11227416992188,
+      "learning_rate": 1.755106364099689e-06,
+      "loss": 41.9647,
+      "step": 93100
+    },
+    {
+      "epoch": 0.37617618183801516,
+      "grad_norm": 804.1954345703125,
+      "learning_rate": 1.7540334766932738e-06,
+      "loss": 60.3842,
+      "step": 93110
+    },
+    {
+      "epoch": 0.3762165831033828,
+      "grad_norm": 446.4356994628906,
+      "learning_rate": 1.7529608475535165e-06,
+      "loss": 40.3711,
+      "step": 93120
+    },
+    {
+      "epoch": 0.37625698436875044,
+      "grad_norm": 681.1730346679688,
+      "learning_rate": 1.7518884767657612e-06,
+      "loss": 73.1313,
+      "step": 93130
+    },
+    {
+      "epoch": 0.3762973856341181,
+      "grad_norm": 540.1654663085938,
+      "learning_rate": 1.7508163644153342e-06,
+      "loss": 40.5221,
+      "step": 93140
+    },
+    {
+      "epoch": 0.3763377868994857,
+      "grad_norm": 765.6929321289062,
+      "learning_rate": 1.7497445105875377e-06,
+      "loss": 50.165,
+      "step": 93150
+    },
+    {
+      "epoch": 0.3763781881648533,
+      "grad_norm": 246.20120239257812,
+      "learning_rate": 1.7486729153676536e-06,
+      "loss": 21.7108,
+      "step": 93160
+    },
+    {
+      "epoch": 0.37641858943022094,
+      "grad_norm": 367.9482727050781,
+      "learning_rate": 1.7476015788409439e-06,
+      "loss": 36.8506,
+      "step": 93170
+    },
+    {
+      "epoch": 0.3764589906955886,
+      "grad_norm": 562.669677734375,
+      "learning_rate": 1.7465305010926503e-06,
+      "loss": 35.8054,
+      "step": 93180
+    },
+    {
+      "epoch": 0.3764993919609562,
+      "grad_norm": 644.0034790039062,
+      "learning_rate": 1.745459682207995e-06,
+      "loss": 35.9354,
+      "step": 93190
+    },
+    {
+      "epoch": 0.37653979322632386,
+      "grad_norm": 548.04052734375,
+      "learning_rate": 1.7443891222721749e-06,
+      "loss": 25.4771,
+      "step": 93200
+    },
+    {
+      "epoch": 0.3765801944916915,
+      "grad_norm": 531.5404663085938,
+      "learning_rate": 1.7433188213703712e-06,
+      "loss": 40.387,
+      "step": 93210
+    },
+    {
+      "epoch": 0.3766205957570591,
+      "grad_norm": 481.58502197265625,
+      "learning_rate": 1.7422487795877424e-06,
+      "loss": 45.3946,
+      "step": 93220
+    },
+    {
+      "epoch": 0.3766609970224267,
+      "grad_norm": 1088.45654296875,
+      "learning_rate": 1.7411789970094257e-06,
+      "loss": 44.2268,
+      "step": 93230
+    },
+    {
+      "epoch": 0.37670139828779436,
+      "grad_norm": 716.853271484375,
+      "learning_rate": 1.7401094737205415e-06,
+      "loss": 70.572,
+      "step": 93240
+    },
+    {
+      "epoch": 0.376741799553162,
+      "grad_norm": 666.8670043945312,
+      "learning_rate": 1.739040209806186e-06,
+      "loss": 56.1082,
+      "step": 93250
+    },
+    {
+      "epoch": 0.37678220081852964,
+      "grad_norm": 255.77133178710938,
+      "learning_rate": 1.7379712053514352e-06,
+      "loss": 28.3571,
+      "step": 93260
+    },
+    {
+      "epoch": 0.3768226020838973,
+      "grad_norm": 473.16607666015625,
+      "learning_rate": 1.736902460441345e-06,
+      "loss": 20.6927,
+      "step": 93270
+    },
+    {
+      "epoch": 0.3768630033492649,
+      "grad_norm": 417.4438171386719,
+      "learning_rate": 1.735833975160952e-06,
+      "loss": 42.4167,
+      "step": 93280
+    },
+    {
+      "epoch": 0.3769034046146325,
+      "grad_norm": 463.9417724609375,
+      "learning_rate": 1.7347657495952675e-06,
+      "loss": 36.2839,
+      "step": 93290
+    },
+    {
+      "epoch": 0.37694380588000015,
+      "grad_norm": 874.1891479492188,
+      "learning_rate": 1.7336977838292867e-06,
+      "loss": 43.4433,
+      "step": 93300
+    },
+    {
+      "epoch": 0.3769842071453678,
+      "grad_norm": 634.763916015625,
+      "learning_rate": 1.7326300779479826e-06,
+      "loss": 37.5954,
+      "step": 93310
+    },
+    {
+      "epoch": 0.37702460841073543,
+      "grad_norm": 394.0238342285156,
+      "learning_rate": 1.731562632036307e-06,
+      "loss": 38.361,
+      "step": 93320
+    },
+    {
+      "epoch": 0.37706500967610307,
+      "grad_norm": 667.3516845703125,
+      "learning_rate": 1.730495446179194e-06,
+      "loss": 39.4061,
+      "step": 93330
+    },
+    {
+      "epoch": 0.3771054109414707,
+      "grad_norm": 703.3050537109375,
+      "learning_rate": 1.7294285204615536e-06,
+      "loss": 57.0635,
+      "step": 93340
+    },
+    {
+      "epoch": 0.3771458122068383,
+      "grad_norm": 562.1600341796875,
+      "learning_rate": 1.7283618549682757e-06,
+      "loss": 33.8003,
+      "step": 93350
+    },
+    {
+      "epoch": 0.37718621347220593,
+      "grad_norm": 707.1860961914062,
+      "learning_rate": 1.727295449784232e-06,
+      "loss": 35.2757,
+      "step": 93360
+    },
+    {
+      "epoch": 0.3772266147375736,
+      "grad_norm": 540.1780395507812,
+      "learning_rate": 1.726229304994268e-06,
+      "loss": 41.1969,
+      "step": 93370
+    },
+    {
+      "epoch": 0.3772670160029412,
+      "grad_norm": 282.990234375,
+      "learning_rate": 1.7251634206832135e-06,
+      "loss": 44.1028,
+      "step": 93380
+    },
+    {
+      "epoch": 0.37730741726830885,
+      "grad_norm": 473.59356689453125,
+      "learning_rate": 1.7240977969358757e-06,
+      "loss": 27.7578,
+      "step": 93390
+    },
+    {
+      "epoch": 0.3773478185336765,
+      "grad_norm": 718.213134765625,
+      "learning_rate": 1.7230324338370425e-06,
+      "loss": 38.0047,
+      "step": 93400
+    },
+    {
+      "epoch": 0.37738821979904413,
+      "grad_norm": 972.40283203125,
+      "learning_rate": 1.721967331471479e-06,
+      "loss": 41.0903,
+      "step": 93410
+    },
+    {
+      "epoch": 0.3774286210644117,
+      "grad_norm": 570.8587036132812,
+      "learning_rate": 1.7209024899239297e-06,
+      "loss": 46.1088,
+      "step": 93420
+    },
+    {
+      "epoch": 0.37746902232977936,
+      "grad_norm": 2588.669189453125,
+      "learning_rate": 1.7198379092791213e-06,
+      "loss": 43.5806,
+      "step": 93430
+    },
+    {
+      "epoch": 0.377509423595147,
+      "grad_norm": 710.2998657226562,
+      "learning_rate": 1.7187735896217567e-06,
+      "loss": 30.8053,
+      "step": 93440
+    },
+    {
+      "epoch": 0.37754982486051464,
+      "grad_norm": 534.6858520507812,
+      "learning_rate": 1.7177095310365205e-06,
+      "loss": 34.1006,
+      "step": 93450
+    },
+    {
+      "epoch": 0.3775902261258823,
+      "grad_norm": 371.9433898925781,
+      "learning_rate": 1.7166457336080716e-06,
+      "loss": 32.8527,
+      "step": 93460
+    },
+    {
+      "epoch": 0.3776306273912499,
+      "grad_norm": 568.0192260742188,
+      "learning_rate": 1.715582197421053e-06,
+      "loss": 39.4708,
+      "step": 93470
+    },
+    {
+      "epoch": 0.3776710286566175,
+      "grad_norm": 374.3291320800781,
+      "learning_rate": 1.7145189225600856e-06,
+      "loss": 35.2344,
+      "step": 93480
+    },
+    {
+      "epoch": 0.37771142992198514,
+      "grad_norm": 666.5590209960938,
+      "learning_rate": 1.7134559091097691e-06,
+      "loss": 33.7738,
+      "step": 93490
+    },
+    {
+      "epoch": 0.3777518311873528,
+      "grad_norm": 950.9321899414062,
+      "learning_rate": 1.7123931571546826e-06,
+      "loss": 37.9408,
+      "step": 93500
+    },
+    {
+      "epoch": 0.3777922324527204,
+      "grad_norm": 535.5814208984375,
+      "learning_rate": 1.711330666779385e-06,
+      "loss": 57.5564,
+      "step": 93510
+    },
+    {
+      "epoch": 0.37783263371808806,
+      "grad_norm": 379.9372253417969,
+      "learning_rate": 1.7102684380684109e-06,
+      "loss": 37.9954,
+      "step": 93520
+    },
+    {
+      "epoch": 0.3778730349834557,
+      "grad_norm": 724.26220703125,
+      "learning_rate": 1.7092064711062816e-06,
+      "loss": 70.6237,
+      "step": 93530
+    },
+    {
+      "epoch": 0.3779134362488233,
+      "grad_norm": 755.5435791015625,
+      "learning_rate": 1.708144765977492e-06,
+      "loss": 33.0022,
+      "step": 93540
+    },
+    {
+      "epoch": 0.3779538375141909,
+      "grad_norm": 487.3462219238281,
+      "learning_rate": 1.7070833227665146e-06,
+      "loss": 53.9496,
+      "step": 93550
+    },
+    {
+      "epoch": 0.37799423877955857,
+      "grad_norm": 503.11114501953125,
+      "learning_rate": 1.7060221415578042e-06,
+      "loss": 33.7394,
+      "step": 93560
+    },
+    {
+      "epoch": 0.3780346400449262,
+      "grad_norm": 771.644287109375,
+      "learning_rate": 1.7049612224357954e-06,
+      "loss": 40.8785,
+      "step": 93570
+    },
+    {
+      "epoch": 0.37807504131029385,
+      "grad_norm": 934.612060546875,
+      "learning_rate": 1.703900565484899e-06,
+      "loss": 39.6854,
+      "step": 93580
+    },
+    {
+      "epoch": 0.3781154425756615,
+      "grad_norm": 567.306640625,
+      "learning_rate": 1.7028401707895082e-06,
+      "loss": 51.0686,
+      "step": 93590
+    },
+    {
+      "epoch": 0.3781558438410291,
+      "grad_norm": 150.63609313964844,
+      "learning_rate": 1.7017800384339928e-06,
+      "loss": 32.8338,
+      "step": 93600
+    },
+    {
+      "epoch": 0.3781962451063967,
+      "grad_norm": 863.6294555664062,
+      "learning_rate": 1.700720168502703e-06,
+      "loss": 52.1523,
+      "step": 93610
+    },
+    {
+      "epoch": 0.37823664637176435,
+      "grad_norm": 652.1657104492188,
+      "learning_rate": 1.6996605610799682e-06,
+      "loss": 30.39,
+      "step": 93620
+    },
+    {
+      "epoch": 0.378277047637132,
+      "grad_norm": 402.42376708984375,
+      "learning_rate": 1.6986012162500953e-06,
+      "loss": 28.3965,
+      "step": 93630
+    },
+    {
+      "epoch": 0.37831744890249963,
+      "grad_norm": 489.7448425292969,
+      "learning_rate": 1.697542134097373e-06,
+      "loss": 60.2797,
+      "step": 93640
+    },
+    {
+      "epoch": 0.37835785016786727,
+      "grad_norm": 696.7704467773438,
+      "learning_rate": 1.6964833147060661e-06,
+      "loss": 42.7316,
+      "step": 93650
+    },
+    {
+      "epoch": 0.3783982514332349,
+      "grad_norm": 561.4448852539062,
+      "learning_rate": 1.6954247581604216e-06,
+      "loss": 37.4174,
+      "step": 93660
+    },
+    {
+      "epoch": 0.3784386526986025,
+      "grad_norm": 652.9007568359375,
+      "learning_rate": 1.6943664645446622e-06,
+      "loss": 50.4149,
+      "step": 93670
+    },
+    {
+      "epoch": 0.37847905396397014,
+      "grad_norm": 342.3523864746094,
+      "learning_rate": 1.6933084339429935e-06,
+      "loss": 40.2572,
+      "step": 93680
+    },
+    {
+      "epoch": 0.3785194552293378,
+      "grad_norm": 428.1909484863281,
+      "learning_rate": 1.692250666439596e-06,
+      "loss": 25.0063,
+      "step": 93690
+    },
+    {
+      "epoch": 0.3785598564947054,
+      "grad_norm": 569.578369140625,
+      "learning_rate": 1.6911931621186329e-06,
+      "loss": 45.803,
+      "step": 93700
+    },
+    {
+      "epoch": 0.37860025776007306,
+      "grad_norm": 1233.8858642578125,
+      "learning_rate": 1.6901359210642444e-06,
+      "loss": 49.5512,
+      "step": 93710
+    },
+    {
+      "epoch": 0.3786406590254407,
+      "grad_norm": 615.4265747070312,
+      "learning_rate": 1.6890789433605508e-06,
+      "loss": 44.6923,
+      "step": 93720
+    },
+    {
+      "epoch": 0.37868106029080834,
+      "grad_norm": 457.422119140625,
+      "learning_rate": 1.6880222290916503e-06,
+      "loss": 51.6429,
+      "step": 93730
+    },
+    {
+      "epoch": 0.3787214615561759,
+      "grad_norm": 483.12017822265625,
+      "learning_rate": 1.686965778341621e-06,
+      "loss": 54.6015,
+      "step": 93740
+    },
+    {
+      "epoch": 0.37876186282154356,
+      "grad_norm": 637.249755859375,
+      "learning_rate": 1.68590959119452e-06,
+      "loss": 38.6967,
+      "step": 93750
+    },
+    {
+      "epoch": 0.3788022640869112,
+      "grad_norm": 549.8311767578125,
+      "learning_rate": 1.6848536677343836e-06,
+      "loss": 43.5703,
+      "step": 93760
+    },
+    {
+      "epoch": 0.37884266535227884,
+      "grad_norm": 885.5505981445312,
+      "learning_rate": 1.683798008045226e-06,
+      "loss": 37.328,
+      "step": 93770
+    },
+    {
+      "epoch": 0.3788830666176465,
+      "grad_norm": 344.5069274902344,
+      "learning_rate": 1.6827426122110412e-06,
+      "loss": 44.6456,
+      "step": 93780
+    },
+    {
+      "epoch": 0.3789234678830141,
+      "grad_norm": 464.1234436035156,
+      "learning_rate": 1.6816874803158034e-06,
+      "loss": 71.6938,
+      "step": 93790
+    },
+    {
+      "epoch": 0.3789638691483817,
+      "grad_norm": 640.6181030273438,
+      "learning_rate": 1.6806326124434634e-06,
+      "loss": 61.3986,
+      "step": 93800
+    },
+    {
+      "epoch": 0.37900427041374934,
+      "grad_norm": 549.3226318359375,
+      "learning_rate": 1.679578008677953e-06,
+      "loss": 46.0698,
+      "step": 93810
+    },
+    {
+      "epoch": 0.379044671679117,
+      "grad_norm": 782.4042358398438,
+      "learning_rate": 1.6785236691031808e-06,
+      "loss": 57.5303,
+      "step": 93820
+    },
+    {
+      "epoch": 0.3790850729444846,
+      "grad_norm": 581.6171875,
+      "learning_rate": 1.6774695938030378e-06,
+      "loss": 54.6885,
+      "step": 93830
+    },
+    {
+      "epoch": 0.37912547420985226,
+      "grad_norm": 858.0625,
+      "learning_rate": 1.6764157828613902e-06,
+      "loss": 40.6032,
+      "step": 93840
+    },
+    {
+      "epoch": 0.3791658754752199,
+      "grad_norm": 550.3126220703125,
+      "learning_rate": 1.675362236362086e-06,
+      "loss": 80.7207,
+      "step": 93850
+    },
+    {
+      "epoch": 0.3792062767405875,
+      "grad_norm": 340.3455810546875,
+      "learning_rate": 1.6743089543889502e-06,
+      "loss": 35.5949,
+      "step": 93860
+    },
+    {
+      "epoch": 0.37924667800595513,
+      "grad_norm": 271.68463134765625,
+      "learning_rate": 1.6732559370257884e-06,
+      "loss": 42.7064,
+      "step": 93870
+    },
+    {
+      "epoch": 0.37928707927132277,
+      "grad_norm": 528.7034912109375,
+      "learning_rate": 1.6722031843563836e-06,
+      "loss": 43.5807,
+      "step": 93880
+    },
+    {
+      "epoch": 0.3793274805366904,
+      "grad_norm": 365.6889953613281,
+      "learning_rate": 1.6711506964644992e-06,
+      "loss": 41.8635,
+      "step": 93890
+    },
+    {
+      "epoch": 0.37936788180205805,
+      "grad_norm": 153.63197326660156,
+      "learning_rate": 1.6700984734338765e-06,
+      "loss": 30.3246,
+      "step": 93900
+    },
+    {
+      "epoch": 0.3794082830674257,
+      "grad_norm": 293.03607177734375,
+      "learning_rate": 1.669046515348236e-06,
+      "loss": 38.6674,
+      "step": 93910
+    },
+    {
+      "epoch": 0.37944868433279333,
+      "grad_norm": 653.3015747070312,
+      "learning_rate": 1.6679948222912773e-06,
+      "loss": 39.9429,
+      "step": 93920
+    },
+    {
+      "epoch": 0.3794890855981609,
+      "grad_norm": 658.6498413085938,
+      "learning_rate": 1.6669433943466789e-06,
+      "loss": 43.2089,
+      "step": 93930
+    },
+    {
+      "epoch": 0.37952948686352855,
+      "grad_norm": 335.7240905761719,
+      "learning_rate": 1.6658922315980975e-06,
+      "loss": 33.1409,
+      "step": 93940
+    },
+    {
+      "epoch": 0.3795698881288962,
+      "grad_norm": 508.2664489746094,
+      "learning_rate": 1.6648413341291703e-06,
+      "loss": 47.2963,
+      "step": 93950
+    },
+    {
+      "epoch": 0.37961028939426383,
+      "grad_norm": 690.1048583984375,
+      "learning_rate": 1.6637907020235117e-06,
+      "loss": 57.5872,
+      "step": 93960
+    },
+    {
+      "epoch": 0.3796506906596315,
+      "grad_norm": 659.0160522460938,
+      "learning_rate": 1.662740335364717e-06,
+      "loss": 39.8609,
+      "step": 93970
+    },
+    {
+      "epoch": 0.3796910919249991,
+      "grad_norm": 615.6521606445312,
+      "learning_rate": 1.661690234236355e-06,
+      "loss": 29.5961,
+      "step": 93980
+    },
+    {
+      "epoch": 0.3797314931903667,
+      "grad_norm": 206.93682861328125,
+      "learning_rate": 1.6606403987219815e-06,
+      "loss": 38.3237,
+      "step": 93990
+    },
+    {
+      "epoch": 0.37977189445573434,
+      "grad_norm": 464.7280578613281,
+      "learning_rate": 1.6595908289051266e-06,
+      "loss": 53.9895,
+      "step": 94000
+    },
+    {
+      "epoch": 0.379812295721102,
+      "grad_norm": 509.655517578125,
+      "learning_rate": 1.6585415248692988e-06,
+      "loss": 30.5599,
+      "step": 94010
+    },
+    {
+      "epoch": 0.3798526969864696,
+      "grad_norm": 773.5880126953125,
+      "learning_rate": 1.6574924866979863e-06,
+      "loss": 32.8057,
+      "step": 94020
+    },
+    {
+      "epoch": 0.37989309825183726,
+      "grad_norm": 350.2411193847656,
+      "learning_rate": 1.6564437144746564e-06,
+      "loss": 35.1408,
+      "step": 94030
+    },
+    {
+      "epoch": 0.3799334995172049,
+      "grad_norm": 837.4682006835938,
+      "learning_rate": 1.6553952082827562e-06,
+      "loss": 49.9956,
+      "step": 94040
+    },
+    {
+      "epoch": 0.37997390078257254,
+      "grad_norm": 509.6433410644531,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 54.7281,
+      "step": 94050
+    },
+    {
+      "epoch": 0.3800143020479401,
+      "grad_norm": 472.0282897949219,
+      "learning_rate": 1.6532989943269207e-06,
+      "loss": 37.2688,
+      "step": 94060
+    },
+    {
+      "epoch": 0.38005470331330776,
+      "grad_norm": 635.9703979492188,
+      "learning_rate": 1.6522512867297707e-06,
+      "loss": 45.623,
+      "step": 94070
+    },
+    {
+      "epoch": 0.3800951045786754,
+      "grad_norm": 397.69342041015625,
+      "learning_rate": 1.6512038454976198e-06,
+      "loss": 43.5267,
+      "step": 94080
+    },
+    {
+      "epoch": 0.38013550584404304,
+      "grad_norm": 46.28896713256836,
+      "learning_rate": 1.6501566707138116e-06,
+      "loss": 22.4504,
+      "step": 94090
+    },
+    {
+      "epoch": 0.3801759071094107,
+      "grad_norm": 543.3253784179688,
+      "learning_rate": 1.6491097624616637e-06,
+      "loss": 35.8103,
+      "step": 94100
+    },
+    {
+      "epoch": 0.3802163083747783,
+      "grad_norm": 1130.6624755859375,
+      "learning_rate": 1.6480631208244735e-06,
+      "loss": 44.4341,
+      "step": 94110
+    },
+    {
+      "epoch": 0.3802567096401459,
+      "grad_norm": 398.57989501953125,
+      "learning_rate": 1.6470167458855174e-06,
+      "loss": 34.3871,
+      "step": 94120
+    },
+    {
+      "epoch": 0.38029711090551355,
+      "grad_norm": 889.4392700195312,
+      "learning_rate": 1.645970637728051e-06,
+      "loss": 75.1114,
+      "step": 94130
+    },
+    {
+      "epoch": 0.3803375121708812,
+      "grad_norm": 800.8850708007812,
+      "learning_rate": 1.6449247964353094e-06,
+      "loss": 43.0828,
+      "step": 94140
+    },
+    {
+      "epoch": 0.3803779134362488,
+      "grad_norm": 571.4644165039062,
+      "learning_rate": 1.643879222090502e-06,
+      "loss": 33.5428,
+      "step": 94150
+    },
+    {
+      "epoch": 0.38041831470161647,
+      "grad_norm": 795.34716796875,
+      "learning_rate": 1.642833914776823e-06,
+      "loss": 34.8854,
+      "step": 94160
+    },
+    {
+      "epoch": 0.3804587159669841,
+      "grad_norm": 645.9882202148438,
+      "learning_rate": 1.6417888745774418e-06,
+      "loss": 35.6082,
+      "step": 94170
+    },
+    {
+      "epoch": 0.3804991172323517,
+      "grad_norm": 530.788330078125,
+      "learning_rate": 1.640744101575506e-06,
+      "loss": 51.9296,
+      "step": 94180
+    },
+    {
+      "epoch": 0.38053951849771933,
+      "grad_norm": 446.3039855957031,
+      "learning_rate": 1.6396995958541468e-06,
+      "loss": 47.5382,
+      "step": 94190
+    },
+    {
+      "epoch": 0.38057991976308697,
+      "grad_norm": 371.99151611328125,
+      "learning_rate": 1.6386553574964691e-06,
+      "loss": 29.8966,
+      "step": 94200
+    },
+    {
+      "epoch": 0.3806203210284546,
+      "grad_norm": 950.2408447265625,
+      "learning_rate": 1.6376113865855585e-06,
+      "loss": 49.887,
+      "step": 94210
+    },
+    {
+      "epoch": 0.38066072229382225,
+      "grad_norm": 445.62725830078125,
+      "learning_rate": 1.6365676832044796e-06,
+      "loss": 45.9934,
+      "step": 94220
+    },
+    {
+      "epoch": 0.3807011235591899,
+      "grad_norm": 718.4339599609375,
+      "learning_rate": 1.6355242474362732e-06,
+      "loss": 30.1935,
+      "step": 94230
+    },
+    {
+      "epoch": 0.38074152482455753,
+      "grad_norm": 719.7804565429688,
+      "learning_rate": 1.634481079363961e-06,
+      "loss": 39.0847,
+      "step": 94240
+    },
+    {
+      "epoch": 0.3807819260899251,
+      "grad_norm": 529.1168212890625,
+      "learning_rate": 1.6334381790705439e-06,
+      "loss": 71.6443,
+      "step": 94250
+    },
+    {
+      "epoch": 0.38082232735529276,
+      "grad_norm": 499.4634094238281,
+      "learning_rate": 1.6323955466390001e-06,
+      "loss": 30.6561,
+      "step": 94260
+    },
+    {
+      "epoch": 0.3808627286206604,
+      "grad_norm": 327.93621826171875,
+      "learning_rate": 1.6313531821522876e-06,
+      "loss": 39.9715,
+      "step": 94270
+    },
+    {
+      "epoch": 0.38090312988602804,
+      "grad_norm": 888.1876831054688,
+      "learning_rate": 1.6303110856933413e-06,
+      "loss": 33.8202,
+      "step": 94280
+    },
+    {
+      "epoch": 0.3809435311513957,
+      "grad_norm": 438.6484680175781,
+      "learning_rate": 1.629269257345078e-06,
+      "loss": 38.3931,
+      "step": 94290
+    },
+    {
+      "epoch": 0.3809839324167633,
+      "grad_norm": 420.45562744140625,
+      "learning_rate": 1.628227697190391e-06,
+      "loss": 42.586,
+      "step": 94300
+    },
+    {
+      "epoch": 0.3810243336821309,
+      "grad_norm": 732.1066284179688,
+      "learning_rate": 1.6271864053121528e-06,
+      "loss": 35.4111,
+      "step": 94310
+    },
+    {
+      "epoch": 0.38106473494749854,
+      "grad_norm": 625.7940063476562,
+      "learning_rate": 1.6261453817932122e-06,
+      "loss": 37.4915,
+      "step": 94320
+    },
+    {
+      "epoch": 0.3811051362128662,
+      "grad_norm": 635.6824340820312,
+      "learning_rate": 1.6251046267163988e-06,
+      "loss": 33.4941,
+      "step": 94330
+    },
+    {
+      "epoch": 0.3811455374782338,
+      "grad_norm": 484.083984375,
+      "learning_rate": 1.6240641401645224e-06,
+      "loss": 42.375,
+      "step": 94340
+    },
+    {
+      "epoch": 0.38118593874360146,
+      "grad_norm": 810.295654296875,
+      "learning_rate": 1.6230239222203687e-06,
+      "loss": 57.4285,
+      "step": 94350
+    },
+    {
+      "epoch": 0.3812263400089691,
+      "grad_norm": 357.74725341796875,
+      "learning_rate": 1.621983972966703e-06,
+      "loss": 44.1577,
+      "step": 94360
+    },
+    {
+      "epoch": 0.3812667412743367,
+      "grad_norm": 466.22796630859375,
+      "learning_rate": 1.6209442924862684e-06,
+      "loss": 48.3063,
+      "step": 94370
+    },
+    {
+      "epoch": 0.3813071425397043,
+      "grad_norm": 590.7434692382812,
+      "learning_rate": 1.6199048808617896e-06,
+      "loss": 39.6602,
+      "step": 94380
+    },
+    {
+      "epoch": 0.38134754380507196,
+      "grad_norm": 316.4775085449219,
+      "learning_rate": 1.6188657381759676e-06,
+      "loss": 31.71,
+      "step": 94390
+    },
+    {
+      "epoch": 0.3813879450704396,
+      "grad_norm": 1969.42626953125,
+      "learning_rate": 1.6178268645114826e-06,
+      "loss": 42.9319,
+      "step": 94400
+    },
+    {
+      "epoch": 0.38142834633580724,
+      "grad_norm": 799.62744140625,
+      "learning_rate": 1.6167882599509904e-06,
+      "loss": 39.8405,
+      "step": 94410
+    },
+    {
+      "epoch": 0.3814687476011749,
+      "grad_norm": 479.5269775390625,
+      "learning_rate": 1.6157499245771296e-06,
+      "loss": 41.1756,
+      "step": 94420
+    },
+    {
+      "epoch": 0.3815091488665425,
+      "grad_norm": 762.2589721679688,
+      "learning_rate": 1.6147118584725163e-06,
+      "loss": 42.6559,
+      "step": 94430
+    },
+    {
+      "epoch": 0.3815495501319101,
+      "grad_norm": 319.53082275390625,
+      "learning_rate": 1.6136740617197433e-06,
+      "loss": 40.6062,
+      "step": 94440
+    },
+    {
+      "epoch": 0.38158995139727775,
+      "grad_norm": 497.3779296875,
+      "learning_rate": 1.612636534401384e-06,
+      "loss": 34.069,
+      "step": 94450
+    },
+    {
+      "epoch": 0.3816303526626454,
+      "grad_norm": 530.7467041015625,
+      "learning_rate": 1.61159927659999e-06,
+      "loss": 28.3256,
+      "step": 94460
+    },
+    {
+      "epoch": 0.38167075392801303,
+      "grad_norm": 1004.8090209960938,
+      "learning_rate": 1.6105622883980893e-06,
+      "loss": 56.0737,
+      "step": 94470
+    },
+    {
+      "epoch": 0.38171115519338067,
+      "grad_norm": 660.3544311523438,
+      "learning_rate": 1.6095255698781954e-06,
+      "loss": 44.8425,
+      "step": 94480
+    },
+    {
+      "epoch": 0.3817515564587483,
+      "grad_norm": 487.4358215332031,
+      "learning_rate": 1.6084891211227899e-06,
+      "loss": 35.9278,
+      "step": 94490
+    },
+    {
+      "epoch": 0.3817919577241159,
+      "grad_norm": 572.1942749023438,
+      "learning_rate": 1.6074529422143398e-06,
+      "loss": 51.2289,
+      "step": 94500
+    },
+    {
+      "epoch": 0.38183235898948353,
+      "grad_norm": 584.8763427734375,
+      "learning_rate": 1.6064170332352897e-06,
+      "loss": 35.8506,
+      "step": 94510
+    },
+    {
+      "epoch": 0.3818727602548512,
+      "grad_norm": 854.034423828125,
+      "learning_rate": 1.6053813942680618e-06,
+      "loss": 60.1789,
+      "step": 94520
+    },
+    {
+      "epoch": 0.3819131615202188,
+      "grad_norm": 959.2164916992188,
+      "learning_rate": 1.604346025395057e-06,
+      "loss": 33.7964,
+      "step": 94530
+    },
+    {
+      "epoch": 0.38195356278558645,
+      "grad_norm": 408.621337890625,
+      "learning_rate": 1.6033109266986552e-06,
+      "loss": 19.004,
+      "step": 94540
+    },
+    {
+      "epoch": 0.3819939640509541,
+      "grad_norm": 713.4255981445312,
+      "learning_rate": 1.602276098261214e-06,
+      "loss": 37.2563,
+      "step": 94550
+    },
+    {
+      "epoch": 0.38203436531632173,
+      "grad_norm": 722.92724609375,
+      "learning_rate": 1.6012415401650706e-06,
+      "loss": 35.3043,
+      "step": 94560
+    },
+    {
+      "epoch": 0.3820747665816893,
+      "grad_norm": 515.4312744140625,
+      "learning_rate": 1.6002072524925395e-06,
+      "loss": 46.4359,
+      "step": 94570
+    },
+    {
+      "epoch": 0.38211516784705696,
+      "grad_norm": 202.4333953857422,
+      "learning_rate": 1.5991732353259142e-06,
+      "loss": 52.5165,
+      "step": 94580
+    },
+    {
+      "epoch": 0.3821555691124246,
+      "grad_norm": 406.05755615234375,
+      "learning_rate": 1.598139488747467e-06,
+      "loss": 48.7197,
+      "step": 94590
+    },
+    {
+      "epoch": 0.38219597037779224,
+      "grad_norm": 581.1624755859375,
+      "learning_rate": 1.5971060128394483e-06,
+      "loss": 34.375,
+      "step": 94600
+    },
+    {
+      "epoch": 0.3822363716431599,
+      "grad_norm": 655.8013916015625,
+      "learning_rate": 1.596072807684087e-06,
+      "loss": 56.5818,
+      "step": 94610
+    },
+    {
+      "epoch": 0.3822767729085275,
+      "grad_norm": 336.11737060546875,
+      "learning_rate": 1.5950398733635903e-06,
+      "loss": 37.6246,
+      "step": 94620
+    },
+    {
+      "epoch": 0.3823171741738951,
+      "grad_norm": 669.8641357421875,
+      "learning_rate": 1.5940072099601446e-06,
+      "loss": 41.239,
+      "step": 94630
+    },
+    {
+      "epoch": 0.38235757543926274,
+      "grad_norm": 644.0601806640625,
+      "learning_rate": 1.5929748175559135e-06,
+      "loss": 30.1086,
+      "step": 94640
+    },
+    {
+      "epoch": 0.3823979767046304,
+      "grad_norm": 625.1327514648438,
+      "learning_rate": 1.5919426962330398e-06,
+      "loss": 35.999,
+      "step": 94650
+    },
+    {
+      "epoch": 0.382438377969998,
+      "grad_norm": 1454.2506103515625,
+      "learning_rate": 1.5909108460736455e-06,
+      "loss": 41.8045,
+      "step": 94660
+    },
+    {
+      "epoch": 0.38247877923536566,
+      "grad_norm": 456.4172668457031,
+      "learning_rate": 1.589879267159829e-06,
+      "loss": 37.1018,
+      "step": 94670
+    },
+    {
+      "epoch": 0.3825191805007333,
+      "grad_norm": 652.0750122070312,
+      "learning_rate": 1.5888479595736695e-06,
+      "loss": 52.8909,
+      "step": 94680
+    },
+    {
+      "epoch": 0.3825595817661009,
+      "grad_norm": 454.7263488769531,
+      "learning_rate": 1.5878169233972218e-06,
+      "loss": 30.1145,
+      "step": 94690
+    },
+    {
+      "epoch": 0.3825999830314685,
+      "grad_norm": 155.27793884277344,
+      "learning_rate": 1.5867861587125228e-06,
+      "loss": 33.1408,
+      "step": 94700
+    },
+    {
+      "epoch": 0.38264038429683617,
+      "grad_norm": 857.6797485351562,
+      "learning_rate": 1.5857556656015837e-06,
+      "loss": 90.8161,
+      "step": 94710
+    },
+    {
+      "epoch": 0.3826807855622038,
+      "grad_norm": 497.7012634277344,
+      "learning_rate": 1.5847254441463978e-06,
+      "loss": 41.9356,
+      "step": 94720
+    },
+    {
+      "epoch": 0.38272118682757145,
+      "grad_norm": 1495.0457763671875,
+      "learning_rate": 1.583695494428934e-06,
+      "loss": 69.8086,
+      "step": 94730
+    },
+    {
+      "epoch": 0.3827615880929391,
+      "grad_norm": 652.8819580078125,
+      "learning_rate": 1.5826658165311409e-06,
+      "loss": 36.6699,
+      "step": 94740
+    },
+    {
+      "epoch": 0.3828019893583067,
+      "grad_norm": 773.2257690429688,
+      "learning_rate": 1.5816364105349451e-06,
+      "loss": 41.6954,
+      "step": 94750
+    },
+    {
+      "epoch": 0.3828423906236743,
+      "grad_norm": 373.1732177734375,
+      "learning_rate": 1.5806072765222524e-06,
+      "loss": 41.5614,
+      "step": 94760
+    },
+    {
+      "epoch": 0.38288279188904195,
+      "grad_norm": 578.3794555664062,
+      "learning_rate": 1.5795784145749453e-06,
+      "loss": 42.0277,
+      "step": 94770
+    },
+    {
+      "epoch": 0.3829231931544096,
+      "grad_norm": 700.9031372070312,
+      "learning_rate": 1.5785498247748864e-06,
+      "loss": 45.0889,
+      "step": 94780
+    },
+    {
+      "epoch": 0.38296359441977723,
+      "grad_norm": 719.7659301757812,
+      "learning_rate": 1.5775215072039157e-06,
+      "loss": 30.6197,
+      "step": 94790
+    },
+    {
+      "epoch": 0.38300399568514487,
+      "grad_norm": 396.6778259277344,
+      "learning_rate": 1.5764934619438515e-06,
+      "loss": 60.4507,
+      "step": 94800
+    },
+    {
+      "epoch": 0.3830443969505125,
+      "grad_norm": 715.6949462890625,
+      "learning_rate": 1.5754656890764912e-06,
+      "loss": 38.4569,
+      "step": 94810
+    },
+    {
+      "epoch": 0.3830847982158801,
+      "grad_norm": 518.1155395507812,
+      "learning_rate": 1.5744381886836091e-06,
+      "loss": 45.3779,
+      "step": 94820
+    },
+    {
+      "epoch": 0.38312519948124774,
+      "grad_norm": 495.7441101074219,
+      "learning_rate": 1.5734109608469612e-06,
+      "loss": 39.1621,
+      "step": 94830
+    },
+    {
+      "epoch": 0.3831656007466154,
+      "grad_norm": 888.2919311523438,
+      "learning_rate": 1.5723840056482731e-06,
+      "loss": 45.9851,
+      "step": 94840
+    },
+    {
+      "epoch": 0.383206002011983,
+      "grad_norm": 582.6542358398438,
+      "learning_rate": 1.5713573231692613e-06,
+      "loss": 31.6152,
+      "step": 94850
+    },
+    {
+      "epoch": 0.38324640327735066,
+      "grad_norm": 399.1988525390625,
+      "learning_rate": 1.5703309134916116e-06,
+      "loss": 29.2875,
+      "step": 94860
+    },
+    {
+      "epoch": 0.3832868045427183,
+      "grad_norm": 770.6976318359375,
+      "learning_rate": 1.5693047766969916e-06,
+      "loss": 44.1505,
+      "step": 94870
+    },
+    {
+      "epoch": 0.38332720580808594,
+      "grad_norm": 67.18172454833984,
+      "learning_rate": 1.568278912867045e-06,
+      "loss": 47.0008,
+      "step": 94880
+    },
+    {
+      "epoch": 0.3833676070734535,
+      "grad_norm": 550.394775390625,
+      "learning_rate": 1.5672533220833962e-06,
+      "loss": 31.0633,
+      "step": 94890
+    },
+    {
+      "epoch": 0.38340800833882116,
+      "grad_norm": 470.9195251464844,
+      "learning_rate": 1.5662280044276467e-06,
+      "loss": 58.1111,
+      "step": 94900
+    },
+    {
+      "epoch": 0.3834484096041888,
+      "grad_norm": 434.5940246582031,
+      "learning_rate": 1.5652029599813773e-06,
+      "loss": 68.5302,
+      "step": 94910
+    },
+    {
+      "epoch": 0.38348881086955644,
+      "grad_norm": 886.968505859375,
+      "learning_rate": 1.564178188826143e-06,
+      "loss": 52.7653,
+      "step": 94920
+    },
+    {
+      "epoch": 0.3835292121349241,
+      "grad_norm": 858.910888671875,
+      "learning_rate": 1.5631536910434807e-06,
+      "loss": 37.843,
+      "step": 94930
+    },
+    {
+      "epoch": 0.3835696134002917,
+      "grad_norm": 523.4095458984375,
+      "learning_rate": 1.5621294667149079e-06,
+      "loss": 44.0606,
+      "step": 94940
+    },
+    {
+      "epoch": 0.3836100146656593,
+      "grad_norm": 576.0836791992188,
+      "learning_rate": 1.561105515921915e-06,
+      "loss": 52.5642,
+      "step": 94950
+    },
+    {
+      "epoch": 0.38365041593102694,
+      "grad_norm": 454.44219970703125,
+      "learning_rate": 1.5600818387459748e-06,
+      "loss": 36.4562,
+      "step": 94960
+    },
+    {
+      "epoch": 0.3836908171963946,
+      "grad_norm": 704.0636596679688,
+      "learning_rate": 1.559058435268535e-06,
+      "loss": 43.8542,
+      "step": 94970
+    },
+    {
+      "epoch": 0.3837312184617622,
+      "grad_norm": 521.6298217773438,
+      "learning_rate": 1.558035305571024e-06,
+      "loss": 38.1112,
+      "step": 94980
+    },
+    {
+      "epoch": 0.38377161972712986,
+      "grad_norm": 363.50469970703125,
+      "learning_rate": 1.5570124497348466e-06,
+      "loss": 38.1875,
+      "step": 94990
+    },
+    {
+      "epoch": 0.3838120209924975,
+      "grad_norm": 851.0789184570312,
+      "learning_rate": 1.5559898678413898e-06,
+      "loss": 43.4976,
+      "step": 95000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 123750,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}