diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,3659 +1,43769 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.0,
+  "epoch": 2.0,
   "eval_steps": 500,
-  "global_step": 15625,
+  "global_step": 62500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.00192,
-      "grad_norm": 3.3767759799957275,
-      "learning_rate": 1.9193857965451055e-07,
-      "loss": 1.9711,
+      "epoch": 0.00032,
+      "grad_norm": 116.36629486083984,
+      "learning_rate": 3.2e-08,
+      "loss": 2.146,
+      "step": 10
+    },
+    {
+      "epoch": 0.00064,
+      "grad_norm": 78.20809173583984,
+      "learning_rate": 6.4e-08,
+      "loss": 2.1238,
+      "step": 20
+    },
+    {
+      "epoch": 0.00096,
+      "grad_norm": 256.7001953125,
+      "learning_rate": 9.6e-08,
+      "loss": 2.124,
       "step": 30
     },
     {
-      "epoch": 0.00384,
-      "grad_norm": 3.4691507816314697,
-      "learning_rate": 3.838771593090211e-07,
-      "loss": 1.9557,
+      "epoch": 0.00128,
+      "grad_norm": 80.30634307861328,
+      "learning_rate": 1.28e-07,
+      "loss": 2.1171,
+      "step": 40
+    },
+    {
+      "epoch": 0.0016,
+      "grad_norm": 130.5838165283203,
+      "learning_rate": 1.6e-07,
+      "loss": 2.1067,
+      "step": 50
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 78.49329376220703,
+      "learning_rate": 1.92e-07,
+      "loss": 2.1578,
       "step": 60
     },
     {
-      "epoch": 0.00576,
-      "grad_norm": 4.0405402183532715,
-      "learning_rate": 5.758157389635317e-07,
-      "loss": 1.9561,
+      "epoch": 0.00224,
+      "grad_norm": 98.15235137939453,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 2.1194,
+      "step": 70
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 87.7013931274414,
+      "learning_rate": 2.56e-07,
+      "loss": 2.1389,
+      "step": 80
+    },
+    {
+      "epoch": 0.00288,
+      "grad_norm": 74.4692153930664,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 2.1174,
       "step": 90
     },
     {
-      "epoch": 0.00768,
-      "grad_norm": 3.4379372596740723,
-      "learning_rate": 7.677543186180422e-07,
-      "loss": 1.9746,
+      "epoch": 0.0032,
+      "grad_norm": 73.72760009765625,
+      "learning_rate": 3.2e-07,
+      "loss": 2.1742,
+      "step": 100
+    },
+    {
+      "epoch": 0.00352,
+      "grad_norm": 73.00900268554688,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 2.1119,
+      "step": 110
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 78.37140655517578,
+      "learning_rate": 3.84e-07,
+      "loss": 2.151,
       "step": 120
     },
     {
-      "epoch": 0.0096,
-      "grad_norm": 3.476238489151001,
-      "learning_rate": 9.59692898272553e-07,
-      "loss": 1.9785,
+      "epoch": 0.00416,
+      "grad_norm": 99.71389770507812,
+      "learning_rate": 4.16e-07,
+      "loss": 2.1146,
+      "step": 130
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 363.2371826171875,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 2.1138,
+      "step": 140
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 86.691650390625,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 2.1319,
       "step": 150
     },
     {
-      "epoch": 0.01152,
-      "grad_norm": 3.3316054344177246,
-      "learning_rate": 1.1516314779270634e-06,
-      "loss": 1.97,
+      "epoch": 0.00512,
+      "grad_norm": 88.12674713134766,
+      "learning_rate": 5.12e-07,
+      "loss": 2.1143,
+      "step": 160
+    },
+    {
+      "epoch": 0.00544,
+      "grad_norm": 68.71428680419922,
+      "learning_rate": 5.44e-07,
+      "loss": 2.1265,
+      "step": 170
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 65.46075439453125,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 2.1211,
       "step": 180
     },
     {
-      "epoch": 0.01344,
-      "grad_norm": 3.4686760902404785,
-      "learning_rate": 1.3435700575815741e-06,
-      "loss": 1.9627,
+      "epoch": 0.00608,
+      "grad_norm": 78.31707000732422,
+      "learning_rate": 6.08e-07,
+      "loss": 2.1096,
+      "step": 190
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 156.94078063964844,
+      "learning_rate": 6.4e-07,
+      "loss": 2.1099,
+      "step": 200
+    },
+    {
+      "epoch": 0.00672,
+      "grad_norm": 82.27240753173828,
+      "learning_rate": 6.72e-07,
+      "loss": 2.0848,
       "step": 210
     },
     {
-      "epoch": 0.01536,
-      "grad_norm": 3.43794322013855,
-      "learning_rate": 1.5355086372360844e-06,
-      "loss": 1.9695,
+      "epoch": 0.00704,
+      "grad_norm": 82.82190704345703,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 2.108,
+      "step": 220
+    },
+    {
+      "epoch": 0.00736,
+      "grad_norm": 72.47757720947266,
+      "learning_rate": 7.36e-07,
+      "loss": 2.071,
+      "step": 230
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 79.20008087158203,
+      "learning_rate": 7.68e-07,
+      "loss": 2.0826,
       "step": 240
     },
     {
-      "epoch": 0.01728,
-      "grad_norm": 3.4568049907684326,
-      "learning_rate": 1.7274472168905951e-06,
-      "loss": 1.9606,
+      "epoch": 0.008,
+      "grad_norm": 62.04548645019531,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.0864,
+      "step": 250
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 71.6567153930664,
+      "learning_rate": 8.32e-07,
+      "loss": 2.133,
+      "step": 260
+    },
+    {
+      "epoch": 0.00864,
+      "grad_norm": 62.64482879638672,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 2.102,
       "step": 270
     },
     {
-      "epoch": 0.0192,
-      "grad_norm": 3.4242303371429443,
-      "learning_rate": 1.919385796545106e-06,
-      "loss": 1.9673,
+      "epoch": 0.00896,
+      "grad_norm": 75.67320251464844,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 2.0992,
+      "step": 280
+    },
+    {
+      "epoch": 0.00928,
+      "grad_norm": 63.39164352416992,
+      "learning_rate": 9.28e-07,
+      "loss": 2.0856,
+      "step": 290
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 99.13253021240234,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.0637,
       "step": 300
     },
     {
-      "epoch": 0.02112,
-      "grad_norm": 3.546679973602295,
-      "learning_rate": 2.1113243761996164e-06,
-      "loss": 1.9323,
+      "epoch": 0.00992,
+      "grad_norm": 274.91937255859375,
+      "learning_rate": 9.92e-07,
+      "loss": 2.0504,
+      "step": 310
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 125.71237182617188,
+      "learning_rate": 1.024e-06,
+      "loss": 2.079,
+      "step": 320
+    },
+    {
+      "epoch": 0.01056,
+      "grad_norm": 61.604496002197266,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 2.0687,
       "step": 330
     },
     {
-      "epoch": 0.02304,
-      "grad_norm": 3.381037473678589,
-      "learning_rate": 2.303262955854127e-06,
-      "loss": 1.9767,
+      "epoch": 0.01088,
+      "grad_norm": 88.08313751220703,
+      "learning_rate": 1.088e-06,
+      "loss": 2.1214,
+      "step": 340
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 64.370849609375,
+      "learning_rate": 1.12e-06,
+      "loss": 2.082,
+      "step": 350
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 60.212215423583984,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 2.0565,
       "step": 360
     },
     {
-      "epoch": 0.02496,
-      "grad_norm": 3.55702543258667,
-      "learning_rate": 2.4952015355086374e-06,
-      "loss": 1.9693,
+      "epoch": 0.01184,
+      "grad_norm": 130.44395446777344,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 2.0608,
+      "step": 370
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 78.3271255493164,
+      "learning_rate": 1.216e-06,
+      "loss": 2.0681,
+      "step": 380
+    },
+    {
+      "epoch": 0.01248,
+      "grad_norm": 64.16400146484375,
+      "learning_rate": 1.248e-06,
+      "loss": 2.0408,
       "step": 390
     },
     {
-      "epoch": 0.02688,
-      "grad_norm": 3.489302396774292,
-      "learning_rate": 2.6871401151631483e-06,
-      "loss": 1.9476,
+      "epoch": 0.0128,
+      "grad_norm": 65.31056213378906,
+      "learning_rate": 1.28e-06,
+      "loss": 2.0729,
+      "step": 400
+    },
+    {
+      "epoch": 0.01312,
+      "grad_norm": 65.5149917602539,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 2.0688,
+      "step": 410
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 58.871517181396484,
+      "learning_rate": 1.344e-06,
+      "loss": 2.0569,
       "step": 420
     },
     {
-      "epoch": 0.0288,
-      "grad_norm": 3.3413147926330566,
-      "learning_rate": 2.8790786948176584e-06,
-      "loss": 1.9696,
+      "epoch": 0.01376,
+      "grad_norm": 63.139610290527344,
+      "learning_rate": 1.376e-06,
+      "loss": 2.099,
+      "step": 430
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 59.644832611083984,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 2.0908,
+      "step": 440
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 67.80413055419922,
+      "learning_rate": 1.44e-06,
+      "loss": 2.0677,
       "step": 450
     },
     {
-      "epoch": 0.03072,
-      "grad_norm": 3.3850185871124268,
-      "learning_rate": 3.071017274472169e-06,
-      "loss": 1.9677,
+      "epoch": 0.01472,
+      "grad_norm": 83.1611328125,
+      "learning_rate": 1.472e-06,
+      "loss": 2.0639,
+      "step": 460
+    },
+    {
+      "epoch": 0.01504,
+      "grad_norm": 60.72829818725586,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 2.0843,
+      "step": 470
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 59.140357971191406,
+      "learning_rate": 1.536e-06,
+      "loss": 2.0815,
       "step": 480
     },
     {
-      "epoch": 0.03264,
-      "grad_norm": 3.4306411743164062,
-      "learning_rate": 3.26295585412668e-06,
-      "loss": 1.96,
+      "epoch": 0.01568,
+      "grad_norm": 62.182960510253906,
+      "learning_rate": 1.568e-06,
+      "loss": 2.0586,
+      "step": 490
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 57.94507598876953,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.0707,
+      "step": 500
+    },
+    {
+      "epoch": 0.01632,
+      "grad_norm": 58.95804977416992,
+      "learning_rate": 1.6320000000000002e-06,
+      "loss": 2.0829,
       "step": 510
     },
     {
-      "epoch": 0.03456,
-      "grad_norm": 3.456747531890869,
-      "learning_rate": 3.4548944337811903e-06,
-      "loss": 1.9523,
+      "epoch": 0.01664,
+      "grad_norm": 60.962276458740234,
+      "learning_rate": 1.664e-06,
+      "loss": 2.1051,
+      "step": 520
+    },
+    {
+      "epoch": 0.01696,
+      "grad_norm": 65.06385803222656,
+      "learning_rate": 1.6960000000000002e-06,
+      "loss": 2.0823,
+      "step": 530
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 97.38597869873047,
+      "learning_rate": 1.7280000000000002e-06,
+      "loss": 2.0762,
       "step": 540
     },
     {
-      "epoch": 0.03648,
-      "grad_norm": 3.3029491901397705,
-      "learning_rate": 3.6468330134357012e-06,
-      "loss": 1.9673,
+      "epoch": 0.0176,
+      "grad_norm": 57.09391403198242,
+      "learning_rate": 1.76e-06,
+      "loss": 2.0419,
+      "step": 550
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 56.89313507080078,
+      "learning_rate": 1.7920000000000002e-06,
+      "loss": 2.0712,
+      "step": 560
+    },
+    {
+      "epoch": 0.01824,
+      "grad_norm": 59.043785095214844,
+      "learning_rate": 1.8240000000000002e-06,
+      "loss": 2.0587,
       "step": 570
     },
     {
-      "epoch": 0.0384,
-      "grad_norm": 3.2382235527038574,
-      "learning_rate": 3.838771593090212e-06,
-      "loss": 1.9626,
+      "epoch": 0.01856,
+      "grad_norm": 62.550270080566406,
+      "learning_rate": 1.856e-06,
+      "loss": 2.054,
+      "step": 580
+    },
+    {
+      "epoch": 0.01888,
+      "grad_norm": 55.80398178100586,
+      "learning_rate": 1.8880000000000002e-06,
+      "loss": 2.0576,
+      "step": 590
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 96.00469207763672,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.0298,
       "step": 600
     },
     {
-      "epoch": 0.04032,
-      "grad_norm": 3.461942195892334,
-      "learning_rate": 4.030710172744722e-06,
-      "loss": 1.9554,
+      "epoch": 0.01952,
+      "grad_norm": 57.13787841796875,
+      "learning_rate": 1.952e-06,
+      "loss": 2.0663,
+      "step": 610
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 77.44837951660156,
+      "learning_rate": 1.984e-06,
+      "loss": 2.0384,
+      "step": 620
+    },
+    {
+      "epoch": 0.02016,
+      "grad_norm": 55.39677429199219,
+      "learning_rate": 2.0160000000000003e-06,
+      "loss": 2.0735,
       "step": 630
     },
     {
-      "epoch": 0.04224,
-      "grad_norm": 3.2762303352355957,
-      "learning_rate": 4.222648752399233e-06,
-      "loss": 1.9723,
+      "epoch": 0.02048,
+      "grad_norm": 56.57113265991211,
+      "learning_rate": 2.048e-06,
+      "loss": 2.0319,
+      "step": 640
+    },
+    {
+      "epoch": 0.0208,
+      "grad_norm": 57.18354034423828,
+      "learning_rate": 2.08e-06,
+      "loss": 2.0575,
+      "step": 650
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 58.37870788574219,
+      "learning_rate": 2.1120000000000003e-06,
+      "loss": 2.069,
       "step": 660
     },
     {
-      "epoch": 0.04416,
-      "grad_norm": 3.558966636657715,
-      "learning_rate": 4.414587332053743e-06,
-      "loss": 1.9635,
+      "epoch": 0.02144,
+      "grad_norm": 65.55963897705078,
+      "learning_rate": 2.144e-06,
+      "loss": 2.0331,
+      "step": 670
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 57.383968353271484,
+      "learning_rate": 2.176e-06,
+      "loss": 2.0384,
+      "step": 680
+    },
+    {
+      "epoch": 0.02208,
+      "grad_norm": 57.483943939208984,
+      "learning_rate": 2.2080000000000003e-06,
+      "loss": 2.0409,
       "step": 690
     },
     {
-      "epoch": 0.04608,
-      "grad_norm": 3.35530948638916,
-      "learning_rate": 4.606525911708254e-06,
-      "loss": 1.9581,
+      "epoch": 0.0224,
+      "grad_norm": 59.38597869873047,
+      "learning_rate": 2.24e-06,
+      "loss": 2.0512,
+      "step": 700
+    },
+    {
+      "epoch": 0.02272,
+      "grad_norm": 57.73863220214844,
+      "learning_rate": 2.2720000000000004e-06,
+      "loss": 2.0403,
+      "step": 710
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 63.45584487915039,
+      "learning_rate": 2.3040000000000003e-06,
+      "loss": 2.0434,
       "step": 720
     },
     {
-      "epoch": 0.048,
-      "grad_norm": 3.4470019340515137,
-      "learning_rate": 4.798464491362765e-06,
-      "loss": 1.9679,
+      "epoch": 0.02336,
+      "grad_norm": 100.67032623291016,
+      "learning_rate": 2.336e-06,
+      "loss": 2.0526,
+      "step": 730
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 57.207584381103516,
+      "learning_rate": 2.3680000000000005e-06,
+      "loss": 2.0304,
+      "step": 740
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 56.702999114990234,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.0786,
       "step": 750
     },
     {
-      "epoch": 0.04992,
-      "grad_norm": 3.437478542327881,
-      "learning_rate": 4.990403071017275e-06,
-      "loss": 1.9739,
+      "epoch": 0.02432,
+      "grad_norm": 63.29140090942383,
+      "learning_rate": 2.432e-06,
+      "loss": 2.0715,
+      "step": 760
+    },
+    {
+      "epoch": 0.02464,
+      "grad_norm": 59.268550872802734,
+      "learning_rate": 2.4640000000000005e-06,
+      "loss": 2.0632,
+      "step": 770
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 55.43419647216797,
+      "learning_rate": 2.496e-06,
+      "loss": 2.0311,
       "step": 780
     },
     {
-      "epoch": 0.05184,
-      "grad_norm": 3.380098342895508,
-      "learning_rate": 5.182341650671785e-06,
-      "loss": 1.9632,
+      "epoch": 0.02528,
+      "grad_norm": 55.27999496459961,
+      "learning_rate": 2.5280000000000006e-06,
+      "loss": 2.046,
+      "step": 790
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 57.40989685058594,
+      "learning_rate": 2.56e-06,
+      "loss": 2.0419,
+      "step": 800
+    },
+    {
+      "epoch": 0.02592,
+      "grad_norm": 67.16029357910156,
+      "learning_rate": 2.592e-06,
+      "loss": 2.03,
       "step": 810
     },
     {
-      "epoch": 0.05376,
-      "grad_norm": 3.612208366394043,
-      "learning_rate": 5.374280230326297e-06,
-      "loss": 1.9585,
+      "epoch": 0.02624,
+      "grad_norm": 57.05144500732422,
+      "learning_rate": 2.6240000000000006e-06,
+      "loss": 2.0313,
+      "step": 820
+    },
+    {
+      "epoch": 0.02656,
+      "grad_norm": 65.95780944824219,
+      "learning_rate": 2.656e-06,
+      "loss": 2.0233,
+      "step": 830
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 57.77890396118164,
+      "learning_rate": 2.688e-06,
+      "loss": 1.993,
       "step": 840
     },
     {
-      "epoch": 0.05568,
-      "grad_norm": 3.526761054992676,
-      "learning_rate": 5.566218809980807e-06,
-      "loss": 1.9775,
+      "epoch": 0.0272,
+      "grad_norm": 55.54998779296875,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.0289,
+      "step": 850
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 55.96319580078125,
+      "learning_rate": 2.752e-06,
+      "loss": 2.0452,
+      "step": 860
+    },
+    {
+      "epoch": 0.02784,
+      "grad_norm": 55.67866897583008,
+      "learning_rate": 2.784e-06,
+      "loss": 2.0498,
       "step": 870
     },
     {
-      "epoch": 0.0576,
-      "grad_norm": 3.5100183486938477,
-      "learning_rate": 5.758157389635317e-06,
-      "loss": 1.9545,
-      "step": 900
+      "epoch": 0.02816,
+      "grad_norm": 55.35938262939453,
+      "learning_rate": 2.8160000000000002e-06,
+      "loss": 2.013,
+      "step": 880
     },
     {
-      "epoch": 0.05952,
-      "grad_norm": 3.3006579875946045,
-      "learning_rate": 5.950095969289828e-06,
-      "loss": 1.9681,
-      "step": 930
+      "epoch": 0.02848,
+      "grad_norm": 55.288124084472656,
+      "learning_rate": 2.848e-06,
+      "loss": 2.0345,
+      "step": 890
     },
     {
-      "epoch": 0.06144,
-      "grad_norm": 3.382730007171631,
-      "learning_rate": 6.142034548944338e-06,
-      "loss": 1.9541,
-      "step": 960
+      "epoch": 0.0288,
+      "grad_norm": 56.8425178527832,
+      "learning_rate": 2.88e-06,
+      "loss": 2.0455,
+      "step": 900
     },
     {
-      "epoch": 0.06336,
-      "grad_norm": 3.4000792503356934,
-      "learning_rate": 6.3339731285988495e-06,
-      "loss": 1.9672,
-      "step": 990
+      "epoch": 0.02912,
+      "grad_norm": 58.22956848144531,
+      "learning_rate": 2.9120000000000002e-06,
+      "loss": 2.038,
+      "step": 910
     },
     {
-      "epoch": 0.06528,
-      "grad_norm": 3.6164541244506836,
-      "learning_rate": 6.52591170825336e-06,
-      "loss": 1.9669,
-      "step": 1020
+      "epoch": 0.02944,
+      "grad_norm": 54.9834098815918,
+      "learning_rate": 2.944e-06,
+      "loss": 2.0466,
+      "step": 920
     },
     {
-      "epoch": 0.0672,
-      "grad_norm": 3.474783420562744,
-      "learning_rate": 6.71785028790787e-06,
-      "loss": 1.9669,
-      "step": 1050
+      "epoch": 0.02976,
+      "grad_norm": 66.05599212646484,
+      "learning_rate": 2.976e-06,
+      "loss": 2.0497,
+      "step": 930
     },
     {
-      "epoch": 0.06912,
-      "grad_norm": 3.2741751670837402,
-      "learning_rate": 6.909788867562381e-06,
-      "loss": 1.9703,
-      "step": 1080
+      "epoch": 0.03008,
+      "grad_norm": 57.55447006225586,
+      "learning_rate": 3.0080000000000003e-06,
+      "loss": 2.0273,
+      "step": 940
     },
     {
-      "epoch": 0.07104,
-      "grad_norm": 3.6716578006744385,
-      "learning_rate": 7.101727447216891e-06,
-      "loss": 1.9562,
-      "step": 1110
+      "epoch": 0.0304,
+      "grad_norm": 54.68487548828125,
+      "learning_rate": 3.04e-06,
+      "loss": 2.0367,
+      "step": 950
     },
     {
-      "epoch": 0.07296,
-      "grad_norm": 3.3192343711853027,
-      "learning_rate": 7.2936660268714024e-06,
-      "loss": 1.9622,
-      "step": 1140
+      "epoch": 0.03072,
+      "grad_norm": 55.63594436645508,
+      "learning_rate": 3.072e-06,
+      "loss": 2.0618,
+      "step": 960
     },
     {
-      "epoch": 0.07488,
-      "grad_norm": 3.3274195194244385,
-      "learning_rate": 7.4856046065259125e-06,
-      "loss": 1.9785,
-      "step": 1170
+      "epoch": 0.03104,
+      "grad_norm": 54.52610778808594,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 2.0254,
+      "step": 970
     },
     {
-      "epoch": 0.0768,
-      "grad_norm": 3.500359058380127,
-      "learning_rate": 7.677543186180423e-06,
-      "loss": 1.9681,
-      "step": 1200
+      "epoch": 0.03136,
+      "grad_norm": 55.98470687866211,
+      "learning_rate": 3.136e-06,
+      "loss": 1.9674,
+      "step": 980
     },
     {
-      "epoch": 0.07872,
-      "grad_norm": 3.3644425868988037,
-      "learning_rate": 7.869481765834934e-06,
-      "loss": 1.9502,
-      "step": 1230
+      "epoch": 0.03168,
+      "grad_norm": 53.209693908691406,
+      "learning_rate": 3.1680000000000004e-06,
+      "loss": 2.0201,
+      "step": 990
     },
     {
-      "epoch": 0.08064,
-      "grad_norm": 3.5763943195343018,
-      "learning_rate": 8.061420345489444e-06,
-      "loss": 1.9943,
-      "step": 1260
+      "epoch": 0.032,
+      "grad_norm": 57.08290100097656,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.0513,
+      "step": 1000
     },
     {
-      "epoch": 0.08256,
-      "grad_norm": 3.329719305038452,
-      "learning_rate": 8.253358925143955e-06,
-      "loss": 1.962,
-      "step": 1290
+      "epoch": 0.03232,
+      "grad_norm": 53.92963409423828,
+      "learning_rate": 3.232e-06,
+      "loss": 2.0478,
+      "step": 1010
     },
     {
-      "epoch": 0.08448,
-      "grad_norm": 3.4307610988616943,
-      "learning_rate": 8.445297504798465e-06,
-      "loss": 1.984,
-      "step": 1320
+      "epoch": 0.03264,
+      "grad_norm": 54.785301208496094,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 2.0312,
+      "step": 1020
     },
     {
-      "epoch": 0.0864,
-      "grad_norm": 3.4413914680480957,
-      "learning_rate": 8.637236084452976e-06,
-      "loss": 1.9733,
-      "step": 1350
+      "epoch": 0.03296,
+      "grad_norm": 56.84122848510742,
+      "learning_rate": 3.2960000000000003e-06,
+      "loss": 2.036,
+      "step": 1030
     },
     {
-      "epoch": 0.08832,
-      "grad_norm": 3.477210521697998,
-      "learning_rate": 8.829174664107486e-06,
-      "loss": 1.9533,
-      "step": 1380
+      "epoch": 0.03328,
+      "grad_norm": 58.097774505615234,
+      "learning_rate": 3.328e-06,
+      "loss": 2.0265,
+      "step": 1040
     },
     {
-      "epoch": 0.09024,
-      "grad_norm": 3.527116060256958,
-      "learning_rate": 9.021113243761996e-06,
-      "loss": 1.9618,
-      "step": 1410
+      "epoch": 0.0336,
+      "grad_norm": 59.78932571411133,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.0173,
+      "step": 1050
     },
     {
-      "epoch": 0.09216,
-      "grad_norm": 3.4810211658477783,
-      "learning_rate": 9.213051823416507e-06,
-      "loss": 1.9618,
-      "step": 1440
+      "epoch": 0.03392,
+      "grad_norm": 53.78879928588867,
+      "learning_rate": 3.3920000000000003e-06,
+      "loss": 2.0493,
+      "step": 1060
     },
     {
-      "epoch": 0.09408,
-      "grad_norm": 3.9999337196350098,
-      "learning_rate": 9.404990403071018e-06,
-      "loss": 1.9655,
-      "step": 1470
+      "epoch": 0.03424,
+      "grad_norm": 58.1050910949707,
+      "learning_rate": 3.424e-06,
+      "loss": 2.0251,
+      "step": 1070
     },
     {
-      "epoch": 0.096,
-      "grad_norm": 3.464796304702759,
-      "learning_rate": 9.59692898272553e-06,
-      "loss": 1.9742,
-      "step": 1500
+      "epoch": 0.03456,
+      "grad_norm": 54.9379768371582,
+      "learning_rate": 3.4560000000000005e-06,
+      "loss": 2.029,
+      "step": 1080
     },
     {
-      "epoch": 0.09792,
-      "grad_norm": 3.4603271484375,
-      "learning_rate": 9.78886756238004e-06,
-      "loss": 1.958,
-      "step": 1530
+      "epoch": 0.03488,
+      "grad_norm": 57.901493072509766,
+      "learning_rate": 3.4880000000000003e-06,
+      "loss": 1.9813,
+      "step": 1090
     },
     {
-      "epoch": 0.09984,
-      "grad_norm": 3.5983641147613525,
-      "learning_rate": 9.98080614203455e-06,
-      "loss": 1.9676,
-      "step": 1560
+      "epoch": 0.0352,
+      "grad_norm": 55.3337516784668,
+      "learning_rate": 3.52e-06,
+      "loss": 2.0027,
+      "step": 1100
     },
     {
-      "epoch": 0.10176,
-      "grad_norm": 3.5228517055511475,
-      "learning_rate": 9.98079931730906e-06,
-      "loss": 1.9645,
-      "step": 1590
+      "epoch": 0.03552,
+      "grad_norm": 57.643306732177734,
+      "learning_rate": 3.5520000000000005e-06,
+      "loss": 2.0194,
+      "step": 1110
     },
     {
-      "epoch": 0.10368,
-      "grad_norm": 3.437927007675171,
-      "learning_rate": 9.959465225430239e-06,
-      "loss": 1.9837,
-      "step": 1620
+      "epoch": 0.03584,
+      "grad_norm": 56.8505744934082,
+      "learning_rate": 3.5840000000000003e-06,
+      "loss": 2.014,
+      "step": 1120
     },
     {
-      "epoch": 0.1056,
-      "grad_norm": 3.389080286026001,
-      "learning_rate": 9.938131133551416e-06,
-      "loss": 1.9816,
-      "step": 1650
+      "epoch": 0.03616,
+      "grad_norm": 54.06476593017578,
+      "learning_rate": 3.616e-06,
+      "loss": 2.0378,
+      "step": 1130
     },
     {
-      "epoch": 0.10752,
-      "grad_norm": 3.3758959770202637,
-      "learning_rate": 9.916797041672593e-06,
-      "loss": 1.9681,
-      "step": 1680
+      "epoch": 0.03648,
+      "grad_norm": 57.77007293701172,
+      "learning_rate": 3.6480000000000005e-06,
+      "loss": 2.0199,
+      "step": 1140
     },
     {
-      "epoch": 0.10944,
-      "grad_norm": 3.3564791679382324,
-      "learning_rate": 9.895462949793771e-06,
-      "loss": 1.954,
-      "step": 1710
+      "epoch": 0.0368,
+      "grad_norm": 57.76017379760742,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.0117,
+      "step": 1150
     },
     {
-      "epoch": 0.11136,
-      "grad_norm": 3.358147144317627,
-      "learning_rate": 9.874128857914948e-06,
-      "loss": 1.948,
-      "step": 1740
+      "epoch": 0.03712,
+      "grad_norm": 55.163944244384766,
+      "learning_rate": 3.712e-06,
+      "loss": 2.0031,
+      "step": 1160
     },
     {
-      "epoch": 0.11328,
-      "grad_norm": 3.390315532684326,
-      "learning_rate": 9.852794766036127e-06,
-      "loss": 1.9639,
-      "step": 1770
+      "epoch": 0.03744,
+      "grad_norm": 55.953636169433594,
+      "learning_rate": 3.7440000000000005e-06,
+      "loss": 1.9844,
+      "step": 1170
     },
     {
-      "epoch": 0.1152,
-      "grad_norm": 3.3553218841552734,
-      "learning_rate": 9.831460674157303e-06,
-      "loss": 1.9574,
-      "step": 1800
+      "epoch": 0.03776,
+      "grad_norm": 54.14808654785156,
+      "learning_rate": 3.7760000000000004e-06,
+      "loss": 2.0015,
+      "step": 1180
     },
     {
-      "epoch": 0.11712,
-      "grad_norm": 3.4452435970306396,
-      "learning_rate": 9.810126582278482e-06,
-      "loss": 1.9398,
-      "step": 1830
+      "epoch": 0.03808,
+      "grad_norm": 55.77479553222656,
+      "learning_rate": 3.8080000000000006e-06,
+      "loss": 2.0269,
+      "step": 1190
     },
     {
-      "epoch": 0.11904,
-      "grad_norm": 3.4306797981262207,
-      "learning_rate": 9.788792490399659e-06,
-      "loss": 1.9558,
-      "step": 1860
+      "epoch": 0.0384,
+      "grad_norm": 58.67176055908203,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.0185,
+      "step": 1200
     },
     {
-      "epoch": 0.12096,
-      "grad_norm": 3.2931087017059326,
-      "learning_rate": 9.767458398520837e-06,
-      "loss": 1.9686,
-      "step": 1890
+      "epoch": 0.03872,
+      "grad_norm": 54.49531555175781,
+      "learning_rate": 3.872e-06,
+      "loss": 2.0179,
+      "step": 1210
     },
     {
-      "epoch": 0.12288,
-      "grad_norm": 3.547395944595337,
-      "learning_rate": 9.746124306642016e-06,
-      "loss": 1.9591,
-      "step": 1920
+      "epoch": 0.03904,
+      "grad_norm": 56.10469436645508,
+      "learning_rate": 3.904e-06,
+      "loss": 1.9929,
+      "step": 1220
     },
     {
-      "epoch": 0.1248,
-      "grad_norm": 3.3633499145507812,
-      "learning_rate": 9.724790214763193e-06,
-      "loss": 1.9754,
+      "epoch": 0.03936,
+      "grad_norm": 53.95834732055664,
+      "learning_rate": 3.936e-06,
+      "loss": 2.03,
+      "step": 1230
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 102.673095703125,
+      "learning_rate": 3.968e-06,
+      "loss": 2.0318,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 54.45889663696289,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.9909,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 55.20240020751953,
+      "learning_rate": 4.0320000000000005e-06,
+      "loss": 2.0459,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04064,
+      "grad_norm": 56.24109649658203,
+      "learning_rate": 4.064e-06,
+      "loss": 2.0029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 56.00358963012695,
+      "learning_rate": 4.096e-06,
+      "loss": 1.9921,
+      "step": 1280
+    },
+    {
+      "epoch": 0.04128,
+      "grad_norm": 59.99430847167969,
+      "learning_rate": 4.128e-06,
+      "loss": 1.975,
+      "step": 1290
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 56.52634811401367,
+      "learning_rate": 4.16e-06,
+      "loss": 1.9956,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04192,
+      "grad_norm": 53.13933181762695,
+      "learning_rate": 4.192000000000001e-06,
+      "loss": 2.0033,
+      "step": 1310
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 54.04265594482422,
+      "learning_rate": 4.2240000000000006e-06,
+      "loss": 2.0051,
+      "step": 1320
+    },
+    {
+      "epoch": 0.04256,
+      "grad_norm": 52.66888427734375,
+      "learning_rate": 4.256e-06,
+      "loss": 1.9798,
+      "step": 1330
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 53.12565994262695,
+      "learning_rate": 4.288e-06,
+      "loss": 1.9773,
+      "step": 1340
+    },
+    {
+      "epoch": 0.0432,
+      "grad_norm": 55.532291412353516,
+      "learning_rate": 4.32e-06,
+      "loss": 1.996,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 55.148712158203125,
+      "learning_rate": 4.352e-06,
+      "loss": 2.0011,
+      "step": 1360
+    },
+    {
+      "epoch": 0.04384,
+      "grad_norm": 55.9796257019043,
+      "learning_rate": 4.384000000000001e-06,
+      "loss": 1.9802,
+      "step": 1370
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 55.14250564575195,
+      "learning_rate": 4.416000000000001e-06,
+      "loss": 2.0021,
+      "step": 1380
+    },
+    {
+      "epoch": 0.04448,
+      "grad_norm": 57.55263137817383,
+      "learning_rate": 4.4480000000000004e-06,
+      "loss": 1.9682,
+      "step": 1390
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 56.29933547973633,
+      "learning_rate": 4.48e-06,
+      "loss": 1.9691,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04512,
+      "grad_norm": 56.33018493652344,
+      "learning_rate": 4.512e-06,
+      "loss": 1.9934,
+      "step": 1410
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 56.64614486694336,
+      "learning_rate": 4.544000000000001e-06,
+      "loss": 2.0083,
+      "step": 1420
+    },
+    {
+      "epoch": 0.04576,
+      "grad_norm": 57.90870666503906,
+      "learning_rate": 4.576000000000001e-06,
+      "loss": 2.0079,
+      "step": 1430
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 53.97960662841797,
+      "learning_rate": 4.608000000000001e-06,
+      "loss": 2.019,
+      "step": 1440
+    },
+    {
+      "epoch": 0.0464,
+      "grad_norm": 54.856109619140625,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.0008,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 54.7287483215332,
+      "learning_rate": 4.672e-06,
+      "loss": 1.9936,
+      "step": 1460
+    },
+    {
+      "epoch": 0.04704,
+      "grad_norm": 54.871925354003906,
+      "learning_rate": 4.704e-06,
+      "loss": 2.0316,
+      "step": 1470
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 57.25019073486328,
+      "learning_rate": 4.736000000000001e-06,
+      "loss": 1.9713,
+      "step": 1480
+    },
+    {
+      "epoch": 0.04768,
+      "grad_norm": 53.96601867675781,
+      "learning_rate": 4.768000000000001e-06,
+      "loss": 1.9888,
+      "step": 1490
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 54.54352569580078,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.9706,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04832,
+      "grad_norm": 56.38828659057617,
+      "learning_rate": 4.8320000000000005e-06,
+      "loss": 1.9915,
+      "step": 1510
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 56.09176254272461,
+      "learning_rate": 4.864e-06,
+      "loss": 1.9934,
+      "step": 1520
+    },
+    {
+      "epoch": 0.04896,
+      "grad_norm": 53.20353317260742,
+      "learning_rate": 4.896e-06,
+      "loss": 1.9913,
+      "step": 1530
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 55.36820983886719,
+      "learning_rate": 4.928000000000001e-06,
+      "loss": 1.9903,
+      "step": 1540
+    },
+    {
+      "epoch": 0.0496,
+      "grad_norm": 53.72861862182617,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 1.992,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 56.07704162597656,
+      "learning_rate": 4.992e-06,
+      "loss": 1.9861,
+      "step": 1560
+    },
+    {
+      "epoch": 0.05024,
+      "grad_norm": 54.132179260253906,
+      "learning_rate": 5.024e-06,
+      "loss": 2.0059,
+      "step": 1570
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 53.731773376464844,
+      "learning_rate": 5.056000000000001e-06,
+      "loss": 1.9821,
+      "step": 1580
+    },
+    {
+      "epoch": 0.05088,
+      "grad_norm": 54.18653869628906,
+      "learning_rate": 5.088000000000001e-06,
+      "loss": 1.981,
+      "step": 1590
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 55.18617248535156,
+      "learning_rate": 5.12e-06,
+      "loss": 2.0078,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05152,
+      "grad_norm": 54.86219787597656,
+      "learning_rate": 5.152e-06,
+      "loss": 1.9672,
+      "step": 1610
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 55.19453048706055,
+      "learning_rate": 5.184e-06,
+      "loss": 1.9968,
+      "step": 1620
+    },
+    {
+      "epoch": 0.05216,
+      "grad_norm": 53.446937561035156,
+      "learning_rate": 5.216e-06,
+      "loss": 1.9704,
+      "step": 1630
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 54.388919830322266,
+      "learning_rate": 5.248000000000001e-06,
+      "loss": 1.9644,
+      "step": 1640
+    },
+    {
+      "epoch": 0.0528,
+      "grad_norm": 53.366363525390625,
+      "learning_rate": 5.28e-06,
+      "loss": 2.016,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 54.797367095947266,
+      "learning_rate": 5.312e-06,
+      "loss": 1.9977,
+      "step": 1660
+    },
+    {
+      "epoch": 0.05344,
+      "grad_norm": 55.27174377441406,
+      "learning_rate": 5.344e-06,
+      "loss": 1.9998,
+      "step": 1670
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 53.69790267944336,
+      "learning_rate": 5.376e-06,
+      "loss": 2.0032,
+      "step": 1680
+    },
+    {
+      "epoch": 0.05408,
+      "grad_norm": 54.89072799682617,
+      "learning_rate": 5.408e-06,
+      "loss": 1.9904,
+      "step": 1690
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 55.35121536254883,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 1.9592,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05472,
+      "grad_norm": 52.87653350830078,
+      "learning_rate": 5.472e-06,
+      "loss": 1.9727,
+      "step": 1710
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 55.05913543701172,
+      "learning_rate": 5.504e-06,
+      "loss": 1.9658,
+      "step": 1720
+    },
+    {
+      "epoch": 0.05536,
+      "grad_norm": 57.34429168701172,
+      "learning_rate": 5.536e-06,
+      "loss": 2.0091,
+      "step": 1730
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 55.97663497924805,
+      "learning_rate": 5.568e-06,
+      "loss": 1.9778,
+      "step": 1740
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 55.70100021362305,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.9616,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 58.278560638427734,
+      "learning_rate": 5.6320000000000005e-06,
+      "loss": 1.9674,
+      "step": 1760
+    },
+    {
+      "epoch": 0.05664,
+      "grad_norm": 57.26869583129883,
+      "learning_rate": 5.664e-06,
+      "loss": 1.9691,
+      "step": 1770
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 54.13764572143555,
+      "learning_rate": 5.696e-06,
+      "loss": 1.9874,
+      "step": 1780
+    },
+    {
+      "epoch": 0.05728,
+      "grad_norm": 51.36130142211914,
+      "learning_rate": 5.728e-06,
+      "loss": 1.9858,
+      "step": 1790
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 55.706329345703125,
+      "learning_rate": 5.76e-06,
+      "loss": 1.9472,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05792,
+      "grad_norm": 53.5456428527832,
+      "learning_rate": 5.792000000000001e-06,
+      "loss": 1.9674,
+      "step": 1810
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 54.37445068359375,
+      "learning_rate": 5.8240000000000005e-06,
+      "loss": 1.9978,
+      "step": 1820
+    },
+    {
+      "epoch": 0.05856,
+      "grad_norm": 54.93683624267578,
+      "learning_rate": 5.856e-06,
+      "loss": 1.9733,
+      "step": 1830
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 54.61006164550781,
+      "learning_rate": 5.888e-06,
+      "loss": 1.9629,
+      "step": 1840
+    },
+    {
+      "epoch": 0.0592,
+      "grad_norm": 55.023983001708984,
+      "learning_rate": 5.92e-06,
+      "loss": 2.0033,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 53.465049743652344,
+      "learning_rate": 5.952e-06,
+      "loss": 1.989,
+      "step": 1860
+    },
+    {
+      "epoch": 0.05984,
+      "grad_norm": 56.193817138671875,
+      "learning_rate": 5.984000000000001e-06,
+      "loss": 1.9739,
+      "step": 1870
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 54.541046142578125,
+      "learning_rate": 6.0160000000000005e-06,
+      "loss": 1.9517,
+      "step": 1880
+    },
+    {
+      "epoch": 0.06048,
+      "grad_norm": 56.678497314453125,
+      "learning_rate": 6.048e-06,
+      "loss": 1.9855,
+      "step": 1890
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 55.969120025634766,
+      "learning_rate": 6.08e-06,
+      "loss": 2.007,
+      "step": 1900
+    },
+    {
+      "epoch": 0.06112,
+      "grad_norm": 57.037635803222656,
+      "learning_rate": 6.112e-06,
+      "loss": 1.97,
+      "step": 1910
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 54.40221405029297,
+      "learning_rate": 6.144e-06,
+      "loss": 1.9743,
+      "step": 1920
+    },
+    {
+      "epoch": 0.06176,
+      "grad_norm": 53.17253494262695,
+      "learning_rate": 6.176000000000001e-06,
+      "loss": 1.9731,
+      "step": 1930
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 54.895198822021484,
+      "learning_rate": 6.2080000000000005e-06,
+      "loss": 1.9538,
+      "step": 1940
+    },
+    {
+      "epoch": 0.0624,
+      "grad_norm": 53.09823226928711,
+      "learning_rate": 6.24e-06,
+      "loss": 1.9552,
       "step": 1950
     },
     {
-      "epoch": 0.12672,
-      "grad_norm": 3.5388097763061523,
-      "learning_rate": 9.703456122884371e-06,
-      "loss": 1.9726,
-      "step": 1980
+      "epoch": 0.06272,
+      "grad_norm": 56.002220153808594,
+      "learning_rate": 6.272e-06,
+      "loss": 1.9599,
+      "step": 1960
+    },
+    {
+      "epoch": 0.06304,
+      "grad_norm": 55.639400482177734,
+      "learning_rate": 6.304e-06,
+      "loss": 1.9609,
+      "step": 1970
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 52.89950942993164,
+      "learning_rate": 6.336000000000001e-06,
+      "loss": 1.9171,
+      "step": 1980
+    },
+    {
+      "epoch": 0.06368,
+      "grad_norm": 53.86906814575195,
+      "learning_rate": 6.368000000000001e-06,
+      "loss": 1.97,
+      "step": 1990
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 55.6588134765625,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.9761,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06432,
+      "grad_norm": 53.31937026977539,
+      "learning_rate": 6.432e-06,
+      "loss": 1.9875,
+      "step": 2010
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 54.0553092956543,
+      "learning_rate": 6.464e-06,
+      "loss": 1.9506,
+      "step": 2020
+    },
+    {
+      "epoch": 0.06496,
+      "grad_norm": 55.71330261230469,
+      "learning_rate": 6.496e-06,
+      "loss": 1.9804,
+      "step": 2030
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 55.74639129638672,
+      "learning_rate": 6.528000000000001e-06,
+      "loss": 1.9477,
+      "step": 2040
+    },
+    {
+      "epoch": 0.0656,
+      "grad_norm": 52.34907913208008,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 1.9649,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 56.01731872558594,
+      "learning_rate": 6.592000000000001e-06,
+      "loss": 1.9424,
+      "step": 2060
+    },
+    {
+      "epoch": 0.06624,
+      "grad_norm": 54.96754455566406,
+      "learning_rate": 6.6240000000000004e-06,
+      "loss": 1.9472,
+      "step": 2070
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 54.93800735473633,
+      "learning_rate": 6.656e-06,
+      "loss": 1.9315,
+      "step": 2080
+    },
+    {
+      "epoch": 0.06688,
+      "grad_norm": 53.28120040893555,
+      "learning_rate": 6.688e-06,
+      "loss": 1.9172,
+      "step": 2090
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 57.81128692626953,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 1.9547,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06752,
+      "grad_norm": 55.10950469970703,
+      "learning_rate": 6.752000000000001e-06,
+      "loss": 1.9546,
+      "step": 2110
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 57.18091583251953,
+      "learning_rate": 6.784000000000001e-06,
+      "loss": 1.9474,
+      "step": 2120
+    },
+    {
+      "epoch": 0.06816,
+      "grad_norm": 55.841949462890625,
+      "learning_rate": 6.8160000000000005e-06,
+      "loss": 1.9447,
+      "step": 2130
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 56.35560989379883,
+      "learning_rate": 6.848e-06,
+      "loss": 1.9712,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0688,
+      "grad_norm": 53.2947883605957,
+      "learning_rate": 6.88e-06,
+      "loss": 1.9627,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 54.127017974853516,
+      "learning_rate": 6.912000000000001e-06,
+      "loss": 1.9434,
+      "step": 2160
+    },
+    {
+      "epoch": 0.06944,
+      "grad_norm": 55.26138687133789,
+      "learning_rate": 6.944000000000001e-06,
+      "loss": 1.9383,
+      "step": 2170
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 53.60810852050781,
+      "learning_rate": 6.976000000000001e-06,
+      "loss": 1.9546,
+      "step": 2180
+    },
+    {
+      "epoch": 0.07008,
+      "grad_norm": 59.719818115234375,
+      "learning_rate": 7.0080000000000005e-06,
+      "loss": 1.942,
+      "step": 2190
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 57.145912170410156,
+      "learning_rate": 7.04e-06,
+      "loss": 1.9615,
+      "step": 2200
+    },
+    {
+      "epoch": 0.07072,
+      "grad_norm": 54.45525360107422,
+      "learning_rate": 7.072000000000001e-06,
+      "loss": 1.985,
+      "step": 2210
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 54.832584381103516,
+      "learning_rate": 7.104000000000001e-06,
+      "loss": 1.962,
+      "step": 2220
+    },
+    {
+      "epoch": 0.07136,
+      "grad_norm": 51.92244338989258,
+      "learning_rate": 7.136000000000001e-06,
+      "loss": 1.9542,
+      "step": 2230
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 56.1202392578125,
+      "learning_rate": 7.168000000000001e-06,
+      "loss": 1.9922,
+      "step": 2240
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 54.495853424072266,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.9604,
+      "step": 2250
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 54.13912582397461,
+      "learning_rate": 7.232e-06,
+      "loss": 1.9325,
+      "step": 2260
+    },
+    {
+      "epoch": 0.07264,
+      "grad_norm": 54.35029220581055,
+      "learning_rate": 7.264000000000001e-06,
+      "loss": 1.9923,
+      "step": 2270
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 54.93330764770508,
+      "learning_rate": 7.296000000000001e-06,
+      "loss": 1.9505,
+      "step": 2280
+    },
+    {
+      "epoch": 0.07328,
+      "grad_norm": 57.12767028808594,
+      "learning_rate": 7.328000000000001e-06,
+      "loss": 1.914,
+      "step": 2290
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 55.84728240966797,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 1.9528,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07392,
+      "grad_norm": 55.132205963134766,
+      "learning_rate": 7.3920000000000005e-06,
+      "loss": 1.9449,
+      "step": 2310
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 56.098594665527344,
+      "learning_rate": 7.424e-06,
+      "loss": 1.9802,
+      "step": 2320
+    },
+    {
+      "epoch": 0.07456,
+      "grad_norm": 56.53871154785156,
+      "learning_rate": 7.456000000000001e-06,
+      "loss": 1.9791,
+      "step": 2330
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 52.63264465332031,
+      "learning_rate": 7.488000000000001e-06,
+      "loss": 1.9377,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0752,
+      "grad_norm": 56.09365463256836,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.9626,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 51.870357513427734,
+      "learning_rate": 7.552000000000001e-06,
+      "loss": 1.973,
+      "step": 2360
+    },
+    {
+      "epoch": 0.07584,
+      "grad_norm": 54.86323165893555,
+      "learning_rate": 7.5840000000000006e-06,
+      "loss": 1.9508,
+      "step": 2370
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 52.37346649169922,
+      "learning_rate": 7.616000000000001e-06,
+      "loss": 1.9633,
+      "step": 2380
+    },
+    {
+      "epoch": 0.07648,
+      "grad_norm": 53.62883758544922,
+      "learning_rate": 7.648e-06,
+      "loss": 1.92,
+      "step": 2390
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 53.85102462768555,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.9566,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07712,
+      "grad_norm": 52.15278244018555,
+      "learning_rate": 7.712e-06,
+      "loss": 1.9368,
+      "step": 2410
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 54.85605239868164,
+      "learning_rate": 7.744e-06,
+      "loss": 1.9585,
+      "step": 2420
+    },
+    {
+      "epoch": 0.07776,
+      "grad_norm": 56.958553314208984,
+      "learning_rate": 7.776e-06,
+      "loss": 1.9739,
+      "step": 2430
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 51.912452697753906,
+      "learning_rate": 7.808e-06,
+      "loss": 1.97,
+      "step": 2440
+    },
+    {
+      "epoch": 0.0784,
+      "grad_norm": 54.37236022949219,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 1.976,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 54.4852409362793,
+      "learning_rate": 7.872e-06,
+      "loss": 1.9596,
+      "step": 2460
+    },
+    {
+      "epoch": 0.07904,
+      "grad_norm": 55.548343658447266,
+      "learning_rate": 7.904000000000001e-06,
+      "loss": 1.9475,
+      "step": 2470
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 53.80973434448242,
+      "learning_rate": 7.936e-06,
+      "loss": 1.9215,
+      "step": 2480
+    },
+    {
+      "epoch": 0.07968,
+      "grad_norm": 54.02414321899414,
+      "learning_rate": 7.968e-06,
+      "loss": 1.9115,
+      "step": 2490
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 54.69728088378906,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.9142,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08032,
+      "grad_norm": 54.138126373291016,
+      "learning_rate": 8.032e-06,
+      "loss": 1.9299,
+      "step": 2510
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 55.51025390625,
+      "learning_rate": 8.064000000000001e-06,
+      "loss": 1.9347,
+      "step": 2520
+    },
+    {
+      "epoch": 0.08096,
+      "grad_norm": 51.047943115234375,
+      "learning_rate": 8.096e-06,
+      "loss": 1.9486,
+      "step": 2530
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 56.506046295166016,
+      "learning_rate": 8.128e-06,
+      "loss": 1.9777,
+      "step": 2540
+    },
+    {
+      "epoch": 0.0816,
+      "grad_norm": 55.341896057128906,
+      "learning_rate": 8.16e-06,
+      "loss": 1.953,
+      "step": 2550
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 52.932979583740234,
+      "learning_rate": 8.192e-06,
+      "loss": 1.9175,
+      "step": 2560
+    },
+    {
+      "epoch": 0.08224,
+      "grad_norm": 53.52081298828125,
+      "learning_rate": 8.224000000000001e-06,
+      "loss": 1.966,
+      "step": 2570
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 51.49814987182617,
+      "learning_rate": 8.256e-06,
+      "loss": 1.9507,
+      "step": 2580
+    },
+    {
+      "epoch": 0.08288,
+      "grad_norm": 57.984745025634766,
+      "learning_rate": 8.288000000000001e-06,
+      "loss": 1.9256,
+      "step": 2590
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 53.47091293334961,
+      "learning_rate": 8.32e-06,
+      "loss": 1.9209,
+      "step": 2600
+    },
+    {
+      "epoch": 0.08352,
+      "grad_norm": 55.28806686401367,
+      "learning_rate": 8.352e-06,
+      "loss": 1.9572,
+      "step": 2610
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 53.81025314331055,
+      "learning_rate": 8.384000000000001e-06,
+      "loss": 1.9052,
+      "step": 2620
+    },
+    {
+      "epoch": 0.08416,
+      "grad_norm": 54.805519104003906,
+      "learning_rate": 8.416e-06,
+      "loss": 1.9068,
+      "step": 2630
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 55.12998580932617,
+      "learning_rate": 8.448000000000001e-06,
+      "loss": 1.9165,
+      "step": 2640
+    },
+    {
+      "epoch": 0.0848,
+      "grad_norm": 54.45305633544922,
+      "learning_rate": 8.48e-06,
+      "loss": 1.9044,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 52.86287307739258,
+      "learning_rate": 8.512e-06,
+      "loss": 1.9557,
+      "step": 2660
+    },
+    {
+      "epoch": 0.08544,
+      "grad_norm": 54.07919692993164,
+      "learning_rate": 8.544000000000002e-06,
+      "loss": 1.9357,
+      "step": 2670
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 53.222190856933594,
+      "learning_rate": 8.576e-06,
+      "loss": 1.9608,
+      "step": 2680
+    },
+    {
+      "epoch": 0.08608,
+      "grad_norm": 53.212642669677734,
+      "learning_rate": 8.608000000000001e-06,
+      "loss": 1.9576,
+      "step": 2690
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 54.31929016113281,
+      "learning_rate": 8.64e-06,
+      "loss": 1.9355,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08672,
+      "grad_norm": 54.23030090332031,
+      "learning_rate": 8.672000000000001e-06,
+      "loss": 1.9085,
+      "step": 2710
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 54.162254333496094,
+      "learning_rate": 8.704e-06,
+      "loss": 1.9607,
+      "step": 2720
+    },
+    {
+      "epoch": 0.08736,
+      "grad_norm": 52.76661682128906,
+      "learning_rate": 8.736e-06,
+      "loss": 1.9672,
+      "step": 2730
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 55.04710006713867,
+      "learning_rate": 8.768000000000001e-06,
+      "loss": 1.9271,
+      "step": 2740
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 50.674922943115234,
+      "learning_rate": 8.8e-06,
+      "loss": 1.9289,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 53.99641799926758,
+      "learning_rate": 8.832000000000001e-06,
+      "loss": 1.9284,
+      "step": 2760
+    },
+    {
+      "epoch": 0.08864,
+      "grad_norm": 54.7790412902832,
+      "learning_rate": 8.864e-06,
+      "loss": 1.9531,
+      "step": 2770
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 55.55375671386719,
+      "learning_rate": 8.896000000000001e-06,
+      "loss": 1.9442,
+      "step": 2780
+    },
+    {
+      "epoch": 0.08928,
+      "grad_norm": 54.156951904296875,
+      "learning_rate": 8.928000000000002e-06,
+      "loss": 1.9336,
+      "step": 2790
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 55.16888427734375,
+      "learning_rate": 8.96e-06,
+      "loss": 1.9637,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08992,
+      "grad_norm": 53.6899528503418,
+      "learning_rate": 8.992000000000001e-06,
+      "loss": 1.9319,
+      "step": 2810
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 53.920284271240234,
+      "learning_rate": 9.024e-06,
+      "loss": 1.9264,
+      "step": 2820
+    },
+    {
+      "epoch": 0.09056,
+      "grad_norm": 56.052589416503906,
+      "learning_rate": 9.056000000000001e-06,
+      "loss": 1.9338,
+      "step": 2830
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 54.092559814453125,
+      "learning_rate": 9.088000000000002e-06,
+      "loss": 1.9374,
+      "step": 2840
+    },
+    {
+      "epoch": 0.0912,
+      "grad_norm": 54.96030044555664,
+      "learning_rate": 9.12e-06,
+      "loss": 1.9035,
+      "step": 2850
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 54.033260345458984,
+      "learning_rate": 9.152000000000001e-06,
+      "loss": 1.9501,
+      "step": 2860
+    },
+    {
+      "epoch": 0.09184,
+      "grad_norm": 56.32831954956055,
+      "learning_rate": 9.184e-06,
+      "loss": 1.9292,
+      "step": 2870
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 56.31962966918945,
+      "learning_rate": 9.216000000000001e-06,
+      "loss": 1.934,
+      "step": 2880
+    },
+    {
+      "epoch": 0.09248,
+      "grad_norm": 52.693443298339844,
+      "learning_rate": 9.248e-06,
+      "loss": 1.9015,
+      "step": 2890
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 58.64433670043945,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 1.9302,
+      "step": 2900
+    },
+    {
+      "epoch": 0.09312,
+      "grad_norm": 54.63075256347656,
+      "learning_rate": 9.312000000000002e-06,
+      "loss": 1.9186,
+      "step": 2910
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 55.12821578979492,
+      "learning_rate": 9.344e-06,
+      "loss": 1.9432,
+      "step": 2920
+    },
+    {
+      "epoch": 0.09376,
+      "grad_norm": 53.73576736450195,
+      "learning_rate": 9.376000000000001e-06,
+      "loss": 1.9453,
+      "step": 2930
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 54.59904098510742,
+      "learning_rate": 9.408e-06,
+      "loss": 1.9144,
+      "step": 2940
+    },
+    {
+      "epoch": 0.0944,
+      "grad_norm": 56.60734558105469,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 1.92,
+      "step": 2950
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 54.28753662109375,
+      "learning_rate": 9.472000000000002e-06,
+      "loss": 1.9247,
+      "step": 2960
+    },
+    {
+      "epoch": 0.09504,
+      "grad_norm": 54.17729568481445,
+      "learning_rate": 9.504e-06,
+      "loss": 1.9287,
+      "step": 2970
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 51.31298828125,
+      "learning_rate": 9.536000000000002e-06,
+      "loss": 1.9311,
+      "step": 2980
+    },
+    {
+      "epoch": 0.09568,
+      "grad_norm": 51.89619827270508,
+      "learning_rate": 9.568e-06,
+      "loss": 1.9267,
+      "step": 2990
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 53.17412567138672,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.9588,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09632,
+      "grad_norm": 54.527095794677734,
+      "learning_rate": 9.632e-06,
+      "loss": 1.9124,
+      "step": 3010
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 53.29350280761719,
+      "learning_rate": 9.664000000000001e-06,
+      "loss": 1.9428,
+      "step": 3020
+    },
+    {
+      "epoch": 0.09696,
+      "grad_norm": 52.03528594970703,
+      "learning_rate": 9.696000000000002e-06,
+      "loss": 1.9349,
+      "step": 3030
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 52.16074752807617,
+      "learning_rate": 9.728e-06,
+      "loss": 1.8923,
+      "step": 3040
+    },
+    {
+      "epoch": 0.0976,
+      "grad_norm": 54.195716857910156,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.948,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 53.85103988647461,
+      "learning_rate": 9.792e-06,
+      "loss": 1.9211,
+      "step": 3060
+    },
+    {
+      "epoch": 0.09824,
+      "grad_norm": 52.96208190917969,
+      "learning_rate": 9.824000000000001e-06,
+      "loss": 1.9633,
+      "step": 3070
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 54.462432861328125,
+      "learning_rate": 9.856000000000002e-06,
+      "loss": 1.8881,
+      "step": 3080
+    },
+    {
+      "epoch": 0.09888,
+      "grad_norm": 52.74999237060547,
+      "learning_rate": 9.888000000000001e-06,
+      "loss": 1.944,
+      "step": 3090
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 56.4122428894043,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 1.9364,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09952,
+      "grad_norm": 51.049678802490234,
+      "learning_rate": 9.952e-06,
+      "loss": 1.9211,
+      "step": 3110
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 52.30768966674805,
+      "learning_rate": 9.984e-06,
+      "loss": 1.9375,
+      "step": 3120
+    },
+    {
+      "epoch": 0.10016,
+      "grad_norm": 54.005855560302734,
+      "learning_rate": 1.0016000000000002e-05,
+      "loss": 1.9178,
+      "step": 3130
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 55.89468765258789,
+      "learning_rate": 1.0048e-05,
+      "loss": 1.9192,
+      "step": 3140
+    },
+    {
+      "epoch": 0.1008,
+      "grad_norm": 52.7632942199707,
+      "learning_rate": 1.008e-05,
+      "loss": 1.9249,
+      "step": 3150
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 54.614105224609375,
+      "learning_rate": 1.0112000000000002e-05,
+      "loss": 1.9498,
+      "step": 3160
+    },
+    {
+      "epoch": 0.10144,
+      "grad_norm": 53.3930549621582,
+      "learning_rate": 1.0144e-05,
+      "loss": 1.9389,
+      "step": 3170
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 52.193302154541016,
+      "learning_rate": 1.0176000000000002e-05,
+      "loss": 1.9007,
+      "step": 3180
+    },
+    {
+      "epoch": 0.10208,
+      "grad_norm": 52.78200912475586,
+      "learning_rate": 1.0208e-05,
+      "loss": 1.9244,
+      "step": 3190
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 56.6796875,
+      "learning_rate": 1.024e-05,
+      "loss": 1.9249,
+      "step": 3200
+    },
+    {
+      "epoch": 0.10272,
+      "grad_norm": 52.93400573730469,
+      "learning_rate": 1.0272e-05,
+      "loss": 1.901,
+      "step": 3210
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 52.60874557495117,
+      "learning_rate": 1.0304e-05,
+      "loss": 1.8986,
+      "step": 3220
+    },
+    {
+      "epoch": 0.10336,
+      "grad_norm": 54.57966232299805,
+      "learning_rate": 1.0336000000000002e-05,
+      "loss": 1.9102,
+      "step": 3230
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 52.31039810180664,
+      "learning_rate": 1.0368e-05,
+      "loss": 1.9536,
+      "step": 3240
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 53.94814682006836,
+      "learning_rate": 1.04e-05,
+      "loss": 1.9062,
+      "step": 3250
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 54.85002517700195,
+      "learning_rate": 1.0432e-05,
+      "loss": 1.9327,
+      "step": 3260
+    },
+    {
+      "epoch": 0.10464,
+      "grad_norm": 54.65288543701172,
+      "learning_rate": 1.0464e-05,
+      "loss": 1.9333,
+      "step": 3270
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 53.56013488769531,
+      "learning_rate": 1.0496000000000003e-05,
+      "loss": 1.9172,
+      "step": 3280
+    },
+    {
+      "epoch": 0.10528,
+      "grad_norm": 51.968772888183594,
+      "learning_rate": 1.0528e-05,
+      "loss": 1.9291,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 53.27824020385742,
+      "learning_rate": 1.056e-05,
+      "loss": 1.9172,
+      "step": 3300
+    },
+    {
+      "epoch": 0.10592,
+      "grad_norm": 52.052886962890625,
+      "learning_rate": 1.0592e-05,
+      "loss": 1.9086,
+      "step": 3310
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 57.092655181884766,
+      "learning_rate": 1.0624e-05,
+      "loss": 1.9064,
+      "step": 3320
+    },
+    {
+      "epoch": 0.10656,
+      "grad_norm": 53.775962829589844,
+      "learning_rate": 1.0656000000000003e-05,
+      "loss": 1.9032,
+      "step": 3330
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 54.14551544189453,
+      "learning_rate": 1.0688e-05,
+      "loss": 1.8882,
+      "step": 3340
+    },
+    {
+      "epoch": 0.1072,
+      "grad_norm": 51.43288040161133,
+      "learning_rate": 1.072e-05,
+      "loss": 1.9218,
+      "step": 3350
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 53.97608947753906,
+      "learning_rate": 1.0752e-05,
+      "loss": 1.9089,
+      "step": 3360
+    },
+    {
+      "epoch": 0.10784,
+      "grad_norm": 54.1445426940918,
+      "learning_rate": 1.0784e-05,
+      "loss": 1.9201,
+      "step": 3370
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 52.34402084350586,
+      "learning_rate": 1.0816e-05,
+      "loss": 1.921,
+      "step": 3380
+    },
+    {
+      "epoch": 0.10848,
+      "grad_norm": 53.012149810791016,
+      "learning_rate": 1.0848e-05,
+      "loss": 1.8994,
+      "step": 3390
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 51.83297348022461,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 1.9445,
+      "step": 3400
+    },
+    {
+      "epoch": 0.10912,
+      "grad_norm": 54.20074462890625,
+      "learning_rate": 1.0912e-05,
+      "loss": 1.9188,
+      "step": 3410
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 53.536834716796875,
+      "learning_rate": 1.0944e-05,
+      "loss": 1.9129,
+      "step": 3420
+    },
+    {
+      "epoch": 0.10976,
+      "grad_norm": 55.08799362182617,
+      "learning_rate": 1.0976e-05,
+      "loss": 1.9033,
+      "step": 3430
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 53.32230758666992,
+      "learning_rate": 1.1008e-05,
+      "loss": 1.8888,
+      "step": 3440
+    },
+    {
+      "epoch": 0.1104,
+      "grad_norm": 51.482086181640625,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 1.929,
+      "step": 3450
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 52.34898376464844,
+      "learning_rate": 1.1072e-05,
+      "loss": 1.9445,
+      "step": 3460
+    },
+    {
+      "epoch": 0.11104,
+      "grad_norm": 51.944236755371094,
+      "learning_rate": 1.1104e-05,
+      "loss": 1.9294,
+      "step": 3470
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 53.45691680908203,
+      "learning_rate": 1.1136e-05,
+      "loss": 1.9127,
+      "step": 3480
+    },
+    {
+      "epoch": 0.11168,
+      "grad_norm": 53.34093475341797,
+      "learning_rate": 1.1168e-05,
+      "loss": 1.9039,
+      "step": 3490
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 53.048362731933594,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.9287,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11232,
+      "grad_norm": 52.95180130004883,
+      "learning_rate": 1.1232e-05,
+      "loss": 1.9277,
+      "step": 3510
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 51.409751892089844,
+      "learning_rate": 1.1264000000000001e-05,
+      "loss": 1.9043,
+      "step": 3520
+    },
+    {
+      "epoch": 0.11296,
+      "grad_norm": 53.68730163574219,
+      "learning_rate": 1.1296e-05,
+      "loss": 1.9118,
+      "step": 3530
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 54.60579299926758,
+      "learning_rate": 1.1328e-05,
+      "loss": 1.9177,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1136,
+      "grad_norm": 52.81605529785156,
+      "learning_rate": 1.136e-05,
+      "loss": 1.899,
+      "step": 3550
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 52.89885711669922,
+      "learning_rate": 1.1392e-05,
+      "loss": 1.8706,
+      "step": 3560
+    },
+    {
+      "epoch": 0.11424,
+      "grad_norm": 53.58485794067383,
+      "learning_rate": 1.1424000000000001e-05,
+      "loss": 1.8861,
+      "step": 3570
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 52.32903289794922,
+      "learning_rate": 1.1456e-05,
+      "loss": 1.9004,
+      "step": 3580
+    },
+    {
+      "epoch": 0.11488,
+      "grad_norm": 51.63499450683594,
+      "learning_rate": 1.1488e-05,
+      "loss": 1.891,
+      "step": 3590
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 56.98868942260742,
+      "learning_rate": 1.152e-05,
+      "loss": 1.8828,
+      "step": 3600
+    },
+    {
+      "epoch": 0.11552,
+      "grad_norm": 52.2146110534668,
+      "learning_rate": 1.1552e-05,
+      "loss": 1.9317,
+      "step": 3610
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 53.047584533691406,
+      "learning_rate": 1.1584000000000001e-05,
+      "loss": 1.9252,
+      "step": 3620
+    },
+    {
+      "epoch": 0.11616,
+      "grad_norm": 54.36164093017578,
+      "learning_rate": 1.1616e-05,
+      "loss": 1.9031,
+      "step": 3630
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 51.535457611083984,
+      "learning_rate": 1.1648000000000001e-05,
+      "loss": 1.8883,
+      "step": 3640
+    },
+    {
+      "epoch": 0.1168,
+      "grad_norm": 51.89107131958008,
+      "learning_rate": 1.168e-05,
+      "loss": 1.9308,
+      "step": 3650
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 52.61616897583008,
+      "learning_rate": 1.1712e-05,
+      "loss": 1.904,
+      "step": 3660
+    },
+    {
+      "epoch": 0.11744,
+      "grad_norm": 54.508609771728516,
+      "learning_rate": 1.1744000000000001e-05,
+      "loss": 1.8981,
+      "step": 3670
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 52.905399322509766,
+      "learning_rate": 1.1776e-05,
+      "loss": 1.8918,
+      "step": 3680
+    },
+    {
+      "epoch": 0.11808,
+      "grad_norm": 54.308448791503906,
+      "learning_rate": 1.1808000000000001e-05,
+      "loss": 1.8762,
+      "step": 3690
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 50.295318603515625,
+      "learning_rate": 1.184e-05,
+      "loss": 1.9246,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11872,
+      "grad_norm": 55.23350524902344,
+      "learning_rate": 1.1872000000000001e-05,
+      "loss": 1.9205,
+      "step": 3710
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 52.099143981933594,
+      "learning_rate": 1.1904e-05,
+      "loss": 1.9082,
+      "step": 3720
+    },
+    {
+      "epoch": 0.11936,
+      "grad_norm": 52.80199432373047,
+      "learning_rate": 1.1936e-05,
+      "loss": 1.8809,
+      "step": 3730
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 50.771202087402344,
+      "learning_rate": 1.1968000000000001e-05,
+      "loss": 1.892,
+      "step": 3740
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 52.11357879638672,
+      "learning_rate": 1.2e-05,
+      "loss": 1.9049,
+      "step": 3750
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 53.48591613769531,
+      "learning_rate": 1.2032000000000001e-05,
+      "loss": 1.874,
+      "step": 3760
+    },
+    {
+      "epoch": 0.12064,
+      "grad_norm": 52.847015380859375,
+      "learning_rate": 1.2064e-05,
+      "loss": 1.8921,
+      "step": 3770
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 53.891414642333984,
+      "learning_rate": 1.2096e-05,
+      "loss": 1.9263,
+      "step": 3780
+    },
+    {
+      "epoch": 0.12128,
+      "grad_norm": 53.7008171081543,
+      "learning_rate": 1.2128000000000001e-05,
+      "loss": 1.8968,
+      "step": 3790
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 56.37105178833008,
+      "learning_rate": 1.216e-05,
+      "loss": 1.8804,
+      "step": 3800
+    },
+    {
+      "epoch": 0.12192,
+      "grad_norm": 49.93663024902344,
+      "learning_rate": 1.2192000000000001e-05,
+      "loss": 1.8962,
+      "step": 3810
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 53.451927185058594,
+      "learning_rate": 1.2224e-05,
+      "loss": 1.8972,
+      "step": 3820
+    },
+    {
+      "epoch": 0.12256,
+      "grad_norm": 51.564979553222656,
+      "learning_rate": 1.2256000000000001e-05,
+      "loss": 1.8889,
+      "step": 3830
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 54.76313400268555,
+      "learning_rate": 1.2288e-05,
+      "loss": 1.8864,
+      "step": 3840
+    },
+    {
+      "epoch": 0.1232,
+      "grad_norm": 52.7327880859375,
+      "learning_rate": 1.232e-05,
+      "loss": 1.9118,
+      "step": 3850
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 54.4197883605957,
+      "learning_rate": 1.2352000000000001e-05,
+      "loss": 1.8859,
+      "step": 3860
+    },
+    {
+      "epoch": 0.12384,
+      "grad_norm": 52.1079216003418,
+      "learning_rate": 1.2384e-05,
+      "loss": 1.919,
+      "step": 3870
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 51.248836517333984,
+      "learning_rate": 1.2416000000000001e-05,
+      "loss": 1.911,
+      "step": 3880
+    },
+    {
+      "epoch": 0.12448,
+      "grad_norm": 52.00767517089844,
+      "learning_rate": 1.2448e-05,
+      "loss": 1.8667,
+      "step": 3890
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 53.36152267456055,
+      "learning_rate": 1.248e-05,
+      "loss": 1.8689,
+      "step": 3900
+    },
+    {
+      "epoch": 0.12512,
+      "grad_norm": 54.70954895019531,
+      "learning_rate": 1.2512000000000002e-05,
+      "loss": 1.9175,
+      "step": 3910
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 53.92921829223633,
+      "learning_rate": 1.2544e-05,
+      "loss": 1.9274,
+      "step": 3920
+    },
+    {
+      "epoch": 0.12576,
+      "grad_norm": 53.22060012817383,
+      "learning_rate": 1.2576000000000001e-05,
+      "loss": 1.8985,
+      "step": 3930
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 53.11907958984375,
+      "learning_rate": 1.2608e-05,
+      "loss": 1.9053,
+      "step": 3940
+    },
+    {
+      "epoch": 0.1264,
+      "grad_norm": 56.41827392578125,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 1.9134,
+      "step": 3950
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 53.10688018798828,
+      "learning_rate": 1.2672000000000002e-05,
+      "loss": 1.8679,
+      "step": 3960
+    },
+    {
+      "epoch": 0.12704,
+      "grad_norm": 54.76327133178711,
+      "learning_rate": 1.2704e-05,
+      "loss": 1.88,
+      "step": 3970
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 51.2487907409668,
+      "learning_rate": 1.2736000000000001e-05,
+      "loss": 1.8733,
+      "step": 3980
+    },
+    {
+      "epoch": 0.12768,
+      "grad_norm": 54.9959602355957,
+      "learning_rate": 1.2768e-05,
+      "loss": 1.9171,
+      "step": 3990
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 51.75206756591797,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.9109,
+      "step": 4000
+    },
+    {
+      "epoch": 0.12832,
+      "grad_norm": 52.249847412109375,
+      "learning_rate": 1.2832e-05,
+      "loss": 1.8478,
+      "step": 4010
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 51.156700134277344,
+      "learning_rate": 1.2864e-05,
+      "loss": 1.8795,
+      "step": 4020
+    },
+    {
+      "epoch": 0.12896,
+      "grad_norm": 52.112060546875,
+      "learning_rate": 1.2896000000000002e-05,
+      "loss": 1.8716,
+      "step": 4030
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 53.839229583740234,
+      "learning_rate": 1.2928e-05,
+      "loss": 1.9113,
+      "step": 4040
+    },
+    {
+      "epoch": 0.1296,
+      "grad_norm": 50.280059814453125,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 1.9032,
+      "step": 4050
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 55.25138473510742,
+      "learning_rate": 1.2992e-05,
+      "loss": 1.9209,
+      "step": 4060
+    },
+    {
+      "epoch": 0.13024,
+      "grad_norm": 52.41676330566406,
+      "learning_rate": 1.3024000000000001e-05,
+      "loss": 1.8936,
+      "step": 4070
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 53.2076416015625,
+      "learning_rate": 1.3056000000000002e-05,
+      "loss": 1.917,
+      "step": 4080
+    },
+    {
+      "epoch": 0.13088,
+      "grad_norm": 53.43117141723633,
+      "learning_rate": 1.3088e-05,
+      "loss": 1.8803,
+      "step": 4090
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 50.76637268066406,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 1.9023,
+      "step": 4100
+    },
+    {
+      "epoch": 0.13152,
+      "grad_norm": 51.317291259765625,
+      "learning_rate": 1.3152e-05,
+      "loss": 1.9109,
+      "step": 4110
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 52.48469924926758,
+      "learning_rate": 1.3184000000000001e-05,
+      "loss": 1.8672,
+      "step": 4120
+    },
+    {
+      "epoch": 0.13216,
+      "grad_norm": 52.752708435058594,
+      "learning_rate": 1.3216000000000002e-05,
+      "loss": 1.9122,
+      "step": 4130
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 53.35075378417969,
+      "learning_rate": 1.3248000000000001e-05,
+      "loss": 1.888,
+      "step": 4140
+    },
+    {
+      "epoch": 0.1328,
+      "grad_norm": 54.060489654541016,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 1.8922,
+      "step": 4150
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 50.84089660644531,
+      "learning_rate": 1.3312e-05,
+      "loss": 1.8884,
+      "step": 4160
+    },
+    {
+      "epoch": 0.13344,
+      "grad_norm": 52.881195068359375,
+      "learning_rate": 1.3344000000000001e-05,
+      "loss": 1.8952,
+      "step": 4170
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 52.58907699584961,
+      "learning_rate": 1.3376e-05,
+      "loss": 1.8752,
+      "step": 4180
+    },
+    {
+      "epoch": 0.13408,
+      "grad_norm": 53.9188346862793,
+      "learning_rate": 1.3408000000000001e-05,
+      "loss": 1.8764,
+      "step": 4190
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 51.16621780395508,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 1.867,
+      "step": 4200
+    },
+    {
+      "epoch": 0.13472,
+      "grad_norm": 48.84093475341797,
+      "learning_rate": 1.3472e-05,
+      "loss": 1.8732,
+      "step": 4210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 52.254451751708984,
+      "learning_rate": 1.3504000000000001e-05,
+      "loss": 1.9032,
+      "step": 4220
+    },
+    {
+      "epoch": 0.13536,
+      "grad_norm": 54.642391204833984,
+      "learning_rate": 1.3536e-05,
+      "loss": 1.8666,
+      "step": 4230
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 53.06638717651367,
+      "learning_rate": 1.3568000000000001e-05,
+      "loss": 1.8792,
+      "step": 4240
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 52.53422927856445,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.889,
+      "step": 4250
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 53.5844612121582,
+      "learning_rate": 1.3632000000000001e-05,
+      "loss": 1.8908,
+      "step": 4260
+    },
+    {
+      "epoch": 0.13664,
+      "grad_norm": 52.1329345703125,
+      "learning_rate": 1.3664000000000002e-05,
+      "loss": 1.8925,
+      "step": 4270
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 52.22524642944336,
+      "learning_rate": 1.3696e-05,
+      "loss": 1.8654,
+      "step": 4280
+    },
+    {
+      "epoch": 0.13728,
+      "grad_norm": 53.75664138793945,
+      "learning_rate": 1.3728000000000001e-05,
+      "loss": 1.8558,
+      "step": 4290
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 55.64548110961914,
+      "learning_rate": 1.376e-05,
+      "loss": 1.878,
+      "step": 4300
+    },
+    {
+      "epoch": 0.13792,
+      "grad_norm": 53.45896911621094,
+      "learning_rate": 1.3792000000000001e-05,
+      "loss": 1.8806,
+      "step": 4310
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 52.175533294677734,
+      "learning_rate": 1.3824000000000002e-05,
+      "loss": 1.8848,
+      "step": 4320
+    },
+    {
+      "epoch": 0.13856,
+      "grad_norm": 50.984527587890625,
+      "learning_rate": 1.3856e-05,
+      "loss": 1.8641,
+      "step": 4330
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 54.03857421875,
+      "learning_rate": 1.3888000000000002e-05,
+      "loss": 1.8907,
+      "step": 4340
+    },
+    {
+      "epoch": 0.1392,
+      "grad_norm": 51.122833251953125,
+      "learning_rate": 1.392e-05,
+      "loss": 1.9,
+      "step": 4350
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 49.632469177246094,
+      "learning_rate": 1.3952000000000001e-05,
+      "loss": 1.8702,
+      "step": 4360
+    },
+    {
+      "epoch": 0.13984,
+      "grad_norm": 51.28369140625,
+      "learning_rate": 1.3984000000000002e-05,
+      "loss": 1.8763,
+      "step": 4370
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 50.211151123046875,
+      "learning_rate": 1.4016000000000001e-05,
+      "loss": 1.85,
+      "step": 4380
+    },
+    {
+      "epoch": 0.14048,
+      "grad_norm": 51.59481430053711,
+      "learning_rate": 1.4048000000000002e-05,
+      "loss": 1.8771,
+      "step": 4390
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 52.014102935791016,
+      "learning_rate": 1.408e-05,
+      "loss": 1.872,
+      "step": 4400
+    },
+    {
+      "epoch": 0.14112,
+      "grad_norm": 55.16132354736328,
+      "learning_rate": 1.4112000000000001e-05,
+      "loss": 1.8373,
+      "step": 4410
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 54.09187316894531,
+      "learning_rate": 1.4144000000000002e-05,
+      "loss": 1.8705,
+      "step": 4420
+    },
+    {
+      "epoch": 0.14176,
+      "grad_norm": 52.59539031982422,
+      "learning_rate": 1.4176000000000001e-05,
+      "loss": 1.8561,
+      "step": 4430
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 49.13827133178711,
+      "learning_rate": 1.4208000000000002e-05,
+      "loss": 1.8817,
+      "step": 4440
+    },
+    {
+      "epoch": 0.1424,
+      "grad_norm": 51.36490249633789,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.8702,
+      "step": 4450
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 50.50772476196289,
+      "learning_rate": 1.4272000000000002e-05,
+      "loss": 1.8779,
+      "step": 4460
+    },
+    {
+      "epoch": 0.14304,
+      "grad_norm": 53.97172546386719,
+      "learning_rate": 1.4304e-05,
+      "loss": 1.9014,
+      "step": 4470
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 52.78620910644531,
+      "learning_rate": 1.4336000000000001e-05,
+      "loss": 1.9061,
+      "step": 4480
+    },
+    {
+      "epoch": 0.14368,
+      "grad_norm": 51.83744430541992,
+      "learning_rate": 1.4368000000000002e-05,
+      "loss": 1.904,
+      "step": 4490
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 52.39995193481445,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.8681,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14432,
+      "grad_norm": 54.1072998046875,
+      "learning_rate": 1.4432000000000002e-05,
+      "loss": 1.9006,
+      "step": 4510
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 50.79708480834961,
+      "learning_rate": 1.4464e-05,
+      "loss": 1.8475,
+      "step": 4520
+    },
+    {
+      "epoch": 0.14496,
+      "grad_norm": 52.832576751708984,
+      "learning_rate": 1.4496000000000001e-05,
+      "loss": 1.8898,
+      "step": 4530
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 52.59731674194336,
+      "learning_rate": 1.4528000000000002e-05,
+      "loss": 1.8694,
+      "step": 4540
+    },
+    {
+      "epoch": 0.1456,
+      "grad_norm": 53.897300720214844,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 1.8701,
+      "step": 4550
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 51.246253967285156,
+      "learning_rate": 1.4592000000000002e-05,
+      "loss": 1.8392,
+      "step": 4560
+    },
+    {
+      "epoch": 0.14624,
+      "grad_norm": 51.34325408935547,
+      "learning_rate": 1.4624000000000001e-05,
+      "loss": 1.8767,
+      "step": 4570
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 52.48518371582031,
+      "learning_rate": 1.4656000000000002e-05,
+      "loss": 1.8632,
+      "step": 4580
+    },
+    {
+      "epoch": 0.14688,
+      "grad_norm": 52.77693557739258,
+      "learning_rate": 1.4688000000000002e-05,
+      "loss": 1.855,
+      "step": 4590
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 51.66023635864258,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 1.888,
+      "step": 4600
+    },
+    {
+      "epoch": 0.14752,
+      "grad_norm": 52.59580612182617,
+      "learning_rate": 1.4752000000000002e-05,
+      "loss": 1.8731,
+      "step": 4610
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 50.908451080322266,
+      "learning_rate": 1.4784000000000001e-05,
+      "loss": 1.9071,
+      "step": 4620
+    },
+    {
+      "epoch": 0.14816,
+      "grad_norm": 54.34564208984375,
+      "learning_rate": 1.4816000000000002e-05,
+      "loss": 1.8867,
+      "step": 4630
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 52.151248931884766,
+      "learning_rate": 1.4848e-05,
+      "loss": 1.8917,
+      "step": 4640
+    },
+    {
+      "epoch": 0.1488,
+      "grad_norm": 51.90532684326172,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 1.8505,
+      "step": 4650
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 53.09203338623047,
+      "learning_rate": 1.4912000000000002e-05,
+      "loss": 1.8405,
+      "step": 4660
+    },
+    {
+      "epoch": 0.14944,
+      "grad_norm": 50.767364501953125,
+      "learning_rate": 1.4944000000000001e-05,
+      "loss": 1.8498,
+      "step": 4670
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 55.40914535522461,
+      "learning_rate": 1.4976000000000002e-05,
+      "loss": 1.8578,
+      "step": 4680
+    },
+    {
+      "epoch": 0.15008,
+      "grad_norm": 50.39368438720703,
+      "learning_rate": 1.5008000000000001e-05,
+      "loss": 1.8433,
+      "step": 4690
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 50.388824462890625,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 1.8765,
+      "step": 4700
+    },
+    {
+      "epoch": 0.15072,
+      "grad_norm": 52.96123504638672,
+      "learning_rate": 1.5072000000000002e-05,
+      "loss": 1.8703,
+      "step": 4710
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 54.47455978393555,
+      "learning_rate": 1.5104000000000001e-05,
+      "loss": 1.9063,
+      "step": 4720
+    },
+    {
+      "epoch": 0.15136,
+      "grad_norm": 53.61328887939453,
+      "learning_rate": 1.5136000000000002e-05,
+      "loss": 1.8588,
+      "step": 4730
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 53.73160934448242,
+      "learning_rate": 1.5168000000000001e-05,
+      "loss": 1.8677,
+      "step": 4740
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 52.09059524536133,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.8523,
+      "step": 4750
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 51.840579986572266,
+      "learning_rate": 1.5232000000000003e-05,
+      "loss": 1.8375,
+      "step": 4760
+    },
+    {
+      "epoch": 0.15264,
+      "grad_norm": 51.018516540527344,
+      "learning_rate": 1.5264e-05,
+      "loss": 1.8733,
+      "step": 4770
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 51.18205261230469,
+      "learning_rate": 1.5296e-05,
+      "loss": 1.8648,
+      "step": 4780
+    },
+    {
+      "epoch": 0.15328,
+      "grad_norm": 52.53403091430664,
+      "learning_rate": 1.5328e-05,
+      "loss": 1.8639,
+      "step": 4790
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 49.89756774902344,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 1.8552,
+      "step": 4800
+    },
+    {
+      "epoch": 0.15392,
+      "grad_norm": 50.81180191040039,
+      "learning_rate": 1.5392e-05,
+      "loss": 1.8875,
+      "step": 4810
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 50.676231384277344,
+      "learning_rate": 1.5424e-05,
+      "loss": 1.865,
+      "step": 4820
+    },
+    {
+      "epoch": 0.15456,
+      "grad_norm": 51.839927673339844,
+      "learning_rate": 1.5456000000000002e-05,
+      "loss": 1.8921,
+      "step": 4830
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 53.45131301879883,
+      "learning_rate": 1.5488e-05,
+      "loss": 1.8541,
+      "step": 4840
+    },
+    {
+      "epoch": 0.1552,
+      "grad_norm": 51.4936637878418,
+      "learning_rate": 1.552e-05,
+      "loss": 1.878,
+      "step": 4850
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 51.89486312866211,
+      "learning_rate": 1.5552e-05,
+      "loss": 1.8731,
+      "step": 4860
+    },
+    {
+      "epoch": 0.15584,
+      "grad_norm": 53.06976318359375,
+      "learning_rate": 1.5584000000000002e-05,
+      "loss": 1.9073,
+      "step": 4870
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 50.86164855957031,
+      "learning_rate": 1.5616e-05,
+      "loss": 1.8745,
+      "step": 4880
+    },
+    {
+      "epoch": 0.15648,
+      "grad_norm": 49.303367614746094,
+      "learning_rate": 1.5648e-05,
+      "loss": 1.8539,
+      "step": 4890
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 52.93231201171875,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 1.868,
+      "step": 4900
+    },
+    {
+      "epoch": 0.15712,
+      "grad_norm": 51.90977478027344,
+      "learning_rate": 1.5712e-05,
+      "loss": 1.8691,
+      "step": 4910
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 52.30866622924805,
+      "learning_rate": 1.5744e-05,
+      "loss": 1.8875,
+      "step": 4920
+    },
+    {
+      "epoch": 0.15776,
+      "grad_norm": 52.9408073425293,
+      "learning_rate": 1.5776e-05,
+      "loss": 1.8858,
+      "step": 4930
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 54.67021560668945,
+      "learning_rate": 1.5808000000000002e-05,
+      "loss": 1.8601,
+      "step": 4940
+    },
+    {
+      "epoch": 0.1584,
+      "grad_norm": 51.81747055053711,
+      "learning_rate": 1.584e-05,
+      "loss": 1.8401,
+      "step": 4950
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 49.072303771972656,
+      "learning_rate": 1.5872e-05,
+      "loss": 1.8616,
+      "step": 4960
+    },
+    {
+      "epoch": 0.15904,
+      "grad_norm": 51.137847900390625,
+      "learning_rate": 1.5904000000000002e-05,
+      "loss": 1.858,
+      "step": 4970
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 52.146732330322266,
+      "learning_rate": 1.5936e-05,
+      "loss": 1.8621,
+      "step": 4980
+    },
+    {
+      "epoch": 0.15968,
+      "grad_norm": 52.12568664550781,
+      "learning_rate": 1.5968e-05,
+      "loss": 1.8653,
+      "step": 4990
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 50.79985809326172,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.8586,
+      "step": 5000
+    },
+    {
+      "epoch": 0.16032,
+      "grad_norm": 51.26686477661133,
+      "learning_rate": 1.6032e-05,
+      "loss": 1.8785,
+      "step": 5010
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 52.268218994140625,
+      "learning_rate": 1.6064e-05,
+      "loss": 1.8588,
+      "step": 5020
+    },
+    {
+      "epoch": 0.16096,
+      "grad_norm": 52.18198776245117,
+      "learning_rate": 1.6096e-05,
+      "loss": 1.854,
+      "step": 5030
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 51.68186569213867,
+      "learning_rate": 1.6128000000000002e-05,
+      "loss": 1.8735,
+      "step": 5040
+    },
+    {
+      "epoch": 0.1616,
+      "grad_norm": 50.604740142822266,
+      "learning_rate": 1.616e-05,
+      "loss": 1.8441,
+      "step": 5050
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 50.60602569580078,
+      "learning_rate": 1.6192e-05,
+      "loss": 1.8835,
+      "step": 5060
+    },
+    {
+      "epoch": 0.16224,
+      "grad_norm": 51.33891677856445,
+      "learning_rate": 1.6224000000000003e-05,
+      "loss": 1.8837,
+      "step": 5070
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 48.54633331298828,
+      "learning_rate": 1.6256e-05,
+      "loss": 1.8548,
+      "step": 5080
+    },
+    {
+      "epoch": 0.16288,
+      "grad_norm": 48.90038299560547,
+      "learning_rate": 1.6288e-05,
+      "loss": 1.8318,
+      "step": 5090
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 50.23739242553711,
+      "learning_rate": 1.632e-05,
+      "loss": 1.8365,
+      "step": 5100
+    },
+    {
+      "epoch": 0.16352,
+      "grad_norm": 49.558319091796875,
+      "learning_rate": 1.6352000000000002e-05,
+      "loss": 1.8786,
+      "step": 5110
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 52.7110481262207,
+      "learning_rate": 1.6384e-05,
+      "loss": 1.8586,
+      "step": 5120
+    },
+    {
+      "epoch": 0.16416,
+      "grad_norm": 50.92003631591797,
+      "learning_rate": 1.6416e-05,
+      "loss": 1.8429,
+      "step": 5130
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 50.95296096801758,
+      "learning_rate": 1.6448000000000002e-05,
+      "loss": 1.8639,
+      "step": 5140
+    },
+    {
+      "epoch": 0.1648,
+      "grad_norm": 52.3206901550293,
+      "learning_rate": 1.648e-05,
+      "loss": 1.8414,
+      "step": 5150
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 51.82664489746094,
+      "learning_rate": 1.6512e-05,
+      "loss": 1.8899,
+      "step": 5160
+    },
+    {
+      "epoch": 0.16544,
+      "grad_norm": 51.14692687988281,
+      "learning_rate": 1.6544000000000003e-05,
+      "loss": 1.8878,
+      "step": 5170
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 49.87936782836914,
+      "learning_rate": 1.6576000000000002e-05,
+      "loss": 1.863,
+      "step": 5180
+    },
+    {
+      "epoch": 0.16608,
+      "grad_norm": 53.458229064941406,
+      "learning_rate": 1.6608e-05,
+      "loss": 1.876,
+      "step": 5190
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 49.172882080078125,
+      "learning_rate": 1.664e-05,
+      "loss": 1.8608,
+      "step": 5200
+    },
+    {
+      "epoch": 0.16672,
+      "grad_norm": 52.48230743408203,
+      "learning_rate": 1.6672000000000002e-05,
+      "loss": 1.8329,
+      "step": 5210
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 50.219295501708984,
+      "learning_rate": 1.6704e-05,
+      "loss": 1.8544,
+      "step": 5220
+    },
+    {
+      "epoch": 0.16736,
+      "grad_norm": 52.36051559448242,
+      "learning_rate": 1.6736e-05,
+      "loss": 1.8895,
+      "step": 5230
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 50.70857620239258,
+      "learning_rate": 1.6768000000000003e-05,
+      "loss": 1.8861,
+      "step": 5240
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 53.20645523071289,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.8769,
+      "step": 5250
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 50.887107849121094,
+      "learning_rate": 1.6832e-05,
+      "loss": 1.8503,
+      "step": 5260
+    },
+    {
+      "epoch": 0.16864,
+      "grad_norm": 52.35952377319336,
+      "learning_rate": 1.6864e-05,
+      "loss": 1.8513,
+      "step": 5270
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 50.2011833190918,
+      "learning_rate": 1.6896000000000002e-05,
+      "loss": 1.8662,
+      "step": 5280
+    },
+    {
+      "epoch": 0.16928,
+      "grad_norm": 49.79100799560547,
+      "learning_rate": 1.6928e-05,
+      "loss": 1.8815,
+      "step": 5290
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 52.19636154174805,
+      "learning_rate": 1.696e-05,
+      "loss": 1.8728,
+      "step": 5300
+    },
+    {
+      "epoch": 0.16992,
+      "grad_norm": 51.62786865234375,
+      "learning_rate": 1.6992000000000003e-05,
+      "loss": 1.8465,
+      "step": 5310
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 51.08530044555664,
+      "learning_rate": 1.7024e-05,
+      "loss": 1.8378,
+      "step": 5320
+    },
+    {
+      "epoch": 0.17056,
+      "grad_norm": 52.85392379760742,
+      "learning_rate": 1.7056e-05,
+      "loss": 1.8521,
+      "step": 5330
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 53.09284973144531,
+      "learning_rate": 1.7088000000000003e-05,
+      "loss": 1.8643,
+      "step": 5340
+    },
+    {
+      "epoch": 0.1712,
+      "grad_norm": 50.43644332885742,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 1.8455,
+      "step": 5350
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 48.954288482666016,
+      "learning_rate": 1.7152e-05,
+      "loss": 1.8727,
+      "step": 5360
+    },
+    {
+      "epoch": 0.17184,
+      "grad_norm": 49.91568374633789,
+      "learning_rate": 1.7184e-05,
+      "loss": 1.8408,
+      "step": 5370
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 52.08123016357422,
+      "learning_rate": 1.7216000000000003e-05,
+      "loss": 1.863,
+      "step": 5380
+    },
+    {
+      "epoch": 0.17248,
+      "grad_norm": 50.53387451171875,
+      "learning_rate": 1.7248e-05,
+      "loss": 1.8408,
+      "step": 5390
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 48.23560333251953,
+      "learning_rate": 1.728e-05,
+      "loss": 1.8469,
+      "step": 5400
+    },
+    {
+      "epoch": 0.17312,
+      "grad_norm": 50.42172622680664,
+      "learning_rate": 1.7312000000000003e-05,
+      "loss": 1.8425,
+      "step": 5410
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 54.85725784301758,
+      "learning_rate": 1.7344000000000002e-05,
+      "loss": 1.832,
+      "step": 5420
+    },
+    {
+      "epoch": 0.17376,
+      "grad_norm": 50.49094772338867,
+      "learning_rate": 1.7376e-05,
+      "loss": 1.8302,
+      "step": 5430
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 49.896392822265625,
+      "learning_rate": 1.7408e-05,
+      "loss": 1.8576,
+      "step": 5440
+    },
+    {
+      "epoch": 0.1744,
+      "grad_norm": 51.21379089355469,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 1.8592,
+      "step": 5450
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 49.87135696411133,
+      "learning_rate": 1.7472e-05,
+      "loss": 1.8228,
+      "step": 5460
+    },
+    {
+      "epoch": 0.17504,
+      "grad_norm": 51.83329772949219,
+      "learning_rate": 1.7504e-05,
+      "loss": 1.8362,
+      "step": 5470
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 50.30447006225586,
+      "learning_rate": 1.7536000000000003e-05,
+      "loss": 1.8256,
+      "step": 5480
+    },
+    {
+      "epoch": 0.17568,
+      "grad_norm": 54.67784118652344,
+      "learning_rate": 1.7568000000000002e-05,
+      "loss": 1.8355,
+      "step": 5490
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 50.64011764526367,
+      "learning_rate": 1.76e-05,
+      "loss": 1.8306,
+      "step": 5500
+    },
+    {
+      "epoch": 0.17632,
+      "grad_norm": 50.306182861328125,
+      "learning_rate": 1.7632000000000003e-05,
+      "loss": 1.8325,
+      "step": 5510
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 51.00660705566406,
+      "learning_rate": 1.7664000000000002e-05,
+      "loss": 1.8385,
+      "step": 5520
+    },
+    {
+      "epoch": 0.17696,
+      "grad_norm": 50.2363395690918,
+      "learning_rate": 1.7696e-05,
+      "loss": 1.8445,
+      "step": 5530
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 49.31149673461914,
+      "learning_rate": 1.7728e-05,
+      "loss": 1.8328,
+      "step": 5540
+    },
+    {
+      "epoch": 0.1776,
+      "grad_norm": 50.22332000732422,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.8329,
+      "step": 5550
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 51.147701263427734,
+      "learning_rate": 1.7792000000000002e-05,
+      "loss": 1.8246,
+      "step": 5560
+    },
+    {
+      "epoch": 0.17824,
+      "grad_norm": 49.7264404296875,
+      "learning_rate": 1.7824e-05,
+      "loss": 1.8628,
+      "step": 5570
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 51.533050537109375,
+      "learning_rate": 1.7856000000000003e-05,
+      "loss": 1.8456,
+      "step": 5580
+    },
+    {
+      "epoch": 0.17888,
+      "grad_norm": 50.259490966796875,
+      "learning_rate": 1.7888000000000002e-05,
+      "loss": 1.8387,
+      "step": 5590
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 49.96138000488281,
+      "learning_rate": 1.792e-05,
+      "loss": 1.9014,
+      "step": 5600
+    },
+    {
+      "epoch": 0.17952,
+      "grad_norm": 50.213592529296875,
+      "learning_rate": 1.7952e-05,
+      "loss": 1.8341,
+      "step": 5610
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 50.193511962890625,
+      "learning_rate": 1.7984000000000003e-05,
+      "loss": 1.8239,
+      "step": 5620
+    },
+    {
+      "epoch": 0.18016,
+      "grad_norm": 52.21843338012695,
+      "learning_rate": 1.8016e-05,
+      "loss": 1.8507,
+      "step": 5630
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 51.92977523803711,
+      "learning_rate": 1.8048e-05,
+      "loss": 1.8356,
+      "step": 5640
+    },
+    {
+      "epoch": 0.1808,
+      "grad_norm": 52.1923713684082,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 1.8525,
+      "step": 5650
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 49.51137161254883,
+      "learning_rate": 1.8112000000000002e-05,
+      "loss": 1.8448,
+      "step": 5660
+    },
+    {
+      "epoch": 0.18144,
+      "grad_norm": 50.198326110839844,
+      "learning_rate": 1.8144e-05,
+      "loss": 1.8553,
+      "step": 5670
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 52.188941955566406,
+      "learning_rate": 1.8176000000000004e-05,
+      "loss": 1.8113,
+      "step": 5680
+    },
+    {
+      "epoch": 0.18208,
+      "grad_norm": 50.63345718383789,
+      "learning_rate": 1.8208000000000003e-05,
+      "loss": 1.8791,
+      "step": 5690
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 50.17583465576172,
+      "learning_rate": 1.824e-05,
+      "loss": 1.8235,
+      "step": 5700
+    },
+    {
+      "epoch": 0.18272,
+      "grad_norm": 51.316585540771484,
+      "learning_rate": 1.8272e-05,
+      "loss": 1.8119,
+      "step": 5710
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 51.15176010131836,
+      "learning_rate": 1.8304000000000003e-05,
+      "loss": 1.8495,
+      "step": 5720
+    },
+    {
+      "epoch": 0.18336,
+      "grad_norm": 48.695106506347656,
+      "learning_rate": 1.8336000000000002e-05,
+      "loss": 1.8428,
+      "step": 5730
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 51.550941467285156,
+      "learning_rate": 1.8368e-05,
+      "loss": 1.8434,
+      "step": 5740
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 51.58259582519531,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.8219,
+      "step": 5750
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 51.678165435791016,
+      "learning_rate": 1.8432000000000002e-05,
+      "loss": 1.8378,
+      "step": 5760
+    },
+    {
+      "epoch": 0.18464,
+      "grad_norm": 51.844886779785156,
+      "learning_rate": 1.8464e-05,
+      "loss": 1.8615,
+      "step": 5770
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 51.19352340698242,
+      "learning_rate": 1.8496e-05,
+      "loss": 1.8592,
+      "step": 5780
+    },
+    {
+      "epoch": 0.18528,
+      "grad_norm": 50.802066802978516,
+      "learning_rate": 1.8528000000000003e-05,
+      "loss": 1.8411,
+      "step": 5790
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 49.88358688354492,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 1.8476,
+      "step": 5800
+    },
+    {
+      "epoch": 0.18592,
+      "grad_norm": 50.59362030029297,
+      "learning_rate": 1.8592e-05,
+      "loss": 1.8264,
+      "step": 5810
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 51.28804016113281,
+      "learning_rate": 1.8624000000000003e-05,
+      "loss": 1.8364,
+      "step": 5820
+    },
+    {
+      "epoch": 0.18656,
+      "grad_norm": 50.7144660949707,
+      "learning_rate": 1.8656000000000002e-05,
+      "loss": 1.8469,
+      "step": 5830
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 51.3600959777832,
+      "learning_rate": 1.8688e-05,
+      "loss": 1.8634,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1872,
+      "grad_norm": 49.02365493774414,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 1.8499,
+      "step": 5850
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 51.212425231933594,
+      "learning_rate": 1.8752000000000003e-05,
+      "loss": 1.8267,
+      "step": 5860
+    },
+    {
+      "epoch": 0.18784,
+      "grad_norm": 50.525779724121094,
+      "learning_rate": 1.8784000000000002e-05,
+      "loss": 1.8386,
+      "step": 5870
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 51.337127685546875,
+      "learning_rate": 1.8816e-05,
+      "loss": 1.8312,
+      "step": 5880
+    },
+    {
+      "epoch": 0.18848,
+      "grad_norm": 50.37173080444336,
+      "learning_rate": 1.8848000000000003e-05,
+      "loss": 1.8558,
+      "step": 5890
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 50.84103012084961,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 1.8531,
+      "step": 5900
+    },
+    {
+      "epoch": 0.18912,
+      "grad_norm": 54.04008865356445,
+      "learning_rate": 1.8912e-05,
+      "loss": 1.8471,
+      "step": 5910
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 51.854923248291016,
+      "learning_rate": 1.8944000000000004e-05,
+      "loss": 1.827,
+      "step": 5920
+    },
+    {
+      "epoch": 0.18976,
+      "grad_norm": 51.3617057800293,
+      "learning_rate": 1.8976000000000003e-05,
+      "loss": 1.8358,
+      "step": 5930
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 49.379112243652344,
+      "learning_rate": 1.9008e-05,
+      "loss": 1.8451,
+      "step": 5940
+    },
+    {
+      "epoch": 0.1904,
+      "grad_norm": 51.578125,
+      "learning_rate": 1.904e-05,
+      "loss": 1.8311,
+      "step": 5950
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 49.91752624511719,
+      "learning_rate": 1.9072000000000003e-05,
+      "loss": 1.8535,
+      "step": 5960
+    },
+    {
+      "epoch": 0.19104,
+      "grad_norm": 48.54228210449219,
+      "learning_rate": 1.9104000000000002e-05,
+      "loss": 1.8538,
+      "step": 5970
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 50.564903259277344,
+      "learning_rate": 1.9136e-05,
+      "loss": 1.8037,
+      "step": 5980
+    },
+    {
+      "epoch": 0.19168,
+      "grad_norm": 51.50353240966797,
+      "learning_rate": 1.9168000000000004e-05,
+      "loss": 1.8539,
+      "step": 5990
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 47.665870666503906,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.8266,
+      "step": 6000
+    },
+    {
+      "epoch": 0.19232,
+      "grad_norm": 51.77862548828125,
+      "learning_rate": 1.9232e-05,
+      "loss": 1.824,
+      "step": 6010
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 48.47139358520508,
+      "learning_rate": 1.9264e-05,
+      "loss": 1.8496,
+      "step": 6020
+    },
+    {
+      "epoch": 0.19296,
+      "grad_norm": 50.459163665771484,
+      "learning_rate": 1.9296000000000003e-05,
+      "loss": 1.86,
+      "step": 6030
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 48.988704681396484,
+      "learning_rate": 1.9328000000000002e-05,
+      "loss": 1.8439,
+      "step": 6040
+    },
+    {
+      "epoch": 0.1936,
+      "grad_norm": 50.94102478027344,
+      "learning_rate": 1.936e-05,
+      "loss": 1.8203,
+      "step": 6050
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 49.791561126708984,
+      "learning_rate": 1.9392000000000003e-05,
+      "loss": 1.8114,
+      "step": 6060
+    },
+    {
+      "epoch": 0.19424,
+      "grad_norm": 52.81388473510742,
+      "learning_rate": 1.9424e-05,
+      "loss": 1.8268,
+      "step": 6070
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 50.54151153564453,
+      "learning_rate": 1.9456e-05,
+      "loss": 1.8154,
+      "step": 6080
+    },
+    {
+      "epoch": 0.19488,
+      "grad_norm": 50.21413040161133,
+      "learning_rate": 1.9488000000000004e-05,
+      "loss": 1.8101,
+      "step": 6090
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 49.325809478759766,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.8272,
+      "step": 6100
+    },
+    {
+      "epoch": 0.19552,
+      "grad_norm": 52.05915069580078,
+      "learning_rate": 1.9552000000000002e-05,
+      "loss": 1.8456,
+      "step": 6110
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 49.70790481567383,
+      "learning_rate": 1.9584e-05,
+      "loss": 1.8197,
+      "step": 6120
+    },
+    {
+      "epoch": 0.19616,
+      "grad_norm": 48.50461959838867,
+      "learning_rate": 1.9616000000000003e-05,
+      "loss": 1.8252,
+      "step": 6130
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 50.48190689086914,
+      "learning_rate": 1.9648000000000002e-05,
+      "loss": 1.8315,
+      "step": 6140
+    },
+    {
+      "epoch": 0.1968,
+      "grad_norm": 48.6129264831543,
+      "learning_rate": 1.968e-05,
+      "loss": 1.837,
+      "step": 6150
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 50.71052169799805,
+      "learning_rate": 1.9712000000000004e-05,
+      "loss": 1.8467,
+      "step": 6160
+    },
+    {
+      "epoch": 0.19744,
+      "grad_norm": 46.237510681152344,
+      "learning_rate": 1.9744e-05,
+      "loss": 1.7948,
+      "step": 6170
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 49.22959518432617,
+      "learning_rate": 1.9776000000000002e-05,
+      "loss": 1.8117,
+      "step": 6180
+    },
+    {
+      "epoch": 0.19808,
+      "grad_norm": 49.88234329223633,
+      "learning_rate": 1.9808e-05,
+      "loss": 1.8186,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 50.502845764160156,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 1.8444,
+      "step": 6200
+    },
+    {
+      "epoch": 0.19872,
+      "grad_norm": 49.359500885009766,
+      "learning_rate": 1.9872000000000002e-05,
+      "loss": 1.8415,
+      "step": 6210
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 48.598777770996094,
+      "learning_rate": 1.9904e-05,
+      "loss": 1.8026,
+      "step": 6220
+    },
+    {
+      "epoch": 0.19936,
+      "grad_norm": 50.7224006652832,
+      "learning_rate": 1.9936000000000004e-05,
+      "loss": 1.8158,
+      "step": 6230
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 49.399681091308594,
+      "learning_rate": 1.9968e-05,
+      "loss": 1.848,
+      "step": 6240
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 48.65672302246094,
+      "learning_rate": 2e-05,
+      "loss": 1.8308,
+      "step": 6250
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 51.61461639404297,
+      "learning_rate": 1.9996444444444446e-05,
+      "loss": 1.8579,
+      "step": 6260
+    },
+    {
+      "epoch": 0.20064,
+      "grad_norm": 49.613704681396484,
+      "learning_rate": 1.999288888888889e-05,
+      "loss": 1.8326,
+      "step": 6270
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 50.30833435058594,
+      "learning_rate": 1.9989333333333335e-05,
+      "loss": 1.82,
+      "step": 6280
+    },
+    {
+      "epoch": 0.20128,
+      "grad_norm": 50.83747482299805,
+      "learning_rate": 1.998577777777778e-05,
+      "loss": 1.8319,
+      "step": 6290
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 49.3170051574707,
+      "learning_rate": 1.9982222222222224e-05,
+      "loss": 1.8233,
+      "step": 6300
+    },
+    {
+      "epoch": 0.20192,
+      "grad_norm": 49.773380279541016,
+      "learning_rate": 1.997866666666667e-05,
+      "loss": 1.8443,
+      "step": 6310
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 49.04866027832031,
+      "learning_rate": 1.997511111111111e-05,
+      "loss": 1.8392,
+      "step": 6320
+    },
+    {
+      "epoch": 0.20256,
+      "grad_norm": 50.339622497558594,
+      "learning_rate": 1.9971555555555558e-05,
+      "loss": 1.8446,
+      "step": 6330
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 52.96565628051758,
+      "learning_rate": 1.9968e-05,
+      "loss": 1.883,
+      "step": 6340
+    },
+    {
+      "epoch": 0.2032,
+      "grad_norm": 48.281497955322266,
+      "learning_rate": 1.9964444444444447e-05,
+      "loss": 1.81,
+      "step": 6350
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 49.1937255859375,
+      "learning_rate": 1.996088888888889e-05,
+      "loss": 1.8321,
+      "step": 6360
+    },
+    {
+      "epoch": 0.20384,
+      "grad_norm": 51.74014663696289,
+      "learning_rate": 1.9957333333333336e-05,
+      "loss": 1.7919,
+      "step": 6370
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 48.33229064941406,
+      "learning_rate": 1.9953777777777777e-05,
+      "loss": 1.8402,
+      "step": 6380
+    },
+    {
+      "epoch": 0.20448,
+      "grad_norm": 51.9826774597168,
+      "learning_rate": 1.9950222222222225e-05,
+      "loss": 1.8295,
+      "step": 6390
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 48.660789489746094,
+      "learning_rate": 1.9946666666666667e-05,
+      "loss": 1.8047,
+      "step": 6400
+    },
+    {
+      "epoch": 0.20512,
+      "grad_norm": 49.63129806518555,
+      "learning_rate": 1.9943111111111114e-05,
+      "loss": 1.8117,
+      "step": 6410
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 51.93993377685547,
+      "learning_rate": 1.9939555555555556e-05,
+      "loss": 1.8335,
+      "step": 6420
+    },
+    {
+      "epoch": 0.20576,
+      "grad_norm": 50.61872482299805,
+      "learning_rate": 1.9936000000000004e-05,
+      "loss": 1.8611,
+      "step": 6430
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 51.35618591308594,
+      "learning_rate": 1.9932444444444445e-05,
+      "loss": 1.8676,
+      "step": 6440
+    },
+    {
+      "epoch": 0.2064,
+      "grad_norm": 53.216251373291016,
+      "learning_rate": 1.992888888888889e-05,
+      "loss": 1.8581,
+      "step": 6450
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 51.05562973022461,
+      "learning_rate": 1.9925333333333334e-05,
+      "loss": 1.8037,
+      "step": 6460
+    },
+    {
+      "epoch": 0.20704,
+      "grad_norm": 48.82215118408203,
+      "learning_rate": 1.992177777777778e-05,
+      "loss": 1.8448,
+      "step": 6470
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 52.66543960571289,
+      "learning_rate": 1.9918222222222223e-05,
+      "loss": 1.8332,
+      "step": 6480
+    },
+    {
+      "epoch": 0.20768,
+      "grad_norm": 49.5877799987793,
+      "learning_rate": 1.9914666666666668e-05,
+      "loss": 1.8237,
+      "step": 6490
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 47.73166275024414,
+      "learning_rate": 1.9911111111111112e-05,
+      "loss": 1.8148,
+      "step": 6500
+    },
+    {
+      "epoch": 0.20832,
+      "grad_norm": 47.3873405456543,
+      "learning_rate": 1.9907555555555557e-05,
+      "loss": 1.8299,
+      "step": 6510
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 49.92698669433594,
+      "learning_rate": 1.9904e-05,
+      "loss": 1.8304,
+      "step": 6520
+    },
+    {
+      "epoch": 0.20896,
+      "grad_norm": 48.46565628051758,
+      "learning_rate": 1.9900444444444446e-05,
+      "loss": 1.8439,
+      "step": 6530
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 50.39113998413086,
+      "learning_rate": 1.989688888888889e-05,
+      "loss": 1.8385,
+      "step": 6540
+    },
+    {
+      "epoch": 0.2096,
+      "grad_norm": 46.97870635986328,
+      "learning_rate": 1.9893333333333335e-05,
+      "loss": 1.8522,
+      "step": 6550
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 52.359954833984375,
+      "learning_rate": 1.988977777777778e-05,
+      "loss": 1.8461,
+      "step": 6560
+    },
+    {
+      "epoch": 0.21024,
+      "grad_norm": 50.89912796020508,
+      "learning_rate": 1.9886222222222224e-05,
+      "loss": 1.8255,
+      "step": 6570
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 49.93498992919922,
+      "learning_rate": 1.988266666666667e-05,
+      "loss": 1.807,
+      "step": 6580
+    },
+    {
+      "epoch": 0.21088,
+      "grad_norm": 51.63752746582031,
+      "learning_rate": 1.9879111111111113e-05,
+      "loss": 1.8633,
+      "step": 6590
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 48.523582458496094,
+      "learning_rate": 1.9875555555555558e-05,
+      "loss": 1.8217,
+      "step": 6600
+    },
+    {
+      "epoch": 0.21152,
+      "grad_norm": 48.10843276977539,
+      "learning_rate": 1.9872000000000002e-05,
+      "loss": 1.8299,
+      "step": 6610
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 49.58185577392578,
+      "learning_rate": 1.9868444444444447e-05,
+      "loss": 1.8587,
+      "step": 6620
+    },
+    {
+      "epoch": 0.21216,
+      "grad_norm": 48.92919158935547,
+      "learning_rate": 1.986488888888889e-05,
+      "loss": 1.8477,
+      "step": 6630
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 51.368221282958984,
+      "learning_rate": 1.9861333333333336e-05,
+      "loss": 1.823,
+      "step": 6640
+    },
+    {
+      "epoch": 0.2128,
+      "grad_norm": 50.350975036621094,
+      "learning_rate": 1.985777777777778e-05,
+      "loss": 1.8338,
+      "step": 6650
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 48.929107666015625,
+      "learning_rate": 1.9854222222222225e-05,
+      "loss": 1.8161,
+      "step": 6660
+    },
+    {
+      "epoch": 0.21344,
+      "grad_norm": 49.51737594604492,
+      "learning_rate": 1.985066666666667e-05,
+      "loss": 1.8434,
+      "step": 6670
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 47.257171630859375,
+      "learning_rate": 1.9847111111111114e-05,
+      "loss": 1.8379,
+      "step": 6680
+    },
+    {
+      "epoch": 0.21408,
+      "grad_norm": 47.63740921020508,
+      "learning_rate": 1.984355555555556e-05,
+      "loss": 1.8138,
+      "step": 6690
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 48.09037780761719,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 1.8487,
+      "step": 6700
+    },
+    {
+      "epoch": 0.21472,
+      "grad_norm": 52.047420501708984,
+      "learning_rate": 1.9836444444444448e-05,
+      "loss": 1.8312,
+      "step": 6710
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 50.519474029541016,
+      "learning_rate": 1.983288888888889e-05,
+      "loss": 1.8581,
+      "step": 6720
+    },
+    {
+      "epoch": 0.21536,
+      "grad_norm": 49.650238037109375,
+      "learning_rate": 1.9829333333333337e-05,
+      "loss": 1.8074,
+      "step": 6730
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 49.53740310668945,
+      "learning_rate": 1.9825777777777778e-05,
+      "loss": 1.8089,
+      "step": 6740
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 50.6927604675293,
+      "learning_rate": 1.9822222222222226e-05,
+      "loss": 1.8014,
+      "step": 6750
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 54.616119384765625,
+      "learning_rate": 1.9818666666666667e-05,
+      "loss": 1.7996,
+      "step": 6760
+    },
+    {
+      "epoch": 0.21664,
+      "grad_norm": 47.97896957397461,
+      "learning_rate": 1.981511111111111e-05,
+      "loss": 1.8241,
+      "step": 6770
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 49.36483383178711,
+      "learning_rate": 1.9811555555555556e-05,
+      "loss": 1.7999,
+      "step": 6780
+    },
+    {
+      "epoch": 0.21728,
+      "grad_norm": 48.21449279785156,
+      "learning_rate": 1.9808e-05,
+      "loss": 1.8411,
+      "step": 6790
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 52.211509704589844,
+      "learning_rate": 1.9804444444444445e-05,
+      "loss": 1.8388,
+      "step": 6800
+    },
+    {
+      "epoch": 0.21792,
+      "grad_norm": 49.41231918334961,
+      "learning_rate": 1.980088888888889e-05,
+      "loss": 1.8171,
+      "step": 6810
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 50.267425537109375,
+      "learning_rate": 1.9797333333333334e-05,
+      "loss": 1.8346,
+      "step": 6820
+    },
+    {
+      "epoch": 0.21856,
+      "grad_norm": 47.68706512451172,
+      "learning_rate": 1.979377777777778e-05,
+      "loss": 1.8136,
+      "step": 6830
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 48.38620376586914,
+      "learning_rate": 1.9790222222222224e-05,
+      "loss": 1.8094,
+      "step": 6840
+    },
+    {
+      "epoch": 0.2192,
+      "grad_norm": 49.339664459228516,
+      "learning_rate": 1.9786666666666668e-05,
+      "loss": 1.8194,
+      "step": 6850
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 51.0460319519043,
+      "learning_rate": 1.9783111111111113e-05,
+      "loss": 1.7893,
+      "step": 6860
+    },
+    {
+      "epoch": 0.21984,
+      "grad_norm": 47.96543502807617,
+      "learning_rate": 1.9779555555555557e-05,
+      "loss": 1.7939,
+      "step": 6870
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 51.15391540527344,
+      "learning_rate": 1.9776000000000002e-05,
+      "loss": 1.8204,
+      "step": 6880
+    },
+    {
+      "epoch": 0.22048,
+      "grad_norm": 47.516658782958984,
+      "learning_rate": 1.9772444444444446e-05,
+      "loss": 1.8251,
+      "step": 6890
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 46.96601486206055,
+      "learning_rate": 1.976888888888889e-05,
+      "loss": 1.7885,
+      "step": 6900
+    },
+    {
+      "epoch": 0.22112,
+      "grad_norm": 50.02362060546875,
+      "learning_rate": 1.9765333333333335e-05,
+      "loss": 1.8253,
+      "step": 6910
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 49.389183044433594,
+      "learning_rate": 1.976177777777778e-05,
+      "loss": 1.808,
+      "step": 6920
+    },
+    {
+      "epoch": 0.22176,
+      "grad_norm": 50.543033599853516,
+      "learning_rate": 1.9758222222222225e-05,
+      "loss": 1.8429,
+      "step": 6930
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 50.78889465332031,
+      "learning_rate": 1.975466666666667e-05,
+      "loss": 1.8296,
+      "step": 6940
+    },
+    {
+      "epoch": 0.2224,
+      "grad_norm": 46.22223663330078,
+      "learning_rate": 1.9751111111111114e-05,
+      "loss": 1.8124,
+      "step": 6950
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 49.157684326171875,
+      "learning_rate": 1.9747555555555558e-05,
+      "loss": 1.8049,
+      "step": 6960
+    },
+    {
+      "epoch": 0.22304,
+      "grad_norm": 48.17796325683594,
+      "learning_rate": 1.9744e-05,
+      "loss": 1.7927,
+      "step": 6970
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 47.83882141113281,
+      "learning_rate": 1.9740444444444447e-05,
+      "loss": 1.7881,
+      "step": 6980
+    },
+    {
+      "epoch": 0.22368,
+      "grad_norm": 48.507755279541016,
+      "learning_rate": 1.973688888888889e-05,
+      "loss": 1.8145,
+      "step": 6990
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 48.463340759277344,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.8137,
+      "step": 7000
+    },
+    {
+      "epoch": 0.22432,
+      "grad_norm": 47.04343032836914,
+      "learning_rate": 1.9729777777777778e-05,
+      "loss": 1.8385,
+      "step": 7010
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 48.241920471191406,
+      "learning_rate": 1.9726222222222226e-05,
+      "loss": 1.7984,
+      "step": 7020
+    },
+    {
+      "epoch": 0.22496,
+      "grad_norm": 50.49192428588867,
+      "learning_rate": 1.9722666666666667e-05,
+      "loss": 1.8091,
+      "step": 7030
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 53.93669128417969,
+      "learning_rate": 1.9719111111111115e-05,
+      "loss": 1.8452,
+      "step": 7040
+    },
+    {
+      "epoch": 0.2256,
+      "grad_norm": 49.90575408935547,
+      "learning_rate": 1.9715555555555556e-05,
+      "loss": 1.8297,
+      "step": 7050
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 48.688812255859375,
+      "learning_rate": 1.9712000000000004e-05,
+      "loss": 1.8348,
+      "step": 7060
+    },
+    {
+      "epoch": 0.22624,
+      "grad_norm": 52.05872344970703,
+      "learning_rate": 1.9708444444444445e-05,
+      "loss": 1.8451,
+      "step": 7070
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 47.86528778076172,
+      "learning_rate": 1.9704888888888893e-05,
+      "loss": 1.8027,
+      "step": 7080
+    },
+    {
+      "epoch": 0.22688,
+      "grad_norm": 48.741554260253906,
+      "learning_rate": 1.9701333333333334e-05,
+      "loss": 1.8219,
+      "step": 7090
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 51.6088752746582,
+      "learning_rate": 1.969777777777778e-05,
+      "loss": 1.827,
+      "step": 7100
+    },
+    {
+      "epoch": 0.22752,
+      "grad_norm": 49.774803161621094,
+      "learning_rate": 1.9694222222222223e-05,
+      "loss": 1.8096,
+      "step": 7110
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 49.89011001586914,
+      "learning_rate": 1.9690666666666668e-05,
+      "loss": 1.8065,
+      "step": 7120
+    },
+    {
+      "epoch": 0.22816,
+      "grad_norm": 48.9777717590332,
+      "learning_rate": 1.9687111111111112e-05,
+      "loss": 1.8163,
+      "step": 7130
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 48.017555236816406,
+      "learning_rate": 1.9683555555555557e-05,
+      "loss": 1.8264,
+      "step": 7140
+    },
+    {
+      "epoch": 0.2288,
+      "grad_norm": 48.99272155761719,
+      "learning_rate": 1.968e-05,
+      "loss": 1.82,
+      "step": 7150
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 49.58944320678711,
+      "learning_rate": 1.9676444444444446e-05,
+      "loss": 1.8333,
+      "step": 7160
+    },
+    {
+      "epoch": 0.22944,
+      "grad_norm": 50.1278190612793,
+      "learning_rate": 1.967288888888889e-05,
+      "loss": 1.8064,
+      "step": 7170
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 51.582210540771484,
+      "learning_rate": 1.9669333333333335e-05,
+      "loss": 1.8158,
+      "step": 7180
+    },
+    {
+      "epoch": 0.23008,
+      "grad_norm": 49.89718246459961,
+      "learning_rate": 1.966577777777778e-05,
+      "loss": 1.8372,
+      "step": 7190
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 49.177001953125,
+      "learning_rate": 1.9662222222222224e-05,
+      "loss": 1.8397,
+      "step": 7200
+    },
+    {
+      "epoch": 0.23072,
+      "grad_norm": 47.591617584228516,
+      "learning_rate": 1.965866666666667e-05,
+      "loss": 1.799,
+      "step": 7210
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 49.274208068847656,
+      "learning_rate": 1.9655111111111113e-05,
+      "loss": 1.8342,
+      "step": 7220
+    },
+    {
+      "epoch": 0.23136,
+      "grad_norm": 47.56116485595703,
+      "learning_rate": 1.9651555555555558e-05,
+      "loss": 1.8098,
+      "step": 7230
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 46.61338424682617,
+      "learning_rate": 1.9648000000000002e-05,
+      "loss": 1.8044,
+      "step": 7240
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 51.38720703125,
+      "learning_rate": 1.9644444444444447e-05,
+      "loss": 1.8376,
+      "step": 7250
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 48.6185417175293,
+      "learning_rate": 1.964088888888889e-05,
+      "loss": 1.8045,
+      "step": 7260
+    },
+    {
+      "epoch": 0.23264,
+      "grad_norm": 46.933162689208984,
+      "learning_rate": 1.9637333333333336e-05,
+      "loss": 1.8378,
+      "step": 7270
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 49.86311721801758,
+      "learning_rate": 1.963377777777778e-05,
+      "loss": 1.8234,
+      "step": 7280
+    },
+    {
+      "epoch": 0.23328,
+      "grad_norm": 50.35856246948242,
+      "learning_rate": 1.9630222222222225e-05,
+      "loss": 1.7944,
+      "step": 7290
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 47.689151763916016,
+      "learning_rate": 1.9626666666666666e-05,
+      "loss": 1.8469,
+      "step": 7300
+    },
+    {
+      "epoch": 0.23392,
+      "grad_norm": 46.43436813354492,
+      "learning_rate": 1.9623111111111114e-05,
+      "loss": 1.8313,
+      "step": 7310
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 49.50898361206055,
+      "learning_rate": 1.9619555555555555e-05,
+      "loss": 1.8028,
+      "step": 7320
+    },
+    {
+      "epoch": 0.23456,
+      "grad_norm": 52.28144454956055,
+      "learning_rate": 1.9616000000000003e-05,
+      "loss": 1.8171,
+      "step": 7330
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 49.656715393066406,
+      "learning_rate": 1.9612444444444444e-05,
+      "loss": 1.8042,
+      "step": 7340
+    },
+    {
+      "epoch": 0.2352,
+      "grad_norm": 49.44095993041992,
+      "learning_rate": 1.960888888888889e-05,
+      "loss": 1.814,
+      "step": 7350
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 48.879371643066406,
+      "learning_rate": 1.9605333333333334e-05,
+      "loss": 1.8068,
+      "step": 7360
+    },
+    {
+      "epoch": 0.23584,
+      "grad_norm": 50.464385986328125,
+      "learning_rate": 1.9601777777777778e-05,
+      "loss": 1.8314,
+      "step": 7370
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 46.931087493896484,
+      "learning_rate": 1.9598222222222223e-05,
+      "loss": 1.817,
+      "step": 7380
+    },
+    {
+      "epoch": 0.23648,
+      "grad_norm": 50.16783905029297,
+      "learning_rate": 1.9594666666666667e-05,
+      "loss": 1.8205,
+      "step": 7390
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 48.354862213134766,
+      "learning_rate": 1.9591111111111112e-05,
+      "loss": 1.7917,
+      "step": 7400
+    },
+    {
+      "epoch": 0.23712,
+      "grad_norm": 48.41815185546875,
+      "learning_rate": 1.9587555555555556e-05,
+      "loss": 1.7994,
+      "step": 7410
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 47.2100715637207,
+      "learning_rate": 1.9584e-05,
+      "loss": 1.8207,
+      "step": 7420
+    },
+    {
+      "epoch": 0.23776,
+      "grad_norm": 50.41408920288086,
+      "learning_rate": 1.9580444444444445e-05,
+      "loss": 1.8124,
+      "step": 7430
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 49.47113037109375,
+      "learning_rate": 1.957688888888889e-05,
+      "loss": 1.7621,
+      "step": 7440
+    },
+    {
+      "epoch": 0.2384,
+      "grad_norm": 49.08128356933594,
+      "learning_rate": 1.9573333333333335e-05,
+      "loss": 1.8061,
+      "step": 7450
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 50.74384307861328,
+      "learning_rate": 1.956977777777778e-05,
+      "loss": 1.8424,
+      "step": 7460
+    },
+    {
+      "epoch": 0.23904,
+      "grad_norm": 52.481388092041016,
+      "learning_rate": 1.9566222222222224e-05,
+      "loss": 1.8255,
+      "step": 7470
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 49.44757080078125,
+      "learning_rate": 1.9562666666666668e-05,
+      "loss": 1.7856,
+      "step": 7480
+    },
+    {
+      "epoch": 0.23968,
+      "grad_norm": 48.70913314819336,
+      "learning_rate": 1.9559111111111113e-05,
+      "loss": 1.808,
+      "step": 7490
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 50.43172836303711,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 1.8109,
+      "step": 7500
+    },
+    {
+      "epoch": 0.24032,
+      "grad_norm": 50.79537582397461,
+      "learning_rate": 1.9552000000000002e-05,
+      "loss": 1.7816,
+      "step": 7510
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 47.54401397705078,
+      "learning_rate": 1.9548444444444446e-05,
+      "loss": 1.8125,
+      "step": 7520
+    },
+    {
+      "epoch": 0.24096,
+      "grad_norm": 48.10938262939453,
+      "learning_rate": 1.954488888888889e-05,
+      "loss": 1.779,
+      "step": 7530
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 49.07471466064453,
+      "learning_rate": 1.9541333333333336e-05,
+      "loss": 1.7964,
+      "step": 7540
+    },
+    {
+      "epoch": 0.2416,
+      "grad_norm": 49.26071548461914,
+      "learning_rate": 1.953777777777778e-05,
+      "loss": 1.8281,
+      "step": 7550
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 47.97166061401367,
+      "learning_rate": 1.9534222222222225e-05,
+      "loss": 1.8303,
+      "step": 7560
+    },
+    {
+      "epoch": 0.24224,
+      "grad_norm": 47.929691314697266,
+      "learning_rate": 1.953066666666667e-05,
+      "loss": 1.8316,
+      "step": 7570
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 50.565853118896484,
+      "learning_rate": 1.9527111111111114e-05,
+      "loss": 1.8414,
+      "step": 7580
+    },
+    {
+      "epoch": 0.24288,
+      "grad_norm": 48.56721878051758,
+      "learning_rate": 1.9523555555555558e-05,
+      "loss": 1.798,
+      "step": 7590
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 48.066680908203125,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 1.8232,
+      "step": 7600
+    },
+    {
+      "epoch": 0.24352,
+      "grad_norm": 48.67830276489258,
+      "learning_rate": 1.9516444444444447e-05,
+      "loss": 1.8281,
+      "step": 7610
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 47.78409194946289,
+      "learning_rate": 1.951288888888889e-05,
+      "loss": 1.8156,
+      "step": 7620
+    },
+    {
+      "epoch": 0.24416,
+      "grad_norm": 46.79526138305664,
+      "learning_rate": 1.9509333333333337e-05,
+      "loss": 1.7844,
+      "step": 7630
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 47.15976333618164,
+      "learning_rate": 1.9505777777777778e-05,
+      "loss": 1.8172,
+      "step": 7640
+    },
+    {
+      "epoch": 0.2448,
+      "grad_norm": 50.70997619628906,
+      "learning_rate": 1.9502222222222226e-05,
+      "loss": 1.809,
+      "step": 7650
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 49.29851150512695,
+      "learning_rate": 1.9498666666666667e-05,
+      "loss": 1.7748,
+      "step": 7660
+    },
+    {
+      "epoch": 0.24544,
+      "grad_norm": 49.05281448364258,
+      "learning_rate": 1.9495111111111115e-05,
+      "loss": 1.8165,
+      "step": 7670
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 48.29036331176758,
+      "learning_rate": 1.9491555555555556e-05,
+      "loss": 1.7774,
+      "step": 7680
+    },
+    {
+      "epoch": 0.24608,
+      "grad_norm": 50.523536682128906,
+      "learning_rate": 1.9488000000000004e-05,
+      "loss": 1.793,
+      "step": 7690
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 48.234649658203125,
+      "learning_rate": 1.9484444444444445e-05,
+      "loss": 1.7864,
+      "step": 7700
+    },
+    {
+      "epoch": 0.24672,
+      "grad_norm": 49.41721725463867,
+      "learning_rate": 1.9480888888888893e-05,
+      "loss": 1.8022,
+      "step": 7710
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 50.055355072021484,
+      "learning_rate": 1.9477333333333334e-05,
+      "loss": 1.792,
+      "step": 7720
+    },
+    {
+      "epoch": 0.24736,
+      "grad_norm": 50.13814926147461,
+      "learning_rate": 1.947377777777778e-05,
+      "loss": 1.7929,
+      "step": 7730
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 47.56300735473633,
+      "learning_rate": 1.9470222222222223e-05,
+      "loss": 1.7964,
+      "step": 7740
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 51.54580307006836,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.7946,
+      "step": 7750
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 47.7603759765625,
+      "learning_rate": 1.9463111111111112e-05,
+      "loss": 1.803,
+      "step": 7760
+    },
+    {
+      "epoch": 0.24864,
+      "grad_norm": 49.26568603515625,
+      "learning_rate": 1.9459555555555557e-05,
+      "loss": 1.8293,
+      "step": 7770
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 47.8105583190918,
+      "learning_rate": 1.9456e-05,
+      "loss": 1.7627,
+      "step": 7780
+    },
+    {
+      "epoch": 0.24928,
+      "grad_norm": 48.05279541015625,
+      "learning_rate": 1.9452444444444446e-05,
+      "loss": 1.7881,
+      "step": 7790
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 50.0302848815918,
+      "learning_rate": 1.944888888888889e-05,
+      "loss": 1.7889,
+      "step": 7800
+    },
+    {
+      "epoch": 0.24992,
+      "grad_norm": 49.33319854736328,
+      "learning_rate": 1.9445333333333335e-05,
+      "loss": 1.807,
+      "step": 7810
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 46.37745666503906,
+      "learning_rate": 1.944177777777778e-05,
+      "loss": 1.807,
+      "step": 7820
+    },
+    {
+      "epoch": 0.25056,
+      "grad_norm": 46.626041412353516,
+      "learning_rate": 1.9438222222222224e-05,
+      "loss": 1.8251,
+      "step": 7830
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 47.2230110168457,
+      "learning_rate": 1.943466666666667e-05,
+      "loss": 1.7891,
+      "step": 7840
+    },
+    {
+      "epoch": 0.2512,
+      "grad_norm": 46.498348236083984,
+      "learning_rate": 1.9431111111111113e-05,
+      "loss": 1.7927,
+      "step": 7850
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 47.701663970947266,
+      "learning_rate": 1.9427555555555558e-05,
+      "loss": 1.8128,
+      "step": 7860
+    },
+    {
+      "epoch": 0.25184,
+      "grad_norm": 46.88845443725586,
+      "learning_rate": 1.9424e-05,
+      "loss": 1.7962,
+      "step": 7870
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 50.29132080078125,
+      "learning_rate": 1.9420444444444447e-05,
+      "loss": 1.7985,
+      "step": 7880
+    },
+    {
+      "epoch": 0.25248,
+      "grad_norm": 48.16657257080078,
+      "learning_rate": 1.9416888888888888e-05,
+      "loss": 1.8203,
+      "step": 7890
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 50.36405944824219,
+      "learning_rate": 1.9413333333333336e-05,
+      "loss": 1.7756,
+      "step": 7900
+    },
+    {
+      "epoch": 0.25312,
+      "grad_norm": 47.91851043701172,
+      "learning_rate": 1.9409777777777777e-05,
+      "loss": 1.7991,
+      "step": 7910
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 49.88837814331055,
+      "learning_rate": 1.9406222222222225e-05,
+      "loss": 1.7981,
+      "step": 7920
+    },
+    {
+      "epoch": 0.25376,
+      "grad_norm": 49.87557601928711,
+      "learning_rate": 1.9402666666666666e-05,
+      "loss": 1.797,
+      "step": 7930
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 50.396419525146484,
+      "learning_rate": 1.9399111111111114e-05,
+      "loss": 1.8224,
+      "step": 7940
+    },
+    {
+      "epoch": 0.2544,
+      "grad_norm": 47.169822692871094,
+      "learning_rate": 1.9395555555555555e-05,
+      "loss": 1.7925,
+      "step": 7950
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 48.06186294555664,
+      "learning_rate": 1.9392000000000003e-05,
+      "loss": 1.7787,
+      "step": 7960
+    },
+    {
+      "epoch": 0.25504,
+      "grad_norm": 46.269710540771484,
+      "learning_rate": 1.9388444444444445e-05,
+      "loss": 1.7816,
+      "step": 7970
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 45.91224670410156,
+      "learning_rate": 1.9384888888888892e-05,
+      "loss": 1.8049,
+      "step": 7980
+    },
+    {
+      "epoch": 0.25568,
+      "grad_norm": 47.012298583984375,
+      "learning_rate": 1.9381333333333334e-05,
+      "loss": 1.7766,
+      "step": 7990
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 48.96611404418945,
+      "learning_rate": 1.9377777777777778e-05,
+      "loss": 1.7769,
+      "step": 8000
+    },
+    {
+      "epoch": 0.25632,
+      "grad_norm": 50.42731857299805,
+      "learning_rate": 1.9374222222222223e-05,
+      "loss": 1.8164,
+      "step": 8010
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 47.28165817260742,
+      "learning_rate": 1.9370666666666667e-05,
+      "loss": 1.7714,
+      "step": 8020
+    },
+    {
+      "epoch": 0.25696,
+      "grad_norm": 47.277191162109375,
+      "learning_rate": 1.9367111111111112e-05,
+      "loss": 1.8067,
+      "step": 8030
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 54.00135040283203,
+      "learning_rate": 1.9363555555555556e-05,
+      "loss": 1.8454,
+      "step": 8040
+    },
+    {
+      "epoch": 0.2576,
+      "grad_norm": 46.93778991699219,
+      "learning_rate": 1.936e-05,
+      "loss": 1.8049,
+      "step": 8050
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 49.47587203979492,
+      "learning_rate": 1.9356444444444446e-05,
+      "loss": 1.8151,
+      "step": 8060
+    },
+    {
+      "epoch": 0.25824,
+      "grad_norm": 47.49064636230469,
+      "learning_rate": 1.935288888888889e-05,
+      "loss": 1.8169,
+      "step": 8070
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 48.66109848022461,
+      "learning_rate": 1.9349333333333335e-05,
+      "loss": 1.7851,
+      "step": 8080
+    },
+    {
+      "epoch": 0.25888,
+      "grad_norm": 46.810081481933594,
+      "learning_rate": 1.934577777777778e-05,
+      "loss": 1.8211,
+      "step": 8090
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 46.94947814941406,
+      "learning_rate": 1.9342222222222224e-05,
+      "loss": 1.8169,
+      "step": 8100
+    },
+    {
+      "epoch": 0.25952,
+      "grad_norm": 48.868526458740234,
+      "learning_rate": 1.933866666666667e-05,
+      "loss": 1.8155,
+      "step": 8110
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 47.07829666137695,
+      "learning_rate": 1.9335111111111113e-05,
+      "loss": 1.7701,
+      "step": 8120
+    },
+    {
+      "epoch": 0.26016,
+      "grad_norm": 49.09501266479492,
+      "learning_rate": 1.9331555555555557e-05,
+      "loss": 1.7843,
+      "step": 8130
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 48.02748489379883,
+      "learning_rate": 1.9328000000000002e-05,
+      "loss": 1.7955,
+      "step": 8140
+    },
+    {
+      "epoch": 0.2608,
+      "grad_norm": 49.24200439453125,
+      "learning_rate": 1.9324444444444447e-05,
+      "loss": 1.788,
+      "step": 8150
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 47.42225646972656,
+      "learning_rate": 1.932088888888889e-05,
+      "loss": 1.7825,
+      "step": 8160
+    },
+    {
+      "epoch": 0.26144,
+      "grad_norm": 47.41923141479492,
+      "learning_rate": 1.9317333333333336e-05,
+      "loss": 1.7687,
+      "step": 8170
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 47.802574157714844,
+      "learning_rate": 1.931377777777778e-05,
+      "loss": 1.7701,
+      "step": 8180
+    },
+    {
+      "epoch": 0.26208,
+      "grad_norm": 48.67831802368164,
+      "learning_rate": 1.9310222222222225e-05,
+      "loss": 1.7282,
+      "step": 8190
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 47.63313293457031,
+      "learning_rate": 1.930666666666667e-05,
+      "loss": 1.7829,
+      "step": 8200
+    },
+    {
+      "epoch": 0.26272,
+      "grad_norm": 46.64168930053711,
+      "learning_rate": 1.9303111111111114e-05,
+      "loss": 1.7832,
+      "step": 8210
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 50.76215362548828,
+      "learning_rate": 1.929955555555556e-05,
+      "loss": 1.8191,
+      "step": 8220
+    },
+    {
+      "epoch": 0.26336,
+      "grad_norm": 47.865840911865234,
+      "learning_rate": 1.9296000000000003e-05,
+      "loss": 1.7964,
+      "step": 8230
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 48.87041473388672,
+      "learning_rate": 1.9292444444444448e-05,
+      "loss": 1.763,
+      "step": 8240
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 49.547462463378906,
+      "learning_rate": 1.928888888888889e-05,
+      "loss": 1.8002,
+      "step": 8250
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 48.55093002319336,
+      "learning_rate": 1.9285333333333337e-05,
+      "loss": 1.7796,
+      "step": 8260
+    },
+    {
+      "epoch": 0.26464,
+      "grad_norm": 48.40106964111328,
+      "learning_rate": 1.9281777777777778e-05,
+      "loss": 1.8064,
+      "step": 8270
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 47.861427307128906,
+      "learning_rate": 1.9278222222222226e-05,
+      "loss": 1.7662,
+      "step": 8280
+    },
+    {
+      "epoch": 0.26528,
+      "grad_norm": 47.32961654663086,
+      "learning_rate": 1.9274666666666667e-05,
+      "loss": 1.7934,
+      "step": 8290
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 48.121803283691406,
+      "learning_rate": 1.9271111111111115e-05,
+      "loss": 1.7799,
+      "step": 8300
+    },
+    {
+      "epoch": 0.26592,
+      "grad_norm": 46.65339279174805,
+      "learning_rate": 1.9267555555555556e-05,
+      "loss": 1.75,
+      "step": 8310
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 47.114437103271484,
+      "learning_rate": 1.9264e-05,
+      "loss": 1.7605,
+      "step": 8320
+    },
+    {
+      "epoch": 0.26656,
+      "grad_norm": 49.223777770996094,
+      "learning_rate": 1.9260444444444445e-05,
+      "loss": 1.7603,
+      "step": 8330
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 47.0412483215332,
+      "learning_rate": 1.925688888888889e-05,
+      "loss": 1.7874,
+      "step": 8340
+    },
+    {
+      "epoch": 0.2672,
+      "grad_norm": 45.218345642089844,
+      "learning_rate": 1.9253333333333334e-05,
+      "loss": 1.7851,
+      "step": 8350
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 48.539791107177734,
+      "learning_rate": 1.924977777777778e-05,
+      "loss": 1.7917,
+      "step": 8360
+    },
+    {
+      "epoch": 0.26784,
+      "grad_norm": 48.42888641357422,
+      "learning_rate": 1.9246222222222223e-05,
+      "loss": 1.7962,
+      "step": 8370
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 48.56917953491211,
+      "learning_rate": 1.9242666666666668e-05,
+      "loss": 1.7911,
+      "step": 8380
+    },
+    {
+      "epoch": 0.26848,
+      "grad_norm": 47.5833625793457,
+      "learning_rate": 1.9239111111111112e-05,
+      "loss": 1.7804,
+      "step": 8390
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 48.058265686035156,
+      "learning_rate": 1.9235555555555557e-05,
+      "loss": 1.7782,
+      "step": 8400
+    },
+    {
+      "epoch": 0.26912,
+      "grad_norm": 48.20596694946289,
+      "learning_rate": 1.9232e-05,
+      "loss": 1.78,
+      "step": 8410
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 48.306278228759766,
+      "learning_rate": 1.9228444444444446e-05,
+      "loss": 1.8114,
+      "step": 8420
+    },
+    {
+      "epoch": 0.26976,
+      "grad_norm": 47.50426483154297,
+      "learning_rate": 1.922488888888889e-05,
+      "loss": 1.7793,
+      "step": 8430
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 46.97312545776367,
+      "learning_rate": 1.9221333333333335e-05,
+      "loss": 1.7618,
+      "step": 8440
+    },
+    {
+      "epoch": 0.2704,
+      "grad_norm": 46.43683624267578,
+      "learning_rate": 1.921777777777778e-05,
+      "loss": 1.7896,
+      "step": 8450
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 46.32859802246094,
+      "learning_rate": 1.9214222222222224e-05,
+      "loss": 1.8147,
+      "step": 8460
+    },
+    {
+      "epoch": 0.27104,
+      "grad_norm": 51.10436248779297,
+      "learning_rate": 1.921066666666667e-05,
+      "loss": 1.8001,
+      "step": 8470
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 47.70284652709961,
+      "learning_rate": 1.9207111111111113e-05,
+      "loss": 1.7932,
+      "step": 8480
+    },
+    {
+      "epoch": 0.27168,
+      "grad_norm": 47.11650848388672,
+      "learning_rate": 1.9203555555555558e-05,
+      "loss": 1.8136,
+      "step": 8490
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 47.973411560058594,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.7823,
+      "step": 8500
+    },
+    {
+      "epoch": 0.27232,
+      "grad_norm": 48.01387023925781,
+      "learning_rate": 1.9196444444444447e-05,
+      "loss": 1.7939,
+      "step": 8510
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 49.093467712402344,
+      "learning_rate": 1.9192888888888888e-05,
+      "loss": 1.7999,
+      "step": 8520
+    },
+    {
+      "epoch": 0.27296,
+      "grad_norm": 48.86354446411133,
+      "learning_rate": 1.9189333333333336e-05,
+      "loss": 1.7981,
+      "step": 8530
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 47.82622146606445,
+      "learning_rate": 1.9185777777777777e-05,
+      "loss": 1.8161,
+      "step": 8540
+    },
+    {
+      "epoch": 0.2736,
+      "grad_norm": 47.5350456237793,
+      "learning_rate": 1.9182222222222225e-05,
+      "loss": 1.7884,
+      "step": 8550
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 46.69621276855469,
+      "learning_rate": 1.9178666666666666e-05,
+      "loss": 1.7977,
+      "step": 8560
+    },
+    {
+      "epoch": 0.27424,
+      "grad_norm": 47.91357421875,
+      "learning_rate": 1.9175111111111114e-05,
+      "loss": 1.7925,
+      "step": 8570
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 47.39646530151367,
+      "learning_rate": 1.9171555555555556e-05,
+      "loss": 1.8325,
+      "step": 8580
+    },
+    {
+      "epoch": 0.27488,
+      "grad_norm": 48.67210006713867,
+      "learning_rate": 1.9168000000000004e-05,
+      "loss": 1.7686,
+      "step": 8590
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 47.34749984741211,
+      "learning_rate": 1.9164444444444445e-05,
+      "loss": 1.7635,
+      "step": 8600
+    },
+    {
+      "epoch": 0.27552,
+      "grad_norm": 46.572601318359375,
+      "learning_rate": 1.9160888888888893e-05,
+      "loss": 1.8069,
+      "step": 8610
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 47.41379928588867,
+      "learning_rate": 1.9157333333333334e-05,
+      "loss": 1.8034,
+      "step": 8620
+    },
+    {
+      "epoch": 0.27616,
+      "grad_norm": 49.406028747558594,
+      "learning_rate": 1.915377777777778e-05,
+      "loss": 1.7773,
+      "step": 8630
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 46.770076751708984,
+      "learning_rate": 1.9150222222222223e-05,
+      "loss": 1.7657,
+      "step": 8640
+    },
+    {
+      "epoch": 0.2768,
+      "grad_norm": 46.42891311645508,
+      "learning_rate": 1.9146666666666667e-05,
+      "loss": 1.7853,
+      "step": 8650
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 48.82978057861328,
+      "learning_rate": 1.9143111111111112e-05,
+      "loss": 1.761,
+      "step": 8660
+    },
+    {
+      "epoch": 0.27744,
+      "grad_norm": 54.949058532714844,
+      "learning_rate": 1.9139555555555557e-05,
+      "loss": 1.8219,
+      "step": 8670
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 49.66971969604492,
+      "learning_rate": 1.9136e-05,
+      "loss": 1.7843,
+      "step": 8680
+    },
+    {
+      "epoch": 0.27808,
+      "grad_norm": 51.274681091308594,
+      "learning_rate": 1.9132444444444446e-05,
+      "loss": 1.7775,
+      "step": 8690
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 52.11601638793945,
+      "learning_rate": 1.912888888888889e-05,
+      "loss": 1.7627,
+      "step": 8700
+    },
+    {
+      "epoch": 0.27872,
+      "grad_norm": 46.86934280395508,
+      "learning_rate": 1.9125333333333335e-05,
+      "loss": 1.7629,
+      "step": 8710
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 48.512325286865234,
+      "learning_rate": 1.912177777777778e-05,
+      "loss": 1.7916,
+      "step": 8720
+    },
+    {
+      "epoch": 0.27936,
+      "grad_norm": 46.8431510925293,
+      "learning_rate": 1.9118222222222224e-05,
+      "loss": 1.8014,
+      "step": 8730
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 46.54253387451172,
+      "learning_rate": 1.911466666666667e-05,
+      "loss": 1.7705,
+      "step": 8740
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 48.69770812988281,
+      "learning_rate": 1.9111111111111113e-05,
+      "loss": 1.7871,
+      "step": 8750
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 47.79988098144531,
+      "learning_rate": 1.9107555555555558e-05,
+      "loss": 1.7642,
+      "step": 8760
+    },
+    {
+      "epoch": 0.28064,
+      "grad_norm": 47.7818489074707,
+      "learning_rate": 1.9104000000000002e-05,
+      "loss": 1.8089,
+      "step": 8770
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 46.9085807800293,
+      "learning_rate": 1.9100444444444447e-05,
+      "loss": 1.8113,
+      "step": 8780
+    },
+    {
+      "epoch": 0.28128,
+      "grad_norm": 51.09097671508789,
+      "learning_rate": 1.909688888888889e-05,
+      "loss": 1.7796,
+      "step": 8790
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 45.09664535522461,
+      "learning_rate": 1.9093333333333336e-05,
+      "loss": 1.7894,
+      "step": 8800
+    },
+    {
+      "epoch": 0.28192,
+      "grad_norm": 46.47522735595703,
+      "learning_rate": 1.908977777777778e-05,
+      "loss": 1.7885,
+      "step": 8810
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 46.489280700683594,
+      "learning_rate": 1.9086222222222225e-05,
+      "loss": 1.7831,
+      "step": 8820
+    },
+    {
+      "epoch": 0.28256,
+      "grad_norm": 46.805702209472656,
+      "learning_rate": 1.908266666666667e-05,
+      "loss": 1.7966,
+      "step": 8830
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 48.01674270629883,
+      "learning_rate": 1.9079111111111114e-05,
+      "loss": 1.7632,
+      "step": 8840
+    },
+    {
+      "epoch": 0.2832,
+      "grad_norm": 48.38603973388672,
+      "learning_rate": 1.9075555555555555e-05,
+      "loss": 1.7791,
+      "step": 8850
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 48.47395706176758,
+      "learning_rate": 1.9072000000000003e-05,
+      "loss": 1.8077,
+      "step": 8860
+    },
+    {
+      "epoch": 0.28384,
+      "grad_norm": 48.59999465942383,
+      "learning_rate": 1.9068444444444444e-05,
+      "loss": 1.769,
+      "step": 8870
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 45.61606979370117,
+      "learning_rate": 1.9064888888888892e-05,
+      "loss": 1.7588,
+      "step": 8880
+    },
+    {
+      "epoch": 0.28448,
+      "grad_norm": 47.03929901123047,
+      "learning_rate": 1.9061333333333333e-05,
+      "loss": 1.7967,
+      "step": 8890
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 50.322166442871094,
+      "learning_rate": 1.9057777777777778e-05,
+      "loss": 1.7673,
+      "step": 8900
+    },
+    {
+      "epoch": 0.28512,
+      "grad_norm": 47.95815658569336,
+      "learning_rate": 1.9054222222222222e-05,
+      "loss": 1.8022,
+      "step": 8910
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 46.3283576965332,
+      "learning_rate": 1.9050666666666667e-05,
+      "loss": 1.7782,
+      "step": 8920
+    },
+    {
+      "epoch": 0.28576,
+      "grad_norm": 50.623783111572266,
+      "learning_rate": 1.904711111111111e-05,
+      "loss": 1.7955,
+      "step": 8930
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 47.69876480102539,
+      "learning_rate": 1.9043555555555556e-05,
+      "loss": 1.7749,
+      "step": 8940
+    },
+    {
+      "epoch": 0.2864,
+      "grad_norm": 46.81853103637695,
+      "learning_rate": 1.904e-05,
+      "loss": 1.8274,
+      "step": 8950
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 46.980613708496094,
+      "learning_rate": 1.9036444444444445e-05,
+      "loss": 1.7823,
+      "step": 8960
+    },
+    {
+      "epoch": 0.28704,
+      "grad_norm": 49.283103942871094,
+      "learning_rate": 1.903288888888889e-05,
+      "loss": 1.7414,
+      "step": 8970
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 48.2234001159668,
+      "learning_rate": 1.9029333333333334e-05,
+      "loss": 1.7708,
+      "step": 8980
+    },
+    {
+      "epoch": 0.28768,
+      "grad_norm": 47.294795989990234,
+      "learning_rate": 1.902577777777778e-05,
+      "loss": 1.8338,
+      "step": 8990
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 48.731834411621094,
+      "learning_rate": 1.9022222222222223e-05,
+      "loss": 1.7925,
+      "step": 9000
+    },
+    {
+      "epoch": 0.28832,
+      "grad_norm": 49.90666198730469,
+      "learning_rate": 1.9018666666666668e-05,
+      "loss": 1.8027,
+      "step": 9010
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 48.464569091796875,
+      "learning_rate": 1.9015111111111113e-05,
+      "loss": 1.7825,
+      "step": 9020
+    },
+    {
+      "epoch": 0.28896,
+      "grad_norm": 47.314823150634766,
+      "learning_rate": 1.9011555555555557e-05,
+      "loss": 1.7472,
+      "step": 9030
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 44.8106689453125,
+      "learning_rate": 1.9008e-05,
+      "loss": 1.7687,
+      "step": 9040
+    },
+    {
+      "epoch": 0.2896,
+      "grad_norm": 47.74726867675781,
+      "learning_rate": 1.9004444444444446e-05,
+      "loss": 1.7818,
+      "step": 9050
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 45.84540557861328,
+      "learning_rate": 1.900088888888889e-05,
+      "loss": 1.7905,
+      "step": 9060
+    },
+    {
+      "epoch": 0.29024,
+      "grad_norm": 46.0303840637207,
+      "learning_rate": 1.8997333333333335e-05,
+      "loss": 1.7701,
+      "step": 9070
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 47.5675048828125,
+      "learning_rate": 1.899377777777778e-05,
+      "loss": 1.7884,
+      "step": 9080
+    },
+    {
+      "epoch": 0.29088,
+      "grad_norm": 49.11726379394531,
+      "learning_rate": 1.8990222222222224e-05,
+      "loss": 1.7676,
+      "step": 9090
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 45.33092498779297,
+      "learning_rate": 1.898666666666667e-05,
+      "loss": 1.7959,
+      "step": 9100
+    },
+    {
+      "epoch": 0.29152,
+      "grad_norm": 45.65995407104492,
+      "learning_rate": 1.8983111111111114e-05,
+      "loss": 1.7636,
+      "step": 9110
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 46.20924758911133,
+      "learning_rate": 1.8979555555555558e-05,
+      "loss": 1.7979,
+      "step": 9120
+    },
+    {
+      "epoch": 0.29216,
+      "grad_norm": 45.72431945800781,
+      "learning_rate": 1.8976000000000003e-05,
+      "loss": 1.7564,
+      "step": 9130
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 46.04757308959961,
+      "learning_rate": 1.8972444444444447e-05,
+      "loss": 1.7588,
+      "step": 9140
+    },
+    {
+      "epoch": 0.2928,
+      "grad_norm": 45.486202239990234,
+      "learning_rate": 1.896888888888889e-05,
+      "loss": 1.7994,
+      "step": 9150
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 48.11737060546875,
+      "learning_rate": 1.8965333333333336e-05,
+      "loss": 1.7839,
+      "step": 9160
+    },
+    {
+      "epoch": 0.29344,
+      "grad_norm": 48.25556564331055,
+      "learning_rate": 1.8961777777777777e-05,
+      "loss": 1.7964,
+      "step": 9170
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 47.92764663696289,
+      "learning_rate": 1.8958222222222225e-05,
+      "loss": 1.759,
+      "step": 9180
+    },
+    {
+      "epoch": 0.29408,
+      "grad_norm": 48.09939956665039,
+      "learning_rate": 1.8954666666666667e-05,
+      "loss": 1.8148,
+      "step": 9190
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 45.122501373291016,
+      "learning_rate": 1.8951111111111115e-05,
+      "loss": 1.7707,
+      "step": 9200
+    },
+    {
+      "epoch": 0.29472,
+      "grad_norm": 47.23085403442383,
+      "learning_rate": 1.8947555555555556e-05,
+      "loss": 1.7575,
+      "step": 9210
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 47.04438018798828,
+      "learning_rate": 1.8944000000000004e-05,
+      "loss": 1.7709,
+      "step": 9220
+    },
+    {
+      "epoch": 0.29536,
+      "grad_norm": 49.08626174926758,
+      "learning_rate": 1.8940444444444445e-05,
+      "loss": 1.7826,
+      "step": 9230
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 45.1053581237793,
+      "learning_rate": 1.8936888888888893e-05,
+      "loss": 1.7963,
+      "step": 9240
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 48.38051986694336,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.7631,
+      "step": 9250
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 47.439598083496094,
+      "learning_rate": 1.8929777777777782e-05,
+      "loss": 1.7905,
+      "step": 9260
+    },
+    {
+      "epoch": 0.29664,
+      "grad_norm": 47.75306701660156,
+      "learning_rate": 1.8926222222222223e-05,
+      "loss": 1.7846,
+      "step": 9270
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 47.29454803466797,
+      "learning_rate": 1.8922666666666668e-05,
+      "loss": 1.7869,
+      "step": 9280
+    },
+    {
+      "epoch": 0.29728,
+      "grad_norm": 50.435028076171875,
+      "learning_rate": 1.8919111111111112e-05,
+      "loss": 1.7748,
+      "step": 9290
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 47.95176696777344,
+      "learning_rate": 1.8915555555555557e-05,
+      "loss": 1.766,
+      "step": 9300
+    },
+    {
+      "epoch": 0.29792,
+      "grad_norm": 48.74436950683594,
+      "learning_rate": 1.8912e-05,
+      "loss": 1.7644,
+      "step": 9310
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 47.629600524902344,
+      "learning_rate": 1.8908444444444446e-05,
+      "loss": 1.7657,
+      "step": 9320
+    },
+    {
+      "epoch": 0.29856,
+      "grad_norm": 48.885963439941406,
+      "learning_rate": 1.890488888888889e-05,
+      "loss": 1.7768,
+      "step": 9330
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 47.101810455322266,
+      "learning_rate": 1.8901333333333335e-05,
+      "loss": 1.7704,
+      "step": 9340
+    },
+    {
+      "epoch": 0.2992,
+      "grad_norm": 45.48361587524414,
+      "learning_rate": 1.889777777777778e-05,
+      "loss": 1.7875,
+      "step": 9350
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 48.26191329956055,
+      "learning_rate": 1.8894222222222224e-05,
+      "loss": 1.7986,
+      "step": 9360
+    },
+    {
+      "epoch": 0.29984,
+      "grad_norm": 47.485687255859375,
+      "learning_rate": 1.889066666666667e-05,
+      "loss": 1.7677,
+      "step": 9370
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 46.89303970336914,
+      "learning_rate": 1.8887111111111113e-05,
+      "loss": 1.7426,
+      "step": 9380
+    },
+    {
+      "epoch": 0.30048,
+      "grad_norm": 46.88542938232422,
+      "learning_rate": 1.8883555555555558e-05,
+      "loss": 1.7756,
+      "step": 9390
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 48.30686950683594,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 1.8225,
+      "step": 9400
+    },
+    {
+      "epoch": 0.30112,
+      "grad_norm": 47.19614028930664,
+      "learning_rate": 1.8876444444444447e-05,
+      "loss": 1.7551,
+      "step": 9410
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 47.09248352050781,
+      "learning_rate": 1.8872888888888888e-05,
+      "loss": 1.7781,
+      "step": 9420
+    },
+    {
+      "epoch": 0.30176,
+      "grad_norm": 47.05224609375,
+      "learning_rate": 1.8869333333333336e-05,
+      "loss": 1.7903,
+      "step": 9430
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 47.02943420410156,
+      "learning_rate": 1.8865777777777777e-05,
+      "loss": 1.7767,
+      "step": 9440
+    },
+    {
+      "epoch": 0.3024,
+      "grad_norm": 47.527587890625,
+      "learning_rate": 1.8862222222222225e-05,
+      "loss": 1.7778,
+      "step": 9450
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 46.53438186645508,
+      "learning_rate": 1.8858666666666666e-05,
+      "loss": 1.7415,
+      "step": 9460
+    },
+    {
+      "epoch": 0.30304,
+      "grad_norm": 46.07378387451172,
+      "learning_rate": 1.8855111111111114e-05,
+      "loss": 1.7837,
+      "step": 9470
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 47.58318328857422,
+      "learning_rate": 1.8851555555555555e-05,
+      "loss": 1.7784,
+      "step": 9480
+    },
+    {
+      "epoch": 0.30368,
+      "grad_norm": 47.040992736816406,
+      "learning_rate": 1.8848000000000003e-05,
+      "loss": 1.7898,
+      "step": 9490
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 47.662567138671875,
+      "learning_rate": 1.8844444444444444e-05,
+      "loss": 1.7556,
+      "step": 9500
+    },
+    {
+      "epoch": 0.30432,
+      "grad_norm": 47.28549575805664,
+      "learning_rate": 1.8840888888888892e-05,
+      "loss": 1.765,
+      "step": 9510
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 46.851097106933594,
+      "learning_rate": 1.8837333333333333e-05,
+      "loss": 1.7466,
+      "step": 9520
+    },
+    {
+      "epoch": 0.30496,
+      "grad_norm": 46.6473388671875,
+      "learning_rate": 1.883377777777778e-05,
+      "loss": 1.7977,
+      "step": 9530
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 48.394996643066406,
+      "learning_rate": 1.8830222222222223e-05,
+      "loss": 1.7194,
+      "step": 9540
+    },
+    {
+      "epoch": 0.3056,
+      "grad_norm": 45.25028610229492,
+      "learning_rate": 1.8826666666666667e-05,
+      "loss": 1.7798,
+      "step": 9550
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 47.76869583129883,
+      "learning_rate": 1.882311111111111e-05,
+      "loss": 1.7691,
+      "step": 9560
+    },
+    {
+      "epoch": 0.30624,
+      "grad_norm": 48.7626838684082,
+      "learning_rate": 1.8819555555555556e-05,
+      "loss": 1.771,
+      "step": 9570
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 46.44325256347656,
+      "learning_rate": 1.8816e-05,
+      "loss": 1.742,
+      "step": 9580
+    },
+    {
+      "epoch": 0.30688,
+      "grad_norm": 46.602508544921875,
+      "learning_rate": 1.8812444444444445e-05,
+      "loss": 1.7771,
+      "step": 9590
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 47.8487663269043,
+      "learning_rate": 1.880888888888889e-05,
+      "loss": 1.7564,
+      "step": 9600
+    },
+    {
+      "epoch": 0.30752,
+      "grad_norm": 47.99014663696289,
+      "learning_rate": 1.8805333333333334e-05,
+      "loss": 1.795,
+      "step": 9610
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 46.99567794799805,
+      "learning_rate": 1.880177777777778e-05,
+      "loss": 1.771,
+      "step": 9620
+    },
+    {
+      "epoch": 0.30816,
+      "grad_norm": 45.05099868774414,
+      "learning_rate": 1.8798222222222224e-05,
+      "loss": 1.7543,
+      "step": 9630
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 44.71341323852539,
+      "learning_rate": 1.8794666666666668e-05,
+      "loss": 1.7704,
+      "step": 9640
+    },
+    {
+      "epoch": 0.3088,
+      "grad_norm": 47.164066314697266,
+      "learning_rate": 1.8791111111111113e-05,
+      "loss": 1.7739,
+      "step": 9650
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 47.73215866088867,
+      "learning_rate": 1.8787555555555557e-05,
+      "loss": 1.7642,
+      "step": 9660
+    },
+    {
+      "epoch": 0.30944,
+      "grad_norm": 45.98970413208008,
+      "learning_rate": 1.8784000000000002e-05,
+      "loss": 1.7849,
+      "step": 9670
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 47.81766128540039,
+      "learning_rate": 1.8780444444444446e-05,
+      "loss": 1.7772,
+      "step": 9680
+    },
+    {
+      "epoch": 0.31008,
+      "grad_norm": 47.27102279663086,
+      "learning_rate": 1.877688888888889e-05,
+      "loss": 1.7741,
+      "step": 9690
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 45.84796142578125,
+      "learning_rate": 1.8773333333333335e-05,
+      "loss": 1.7645,
+      "step": 9700
+    },
+    {
+      "epoch": 0.31072,
+      "grad_norm": 49.074859619140625,
+      "learning_rate": 1.876977777777778e-05,
+      "loss": 1.777,
+      "step": 9710
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 46.83264923095703,
+      "learning_rate": 1.8766222222222225e-05,
+      "loss": 1.7312,
+      "step": 9720
+    },
+    {
+      "epoch": 0.31136,
+      "grad_norm": 48.32571792602539,
+      "learning_rate": 1.876266666666667e-05,
+      "loss": 1.7871,
+      "step": 9730
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 45.8250846862793,
+      "learning_rate": 1.8759111111111114e-05,
+      "loss": 1.7578,
+      "step": 9740
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 45.85641860961914,
+      "learning_rate": 1.8755555555555558e-05,
+      "loss": 1.8071,
+      "step": 9750
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 48.53336715698242,
+      "learning_rate": 1.8752000000000003e-05,
+      "loss": 1.7663,
+      "step": 9760
+    },
+    {
+      "epoch": 0.31264,
+      "grad_norm": 47.985836029052734,
+      "learning_rate": 1.8748444444444447e-05,
+      "loss": 1.7587,
+      "step": 9770
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 46.00651550292969,
+      "learning_rate": 1.8744888888888892e-05,
+      "loss": 1.7713,
+      "step": 9780
+    },
+    {
+      "epoch": 0.31328,
+      "grad_norm": 44.8219108581543,
+      "learning_rate": 1.8741333333333336e-05,
+      "loss": 1.8027,
+      "step": 9790
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 48.181697845458984,
+      "learning_rate": 1.8737777777777778e-05,
+      "loss": 1.7904,
+      "step": 9800
+    },
+    {
+      "epoch": 0.31392,
+      "grad_norm": 46.20647430419922,
+      "learning_rate": 1.8734222222222226e-05,
+      "loss": 1.7507,
+      "step": 9810
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 46.36750793457031,
+      "learning_rate": 1.8730666666666667e-05,
+      "loss": 1.7458,
+      "step": 9820
+    },
+    {
+      "epoch": 0.31456,
+      "grad_norm": 48.03936004638672,
+      "learning_rate": 1.8727111111111115e-05,
+      "loss": 1.7935,
+      "step": 9830
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 47.40336990356445,
+      "learning_rate": 1.8723555555555556e-05,
+      "loss": 1.7613,
+      "step": 9840
+    },
+    {
+      "epoch": 0.3152,
+      "grad_norm": 44.89904022216797,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 1.7486,
+      "step": 9850
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 46.88365936279297,
+      "learning_rate": 1.8716444444444445e-05,
+      "loss": 1.7897,
+      "step": 9860
+    },
+    {
+      "epoch": 0.31584,
+      "grad_norm": 45.93436050415039,
+      "learning_rate": 1.871288888888889e-05,
+      "loss": 1.7787,
+      "step": 9870
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 45.79889678955078,
+      "learning_rate": 1.8709333333333334e-05,
+      "loss": 1.7857,
+      "step": 9880
+    },
+    {
+      "epoch": 0.31648,
+      "grad_norm": 48.29885482788086,
+      "learning_rate": 1.870577777777778e-05,
+      "loss": 1.782,
+      "step": 9890
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 46.49193572998047,
+      "learning_rate": 1.8702222222222223e-05,
+      "loss": 1.7554,
+      "step": 9900
+    },
+    {
+      "epoch": 0.31712,
+      "grad_norm": 44.27806091308594,
+      "learning_rate": 1.8698666666666668e-05,
+      "loss": 1.773,
+      "step": 9910
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 48.690162658691406,
+      "learning_rate": 1.8695111111111112e-05,
+      "loss": 1.7727,
+      "step": 9920
+    },
+    {
+      "epoch": 0.31776,
+      "grad_norm": 47.276405334472656,
+      "learning_rate": 1.8691555555555557e-05,
+      "loss": 1.7563,
+      "step": 9930
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 48.51708221435547,
+      "learning_rate": 1.8688e-05,
+      "loss": 1.7822,
+      "step": 9940
+    },
+    {
+      "epoch": 0.3184,
+      "grad_norm": 48.64371109008789,
+      "learning_rate": 1.8684444444444446e-05,
+      "loss": 1.7525,
+      "step": 9950
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 45.173187255859375,
+      "learning_rate": 1.868088888888889e-05,
+      "loss": 1.7504,
+      "step": 9960
+    },
+    {
+      "epoch": 0.31904,
+      "grad_norm": 47.1632194519043,
+      "learning_rate": 1.8677333333333335e-05,
+      "loss": 1.784,
+      "step": 9970
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 46.15009307861328,
+      "learning_rate": 1.867377777777778e-05,
+      "loss": 1.7506,
+      "step": 9980
+    },
+    {
+      "epoch": 0.31968,
+      "grad_norm": 46.347412109375,
+      "learning_rate": 1.8670222222222224e-05,
+      "loss": 1.7905,
+      "step": 9990
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 49.578147888183594,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.7491,
+      "step": 10000
+    },
+    {
+      "epoch": 0.32032,
+      "grad_norm": 48.711246490478516,
+      "learning_rate": 1.8663111111111113e-05,
+      "loss": 1.7571,
+      "step": 10010
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 47.9279899597168,
+      "learning_rate": 1.8659555555555558e-05,
+      "loss": 1.764,
+      "step": 10020
+    },
+    {
+      "epoch": 0.32096,
+      "grad_norm": 47.64333724975586,
+      "learning_rate": 1.8656000000000002e-05,
+      "loss": 1.7664,
+      "step": 10030
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 46.161312103271484,
+      "learning_rate": 1.8652444444444447e-05,
+      "loss": 1.762,
+      "step": 10040
+    },
+    {
+      "epoch": 0.3216,
+      "grad_norm": 46.62035369873047,
+      "learning_rate": 1.8648888888888888e-05,
+      "loss": 1.7595,
+      "step": 10050
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 46.48634338378906,
+      "learning_rate": 1.8645333333333336e-05,
+      "loss": 1.7663,
+      "step": 10060
+    },
+    {
+      "epoch": 0.32224,
+      "grad_norm": 44.407623291015625,
+      "learning_rate": 1.8641777777777777e-05,
+      "loss": 1.7764,
+      "step": 10070
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 47.40081787109375,
+      "learning_rate": 1.8638222222222225e-05,
+      "loss": 1.7533,
+      "step": 10080
+    },
+    {
+      "epoch": 0.32288,
+      "grad_norm": 44.79414749145508,
+      "learning_rate": 1.8634666666666666e-05,
+      "loss": 1.7668,
+      "step": 10090
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 49.381526947021484,
+      "learning_rate": 1.8631111111111114e-05,
+      "loss": 1.7661,
+      "step": 10100
+    },
+    {
+      "epoch": 0.32352,
+      "grad_norm": 46.1864128112793,
+      "learning_rate": 1.8627555555555555e-05,
+      "loss": 1.7585,
+      "step": 10110
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 44.101497650146484,
+      "learning_rate": 1.8624000000000003e-05,
+      "loss": 1.7463,
+      "step": 10120
+    },
+    {
+      "epoch": 0.32416,
+      "grad_norm": 48.681636810302734,
+      "learning_rate": 1.8620444444444444e-05,
+      "loss": 1.7981,
+      "step": 10130
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 46.09796142578125,
+      "learning_rate": 1.8616888888888892e-05,
+      "loss": 1.7711,
+      "step": 10140
+    },
+    {
+      "epoch": 0.3248,
+      "grad_norm": 45.78500747680664,
+      "learning_rate": 1.8613333333333334e-05,
+      "loss": 1.7613,
+      "step": 10150
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 46.612937927246094,
+      "learning_rate": 1.860977777777778e-05,
+      "loss": 1.7646,
+      "step": 10160
+    },
+    {
+      "epoch": 0.32544,
+      "grad_norm": 46.29045867919922,
+      "learning_rate": 1.8606222222222223e-05,
+      "loss": 1.7676,
+      "step": 10170
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 47.204795837402344,
+      "learning_rate": 1.8602666666666667e-05,
+      "loss": 1.7667,
+      "step": 10180
+    },
+    {
+      "epoch": 0.32608,
+      "grad_norm": 49.51637268066406,
+      "learning_rate": 1.8599111111111112e-05,
+      "loss": 1.7439,
+      "step": 10190
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 46.27983856201172,
+      "learning_rate": 1.8595555555555556e-05,
+      "loss": 1.7464,
+      "step": 10200
+    },
+    {
+      "epoch": 0.32672,
+      "grad_norm": 50.15081024169922,
+      "learning_rate": 1.8592e-05,
+      "loss": 1.737,
+      "step": 10210
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 48.33657455444336,
+      "learning_rate": 1.8588444444444445e-05,
+      "loss": 1.7338,
+      "step": 10220
+    },
+    {
+      "epoch": 0.32736,
+      "grad_norm": 45.548072814941406,
+      "learning_rate": 1.858488888888889e-05,
+      "loss": 1.7796,
+      "step": 10230
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 46.754878997802734,
+      "learning_rate": 1.8581333333333335e-05,
+      "loss": 1.7664,
+      "step": 10240
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 47.31142807006836,
+      "learning_rate": 1.857777777777778e-05,
+      "loss": 1.7729,
+      "step": 10250
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 48.030948638916016,
+      "learning_rate": 1.8574222222222224e-05,
+      "loss": 1.7665,
+      "step": 10260
+    },
+    {
+      "epoch": 0.32864,
+      "grad_norm": 47.46772003173828,
+      "learning_rate": 1.8570666666666668e-05,
+      "loss": 1.731,
+      "step": 10270
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 46.123939514160156,
+      "learning_rate": 1.8567111111111113e-05,
+      "loss": 1.7825,
+      "step": 10280
+    },
+    {
+      "epoch": 0.32928,
+      "grad_norm": 46.65377426147461,
+      "learning_rate": 1.8563555555555557e-05,
+      "loss": 1.7956,
+      "step": 10290
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 51.14271926879883,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 1.7583,
+      "step": 10300
+    },
+    {
+      "epoch": 0.32992,
+      "grad_norm": 45.32281494140625,
+      "learning_rate": 1.8556444444444446e-05,
+      "loss": 1.7874,
+      "step": 10310
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 45.66364669799805,
+      "learning_rate": 1.855288888888889e-05,
+      "loss": 1.7917,
+      "step": 10320
+    },
+    {
+      "epoch": 0.33056,
+      "grad_norm": 47.25345993041992,
+      "learning_rate": 1.8549333333333336e-05,
+      "loss": 1.7249,
+      "step": 10330
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 45.0384635925293,
+      "learning_rate": 1.854577777777778e-05,
+      "loss": 1.7708,
+      "step": 10340
+    },
+    {
+      "epoch": 0.3312,
+      "grad_norm": 47.916831970214844,
+      "learning_rate": 1.8542222222222225e-05,
+      "loss": 1.7726,
+      "step": 10350
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 45.50624084472656,
+      "learning_rate": 1.853866666666667e-05,
+      "loss": 1.7704,
+      "step": 10360
+    },
+    {
+      "epoch": 0.33184,
+      "grad_norm": 49.995033264160156,
+      "learning_rate": 1.8535111111111114e-05,
+      "loss": 1.7448,
+      "step": 10370
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 43.8958625793457,
+      "learning_rate": 1.8531555555555555e-05,
+      "loss": 1.7621,
+      "step": 10380
+    },
+    {
+      "epoch": 0.33248,
+      "grad_norm": 47.6920280456543,
+      "learning_rate": 1.8528000000000003e-05,
+      "loss": 1.7757,
+      "step": 10390
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 46.501773834228516,
+      "learning_rate": 1.8524444444444444e-05,
+      "loss": 1.757,
+      "step": 10400
+    },
+    {
+      "epoch": 0.33312,
+      "grad_norm": 49.876014709472656,
+      "learning_rate": 1.8520888888888892e-05,
+      "loss": 1.7676,
+      "step": 10410
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 45.7028694152832,
+      "learning_rate": 1.8517333333333333e-05,
+      "loss": 1.7685,
+      "step": 10420
+    },
+    {
+      "epoch": 0.33376,
+      "grad_norm": 47.121124267578125,
+      "learning_rate": 1.851377777777778e-05,
+      "loss": 1.7713,
+      "step": 10430
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 43.50001525878906,
+      "learning_rate": 1.8510222222222222e-05,
+      "loss": 1.7629,
+      "step": 10440
+    },
+    {
+      "epoch": 0.3344,
+      "grad_norm": 45.10226821899414,
+      "learning_rate": 1.8506666666666667e-05,
+      "loss": 1.7892,
+      "step": 10450
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 45.64974594116211,
+      "learning_rate": 1.850311111111111e-05,
+      "loss": 1.7734,
+      "step": 10460
+    },
+    {
+      "epoch": 0.33504,
+      "grad_norm": 48.50352096557617,
+      "learning_rate": 1.8499555555555556e-05,
+      "loss": 1.7701,
+      "step": 10470
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 47.27094268798828,
+      "learning_rate": 1.8496e-05,
+      "loss": 1.7888,
+      "step": 10480
+    },
+    {
+      "epoch": 0.33568,
+      "grad_norm": 44.90556335449219,
+      "learning_rate": 1.8492444444444445e-05,
+      "loss": 1.7535,
+      "step": 10490
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 47.575096130371094,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 1.7657,
+      "step": 10500
+    },
+    {
+      "epoch": 0.33632,
+      "grad_norm": 49.730674743652344,
+      "learning_rate": 1.8485333333333334e-05,
+      "loss": 1.7859,
+      "step": 10510
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 45.4715690612793,
+      "learning_rate": 1.848177777777778e-05,
+      "loss": 1.7721,
+      "step": 10520
+    },
+    {
+      "epoch": 0.33696,
+      "grad_norm": 45.48630905151367,
+      "learning_rate": 1.8478222222222223e-05,
+      "loss": 1.7347,
+      "step": 10530
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 46.90533447265625,
+      "learning_rate": 1.8474666666666668e-05,
+      "loss": 1.7278,
+      "step": 10540
+    },
+    {
+      "epoch": 0.3376,
+      "grad_norm": 47.98374557495117,
+      "learning_rate": 1.8471111111111112e-05,
+      "loss": 1.7522,
+      "step": 10550
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 45.95107650756836,
+      "learning_rate": 1.8467555555555557e-05,
+      "loss": 1.7476,
+      "step": 10560
+    },
+    {
+      "epoch": 0.33824,
+      "grad_norm": 47.6331672668457,
+      "learning_rate": 1.8464e-05,
+      "loss": 1.7487,
+      "step": 10570
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 44.778446197509766,
+      "learning_rate": 1.8460444444444446e-05,
+      "loss": 1.7472,
+      "step": 10580
+    },
+    {
+      "epoch": 0.33888,
+      "grad_norm": 47.06455993652344,
+      "learning_rate": 1.845688888888889e-05,
+      "loss": 1.7799,
+      "step": 10590
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 47.35928726196289,
+      "learning_rate": 1.8453333333333335e-05,
+      "loss": 1.7418,
+      "step": 10600
+    },
+    {
+      "epoch": 0.33952,
+      "grad_norm": 44.61867141723633,
+      "learning_rate": 1.844977777777778e-05,
+      "loss": 1.7388,
+      "step": 10610
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 45.826133728027344,
+      "learning_rate": 1.8446222222222224e-05,
+      "loss": 1.7485,
+      "step": 10620
+    },
+    {
+      "epoch": 0.34016,
+      "grad_norm": 48.199649810791016,
+      "learning_rate": 1.844266666666667e-05,
+      "loss": 1.7742,
+      "step": 10630
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 48.93040466308594,
+      "learning_rate": 1.8439111111111113e-05,
+      "loss": 1.7667,
+      "step": 10640
+    },
+    {
+      "epoch": 0.3408,
+      "grad_norm": 46.984500885009766,
+      "learning_rate": 1.8435555555555558e-05,
+      "loss": 1.7965,
+      "step": 10650
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 48.803184509277344,
+      "learning_rate": 1.8432000000000002e-05,
+      "loss": 1.7528,
+      "step": 10660
+    },
+    {
+      "epoch": 0.34144,
+      "grad_norm": 47.56668472290039,
+      "learning_rate": 1.8428444444444447e-05,
+      "loss": 1.7542,
+      "step": 10670
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 47.09225082397461,
+      "learning_rate": 1.842488888888889e-05,
+      "loss": 1.7825,
+      "step": 10680
+    },
+    {
+      "epoch": 0.34208,
+      "grad_norm": 47.98640060424805,
+      "learning_rate": 1.8421333333333336e-05,
+      "loss": 1.7399,
+      "step": 10690
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 45.85944366455078,
+      "learning_rate": 1.8417777777777777e-05,
+      "loss": 1.7566,
+      "step": 10700
+    },
+    {
+      "epoch": 0.34272,
+      "grad_norm": 47.46086120605469,
+      "learning_rate": 1.8414222222222225e-05,
+      "loss": 1.7667,
+      "step": 10710
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 46.52635192871094,
+      "learning_rate": 1.8410666666666666e-05,
+      "loss": 1.7813,
+      "step": 10720
+    },
+    {
+      "epoch": 0.34336,
+      "grad_norm": 44.20283126831055,
+      "learning_rate": 1.8407111111111114e-05,
+      "loss": 1.7366,
+      "step": 10730
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 47.95680618286133,
+      "learning_rate": 1.8403555555555555e-05,
+      "loss": 1.7492,
+      "step": 10740
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 44.47868728637695,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.7709,
+      "step": 10750
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 45.7765998840332,
+      "learning_rate": 1.8396444444444445e-05,
+      "loss": 1.7466,
+      "step": 10760
+    },
+    {
+      "epoch": 0.34464,
+      "grad_norm": 49.115234375,
+      "learning_rate": 1.8392888888888893e-05,
+      "loss": 1.7933,
+      "step": 10770
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 45.5890007019043,
+      "learning_rate": 1.8389333333333334e-05,
+      "loss": 1.7436,
+      "step": 10780
+    },
+    {
+      "epoch": 0.34528,
+      "grad_norm": 46.42338562011719,
+      "learning_rate": 1.838577777777778e-05,
+      "loss": 1.7747,
+      "step": 10790
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 45.16661071777344,
+      "learning_rate": 1.8382222222222223e-05,
+      "loss": 1.7467,
+      "step": 10800
+    },
+    {
+      "epoch": 0.34592,
+      "grad_norm": 45.05027770996094,
+      "learning_rate": 1.837866666666667e-05,
+      "loss": 1.7571,
+      "step": 10810
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 44.80595397949219,
+      "learning_rate": 1.8375111111111112e-05,
+      "loss": 1.7643,
+      "step": 10820
+    },
+    {
+      "epoch": 0.34656,
+      "grad_norm": 44.93939971923828,
+      "learning_rate": 1.8371555555555556e-05,
+      "loss": 1.7336,
+      "step": 10830
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 46.136112213134766,
+      "learning_rate": 1.8368e-05,
+      "loss": 1.7521,
+      "step": 10840
+    },
+    {
+      "epoch": 0.3472,
+      "grad_norm": 45.64118957519531,
+      "learning_rate": 1.8364444444444446e-05,
+      "loss": 1.7721,
+      "step": 10850
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 47.01308822631836,
+      "learning_rate": 1.836088888888889e-05,
+      "loss": 1.7538,
+      "step": 10860
+    },
+    {
+      "epoch": 0.34784,
+      "grad_norm": 48.4968147277832,
+      "learning_rate": 1.8357333333333335e-05,
+      "loss": 1.7721,
+      "step": 10870
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 45.39475631713867,
+      "learning_rate": 1.835377777777778e-05,
+      "loss": 1.7198,
+      "step": 10880
+    },
+    {
+      "epoch": 0.34848,
+      "grad_norm": 48.02215576171875,
+      "learning_rate": 1.8350222222222224e-05,
+      "loss": 1.7364,
+      "step": 10890
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 45.59247970581055,
+      "learning_rate": 1.834666666666667e-05,
+      "loss": 1.7428,
+      "step": 10900
+    },
+    {
+      "epoch": 0.34912,
+      "grad_norm": 46.79558181762695,
+      "learning_rate": 1.8343111111111113e-05,
+      "loss": 1.7639,
+      "step": 10910
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 45.718727111816406,
+      "learning_rate": 1.8339555555555557e-05,
+      "loss": 1.7313,
+      "step": 10920
+    },
+    {
+      "epoch": 0.34976,
+      "grad_norm": 48.2978630065918,
+      "learning_rate": 1.8336000000000002e-05,
+      "loss": 1.7908,
+      "step": 10930
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 46.27851104736328,
+      "learning_rate": 1.8332444444444447e-05,
+      "loss": 1.7485,
+      "step": 10940
+    },
+    {
+      "epoch": 0.3504,
+      "grad_norm": 46.8973274230957,
+      "learning_rate": 1.832888888888889e-05,
+      "loss": 1.7244,
+      "step": 10950
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 45.576988220214844,
+      "learning_rate": 1.8325333333333336e-05,
+      "loss": 1.7575,
+      "step": 10960
+    },
+    {
+      "epoch": 0.35104,
+      "grad_norm": 45.916908264160156,
+      "learning_rate": 1.8321777777777777e-05,
+      "loss": 1.722,
+      "step": 10970
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 44.97768783569336,
+      "learning_rate": 1.8318222222222225e-05,
+      "loss": 1.7588,
+      "step": 10980
+    },
+    {
+      "epoch": 0.35168,
+      "grad_norm": 45.55659484863281,
+      "learning_rate": 1.8314666666666666e-05,
+      "loss": 1.7565,
+      "step": 10990
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 45.32796096801758,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 1.7627,
+      "step": 11000
+    },
+    {
+      "epoch": 0.35232,
+      "grad_norm": 46.5700798034668,
+      "learning_rate": 1.8307555555555555e-05,
+      "loss": 1.7332,
+      "step": 11010
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 46.67179489135742,
+      "learning_rate": 1.8304000000000003e-05,
+      "loss": 1.764,
+      "step": 11020
+    },
+    {
+      "epoch": 0.35296,
+      "grad_norm": 44.31849670410156,
+      "learning_rate": 1.8300444444444444e-05,
+      "loss": 1.7501,
+      "step": 11030
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 47.23154067993164,
+      "learning_rate": 1.8296888888888892e-05,
+      "loss": 1.7756,
+      "step": 11040
+    },
+    {
+      "epoch": 0.3536,
+      "grad_norm": 45.82375717163086,
+      "learning_rate": 1.8293333333333333e-05,
+      "loss": 1.758,
+      "step": 11050
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 45.44552993774414,
+      "learning_rate": 1.828977777777778e-05,
+      "loss": 1.7307,
+      "step": 11060
+    },
+    {
+      "epoch": 0.35424,
+      "grad_norm": 44.66324234008789,
+      "learning_rate": 1.8286222222222222e-05,
+      "loss": 1.7543,
+      "step": 11070
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 45.85771560668945,
+      "learning_rate": 1.8282666666666667e-05,
+      "loss": 1.7505,
+      "step": 11080
+    },
+    {
+      "epoch": 0.35488,
+      "grad_norm": 44.97885513305664,
+      "learning_rate": 1.827911111111111e-05,
+      "loss": 1.7334,
+      "step": 11090
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 44.4973258972168,
+      "learning_rate": 1.8275555555555556e-05,
+      "loss": 1.7625,
+      "step": 11100
+    },
+    {
+      "epoch": 0.35552,
+      "grad_norm": 48.64956283569336,
+      "learning_rate": 1.8272e-05,
+      "loss": 1.7664,
+      "step": 11110
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 46.956947326660156,
+      "learning_rate": 1.8268444444444445e-05,
+      "loss": 1.7611,
+      "step": 11120
+    },
+    {
+      "epoch": 0.35616,
+      "grad_norm": 45.659523010253906,
+      "learning_rate": 1.826488888888889e-05,
+      "loss": 1.7577,
+      "step": 11130
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 46.591251373291016,
+      "learning_rate": 1.8261333333333334e-05,
+      "loss": 1.7543,
+      "step": 11140
+    },
+    {
+      "epoch": 0.3568,
+      "grad_norm": 44.3946533203125,
+      "learning_rate": 1.825777777777778e-05,
+      "loss": 1.7314,
+      "step": 11150
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 45.77732849121094,
+      "learning_rate": 1.8254222222222223e-05,
+      "loss": 1.7455,
+      "step": 11160
+    },
+    {
+      "epoch": 0.35744,
+      "grad_norm": 44.791587829589844,
+      "learning_rate": 1.8250666666666668e-05,
+      "loss": 1.7583,
+      "step": 11170
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 45.5029296875,
+      "learning_rate": 1.8247111111111112e-05,
+      "loss": 1.757,
+      "step": 11180
+    },
+    {
+      "epoch": 0.35808,
+      "grad_norm": 46.15699768066406,
+      "learning_rate": 1.8243555555555557e-05,
+      "loss": 1.7399,
+      "step": 11190
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 44.94160461425781,
+      "learning_rate": 1.824e-05,
+      "loss": 1.7639,
+      "step": 11200
+    },
+    {
+      "epoch": 0.35872,
+      "grad_norm": 45.93684005737305,
+      "learning_rate": 1.8236444444444446e-05,
+      "loss": 1.763,
+      "step": 11210
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 44.380393981933594,
+      "learning_rate": 1.823288888888889e-05,
+      "loss": 1.7609,
+      "step": 11220
+    },
+    {
+      "epoch": 0.35936,
+      "grad_norm": 46.544795989990234,
+      "learning_rate": 1.8229333333333335e-05,
+      "loss": 1.7309,
+      "step": 11230
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 45.34314727783203,
+      "learning_rate": 1.822577777777778e-05,
+      "loss": 1.7389,
+      "step": 11240
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 47.537864685058594,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 1.75,
+      "step": 11250
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 48.27071762084961,
+      "learning_rate": 1.821866666666667e-05,
+      "loss": 1.7758,
+      "step": 11260
+    },
+    {
+      "epoch": 0.36064,
+      "grad_norm": 50.97309875488281,
+      "learning_rate": 1.8215111111111113e-05,
+      "loss": 1.76,
+      "step": 11270
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 45.12947463989258,
+      "learning_rate": 1.8211555555555558e-05,
+      "loss": 1.7682,
+      "step": 11280
+    },
+    {
+      "epoch": 0.36128,
+      "grad_norm": 48.3563117980957,
+      "learning_rate": 1.8208000000000003e-05,
+      "loss": 1.7378,
+      "step": 11290
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 45.731014251708984,
+      "learning_rate": 1.8204444444444447e-05,
+      "loss": 1.7396,
+      "step": 11300
+    },
+    {
+      "epoch": 0.36192,
+      "grad_norm": 45.1165771484375,
+      "learning_rate": 1.820088888888889e-05,
+      "loss": 1.7387,
+      "step": 11310
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 45.07907485961914,
+      "learning_rate": 1.8197333333333336e-05,
+      "loss": 1.8069,
+      "step": 11320
+    },
+    {
+      "epoch": 0.36256,
+      "grad_norm": 45.680091857910156,
+      "learning_rate": 1.819377777777778e-05,
+      "loss": 1.7286,
+      "step": 11330
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 47.89656066894531,
+      "learning_rate": 1.8190222222222225e-05,
+      "loss": 1.746,
+      "step": 11340
+    },
+    {
+      "epoch": 0.3632,
+      "grad_norm": 44.2681999206543,
+      "learning_rate": 1.8186666666666666e-05,
+      "loss": 1.7426,
+      "step": 11350
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 46.58122634887695,
+      "learning_rate": 1.8183111111111114e-05,
+      "loss": 1.7493,
+      "step": 11360
+    },
+    {
+      "epoch": 0.36384,
+      "grad_norm": 45.0501823425293,
+      "learning_rate": 1.8179555555555556e-05,
+      "loss": 1.7392,
+      "step": 11370
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 46.51576232910156,
+      "learning_rate": 1.8176000000000004e-05,
+      "loss": 1.7417,
+      "step": 11380
+    },
+    {
+      "epoch": 0.36448,
+      "grad_norm": 44.66267776489258,
+      "learning_rate": 1.8172444444444445e-05,
+      "loss": 1.7547,
+      "step": 11390
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 45.7799186706543,
+      "learning_rate": 1.8168888888888893e-05,
+      "loss": 1.7624,
+      "step": 11400
+    },
+    {
+      "epoch": 0.36512,
+      "grad_norm": 47.37014389038086,
+      "learning_rate": 1.8165333333333334e-05,
+      "loss": 1.767,
+      "step": 11410
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 47.13766098022461,
+      "learning_rate": 1.816177777777778e-05,
+      "loss": 1.7526,
+      "step": 11420
+    },
+    {
+      "epoch": 0.36576,
+      "grad_norm": 47.16769790649414,
+      "learning_rate": 1.8158222222222223e-05,
+      "loss": 1.7376,
+      "step": 11430
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 46.399654388427734,
+      "learning_rate": 1.8154666666666667e-05,
+      "loss": 1.75,
+      "step": 11440
+    },
+    {
+      "epoch": 0.3664,
+      "grad_norm": 46.89807891845703,
+      "learning_rate": 1.8151111111111112e-05,
+      "loss": 1.7607,
+      "step": 11450
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 46.25182342529297,
+      "learning_rate": 1.8147555555555557e-05,
+      "loss": 1.7196,
+      "step": 11460
+    },
+    {
+      "epoch": 0.36704,
+      "grad_norm": 45.322322845458984,
+      "learning_rate": 1.8144e-05,
+      "loss": 1.7141,
+      "step": 11470
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 46.26806640625,
+      "learning_rate": 1.8140444444444446e-05,
+      "loss": 1.7524,
+      "step": 11480
+    },
+    {
+      "epoch": 0.36768,
+      "grad_norm": 43.700660705566406,
+      "learning_rate": 1.813688888888889e-05,
+      "loss": 1.7206,
+      "step": 11490
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 46.64167785644531,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.7522,
+      "step": 11500
+    },
+    {
+      "epoch": 0.36832,
+      "grad_norm": 45.604671478271484,
+      "learning_rate": 1.812977777777778e-05,
+      "loss": 1.7551,
+      "step": 11510
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 46.66270446777344,
+      "learning_rate": 1.8126222222222224e-05,
+      "loss": 1.7218,
+      "step": 11520
+    },
+    {
+      "epoch": 0.36896,
+      "grad_norm": 47.54534149169922,
+      "learning_rate": 1.812266666666667e-05,
+      "loss": 1.7246,
+      "step": 11530
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 48.792396545410156,
+      "learning_rate": 1.8119111111111113e-05,
+      "loss": 1.7373,
+      "step": 11540
+    },
+    {
+      "epoch": 0.3696,
+      "grad_norm": 45.595306396484375,
+      "learning_rate": 1.8115555555555558e-05,
+      "loss": 1.7531,
+      "step": 11550
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 45.06191635131836,
+      "learning_rate": 1.8112000000000002e-05,
+      "loss": 1.7465,
+      "step": 11560
+    },
+    {
+      "epoch": 0.37024,
+      "grad_norm": 47.39137649536133,
+      "learning_rate": 1.8108444444444447e-05,
+      "loss": 1.7441,
+      "step": 11570
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 44.804264068603516,
+      "learning_rate": 1.810488888888889e-05,
+      "loss": 1.7558,
+      "step": 11580
+    },
+    {
+      "epoch": 0.37088,
+      "grad_norm": 47.24115753173828,
+      "learning_rate": 1.8101333333333336e-05,
+      "loss": 1.735,
+      "step": 11590
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 47.79445266723633,
+      "learning_rate": 1.8097777777777777e-05,
+      "loss": 1.7455,
+      "step": 11600
+    },
+    {
+      "epoch": 0.37152,
+      "grad_norm": 45.69266891479492,
+      "learning_rate": 1.8094222222222225e-05,
+      "loss": 1.7525,
+      "step": 11610
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 46.884437561035156,
+      "learning_rate": 1.8090666666666666e-05,
+      "loss": 1.73,
+      "step": 11620
+    },
+    {
+      "epoch": 0.37216,
+      "grad_norm": 46.90225601196289,
+      "learning_rate": 1.8087111111111114e-05,
+      "loss": 1.7528,
+      "step": 11630
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 48.71936798095703,
+      "learning_rate": 1.8083555555555555e-05,
+      "loss": 1.7683,
+      "step": 11640
+    },
+    {
+      "epoch": 0.3728,
+      "grad_norm": 44.3066291809082,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 1.7455,
+      "step": 11650
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 45.78877639770508,
+      "learning_rate": 1.8076444444444444e-05,
+      "loss": 1.78,
+      "step": 11660
+    },
+    {
+      "epoch": 0.37344,
+      "grad_norm": 48.4770622253418,
+      "learning_rate": 1.8072888888888892e-05,
+      "loss": 1.7724,
+      "step": 11670
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 44.937191009521484,
+      "learning_rate": 1.8069333333333333e-05,
+      "loss": 1.7467,
+      "step": 11680
+    },
+    {
+      "epoch": 0.37408,
+      "grad_norm": 45.83155059814453,
+      "learning_rate": 1.806577777777778e-05,
+      "loss": 1.7419,
+      "step": 11690
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 47.36248779296875,
+      "learning_rate": 1.8062222222222222e-05,
+      "loss": 1.7215,
+      "step": 11700
+    },
+    {
+      "epoch": 0.37472,
+      "grad_norm": 48.68720626831055,
+      "learning_rate": 1.805866666666667e-05,
+      "loss": 1.7201,
+      "step": 11710
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 44.902809143066406,
+      "learning_rate": 1.805511111111111e-05,
+      "loss": 1.7343,
+      "step": 11720
+    },
+    {
+      "epoch": 0.37536,
+      "grad_norm": 45.616302490234375,
+      "learning_rate": 1.8051555555555556e-05,
+      "loss": 1.7533,
+      "step": 11730
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 48.17421340942383,
+      "learning_rate": 1.8048e-05,
+      "loss": 1.7369,
+      "step": 11740
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 46.45351791381836,
+      "learning_rate": 1.8044444444444445e-05,
+      "loss": 1.7455,
+      "step": 11750
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 47.1370849609375,
+      "learning_rate": 1.804088888888889e-05,
+      "loss": 1.7546,
+      "step": 11760
+    },
+    {
+      "epoch": 0.37664,
+      "grad_norm": 47.52964401245117,
+      "learning_rate": 1.8037333333333334e-05,
+      "loss": 1.7315,
+      "step": 11770
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 46.355072021484375,
+      "learning_rate": 1.803377777777778e-05,
+      "loss": 1.7806,
+      "step": 11780
+    },
+    {
+      "epoch": 0.37728,
+      "grad_norm": 47.648521423339844,
+      "learning_rate": 1.8030222222222223e-05,
+      "loss": 1.7521,
+      "step": 11790
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 46.333187103271484,
+      "learning_rate": 1.8026666666666668e-05,
+      "loss": 1.7419,
+      "step": 11800
+    },
+    {
+      "epoch": 0.37792,
+      "grad_norm": 47.49595260620117,
+      "learning_rate": 1.8023111111111113e-05,
+      "loss": 1.7417,
+      "step": 11810
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 44.231197357177734,
+      "learning_rate": 1.8019555555555557e-05,
+      "loss": 1.7185,
+      "step": 11820
+    },
+    {
+      "epoch": 0.37856,
+      "grad_norm": 46.03631591796875,
+      "learning_rate": 1.8016e-05,
+      "loss": 1.7129,
+      "step": 11830
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 44.27299499511719,
+      "learning_rate": 1.8012444444444446e-05,
+      "loss": 1.752,
+      "step": 11840
+    },
+    {
+      "epoch": 0.3792,
+      "grad_norm": 45.1333122253418,
+      "learning_rate": 1.800888888888889e-05,
+      "loss": 1.7286,
+      "step": 11850
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 46.22859573364258,
+      "learning_rate": 1.8005333333333335e-05,
+      "loss": 1.7488,
+      "step": 11860
+    },
+    {
+      "epoch": 0.37984,
+      "grad_norm": 44.65610885620117,
+      "learning_rate": 1.800177777777778e-05,
+      "loss": 1.7543,
+      "step": 11870
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 45.86985778808594,
+      "learning_rate": 1.7998222222222224e-05,
+      "loss": 1.7556,
+      "step": 11880
+    },
+    {
+      "epoch": 0.38048,
+      "grad_norm": 45.231658935546875,
+      "learning_rate": 1.799466666666667e-05,
+      "loss": 1.7255,
+      "step": 11890
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 44.76812744140625,
+      "learning_rate": 1.7991111111111114e-05,
+      "loss": 1.7467,
+      "step": 11900
+    },
+    {
+      "epoch": 0.38112,
+      "grad_norm": 45.64585876464844,
+      "learning_rate": 1.7987555555555558e-05,
+      "loss": 1.7702,
+      "step": 11910
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 47.54214859008789,
+      "learning_rate": 1.7984000000000003e-05,
+      "loss": 1.7646,
+      "step": 11920
+    },
+    {
+      "epoch": 0.38176,
+      "grad_norm": 44.36042785644531,
+      "learning_rate": 1.7980444444444444e-05,
+      "loss": 1.7226,
+      "step": 11930
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 46.86466979980469,
+      "learning_rate": 1.7976888888888892e-05,
+      "loss": 1.7231,
+      "step": 11940
+    },
+    {
+      "epoch": 0.3824,
+      "grad_norm": 44.66880416870117,
+      "learning_rate": 1.7973333333333333e-05,
+      "loss": 1.7667,
+      "step": 11950
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 46.481754302978516,
+      "learning_rate": 1.796977777777778e-05,
+      "loss": 1.7423,
+      "step": 11960
+    },
+    {
+      "epoch": 0.38304,
+      "grad_norm": 47.537837982177734,
+      "learning_rate": 1.7966222222222222e-05,
+      "loss": 1.7087,
+      "step": 11970
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 50.210052490234375,
+      "learning_rate": 1.7962666666666667e-05,
+      "loss": 1.7374,
+      "step": 11980
+    },
+    {
+      "epoch": 0.38368,
+      "grad_norm": 45.741024017333984,
+      "learning_rate": 1.795911111111111e-05,
+      "loss": 1.7655,
+      "step": 11990
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 45.897953033447266,
+      "learning_rate": 1.7955555555555556e-05,
+      "loss": 1.7656,
+      "step": 12000
+    },
+    {
+      "epoch": 0.38432,
+      "grad_norm": 45.622100830078125,
+      "learning_rate": 1.7952e-05,
+      "loss": 1.762,
+      "step": 12010
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 46.58403396606445,
+      "learning_rate": 1.7948444444444445e-05,
+      "loss": 1.7246,
+      "step": 12020
+    },
+    {
+      "epoch": 0.38496,
+      "grad_norm": 45.18146514892578,
+      "learning_rate": 1.794488888888889e-05,
+      "loss": 1.7601,
+      "step": 12030
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 45.685115814208984,
+      "learning_rate": 1.7941333333333334e-05,
+      "loss": 1.7703,
+      "step": 12040
+    },
+    {
+      "epoch": 0.3856,
+      "grad_norm": 45.952911376953125,
+      "learning_rate": 1.793777777777778e-05,
+      "loss": 1.7499,
+      "step": 12050
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 45.1624755859375,
+      "learning_rate": 1.7934222222222223e-05,
+      "loss": 1.767,
+      "step": 12060
+    },
+    {
+      "epoch": 0.38624,
+      "grad_norm": 46.44557571411133,
+      "learning_rate": 1.7930666666666668e-05,
+      "loss": 1.7551,
+      "step": 12070
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 46.18535232543945,
+      "learning_rate": 1.7927111111111112e-05,
+      "loss": 1.7062,
+      "step": 12080
+    },
+    {
+      "epoch": 0.38688,
+      "grad_norm": 46.5549430847168,
+      "learning_rate": 1.7923555555555557e-05,
+      "loss": 1.739,
+      "step": 12090
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 45.92256164550781,
+      "learning_rate": 1.792e-05,
+      "loss": 1.7616,
+      "step": 12100
+    },
+    {
+      "epoch": 0.38752,
+      "grad_norm": 47.67626953125,
+      "learning_rate": 1.7916444444444446e-05,
+      "loss": 1.7356,
+      "step": 12110
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 46.13505935668945,
+      "learning_rate": 1.791288888888889e-05,
+      "loss": 1.7371,
+      "step": 12120
+    },
+    {
+      "epoch": 0.38816,
+      "grad_norm": 46.67782974243164,
+      "learning_rate": 1.7909333333333335e-05,
+      "loss": 1.7342,
+      "step": 12130
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 45.36748123168945,
+      "learning_rate": 1.790577777777778e-05,
+      "loss": 1.7383,
+      "step": 12140
+    },
+    {
+      "epoch": 0.3888,
+      "grad_norm": 47.25735092163086,
+      "learning_rate": 1.7902222222222224e-05,
+      "loss": 1.7343,
+      "step": 12150
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 46.73722457885742,
+      "learning_rate": 1.789866666666667e-05,
+      "loss": 1.7394,
+      "step": 12160
+    },
+    {
+      "epoch": 0.38944,
+      "grad_norm": 45.669830322265625,
+      "learning_rate": 1.7895111111111113e-05,
+      "loss": 1.7294,
+      "step": 12170
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 43.606658935546875,
+      "learning_rate": 1.7891555555555558e-05,
+      "loss": 1.741,
+      "step": 12180
+    },
+    {
+      "epoch": 0.39008,
+      "grad_norm": 46.752349853515625,
+      "learning_rate": 1.7888000000000002e-05,
+      "loss": 1.7201,
+      "step": 12190
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 45.13085174560547,
+      "learning_rate": 1.7884444444444447e-05,
+      "loss": 1.6933,
+      "step": 12200
+    },
+    {
+      "epoch": 0.39072,
+      "grad_norm": 47.74692916870117,
+      "learning_rate": 1.788088888888889e-05,
+      "loss": 1.7298,
+      "step": 12210
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 47.35121536254883,
+      "learning_rate": 1.7877333333333336e-05,
+      "loss": 1.7401,
+      "step": 12220
+    },
+    {
+      "epoch": 0.39136,
+      "grad_norm": 47.107688903808594,
+      "learning_rate": 1.787377777777778e-05,
+      "loss": 1.736,
+      "step": 12230
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 46.06573486328125,
+      "learning_rate": 1.7870222222222225e-05,
+      "loss": 1.7439,
+      "step": 12240
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 44.885154724121094,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.7402,
+      "step": 12250
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 44.785648345947266,
+      "learning_rate": 1.7863111111111114e-05,
+      "loss": 1.7475,
+      "step": 12260
+    },
+    {
+      "epoch": 0.39264,
+      "grad_norm": 47.50531768798828,
+      "learning_rate": 1.7859555555555555e-05,
+      "loss": 1.7689,
+      "step": 12270
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 46.02947998046875,
+      "learning_rate": 1.7856000000000003e-05,
+      "loss": 1.7424,
+      "step": 12280
+    },
+    {
+      "epoch": 0.39328,
+      "grad_norm": 47.10367965698242,
+      "learning_rate": 1.7852444444444444e-05,
+      "loss": 1.7345,
+      "step": 12290
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 49.23219680786133,
+      "learning_rate": 1.7848888888888892e-05,
+      "loss": 1.7514,
+      "step": 12300
+    },
+    {
+      "epoch": 0.39392,
+      "grad_norm": 47.4083251953125,
+      "learning_rate": 1.7845333333333333e-05,
+      "loss": 1.751,
+      "step": 12310
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 47.23759078979492,
+      "learning_rate": 1.784177777777778e-05,
+      "loss": 1.7486,
+      "step": 12320
+    },
+    {
+      "epoch": 0.39456,
+      "grad_norm": 47.71400451660156,
+      "learning_rate": 1.7838222222222223e-05,
+      "loss": 1.7515,
+      "step": 12330
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 47.39936065673828,
+      "learning_rate": 1.783466666666667e-05,
+      "loss": 1.7542,
+      "step": 12340
+    },
+    {
+      "epoch": 0.3952,
+      "grad_norm": 47.184940338134766,
+      "learning_rate": 1.783111111111111e-05,
+      "loss": 1.7303,
+      "step": 12350
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 43.96541976928711,
+      "learning_rate": 1.782755555555556e-05,
+      "loss": 1.7178,
+      "step": 12360
+    },
+    {
+      "epoch": 0.39584,
+      "grad_norm": 48.6561393737793,
+      "learning_rate": 1.7824e-05,
+      "loss": 1.7481,
+      "step": 12370
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 45.21891784667969,
+      "learning_rate": 1.7820444444444445e-05,
+      "loss": 1.7518,
+      "step": 12380
+    },
+    {
+      "epoch": 0.39648,
+      "grad_norm": 44.285064697265625,
+      "learning_rate": 1.781688888888889e-05,
+      "loss": 1.7334,
+      "step": 12390
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 45.4031867980957,
+      "learning_rate": 1.7813333333333334e-05,
+      "loss": 1.7392,
+      "step": 12400
+    },
+    {
+      "epoch": 0.39712,
+      "grad_norm": 46.44404983520508,
+      "learning_rate": 1.780977777777778e-05,
+      "loss": 1.7558,
+      "step": 12410
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 46.931880950927734,
+      "learning_rate": 1.7806222222222224e-05,
+      "loss": 1.757,
+      "step": 12420
+    },
+    {
+      "epoch": 0.39776,
+      "grad_norm": 45.4690055847168,
+      "learning_rate": 1.7802666666666668e-05,
+      "loss": 1.7441,
+      "step": 12430
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 47.687198638916016,
+      "learning_rate": 1.7799111111111113e-05,
+      "loss": 1.7388,
+      "step": 12440
+    },
+    {
+      "epoch": 0.3984,
+      "grad_norm": 45.4809455871582,
+      "learning_rate": 1.7795555555555557e-05,
+      "loss": 1.7613,
+      "step": 12450
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 44.98910140991211,
+      "learning_rate": 1.7792000000000002e-05,
+      "loss": 1.7376,
+      "step": 12460
+    },
+    {
+      "epoch": 0.39904,
+      "grad_norm": 45.352169036865234,
+      "learning_rate": 1.7788444444444446e-05,
+      "loss": 1.7192,
+      "step": 12470
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 45.38097381591797,
+      "learning_rate": 1.778488888888889e-05,
+      "loss": 1.7749,
+      "step": 12480
+    },
+    {
+      "epoch": 0.39968,
+      "grad_norm": 47.00810241699219,
+      "learning_rate": 1.7781333333333335e-05,
+      "loss": 1.7183,
+      "step": 12490
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 46.29263687133789,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.6969,
+      "step": 12500
+    },
+    {
+      "epoch": 0.40032,
+      "grad_norm": 46.169193267822266,
+      "learning_rate": 1.7774222222222225e-05,
+      "loss": 1.7214,
+      "step": 12510
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 47.48134231567383,
+      "learning_rate": 1.7770666666666666e-05,
+      "loss": 1.7373,
+      "step": 12520
+    },
+    {
+      "epoch": 0.40096,
+      "grad_norm": 46.51030731201172,
+      "learning_rate": 1.7767111111111114e-05,
+      "loss": 1.734,
+      "step": 12530
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 46.44158172607422,
+      "learning_rate": 1.7763555555555555e-05,
+      "loss": 1.7533,
+      "step": 12540
+    },
+    {
+      "epoch": 0.4016,
+      "grad_norm": 44.81826400756836,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 1.7474,
+      "step": 12550
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 47.05168533325195,
+      "learning_rate": 1.7756444444444444e-05,
+      "loss": 1.7421,
+      "step": 12560
+    },
+    {
+      "epoch": 0.40224,
+      "grad_norm": 45.72329330444336,
+      "learning_rate": 1.7752888888888892e-05,
+      "loss": 1.7243,
+      "step": 12570
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 48.39426040649414,
+      "learning_rate": 1.7749333333333333e-05,
+      "loss": 1.7181,
+      "step": 12580
+    },
+    {
+      "epoch": 0.40288,
+      "grad_norm": 45.663997650146484,
+      "learning_rate": 1.774577777777778e-05,
+      "loss": 1.7138,
+      "step": 12590
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 46.39791488647461,
+      "learning_rate": 1.7742222222222222e-05,
+      "loss": 1.7452,
+      "step": 12600
+    },
+    {
+      "epoch": 0.40352,
+      "grad_norm": 45.58061218261719,
+      "learning_rate": 1.773866666666667e-05,
+      "loss": 1.7456,
+      "step": 12610
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 47.24848175048828,
+      "learning_rate": 1.773511111111111e-05,
+      "loss": 1.7502,
+      "step": 12620
+    },
+    {
+      "epoch": 0.40416,
+      "grad_norm": 45.21625900268555,
+      "learning_rate": 1.7731555555555556e-05,
+      "loss": 1.7408,
+      "step": 12630
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 43.57041931152344,
+      "learning_rate": 1.7728e-05,
+      "loss": 1.7407,
+      "step": 12640
+    },
+    {
+      "epoch": 0.4048,
+      "grad_norm": 45.896942138671875,
+      "learning_rate": 1.7724444444444445e-05,
+      "loss": 1.7327,
+      "step": 12650
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 45.211212158203125,
+      "learning_rate": 1.772088888888889e-05,
+      "loss": 1.7506,
+      "step": 12660
+    },
+    {
+      "epoch": 0.40544,
+      "grad_norm": 45.28093719482422,
+      "learning_rate": 1.7717333333333334e-05,
+      "loss": 1.7308,
+      "step": 12670
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 44.34235382080078,
+      "learning_rate": 1.771377777777778e-05,
+      "loss": 1.7177,
+      "step": 12680
+    },
+    {
+      "epoch": 0.40608,
+      "grad_norm": 44.18003463745117,
+      "learning_rate": 1.7710222222222223e-05,
+      "loss": 1.7427,
+      "step": 12690
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 45.78373336791992,
+      "learning_rate": 1.7706666666666668e-05,
+      "loss": 1.736,
+      "step": 12700
+    },
+    {
+      "epoch": 0.40672,
+      "grad_norm": 47.60939025878906,
+      "learning_rate": 1.7703111111111112e-05,
+      "loss": 1.7644,
+      "step": 12710
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 47.69173812866211,
+      "learning_rate": 1.7699555555555557e-05,
+      "loss": 1.7165,
+      "step": 12720
+    },
+    {
+      "epoch": 0.40736,
+      "grad_norm": 47.19498825073242,
+      "learning_rate": 1.7696e-05,
+      "loss": 1.7511,
+      "step": 12730
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 45.9237060546875,
+      "learning_rate": 1.7692444444444446e-05,
+      "loss": 1.7394,
+      "step": 12740
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 48.24811553955078,
+      "learning_rate": 1.768888888888889e-05,
+      "loss": 1.7615,
+      "step": 12750
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 46.84287643432617,
+      "learning_rate": 1.7685333333333335e-05,
+      "loss": 1.7241,
+      "step": 12760
+    },
+    {
+      "epoch": 0.40864,
+      "grad_norm": 44.28770065307617,
+      "learning_rate": 1.768177777777778e-05,
+      "loss": 1.7564,
+      "step": 12770
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 46.839080810546875,
+      "learning_rate": 1.7678222222222224e-05,
+      "loss": 1.7411,
+      "step": 12780
+    },
+    {
+      "epoch": 0.40928,
+      "grad_norm": 47.01158905029297,
+      "learning_rate": 1.767466666666667e-05,
+      "loss": 1.7154,
+      "step": 12790
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 47.18655776977539,
+      "learning_rate": 1.7671111111111113e-05,
+      "loss": 1.7384,
+      "step": 12800
+    },
+    {
+      "epoch": 0.40992,
+      "grad_norm": 43.830631256103516,
+      "learning_rate": 1.7667555555555558e-05,
+      "loss": 1.7559,
+      "step": 12810
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 46.170230865478516,
+      "learning_rate": 1.7664000000000002e-05,
+      "loss": 1.7664,
+      "step": 12820
+    },
+    {
+      "epoch": 0.41056,
+      "grad_norm": 44.55336380004883,
+      "learning_rate": 1.7660444444444447e-05,
+      "loss": 1.746,
+      "step": 12830
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 44.43898010253906,
+      "learning_rate": 1.765688888888889e-05,
+      "loss": 1.7448,
+      "step": 12840
+    },
+    {
+      "epoch": 0.4112,
+      "grad_norm": 46.56974411010742,
+      "learning_rate": 1.7653333333333336e-05,
+      "loss": 1.7177,
+      "step": 12850
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 45.02238845825195,
+      "learning_rate": 1.764977777777778e-05,
+      "loss": 1.7224,
+      "step": 12860
+    },
+    {
+      "epoch": 0.41184,
+      "grad_norm": 43.810420989990234,
+      "learning_rate": 1.7646222222222225e-05,
+      "loss": 1.7423,
+      "step": 12870
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 45.10750961303711,
+      "learning_rate": 1.7642666666666666e-05,
+      "loss": 1.7277,
+      "step": 12880
+    },
+    {
+      "epoch": 0.41248,
+      "grad_norm": 44.9813232421875,
+      "learning_rate": 1.7639111111111114e-05,
+      "loss": 1.7396,
+      "step": 12890
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 43.90498352050781,
+      "learning_rate": 1.7635555555555555e-05,
+      "loss": 1.7491,
+      "step": 12900
+    },
+    {
+      "epoch": 0.41312,
+      "grad_norm": 45.96403503417969,
+      "learning_rate": 1.7632000000000003e-05,
+      "loss": 1.7272,
+      "step": 12910
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 48.598793029785156,
+      "learning_rate": 1.7628444444444444e-05,
+      "loss": 1.7497,
+      "step": 12920
+    },
+    {
+      "epoch": 0.41376,
+      "grad_norm": 47.302711486816406,
+      "learning_rate": 1.7624888888888892e-05,
+      "loss": 1.7199,
+      "step": 12930
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 44.98503112792969,
+      "learning_rate": 1.7621333333333334e-05,
+      "loss": 1.717,
+      "step": 12940
+    },
+    {
+      "epoch": 0.4144,
+      "grad_norm": 45.62991714477539,
+      "learning_rate": 1.761777777777778e-05,
+      "loss": 1.7229,
+      "step": 12950
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 47.17755126953125,
+      "learning_rate": 1.7614222222222223e-05,
+      "loss": 1.7291,
+      "step": 12960
+    },
+    {
+      "epoch": 0.41504,
+      "grad_norm": 47.77396011352539,
+      "learning_rate": 1.7610666666666667e-05,
+      "loss": 1.7388,
+      "step": 12970
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 45.62364196777344,
+      "learning_rate": 1.7607111111111112e-05,
+      "loss": 1.7444,
+      "step": 12980
+    },
+    {
+      "epoch": 0.41568,
+      "grad_norm": 46.032386779785156,
+      "learning_rate": 1.7603555555555556e-05,
+      "loss": 1.7229,
+      "step": 12990
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 44.89250946044922,
+      "learning_rate": 1.76e-05,
+      "loss": 1.748,
+      "step": 13000
+    },
+    {
+      "epoch": 0.41632,
+      "grad_norm": 45.50238800048828,
+      "learning_rate": 1.7596444444444445e-05,
+      "loss": 1.7233,
+      "step": 13010
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 44.66646194458008,
+      "learning_rate": 1.759288888888889e-05,
+      "loss": 1.7281,
+      "step": 13020
+    },
+    {
+      "epoch": 0.41696,
+      "grad_norm": 45.436370849609375,
+      "learning_rate": 1.7589333333333335e-05,
+      "loss": 1.7203,
+      "step": 13030
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 44.28810119628906,
+      "learning_rate": 1.758577777777778e-05,
+      "loss": 1.7491,
+      "step": 13040
+    },
+    {
+      "epoch": 0.4176,
+      "grad_norm": 45.3896369934082,
+      "learning_rate": 1.7582222222222224e-05,
+      "loss": 1.7098,
+      "step": 13050
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 46.0196533203125,
+      "learning_rate": 1.7578666666666668e-05,
+      "loss": 1.7428,
+      "step": 13060
+    },
+    {
+      "epoch": 0.41824,
+      "grad_norm": 48.799415588378906,
+      "learning_rate": 1.7575111111111113e-05,
+      "loss": 1.7505,
+      "step": 13070
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 46.70732116699219,
+      "learning_rate": 1.7571555555555557e-05,
+      "loss": 1.7166,
+      "step": 13080
+    },
+    {
+      "epoch": 0.41888,
+      "grad_norm": 43.49319839477539,
+      "learning_rate": 1.7568000000000002e-05,
+      "loss": 1.7557,
+      "step": 13090
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 46.664398193359375,
+      "learning_rate": 1.7564444444444446e-05,
+      "loss": 1.7581,
+      "step": 13100
+    },
+    {
+      "epoch": 0.41952,
+      "grad_norm": 48.30585479736328,
+      "learning_rate": 1.756088888888889e-05,
+      "loss": 1.7517,
+      "step": 13110
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 45.69898223876953,
+      "learning_rate": 1.7557333333333336e-05,
+      "loss": 1.7352,
+      "step": 13120
+    },
+    {
+      "epoch": 0.42016,
+      "grad_norm": 44.632633209228516,
+      "learning_rate": 1.755377777777778e-05,
+      "loss": 1.7386,
+      "step": 13130
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 46.91196060180664,
+      "learning_rate": 1.7550222222222225e-05,
+      "loss": 1.729,
+      "step": 13140
+    },
+    {
+      "epoch": 0.4208,
+      "grad_norm": 45.67945861816406,
+      "learning_rate": 1.7546666666666666e-05,
+      "loss": 1.7209,
+      "step": 13150
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 46.89979934692383,
+      "learning_rate": 1.7543111111111114e-05,
+      "loss": 1.7585,
+      "step": 13160
+    },
+    {
+      "epoch": 0.42144,
+      "grad_norm": 47.15142059326172,
+      "learning_rate": 1.7539555555555555e-05,
+      "loss": 1.7552,
+      "step": 13170
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 46.78049850463867,
+      "learning_rate": 1.7536000000000003e-05,
+      "loss": 1.7493,
+      "step": 13180
+    },
+    {
+      "epoch": 0.42208,
+      "grad_norm": 45.9544563293457,
+      "learning_rate": 1.7532444444444444e-05,
+      "loss": 1.7206,
+      "step": 13190
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 45.94736099243164,
+      "learning_rate": 1.7528888888888892e-05,
+      "loss": 1.7026,
+      "step": 13200
+    },
+    {
+      "epoch": 0.42272,
+      "grad_norm": 44.32487106323242,
+      "learning_rate": 1.7525333333333333e-05,
+      "loss": 1.7201,
+      "step": 13210
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 45.86711883544922,
+      "learning_rate": 1.752177777777778e-05,
+      "loss": 1.7331,
+      "step": 13220
+    },
+    {
+      "epoch": 0.42336,
+      "grad_norm": 44.53461837768555,
+      "learning_rate": 1.7518222222222222e-05,
+      "loss": 1.7525,
+      "step": 13230
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 44.8212776184082,
+      "learning_rate": 1.751466666666667e-05,
+      "loss": 1.758,
+      "step": 13240
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 45.448646545410156,
+      "learning_rate": 1.751111111111111e-05,
+      "loss": 1.7499,
+      "step": 13250
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 48.57109069824219,
+      "learning_rate": 1.750755555555556e-05,
+      "loss": 1.7272,
+      "step": 13260
+    },
+    {
+      "epoch": 0.42464,
+      "grad_norm": 44.58144760131836,
+      "learning_rate": 1.7504e-05,
+      "loss": 1.7274,
+      "step": 13270
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 44.63924026489258,
+      "learning_rate": 1.7500444444444445e-05,
+      "loss": 1.7055,
+      "step": 13280
+    },
+    {
+      "epoch": 0.42528,
+      "grad_norm": 46.89718246459961,
+      "learning_rate": 1.749688888888889e-05,
+      "loss": 1.7027,
+      "step": 13290
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 43.74995422363281,
+      "learning_rate": 1.7493333333333334e-05,
+      "loss": 1.7279,
+      "step": 13300
+    },
+    {
+      "epoch": 0.42592,
+      "grad_norm": 44.514400482177734,
+      "learning_rate": 1.748977777777778e-05,
+      "loss": 1.7189,
+      "step": 13310
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 45.729801177978516,
+      "learning_rate": 1.7486222222222223e-05,
+      "loss": 1.7564,
+      "step": 13320
+    },
+    {
+      "epoch": 0.42656,
+      "grad_norm": 46.24195861816406,
+      "learning_rate": 1.7482666666666668e-05,
+      "loss": 1.7214,
+      "step": 13330
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 43.434078216552734,
+      "learning_rate": 1.7479111111111112e-05,
+      "loss": 1.7219,
+      "step": 13340
+    },
+    {
+      "epoch": 0.4272,
+      "grad_norm": 46.25753402709961,
+      "learning_rate": 1.7475555555555557e-05,
+      "loss": 1.7537,
+      "step": 13350
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 46.87751388549805,
+      "learning_rate": 1.7472e-05,
+      "loss": 1.7413,
+      "step": 13360
+    },
+    {
+      "epoch": 0.42784,
+      "grad_norm": 45.92871856689453,
+      "learning_rate": 1.7468444444444446e-05,
+      "loss": 1.7182,
+      "step": 13370
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 45.935028076171875,
+      "learning_rate": 1.746488888888889e-05,
+      "loss": 1.7382,
+      "step": 13380
+    },
+    {
+      "epoch": 0.42848,
+      "grad_norm": 49.45673370361328,
+      "learning_rate": 1.7461333333333335e-05,
+      "loss": 1.7248,
+      "step": 13390
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 46.06485366821289,
+      "learning_rate": 1.745777777777778e-05,
+      "loss": 1.7203,
+      "step": 13400
+    },
+    {
+      "epoch": 0.42912,
+      "grad_norm": 44.7425537109375,
+      "learning_rate": 1.7454222222222224e-05,
+      "loss": 1.741,
+      "step": 13410
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 45.23655700683594,
+      "learning_rate": 1.745066666666667e-05,
+      "loss": 1.7522,
+      "step": 13420
+    },
+    {
+      "epoch": 0.42976,
+      "grad_norm": 44.002384185791016,
+      "learning_rate": 1.7447111111111113e-05,
+      "loss": 1.7217,
+      "step": 13430
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 46.94925308227539,
+      "learning_rate": 1.7443555555555558e-05,
+      "loss": 1.7299,
+      "step": 13440
+    },
+    {
+      "epoch": 0.4304,
+      "grad_norm": 45.58026123046875,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 1.7075,
+      "step": 13450
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 52.205322265625,
+      "learning_rate": 1.7436444444444447e-05,
+      "loss": 1.732,
+      "step": 13460
+    },
+    {
+      "epoch": 0.43104,
+      "grad_norm": 42.87519836425781,
+      "learning_rate": 1.743288888888889e-05,
+      "loss": 1.7388,
+      "step": 13470
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 45.98421096801758,
+      "learning_rate": 1.7429333333333333e-05,
+      "loss": 1.7596,
+      "step": 13480
+    },
+    {
+      "epoch": 0.43168,
+      "grad_norm": 45.058380126953125,
+      "learning_rate": 1.742577777777778e-05,
+      "loss": 1.7358,
+      "step": 13490
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 45.228031158447266,
+      "learning_rate": 1.7422222222222222e-05,
+      "loss": 1.7616,
+      "step": 13500
+    },
+    {
+      "epoch": 0.43232,
+      "grad_norm": 45.51875305175781,
+      "learning_rate": 1.741866666666667e-05,
+      "loss": 1.7531,
+      "step": 13510
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 46.320167541503906,
+      "learning_rate": 1.741511111111111e-05,
+      "loss": 1.7092,
+      "step": 13520
+    },
+    {
+      "epoch": 0.43296,
+      "grad_norm": 44.57635498046875,
+      "learning_rate": 1.7411555555555555e-05,
+      "loss": 1.7187,
+      "step": 13530
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 44.553035736083984,
+      "learning_rate": 1.7408e-05,
+      "loss": 1.7103,
+      "step": 13540
+    },
+    {
+      "epoch": 0.4336,
+      "grad_norm": 45.63328170776367,
+      "learning_rate": 1.7404444444444445e-05,
+      "loss": 1.7485,
+      "step": 13550
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 46.18551254272461,
+      "learning_rate": 1.740088888888889e-05,
+      "loss": 1.7419,
+      "step": 13560
+    },
+    {
+      "epoch": 0.43424,
+      "grad_norm": 44.467525482177734,
+      "learning_rate": 1.7397333333333334e-05,
+      "loss": 1.7,
+      "step": 13570
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 46.38378143310547,
+      "learning_rate": 1.7393777777777778e-05,
+      "loss": 1.7244,
+      "step": 13580
+    },
+    {
+      "epoch": 0.43488,
+      "grad_norm": 45.0339469909668,
+      "learning_rate": 1.7390222222222223e-05,
+      "loss": 1.7539,
+      "step": 13590
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 44.02116394042969,
+      "learning_rate": 1.7386666666666667e-05,
+      "loss": 1.7397,
+      "step": 13600
+    },
+    {
+      "epoch": 0.43552,
+      "grad_norm": 49.48076248168945,
+      "learning_rate": 1.7383111111111112e-05,
+      "loss": 1.7492,
+      "step": 13610
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 45.204132080078125,
+      "learning_rate": 1.7379555555555556e-05,
+      "loss": 1.7168,
+      "step": 13620
+    },
+    {
+      "epoch": 0.43616,
+      "grad_norm": 46.511878967285156,
+      "learning_rate": 1.7376e-05,
+      "loss": 1.7473,
+      "step": 13630
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 45.37043762207031,
+      "learning_rate": 1.7372444444444446e-05,
+      "loss": 1.7164,
+      "step": 13640
+    },
+    {
+      "epoch": 0.4368,
+      "grad_norm": 44.92662048339844,
+      "learning_rate": 1.736888888888889e-05,
+      "loss": 1.7352,
+      "step": 13650
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 45.94477844238281,
+      "learning_rate": 1.7365333333333335e-05,
+      "loss": 1.7325,
+      "step": 13660
+    },
+    {
+      "epoch": 0.43744,
+      "grad_norm": 44.44512176513672,
+      "learning_rate": 1.736177777777778e-05,
+      "loss": 1.7126,
+      "step": 13670
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 44.867652893066406,
+      "learning_rate": 1.7358222222222224e-05,
+      "loss": 1.7423,
+      "step": 13680
+    },
+    {
+      "epoch": 0.43808,
+      "grad_norm": 42.79388427734375,
+      "learning_rate": 1.735466666666667e-05,
+      "loss": 1.7241,
+      "step": 13690
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 45.237342834472656,
+      "learning_rate": 1.7351111111111113e-05,
+      "loss": 1.7428,
+      "step": 13700
+    },
+    {
+      "epoch": 0.43872,
+      "grad_norm": 45.06656265258789,
+      "learning_rate": 1.7347555555555557e-05,
+      "loss": 1.7279,
+      "step": 13710
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 44.70061111450195,
+      "learning_rate": 1.7344000000000002e-05,
+      "loss": 1.7414,
+      "step": 13720
+    },
+    {
+      "epoch": 0.43936,
+      "grad_norm": 43.81660842895508,
+      "learning_rate": 1.7340444444444447e-05,
+      "loss": 1.7378,
+      "step": 13730
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 45.633880615234375,
+      "learning_rate": 1.733688888888889e-05,
+      "loss": 1.7447,
+      "step": 13740
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 46.27029800415039,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.7176,
+      "step": 13750
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 47.65945053100586,
+      "learning_rate": 1.732977777777778e-05,
+      "loss": 1.7371,
+      "step": 13760
+    },
+    {
+      "epoch": 0.44064,
+      "grad_norm": 45.97589111328125,
+      "learning_rate": 1.7326222222222225e-05,
+      "loss": 1.6964,
+      "step": 13770
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 44.704036712646484,
+      "learning_rate": 1.732266666666667e-05,
+      "loss": 1.7066,
+      "step": 13780
+    },
+    {
+      "epoch": 0.44128,
+      "grad_norm": 46.602542877197266,
+      "learning_rate": 1.7319111111111114e-05,
+      "loss": 1.7192,
+      "step": 13790
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 46.97726821899414,
+      "learning_rate": 1.7315555555555555e-05,
+      "loss": 1.7008,
+      "step": 13800
+    },
+    {
+      "epoch": 0.44192,
+      "grad_norm": 46.14274978637695,
+      "learning_rate": 1.7312000000000003e-05,
+      "loss": 1.7701,
+      "step": 13810
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 45.656639099121094,
+      "learning_rate": 1.7308444444444444e-05,
+      "loss": 1.7404,
+      "step": 13820
+    },
+    {
+      "epoch": 0.44256,
+      "grad_norm": 45.84688949584961,
+      "learning_rate": 1.7304888888888892e-05,
+      "loss": 1.7281,
+      "step": 13830
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 44.40279006958008,
+      "learning_rate": 1.7301333333333333e-05,
+      "loss": 1.7412,
+      "step": 13840
+    },
+    {
+      "epoch": 0.4432,
+      "grad_norm": 44.84839630126953,
+      "learning_rate": 1.729777777777778e-05,
+      "loss": 1.7568,
+      "step": 13850
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 45.91680145263672,
+      "learning_rate": 1.7294222222222222e-05,
+      "loss": 1.7247,
+      "step": 13860
+    },
+    {
+      "epoch": 0.44384,
+      "grad_norm": 43.44932556152344,
+      "learning_rate": 1.729066666666667e-05,
+      "loss": 1.7638,
+      "step": 13870
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 45.238861083984375,
+      "learning_rate": 1.728711111111111e-05,
+      "loss": 1.7215,
+      "step": 13880
+    },
+    {
+      "epoch": 0.44448,
+      "grad_norm": 45.49114990234375,
+      "learning_rate": 1.728355555555556e-05,
+      "loss": 1.7085,
+      "step": 13890
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 44.86478805541992,
+      "learning_rate": 1.728e-05,
+      "loss": 1.713,
+      "step": 13900
+    },
+    {
+      "epoch": 0.44512,
+      "grad_norm": 45.341129302978516,
+      "learning_rate": 1.7276444444444445e-05,
+      "loss": 1.7263,
+      "step": 13910
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 44.82181167602539,
+      "learning_rate": 1.727288888888889e-05,
+      "loss": 1.7458,
+      "step": 13920
+    },
+    {
+      "epoch": 0.44576,
+      "grad_norm": 42.75132751464844,
+      "learning_rate": 1.7269333333333334e-05,
+      "loss": 1.7263,
+      "step": 13930
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 46.062068939208984,
+      "learning_rate": 1.726577777777778e-05,
+      "loss": 1.7592,
+      "step": 13940
+    },
+    {
+      "epoch": 0.4464,
+      "grad_norm": 43.95463562011719,
+      "learning_rate": 1.7262222222222223e-05,
+      "loss": 1.7476,
+      "step": 13950
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 44.0671272277832,
+      "learning_rate": 1.7258666666666668e-05,
+      "loss": 1.7189,
+      "step": 13960
+    },
+    {
+      "epoch": 0.44704,
+      "grad_norm": 46.24520492553711,
+      "learning_rate": 1.7255111111111112e-05,
+      "loss": 1.7354,
+      "step": 13970
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 43.28176498413086,
+      "learning_rate": 1.7251555555555557e-05,
+      "loss": 1.6807,
+      "step": 13980
+    },
+    {
+      "epoch": 0.44768,
+      "grad_norm": 44.17888259887695,
+      "learning_rate": 1.7248e-05,
+      "loss": 1.7,
+      "step": 13990
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 45.7882080078125,
+      "learning_rate": 1.7244444444444446e-05,
+      "loss": 1.7423,
+      "step": 14000
+    },
+    {
+      "epoch": 0.44832,
+      "grad_norm": 47.963966369628906,
+      "learning_rate": 1.724088888888889e-05,
+      "loss": 1.753,
+      "step": 14010
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 47.81916427612305,
+      "learning_rate": 1.7237333333333335e-05,
+      "loss": 1.7314,
+      "step": 14020
+    },
+    {
+      "epoch": 0.44896,
+      "grad_norm": 43.19534683227539,
+      "learning_rate": 1.723377777777778e-05,
+      "loss": 1.7115,
+      "step": 14030
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 44.247955322265625,
+      "learning_rate": 1.7230222222222224e-05,
+      "loss": 1.7209,
+      "step": 14040
+    },
+    {
+      "epoch": 0.4496,
+      "grad_norm": 45.006771087646484,
+      "learning_rate": 1.7226666666666665e-05,
+      "loss": 1.6751,
+      "step": 14050
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 45.61868667602539,
+      "learning_rate": 1.7223111111111113e-05,
+      "loss": 1.7024,
+      "step": 14060
+    },
+    {
+      "epoch": 0.45024,
+      "grad_norm": 43.44654083251953,
+      "learning_rate": 1.7219555555555555e-05,
+      "loss": 1.6982,
+      "step": 14070
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 45.38816833496094,
+      "learning_rate": 1.7216000000000003e-05,
+      "loss": 1.7487,
+      "step": 14080
+    },
+    {
+      "epoch": 0.45088,
+      "grad_norm": 46.881126403808594,
+      "learning_rate": 1.7212444444444444e-05,
+      "loss": 1.6988,
+      "step": 14090
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 45.429744720458984,
+      "learning_rate": 1.720888888888889e-05,
+      "loss": 1.7508,
+      "step": 14100
+    },
+    {
+      "epoch": 0.45152,
+      "grad_norm": 46.08954620361328,
+      "learning_rate": 1.7205333333333333e-05,
+      "loss": 1.7342,
+      "step": 14110
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 43.329837799072266,
+      "learning_rate": 1.720177777777778e-05,
+      "loss": 1.7219,
+      "step": 14120
+    },
+    {
+      "epoch": 0.45216,
+      "grad_norm": 45.99585723876953,
+      "learning_rate": 1.7198222222222222e-05,
+      "loss": 1.7176,
+      "step": 14130
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 43.80538558959961,
+      "learning_rate": 1.719466666666667e-05,
+      "loss": 1.7207,
+      "step": 14140
+    },
+    {
+      "epoch": 0.4528,
+      "grad_norm": 43.61793518066406,
+      "learning_rate": 1.719111111111111e-05,
+      "loss": 1.706,
+      "step": 14150
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 46.47148513793945,
+      "learning_rate": 1.718755555555556e-05,
+      "loss": 1.7146,
+      "step": 14160
+    },
+    {
+      "epoch": 0.45344,
+      "grad_norm": 48.728145599365234,
+      "learning_rate": 1.7184e-05,
+      "loss": 1.7175,
+      "step": 14170
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 43.15789031982422,
+      "learning_rate": 1.7180444444444445e-05,
+      "loss": 1.7148,
+      "step": 14180
+    },
+    {
+      "epoch": 0.45408,
+      "grad_norm": 46.423492431640625,
+      "learning_rate": 1.717688888888889e-05,
+      "loss": 1.725,
+      "step": 14190
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 46.454776763916016,
+      "learning_rate": 1.7173333333333334e-05,
+      "loss": 1.6916,
+      "step": 14200
+    },
+    {
+      "epoch": 0.45472,
+      "grad_norm": 48.568687438964844,
+      "learning_rate": 1.716977777777778e-05,
+      "loss": 1.7083,
+      "step": 14210
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 44.02393341064453,
+      "learning_rate": 1.7166222222222223e-05,
+      "loss": 1.7551,
+      "step": 14220
+    },
+    {
+      "epoch": 0.45536,
+      "grad_norm": 47.03176498413086,
+      "learning_rate": 1.7162666666666667e-05,
+      "loss": 1.7424,
+      "step": 14230
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 43.364501953125,
+      "learning_rate": 1.7159111111111112e-05,
+      "loss": 1.7158,
+      "step": 14240
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 45.3157844543457,
+      "learning_rate": 1.7155555555555557e-05,
+      "loss": 1.7286,
+      "step": 14250
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 45.80545425415039,
+      "learning_rate": 1.7152e-05,
+      "loss": 1.7186,
+      "step": 14260
+    },
+    {
+      "epoch": 0.45664,
+      "grad_norm": 43.9619140625,
+      "learning_rate": 1.7148444444444446e-05,
+      "loss": 1.7541,
+      "step": 14270
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 45.12228012084961,
+      "learning_rate": 1.714488888888889e-05,
+      "loss": 1.7254,
+      "step": 14280
+    },
+    {
+      "epoch": 0.45728,
+      "grad_norm": 47.905296325683594,
+      "learning_rate": 1.7141333333333335e-05,
+      "loss": 1.7621,
+      "step": 14290
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 45.022823333740234,
+      "learning_rate": 1.713777777777778e-05,
+      "loss": 1.7174,
+      "step": 14300
+    },
+    {
+      "epoch": 0.45792,
+      "grad_norm": 43.46538162231445,
+      "learning_rate": 1.7134222222222224e-05,
+      "loss": 1.6973,
+      "step": 14310
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 46.357234954833984,
+      "learning_rate": 1.713066666666667e-05,
+      "loss": 1.731,
+      "step": 14320
+    },
+    {
+      "epoch": 0.45856,
+      "grad_norm": 45.371124267578125,
+      "learning_rate": 1.7127111111111113e-05,
+      "loss": 1.7153,
+      "step": 14330
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 44.28989028930664,
+      "learning_rate": 1.7123555555555558e-05,
+      "loss": 1.6944,
+      "step": 14340
+    },
+    {
+      "epoch": 0.4592,
+      "grad_norm": 45.44322967529297,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 1.7328,
+      "step": 14350
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 45.2812614440918,
+      "learning_rate": 1.7116444444444447e-05,
+      "loss": 1.7483,
+      "step": 14360
+    },
+    {
+      "epoch": 0.45984,
+      "grad_norm": 44.76524353027344,
+      "learning_rate": 1.711288888888889e-05,
+      "loss": 1.7532,
+      "step": 14370
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 44.747642517089844,
+      "learning_rate": 1.7109333333333336e-05,
+      "loss": 1.7378,
+      "step": 14380
+    },
+    {
+      "epoch": 0.46048,
+      "grad_norm": 46.987571716308594,
+      "learning_rate": 1.710577777777778e-05,
+      "loss": 1.7158,
+      "step": 14390
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 43.8972053527832,
+      "learning_rate": 1.7102222222222225e-05,
+      "loss": 1.6894,
+      "step": 14400
+    },
+    {
+      "epoch": 0.46112,
+      "grad_norm": 44.63825225830078,
+      "learning_rate": 1.709866666666667e-05,
+      "loss": 1.7255,
+      "step": 14410
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 45.58122253417969,
+      "learning_rate": 1.7095111111111114e-05,
+      "loss": 1.7339,
+      "step": 14420
+    },
+    {
+      "epoch": 0.46176,
+      "grad_norm": 45.17179489135742,
+      "learning_rate": 1.7091555555555555e-05,
+      "loss": 1.729,
+      "step": 14430
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 44.413856506347656,
+      "learning_rate": 1.7088000000000003e-05,
+      "loss": 1.6972,
+      "step": 14440
+    },
+    {
+      "epoch": 0.4624,
+      "grad_norm": 46.393951416015625,
+      "learning_rate": 1.7084444444444444e-05,
+      "loss": 1.7067,
+      "step": 14450
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 44.711326599121094,
+      "learning_rate": 1.7080888888888892e-05,
+      "loss": 1.7007,
+      "step": 14460
+    },
+    {
+      "epoch": 0.46304,
+      "grad_norm": 45.364009857177734,
+      "learning_rate": 1.7077333333333333e-05,
+      "loss": 1.728,
+      "step": 14470
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 45.55656433105469,
+      "learning_rate": 1.707377777777778e-05,
+      "loss": 1.7087,
+      "step": 14480
+    },
+    {
+      "epoch": 0.46368,
+      "grad_norm": 44.69358444213867,
+      "learning_rate": 1.7070222222222222e-05,
+      "loss": 1.7276,
+      "step": 14490
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 43.40841293334961,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.7234,
+      "step": 14500
+    },
+    {
+      "epoch": 0.46432,
+      "grad_norm": 46.0664176940918,
+      "learning_rate": 1.706311111111111e-05,
+      "loss": 1.7575,
+      "step": 14510
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 45.3603401184082,
+      "learning_rate": 1.7059555555555556e-05,
+      "loss": 1.7083,
+      "step": 14520
+    },
+    {
+      "epoch": 0.46496,
+      "grad_norm": 43.96553039550781,
+      "learning_rate": 1.7056e-05,
+      "loss": 1.7176,
+      "step": 14530
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 46.85286331176758,
+      "learning_rate": 1.7052444444444445e-05,
+      "loss": 1.7251,
+      "step": 14540
+    },
+    {
+      "epoch": 0.4656,
+      "grad_norm": 45.323936462402344,
+      "learning_rate": 1.704888888888889e-05,
+      "loss": 1.6989,
+      "step": 14550
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 46.17856979370117,
+      "learning_rate": 1.7045333333333334e-05,
+      "loss": 1.7122,
+      "step": 14560
+    },
+    {
+      "epoch": 0.46624,
+      "grad_norm": 45.23008728027344,
+      "learning_rate": 1.704177777777778e-05,
+      "loss": 1.7531,
+      "step": 14570
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 96.77664184570312,
+      "learning_rate": 1.7038222222222223e-05,
+      "loss": 1.7109,
+      "step": 14580
+    },
+    {
+      "epoch": 0.46688,
+      "grad_norm": 48.62037658691406,
+      "learning_rate": 1.7034666666666668e-05,
+      "loss": 1.7294,
+      "step": 14590
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 46.559295654296875,
+      "learning_rate": 1.7031111111111113e-05,
+      "loss": 1.7075,
+      "step": 14600
+    },
+    {
+      "epoch": 0.46752,
+      "grad_norm": 49.41647720336914,
+      "learning_rate": 1.7027555555555557e-05,
+      "loss": 1.7256,
+      "step": 14610
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 46.767372131347656,
+      "learning_rate": 1.7024e-05,
+      "loss": 1.7183,
+      "step": 14620
+    },
+    {
+      "epoch": 0.46816,
+      "grad_norm": 46.65950012207031,
+      "learning_rate": 1.7020444444444446e-05,
+      "loss": 1.7412,
+      "step": 14630
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 46.2630615234375,
+      "learning_rate": 1.701688888888889e-05,
+      "loss": 1.7295,
+      "step": 14640
+    },
+    {
+      "epoch": 0.4688,
+      "grad_norm": 43.83213424682617,
+      "learning_rate": 1.7013333333333335e-05,
+      "loss": 1.7162,
+      "step": 14650
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 44.1187858581543,
+      "learning_rate": 1.700977777777778e-05,
+      "loss": 1.7099,
+      "step": 14660
+    },
+    {
+      "epoch": 0.46944,
+      "grad_norm": 46.37276077270508,
+      "learning_rate": 1.7006222222222224e-05,
+      "loss": 1.6949,
+      "step": 14670
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 44.964237213134766,
+      "learning_rate": 1.700266666666667e-05,
+      "loss": 1.7339,
+      "step": 14680
+    },
+    {
+      "epoch": 0.47008,
+      "grad_norm": 44.88374328613281,
+      "learning_rate": 1.6999111111111114e-05,
+      "loss": 1.7327,
+      "step": 14690
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 46.43785095214844,
+      "learning_rate": 1.6995555555555555e-05,
+      "loss": 1.7196,
+      "step": 14700
+    },
+    {
+      "epoch": 0.47072,
+      "grad_norm": 44.99215316772461,
+      "learning_rate": 1.6992000000000003e-05,
+      "loss": 1.7084,
+      "step": 14710
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 43.4759521484375,
+      "learning_rate": 1.6988444444444444e-05,
+      "loss": 1.6985,
+      "step": 14720
+    },
+    {
+      "epoch": 0.47136,
+      "grad_norm": 43.7542839050293,
+      "learning_rate": 1.6984888888888892e-05,
+      "loss": 1.7298,
+      "step": 14730
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 44.91163635253906,
+      "learning_rate": 1.6981333333333333e-05,
+      "loss": 1.7201,
+      "step": 14740
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 49.6381950378418,
+      "learning_rate": 1.697777777777778e-05,
+      "loss": 1.7273,
+      "step": 14750
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 46.76472473144531,
+      "learning_rate": 1.6974222222222222e-05,
+      "loss": 1.7034,
+      "step": 14760
+    },
+    {
+      "epoch": 0.47264,
+      "grad_norm": 44.74431610107422,
+      "learning_rate": 1.697066666666667e-05,
+      "loss": 1.7266,
+      "step": 14770
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 45.28511428833008,
+      "learning_rate": 1.696711111111111e-05,
+      "loss": 1.7461,
+      "step": 14780
+    },
+    {
+      "epoch": 0.47328,
+      "grad_norm": 44.02510452270508,
+      "learning_rate": 1.696355555555556e-05,
+      "loss": 1.7089,
+      "step": 14790
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 45.76411437988281,
+      "learning_rate": 1.696e-05,
+      "loss": 1.7251,
+      "step": 14800
+    },
+    {
+      "epoch": 0.47392,
+      "grad_norm": 44.46986770629883,
+      "learning_rate": 1.6956444444444445e-05,
+      "loss": 1.6815,
+      "step": 14810
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 45.384033203125,
+      "learning_rate": 1.695288888888889e-05,
+      "loss": 1.7373,
+      "step": 14820
+    },
+    {
+      "epoch": 0.47456,
+      "grad_norm": 43.98371887207031,
+      "learning_rate": 1.6949333333333334e-05,
+      "loss": 1.7153,
+      "step": 14830
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 44.387054443359375,
+      "learning_rate": 1.694577777777778e-05,
+      "loss": 1.7194,
+      "step": 14840
+    },
+    {
+      "epoch": 0.4752,
+      "grad_norm": 44.254051208496094,
+      "learning_rate": 1.6942222222222223e-05,
+      "loss": 1.6953,
+      "step": 14850
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 47.075626373291016,
+      "learning_rate": 1.6938666666666668e-05,
+      "loss": 1.7409,
+      "step": 14860
+    },
+    {
+      "epoch": 0.47584,
+      "grad_norm": 48.30929946899414,
+      "learning_rate": 1.6935111111111112e-05,
+      "loss": 1.7164,
+      "step": 14870
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 45.149024963378906,
+      "learning_rate": 1.6931555555555557e-05,
+      "loss": 1.6862,
+      "step": 14880
+    },
+    {
+      "epoch": 0.47648,
+      "grad_norm": 45.58299255371094,
+      "learning_rate": 1.6928e-05,
+      "loss": 1.6864,
+      "step": 14890
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 44.711151123046875,
+      "learning_rate": 1.6924444444444446e-05,
+      "loss": 1.7183,
+      "step": 14900
+    },
+    {
+      "epoch": 0.47712,
+      "grad_norm": 47.265140533447266,
+      "learning_rate": 1.692088888888889e-05,
+      "loss": 1.7222,
+      "step": 14910
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 45.18415069580078,
+      "learning_rate": 1.6917333333333335e-05,
+      "loss": 1.704,
+      "step": 14920
+    },
+    {
+      "epoch": 0.47776,
+      "grad_norm": 45.706817626953125,
+      "learning_rate": 1.691377777777778e-05,
+      "loss": 1.6791,
+      "step": 14930
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 46.975589752197266,
+      "learning_rate": 1.6910222222222224e-05,
+      "loss": 1.7106,
+      "step": 14940
+    },
+    {
+      "epoch": 0.4784,
+      "grad_norm": 42.881412506103516,
+      "learning_rate": 1.690666666666667e-05,
+      "loss": 1.7162,
+      "step": 14950
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 45.331024169921875,
+      "learning_rate": 1.6903111111111113e-05,
+      "loss": 1.7089,
+      "step": 14960
+    },
+    {
+      "epoch": 0.47904,
+      "grad_norm": 45.485511779785156,
+      "learning_rate": 1.6899555555555558e-05,
+      "loss": 1.6941,
+      "step": 14970
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 44.187564849853516,
+      "learning_rate": 1.6896000000000002e-05,
+      "loss": 1.7062,
+      "step": 14980
+    },
+    {
+      "epoch": 0.47968,
+      "grad_norm": 44.031944274902344,
+      "learning_rate": 1.6892444444444447e-05,
+      "loss": 1.6959,
+      "step": 14990
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 44.38731384277344,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 1.7016,
+      "step": 15000
+    },
+    {
+      "epoch": 0.48032,
+      "grad_norm": 43.112335205078125,
+      "learning_rate": 1.6885333333333336e-05,
+      "loss": 1.6804,
+      "step": 15010
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 47.42929458618164,
+      "learning_rate": 1.688177777777778e-05,
+      "loss": 1.7187,
+      "step": 15020
+    },
+    {
+      "epoch": 0.48096,
+      "grad_norm": 46.1917610168457,
+      "learning_rate": 1.687822222222222e-05,
+      "loss": 1.728,
+      "step": 15030
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 47.63662338256836,
+      "learning_rate": 1.687466666666667e-05,
+      "loss": 1.7058,
+      "step": 15040
+    },
+    {
+      "epoch": 0.4816,
+      "grad_norm": 45.0674934387207,
+      "learning_rate": 1.687111111111111e-05,
+      "loss": 1.7012,
+      "step": 15050
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 43.3830451965332,
+      "learning_rate": 1.686755555555556e-05,
+      "loss": 1.7319,
+      "step": 15060
+    },
+    {
+      "epoch": 0.48224,
+      "grad_norm": 48.20471954345703,
+      "learning_rate": 1.6864e-05,
+      "loss": 1.7366,
+      "step": 15070
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 47.284584045410156,
+      "learning_rate": 1.6860444444444444e-05,
+      "loss": 1.7092,
+      "step": 15080
+    },
+    {
+      "epoch": 0.48288,
+      "grad_norm": 45.17268753051758,
+      "learning_rate": 1.685688888888889e-05,
+      "loss": 1.7328,
+      "step": 15090
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 45.82081985473633,
+      "learning_rate": 1.6853333333333333e-05,
+      "loss": 1.7183,
+      "step": 15100
+    },
+    {
+      "epoch": 0.48352,
+      "grad_norm": 46.140403747558594,
+      "learning_rate": 1.6849777777777778e-05,
+      "loss": 1.7129,
+      "step": 15110
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 44.811729431152344,
+      "learning_rate": 1.6846222222222223e-05,
+      "loss": 1.6577,
+      "step": 15120
+    },
+    {
+      "epoch": 0.48416,
+      "grad_norm": 44.09989547729492,
+      "learning_rate": 1.6842666666666667e-05,
+      "loss": 1.7063,
+      "step": 15130
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 45.67482376098633,
+      "learning_rate": 1.683911111111111e-05,
+      "loss": 1.7513,
+      "step": 15140
+    },
+    {
+      "epoch": 0.4848,
+      "grad_norm": 44.314327239990234,
+      "learning_rate": 1.6835555555555556e-05,
+      "loss": 1.7474,
+      "step": 15150
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 45.371646881103516,
+      "learning_rate": 1.6832e-05,
+      "loss": 1.6844,
+      "step": 15160
+    },
+    {
+      "epoch": 0.48544,
+      "grad_norm": 44.82754135131836,
+      "learning_rate": 1.6828444444444445e-05,
+      "loss": 1.7245,
+      "step": 15170
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 43.516319274902344,
+      "learning_rate": 1.682488888888889e-05,
+      "loss": 1.713,
+      "step": 15180
+    },
+    {
+      "epoch": 0.48608,
+      "grad_norm": 44.03482437133789,
+      "learning_rate": 1.6821333333333334e-05,
+      "loss": 1.6786,
+      "step": 15190
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 43.02853012084961,
+      "learning_rate": 1.681777777777778e-05,
+      "loss": 1.7134,
+      "step": 15200
+    },
+    {
+      "epoch": 0.48672,
+      "grad_norm": 43.97295379638672,
+      "learning_rate": 1.6814222222222224e-05,
+      "loss": 1.7292,
+      "step": 15210
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 43.18336868286133,
+      "learning_rate": 1.6810666666666668e-05,
+      "loss": 1.7036,
+      "step": 15220
+    },
+    {
+      "epoch": 0.48736,
+      "grad_norm": 45.938255310058594,
+      "learning_rate": 1.6807111111111113e-05,
+      "loss": 1.6985,
+      "step": 15230
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 45.43522262573242,
+      "learning_rate": 1.6803555555555557e-05,
+      "loss": 1.6988,
+      "step": 15240
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 44.32875442504883,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.7163,
+      "step": 15250
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 45.96543502807617,
+      "learning_rate": 1.6796444444444446e-05,
+      "loss": 1.7364,
+      "step": 15260
+    },
+    {
+      "epoch": 0.48864,
+      "grad_norm": 44.921085357666016,
+      "learning_rate": 1.679288888888889e-05,
+      "loss": 1.6942,
+      "step": 15270
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 46.332210540771484,
+      "learning_rate": 1.6789333333333335e-05,
+      "loss": 1.7332,
+      "step": 15280
+    },
+    {
+      "epoch": 0.48928,
+      "grad_norm": 45.14290237426758,
+      "learning_rate": 1.678577777777778e-05,
+      "loss": 1.7133,
+      "step": 15290
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 43.08831787109375,
+      "learning_rate": 1.6782222222222225e-05,
+      "loss": 1.763,
+      "step": 15300
+    },
+    {
+      "epoch": 0.48992,
+      "grad_norm": 44.531578063964844,
+      "learning_rate": 1.677866666666667e-05,
+      "loss": 1.7257,
+      "step": 15310
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 42.90713119506836,
+      "learning_rate": 1.6775111111111114e-05,
+      "loss": 1.7057,
+      "step": 15320
+    },
+    {
+      "epoch": 0.49056,
+      "grad_norm": 44.45404052734375,
+      "learning_rate": 1.6771555555555555e-05,
+      "loss": 1.6935,
+      "step": 15330
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 43.76084899902344,
+      "learning_rate": 1.6768000000000003e-05,
+      "loss": 1.7232,
+      "step": 15340
+    },
+    {
+      "epoch": 0.4912,
+      "grad_norm": 44.55863952636719,
+      "learning_rate": 1.6764444444444444e-05,
+      "loss": 1.6916,
+      "step": 15350
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 43.15925979614258,
+      "learning_rate": 1.6760888888888892e-05,
+      "loss": 1.7028,
+      "step": 15360
+    },
+    {
+      "epoch": 0.49184,
+      "grad_norm": 45.47795104980469,
+      "learning_rate": 1.6757333333333333e-05,
+      "loss": 1.6957,
+      "step": 15370
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 45.608734130859375,
+      "learning_rate": 1.675377777777778e-05,
+      "loss": 1.6955,
+      "step": 15380
+    },
+    {
+      "epoch": 0.49248,
+      "grad_norm": 43.81676483154297,
+      "learning_rate": 1.6750222222222222e-05,
+      "loss": 1.7257,
+      "step": 15390
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 45.095298767089844,
+      "learning_rate": 1.674666666666667e-05,
+      "loss": 1.7381,
+      "step": 15400
+    },
+    {
+      "epoch": 0.49312,
+      "grad_norm": 43.94649887084961,
+      "learning_rate": 1.674311111111111e-05,
+      "loss": 1.6942,
+      "step": 15410
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 45.50575256347656,
+      "learning_rate": 1.673955555555556e-05,
+      "loss": 1.6912,
+      "step": 15420
+    },
+    {
+      "epoch": 0.49376,
+      "grad_norm": 48.223209381103516,
+      "learning_rate": 1.6736e-05,
+      "loss": 1.7267,
+      "step": 15430
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 44.392372131347656,
+      "learning_rate": 1.6732444444444448e-05,
+      "loss": 1.7077,
+      "step": 15440
+    },
+    {
+      "epoch": 0.4944,
+      "grad_norm": 44.331565856933594,
+      "learning_rate": 1.672888888888889e-05,
+      "loss": 1.7118,
+      "step": 15450
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 44.65738296508789,
+      "learning_rate": 1.6725333333333334e-05,
+      "loss": 1.6891,
+      "step": 15460
+    },
+    {
+      "epoch": 0.49504,
+      "grad_norm": 43.0691032409668,
+      "learning_rate": 1.672177777777778e-05,
+      "loss": 1.7015,
+      "step": 15470
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 44.851234436035156,
+      "learning_rate": 1.6718222222222223e-05,
+      "loss": 1.7186,
+      "step": 15480
+    },
+    {
+      "epoch": 0.49568,
+      "grad_norm": 44.26792907714844,
+      "learning_rate": 1.6714666666666668e-05,
+      "loss": 1.6874,
+      "step": 15490
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 45.67174530029297,
+      "learning_rate": 1.6711111111111112e-05,
+      "loss": 1.6646,
+      "step": 15500
+    },
+    {
+      "epoch": 0.49632,
+      "grad_norm": 46.26319885253906,
+      "learning_rate": 1.6707555555555557e-05,
+      "loss": 1.7374,
+      "step": 15510
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 45.9160041809082,
+      "learning_rate": 1.6704e-05,
+      "loss": 1.6884,
+      "step": 15520
+    },
+    {
+      "epoch": 0.49696,
+      "grad_norm": 43.73548126220703,
+      "learning_rate": 1.6700444444444446e-05,
+      "loss": 1.704,
+      "step": 15530
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 46.672542572021484,
+      "learning_rate": 1.669688888888889e-05,
+      "loss": 1.7175,
+      "step": 15540
+    },
+    {
+      "epoch": 0.4976,
+      "grad_norm": 43.42729949951172,
+      "learning_rate": 1.6693333333333335e-05,
+      "loss": 1.7285,
+      "step": 15550
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 43.43124008178711,
+      "learning_rate": 1.668977777777778e-05,
+      "loss": 1.7281,
+      "step": 15560
+    },
+    {
+      "epoch": 0.49824,
+      "grad_norm": 43.26439666748047,
+      "learning_rate": 1.6686222222222224e-05,
+      "loss": 1.7142,
+      "step": 15570
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 46.92955780029297,
+      "learning_rate": 1.668266666666667e-05,
+      "loss": 1.7141,
+      "step": 15580
+    },
+    {
+      "epoch": 0.49888,
+      "grad_norm": 45.263893127441406,
+      "learning_rate": 1.6679111111111113e-05,
+      "loss": 1.7077,
+      "step": 15590
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 45.84095001220703,
+      "learning_rate": 1.6675555555555554e-05,
+      "loss": 1.726,
+      "step": 15600
+    },
+    {
+      "epoch": 0.49952,
+      "grad_norm": 44.74760055541992,
+      "learning_rate": 1.6672000000000002e-05,
+      "loss": 1.6628,
+      "step": 15610
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 45.276981353759766,
+      "learning_rate": 1.6668444444444443e-05,
+      "loss": 1.7268,
+      "step": 15620
+    },
+    {
+      "epoch": 0.50016,
+      "grad_norm": 44.45388412475586,
+      "learning_rate": 1.666488888888889e-05,
+      "loss": 1.7051,
+      "step": 15630
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 43.3873291015625,
+      "learning_rate": 1.6661333333333333e-05,
+      "loss": 1.66,
+      "step": 15640
+    },
+    {
+      "epoch": 0.5008,
+      "grad_norm": 45.358951568603516,
+      "learning_rate": 1.665777777777778e-05,
+      "loss": 1.7032,
+      "step": 15650
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 45.18826675415039,
+      "learning_rate": 1.6654222222222222e-05,
+      "loss": 1.7122,
+      "step": 15660
+    },
+    {
+      "epoch": 0.50144,
+      "grad_norm": 47.63956832885742,
+      "learning_rate": 1.665066666666667e-05,
+      "loss": 1.7176,
+      "step": 15670
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 45.27025604248047,
+      "learning_rate": 1.664711111111111e-05,
+      "loss": 1.6904,
+      "step": 15680
+    },
+    {
+      "epoch": 0.50208,
+      "grad_norm": 44.5028190612793,
+      "learning_rate": 1.664355555555556e-05,
+      "loss": 1.6702,
+      "step": 15690
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 43.3703727722168,
+      "learning_rate": 1.664e-05,
+      "loss": 1.7137,
+      "step": 15700
+    },
+    {
+      "epoch": 0.50272,
+      "grad_norm": 43.58053207397461,
+      "learning_rate": 1.6636444444444448e-05,
+      "loss": 1.7239,
+      "step": 15710
+    },
+    {
+      "epoch": 0.50304,
+      "grad_norm": 44.74704360961914,
+      "learning_rate": 1.663288888888889e-05,
+      "loss": 1.6596,
+      "step": 15720
+    },
+    {
+      "epoch": 0.50336,
+      "grad_norm": 43.05439376831055,
+      "learning_rate": 1.6629333333333334e-05,
+      "loss": 1.6869,
+      "step": 15730
+    },
+    {
+      "epoch": 0.50368,
+      "grad_norm": 45.10047149658203,
+      "learning_rate": 1.6625777777777778e-05,
+      "loss": 1.7405,
+      "step": 15740
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 44.82457733154297,
+      "learning_rate": 1.6622222222222223e-05,
+      "loss": 1.6951,
+      "step": 15750
+    },
+    {
+      "epoch": 0.50432,
+      "grad_norm": 46.221214294433594,
+      "learning_rate": 1.6618666666666667e-05,
+      "loss": 1.721,
+      "step": 15760
+    },
+    {
+      "epoch": 0.50464,
+      "grad_norm": 41.99622344970703,
+      "learning_rate": 1.6615111111111112e-05,
+      "loss": 1.6501,
+      "step": 15770
+    },
+    {
+      "epoch": 0.50496,
+      "grad_norm": 44.5416145324707,
+      "learning_rate": 1.6611555555555556e-05,
+      "loss": 1.6827,
+      "step": 15780
+    },
+    {
+      "epoch": 0.50528,
+      "grad_norm": 43.418243408203125,
+      "learning_rate": 1.6608e-05,
+      "loss": 1.7221,
+      "step": 15790
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 44.96596908569336,
+      "learning_rate": 1.6604444444444445e-05,
+      "loss": 1.6984,
+      "step": 15800
+    },
+    {
+      "epoch": 0.50592,
+      "grad_norm": 47.30256652832031,
+      "learning_rate": 1.660088888888889e-05,
+      "loss": 1.7409,
+      "step": 15810
+    },
+    {
+      "epoch": 0.50624,
+      "grad_norm": 45.20161819458008,
+      "learning_rate": 1.6597333333333335e-05,
+      "loss": 1.7211,
+      "step": 15820
+    },
+    {
+      "epoch": 0.50656,
+      "grad_norm": 45.83244705200195,
+      "learning_rate": 1.659377777777778e-05,
+      "loss": 1.6697,
+      "step": 15830
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 46.3140754699707,
+      "learning_rate": 1.6590222222222224e-05,
+      "loss": 1.7022,
+      "step": 15840
+    },
+    {
+      "epoch": 0.5072,
+      "grad_norm": 45.87006378173828,
+      "learning_rate": 1.6586666666666668e-05,
+      "loss": 1.722,
+      "step": 15850
+    },
+    {
+      "epoch": 0.50752,
+      "grad_norm": 45.3338508605957,
+      "learning_rate": 1.6583111111111113e-05,
+      "loss": 1.7095,
+      "step": 15860
+    },
+    {
+      "epoch": 0.50784,
+      "grad_norm": 44.24995422363281,
+      "learning_rate": 1.6579555555555557e-05,
+      "loss": 1.7225,
+      "step": 15870
+    },
+    {
+      "epoch": 0.50816,
+      "grad_norm": 45.53012466430664,
+      "learning_rate": 1.6576000000000002e-05,
+      "loss": 1.6879,
+      "step": 15880
+    },
+    {
+      "epoch": 0.50848,
+      "grad_norm": 46.71226119995117,
+      "learning_rate": 1.6572444444444446e-05,
+      "loss": 1.6784,
+      "step": 15890
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 44.13043212890625,
+      "learning_rate": 1.656888888888889e-05,
+      "loss": 1.6958,
+      "step": 15900
+    },
+    {
+      "epoch": 0.50912,
+      "grad_norm": 44.21406555175781,
+      "learning_rate": 1.6565333333333336e-05,
+      "loss": 1.6638,
+      "step": 15910
+    },
+    {
+      "epoch": 0.50944,
+      "grad_norm": 43.99684143066406,
+      "learning_rate": 1.656177777777778e-05,
+      "loss": 1.7091,
+      "step": 15920
+    },
+    {
+      "epoch": 0.50976,
+      "grad_norm": 44.244503021240234,
+      "learning_rate": 1.6558222222222225e-05,
+      "loss": 1.7032,
+      "step": 15930
+    },
+    {
+      "epoch": 0.51008,
+      "grad_norm": 45.38497543334961,
+      "learning_rate": 1.655466666666667e-05,
+      "loss": 1.6894,
+      "step": 15940
+    },
+    {
+      "epoch": 0.5104,
+      "grad_norm": 44.65730667114258,
+      "learning_rate": 1.6551111111111114e-05,
+      "loss": 1.7331,
+      "step": 15950
+    },
+    {
+      "epoch": 0.51072,
+      "grad_norm": 44.94163513183594,
+      "learning_rate": 1.654755555555556e-05,
+      "loss": 1.7253,
+      "step": 15960
+    },
+    {
+      "epoch": 0.51104,
+      "grad_norm": 43.73176193237305,
+      "learning_rate": 1.6544000000000003e-05,
+      "loss": 1.7087,
+      "step": 15970
+    },
+    {
+      "epoch": 0.51136,
+      "grad_norm": 45.248966217041016,
+      "learning_rate": 1.6540444444444444e-05,
+      "loss": 1.7013,
+      "step": 15980
+    },
+    {
+      "epoch": 0.51168,
+      "grad_norm": 43.847721099853516,
+      "learning_rate": 1.6536888888888892e-05,
+      "loss": 1.7007,
+      "step": 15990
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 43.88396453857422,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 1.6882,
+      "step": 16000
+    },
+    {
+      "epoch": 0.51232,
+      "grad_norm": 43.48516845703125,
+      "learning_rate": 1.652977777777778e-05,
+      "loss": 1.7187,
+      "step": 16010
+    },
+    {
+      "epoch": 0.51264,
+      "grad_norm": 44.9348030090332,
+      "learning_rate": 1.6526222222222222e-05,
+      "loss": 1.6971,
+      "step": 16020
+    },
+    {
+      "epoch": 0.51296,
+      "grad_norm": 48.24411392211914,
+      "learning_rate": 1.652266666666667e-05,
+      "loss": 1.6926,
+      "step": 16030
+    },
+    {
+      "epoch": 0.51328,
+      "grad_norm": 45.70420455932617,
+      "learning_rate": 1.651911111111111e-05,
+      "loss": 1.7344,
+      "step": 16040
+    },
+    {
+      "epoch": 0.5136,
+      "grad_norm": 43.00725555419922,
+      "learning_rate": 1.651555555555556e-05,
+      "loss": 1.702,
+      "step": 16050
+    },
+    {
+      "epoch": 0.51392,
+      "grad_norm": 43.134979248046875,
+      "learning_rate": 1.6512e-05,
+      "loss": 1.7159,
+      "step": 16060
+    },
+    {
+      "epoch": 0.51424,
+      "grad_norm": 43.83757781982422,
+      "learning_rate": 1.6508444444444445e-05,
+      "loss": 1.6921,
+      "step": 16070
+    },
+    {
+      "epoch": 0.51456,
+      "grad_norm": 44.838863372802734,
+      "learning_rate": 1.650488888888889e-05,
+      "loss": 1.7255,
+      "step": 16080
+    },
+    {
+      "epoch": 0.51488,
+      "grad_norm": 46.207984924316406,
+      "learning_rate": 1.6501333333333334e-05,
+      "loss": 1.697,
+      "step": 16090
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 42.93474578857422,
+      "learning_rate": 1.649777777777778e-05,
+      "loss": 1.7225,
+      "step": 16100
+    },
+    {
+      "epoch": 0.51552,
+      "grad_norm": 44.5800895690918,
+      "learning_rate": 1.6494222222222223e-05,
+      "loss": 1.7124,
+      "step": 16110
+    },
+    {
+      "epoch": 0.51584,
+      "grad_norm": 45.92104721069336,
+      "learning_rate": 1.6490666666666668e-05,
+      "loss": 1.718,
+      "step": 16120
+    },
+    {
+      "epoch": 0.51616,
+      "grad_norm": 45.22180938720703,
+      "learning_rate": 1.6487111111111112e-05,
+      "loss": 1.7042,
+      "step": 16130
+    },
+    {
+      "epoch": 0.51648,
+      "grad_norm": 45.1376838684082,
+      "learning_rate": 1.6483555555555557e-05,
+      "loss": 1.6804,
+      "step": 16140
+    },
+    {
+      "epoch": 0.5168,
+      "grad_norm": 44.4555778503418,
+      "learning_rate": 1.648e-05,
+      "loss": 1.7167,
+      "step": 16150
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 46.41279602050781,
+      "learning_rate": 1.6476444444444446e-05,
+      "loss": 1.7113,
+      "step": 16160
+    },
+    {
+      "epoch": 0.51744,
+      "grad_norm": 45.017879486083984,
+      "learning_rate": 1.647288888888889e-05,
+      "loss": 1.7323,
+      "step": 16170
+    },
+    {
+      "epoch": 0.51776,
+      "grad_norm": 44.85575485229492,
+      "learning_rate": 1.6469333333333335e-05,
+      "loss": 1.7163,
+      "step": 16180
+    },
+    {
+      "epoch": 0.51808,
+      "grad_norm": 43.125179290771484,
+      "learning_rate": 1.646577777777778e-05,
+      "loss": 1.7103,
+      "step": 16190
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 47.32752227783203,
+      "learning_rate": 1.6462222222222224e-05,
+      "loss": 1.7208,
+      "step": 16200
+    },
+    {
+      "epoch": 0.51872,
+      "grad_norm": 44.214813232421875,
+      "learning_rate": 1.645866666666667e-05,
+      "loss": 1.6869,
+      "step": 16210
+    },
+    {
+      "epoch": 0.51904,
+      "grad_norm": 43.823272705078125,
+      "learning_rate": 1.6455111111111113e-05,
+      "loss": 1.7107,
+      "step": 16220
+    },
+    {
+      "epoch": 0.51936,
+      "grad_norm": 44.63423538208008,
+      "learning_rate": 1.6451555555555554e-05,
+      "loss": 1.665,
+      "step": 16230
+    },
+    {
+      "epoch": 0.51968,
+      "grad_norm": 43.06886672973633,
+      "learning_rate": 1.6448000000000002e-05,
+      "loss": 1.7051,
+      "step": 16240
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 43.636932373046875,
+      "learning_rate": 1.6444444444444444e-05,
+      "loss": 1.7217,
+      "step": 16250
+    },
+    {
+      "epoch": 0.52032,
+      "grad_norm": 44.1069221496582,
+      "learning_rate": 1.644088888888889e-05,
+      "loss": 1.7063,
+      "step": 16260
+    },
+    {
+      "epoch": 0.52064,
+      "grad_norm": 46.94747543334961,
+      "learning_rate": 1.6437333333333333e-05,
+      "loss": 1.699,
+      "step": 16270
+    },
+    {
+      "epoch": 0.52096,
+      "grad_norm": 45.03914260864258,
+      "learning_rate": 1.643377777777778e-05,
+      "loss": 1.7257,
+      "step": 16280
+    },
+    {
+      "epoch": 0.52128,
+      "grad_norm": 45.86468505859375,
+      "learning_rate": 1.6430222222222222e-05,
+      "loss": 1.722,
+      "step": 16290
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 43.51268005371094,
+      "learning_rate": 1.642666666666667e-05,
+      "loss": 1.676,
+      "step": 16300
+    },
+    {
+      "epoch": 0.52192,
+      "grad_norm": 42.855106353759766,
+      "learning_rate": 1.642311111111111e-05,
+      "loss": 1.6782,
+      "step": 16310
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 48.07331466674805,
+      "learning_rate": 1.641955555555556e-05,
+      "loss": 1.7173,
+      "step": 16320
+    },
+    {
+      "epoch": 0.52256,
+      "grad_norm": 43.21236038208008,
+      "learning_rate": 1.6416e-05,
+      "loss": 1.7336,
+      "step": 16330
+    },
+    {
+      "epoch": 0.52288,
+      "grad_norm": 44.75568771362305,
+      "learning_rate": 1.6412444444444448e-05,
+      "loss": 1.6918,
+      "step": 16340
+    },
+    {
+      "epoch": 0.5232,
+      "grad_norm": 44.176422119140625,
+      "learning_rate": 1.640888888888889e-05,
+      "loss": 1.7002,
+      "step": 16350
+    },
+    {
+      "epoch": 0.52352,
+      "grad_norm": 44.09846115112305,
+      "learning_rate": 1.6405333333333334e-05,
+      "loss": 1.6942,
+      "step": 16360
+    },
+    {
+      "epoch": 0.52384,
+      "grad_norm": 42.87156295776367,
+      "learning_rate": 1.6401777777777778e-05,
+      "loss": 1.7208,
+      "step": 16370
+    },
+    {
+      "epoch": 0.52416,
+      "grad_norm": 46.28589630126953,
+      "learning_rate": 1.6398222222222223e-05,
+      "loss": 1.6768,
+      "step": 16380
+    },
+    {
+      "epoch": 0.52448,
+      "grad_norm": 45.45729064941406,
+      "learning_rate": 1.6394666666666667e-05,
+      "loss": 1.6837,
+      "step": 16390
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 45.6191520690918,
+      "learning_rate": 1.6391111111111112e-05,
+      "loss": 1.7055,
+      "step": 16400
+    },
+    {
+      "epoch": 0.52512,
+      "grad_norm": 45.490787506103516,
+      "learning_rate": 1.6387555555555556e-05,
+      "loss": 1.7033,
+      "step": 16410
+    },
+    {
+      "epoch": 0.52544,
+      "grad_norm": 44.59261703491211,
+      "learning_rate": 1.6384e-05,
+      "loss": 1.7125,
+      "step": 16420
+    },
+    {
+      "epoch": 0.52576,
+      "grad_norm": 45.315059661865234,
+      "learning_rate": 1.6380444444444446e-05,
+      "loss": 1.7056,
+      "step": 16430
+    },
+    {
+      "epoch": 0.52608,
+      "grad_norm": 45.0549430847168,
+      "learning_rate": 1.637688888888889e-05,
+      "loss": 1.7085,
+      "step": 16440
+    },
+    {
+      "epoch": 0.5264,
+      "grad_norm": 44.14567565917969,
+      "learning_rate": 1.6373333333333335e-05,
+      "loss": 1.6874,
+      "step": 16450
+    },
+    {
+      "epoch": 0.52672,
+      "grad_norm": 45.77822494506836,
+      "learning_rate": 1.636977777777778e-05,
+      "loss": 1.7104,
+      "step": 16460
+    },
+    {
+      "epoch": 0.52704,
+      "grad_norm": 46.14954376220703,
+      "learning_rate": 1.6366222222222224e-05,
+      "loss": 1.6697,
+      "step": 16470
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 45.201969146728516,
+      "learning_rate": 1.636266666666667e-05,
+      "loss": 1.7316,
+      "step": 16480
+    },
+    {
+      "epoch": 0.52768,
+      "grad_norm": 47.41619110107422,
+      "learning_rate": 1.6359111111111113e-05,
+      "loss": 1.6999,
+      "step": 16490
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 43.96504211425781,
+      "learning_rate": 1.6355555555555557e-05,
+      "loss": 1.7277,
+      "step": 16500
+    },
+    {
+      "epoch": 0.52832,
+      "grad_norm": 45.96023178100586,
+      "learning_rate": 1.6352000000000002e-05,
+      "loss": 1.703,
+      "step": 16510
+    },
+    {
+      "epoch": 0.52864,
+      "grad_norm": 46.28660583496094,
+      "learning_rate": 1.6348444444444447e-05,
+      "loss": 1.6972,
+      "step": 16520
+    },
+    {
+      "epoch": 0.52896,
+      "grad_norm": 43.34731674194336,
+      "learning_rate": 1.634488888888889e-05,
+      "loss": 1.6852,
+      "step": 16530
+    },
+    {
+      "epoch": 0.52928,
+      "grad_norm": 45.01405334472656,
+      "learning_rate": 1.6341333333333336e-05,
+      "loss": 1.6812,
+      "step": 16540
+    },
+    {
+      "epoch": 0.5296,
+      "grad_norm": 44.526119232177734,
+      "learning_rate": 1.633777777777778e-05,
+      "loss": 1.7143,
+      "step": 16550
+    },
+    {
+      "epoch": 0.52992,
+      "grad_norm": 45.97877883911133,
+      "learning_rate": 1.6334222222222225e-05,
+      "loss": 1.7182,
+      "step": 16560
+    },
+    {
+      "epoch": 0.53024,
+      "grad_norm": 47.06319046020508,
+      "learning_rate": 1.633066666666667e-05,
+      "loss": 1.704,
+      "step": 16570
+    },
+    {
+      "epoch": 0.53056,
+      "grad_norm": 44.56294250488281,
+      "learning_rate": 1.632711111111111e-05,
+      "loss": 1.7028,
+      "step": 16580
+    },
+    {
+      "epoch": 0.53088,
+      "grad_norm": 43.73545837402344,
+      "learning_rate": 1.632355555555556e-05,
+      "loss": 1.6899,
+      "step": 16590
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 45.30746841430664,
+      "learning_rate": 1.632e-05,
+      "loss": 1.687,
+      "step": 16600
+    },
+    {
+      "epoch": 0.53152,
+      "grad_norm": 44.87981414794922,
+      "learning_rate": 1.6316444444444448e-05,
+      "loss": 1.6603,
+      "step": 16610
+    },
+    {
+      "epoch": 0.53184,
+      "grad_norm": 44.816471099853516,
+      "learning_rate": 1.631288888888889e-05,
+      "loss": 1.7085,
+      "step": 16620
+    },
+    {
+      "epoch": 0.53216,
+      "grad_norm": 44.822017669677734,
+      "learning_rate": 1.6309333333333333e-05,
+      "loss": 1.717,
+      "step": 16630
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 46.93596649169922,
+      "learning_rate": 1.6305777777777778e-05,
+      "loss": 1.6723,
+      "step": 16640
+    },
+    {
+      "epoch": 0.5328,
+      "grad_norm": 45.41704177856445,
+      "learning_rate": 1.6302222222222222e-05,
+      "loss": 1.7343,
+      "step": 16650
+    },
+    {
+      "epoch": 0.53312,
+      "grad_norm": 45.636146545410156,
+      "learning_rate": 1.6298666666666667e-05,
+      "loss": 1.698,
+      "step": 16660
+    },
+    {
+      "epoch": 0.53344,
+      "grad_norm": 43.23383712768555,
+      "learning_rate": 1.629511111111111e-05,
+      "loss": 1.6966,
+      "step": 16670
+    },
+    {
+      "epoch": 0.53376,
+      "grad_norm": 42.99844741821289,
+      "learning_rate": 1.6291555555555556e-05,
+      "loss": 1.7189,
+      "step": 16680
+    },
+    {
+      "epoch": 0.53408,
+      "grad_norm": 42.8013801574707,
+      "learning_rate": 1.6288e-05,
+      "loss": 1.7105,
+      "step": 16690
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 44.297874450683594,
+      "learning_rate": 1.6284444444444445e-05,
+      "loss": 1.6963,
+      "step": 16700
+    },
+    {
+      "epoch": 0.53472,
+      "grad_norm": 45.40163803100586,
+      "learning_rate": 1.628088888888889e-05,
+      "loss": 1.7061,
+      "step": 16710
+    },
+    {
+      "epoch": 0.53504,
+      "grad_norm": 44.55815887451172,
+      "learning_rate": 1.6277333333333334e-05,
+      "loss": 1.686,
+      "step": 16720
+    },
+    {
+      "epoch": 0.53536,
+      "grad_norm": 47.977874755859375,
+      "learning_rate": 1.627377777777778e-05,
+      "loss": 1.7157,
+      "step": 16730
+    },
+    {
+      "epoch": 0.53568,
+      "grad_norm": 44.50165557861328,
+      "learning_rate": 1.6270222222222223e-05,
+      "loss": 1.6814,
+      "step": 16740
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 42.11151885986328,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.7029,
+      "step": 16750
+    },
+    {
+      "epoch": 0.53632,
+      "grad_norm": 48.29054641723633,
+      "learning_rate": 1.6263111111111112e-05,
+      "loss": 1.6978,
+      "step": 16760
+    },
+    {
+      "epoch": 0.53664,
+      "grad_norm": 44.33515930175781,
+      "learning_rate": 1.6259555555555557e-05,
+      "loss": 1.6746,
+      "step": 16770
+    },
+    {
+      "epoch": 0.53696,
+      "grad_norm": 49.40854263305664,
+      "learning_rate": 1.6256e-05,
+      "loss": 1.7046,
+      "step": 16780
+    },
+    {
+      "epoch": 0.53728,
+      "grad_norm": 44.964656829833984,
+      "learning_rate": 1.6252444444444446e-05,
+      "loss": 1.7062,
+      "step": 16790
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 45.332523345947266,
+      "learning_rate": 1.624888888888889e-05,
+      "loss": 1.7132,
+      "step": 16800
+    },
+    {
+      "epoch": 0.53792,
+      "grad_norm": 44.173912048339844,
+      "learning_rate": 1.6245333333333335e-05,
+      "loss": 1.7039,
+      "step": 16810
+    },
+    {
+      "epoch": 0.53824,
+      "grad_norm": 45.531803131103516,
+      "learning_rate": 1.624177777777778e-05,
+      "loss": 1.6931,
+      "step": 16820
+    },
+    {
+      "epoch": 0.53856,
+      "grad_norm": 45.48138427734375,
+      "learning_rate": 1.6238222222222224e-05,
+      "loss": 1.7275,
+      "step": 16830
+    },
+    {
+      "epoch": 0.53888,
+      "grad_norm": 47.76054763793945,
+      "learning_rate": 1.623466666666667e-05,
+      "loss": 1.6815,
+      "step": 16840
+    },
+    {
+      "epoch": 0.5392,
+      "grad_norm": 44.69922637939453,
+      "learning_rate": 1.6231111111111113e-05,
+      "loss": 1.6909,
+      "step": 16850
+    },
+    {
+      "epoch": 0.53952,
+      "grad_norm": 44.52599334716797,
+      "learning_rate": 1.6227555555555558e-05,
+      "loss": 1.6969,
+      "step": 16860
+    },
+    {
+      "epoch": 0.53984,
+      "grad_norm": 45.41663360595703,
+      "learning_rate": 1.6224000000000003e-05,
+      "loss": 1.6778,
+      "step": 16870
+    },
+    {
+      "epoch": 0.54016,
+      "grad_norm": 44.901668548583984,
+      "learning_rate": 1.6220444444444444e-05,
+      "loss": 1.6551,
+      "step": 16880
+    },
+    {
+      "epoch": 0.54048,
+      "grad_norm": 43.41035461425781,
+      "learning_rate": 1.621688888888889e-05,
+      "loss": 1.7107,
+      "step": 16890
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 48.324066162109375,
+      "learning_rate": 1.6213333333333333e-05,
+      "loss": 1.6959,
+      "step": 16900
+    },
+    {
+      "epoch": 0.54112,
+      "grad_norm": 43.75379943847656,
+      "learning_rate": 1.620977777777778e-05,
+      "loss": 1.7029,
+      "step": 16910
+    },
+    {
+      "epoch": 0.54144,
+      "grad_norm": 43.38237380981445,
+      "learning_rate": 1.6206222222222222e-05,
+      "loss": 1.6957,
+      "step": 16920
+    },
+    {
+      "epoch": 0.54176,
+      "grad_norm": 44.03130340576172,
+      "learning_rate": 1.620266666666667e-05,
+      "loss": 1.6884,
+      "step": 16930
+    },
+    {
+      "epoch": 0.54208,
+      "grad_norm": 45.490447998046875,
+      "learning_rate": 1.619911111111111e-05,
+      "loss": 1.6706,
+      "step": 16940
+    },
+    {
+      "epoch": 0.5424,
+      "grad_norm": 45.389976501464844,
+      "learning_rate": 1.619555555555556e-05,
+      "loss": 1.7352,
+      "step": 16950
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 44.04669189453125,
+      "learning_rate": 1.6192e-05,
+      "loss": 1.6794,
+      "step": 16960
+    },
+    {
+      "epoch": 0.54304,
+      "grad_norm": 44.17753601074219,
+      "learning_rate": 1.6188444444444448e-05,
+      "loss": 1.6918,
+      "step": 16970
+    },
+    {
+      "epoch": 0.54336,
+      "grad_norm": 43.18086242675781,
+      "learning_rate": 1.618488888888889e-05,
+      "loss": 1.6545,
+      "step": 16980
+    },
+    {
+      "epoch": 0.54368,
+      "grad_norm": 43.751461029052734,
+      "learning_rate": 1.6181333333333337e-05,
+      "loss": 1.689,
+      "step": 16990
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 44.13274002075195,
+      "learning_rate": 1.617777777777778e-05,
+      "loss": 1.7332,
+      "step": 17000
+    },
+    {
+      "epoch": 0.54432,
+      "grad_norm": 45.5268440246582,
+      "learning_rate": 1.6174222222222223e-05,
+      "loss": 1.701,
+      "step": 17010
+    },
+    {
+      "epoch": 0.54464,
+      "grad_norm": 45.75001525878906,
+      "learning_rate": 1.6170666666666667e-05,
+      "loss": 1.711,
+      "step": 17020
+    },
+    {
+      "epoch": 0.54496,
+      "grad_norm": 43.565834045410156,
+      "learning_rate": 1.6167111111111112e-05,
+      "loss": 1.6872,
+      "step": 17030
+    },
+    {
+      "epoch": 0.54528,
+      "grad_norm": 45.25817108154297,
+      "learning_rate": 1.6163555555555557e-05,
+      "loss": 1.6778,
+      "step": 17040
+    },
+    {
+      "epoch": 0.5456,
+      "grad_norm": 43.177886962890625,
+      "learning_rate": 1.616e-05,
+      "loss": 1.7062,
+      "step": 17050
+    },
+    {
+      "epoch": 0.54592,
+      "grad_norm": 45.22313690185547,
+      "learning_rate": 1.6156444444444446e-05,
+      "loss": 1.6735,
+      "step": 17060
+    },
+    {
+      "epoch": 0.54624,
+      "grad_norm": 45.08768844604492,
+      "learning_rate": 1.615288888888889e-05,
+      "loss": 1.7249,
+      "step": 17070
+    },
+    {
+      "epoch": 0.54656,
+      "grad_norm": 44.014251708984375,
+      "learning_rate": 1.6149333333333335e-05,
+      "loss": 1.6784,
+      "step": 17080
+    },
+    {
+      "epoch": 0.54688,
+      "grad_norm": 45.65128707885742,
+      "learning_rate": 1.614577777777778e-05,
+      "loss": 1.6982,
+      "step": 17090
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 44.55026626586914,
+      "learning_rate": 1.6142222222222224e-05,
+      "loss": 1.7088,
+      "step": 17100
+    },
+    {
+      "epoch": 0.54752,
+      "grad_norm": 63.36881637573242,
+      "learning_rate": 1.613866666666667e-05,
+      "loss": 1.6831,
+      "step": 17110
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 45.86473083496094,
+      "learning_rate": 1.6135111111111113e-05,
+      "loss": 1.7168,
+      "step": 17120
+    },
+    {
+      "epoch": 0.54816,
+      "grad_norm": 44.3762092590332,
+      "learning_rate": 1.6131555555555554e-05,
+      "loss": 1.6983,
+      "step": 17130
+    },
+    {
+      "epoch": 0.54848,
+      "grad_norm": 45.90697479248047,
+      "learning_rate": 1.6128000000000002e-05,
+      "loss": 1.6779,
+      "step": 17140
+    },
+    {
+      "epoch": 0.5488,
+      "grad_norm": 47.833343505859375,
+      "learning_rate": 1.6124444444444443e-05,
+      "loss": 1.6911,
+      "step": 17150
+    },
+    {
+      "epoch": 0.54912,
+      "grad_norm": 43.13536834716797,
+      "learning_rate": 1.612088888888889e-05,
+      "loss": 1.7273,
+      "step": 17160
+    },
+    {
+      "epoch": 0.54944,
+      "grad_norm": 45.21577453613281,
+      "learning_rate": 1.6117333333333332e-05,
+      "loss": 1.695,
+      "step": 17170
+    },
+    {
+      "epoch": 0.54976,
+      "grad_norm": 48.69087219238281,
+      "learning_rate": 1.611377777777778e-05,
+      "loss": 1.7112,
+      "step": 17180
+    },
+    {
+      "epoch": 0.55008,
+      "grad_norm": 45.58196258544922,
+      "learning_rate": 1.611022222222222e-05,
+      "loss": 1.6941,
+      "step": 17190
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 45.56036376953125,
+      "learning_rate": 1.610666666666667e-05,
+      "loss": 1.6949,
+      "step": 17200
+    },
+    {
+      "epoch": 0.55072,
+      "grad_norm": 49.056949615478516,
+      "learning_rate": 1.610311111111111e-05,
+      "loss": 1.7246,
+      "step": 17210
+    },
+    {
+      "epoch": 0.55104,
+      "grad_norm": 45.82423782348633,
+      "learning_rate": 1.609955555555556e-05,
+      "loss": 1.6986,
+      "step": 17220
+    },
+    {
+      "epoch": 0.55136,
+      "grad_norm": 44.68406677246094,
+      "learning_rate": 1.6096e-05,
+      "loss": 1.7044,
+      "step": 17230
+    },
+    {
+      "epoch": 0.55168,
+      "grad_norm": 45.451416015625,
+      "learning_rate": 1.6092444444444448e-05,
+      "loss": 1.686,
+      "step": 17240
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 47.66313552856445,
+      "learning_rate": 1.608888888888889e-05,
+      "loss": 1.7049,
+      "step": 17250
+    },
+    {
+      "epoch": 0.55232,
+      "grad_norm": 42.10512924194336,
+      "learning_rate": 1.6085333333333333e-05,
+      "loss": 1.6732,
+      "step": 17260
+    },
+    {
+      "epoch": 0.55264,
+      "grad_norm": 44.73134231567383,
+      "learning_rate": 1.6081777777777778e-05,
+      "loss": 1.6905,
+      "step": 17270
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 43.184906005859375,
+      "learning_rate": 1.6078222222222222e-05,
+      "loss": 1.6986,
+      "step": 17280
+    },
+    {
+      "epoch": 0.55328,
+      "grad_norm": 44.09386444091797,
+      "learning_rate": 1.6074666666666667e-05,
+      "loss": 1.677,
+      "step": 17290
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 42.74454116821289,
+      "learning_rate": 1.607111111111111e-05,
+      "loss": 1.7163,
+      "step": 17300
+    },
+    {
+      "epoch": 0.55392,
+      "grad_norm": 43.98388671875,
+      "learning_rate": 1.6067555555555556e-05,
+      "loss": 1.713,
+      "step": 17310
+    },
+    {
+      "epoch": 0.55424,
+      "grad_norm": 45.99345397949219,
+      "learning_rate": 1.6064e-05,
+      "loss": 1.6824,
+      "step": 17320
+    },
+    {
+      "epoch": 0.55456,
+      "grad_norm": 43.22311019897461,
+      "learning_rate": 1.6060444444444445e-05,
+      "loss": 1.7054,
+      "step": 17330
+    },
+    {
+      "epoch": 0.55488,
+      "grad_norm": 45.42384338378906,
+      "learning_rate": 1.605688888888889e-05,
+      "loss": 1.689,
+      "step": 17340
+    },
+    {
+      "epoch": 0.5552,
+      "grad_norm": 43.916080474853516,
+      "learning_rate": 1.6053333333333334e-05,
+      "loss": 1.6624,
+      "step": 17350
+    },
+    {
+      "epoch": 0.55552,
+      "grad_norm": 46.04616165161133,
+      "learning_rate": 1.604977777777778e-05,
+      "loss": 1.6876,
+      "step": 17360
+    },
+    {
+      "epoch": 0.55584,
+      "grad_norm": 44.111244201660156,
+      "learning_rate": 1.6046222222222223e-05,
+      "loss": 1.713,
+      "step": 17370
+    },
+    {
+      "epoch": 0.55616,
+      "grad_norm": 44.648834228515625,
+      "learning_rate": 1.6042666666666668e-05,
+      "loss": 1.7224,
+      "step": 17380
+    },
+    {
+      "epoch": 0.55648,
+      "grad_norm": 44.2633171081543,
+      "learning_rate": 1.6039111111111113e-05,
+      "loss": 1.6621,
+      "step": 17390
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 45.79343795776367,
+      "learning_rate": 1.6035555555555557e-05,
+      "loss": 1.6911,
+      "step": 17400
+    },
+    {
+      "epoch": 0.55712,
+      "grad_norm": 44.808135986328125,
+      "learning_rate": 1.6032e-05,
+      "loss": 1.6654,
+      "step": 17410
+    },
+    {
+      "epoch": 0.55744,
+      "grad_norm": 45.78480529785156,
+      "learning_rate": 1.6028444444444446e-05,
+      "loss": 1.7339,
+      "step": 17420
+    },
+    {
+      "epoch": 0.55776,
+      "grad_norm": 44.06301498413086,
+      "learning_rate": 1.602488888888889e-05,
+      "loss": 1.6808,
+      "step": 17430
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 43.80913162231445,
+      "learning_rate": 1.6021333333333335e-05,
+      "loss": 1.709,
+      "step": 17440
+    },
+    {
+      "epoch": 0.5584,
+      "grad_norm": 43.52967071533203,
+      "learning_rate": 1.601777777777778e-05,
+      "loss": 1.7063,
+      "step": 17450
+    },
+    {
+      "epoch": 0.55872,
+      "grad_norm": 45.998050689697266,
+      "learning_rate": 1.6014222222222224e-05,
+      "loss": 1.6858,
+      "step": 17460
+    },
+    {
+      "epoch": 0.55904,
+      "grad_norm": 44.415245056152344,
+      "learning_rate": 1.601066666666667e-05,
+      "loss": 1.7001,
+      "step": 17470
+    },
+    {
+      "epoch": 0.55936,
+      "grad_norm": 43.485721588134766,
+      "learning_rate": 1.6007111111111114e-05,
+      "loss": 1.695,
+      "step": 17480
+    },
+    {
+      "epoch": 0.55968,
+      "grad_norm": 44.18275833129883,
+      "learning_rate": 1.6003555555555558e-05,
+      "loss": 1.6661,
+      "step": 17490
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 43.11178207397461,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.7007,
+      "step": 17500
+    },
+    {
+      "epoch": 0.56032,
+      "grad_norm": 44.23533630371094,
+      "learning_rate": 1.5996444444444447e-05,
+      "loss": 1.6794,
+      "step": 17510
+    },
+    {
+      "epoch": 0.56064,
+      "grad_norm": 47.07341003417969,
+      "learning_rate": 1.5992888888888892e-05,
+      "loss": 1.6812,
+      "step": 17520
+    },
+    {
+      "epoch": 0.56096,
+      "grad_norm": 44.18885040283203,
+      "learning_rate": 1.5989333333333333e-05,
+      "loss": 1.6781,
+      "step": 17530
+    },
+    {
+      "epoch": 0.56128,
+      "grad_norm": 43.99700927734375,
+      "learning_rate": 1.598577777777778e-05,
+      "loss": 1.6692,
+      "step": 17540
+    },
+    {
+      "epoch": 0.5616,
+      "grad_norm": 43.02412796020508,
+      "learning_rate": 1.5982222222222222e-05,
+      "loss": 1.6889,
+      "step": 17550
+    },
+    {
+      "epoch": 0.56192,
+      "grad_norm": 43.83738327026367,
+      "learning_rate": 1.597866666666667e-05,
+      "loss": 1.7167,
+      "step": 17560
+    },
+    {
+      "epoch": 0.56224,
+      "grad_norm": 45.4013671875,
+      "learning_rate": 1.597511111111111e-05,
+      "loss": 1.6789,
+      "step": 17570
+    },
+    {
+      "epoch": 0.56256,
+      "grad_norm": 43.10468673706055,
+      "learning_rate": 1.597155555555556e-05,
+      "loss": 1.7157,
+      "step": 17580
+    },
+    {
+      "epoch": 0.56288,
+      "grad_norm": 47.6757698059082,
+      "learning_rate": 1.5968e-05,
+      "loss": 1.7217,
+      "step": 17590
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 43.0317268371582,
+      "learning_rate": 1.5964444444444448e-05,
+      "loss": 1.7004,
+      "step": 17600
+    },
+    {
+      "epoch": 0.56352,
+      "grad_norm": 46.82234191894531,
+      "learning_rate": 1.596088888888889e-05,
+      "loss": 1.6829,
+      "step": 17610
+    },
+    {
+      "epoch": 0.56384,
+      "grad_norm": 44.28413009643555,
+      "learning_rate": 1.5957333333333334e-05,
+      "loss": 1.7082,
+      "step": 17620
+    },
+    {
+      "epoch": 0.56416,
+      "grad_norm": 45.0130615234375,
+      "learning_rate": 1.595377777777778e-05,
+      "loss": 1.7367,
+      "step": 17630
+    },
+    {
+      "epoch": 0.56448,
+      "grad_norm": 46.60377883911133,
+      "learning_rate": 1.5950222222222223e-05,
+      "loss": 1.6681,
+      "step": 17640
+    },
+    {
+      "epoch": 0.5648,
+      "grad_norm": 45.25273513793945,
+      "learning_rate": 1.5946666666666668e-05,
+      "loss": 1.6995,
+      "step": 17650
+    },
+    {
+      "epoch": 0.56512,
+      "grad_norm": 43.679325103759766,
+      "learning_rate": 1.5943111111111112e-05,
+      "loss": 1.6902,
+      "step": 17660
+    },
+    {
+      "epoch": 0.56544,
+      "grad_norm": 44.908199310302734,
+      "learning_rate": 1.5939555555555557e-05,
+      "loss": 1.7095,
+      "step": 17670
+    },
+    {
+      "epoch": 0.56576,
+      "grad_norm": 45.180484771728516,
+      "learning_rate": 1.5936e-05,
+      "loss": 1.7037,
+      "step": 17680
+    },
+    {
+      "epoch": 0.56608,
+      "grad_norm": 46.96930694580078,
+      "learning_rate": 1.5932444444444446e-05,
+      "loss": 1.7447,
+      "step": 17690
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 45.75835037231445,
+      "learning_rate": 1.592888888888889e-05,
+      "loss": 1.7116,
+      "step": 17700
+    },
+    {
+      "epoch": 0.56672,
+      "grad_norm": 43.56595230102539,
+      "learning_rate": 1.5925333333333335e-05,
+      "loss": 1.7121,
+      "step": 17710
+    },
+    {
+      "epoch": 0.56704,
+      "grad_norm": 44.589962005615234,
+      "learning_rate": 1.592177777777778e-05,
+      "loss": 1.7161,
+      "step": 17720
+    },
+    {
+      "epoch": 0.56736,
+      "grad_norm": 42.827396392822266,
+      "learning_rate": 1.5918222222222224e-05,
+      "loss": 1.6813,
+      "step": 17730
+    },
+    {
+      "epoch": 0.56768,
+      "grad_norm": 42.656185150146484,
+      "learning_rate": 1.591466666666667e-05,
+      "loss": 1.6775,
+      "step": 17740
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 45.042747497558594,
+      "learning_rate": 1.5911111111111113e-05,
+      "loss": 1.6921,
+      "step": 17750
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 44.19981384277344,
+      "learning_rate": 1.5907555555555558e-05,
+      "loss": 1.6934,
+      "step": 17760
+    },
+    {
+      "epoch": 0.56864,
+      "grad_norm": 44.734893798828125,
+      "learning_rate": 1.5904000000000002e-05,
+      "loss": 1.686,
+      "step": 17770
+    },
+    {
+      "epoch": 0.56896,
+      "grad_norm": 43.76615905761719,
+      "learning_rate": 1.5900444444444443e-05,
+      "loss": 1.7312,
+      "step": 17780
+    },
+    {
+      "epoch": 0.56928,
+      "grad_norm": 44.63469696044922,
+      "learning_rate": 1.589688888888889e-05,
+      "loss": 1.7115,
+      "step": 17790
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 43.65608215332031,
+      "learning_rate": 1.5893333333333333e-05,
+      "loss": 1.6742,
+      "step": 17800
+    },
+    {
+      "epoch": 0.56992,
+      "grad_norm": 45.9371337890625,
+      "learning_rate": 1.588977777777778e-05,
+      "loss": 1.7102,
+      "step": 17810
+    },
+    {
+      "epoch": 0.57024,
+      "grad_norm": 44.41642379760742,
+      "learning_rate": 1.588622222222222e-05,
+      "loss": 1.7064,
+      "step": 17820
+    },
+    {
+      "epoch": 0.57056,
+      "grad_norm": 45.774192810058594,
+      "learning_rate": 1.588266666666667e-05,
+      "loss": 1.709,
+      "step": 17830
+    },
+    {
+      "epoch": 0.57088,
+      "grad_norm": 46.056217193603516,
+      "learning_rate": 1.587911111111111e-05,
+      "loss": 1.6782,
+      "step": 17840
+    },
+    {
+      "epoch": 0.5712,
+      "grad_norm": 50.94669723510742,
+      "learning_rate": 1.587555555555556e-05,
+      "loss": 1.6707,
+      "step": 17850
+    },
+    {
+      "epoch": 0.57152,
+      "grad_norm": 44.1059455871582,
+      "learning_rate": 1.5872e-05,
+      "loss": 1.67,
+      "step": 17860
+    },
+    {
+      "epoch": 0.57184,
+      "grad_norm": 44.278621673583984,
+      "learning_rate": 1.5868444444444448e-05,
+      "loss": 1.7092,
+      "step": 17870
+    },
+    {
+      "epoch": 0.57216,
+      "grad_norm": 46.264774322509766,
+      "learning_rate": 1.586488888888889e-05,
+      "loss": 1.6798,
+      "step": 17880
+    },
+    {
+      "epoch": 0.57248,
+      "grad_norm": 42.81804656982422,
+      "learning_rate": 1.5861333333333337e-05,
+      "loss": 1.6711,
+      "step": 17890
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 42.7204475402832,
+      "learning_rate": 1.5857777777777778e-05,
+      "loss": 1.7242,
+      "step": 17900
+    },
+    {
+      "epoch": 0.57312,
+      "grad_norm": 47.116390228271484,
+      "learning_rate": 1.5854222222222223e-05,
+      "loss": 1.6984,
+      "step": 17910
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 43.8947639465332,
+      "learning_rate": 1.5850666666666667e-05,
+      "loss": 1.6769,
+      "step": 17920
+    },
+    {
+      "epoch": 0.57376,
+      "grad_norm": 44.71806716918945,
+      "learning_rate": 1.5847111111111112e-05,
+      "loss": 1.7098,
+      "step": 17930
+    },
+    {
+      "epoch": 0.57408,
+      "grad_norm": 44.0560302734375,
+      "learning_rate": 1.5843555555555556e-05,
+      "loss": 1.7095,
+      "step": 17940
+    },
+    {
+      "epoch": 0.5744,
+      "grad_norm": 44.2470817565918,
+      "learning_rate": 1.584e-05,
+      "loss": 1.7066,
+      "step": 17950
+    },
+    {
+      "epoch": 0.57472,
+      "grad_norm": 43.45963668823242,
+      "learning_rate": 1.5836444444444445e-05,
+      "loss": 1.6706,
+      "step": 17960
+    },
+    {
+      "epoch": 0.57504,
+      "grad_norm": 45.13947296142578,
+      "learning_rate": 1.583288888888889e-05,
+      "loss": 1.6806,
+      "step": 17970
+    },
+    {
+      "epoch": 0.57536,
+      "grad_norm": 45.662574768066406,
+      "learning_rate": 1.5829333333333334e-05,
+      "loss": 1.7123,
+      "step": 17980
+    },
+    {
+      "epoch": 0.57568,
+      "grad_norm": 44.67338562011719,
+      "learning_rate": 1.582577777777778e-05,
+      "loss": 1.6867,
+      "step": 17990
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 44.79176712036133,
+      "learning_rate": 1.5822222222222224e-05,
+      "loss": 1.7151,
+      "step": 18000
+    },
+    {
+      "epoch": 0.57632,
+      "grad_norm": 46.5061149597168,
+      "learning_rate": 1.5818666666666668e-05,
+      "loss": 1.7163,
+      "step": 18010
+    },
+    {
+      "epoch": 0.57664,
+      "grad_norm": 45.3877067565918,
+      "learning_rate": 1.5815111111111113e-05,
+      "loss": 1.7154,
+      "step": 18020
+    },
+    {
+      "epoch": 0.57696,
+      "grad_norm": 45.97566223144531,
+      "learning_rate": 1.5811555555555557e-05,
+      "loss": 1.6957,
+      "step": 18030
+    },
+    {
+      "epoch": 0.57728,
+      "grad_norm": 44.81576919555664,
+      "learning_rate": 1.5808000000000002e-05,
+      "loss": 1.6721,
+      "step": 18040
+    },
+    {
+      "epoch": 0.5776,
+      "grad_norm": 43.295509338378906,
+      "learning_rate": 1.5804444444444446e-05,
+      "loss": 1.6782,
+      "step": 18050
+    },
+    {
+      "epoch": 0.57792,
+      "grad_norm": 44.87993621826172,
+      "learning_rate": 1.580088888888889e-05,
+      "loss": 1.6799,
+      "step": 18060
+    },
+    {
+      "epoch": 0.57824,
+      "grad_norm": 43.835365295410156,
+      "learning_rate": 1.5797333333333335e-05,
+      "loss": 1.6654,
+      "step": 18070
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 46.73650360107422,
+      "learning_rate": 1.579377777777778e-05,
+      "loss": 1.6948,
+      "step": 18080
+    },
+    {
+      "epoch": 0.57888,
+      "grad_norm": 43.62708282470703,
+      "learning_rate": 1.5790222222222225e-05,
+      "loss": 1.6844,
+      "step": 18090
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 42.22301483154297,
+      "learning_rate": 1.578666666666667e-05,
+      "loss": 1.6359,
+      "step": 18100
+    },
+    {
+      "epoch": 0.57952,
+      "grad_norm": 43.23906326293945,
+      "learning_rate": 1.5783111111111114e-05,
+      "loss": 1.701,
+      "step": 18110
+    },
+    {
+      "epoch": 0.57984,
+      "grad_norm": 44.715187072753906,
+      "learning_rate": 1.5779555555555558e-05,
+      "loss": 1.6639,
+      "step": 18120
+    },
+    {
+      "epoch": 0.58016,
+      "grad_norm": 42.52825164794922,
+      "learning_rate": 1.5776e-05,
+      "loss": 1.6719,
+      "step": 18130
+    },
+    {
+      "epoch": 0.58048,
+      "grad_norm": 44.43534469604492,
+      "learning_rate": 1.5772444444444447e-05,
+      "loss": 1.7142,
+      "step": 18140
+    },
+    {
+      "epoch": 0.5808,
+      "grad_norm": 41.77783966064453,
+      "learning_rate": 1.576888888888889e-05,
+      "loss": 1.7062,
+      "step": 18150
+    },
+    {
+      "epoch": 0.58112,
+      "grad_norm": 44.75507354736328,
+      "learning_rate": 1.5765333333333333e-05,
+      "loss": 1.6819,
+      "step": 18160
+    },
+    {
+      "epoch": 0.58144,
+      "grad_norm": 47.013145446777344,
+      "learning_rate": 1.5761777777777778e-05,
+      "loss": 1.7089,
+      "step": 18170
+    },
+    {
+      "epoch": 0.58176,
+      "grad_norm": 45.202571868896484,
+      "learning_rate": 1.5758222222222222e-05,
+      "loss": 1.6876,
+      "step": 18180
+    },
+    {
+      "epoch": 0.58208,
+      "grad_norm": 43.15135192871094,
+      "learning_rate": 1.5754666666666667e-05,
+      "loss": 1.6999,
+      "step": 18190
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 44.06850051879883,
+      "learning_rate": 1.575111111111111e-05,
+      "loss": 1.7127,
+      "step": 18200
+    },
+    {
+      "epoch": 0.58272,
+      "grad_norm": 44.04061508178711,
+      "learning_rate": 1.5747555555555556e-05,
+      "loss": 1.6819,
+      "step": 18210
+    },
+    {
+      "epoch": 0.58304,
+      "grad_norm": 46.223114013671875,
+      "learning_rate": 1.5744e-05,
+      "loss": 1.6726,
+      "step": 18220
+    },
+    {
+      "epoch": 0.58336,
+      "grad_norm": 44.79490280151367,
+      "learning_rate": 1.5740444444444445e-05,
+      "loss": 1.6703,
+      "step": 18230
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 44.74020767211914,
+      "learning_rate": 1.573688888888889e-05,
+      "loss": 1.676,
+      "step": 18240
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 45.662872314453125,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.6959,
+      "step": 18250
+    },
+    {
+      "epoch": 0.58432,
+      "grad_norm": 43.779197692871094,
+      "learning_rate": 1.572977777777778e-05,
+      "loss": 1.7072,
+      "step": 18260
+    },
+    {
+      "epoch": 0.58464,
+      "grad_norm": 46.27597427368164,
+      "learning_rate": 1.5726222222222223e-05,
+      "loss": 1.7126,
+      "step": 18270
+    },
+    {
+      "epoch": 0.58496,
+      "grad_norm": 44.4024658203125,
+      "learning_rate": 1.5722666666666668e-05,
+      "loss": 1.6876,
+      "step": 18280
+    },
+    {
+      "epoch": 0.58528,
+      "grad_norm": 44.50633239746094,
+      "learning_rate": 1.5719111111111112e-05,
+      "loss": 1.7045,
+      "step": 18290
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 45.14801025390625,
+      "learning_rate": 1.5715555555555557e-05,
+      "loss": 1.6796,
+      "step": 18300
+    },
+    {
+      "epoch": 0.58592,
+      "grad_norm": 43.62449264526367,
+      "learning_rate": 1.5712e-05,
+      "loss": 1.6682,
+      "step": 18310
+    },
+    {
+      "epoch": 0.58624,
+      "grad_norm": 43.99118423461914,
+      "learning_rate": 1.5708444444444446e-05,
+      "loss": 1.6998,
+      "step": 18320
+    },
+    {
+      "epoch": 0.58656,
+      "grad_norm": 44.510032653808594,
+      "learning_rate": 1.570488888888889e-05,
+      "loss": 1.7034,
+      "step": 18330
+    },
+    {
+      "epoch": 0.58688,
+      "grad_norm": 44.636871337890625,
+      "learning_rate": 1.5701333333333335e-05,
+      "loss": 1.704,
+      "step": 18340
+    },
+    {
+      "epoch": 0.5872,
+      "grad_norm": 43.79024887084961,
+      "learning_rate": 1.569777777777778e-05,
+      "loss": 1.6864,
+      "step": 18350
+    },
+    {
+      "epoch": 0.58752,
+      "grad_norm": 45.9245719909668,
+      "learning_rate": 1.5694222222222224e-05,
+      "loss": 1.6936,
+      "step": 18360
+    },
+    {
+      "epoch": 0.58784,
+      "grad_norm": 44.46293640136719,
+      "learning_rate": 1.569066666666667e-05,
+      "loss": 1.7028,
+      "step": 18370
+    },
+    {
+      "epoch": 0.58816,
+      "grad_norm": 43.017608642578125,
+      "learning_rate": 1.5687111111111113e-05,
+      "loss": 1.6934,
+      "step": 18380
+    },
+    {
+      "epoch": 0.58848,
+      "grad_norm": 44.94241714477539,
+      "learning_rate": 1.5683555555555558e-05,
+      "loss": 1.6954,
+      "step": 18390
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 44.682857513427734,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 1.6912,
+      "step": 18400
+    },
+    {
+      "epoch": 0.58912,
+      "grad_norm": 46.192054748535156,
+      "learning_rate": 1.5676444444444447e-05,
+      "loss": 1.7088,
+      "step": 18410
+    },
+    {
+      "epoch": 0.58944,
+      "grad_norm": 44.73866271972656,
+      "learning_rate": 1.567288888888889e-05,
+      "loss": 1.6758,
+      "step": 18420
+    },
+    {
+      "epoch": 0.58976,
+      "grad_norm": 44.03883743286133,
+      "learning_rate": 1.5669333333333333e-05,
+      "loss": 1.6583,
+      "step": 18430
+    },
+    {
+      "epoch": 0.59008,
+      "grad_norm": 44.191524505615234,
+      "learning_rate": 1.566577777777778e-05,
+      "loss": 1.6804,
+      "step": 18440
+    },
+    {
+      "epoch": 0.5904,
+      "grad_norm": 45.07223892211914,
+      "learning_rate": 1.5662222222222222e-05,
+      "loss": 1.6737,
+      "step": 18450
+    },
+    {
+      "epoch": 0.59072,
+      "grad_norm": 45.84001541137695,
+      "learning_rate": 1.565866666666667e-05,
+      "loss": 1.6878,
+      "step": 18460
+    },
+    {
+      "epoch": 0.59104,
+      "grad_norm": 43.74607467651367,
+      "learning_rate": 1.565511111111111e-05,
+      "loss": 1.6927,
+      "step": 18470
+    },
+    {
+      "epoch": 0.59136,
+      "grad_norm": 43.94548034667969,
+      "learning_rate": 1.565155555555556e-05,
+      "loss": 1.671,
+      "step": 18480
+    },
+    {
+      "epoch": 0.59168,
+      "grad_norm": 45.99900817871094,
+      "learning_rate": 1.5648e-05,
+      "loss": 1.6671,
+      "step": 18490
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 45.90909194946289,
+      "learning_rate": 1.5644444444444448e-05,
+      "loss": 1.7005,
+      "step": 18500
+    },
+    {
+      "epoch": 0.59232,
+      "grad_norm": 45.09535217285156,
+      "learning_rate": 1.564088888888889e-05,
+      "loss": 1.6521,
+      "step": 18510
+    },
+    {
+      "epoch": 0.59264,
+      "grad_norm": 45.49460220336914,
+      "learning_rate": 1.5637333333333337e-05,
+      "loss": 1.6695,
+      "step": 18520
+    },
+    {
+      "epoch": 0.59296,
+      "grad_norm": 46.332801818847656,
+      "learning_rate": 1.5633777777777778e-05,
+      "loss": 1.7065,
+      "step": 18530
+    },
+    {
+      "epoch": 0.59328,
+      "grad_norm": 46.69112777709961,
+      "learning_rate": 1.5630222222222226e-05,
+      "loss": 1.7032,
+      "step": 18540
+    },
+    {
+      "epoch": 0.5936,
+      "grad_norm": 46.828651428222656,
+      "learning_rate": 1.5626666666666667e-05,
+      "loss": 1.7173,
+      "step": 18550
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 44.735477447509766,
+      "learning_rate": 1.5623111111111112e-05,
+      "loss": 1.6966,
+      "step": 18560
+    },
+    {
+      "epoch": 0.59424,
+      "grad_norm": 43.93488693237305,
+      "learning_rate": 1.5619555555555556e-05,
+      "loss": 1.6617,
+      "step": 18570
+    },
+    {
+      "epoch": 0.59456,
+      "grad_norm": 44.45972442626953,
+      "learning_rate": 1.5616e-05,
+      "loss": 1.6642,
+      "step": 18580
+    },
+    {
+      "epoch": 0.59488,
+      "grad_norm": 47.216346740722656,
+      "learning_rate": 1.5612444444444445e-05,
+      "loss": 1.7012,
+      "step": 18590
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 41.691734313964844,
+      "learning_rate": 1.560888888888889e-05,
+      "loss": 1.6542,
+      "step": 18600
+    },
+    {
+      "epoch": 0.59552,
+      "grad_norm": 46.06924057006836,
+      "learning_rate": 1.5605333333333335e-05,
+      "loss": 1.7038,
+      "step": 18610
+    },
+    {
+      "epoch": 0.59584,
+      "grad_norm": 42.43273162841797,
+      "learning_rate": 1.560177777777778e-05,
+      "loss": 1.6794,
+      "step": 18620
+    },
+    {
+      "epoch": 0.59616,
+      "grad_norm": 43.787010192871094,
+      "learning_rate": 1.5598222222222224e-05,
+      "loss": 1.703,
+      "step": 18630
+    },
+    {
+      "epoch": 0.59648,
+      "grad_norm": 43.158817291259766,
+      "learning_rate": 1.5594666666666668e-05,
+      "loss": 1.6894,
+      "step": 18640
+    },
+    {
+      "epoch": 0.5968,
+      "grad_norm": 44.50301742553711,
+      "learning_rate": 1.5591111111111113e-05,
+      "loss": 1.6775,
+      "step": 18650
+    },
+    {
+      "epoch": 0.59712,
+      "grad_norm": 43.681915283203125,
+      "learning_rate": 1.5587555555555557e-05,
+      "loss": 1.6754,
+      "step": 18660
+    },
+    {
+      "epoch": 0.59744,
+      "grad_norm": 43.70466995239258,
+      "learning_rate": 1.5584000000000002e-05,
+      "loss": 1.6803,
+      "step": 18670
+    },
+    {
+      "epoch": 0.59776,
+      "grad_norm": 45.950775146484375,
+      "learning_rate": 1.5580444444444443e-05,
+      "loss": 1.7145,
+      "step": 18680
+    },
+    {
+      "epoch": 0.59808,
+      "grad_norm": 44.928184509277344,
+      "learning_rate": 1.557688888888889e-05,
+      "loss": 1.6855,
+      "step": 18690
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 47.1044807434082,
+      "learning_rate": 1.5573333333333332e-05,
+      "loss": 1.6987,
+      "step": 18700
+    },
+    {
+      "epoch": 0.59872,
+      "grad_norm": 46.54291915893555,
+      "learning_rate": 1.556977777777778e-05,
+      "loss": 1.6644,
+      "step": 18710
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 44.95602035522461,
+      "learning_rate": 1.556622222222222e-05,
+      "loss": 1.6422,
+      "step": 18720
+    },
+    {
+      "epoch": 0.59936,
+      "grad_norm": 48.815269470214844,
+      "learning_rate": 1.556266666666667e-05,
+      "loss": 1.6894,
+      "step": 18730
+    },
+    {
+      "epoch": 0.59968,
+      "grad_norm": 44.7770881652832,
+      "learning_rate": 1.555911111111111e-05,
+      "loss": 1.707,
+      "step": 18740
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 44.44867706298828,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.6788,
+      "step": 18750
+    },
+    {
+      "epoch": 0.60032,
+      "grad_norm": 44.379432678222656,
+      "learning_rate": 1.5552e-05,
+      "loss": 1.6692,
+      "step": 18760
+    },
+    {
+      "epoch": 0.60064,
+      "grad_norm": 44.882110595703125,
+      "learning_rate": 1.5548444444444447e-05,
+      "loss": 1.7007,
+      "step": 18770
+    },
+    {
+      "epoch": 0.60096,
+      "grad_norm": 43.32308578491211,
+      "learning_rate": 1.554488888888889e-05,
+      "loss": 1.6706,
+      "step": 18780
+    },
+    {
+      "epoch": 0.60128,
+      "grad_norm": 45.213165283203125,
+      "learning_rate": 1.5541333333333337e-05,
+      "loss": 1.6923,
+      "step": 18790
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 45.2623176574707,
+      "learning_rate": 1.5537777777777778e-05,
+      "loss": 1.6939,
+      "step": 18800
+    },
+    {
+      "epoch": 0.60192,
+      "grad_norm": 43.48740005493164,
+      "learning_rate": 1.5534222222222222e-05,
+      "loss": 1.6966,
+      "step": 18810
+    },
+    {
+      "epoch": 0.60224,
+      "grad_norm": 46.13747787475586,
+      "learning_rate": 1.5530666666666667e-05,
+      "loss": 1.7097,
+      "step": 18820
+    },
+    {
+      "epoch": 0.60256,
+      "grad_norm": 47.08616638183594,
+      "learning_rate": 1.552711111111111e-05,
+      "loss": 1.6467,
+      "step": 18830
+    },
+    {
+      "epoch": 0.60288,
+      "grad_norm": 43.652793884277344,
+      "learning_rate": 1.5523555555555556e-05,
+      "loss": 1.673,
+      "step": 18840
+    },
+    {
+      "epoch": 0.6032,
+      "grad_norm": 45.95335006713867,
+      "learning_rate": 1.552e-05,
+      "loss": 1.6922,
+      "step": 18850
+    },
+    {
+      "epoch": 0.60352,
+      "grad_norm": 43.87394332885742,
+      "learning_rate": 1.5516444444444445e-05,
+      "loss": 1.6931,
+      "step": 18860
+    },
+    {
+      "epoch": 0.60384,
+      "grad_norm": 44.94473648071289,
+      "learning_rate": 1.551288888888889e-05,
+      "loss": 1.6792,
+      "step": 18870
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 42.34600830078125,
+      "learning_rate": 1.5509333333333334e-05,
+      "loss": 1.6824,
+      "step": 18880
+    },
+    {
+      "epoch": 0.60448,
+      "grad_norm": 47.03275680541992,
+      "learning_rate": 1.550577777777778e-05,
+      "loss": 1.6911,
+      "step": 18890
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 45.630332946777344,
+      "learning_rate": 1.5502222222222223e-05,
+      "loss": 1.6909,
+      "step": 18900
+    },
+    {
+      "epoch": 0.60512,
+      "grad_norm": 42.56125259399414,
+      "learning_rate": 1.5498666666666668e-05,
+      "loss": 1.6462,
+      "step": 18910
+    },
+    {
+      "epoch": 0.60544,
+      "grad_norm": 43.68327331542969,
+      "learning_rate": 1.5495111111111112e-05,
+      "loss": 1.691,
+      "step": 18920
+    },
+    {
+      "epoch": 0.60576,
+      "grad_norm": 46.32207107543945,
+      "learning_rate": 1.5491555555555557e-05,
+      "loss": 1.7196,
+      "step": 18930
+    },
+    {
+      "epoch": 0.60608,
+      "grad_norm": 46.93421936035156,
+      "learning_rate": 1.5488e-05,
+      "loss": 1.684,
+      "step": 18940
+    },
+    {
+      "epoch": 0.6064,
+      "grad_norm": 42.83504867553711,
+      "learning_rate": 1.5484444444444446e-05,
+      "loss": 1.659,
+      "step": 18950
+    },
+    {
+      "epoch": 0.60672,
+      "grad_norm": 44.189720153808594,
+      "learning_rate": 1.548088888888889e-05,
+      "loss": 1.6882,
+      "step": 18960
+    },
+    {
+      "epoch": 0.60704,
+      "grad_norm": 43.321807861328125,
+      "learning_rate": 1.5477333333333335e-05,
+      "loss": 1.7102,
+      "step": 18970
+    },
+    {
+      "epoch": 0.60736,
+      "grad_norm": 43.58332061767578,
+      "learning_rate": 1.547377777777778e-05,
+      "loss": 1.6748,
+      "step": 18980
+    },
+    {
+      "epoch": 0.60768,
+      "grad_norm": 43.99165344238281,
+      "learning_rate": 1.5470222222222224e-05,
+      "loss": 1.6678,
+      "step": 18990
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 43.6734619140625,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 1.6676,
+      "step": 19000
+    },
+    {
+      "epoch": 0.60832,
+      "grad_norm": 44.50790023803711,
+      "learning_rate": 1.5463111111111113e-05,
+      "loss": 1.6732,
+      "step": 19010
+    },
+    {
+      "epoch": 0.60864,
+      "grad_norm": 45.299781799316406,
+      "learning_rate": 1.5459555555555558e-05,
+      "loss": 1.7008,
+      "step": 19020
+    },
+    {
+      "epoch": 0.60896,
+      "grad_norm": 45.65017318725586,
+      "learning_rate": 1.5456000000000002e-05,
+      "loss": 1.7073,
+      "step": 19030
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 45.88054656982422,
+      "learning_rate": 1.5452444444444447e-05,
+      "loss": 1.6941,
+      "step": 19040
+    },
+    {
+      "epoch": 0.6096,
+      "grad_norm": 44.15474319458008,
+      "learning_rate": 1.544888888888889e-05,
+      "loss": 1.6764,
+      "step": 19050
+    },
+    {
+      "epoch": 0.60992,
+      "grad_norm": 43.338172912597656,
+      "learning_rate": 1.5445333333333333e-05,
+      "loss": 1.6635,
+      "step": 19060
+    },
+    {
+      "epoch": 0.61024,
+      "grad_norm": 45.02217483520508,
+      "learning_rate": 1.544177777777778e-05,
+      "loss": 1.7024,
+      "step": 19070
+    },
+    {
+      "epoch": 0.61056,
+      "grad_norm": 44.492401123046875,
+      "learning_rate": 1.5438222222222222e-05,
+      "loss": 1.7113,
+      "step": 19080
+    },
+    {
+      "epoch": 0.61088,
+      "grad_norm": 46.0334587097168,
+      "learning_rate": 1.543466666666667e-05,
+      "loss": 1.6678,
+      "step": 19090
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 44.86727523803711,
+      "learning_rate": 1.543111111111111e-05,
+      "loss": 1.6919,
+      "step": 19100
+    },
+    {
+      "epoch": 0.61152,
+      "grad_norm": 41.22093963623047,
+      "learning_rate": 1.542755555555556e-05,
+      "loss": 1.6657,
+      "step": 19110
+    },
+    {
+      "epoch": 0.61184,
+      "grad_norm": 43.157474517822266,
+      "learning_rate": 1.5424e-05,
+      "loss": 1.6803,
+      "step": 19120
+    },
+    {
+      "epoch": 0.61216,
+      "grad_norm": 44.455142974853516,
+      "learning_rate": 1.5420444444444448e-05,
+      "loss": 1.6962,
+      "step": 19130
+    },
+    {
+      "epoch": 0.61248,
+      "grad_norm": 44.34128189086914,
+      "learning_rate": 1.541688888888889e-05,
+      "loss": 1.6648,
+      "step": 19140
+    },
+    {
+      "epoch": 0.6128,
+      "grad_norm": 46.374366760253906,
+      "learning_rate": 1.5413333333333337e-05,
+      "loss": 1.6898,
+      "step": 19150
+    },
+    {
+      "epoch": 0.61312,
+      "grad_norm": 45.73741149902344,
+      "learning_rate": 1.5409777777777778e-05,
+      "loss": 1.6631,
+      "step": 19160
+    },
+    {
+      "epoch": 0.61344,
+      "grad_norm": 43.768558502197266,
+      "learning_rate": 1.5406222222222223e-05,
+      "loss": 1.6703,
+      "step": 19170
+    },
+    {
+      "epoch": 0.61376,
+      "grad_norm": 44.202598571777344,
+      "learning_rate": 1.5402666666666667e-05,
+      "loss": 1.6574,
+      "step": 19180
+    },
+    {
+      "epoch": 0.61408,
+      "grad_norm": 43.005733489990234,
+      "learning_rate": 1.5399111111111112e-05,
+      "loss": 1.6656,
+      "step": 19190
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 44.95966339111328,
+      "learning_rate": 1.5395555555555556e-05,
+      "loss": 1.6811,
+      "step": 19200
+    },
+    {
+      "epoch": 0.61472,
+      "grad_norm": 46.06792449951172,
+      "learning_rate": 1.5392e-05,
+      "loss": 1.6811,
+      "step": 19210
+    },
+    {
+      "epoch": 0.61504,
+      "grad_norm": 43.61574172973633,
+      "learning_rate": 1.5388444444444446e-05,
+      "loss": 1.6939,
+      "step": 19220
+    },
+    {
+      "epoch": 0.61536,
+      "grad_norm": 45.055580139160156,
+      "learning_rate": 1.538488888888889e-05,
+      "loss": 1.7022,
+      "step": 19230
+    },
+    {
+      "epoch": 0.61568,
+      "grad_norm": 43.75603485107422,
+      "learning_rate": 1.5381333333333335e-05,
+      "loss": 1.6823,
+      "step": 19240
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 44.875606536865234,
+      "learning_rate": 1.537777777777778e-05,
+      "loss": 1.6648,
+      "step": 19250
+    },
+    {
+      "epoch": 0.61632,
+      "grad_norm": 45.77470016479492,
+      "learning_rate": 1.5374222222222224e-05,
+      "loss": 1.6661,
+      "step": 19260
+    },
+    {
+      "epoch": 0.61664,
+      "grad_norm": 43.80479431152344,
+      "learning_rate": 1.537066666666667e-05,
+      "loss": 1.6986,
+      "step": 19270
+    },
+    {
+      "epoch": 0.61696,
+      "grad_norm": 44.32920837402344,
+      "learning_rate": 1.5367111111111113e-05,
+      "loss": 1.7064,
+      "step": 19280
+    },
+    {
+      "epoch": 0.61728,
+      "grad_norm": 41.32722091674805,
+      "learning_rate": 1.5363555555555557e-05,
+      "loss": 1.69,
+      "step": 19290
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 43.27383041381836,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 1.6953,
+      "step": 19300
+    },
+    {
+      "epoch": 0.61792,
+      "grad_norm": 44.69233322143555,
+      "learning_rate": 1.5356444444444447e-05,
+      "loss": 1.6642,
+      "step": 19310
+    },
+    {
+      "epoch": 0.61824,
+      "grad_norm": 44.23918914794922,
+      "learning_rate": 1.535288888888889e-05,
+      "loss": 1.6615,
+      "step": 19320
+    },
+    {
+      "epoch": 0.61856,
+      "grad_norm": 43.73684310913086,
+      "learning_rate": 1.5349333333333332e-05,
+      "loss": 1.6781,
+      "step": 19330
+    },
+    {
+      "epoch": 0.61888,
+      "grad_norm": 42.925804138183594,
+      "learning_rate": 1.534577777777778e-05,
+      "loss": 1.6809,
+      "step": 19340
+    },
+    {
+      "epoch": 0.6192,
+      "grad_norm": 43.84880065917969,
+      "learning_rate": 1.534222222222222e-05,
+      "loss": 1.708,
+      "step": 19350
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 45.829620361328125,
+      "learning_rate": 1.533866666666667e-05,
+      "loss": 1.6932,
+      "step": 19360
+    },
+    {
+      "epoch": 0.61984,
+      "grad_norm": 43.478153228759766,
+      "learning_rate": 1.533511111111111e-05,
+      "loss": 1.6811,
+      "step": 19370
+    },
+    {
+      "epoch": 0.62016,
+      "grad_norm": 45.014156341552734,
+      "learning_rate": 1.533155555555556e-05,
+      "loss": 1.7159,
+      "step": 19380
+    },
+    {
+      "epoch": 0.62048,
+      "grad_norm": 43.75464630126953,
+      "learning_rate": 1.5328e-05,
+      "loss": 1.6799,
+      "step": 19390
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 42.851783752441406,
+      "learning_rate": 1.5324444444444448e-05,
+      "loss": 1.6647,
+      "step": 19400
+    },
+    {
+      "epoch": 0.62112,
+      "grad_norm": 42.745121002197266,
+      "learning_rate": 1.532088888888889e-05,
+      "loss": 1.6471,
+      "step": 19410
+    },
+    {
+      "epoch": 0.62144,
+      "grad_norm": 44.881351470947266,
+      "learning_rate": 1.5317333333333337e-05,
+      "loss": 1.7028,
+      "step": 19420
+    },
+    {
+      "epoch": 0.62176,
+      "grad_norm": 45.286781311035156,
+      "learning_rate": 1.5313777777777778e-05,
+      "loss": 1.7154,
+      "step": 19430
+    },
+    {
+      "epoch": 0.62208,
+      "grad_norm": 44.99727249145508,
+      "learning_rate": 1.5310222222222226e-05,
+      "loss": 1.6808,
+      "step": 19440
+    },
+    {
+      "epoch": 0.6224,
+      "grad_norm": 44.824974060058594,
+      "learning_rate": 1.5306666666666667e-05,
+      "loss": 1.6674,
+      "step": 19450
+    },
+    {
+      "epoch": 0.62272,
+      "grad_norm": 44.57046127319336,
+      "learning_rate": 1.530311111111111e-05,
+      "loss": 1.6796,
+      "step": 19460
+    },
+    {
+      "epoch": 0.62304,
+      "grad_norm": 43.56769561767578,
+      "learning_rate": 1.5299555555555556e-05,
+      "loss": 1.7004,
+      "step": 19470
+    },
+    {
+      "epoch": 0.62336,
+      "grad_norm": 44.90610885620117,
+      "learning_rate": 1.5296e-05,
+      "loss": 1.6807,
+      "step": 19480
+    },
+    {
+      "epoch": 0.62368,
+      "grad_norm": 43.514404296875,
+      "learning_rate": 1.5292444444444445e-05,
+      "loss": 1.6982,
+      "step": 19490
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 46.587955474853516,
+      "learning_rate": 1.528888888888889e-05,
+      "loss": 1.7377,
+      "step": 19500
+    },
+    {
+      "epoch": 0.62432,
+      "grad_norm": 44.5979118347168,
+      "learning_rate": 1.5285333333333334e-05,
+      "loss": 1.6597,
+      "step": 19510
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 44.230716705322266,
+      "learning_rate": 1.528177777777778e-05,
+      "loss": 1.6646,
+      "step": 19520
+    },
+    {
+      "epoch": 0.62496,
+      "grad_norm": 44.42688751220703,
+      "learning_rate": 1.5278222222222223e-05,
+      "loss": 1.6968,
+      "step": 19530
+    },
+    {
+      "epoch": 0.62528,
+      "grad_norm": 41.63092803955078,
+      "learning_rate": 1.5274666666666668e-05,
+      "loss": 1.6577,
+      "step": 19540
+    },
+    {
+      "epoch": 0.6256,
+      "grad_norm": 44.5085563659668,
+      "learning_rate": 1.5271111111111112e-05,
+      "loss": 1.6844,
+      "step": 19550
+    },
+    {
+      "epoch": 0.62592,
+      "grad_norm": 44.731101989746094,
+      "learning_rate": 1.5267555555555557e-05,
+      "loss": 1.649,
+      "step": 19560
+    },
+    {
+      "epoch": 0.62624,
+      "grad_norm": 44.423377990722656,
+      "learning_rate": 1.5264e-05,
+      "loss": 1.6601,
+      "step": 19570
+    },
+    {
+      "epoch": 0.62656,
+      "grad_norm": 43.00068283081055,
+      "learning_rate": 1.5260444444444446e-05,
+      "loss": 1.6932,
+      "step": 19580
+    },
+    {
+      "epoch": 0.62688,
+      "grad_norm": 43.62189865112305,
+      "learning_rate": 1.525688888888889e-05,
+      "loss": 1.693,
+      "step": 19590
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 45.53034973144531,
+      "learning_rate": 1.5253333333333335e-05,
+      "loss": 1.6802,
+      "step": 19600
+    },
+    {
+      "epoch": 0.62752,
+      "grad_norm": 43.723388671875,
+      "learning_rate": 1.5249777777777778e-05,
+      "loss": 1.6553,
+      "step": 19610
+    },
+    {
+      "epoch": 0.62784,
+      "grad_norm": 42.24656295776367,
+      "learning_rate": 1.5246222222222224e-05,
+      "loss": 1.674,
+      "step": 19620
+    },
+    {
+      "epoch": 0.62816,
+      "grad_norm": 44.599510192871094,
+      "learning_rate": 1.5242666666666667e-05,
+      "loss": 1.6999,
+      "step": 19630
+    },
+    {
+      "epoch": 0.62848,
+      "grad_norm": 46.729061126708984,
+      "learning_rate": 1.5239111111111113e-05,
+      "loss": 1.6938,
+      "step": 19640
+    },
+    {
+      "epoch": 0.6288,
+      "grad_norm": 44.45905685424805,
+      "learning_rate": 1.5235555555555556e-05,
+      "loss": 1.6951,
+      "step": 19650
+    },
+    {
+      "epoch": 0.62912,
+      "grad_norm": 45.92387390136719,
+      "learning_rate": 1.5232000000000003e-05,
+      "loss": 1.6967,
+      "step": 19660
+    },
+    {
+      "epoch": 0.62944,
+      "grad_norm": 44.020652770996094,
+      "learning_rate": 1.5228444444444445e-05,
+      "loss": 1.683,
+      "step": 19670
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 44.318145751953125,
+      "learning_rate": 1.5224888888888888e-05,
+      "loss": 1.6638,
+      "step": 19680
+    },
+    {
+      "epoch": 0.63008,
+      "grad_norm": 42.7271842956543,
+      "learning_rate": 1.5221333333333335e-05,
+      "loss": 1.6636,
+      "step": 19690
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 42.00227355957031,
+      "learning_rate": 1.5217777777777777e-05,
+      "loss": 1.6705,
+      "step": 19700
+    },
+    {
+      "epoch": 0.63072,
+      "grad_norm": 41.81639862060547,
+      "learning_rate": 1.5214222222222224e-05,
+      "loss": 1.6666,
+      "step": 19710
+    },
+    {
+      "epoch": 0.63104,
+      "grad_norm": 45.459503173828125,
+      "learning_rate": 1.5210666666666666e-05,
+      "loss": 1.6632,
+      "step": 19720
+    },
+    {
+      "epoch": 0.63136,
+      "grad_norm": 43.953548431396484,
+      "learning_rate": 1.5207111111111113e-05,
+      "loss": 1.6466,
+      "step": 19730
+    },
+    {
+      "epoch": 0.63168,
+      "grad_norm": 43.911537170410156,
+      "learning_rate": 1.5203555555555556e-05,
+      "loss": 1.67,
+      "step": 19740
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 45.1783332824707,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.6857,
+      "step": 19750
+    },
+    {
+      "epoch": 0.63232,
+      "grad_norm": 43.7866096496582,
+      "learning_rate": 1.5196444444444445e-05,
+      "loss": 1.6848,
+      "step": 19760
+    },
+    {
+      "epoch": 0.63264,
+      "grad_norm": 43.66336441040039,
+      "learning_rate": 1.5192888888888891e-05,
+      "loss": 1.6834,
+      "step": 19770
+    },
+    {
+      "epoch": 0.63296,
+      "grad_norm": 44.24501037597656,
+      "learning_rate": 1.5189333333333334e-05,
+      "loss": 1.6539,
+      "step": 19780
+    },
+    {
+      "epoch": 0.63328,
+      "grad_norm": 44.39887619018555,
+      "learning_rate": 1.518577777777778e-05,
+      "loss": 1.6713,
+      "step": 19790
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 45.32782745361328,
+      "learning_rate": 1.5182222222222223e-05,
+      "loss": 1.6631,
+      "step": 19800
+    },
+    {
+      "epoch": 0.63392,
+      "grad_norm": 45.3266716003418,
+      "learning_rate": 1.5178666666666667e-05,
+      "loss": 1.6648,
+      "step": 19810
+    },
+    {
+      "epoch": 0.63424,
+      "grad_norm": 43.484432220458984,
+      "learning_rate": 1.5175111111111112e-05,
+      "loss": 1.6613,
+      "step": 19820
+    },
+    {
+      "epoch": 0.63456,
+      "grad_norm": 43.348140716552734,
+      "learning_rate": 1.5171555555555557e-05,
+      "loss": 1.671,
+      "step": 19830
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 45.595542907714844,
+      "learning_rate": 1.5168000000000001e-05,
+      "loss": 1.662,
+      "step": 19840
+    },
+    {
+      "epoch": 0.6352,
+      "grad_norm": 45.99067306518555,
+      "learning_rate": 1.5164444444444446e-05,
+      "loss": 1.6731,
+      "step": 19850
+    },
+    {
+      "epoch": 0.63552,
+      "grad_norm": 43.89120864868164,
+      "learning_rate": 1.516088888888889e-05,
+      "loss": 1.6906,
+      "step": 19860
+    },
+    {
+      "epoch": 0.63584,
+      "grad_norm": 44.42757797241211,
+      "learning_rate": 1.5157333333333335e-05,
+      "loss": 1.6891,
+      "step": 19870
+    },
+    {
+      "epoch": 0.63616,
+      "grad_norm": 42.13017654418945,
+      "learning_rate": 1.5153777777777778e-05,
+      "loss": 1.6918,
+      "step": 19880
+    },
+    {
+      "epoch": 0.63648,
+      "grad_norm": 43.843170166015625,
+      "learning_rate": 1.5150222222222224e-05,
+      "loss": 1.6718,
+      "step": 19890
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 44.21430587768555,
+      "learning_rate": 1.5146666666666667e-05,
+      "loss": 1.6957,
+      "step": 19900
+    },
+    {
+      "epoch": 0.63712,
+      "grad_norm": 41.84242630004883,
+      "learning_rate": 1.5143111111111113e-05,
+      "loss": 1.6794,
+      "step": 19910
+    },
+    {
+      "epoch": 0.63744,
+      "grad_norm": 43.27943801879883,
+      "learning_rate": 1.5139555555555556e-05,
+      "loss": 1.7045,
+      "step": 19920
+    },
+    {
+      "epoch": 0.63776,
+      "grad_norm": 45.409423828125,
+      "learning_rate": 1.5136000000000002e-05,
+      "loss": 1.7081,
+      "step": 19930
+    },
+    {
+      "epoch": 0.63808,
+      "grad_norm": 42.21131896972656,
+      "learning_rate": 1.5132444444444445e-05,
+      "loss": 1.7003,
+      "step": 19940
+    },
+    {
+      "epoch": 0.6384,
+      "grad_norm": 44.100101470947266,
+      "learning_rate": 1.5128888888888891e-05,
+      "loss": 1.6741,
+      "step": 19950
+    },
+    {
+      "epoch": 0.63872,
+      "grad_norm": 46.517189025878906,
+      "learning_rate": 1.5125333333333334e-05,
+      "loss": 1.6851,
+      "step": 19960
+    },
+    {
+      "epoch": 0.63904,
+      "grad_norm": 47.045021057128906,
+      "learning_rate": 1.512177777777778e-05,
+      "loss": 1.6922,
+      "step": 19970
+    },
+    {
+      "epoch": 0.63936,
+      "grad_norm": 44.45469665527344,
+      "learning_rate": 1.5118222222222223e-05,
+      "loss": 1.6872,
+      "step": 19980
+    },
+    {
+      "epoch": 0.63968,
+      "grad_norm": 43.917457580566406,
+      "learning_rate": 1.5114666666666668e-05,
+      "loss": 1.6584,
+      "step": 19990
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 45.960819244384766,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 1.7114,
+      "step": 20000
+    },
+    {
+      "epoch": 0.64032,
+      "grad_norm": 45.35646438598633,
+      "learning_rate": 1.5107555555555557e-05,
+      "loss": 1.6932,
+      "step": 20010
+    },
+    {
+      "epoch": 0.64064,
+      "grad_norm": 42.95730972290039,
+      "learning_rate": 1.5104000000000001e-05,
+      "loss": 1.6928,
+      "step": 20020
+    },
+    {
+      "epoch": 0.64096,
+      "grad_norm": 43.41942596435547,
+      "learning_rate": 1.5100444444444446e-05,
+      "loss": 1.6895,
+      "step": 20030
+    },
+    {
+      "epoch": 0.64128,
+      "grad_norm": 44.41860580444336,
+      "learning_rate": 1.509688888888889e-05,
+      "loss": 1.6505,
+      "step": 20040
+    },
+    {
+      "epoch": 0.6416,
+      "grad_norm": 46.77073287963867,
+      "learning_rate": 1.5093333333333335e-05,
+      "loss": 1.6877,
+      "step": 20050
+    },
+    {
+      "epoch": 0.64192,
+      "grad_norm": 43.96682357788086,
+      "learning_rate": 1.5089777777777778e-05,
+      "loss": 1.6668,
+      "step": 20060
+    },
+    {
+      "epoch": 0.64224,
+      "grad_norm": 41.405120849609375,
+      "learning_rate": 1.5086222222222224e-05,
+      "loss": 1.681,
+      "step": 20070
+    },
+    {
+      "epoch": 0.64256,
+      "grad_norm": 44.63652038574219,
+      "learning_rate": 1.5082666666666667e-05,
+      "loss": 1.7161,
+      "step": 20080
+    },
+    {
+      "epoch": 0.64288,
+      "grad_norm": 44.282630920410156,
+      "learning_rate": 1.5079111111111113e-05,
+      "loss": 1.6709,
+      "step": 20090
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 47.07522964477539,
+      "learning_rate": 1.5075555555555556e-05,
+      "loss": 1.6955,
+      "step": 20100
+    },
+    {
+      "epoch": 0.64352,
+      "grad_norm": 45.80629348754883,
+      "learning_rate": 1.5072000000000002e-05,
+      "loss": 1.6828,
+      "step": 20110
+    },
+    {
+      "epoch": 0.64384,
+      "grad_norm": 45.871124267578125,
+      "learning_rate": 1.5068444444444445e-05,
+      "loss": 1.6864,
+      "step": 20120
+    },
+    {
+      "epoch": 0.64416,
+      "grad_norm": 44.659149169921875,
+      "learning_rate": 1.5064888888888892e-05,
+      "loss": 1.6907,
+      "step": 20130
+    },
+    {
+      "epoch": 0.64448,
+      "grad_norm": 43.87977600097656,
+      "learning_rate": 1.5061333333333334e-05,
+      "loss": 1.6988,
+      "step": 20140
+    },
+    {
+      "epoch": 0.6448,
+      "grad_norm": 44.692874908447266,
+      "learning_rate": 1.505777777777778e-05,
+      "loss": 1.683,
+      "step": 20150
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 43.429168701171875,
+      "learning_rate": 1.5054222222222223e-05,
+      "loss": 1.6656,
+      "step": 20160
+    },
+    {
+      "epoch": 0.64544,
+      "grad_norm": 43.82966613769531,
+      "learning_rate": 1.505066666666667e-05,
+      "loss": 1.6797,
+      "step": 20170
+    },
+    {
+      "epoch": 0.64576,
+      "grad_norm": 42.469825744628906,
+      "learning_rate": 1.5047111111111113e-05,
+      "loss": 1.6828,
+      "step": 20180
+    },
+    {
+      "epoch": 0.64608,
+      "grad_norm": 45.71176528930664,
+      "learning_rate": 1.5043555555555555e-05,
+      "loss": 1.7081,
+      "step": 20190
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 43.655677795410156,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 1.7044,
+      "step": 20200
+    },
+    {
+      "epoch": 0.64672,
+      "grad_norm": 45.45912170410156,
+      "learning_rate": 1.5036444444444445e-05,
+      "loss": 1.6965,
+      "step": 20210
+    },
+    {
+      "epoch": 0.64704,
+      "grad_norm": 43.07465362548828,
+      "learning_rate": 1.503288888888889e-05,
+      "loss": 1.6755,
+      "step": 20220
+    },
+    {
+      "epoch": 0.64736,
+      "grad_norm": 44.67756652832031,
+      "learning_rate": 1.5029333333333334e-05,
+      "loss": 1.7036,
+      "step": 20230
+    },
+    {
+      "epoch": 0.64768,
+      "grad_norm": 45.22612762451172,
+      "learning_rate": 1.502577777777778e-05,
+      "loss": 1.6593,
+      "step": 20240
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 43.58930587768555,
+      "learning_rate": 1.5022222222222223e-05,
+      "loss": 1.6873,
+      "step": 20250
+    },
+    {
+      "epoch": 0.64832,
+      "grad_norm": 42.889339447021484,
+      "learning_rate": 1.5018666666666667e-05,
+      "loss": 1.6751,
+      "step": 20260
+    },
+    {
+      "epoch": 0.64864,
+      "grad_norm": 43.393394470214844,
+      "learning_rate": 1.5015111111111112e-05,
+      "loss": 1.6846,
+      "step": 20270
+    },
+    {
+      "epoch": 0.64896,
+      "grad_norm": 47.57592010498047,
+      "learning_rate": 1.5011555555555556e-05,
+      "loss": 1.6632,
+      "step": 20280
+    },
+    {
+      "epoch": 0.64928,
+      "grad_norm": 43.86087417602539,
+      "learning_rate": 1.5008000000000001e-05,
+      "loss": 1.6643,
+      "step": 20290
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 43.808414459228516,
+      "learning_rate": 1.5004444444444446e-05,
+      "loss": 1.7019,
+      "step": 20300
+    },
+    {
+      "epoch": 0.64992,
+      "grad_norm": 44.32374572753906,
+      "learning_rate": 1.500088888888889e-05,
+      "loss": 1.6773,
+      "step": 20310
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 45.46870422363281,
+      "learning_rate": 1.4997333333333335e-05,
+      "loss": 1.6681,
+      "step": 20320
+    },
+    {
+      "epoch": 0.65056,
+      "grad_norm": 45.13983917236328,
+      "learning_rate": 1.4993777777777778e-05,
+      "loss": 1.6726,
+      "step": 20330
+    },
+    {
+      "epoch": 0.65088,
+      "grad_norm": 44.52128982543945,
+      "learning_rate": 1.4990222222222224e-05,
+      "loss": 1.6837,
+      "step": 20340
+    },
+    {
+      "epoch": 0.6512,
+      "grad_norm": 45.18207931518555,
+      "learning_rate": 1.4986666666666667e-05,
+      "loss": 1.6595,
+      "step": 20350
+    },
+    {
+      "epoch": 0.65152,
+      "grad_norm": 44.02824020385742,
+      "learning_rate": 1.4983111111111113e-05,
+      "loss": 1.6353,
+      "step": 20360
+    },
+    {
+      "epoch": 0.65184,
+      "grad_norm": 42.066959381103516,
+      "learning_rate": 1.4979555555555556e-05,
+      "loss": 1.6624,
+      "step": 20370
+    },
+    {
+      "epoch": 0.65216,
+      "grad_norm": 45.960731506347656,
+      "learning_rate": 1.4976000000000002e-05,
+      "loss": 1.649,
+      "step": 20380
+    },
+    {
+      "epoch": 0.65248,
+      "grad_norm": 43.551265716552734,
+      "learning_rate": 1.4972444444444445e-05,
+      "loss": 1.6984,
+      "step": 20390
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 44.72035217285156,
+      "learning_rate": 1.4968888888888891e-05,
+      "loss": 1.6866,
+      "step": 20400
+    },
+    {
+      "epoch": 0.65312,
+      "grad_norm": 44.827938079833984,
+      "learning_rate": 1.4965333333333334e-05,
+      "loss": 1.6996,
+      "step": 20410
+    },
+    {
+      "epoch": 0.65344,
+      "grad_norm": 43.63932800292969,
+      "learning_rate": 1.496177777777778e-05,
+      "loss": 1.6786,
+      "step": 20420
+    },
+    {
+      "epoch": 0.65376,
+      "grad_norm": 44.68216323852539,
+      "learning_rate": 1.4958222222222223e-05,
+      "loss": 1.6908,
+      "step": 20430
+    },
+    {
+      "epoch": 0.65408,
+      "grad_norm": 44.0313835144043,
+      "learning_rate": 1.495466666666667e-05,
+      "loss": 1.7021,
+      "step": 20440
+    },
+    {
+      "epoch": 0.6544,
+      "grad_norm": 44.26289367675781,
+      "learning_rate": 1.4951111111111112e-05,
+      "loss": 1.6965,
+      "step": 20450
+    },
+    {
+      "epoch": 0.65472,
+      "grad_norm": 46.12016296386719,
+      "learning_rate": 1.4947555555555557e-05,
+      "loss": 1.6913,
+      "step": 20460
+    },
+    {
+      "epoch": 0.65504,
+      "grad_norm": 43.7845458984375,
+      "learning_rate": 1.4944000000000001e-05,
+      "loss": 1.6764,
+      "step": 20470
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 43.606197357177734,
+      "learning_rate": 1.4940444444444446e-05,
+      "loss": 1.6903,
+      "step": 20480
+    },
+    {
+      "epoch": 0.65568,
+      "grad_norm": 44.065025329589844,
+      "learning_rate": 1.493688888888889e-05,
+      "loss": 1.6771,
+      "step": 20490
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 44.80516052246094,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 1.6638,
+      "step": 20500
+    },
+    {
+      "epoch": 0.65632,
+      "grad_norm": 45.39891052246094,
+      "learning_rate": 1.4929777777777778e-05,
+      "loss": 1.6808,
+      "step": 20510
+    },
+    {
+      "epoch": 0.65664,
+      "grad_norm": 44.93928146362305,
+      "learning_rate": 1.4926222222222224e-05,
+      "loss": 1.6991,
+      "step": 20520
+    },
+    {
+      "epoch": 0.65696,
+      "grad_norm": 42.415870666503906,
+      "learning_rate": 1.4922666666666667e-05,
+      "loss": 1.6825,
+      "step": 20530
+    },
+    {
+      "epoch": 0.65728,
+      "grad_norm": 44.756160736083984,
+      "learning_rate": 1.4919111111111113e-05,
+      "loss": 1.6651,
+      "step": 20540
+    },
+    {
+      "epoch": 0.6576,
+      "grad_norm": 43.42251968383789,
+      "learning_rate": 1.4915555555555556e-05,
+      "loss": 1.6734,
+      "step": 20550
+    },
+    {
+      "epoch": 0.65792,
+      "grad_norm": 43.07693099975586,
+      "learning_rate": 1.4912000000000002e-05,
+      "loss": 1.673,
+      "step": 20560
+    },
+    {
+      "epoch": 0.65824,
+      "grad_norm": 45.04411697387695,
+      "learning_rate": 1.4908444444444445e-05,
+      "loss": 1.6723,
+      "step": 20570
+    },
+    {
+      "epoch": 0.65856,
+      "grad_norm": 44.08964920043945,
+      "learning_rate": 1.4904888888888891e-05,
+      "loss": 1.675,
+      "step": 20580
+    },
+    {
+      "epoch": 0.65888,
+      "grad_norm": 41.829559326171875,
+      "learning_rate": 1.4901333333333334e-05,
+      "loss": 1.6365,
+      "step": 20590
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 43.900211334228516,
+      "learning_rate": 1.489777777777778e-05,
+      "loss": 1.6787,
+      "step": 20600
+    },
+    {
+      "epoch": 0.65952,
+      "grad_norm": 47.46596908569336,
+      "learning_rate": 1.4894222222222223e-05,
+      "loss": 1.6487,
+      "step": 20610
+    },
+    {
+      "epoch": 0.65984,
+      "grad_norm": 43.5471076965332,
+      "learning_rate": 1.489066666666667e-05,
+      "loss": 1.7001,
+      "step": 20620
+    },
+    {
+      "epoch": 0.66016,
+      "grad_norm": 44.62773895263672,
+      "learning_rate": 1.4887111111111112e-05,
+      "loss": 1.7028,
+      "step": 20630
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 43.4677619934082,
+      "learning_rate": 1.4883555555555557e-05,
+      "loss": 1.6899,
+      "step": 20640
+    },
+    {
+      "epoch": 0.6608,
+      "grad_norm": 43.40306854248047,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 1.6764,
+      "step": 20650
+    },
+    {
+      "epoch": 0.66112,
+      "grad_norm": 46.035621643066406,
+      "learning_rate": 1.4876444444444446e-05,
+      "loss": 1.676,
+      "step": 20660
+    },
+    {
+      "epoch": 0.66144,
+      "grad_norm": 45.9166145324707,
+      "learning_rate": 1.487288888888889e-05,
+      "loss": 1.6585,
+      "step": 20670
+    },
+    {
+      "epoch": 0.66176,
+      "grad_norm": 44.903480529785156,
+      "learning_rate": 1.4869333333333335e-05,
+      "loss": 1.6823,
+      "step": 20680
+    },
+    {
+      "epoch": 0.66208,
+      "grad_norm": 44.58999252319336,
+      "learning_rate": 1.486577777777778e-05,
+      "loss": 1.6352,
+      "step": 20690
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 44.45679473876953,
+      "learning_rate": 1.4862222222222223e-05,
+      "loss": 1.6689,
+      "step": 20700
+    },
+    {
+      "epoch": 0.66272,
+      "grad_norm": 46.027244567871094,
+      "learning_rate": 1.4858666666666667e-05,
+      "loss": 1.6583,
+      "step": 20710
+    },
+    {
+      "epoch": 0.66304,
+      "grad_norm": 46.073158264160156,
+      "learning_rate": 1.4855111111111112e-05,
+      "loss": 1.7041,
+      "step": 20720
+    },
+    {
+      "epoch": 0.66336,
+      "grad_norm": 46.57586669921875,
+      "learning_rate": 1.4851555555555556e-05,
+      "loss": 1.7034,
+      "step": 20730
+    },
+    {
+      "epoch": 0.66368,
+      "grad_norm": 42.948909759521484,
+      "learning_rate": 1.4848e-05,
+      "loss": 1.6693,
+      "step": 20740
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 45.48905944824219,
+      "learning_rate": 1.4844444444444445e-05,
+      "loss": 1.7216,
+      "step": 20750
+    },
+    {
+      "epoch": 0.66432,
+      "grad_norm": 44.7623405456543,
+      "learning_rate": 1.484088888888889e-05,
+      "loss": 1.6741,
+      "step": 20760
+    },
+    {
+      "epoch": 0.66464,
+      "grad_norm": 45.26624298095703,
+      "learning_rate": 1.4837333333333334e-05,
+      "loss": 1.6803,
+      "step": 20770
+    },
+    {
+      "epoch": 0.66496,
+      "grad_norm": 43.05164337158203,
+      "learning_rate": 1.4833777777777777e-05,
+      "loss": 1.6731,
+      "step": 20780
+    },
+    {
+      "epoch": 0.66528,
+      "grad_norm": 43.42141342163086,
+      "learning_rate": 1.4830222222222224e-05,
+      "loss": 1.6429,
+      "step": 20790
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 43.655452728271484,
+      "learning_rate": 1.4826666666666666e-05,
+      "loss": 1.6912,
+      "step": 20800
+    },
+    {
+      "epoch": 0.66592,
+      "grad_norm": 44.76203536987305,
+      "learning_rate": 1.4823111111111113e-05,
+      "loss": 1.6713,
+      "step": 20810
+    },
+    {
+      "epoch": 0.66624,
+      "grad_norm": 44.26459503173828,
+      "learning_rate": 1.4819555555555556e-05,
+      "loss": 1.6916,
+      "step": 20820
+    },
+    {
+      "epoch": 0.66656,
+      "grad_norm": 42.33543014526367,
+      "learning_rate": 1.4816000000000002e-05,
+      "loss": 1.6932,
+      "step": 20830
+    },
+    {
+      "epoch": 0.66688,
+      "grad_norm": 43.858802795410156,
+      "learning_rate": 1.4812444444444445e-05,
+      "loss": 1.6687,
+      "step": 20840
+    },
+    {
+      "epoch": 0.6672,
+      "grad_norm": 45.832191467285156,
+      "learning_rate": 1.4808888888888891e-05,
+      "loss": 1.6923,
+      "step": 20850
+    },
+    {
+      "epoch": 0.66752,
+      "grad_norm": 44.79924774169922,
+      "learning_rate": 1.4805333333333334e-05,
+      "loss": 1.6613,
+      "step": 20860
+    },
+    {
+      "epoch": 0.66784,
+      "grad_norm": 45.21271514892578,
+      "learning_rate": 1.480177777777778e-05,
+      "loss": 1.6804,
+      "step": 20870
+    },
+    {
+      "epoch": 0.66816,
+      "grad_norm": 45.01856994628906,
+      "learning_rate": 1.4798222222222223e-05,
+      "loss": 1.6578,
+      "step": 20880
+    },
+    {
+      "epoch": 0.66848,
+      "grad_norm": 43.530738830566406,
+      "learning_rate": 1.4794666666666669e-05,
+      "loss": 1.6844,
+      "step": 20890
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 46.218666076660156,
+      "learning_rate": 1.4791111111111112e-05,
+      "loss": 1.6917,
+      "step": 20900
+    },
+    {
+      "epoch": 0.66912,
+      "grad_norm": 47.14932632446289,
+      "learning_rate": 1.4787555555555557e-05,
+      "loss": 1.6595,
+      "step": 20910
+    },
+    {
+      "epoch": 0.66944,
+      "grad_norm": 47.606266021728516,
+      "learning_rate": 1.4784000000000001e-05,
+      "loss": 1.6732,
+      "step": 20920
+    },
+    {
+      "epoch": 0.66976,
+      "grad_norm": 43.61834716796875,
+      "learning_rate": 1.4780444444444446e-05,
+      "loss": 1.668,
+      "step": 20930
+    },
+    {
+      "epoch": 0.67008,
+      "grad_norm": 43.1433219909668,
+      "learning_rate": 1.477688888888889e-05,
+      "loss": 1.6584,
+      "step": 20940
+    },
+    {
+      "epoch": 0.6704,
+      "grad_norm": 43.34695053100586,
+      "learning_rate": 1.4773333333333335e-05,
+      "loss": 1.6478,
+      "step": 20950
+    },
+    {
+      "epoch": 0.67072,
+      "grad_norm": 44.85986328125,
+      "learning_rate": 1.4769777777777778e-05,
+      "loss": 1.6849,
+      "step": 20960
+    },
+    {
+      "epoch": 0.67104,
+      "grad_norm": 42.863189697265625,
+      "learning_rate": 1.4766222222222224e-05,
+      "loss": 1.6944,
+      "step": 20970
+    },
+    {
+      "epoch": 0.67136,
+      "grad_norm": 43.975914001464844,
+      "learning_rate": 1.4762666666666667e-05,
+      "loss": 1.6611,
+      "step": 20980
+    },
+    {
+      "epoch": 0.67168,
+      "grad_norm": 44.24748992919922,
+      "learning_rate": 1.4759111111111113e-05,
+      "loss": 1.6517,
+      "step": 20990
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 43.405853271484375,
+      "learning_rate": 1.4755555555555556e-05,
+      "loss": 1.6585,
+      "step": 21000
+    },
+    {
+      "epoch": 0.67232,
+      "grad_norm": 43.27943801879883,
+      "learning_rate": 1.4752000000000002e-05,
+      "loss": 1.6554,
+      "step": 21010
+    },
+    {
+      "epoch": 0.67264,
+      "grad_norm": 46.6523323059082,
+      "learning_rate": 1.4748444444444445e-05,
+      "loss": 1.6923,
+      "step": 21020
+    },
+    {
+      "epoch": 0.67296,
+      "grad_norm": 44.580291748046875,
+      "learning_rate": 1.4744888888888891e-05,
+      "loss": 1.6568,
+      "step": 21030
+    },
+    {
+      "epoch": 0.67328,
+      "grad_norm": 43.5629768371582,
+      "learning_rate": 1.4741333333333334e-05,
+      "loss": 1.6756,
+      "step": 21040
+    },
+    {
+      "epoch": 0.6736,
+      "grad_norm": 43.935951232910156,
+      "learning_rate": 1.473777777777778e-05,
+      "loss": 1.6426,
+      "step": 21050
+    },
+    {
+      "epoch": 0.67392,
+      "grad_norm": 43.343994140625,
+      "learning_rate": 1.4734222222222223e-05,
+      "loss": 1.6679,
+      "step": 21060
+    },
+    {
+      "epoch": 0.67424,
+      "grad_norm": 45.69078063964844,
+      "learning_rate": 1.473066666666667e-05,
+      "loss": 1.6671,
+      "step": 21070
+    },
+    {
+      "epoch": 0.67456,
+      "grad_norm": 43.63746643066406,
+      "learning_rate": 1.4727111111111112e-05,
+      "loss": 1.721,
+      "step": 21080
+    },
+    {
+      "epoch": 0.67488,
+      "grad_norm": 46.481971740722656,
+      "learning_rate": 1.4723555555555557e-05,
+      "loss": 1.6843,
+      "step": 21090
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 43.81696319580078,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 1.6857,
+      "step": 21100
+    },
+    {
+      "epoch": 0.67552,
+      "grad_norm": 44.01736831665039,
+      "learning_rate": 1.4716444444444446e-05,
+      "loss": 1.6955,
+      "step": 21110
+    },
+    {
+      "epoch": 0.67584,
+      "grad_norm": 42.9880256652832,
+      "learning_rate": 1.471288888888889e-05,
+      "loss": 1.7147,
+      "step": 21120
+    },
+    {
+      "epoch": 0.67616,
+      "grad_norm": 42.299346923828125,
+      "learning_rate": 1.4709333333333335e-05,
+      "loss": 1.6807,
+      "step": 21130
+    },
+    {
+      "epoch": 0.67648,
+      "grad_norm": 43.38955307006836,
+      "learning_rate": 1.470577777777778e-05,
+      "loss": 1.6958,
+      "step": 21140
+    },
+    {
+      "epoch": 0.6768,
+      "grad_norm": 42.738651275634766,
+      "learning_rate": 1.4702222222222224e-05,
+      "loss": 1.6508,
+      "step": 21150
+    },
+    {
+      "epoch": 0.67712,
+      "grad_norm": 44.31309509277344,
+      "learning_rate": 1.4698666666666667e-05,
+      "loss": 1.6497,
+      "step": 21160
+    },
+    {
+      "epoch": 0.67744,
+      "grad_norm": 42.17631912231445,
+      "learning_rate": 1.4695111111111113e-05,
+      "loss": 1.7003,
+      "step": 21170
+    },
+    {
+      "epoch": 0.67776,
+      "grad_norm": 43.73914337158203,
+      "learning_rate": 1.4691555555555556e-05,
+      "loss": 1.6769,
+      "step": 21180
+    },
+    {
+      "epoch": 0.67808,
+      "grad_norm": 43.96467590332031,
+      "learning_rate": 1.4688000000000002e-05,
+      "loss": 1.6889,
+      "step": 21190
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 46.01679611206055,
+      "learning_rate": 1.4684444444444445e-05,
+      "loss": 1.6636,
+      "step": 21200
+    },
+    {
+      "epoch": 0.67872,
+      "grad_norm": 41.097747802734375,
+      "learning_rate": 1.4680888888888891e-05,
+      "loss": 1.681,
+      "step": 21210
+    },
+    {
+      "epoch": 0.67904,
+      "grad_norm": 44.2218017578125,
+      "learning_rate": 1.4677333333333334e-05,
+      "loss": 1.6958,
+      "step": 21220
+    },
+    {
+      "epoch": 0.67936,
+      "grad_norm": 52.44313049316406,
+      "learning_rate": 1.4673777777777777e-05,
+      "loss": 1.6334,
+      "step": 21230
+    },
+    {
+      "epoch": 0.67968,
+      "grad_norm": 42.90603256225586,
+      "learning_rate": 1.4670222222222223e-05,
+      "loss": 1.6467,
+      "step": 21240
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 43.24466323852539,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.6957,
+      "step": 21250
+    },
+    {
+      "epoch": 0.68032,
+      "grad_norm": 44.07612609863281,
+      "learning_rate": 1.4663111111111113e-05,
+      "loss": 1.6713,
+      "step": 21260
+    },
+    {
+      "epoch": 0.68064,
+      "grad_norm": 44.573612213134766,
+      "learning_rate": 1.4659555555555555e-05,
+      "loss": 1.6881,
+      "step": 21270
+    },
+    {
+      "epoch": 0.68096,
+      "grad_norm": 44.9393424987793,
+      "learning_rate": 1.4656000000000002e-05,
+      "loss": 1.7043,
+      "step": 21280
+    },
+    {
+      "epoch": 0.68128,
+      "grad_norm": 41.96256637573242,
+      "learning_rate": 1.4652444444444445e-05,
+      "loss": 1.7014,
+      "step": 21290
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 42.43061065673828,
+      "learning_rate": 1.464888888888889e-05,
+      "loss": 1.66,
+      "step": 21300
+    },
+    {
+      "epoch": 0.68192,
+      "grad_norm": 43.678466796875,
+      "learning_rate": 1.4645333333333334e-05,
+      "loss": 1.6978,
+      "step": 21310
+    },
+    {
+      "epoch": 0.68224,
+      "grad_norm": 44.234283447265625,
+      "learning_rate": 1.464177777777778e-05,
+      "loss": 1.6687,
+      "step": 21320
+    },
+    {
+      "epoch": 0.68256,
+      "grad_norm": 45.48586654663086,
+      "learning_rate": 1.4638222222222223e-05,
+      "loss": 1.637,
+      "step": 21330
+    },
+    {
+      "epoch": 0.68288,
+      "grad_norm": 45.24445724487305,
+      "learning_rate": 1.4634666666666669e-05,
+      "loss": 1.685,
+      "step": 21340
+    },
+    {
+      "epoch": 0.6832,
+      "grad_norm": 44.61650085449219,
+      "learning_rate": 1.4631111111111112e-05,
+      "loss": 1.6878,
+      "step": 21350
+    },
+    {
+      "epoch": 0.68352,
+      "grad_norm": 45.00833511352539,
+      "learning_rate": 1.4627555555555556e-05,
+      "loss": 1.6808,
+      "step": 21360
+    },
+    {
+      "epoch": 0.68384,
+      "grad_norm": 42.275150299072266,
+      "learning_rate": 1.4624000000000001e-05,
+      "loss": 1.6747,
+      "step": 21370
+    },
+    {
+      "epoch": 0.68416,
+      "grad_norm": 42.742698669433594,
+      "learning_rate": 1.4620444444444445e-05,
+      "loss": 1.6608,
+      "step": 21380
+    },
+    {
+      "epoch": 0.68448,
+      "grad_norm": 44.61221694946289,
+      "learning_rate": 1.461688888888889e-05,
+      "loss": 1.6709,
+      "step": 21390
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 45.47736740112305,
+      "learning_rate": 1.4613333333333335e-05,
+      "loss": 1.675,
+      "step": 21400
+    },
+    {
+      "epoch": 0.68512,
+      "grad_norm": 47.367313385009766,
+      "learning_rate": 1.4609777777777777e-05,
+      "loss": 1.6819,
+      "step": 21410
+    },
+    {
+      "epoch": 0.68544,
+      "grad_norm": 42.664772033691406,
+      "learning_rate": 1.4606222222222224e-05,
+      "loss": 1.6893,
+      "step": 21420
+    },
+    {
+      "epoch": 0.68576,
+      "grad_norm": 45.12783432006836,
+      "learning_rate": 1.4602666666666667e-05,
+      "loss": 1.7108,
+      "step": 21430
+    },
+    {
+      "epoch": 0.68608,
+      "grad_norm": 42.89619827270508,
+      "learning_rate": 1.4599111111111113e-05,
+      "loss": 1.6802,
+      "step": 21440
+    },
+    {
+      "epoch": 0.6864,
+      "grad_norm": 45.1823844909668,
+      "learning_rate": 1.4595555555555556e-05,
+      "loss": 1.6557,
+      "step": 21450
+    },
+    {
+      "epoch": 0.68672,
+      "grad_norm": 46.50447463989258,
+      "learning_rate": 1.4592000000000002e-05,
+      "loss": 1.6734,
+      "step": 21460
+    },
+    {
+      "epoch": 0.68704,
+      "grad_norm": 43.853580474853516,
+      "learning_rate": 1.4588444444444445e-05,
+      "loss": 1.6702,
+      "step": 21470
+    },
+    {
+      "epoch": 0.68736,
+      "grad_norm": 43.31742477416992,
+      "learning_rate": 1.4584888888888891e-05,
+      "loss": 1.6708,
+      "step": 21480
+    },
+    {
+      "epoch": 0.68768,
+      "grad_norm": 47.49016571044922,
+      "learning_rate": 1.4581333333333334e-05,
+      "loss": 1.6976,
+      "step": 21490
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 42.12497329711914,
+      "learning_rate": 1.457777777777778e-05,
+      "loss": 1.6694,
+      "step": 21500
+    },
+    {
+      "epoch": 0.68832,
+      "grad_norm": 45.087188720703125,
+      "learning_rate": 1.4574222222222223e-05,
+      "loss": 1.7175,
+      "step": 21510
+    },
+    {
+      "epoch": 0.68864,
+      "grad_norm": 45.94072341918945,
+      "learning_rate": 1.457066666666667e-05,
+      "loss": 1.6674,
+      "step": 21520
+    },
+    {
+      "epoch": 0.68896,
+      "grad_norm": 45.907684326171875,
+      "learning_rate": 1.4567111111111112e-05,
+      "loss": 1.6998,
+      "step": 21530
+    },
+    {
+      "epoch": 0.68928,
+      "grad_norm": 43.56646728515625,
+      "learning_rate": 1.4563555555555557e-05,
+      "loss": 1.6627,
+      "step": 21540
+    },
+    {
+      "epoch": 0.6896,
+      "grad_norm": 44.42494201660156,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 1.6713,
+      "step": 21550
+    },
+    {
+      "epoch": 0.68992,
+      "grad_norm": 43.25820541381836,
+      "learning_rate": 1.4556444444444446e-05,
+      "loss": 1.6823,
+      "step": 21560
+    },
+    {
+      "epoch": 0.69024,
+      "grad_norm": 44.242862701416016,
+      "learning_rate": 1.455288888888889e-05,
+      "loss": 1.6743,
+      "step": 21570
+    },
+    {
+      "epoch": 0.69056,
+      "grad_norm": 43.78083801269531,
+      "learning_rate": 1.4549333333333335e-05,
+      "loss": 1.6951,
+      "step": 21580
+    },
+    {
+      "epoch": 0.69088,
+      "grad_norm": 44.01467514038086,
+      "learning_rate": 1.454577777777778e-05,
+      "loss": 1.6622,
+      "step": 21590
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 44.382904052734375,
+      "learning_rate": 1.4542222222222224e-05,
+      "loss": 1.7014,
+      "step": 21600
+    },
+    {
+      "epoch": 0.69152,
+      "grad_norm": 43.727657318115234,
+      "learning_rate": 1.4538666666666667e-05,
+      "loss": 1.6609,
+      "step": 21610
+    },
+    {
+      "epoch": 0.69184,
+      "grad_norm": 44.795719146728516,
+      "learning_rate": 1.4535111111111113e-05,
+      "loss": 1.6852,
+      "step": 21620
+    },
+    {
+      "epoch": 0.69216,
+      "grad_norm": 45.50511932373047,
+      "learning_rate": 1.4531555555555556e-05,
+      "loss": 1.7151,
+      "step": 21630
+    },
+    {
+      "epoch": 0.69248,
+      "grad_norm": 44.66515350341797,
+      "learning_rate": 1.4528000000000002e-05,
+      "loss": 1.6791,
+      "step": 21640
+    },
+    {
+      "epoch": 0.6928,
+      "grad_norm": 43.8026008605957,
+      "learning_rate": 1.4524444444444445e-05,
+      "loss": 1.6504,
+      "step": 21650
+    },
+    {
+      "epoch": 0.69312,
+      "grad_norm": 44.98323440551758,
+      "learning_rate": 1.4520888888888891e-05,
+      "loss": 1.6957,
+      "step": 21660
+    },
+    {
+      "epoch": 0.69344,
+      "grad_norm": 44.08596420288086,
+      "learning_rate": 1.4517333333333334e-05,
+      "loss": 1.6781,
+      "step": 21670
+    },
+    {
+      "epoch": 0.69376,
+      "grad_norm": 43.75856018066406,
+      "learning_rate": 1.451377777777778e-05,
+      "loss": 1.6984,
+      "step": 21680
+    },
+    {
+      "epoch": 0.69408,
+      "grad_norm": 42.918182373046875,
+      "learning_rate": 1.4510222222222223e-05,
+      "loss": 1.6458,
+      "step": 21690
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 43.657752990722656,
+      "learning_rate": 1.450666666666667e-05,
+      "loss": 1.6763,
+      "step": 21700
+    },
+    {
+      "epoch": 0.69472,
+      "grad_norm": 44.31597900390625,
+      "learning_rate": 1.4503111111111112e-05,
+      "loss": 1.6781,
+      "step": 21710
+    },
+    {
+      "epoch": 0.69504,
+      "grad_norm": 45.91122817993164,
+      "learning_rate": 1.4499555555555559e-05,
+      "loss": 1.6734,
+      "step": 21720
+    },
+    {
+      "epoch": 0.69536,
+      "grad_norm": 43.836753845214844,
+      "learning_rate": 1.4496000000000001e-05,
+      "loss": 1.6747,
+      "step": 21730
+    },
+    {
+      "epoch": 0.69568,
+      "grad_norm": 44.824256896972656,
+      "learning_rate": 1.4492444444444444e-05,
+      "loss": 1.6594,
+      "step": 21740
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 45.5569953918457,
+      "learning_rate": 1.448888888888889e-05,
+      "loss": 1.6569,
+      "step": 21750
+    },
+    {
+      "epoch": 0.69632,
+      "grad_norm": 42.6880989074707,
+      "learning_rate": 1.4485333333333333e-05,
+      "loss": 1.7061,
+      "step": 21760
+    },
+    {
+      "epoch": 0.69664,
+      "grad_norm": 45.63669967651367,
+      "learning_rate": 1.448177777777778e-05,
+      "loss": 1.6642,
+      "step": 21770
+    },
+    {
+      "epoch": 0.69696,
+      "grad_norm": 46.666629791259766,
+      "learning_rate": 1.4478222222222223e-05,
+      "loss": 1.661,
+      "step": 21780
+    },
+    {
+      "epoch": 0.69728,
+      "grad_norm": 43.4208984375,
+      "learning_rate": 1.4474666666666669e-05,
+      "loss": 1.6883,
+      "step": 21790
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 43.890933990478516,
+      "learning_rate": 1.4471111111111112e-05,
+      "loss": 1.6596,
+      "step": 21800
+    },
+    {
+      "epoch": 0.69792,
+      "grad_norm": 42.00779724121094,
+      "learning_rate": 1.4467555555555556e-05,
+      "loss": 1.6665,
+      "step": 21810
+    },
+    {
+      "epoch": 0.69824,
+      "grad_norm": 41.96549987792969,
+      "learning_rate": 1.4464e-05,
+      "loss": 1.6505,
+      "step": 21820
+    },
+    {
+      "epoch": 0.69856,
+      "grad_norm": 43.60977554321289,
+      "learning_rate": 1.4460444444444445e-05,
+      "loss": 1.6747,
+      "step": 21830
+    },
+    {
+      "epoch": 0.69888,
+      "grad_norm": 43.725929260253906,
+      "learning_rate": 1.445688888888889e-05,
+      "loss": 1.6546,
+      "step": 21840
+    },
+    {
+      "epoch": 0.6992,
+      "grad_norm": 45.897518157958984,
+      "learning_rate": 1.4453333333333334e-05,
+      "loss": 1.669,
+      "step": 21850
+    },
+    {
+      "epoch": 0.69952,
+      "grad_norm": 42.812721252441406,
+      "learning_rate": 1.4449777777777777e-05,
+      "loss": 1.6681,
+      "step": 21860
+    },
+    {
+      "epoch": 0.69984,
+      "grad_norm": 42.7755126953125,
+      "learning_rate": 1.4446222222222224e-05,
+      "loss": 1.6507,
+      "step": 21870
+    },
+    {
+      "epoch": 0.70016,
+      "grad_norm": 43.823123931884766,
+      "learning_rate": 1.4442666666666666e-05,
+      "loss": 1.6878,
+      "step": 21880
+    },
+    {
+      "epoch": 0.70048,
+      "grad_norm": 46.74770736694336,
+      "learning_rate": 1.4439111111111113e-05,
+      "loss": 1.6812,
+      "step": 21890
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 45.07465362548828,
+      "learning_rate": 1.4435555555555556e-05,
+      "loss": 1.6529,
+      "step": 21900
+    },
+    {
+      "epoch": 0.70112,
+      "grad_norm": 45.174041748046875,
+      "learning_rate": 1.4432000000000002e-05,
+      "loss": 1.6818,
+      "step": 21910
+    },
+    {
+      "epoch": 0.70144,
+      "grad_norm": 43.29008483886719,
+      "learning_rate": 1.4428444444444445e-05,
+      "loss": 1.6575,
+      "step": 21920
+    },
+    {
+      "epoch": 0.70176,
+      "grad_norm": 44.14530563354492,
+      "learning_rate": 1.4424888888888891e-05,
+      "loss": 1.6732,
+      "step": 21930
+    },
+    {
+      "epoch": 0.70208,
+      "grad_norm": 44.139373779296875,
+      "learning_rate": 1.4421333333333334e-05,
+      "loss": 1.6953,
+      "step": 21940
+    },
+    {
+      "epoch": 0.7024,
+      "grad_norm": 42.85390090942383,
+      "learning_rate": 1.441777777777778e-05,
+      "loss": 1.6489,
+      "step": 21950
+    },
+    {
+      "epoch": 0.70272,
+      "grad_norm": 44.265907287597656,
+      "learning_rate": 1.4414222222222223e-05,
+      "loss": 1.683,
+      "step": 21960
+    },
+    {
+      "epoch": 0.70304,
+      "grad_norm": 43.23847961425781,
+      "learning_rate": 1.4410666666666669e-05,
+      "loss": 1.6433,
+      "step": 21970
+    },
+    {
+      "epoch": 0.70336,
+      "grad_norm": 44.420169830322266,
+      "learning_rate": 1.4407111111111112e-05,
+      "loss": 1.6664,
+      "step": 21980
+    },
+    {
+      "epoch": 0.70368,
+      "grad_norm": 43.698692321777344,
+      "learning_rate": 1.4403555555555556e-05,
+      "loss": 1.6422,
+      "step": 21990
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 43.02571105957031,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.6612,
+      "step": 22000
+    },
+    {
+      "epoch": 0.70432,
+      "grad_norm": 43.11848068237305,
+      "learning_rate": 1.4396444444444446e-05,
+      "loss": 1.6877,
+      "step": 22010
+    },
+    {
+      "epoch": 0.70464,
+      "grad_norm": 44.298397064208984,
+      "learning_rate": 1.439288888888889e-05,
+      "loss": 1.6554,
+      "step": 22020
+    },
+    {
+      "epoch": 0.70496,
+      "grad_norm": 43.6672248840332,
+      "learning_rate": 1.4389333333333335e-05,
+      "loss": 1.6756,
+      "step": 22030
+    },
+    {
+      "epoch": 0.70528,
+      "grad_norm": 45.8585205078125,
+      "learning_rate": 1.438577777777778e-05,
+      "loss": 1.6875,
+      "step": 22040
+    },
+    {
+      "epoch": 0.7056,
+      "grad_norm": 44.58899688720703,
+      "learning_rate": 1.4382222222222224e-05,
+      "loss": 1.65,
+      "step": 22050
+    },
+    {
+      "epoch": 0.70592,
+      "grad_norm": 43.89964294433594,
+      "learning_rate": 1.4378666666666667e-05,
+      "loss": 1.6596,
+      "step": 22060
+    },
+    {
+      "epoch": 0.70624,
+      "grad_norm": 43.34562683105469,
+      "learning_rate": 1.4375111111111113e-05,
+      "loss": 1.6597,
+      "step": 22070
+    },
+    {
+      "epoch": 0.70656,
+      "grad_norm": 42.41712188720703,
+      "learning_rate": 1.4371555555555556e-05,
+      "loss": 1.6537,
+      "step": 22080
+    },
+    {
+      "epoch": 0.70688,
+      "grad_norm": 44.702484130859375,
+      "learning_rate": 1.4368000000000002e-05,
+      "loss": 1.6462,
+      "step": 22090
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 44.63224792480469,
+      "learning_rate": 1.4364444444444445e-05,
+      "loss": 1.669,
+      "step": 22100
+    },
+    {
+      "epoch": 0.70752,
+      "grad_norm": 44.96320724487305,
+      "learning_rate": 1.4360888888888891e-05,
+      "loss": 1.6802,
+      "step": 22110
+    },
+    {
+      "epoch": 0.70784,
+      "grad_norm": 45.22520065307617,
+      "learning_rate": 1.4357333333333334e-05,
+      "loss": 1.668,
+      "step": 22120
+    },
+    {
+      "epoch": 0.70816,
+      "grad_norm": 43.496524810791016,
+      "learning_rate": 1.435377777777778e-05,
+      "loss": 1.6822,
+      "step": 22130
+    },
+    {
+      "epoch": 0.70848,
+      "grad_norm": 45.04349136352539,
+      "learning_rate": 1.4350222222222223e-05,
+      "loss": 1.6227,
+      "step": 22140
+    },
+    {
+      "epoch": 0.7088,
+      "grad_norm": 43.959861755371094,
+      "learning_rate": 1.434666666666667e-05,
+      "loss": 1.6815,
+      "step": 22150
+    },
+    {
+      "epoch": 0.70912,
+      "grad_norm": 44.3282470703125,
+      "learning_rate": 1.4343111111111112e-05,
+      "loss": 1.6754,
+      "step": 22160
+    },
+    {
+      "epoch": 0.70944,
+      "grad_norm": 42.763973236083984,
+      "learning_rate": 1.4339555555555558e-05,
+      "loss": 1.6731,
+      "step": 22170
+    },
+    {
+      "epoch": 0.70976,
+      "grad_norm": 44.765254974365234,
+      "learning_rate": 1.4336000000000001e-05,
+      "loss": 1.6602,
+      "step": 22180
+    },
+    {
+      "epoch": 0.71008,
+      "grad_norm": 43.3527946472168,
+      "learning_rate": 1.4332444444444446e-05,
+      "loss": 1.665,
+      "step": 22190
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 43.666568756103516,
+      "learning_rate": 1.432888888888889e-05,
+      "loss": 1.6816,
+      "step": 22200
+    },
+    {
+      "epoch": 0.71072,
+      "grad_norm": 46.126827239990234,
+      "learning_rate": 1.4325333333333335e-05,
+      "loss": 1.6815,
+      "step": 22210
+    },
+    {
+      "epoch": 0.71104,
+      "grad_norm": 44.10695266723633,
+      "learning_rate": 1.432177777777778e-05,
+      "loss": 1.6675,
+      "step": 22220
+    },
+    {
+      "epoch": 0.71136,
+      "grad_norm": 45.392333984375,
+      "learning_rate": 1.4318222222222224e-05,
+      "loss": 1.6706,
+      "step": 22230
+    },
+    {
+      "epoch": 0.71168,
+      "grad_norm": 45.036781311035156,
+      "learning_rate": 1.4314666666666669e-05,
+      "loss": 1.6644,
+      "step": 22240
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 44.691246032714844,
+      "learning_rate": 1.4311111111111111e-05,
+      "loss": 1.6886,
+      "step": 22250
+    },
+    {
+      "epoch": 0.71232,
+      "grad_norm": 44.20449447631836,
+      "learning_rate": 1.4307555555555556e-05,
+      "loss": 1.6956,
+      "step": 22260
+    },
+    {
+      "epoch": 0.71264,
+      "grad_norm": 43.10037612915039,
+      "learning_rate": 1.4304e-05,
+      "loss": 1.6857,
+      "step": 22270
+    },
+    {
+      "epoch": 0.71296,
+      "grad_norm": 43.95088577270508,
+      "learning_rate": 1.4300444444444445e-05,
+      "loss": 1.6607,
+      "step": 22280
+    },
+    {
+      "epoch": 0.71328,
+      "grad_norm": 44.31607437133789,
+      "learning_rate": 1.429688888888889e-05,
+      "loss": 1.6689,
+      "step": 22290
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 41.93162536621094,
+      "learning_rate": 1.4293333333333334e-05,
+      "loss": 1.6705,
+      "step": 22300
+    },
+    {
+      "epoch": 0.71392,
+      "grad_norm": 45.12978744506836,
+      "learning_rate": 1.4289777777777777e-05,
+      "loss": 1.6745,
+      "step": 22310
+    },
+    {
+      "epoch": 0.71424,
+      "grad_norm": 46.437583923339844,
+      "learning_rate": 1.4286222222222223e-05,
+      "loss": 1.6456,
+      "step": 22320
+    },
+    {
+      "epoch": 0.71456,
+      "grad_norm": 42.88267517089844,
+      "learning_rate": 1.4282666666666666e-05,
+      "loss": 1.6257,
+      "step": 22330
+    },
+    {
+      "epoch": 0.71488,
+      "grad_norm": 46.049652099609375,
+      "learning_rate": 1.4279111111111112e-05,
+      "loss": 1.6578,
+      "step": 22340
+    },
+    {
+      "epoch": 0.7152,
+      "grad_norm": 45.01377868652344,
+      "learning_rate": 1.4275555555555555e-05,
+      "loss": 1.678,
+      "step": 22350
+    },
+    {
+      "epoch": 0.71552,
+      "grad_norm": 44.42237091064453,
+      "learning_rate": 1.4272000000000002e-05,
+      "loss": 1.6648,
+      "step": 22360
+    },
+    {
+      "epoch": 0.71584,
+      "grad_norm": 43.99474334716797,
+      "learning_rate": 1.4268444444444444e-05,
+      "loss": 1.6889,
+      "step": 22370
+    },
+    {
+      "epoch": 0.71616,
+      "grad_norm": 41.76171875,
+      "learning_rate": 1.426488888888889e-05,
+      "loss": 1.6735,
+      "step": 22380
+    },
+    {
+      "epoch": 0.71648,
+      "grad_norm": 43.52986526489258,
+      "learning_rate": 1.4261333333333334e-05,
+      "loss": 1.6696,
+      "step": 22390
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 43.7346076965332,
+      "learning_rate": 1.425777777777778e-05,
+      "loss": 1.6738,
+      "step": 22400
+    },
+    {
+      "epoch": 0.71712,
+      "grad_norm": 46.53422546386719,
+      "learning_rate": 1.4254222222222223e-05,
+      "loss": 1.6429,
+      "step": 22410
+    },
+    {
+      "epoch": 0.71744,
+      "grad_norm": 43.2796516418457,
+      "learning_rate": 1.4250666666666669e-05,
+      "loss": 1.661,
+      "step": 22420
+    },
+    {
+      "epoch": 0.71776,
+      "grad_norm": 45.7249641418457,
+      "learning_rate": 1.4247111111111112e-05,
+      "loss": 1.6454,
+      "step": 22430
+    },
+    {
+      "epoch": 0.71808,
+      "grad_norm": 44.16270446777344,
+      "learning_rate": 1.4243555555555556e-05,
+      "loss": 1.654,
+      "step": 22440
+    },
+    {
+      "epoch": 0.7184,
+      "grad_norm": 46.57696533203125,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.6901,
+      "step": 22450
+    },
+    {
+      "epoch": 0.71872,
+      "grad_norm": 43.533870697021484,
+      "learning_rate": 1.4236444444444445e-05,
+      "loss": 1.6344,
+      "step": 22460
+    },
+    {
+      "epoch": 0.71904,
+      "grad_norm": 43.986446380615234,
+      "learning_rate": 1.423288888888889e-05,
+      "loss": 1.6738,
+      "step": 22470
+    },
+    {
+      "epoch": 0.71936,
+      "grad_norm": 43.613040924072266,
+      "learning_rate": 1.4229333333333335e-05,
+      "loss": 1.6628,
+      "step": 22480
+    },
+    {
+      "epoch": 0.71968,
+      "grad_norm": 43.533058166503906,
+      "learning_rate": 1.4225777777777779e-05,
+      "loss": 1.6739,
+      "step": 22490
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 46.17250442504883,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 1.6625,
+      "step": 22500
+    },
+    {
+      "epoch": 0.72032,
+      "grad_norm": 44.67798614501953,
+      "learning_rate": 1.4218666666666667e-05,
+      "loss": 1.6553,
+      "step": 22510
+    },
+    {
+      "epoch": 0.72064,
+      "grad_norm": 43.80499267578125,
+      "learning_rate": 1.4215111111111113e-05,
+      "loss": 1.6733,
+      "step": 22520
+    },
+    {
+      "epoch": 0.72096,
+      "grad_norm": 45.787498474121094,
+      "learning_rate": 1.4211555555555556e-05,
+      "loss": 1.6607,
+      "step": 22530
+    },
+    {
+      "epoch": 0.72128,
+      "grad_norm": 43.61078643798828,
+      "learning_rate": 1.4208000000000002e-05,
+      "loss": 1.6785,
+      "step": 22540
+    },
+    {
+      "epoch": 0.7216,
+      "grad_norm": 44.815433502197266,
+      "learning_rate": 1.4204444444444445e-05,
+      "loss": 1.6646,
+      "step": 22550
+    },
+    {
+      "epoch": 0.72192,
+      "grad_norm": 44.859615325927734,
+      "learning_rate": 1.4200888888888891e-05,
+      "loss": 1.6768,
+      "step": 22560
+    },
+    {
+      "epoch": 0.72224,
+      "grad_norm": 47.96544647216797,
+      "learning_rate": 1.4197333333333334e-05,
+      "loss": 1.6902,
+      "step": 22570
+    },
+    {
+      "epoch": 0.72256,
+      "grad_norm": 44.82842254638672,
+      "learning_rate": 1.419377777777778e-05,
+      "loss": 1.6897,
+      "step": 22580
+    },
+    {
+      "epoch": 0.72288,
+      "grad_norm": 45.776710510253906,
+      "learning_rate": 1.4190222222222223e-05,
+      "loss": 1.6566,
+      "step": 22590
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 42.437870025634766,
+      "learning_rate": 1.418666666666667e-05,
+      "loss": 1.6823,
+      "step": 22600
+    },
+    {
+      "epoch": 0.72352,
+      "grad_norm": 43.7807731628418,
+      "learning_rate": 1.4183111111111112e-05,
+      "loss": 1.6802,
+      "step": 22610
+    },
+    {
+      "epoch": 0.72384,
+      "grad_norm": 43.51747512817383,
+      "learning_rate": 1.4179555555555558e-05,
+      "loss": 1.6449,
+      "step": 22620
+    },
+    {
+      "epoch": 0.72416,
+      "grad_norm": 42.205467224121094,
+      "learning_rate": 1.4176000000000001e-05,
+      "loss": 1.6829,
+      "step": 22630
+    },
+    {
+      "epoch": 0.72448,
+      "grad_norm": 43.446014404296875,
+      "learning_rate": 1.4172444444444446e-05,
+      "loss": 1.6902,
+      "step": 22640
+    },
+    {
+      "epoch": 0.7248,
+      "grad_norm": 43.7180061340332,
+      "learning_rate": 1.416888888888889e-05,
+      "loss": 1.6697,
+      "step": 22650
+    },
+    {
+      "epoch": 0.72512,
+      "grad_norm": 44.47442626953125,
+      "learning_rate": 1.4165333333333335e-05,
+      "loss": 1.6647,
+      "step": 22660
+    },
+    {
+      "epoch": 0.72544,
+      "grad_norm": 46.07502746582031,
+      "learning_rate": 1.416177777777778e-05,
+      "loss": 1.6755,
+      "step": 22670
+    },
+    {
+      "epoch": 0.72576,
+      "grad_norm": 44.2828369140625,
+      "learning_rate": 1.4158222222222224e-05,
+      "loss": 1.6874,
+      "step": 22680
+    },
+    {
+      "epoch": 0.72608,
+      "grad_norm": 42.204612731933594,
+      "learning_rate": 1.4154666666666668e-05,
+      "loss": 1.6864,
+      "step": 22690
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 42.07124710083008,
+      "learning_rate": 1.4151111111111113e-05,
+      "loss": 1.6453,
+      "step": 22700
+    },
+    {
+      "epoch": 0.72672,
+      "grad_norm": 49.05067825317383,
+      "learning_rate": 1.4147555555555556e-05,
+      "loss": 1.6727,
+      "step": 22710
+    },
+    {
+      "epoch": 0.72704,
+      "grad_norm": 45.1103401184082,
+      "learning_rate": 1.4144000000000002e-05,
+      "loss": 1.6688,
+      "step": 22720
+    },
+    {
+      "epoch": 0.72736,
+      "grad_norm": 44.48417282104492,
+      "learning_rate": 1.4140444444444445e-05,
+      "loss": 1.6702,
+      "step": 22730
+    },
+    {
+      "epoch": 0.72768,
+      "grad_norm": 45.34505844116211,
+      "learning_rate": 1.4136888888888891e-05,
+      "loss": 1.6479,
+      "step": 22740
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 46.812767028808594,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 1.6598,
+      "step": 22750
+    },
+    {
+      "epoch": 0.72832,
+      "grad_norm": 46.07847595214844,
+      "learning_rate": 1.412977777777778e-05,
+      "loss": 1.6663,
+      "step": 22760
+    },
+    {
+      "epoch": 0.72864,
+      "grad_norm": 47.190879821777344,
+      "learning_rate": 1.4126222222222223e-05,
+      "loss": 1.6324,
+      "step": 22770
+    },
+    {
+      "epoch": 0.72896,
+      "grad_norm": 43.76619338989258,
+      "learning_rate": 1.4122666666666666e-05,
+      "loss": 1.6851,
+      "step": 22780
+    },
+    {
+      "epoch": 0.72928,
+      "grad_norm": 45.95934295654297,
+      "learning_rate": 1.4119111111111112e-05,
+      "loss": 1.6608,
+      "step": 22790
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 43.19196701049805,
+      "learning_rate": 1.4115555555555555e-05,
+      "loss": 1.6564,
+      "step": 22800
+    },
+    {
+      "epoch": 0.72992,
+      "grad_norm": 43.22726821899414,
+      "learning_rate": 1.4112000000000001e-05,
+      "loss": 1.6412,
+      "step": 22810
+    },
+    {
+      "epoch": 0.73024,
+      "grad_norm": 45.99415588378906,
+      "learning_rate": 1.4108444444444444e-05,
+      "loss": 1.6814,
+      "step": 22820
+    },
+    {
+      "epoch": 0.73056,
+      "grad_norm": 43.976158142089844,
+      "learning_rate": 1.410488888888889e-05,
+      "loss": 1.6803,
+      "step": 22830
+    },
+    {
+      "epoch": 0.73088,
+      "grad_norm": 44.73454284667969,
+      "learning_rate": 1.4101333333333333e-05,
+      "loss": 1.669,
+      "step": 22840
+    },
+    {
+      "epoch": 0.7312,
+      "grad_norm": 42.69565963745117,
+      "learning_rate": 1.409777777777778e-05,
+      "loss": 1.667,
+      "step": 22850
+    },
+    {
+      "epoch": 0.73152,
+      "grad_norm": 43.98191452026367,
+      "learning_rate": 1.4094222222222223e-05,
+      "loss": 1.6988,
+      "step": 22860
+    },
+    {
+      "epoch": 0.73184,
+      "grad_norm": 44.971351623535156,
+      "learning_rate": 1.4090666666666669e-05,
+      "loss": 1.6534,
+      "step": 22870
+    },
+    {
+      "epoch": 0.73216,
+      "grad_norm": 45.63749694824219,
+      "learning_rate": 1.4087111111111112e-05,
+      "loss": 1.651,
+      "step": 22880
+    },
+    {
+      "epoch": 0.73248,
+      "grad_norm": 44.00672912597656,
+      "learning_rate": 1.4083555555555556e-05,
+      "loss": 1.6516,
+      "step": 22890
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 44.87765884399414,
+      "learning_rate": 1.408e-05,
+      "loss": 1.6654,
+      "step": 22900
+    },
+    {
+      "epoch": 0.73312,
+      "grad_norm": 43.8448371887207,
+      "learning_rate": 1.4076444444444445e-05,
+      "loss": 1.6614,
+      "step": 22910
+    },
+    {
+      "epoch": 0.73344,
+      "grad_norm": 45.86703872680664,
+      "learning_rate": 1.407288888888889e-05,
+      "loss": 1.6794,
+      "step": 22920
+    },
+    {
+      "epoch": 0.73376,
+      "grad_norm": 45.04334259033203,
+      "learning_rate": 1.4069333333333334e-05,
+      "loss": 1.6469,
+      "step": 22930
+    },
+    {
+      "epoch": 0.73408,
+      "grad_norm": 43.51724624633789,
+      "learning_rate": 1.4065777777777779e-05,
+      "loss": 1.6847,
+      "step": 22940
+    },
+    {
+      "epoch": 0.7344,
+      "grad_norm": 45.09579086303711,
+      "learning_rate": 1.4062222222222223e-05,
+      "loss": 1.6817,
+      "step": 22950
+    },
+    {
+      "epoch": 0.73472,
+      "grad_norm": 45.397491455078125,
+      "learning_rate": 1.4058666666666666e-05,
+      "loss": 1.684,
+      "step": 22960
+    },
+    {
+      "epoch": 0.73504,
+      "grad_norm": 43.79633712768555,
+      "learning_rate": 1.4055111111111113e-05,
+      "loss": 1.6278,
+      "step": 22970
+    },
+    {
+      "epoch": 0.73536,
+      "grad_norm": 43.882774353027344,
+      "learning_rate": 1.4051555555555555e-05,
+      "loss": 1.6492,
+      "step": 22980
+    },
+    {
+      "epoch": 0.73568,
+      "grad_norm": 44.50541687011719,
+      "learning_rate": 1.4048000000000002e-05,
+      "loss": 1.659,
+      "step": 22990
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 45.097389221191406,
+      "learning_rate": 1.4044444444444445e-05,
+      "loss": 1.6857,
+      "step": 23000
+    },
+    {
+      "epoch": 0.73632,
+      "grad_norm": 45.81336212158203,
+      "learning_rate": 1.404088888888889e-05,
+      "loss": 1.677,
+      "step": 23010
+    },
+    {
+      "epoch": 0.73664,
+      "grad_norm": 44.712608337402344,
+      "learning_rate": 1.4037333333333334e-05,
+      "loss": 1.6642,
+      "step": 23020
+    },
+    {
+      "epoch": 0.73696,
+      "grad_norm": 44.72639465332031,
+      "learning_rate": 1.403377777777778e-05,
+      "loss": 1.6722,
+      "step": 23030
+    },
+    {
+      "epoch": 0.73728,
+      "grad_norm": 43.99245071411133,
+      "learning_rate": 1.4030222222222223e-05,
+      "loss": 1.6658,
+      "step": 23040
+    },
+    {
+      "epoch": 0.7376,
+      "grad_norm": 43.95591354370117,
+      "learning_rate": 1.4026666666666669e-05,
+      "loss": 1.6606,
+      "step": 23050
+    },
+    {
+      "epoch": 0.73792,
+      "grad_norm": 46.36194610595703,
+      "learning_rate": 1.4023111111111112e-05,
+      "loss": 1.6521,
+      "step": 23060
+    },
+    {
+      "epoch": 0.73824,
+      "grad_norm": 44.344261169433594,
+      "learning_rate": 1.4019555555555558e-05,
+      "loss": 1.6683,
+      "step": 23070
+    },
+    {
+      "epoch": 0.73856,
+      "grad_norm": 44.6585807800293,
+      "learning_rate": 1.4016000000000001e-05,
+      "loss": 1.6445,
+      "step": 23080
+    },
+    {
+      "epoch": 0.73888,
+      "grad_norm": 45.141300201416016,
+      "learning_rate": 1.4012444444444446e-05,
+      "loss": 1.6738,
+      "step": 23090
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 44.751033782958984,
+      "learning_rate": 1.400888888888889e-05,
+      "loss": 1.6745,
+      "step": 23100
+    },
+    {
+      "epoch": 0.73952,
+      "grad_norm": 45.1855354309082,
+      "learning_rate": 1.4005333333333335e-05,
+      "loss": 1.6641,
+      "step": 23110
+    },
+    {
+      "epoch": 0.73984,
+      "grad_norm": 42.032108306884766,
+      "learning_rate": 1.400177777777778e-05,
+      "loss": 1.6633,
+      "step": 23120
+    },
+    {
+      "epoch": 0.74016,
+      "grad_norm": 43.98733139038086,
+      "learning_rate": 1.3998222222222224e-05,
+      "loss": 1.6436,
+      "step": 23130
+    },
+    {
+      "epoch": 0.74048,
+      "grad_norm": 44.74616622924805,
+      "learning_rate": 1.3994666666666668e-05,
+      "loss": 1.6801,
+      "step": 23140
+    },
+    {
+      "epoch": 0.7408,
+      "grad_norm": 42.1522102355957,
+      "learning_rate": 1.3991111111111113e-05,
+      "loss": 1.646,
+      "step": 23150
+    },
+    {
+      "epoch": 0.74112,
+      "grad_norm": 43.740299224853516,
+      "learning_rate": 1.3987555555555556e-05,
+      "loss": 1.6409,
+      "step": 23160
+    },
+    {
+      "epoch": 0.74144,
+      "grad_norm": 43.136390686035156,
+      "learning_rate": 1.3984000000000002e-05,
+      "loss": 1.668,
+      "step": 23170
+    },
+    {
+      "epoch": 0.74176,
+      "grad_norm": 43.768821716308594,
+      "learning_rate": 1.3980444444444445e-05,
+      "loss": 1.681,
+      "step": 23180
+    },
+    {
+      "epoch": 0.74208,
+      "grad_norm": 44.26373291015625,
+      "learning_rate": 1.3976888888888891e-05,
+      "loss": 1.6726,
+      "step": 23190
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 44.403743743896484,
+      "learning_rate": 1.3973333333333334e-05,
+      "loss": 1.6767,
+      "step": 23200
+    },
+    {
+      "epoch": 0.74272,
+      "grad_norm": 43.6219482421875,
+      "learning_rate": 1.396977777777778e-05,
+      "loss": 1.6611,
+      "step": 23210
+    },
+    {
+      "epoch": 0.74304,
+      "grad_norm": 43.950042724609375,
+      "learning_rate": 1.3966222222222223e-05,
+      "loss": 1.6651,
+      "step": 23220
+    },
+    {
+      "epoch": 0.74336,
+      "grad_norm": 44.61774444580078,
+      "learning_rate": 1.396266666666667e-05,
+      "loss": 1.6349,
+      "step": 23230
+    },
+    {
+      "epoch": 0.74368,
+      "grad_norm": 43.243309020996094,
+      "learning_rate": 1.3959111111111112e-05,
+      "loss": 1.6953,
+      "step": 23240
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 43.35283279418945,
+      "learning_rate": 1.3955555555555558e-05,
+      "loss": 1.6393,
+      "step": 23250
+    },
+    {
+      "epoch": 0.74432,
+      "grad_norm": 43.880435943603516,
+      "learning_rate": 1.3952000000000001e-05,
+      "loss": 1.6631,
+      "step": 23260
+    },
+    {
+      "epoch": 0.74464,
+      "grad_norm": 45.62138366699219,
+      "learning_rate": 1.3948444444444448e-05,
+      "loss": 1.6765,
+      "step": 23270
+    },
+    {
+      "epoch": 0.74496,
+      "grad_norm": 43.951446533203125,
+      "learning_rate": 1.394488888888889e-05,
+      "loss": 1.6431,
+      "step": 23280
+    },
+    {
+      "epoch": 0.74528,
+      "grad_norm": 45.40299606323242,
+      "learning_rate": 1.3941333333333333e-05,
+      "loss": 1.6572,
+      "step": 23290
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 44.030738830566406,
+      "learning_rate": 1.393777777777778e-05,
+      "loss": 1.6635,
+      "step": 23300
+    },
+    {
+      "epoch": 0.74592,
+      "grad_norm": 43.0764274597168,
+      "learning_rate": 1.3934222222222222e-05,
+      "loss": 1.7001,
+      "step": 23310
+    },
+    {
+      "epoch": 0.74624,
+      "grad_norm": 44.9112548828125,
+      "learning_rate": 1.3930666666666669e-05,
+      "loss": 1.6681,
+      "step": 23320
+    },
+    {
+      "epoch": 0.74656,
+      "grad_norm": 42.09294509887695,
+      "learning_rate": 1.3927111111111111e-05,
+      "loss": 1.6978,
+      "step": 23330
+    },
+    {
+      "epoch": 0.74688,
+      "grad_norm": 42.35454177856445,
+      "learning_rate": 1.3923555555555556e-05,
+      "loss": 1.681,
+      "step": 23340
+    },
+    {
+      "epoch": 0.7472,
+      "grad_norm": 45.9742317199707,
+      "learning_rate": 1.392e-05,
+      "loss": 1.7039,
+      "step": 23350
+    },
+    {
+      "epoch": 0.74752,
+      "grad_norm": 43.095428466796875,
+      "learning_rate": 1.3916444444444445e-05,
+      "loss": 1.6668,
+      "step": 23360
+    },
+    {
+      "epoch": 0.74784,
+      "grad_norm": 44.52104187011719,
+      "learning_rate": 1.391288888888889e-05,
+      "loss": 1.6634,
+      "step": 23370
+    },
+    {
+      "epoch": 0.74816,
+      "grad_norm": 42.80414962768555,
+      "learning_rate": 1.3909333333333334e-05,
+      "loss": 1.6686,
+      "step": 23380
+    },
+    {
+      "epoch": 0.74848,
+      "grad_norm": 43.77104568481445,
+      "learning_rate": 1.3905777777777779e-05,
+      "loss": 1.667,
+      "step": 23390
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 44.292640686035156,
+      "learning_rate": 1.3902222222222223e-05,
+      "loss": 1.6475,
+      "step": 23400
+    },
+    {
+      "epoch": 0.74912,
+      "grad_norm": 45.744869232177734,
+      "learning_rate": 1.3898666666666666e-05,
+      "loss": 1.6634,
+      "step": 23410
+    },
+    {
+      "epoch": 0.74944,
+      "grad_norm": 45.20486831665039,
+      "learning_rate": 1.3895111111111112e-05,
+      "loss": 1.614,
+      "step": 23420
+    },
+    {
+      "epoch": 0.74976,
+      "grad_norm": 44.7941780090332,
+      "learning_rate": 1.3891555555555555e-05,
+      "loss": 1.6562,
+      "step": 23430
+    },
+    {
+      "epoch": 0.75008,
+      "grad_norm": 45.210330963134766,
+      "learning_rate": 1.3888000000000002e-05,
+      "loss": 1.6645,
+      "step": 23440
+    },
+    {
+      "epoch": 0.7504,
+      "grad_norm": 43.68789291381836,
+      "learning_rate": 1.3884444444444444e-05,
+      "loss": 1.6724,
+      "step": 23450
+    },
+    {
+      "epoch": 0.75072,
+      "grad_norm": 44.261295318603516,
+      "learning_rate": 1.388088888888889e-05,
+      "loss": 1.6311,
+      "step": 23460
+    },
+    {
+      "epoch": 0.75104,
+      "grad_norm": 43.10234832763672,
+      "learning_rate": 1.3877333333333334e-05,
+      "loss": 1.664,
+      "step": 23470
+    },
+    {
+      "epoch": 0.75136,
+      "grad_norm": 44.797393798828125,
+      "learning_rate": 1.387377777777778e-05,
+      "loss": 1.6604,
+      "step": 23480
+    },
+    {
+      "epoch": 0.75168,
+      "grad_norm": 44.64009094238281,
+      "learning_rate": 1.3870222222222223e-05,
+      "loss": 1.6917,
+      "step": 23490
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 44.40959167480469,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 1.6451,
+      "step": 23500
+    },
+    {
+      "epoch": 0.75232,
+      "grad_norm": 45.48454284667969,
+      "learning_rate": 1.3863111111111112e-05,
+      "loss": 1.6607,
+      "step": 23510
+    },
+    {
+      "epoch": 0.75264,
+      "grad_norm": 42.83772277832031,
+      "learning_rate": 1.3859555555555558e-05,
+      "loss": 1.6673,
+      "step": 23520
+    },
+    {
+      "epoch": 0.75296,
+      "grad_norm": 43.72276306152344,
+      "learning_rate": 1.3856e-05,
+      "loss": 1.6506,
+      "step": 23530
+    },
+    {
+      "epoch": 0.75328,
+      "grad_norm": 45.616329193115234,
+      "learning_rate": 1.3852444444444445e-05,
+      "loss": 1.6674,
+      "step": 23540
+    },
+    {
+      "epoch": 0.7536,
+      "grad_norm": 42.29286193847656,
+      "learning_rate": 1.384888888888889e-05,
+      "loss": 1.6449,
+      "step": 23550
+    },
+    {
+      "epoch": 0.75392,
+      "grad_norm": 43.88602066040039,
+      "learning_rate": 1.3845333333333334e-05,
+      "loss": 1.6775,
+      "step": 23560
+    },
+    {
+      "epoch": 0.75424,
+      "grad_norm": 41.9892578125,
+      "learning_rate": 1.3841777777777779e-05,
+      "loss": 1.6498,
+      "step": 23570
+    },
+    {
+      "epoch": 0.75456,
+      "grad_norm": 44.01221466064453,
+      "learning_rate": 1.3838222222222224e-05,
+      "loss": 1.6794,
+      "step": 23580
+    },
+    {
+      "epoch": 0.75488,
+      "grad_norm": 47.409400939941406,
+      "learning_rate": 1.3834666666666668e-05,
+      "loss": 1.692,
+      "step": 23590
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 47.6396484375,
+      "learning_rate": 1.3831111111111113e-05,
+      "loss": 1.652,
+      "step": 23600
+    },
+    {
+      "epoch": 0.75552,
+      "grad_norm": 44.32497024536133,
+      "learning_rate": 1.3827555555555556e-05,
+      "loss": 1.6507,
+      "step": 23610
+    },
+    {
+      "epoch": 0.75584,
+      "grad_norm": 43.5431022644043,
+      "learning_rate": 1.3824000000000002e-05,
+      "loss": 1.6976,
+      "step": 23620
+    },
+    {
+      "epoch": 0.75616,
+      "grad_norm": 42.28642272949219,
+      "learning_rate": 1.3820444444444445e-05,
+      "loss": 1.6915,
+      "step": 23630
+    },
+    {
+      "epoch": 0.75648,
+      "grad_norm": 42.09159469604492,
+      "learning_rate": 1.3816888888888891e-05,
+      "loss": 1.6478,
+      "step": 23640
+    },
+    {
+      "epoch": 0.7568,
+      "grad_norm": 44.362796783447266,
+      "learning_rate": 1.3813333333333334e-05,
+      "loss": 1.6668,
+      "step": 23650
+    },
+    {
+      "epoch": 0.75712,
+      "grad_norm": 43.37620544433594,
+      "learning_rate": 1.380977777777778e-05,
+      "loss": 1.6616,
+      "step": 23660
+    },
+    {
+      "epoch": 0.75744,
+      "grad_norm": 42.7562141418457,
+      "learning_rate": 1.3806222222222223e-05,
+      "loss": 1.6511,
+      "step": 23670
+    },
+    {
+      "epoch": 0.75776,
+      "grad_norm": 43.32087707519531,
+      "learning_rate": 1.3802666666666669e-05,
+      "loss": 1.6735,
+      "step": 23680
+    },
+    {
+      "epoch": 0.75808,
+      "grad_norm": 43.88164520263672,
+      "learning_rate": 1.3799111111111112e-05,
+      "loss": 1.6521,
+      "step": 23690
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 44.2587776184082,
+      "learning_rate": 1.3795555555555558e-05,
+      "loss": 1.6539,
+      "step": 23700
+    },
+    {
+      "epoch": 0.75872,
+      "grad_norm": 43.35541915893555,
+      "learning_rate": 1.3792000000000001e-05,
+      "loss": 1.6679,
+      "step": 23710
+    },
+    {
+      "epoch": 0.75904,
+      "grad_norm": 44.739402770996094,
+      "learning_rate": 1.3788444444444447e-05,
+      "loss": 1.6677,
+      "step": 23720
+    },
+    {
+      "epoch": 0.75936,
+      "grad_norm": 42.22761154174805,
+      "learning_rate": 1.378488888888889e-05,
+      "loss": 1.6598,
+      "step": 23730
+    },
+    {
+      "epoch": 0.75968,
+      "grad_norm": 45.30773162841797,
+      "learning_rate": 1.3781333333333335e-05,
+      "loss": 1.6386,
+      "step": 23740
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 46.231231689453125,
+      "learning_rate": 1.377777777777778e-05,
+      "loss": 1.6652,
+      "step": 23750
+    },
+    {
+      "epoch": 0.76032,
+      "grad_norm": 43.435951232910156,
+      "learning_rate": 1.3774222222222224e-05,
+      "loss": 1.6611,
+      "step": 23760
+    },
+    {
+      "epoch": 0.76064,
+      "grad_norm": 45.195953369140625,
+      "learning_rate": 1.3770666666666668e-05,
+      "loss": 1.6879,
+      "step": 23770
+    },
+    {
+      "epoch": 0.76096,
+      "grad_norm": 44.517269134521484,
+      "learning_rate": 1.3767111111111113e-05,
+      "loss": 1.6266,
+      "step": 23780
+    },
+    {
+      "epoch": 0.76128,
+      "grad_norm": 47.46632766723633,
+      "learning_rate": 1.3763555555555556e-05,
+      "loss": 1.693,
+      "step": 23790
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 44.885990142822266,
+      "learning_rate": 1.376e-05,
+      "loss": 1.651,
+      "step": 23800
+    },
+    {
+      "epoch": 0.76192,
+      "grad_norm": 45.48068618774414,
+      "learning_rate": 1.3756444444444445e-05,
+      "loss": 1.6616,
+      "step": 23810
+    },
+    {
+      "epoch": 0.76224,
+      "grad_norm": 44.51014709472656,
+      "learning_rate": 1.375288888888889e-05,
+      "loss": 1.7018,
+      "step": 23820
+    },
+    {
+      "epoch": 0.76256,
+      "grad_norm": 43.273250579833984,
+      "learning_rate": 1.3749333333333334e-05,
+      "loss": 1.658,
+      "step": 23830
+    },
+    {
+      "epoch": 0.76288,
+      "grad_norm": 44.35025405883789,
+      "learning_rate": 1.3745777777777779e-05,
+      "loss": 1.6565,
+      "step": 23840
+    },
+    {
+      "epoch": 0.7632,
+      "grad_norm": 46.16026306152344,
+      "learning_rate": 1.3742222222222223e-05,
+      "loss": 1.6601,
+      "step": 23850
+    },
+    {
+      "epoch": 0.76352,
+      "grad_norm": 45.772403717041016,
+      "learning_rate": 1.3738666666666666e-05,
+      "loss": 1.6429,
+      "step": 23860
+    },
+    {
+      "epoch": 0.76384,
+      "grad_norm": 43.03438949584961,
+      "learning_rate": 1.3735111111111112e-05,
+      "loss": 1.6594,
+      "step": 23870
+    },
+    {
+      "epoch": 0.76416,
+      "grad_norm": 45.24376678466797,
+      "learning_rate": 1.3731555555555555e-05,
+      "loss": 1.6646,
+      "step": 23880
+    },
+    {
+      "epoch": 0.76448,
+      "grad_norm": 43.241172790527344,
+      "learning_rate": 1.3728000000000001e-05,
+      "loss": 1.652,
+      "step": 23890
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 43.52678298950195,
+      "learning_rate": 1.3724444444444444e-05,
+      "loss": 1.6588,
+      "step": 23900
+    },
+    {
+      "epoch": 0.76512,
+      "grad_norm": 44.89611053466797,
+      "learning_rate": 1.372088888888889e-05,
+      "loss": 1.6523,
+      "step": 23910
+    },
+    {
+      "epoch": 0.76544,
+      "grad_norm": 43.12721252441406,
+      "learning_rate": 1.3717333333333333e-05,
+      "loss": 1.6595,
+      "step": 23920
+    },
+    {
+      "epoch": 0.76576,
+      "grad_norm": 47.17388153076172,
+      "learning_rate": 1.371377777777778e-05,
+      "loss": 1.6648,
+      "step": 23930
+    },
+    {
+      "epoch": 0.76608,
+      "grad_norm": 43.1567268371582,
+      "learning_rate": 1.3710222222222222e-05,
+      "loss": 1.6707,
+      "step": 23940
+    },
+    {
+      "epoch": 0.7664,
+      "grad_norm": 44.261985778808594,
+      "learning_rate": 1.3706666666666669e-05,
+      "loss": 1.6733,
+      "step": 23950
+    },
+    {
+      "epoch": 0.76672,
+      "grad_norm": 46.473670959472656,
+      "learning_rate": 1.3703111111111112e-05,
+      "loss": 1.6944,
+      "step": 23960
+    },
+    {
+      "epoch": 0.76704,
+      "grad_norm": 44.448974609375,
+      "learning_rate": 1.3699555555555558e-05,
+      "loss": 1.6654,
+      "step": 23970
+    },
+    {
+      "epoch": 0.76736,
+      "grad_norm": 42.94993209838867,
+      "learning_rate": 1.3696e-05,
+      "loss": 1.6726,
+      "step": 23980
+    },
+    {
+      "epoch": 0.76768,
+      "grad_norm": 43.04637908935547,
+      "learning_rate": 1.3692444444444445e-05,
+      "loss": 1.6726,
+      "step": 23990
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 45.355377197265625,
+      "learning_rate": 1.368888888888889e-05,
+      "loss": 1.6507,
+      "step": 24000
+    },
+    {
+      "epoch": 0.76832,
+      "grad_norm": 44.12913131713867,
+      "learning_rate": 1.3685333333333334e-05,
+      "loss": 1.6675,
+      "step": 24010
+    },
+    {
+      "epoch": 0.76864,
+      "grad_norm": 44.19668197631836,
+      "learning_rate": 1.3681777777777779e-05,
+      "loss": 1.6521,
+      "step": 24020
+    },
+    {
+      "epoch": 0.76896,
+      "grad_norm": 43.70733642578125,
+      "learning_rate": 1.3678222222222223e-05,
+      "loss": 1.6814,
+      "step": 24030
+    },
+    {
+      "epoch": 0.76928,
+      "grad_norm": 45.314842224121094,
+      "learning_rate": 1.3674666666666668e-05,
+      "loss": 1.6672,
+      "step": 24040
+    },
+    {
+      "epoch": 0.7696,
+      "grad_norm": 43.1247444152832,
+      "learning_rate": 1.3671111111111113e-05,
+      "loss": 1.6249,
+      "step": 24050
+    },
+    {
+      "epoch": 0.76992,
+      "grad_norm": 42.034088134765625,
+      "learning_rate": 1.3667555555555555e-05,
+      "loss": 1.618,
+      "step": 24060
+    },
+    {
+      "epoch": 0.77024,
+      "grad_norm": 42.96197509765625,
+      "learning_rate": 1.3664000000000002e-05,
+      "loss": 1.6626,
+      "step": 24070
+    },
+    {
+      "epoch": 0.77056,
+      "grad_norm": 44.40598678588867,
+      "learning_rate": 1.3660444444444445e-05,
+      "loss": 1.6692,
+      "step": 24080
+    },
+    {
+      "epoch": 0.77088,
+      "grad_norm": 44.56721878051758,
+      "learning_rate": 1.365688888888889e-05,
+      "loss": 1.6664,
+      "step": 24090
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 43.622642517089844,
+      "learning_rate": 1.3653333333333334e-05,
+      "loss": 1.6357,
+      "step": 24100
+    },
+    {
+      "epoch": 0.77152,
+      "grad_norm": 44.824275970458984,
+      "learning_rate": 1.364977777777778e-05,
+      "loss": 1.6538,
+      "step": 24110
+    },
+    {
+      "epoch": 0.77184,
+      "grad_norm": 46.225868225097656,
+      "learning_rate": 1.3646222222222223e-05,
+      "loss": 1.6541,
+      "step": 24120
+    },
+    {
+      "epoch": 0.77216,
+      "grad_norm": 44.35313034057617,
+      "learning_rate": 1.3642666666666669e-05,
+      "loss": 1.6592,
+      "step": 24130
+    },
+    {
+      "epoch": 0.77248,
+      "grad_norm": 43.661659240722656,
+      "learning_rate": 1.3639111111111112e-05,
+      "loss": 1.6533,
+      "step": 24140
+    },
+    {
+      "epoch": 0.7728,
+      "grad_norm": 42.956539154052734,
+      "learning_rate": 1.3635555555555558e-05,
+      "loss": 1.6678,
+      "step": 24150
+    },
+    {
+      "epoch": 0.77312,
+      "grad_norm": 42.41958236694336,
+      "learning_rate": 1.3632000000000001e-05,
+      "loss": 1.6558,
+      "step": 24160
+    },
+    {
+      "epoch": 0.77344,
+      "grad_norm": 44.25238037109375,
+      "learning_rate": 1.3628444444444447e-05,
+      "loss": 1.6258,
+      "step": 24170
+    },
+    {
+      "epoch": 0.77376,
+      "grad_norm": 44.08992004394531,
+      "learning_rate": 1.362488888888889e-05,
+      "loss": 1.669,
+      "step": 24180
+    },
+    {
+      "epoch": 0.77408,
+      "grad_norm": 44.67106246948242,
+      "learning_rate": 1.3621333333333335e-05,
+      "loss": 1.6765,
+      "step": 24190
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 43.50186538696289,
+      "learning_rate": 1.361777777777778e-05,
+      "loss": 1.6386,
+      "step": 24200
+    },
+    {
+      "epoch": 0.77472,
+      "grad_norm": 45.129730224609375,
+      "learning_rate": 1.3614222222222224e-05,
+      "loss": 1.6417,
+      "step": 24210
+    },
+    {
+      "epoch": 0.77504,
+      "grad_norm": 43.64735794067383,
+      "learning_rate": 1.3610666666666668e-05,
+      "loss": 1.6336,
+      "step": 24220
+    },
+    {
+      "epoch": 0.77536,
+      "grad_norm": 45.431678771972656,
+      "learning_rate": 1.3607111111111113e-05,
+      "loss": 1.6866,
+      "step": 24230
+    },
+    {
+      "epoch": 0.77568,
+      "grad_norm": 45.0831298828125,
+      "learning_rate": 1.3603555555555556e-05,
+      "loss": 1.6499,
+      "step": 24240
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 43.368614196777344,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.6655,
+      "step": 24250
+    },
+    {
+      "epoch": 0.77632,
+      "grad_norm": 42.33442687988281,
+      "learning_rate": 1.3596444444444445e-05,
+      "loss": 1.6556,
+      "step": 24260
+    },
+    {
+      "epoch": 0.77664,
+      "grad_norm": 45.05806350708008,
+      "learning_rate": 1.3592888888888891e-05,
+      "loss": 1.6383,
+      "step": 24270
+    },
+    {
+      "epoch": 0.77696,
+      "grad_norm": 43.16596603393555,
+      "learning_rate": 1.3589333333333334e-05,
+      "loss": 1.663,
+      "step": 24280
+    },
+    {
+      "epoch": 0.77728,
+      "grad_norm": 46.41241455078125,
+      "learning_rate": 1.358577777777778e-05,
+      "loss": 1.6907,
+      "step": 24290
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 46.05997848510742,
+      "learning_rate": 1.3582222222222223e-05,
+      "loss": 1.6556,
+      "step": 24300
+    },
+    {
+      "epoch": 0.77792,
+      "grad_norm": 44.43130111694336,
+      "learning_rate": 1.357866666666667e-05,
+      "loss": 1.6517,
+      "step": 24310
+    },
+    {
+      "epoch": 0.77824,
+      "grad_norm": 43.29718780517578,
+      "learning_rate": 1.3575111111111112e-05,
+      "loss": 1.651,
+      "step": 24320
+    },
+    {
+      "epoch": 0.77856,
+      "grad_norm": 45.74983596801758,
+      "learning_rate": 1.3571555555555555e-05,
+      "loss": 1.6475,
+      "step": 24330
+    },
+    {
+      "epoch": 0.77888,
+      "grad_norm": 43.65877914428711,
+      "learning_rate": 1.3568000000000001e-05,
+      "loss": 1.6843,
+      "step": 24340
+    },
+    {
+      "epoch": 0.7792,
+      "grad_norm": 47.665771484375,
+      "learning_rate": 1.3564444444444444e-05,
+      "loss": 1.6628,
+      "step": 24350
+    },
+    {
+      "epoch": 0.77952,
+      "grad_norm": 43.86433029174805,
+      "learning_rate": 1.356088888888889e-05,
+      "loss": 1.6532,
+      "step": 24360
+    },
+    {
+      "epoch": 0.77984,
+      "grad_norm": 43.74129867553711,
+      "learning_rate": 1.3557333333333333e-05,
+      "loss": 1.6488,
+      "step": 24370
+    },
+    {
+      "epoch": 0.78016,
+      "grad_norm": 43.49934387207031,
+      "learning_rate": 1.355377777777778e-05,
+      "loss": 1.6495,
+      "step": 24380
+    },
+    {
+      "epoch": 0.78048,
+      "grad_norm": 44.210689544677734,
+      "learning_rate": 1.3550222222222222e-05,
+      "loss": 1.6597,
+      "step": 24390
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 42.381004333496094,
+      "learning_rate": 1.3546666666666669e-05,
+      "loss": 1.6672,
+      "step": 24400
+    },
+    {
+      "epoch": 0.78112,
+      "grad_norm": 44.646244049072266,
+      "learning_rate": 1.3543111111111111e-05,
+      "loss": 1.6598,
+      "step": 24410
+    },
+    {
+      "epoch": 0.78144,
+      "grad_norm": 43.49140167236328,
+      "learning_rate": 1.3539555555555558e-05,
+      "loss": 1.6601,
+      "step": 24420
+    },
+    {
+      "epoch": 0.78176,
+      "grad_norm": 43.86427688598633,
+      "learning_rate": 1.3536e-05,
+      "loss": 1.625,
+      "step": 24430
+    },
+    {
+      "epoch": 0.78208,
+      "grad_norm": 44.45777893066406,
+      "learning_rate": 1.3532444444444445e-05,
+      "loss": 1.6933,
+      "step": 24440
+    },
+    {
+      "epoch": 0.7824,
+      "grad_norm": 45.80703353881836,
+      "learning_rate": 1.352888888888889e-05,
+      "loss": 1.6606,
+      "step": 24450
+    },
+    {
+      "epoch": 0.78272,
+      "grad_norm": 43.04664993286133,
+      "learning_rate": 1.3525333333333334e-05,
+      "loss": 1.6555,
+      "step": 24460
+    },
+    {
+      "epoch": 0.78304,
+      "grad_norm": 44.184017181396484,
+      "learning_rate": 1.3521777777777779e-05,
+      "loss": 1.6533,
+      "step": 24470
+    },
+    {
+      "epoch": 0.78336,
+      "grad_norm": 45.938289642333984,
+      "learning_rate": 1.3518222222222223e-05,
+      "loss": 1.6599,
+      "step": 24480
+    },
+    {
+      "epoch": 0.78368,
+      "grad_norm": 43.94887924194336,
+      "learning_rate": 1.3514666666666668e-05,
+      "loss": 1.6648,
+      "step": 24490
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 43.67192459106445,
+      "learning_rate": 1.3511111111111112e-05,
+      "loss": 1.6821,
+      "step": 24500
+    },
+    {
+      "epoch": 0.78432,
+      "grad_norm": 44.66633987426758,
+      "learning_rate": 1.3507555555555555e-05,
+      "loss": 1.6611,
+      "step": 24510
+    },
+    {
+      "epoch": 0.78464,
+      "grad_norm": 43.301578521728516,
+      "learning_rate": 1.3504000000000001e-05,
+      "loss": 1.6827,
+      "step": 24520
+    },
+    {
+      "epoch": 0.78496,
+      "grad_norm": 44.07039260864258,
+      "learning_rate": 1.3500444444444444e-05,
+      "loss": 1.6637,
+      "step": 24530
+    },
+    {
+      "epoch": 0.78528,
+      "grad_norm": 43.60643768310547,
+      "learning_rate": 1.349688888888889e-05,
+      "loss": 1.6485,
+      "step": 24540
+    },
+    {
+      "epoch": 0.7856,
+      "grad_norm": 42.864803314208984,
+      "learning_rate": 1.3493333333333333e-05,
+      "loss": 1.6656,
+      "step": 24550
+    },
+    {
+      "epoch": 0.78592,
+      "grad_norm": 44.05122375488281,
+      "learning_rate": 1.348977777777778e-05,
+      "loss": 1.6528,
+      "step": 24560
+    },
+    {
+      "epoch": 0.78624,
+      "grad_norm": 44.83565139770508,
+      "learning_rate": 1.3486222222222223e-05,
+      "loss": 1.6626,
+      "step": 24570
+    },
+    {
+      "epoch": 0.78656,
+      "grad_norm": 44.86031723022461,
+      "learning_rate": 1.3482666666666669e-05,
+      "loss": 1.6658,
+      "step": 24580
+    },
+    {
+      "epoch": 0.78688,
+      "grad_norm": 43.807945251464844,
+      "learning_rate": 1.3479111111111112e-05,
+      "loss": 1.6325,
+      "step": 24590
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 43.93977355957031,
+      "learning_rate": 1.3475555555555558e-05,
+      "loss": 1.6677,
+      "step": 24600
+    },
+    {
+      "epoch": 0.78752,
+      "grad_norm": 43.10548400878906,
+      "learning_rate": 1.3472e-05,
+      "loss": 1.6611,
+      "step": 24610
+    },
+    {
+      "epoch": 0.78784,
+      "grad_norm": 42.78868865966797,
+      "learning_rate": 1.3468444444444447e-05,
+      "loss": 1.6303,
+      "step": 24620
+    },
+    {
+      "epoch": 0.78816,
+      "grad_norm": 46.41221618652344,
+      "learning_rate": 1.346488888888889e-05,
+      "loss": 1.643,
+      "step": 24630
+    },
+    {
+      "epoch": 0.78848,
+      "grad_norm": 43.43020248413086,
+      "learning_rate": 1.3461333333333334e-05,
+      "loss": 1.6406,
+      "step": 24640
+    },
+    {
+      "epoch": 0.7888,
+      "grad_norm": 45.2264404296875,
+      "learning_rate": 1.3457777777777779e-05,
+      "loss": 1.6771,
+      "step": 24650
+    },
+    {
+      "epoch": 0.78912,
+      "grad_norm": 41.3878288269043,
+      "learning_rate": 1.3454222222222224e-05,
+      "loss": 1.6309,
+      "step": 24660
+    },
+    {
+      "epoch": 0.78944,
+      "grad_norm": 44.629066467285156,
+      "learning_rate": 1.3450666666666668e-05,
+      "loss": 1.6698,
+      "step": 24670
+    },
+    {
+      "epoch": 0.78976,
+      "grad_norm": 43.920928955078125,
+      "learning_rate": 1.3447111111111113e-05,
+      "loss": 1.6432,
+      "step": 24680
+    },
+    {
+      "epoch": 0.79008,
+      "grad_norm": 44.86647033691406,
+      "learning_rate": 1.3443555555555556e-05,
+      "loss": 1.6829,
+      "step": 24690
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 44.51643371582031,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 1.6599,
+      "step": 24700
+    },
+    {
+      "epoch": 0.79072,
+      "grad_norm": 42.92637252807617,
+      "learning_rate": 1.3436444444444445e-05,
+      "loss": 1.6457,
+      "step": 24710
+    },
+    {
+      "epoch": 0.79104,
+      "grad_norm": 44.864200592041016,
+      "learning_rate": 1.3432888888888891e-05,
+      "loss": 1.6647,
+      "step": 24720
+    },
+    {
+      "epoch": 0.79136,
+      "grad_norm": 43.50864028930664,
+      "learning_rate": 1.3429333333333334e-05,
+      "loss": 1.6836,
+      "step": 24730
+    },
+    {
+      "epoch": 0.79168,
+      "grad_norm": 44.62223434448242,
+      "learning_rate": 1.342577777777778e-05,
+      "loss": 1.6503,
+      "step": 24740
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 44.60259246826172,
+      "learning_rate": 1.3422222222222223e-05,
+      "loss": 1.6764,
+      "step": 24750
+    },
+    {
+      "epoch": 0.79232,
+      "grad_norm": 43.178924560546875,
+      "learning_rate": 1.3418666666666669e-05,
+      "loss": 1.6601,
+      "step": 24760
+    },
+    {
+      "epoch": 0.79264,
+      "grad_norm": 44.4167594909668,
+      "learning_rate": 1.3415111111111112e-05,
+      "loss": 1.654,
+      "step": 24770
+    },
+    {
+      "epoch": 0.79296,
+      "grad_norm": 45.009437561035156,
+      "learning_rate": 1.3411555555555558e-05,
+      "loss": 1.6622,
+      "step": 24780
+    },
+    {
+      "epoch": 0.79328,
+      "grad_norm": 44.56413269042969,
+      "learning_rate": 1.3408000000000001e-05,
+      "loss": 1.6705,
+      "step": 24790
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 42.48627853393555,
+      "learning_rate": 1.3404444444444447e-05,
+      "loss": 1.6634,
+      "step": 24800
+    },
+    {
+      "epoch": 0.79392,
+      "grad_norm": 44.3420524597168,
+      "learning_rate": 1.340088888888889e-05,
+      "loss": 1.6268,
+      "step": 24810
+    },
+    {
+      "epoch": 0.79424,
+      "grad_norm": 42.205543518066406,
+      "learning_rate": 1.3397333333333335e-05,
+      "loss": 1.6493,
+      "step": 24820
+    },
+    {
+      "epoch": 0.79456,
+      "grad_norm": 44.827919006347656,
+      "learning_rate": 1.339377777777778e-05,
+      "loss": 1.637,
+      "step": 24830
+    },
+    {
+      "epoch": 0.79488,
+      "grad_norm": 43.14591979980469,
+      "learning_rate": 1.3390222222222222e-05,
+      "loss": 1.6362,
+      "step": 24840
+    },
+    {
+      "epoch": 0.7952,
+      "grad_norm": 43.567264556884766,
+      "learning_rate": 1.3386666666666668e-05,
+      "loss": 1.6551,
+      "step": 24850
+    },
+    {
+      "epoch": 0.79552,
+      "grad_norm": 44.323707580566406,
+      "learning_rate": 1.3383111111111111e-05,
+      "loss": 1.6591,
+      "step": 24860
+    },
+    {
+      "epoch": 0.79584,
+      "grad_norm": 41.868961334228516,
+      "learning_rate": 1.3379555555555557e-05,
+      "loss": 1.6621,
+      "step": 24870
+    },
+    {
+      "epoch": 0.79616,
+      "grad_norm": 43.79658126831055,
+      "learning_rate": 1.3376e-05,
+      "loss": 1.6515,
+      "step": 24880
+    },
+    {
+      "epoch": 0.79648,
+      "grad_norm": 48.700313568115234,
+      "learning_rate": 1.3372444444444445e-05,
+      "loss": 1.6455,
+      "step": 24890
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 42.93735122680664,
+      "learning_rate": 1.336888888888889e-05,
+      "loss": 1.6396,
+      "step": 24900
+    },
+    {
+      "epoch": 0.79712,
+      "grad_norm": 46.09784698486328,
+      "learning_rate": 1.3365333333333334e-05,
+      "loss": 1.6419,
+      "step": 24910
+    },
+    {
+      "epoch": 0.79744,
+      "grad_norm": 42.79138946533203,
+      "learning_rate": 1.3361777777777779e-05,
+      "loss": 1.6685,
+      "step": 24920
+    },
+    {
+      "epoch": 0.79776,
+      "grad_norm": 46.662940979003906,
+      "learning_rate": 1.3358222222222223e-05,
+      "loss": 1.6568,
+      "step": 24930
+    },
+    {
+      "epoch": 0.79808,
+      "grad_norm": 46.082244873046875,
+      "learning_rate": 1.3354666666666668e-05,
+      "loss": 1.6478,
+      "step": 24940
+    },
+    {
+      "epoch": 0.7984,
+      "grad_norm": 45.16201400756836,
+      "learning_rate": 1.3351111111111112e-05,
+      "loss": 1.6585,
+      "step": 24950
+    },
+    {
+      "epoch": 0.79872,
+      "grad_norm": 44.113380432128906,
+      "learning_rate": 1.3347555555555555e-05,
+      "loss": 1.6702,
+      "step": 24960
+    },
+    {
+      "epoch": 0.79904,
+      "grad_norm": 44.27860641479492,
+      "learning_rate": 1.3344000000000001e-05,
+      "loss": 1.642,
+      "step": 24970
+    },
+    {
+      "epoch": 0.79936,
+      "grad_norm": 45.675655364990234,
+      "learning_rate": 1.3340444444444444e-05,
+      "loss": 1.6243,
+      "step": 24980
+    },
+    {
+      "epoch": 0.79968,
+      "grad_norm": 45.01850891113281,
+      "learning_rate": 1.333688888888889e-05,
+      "loss": 1.6773,
+      "step": 24990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 44.070579528808594,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.6678,
+      "step": 25000
+    },
+    {
+      "epoch": 0.80032,
+      "grad_norm": 44.78519058227539,
+      "learning_rate": 1.332977777777778e-05,
+      "loss": 1.6671,
+      "step": 25010
+    },
+    {
+      "epoch": 0.80064,
+      "grad_norm": 45.15095138549805,
+      "learning_rate": 1.3326222222222222e-05,
+      "loss": 1.6402,
+      "step": 25020
+    },
+    {
+      "epoch": 0.80096,
+      "grad_norm": 43.276493072509766,
+      "learning_rate": 1.3322666666666669e-05,
+      "loss": 1.6672,
+      "step": 25030
+    },
+    {
+      "epoch": 0.80128,
+      "grad_norm": 44.25715255737305,
+      "learning_rate": 1.3319111111111112e-05,
+      "loss": 1.6302,
+      "step": 25040
+    },
+    {
+      "epoch": 0.8016,
+      "grad_norm": 43.42182540893555,
+      "learning_rate": 1.3315555555555558e-05,
+      "loss": 1.6653,
+      "step": 25050
+    },
+    {
+      "epoch": 0.80192,
+      "grad_norm": 46.8543586730957,
+      "learning_rate": 1.3312e-05,
+      "loss": 1.687,
+      "step": 25060
+    },
+    {
+      "epoch": 0.80224,
+      "grad_norm": 44.426082611083984,
+      "learning_rate": 1.3308444444444447e-05,
+      "loss": 1.6328,
+      "step": 25070
+    },
+    {
+      "epoch": 0.80256,
+      "grad_norm": 46.286983489990234,
+      "learning_rate": 1.330488888888889e-05,
+      "loss": 1.6477,
+      "step": 25080
+    },
+    {
+      "epoch": 0.80288,
+      "grad_norm": 44.83964920043945,
+      "learning_rate": 1.3301333333333334e-05,
+      "loss": 1.6905,
+      "step": 25090
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 44.6434326171875,
+      "learning_rate": 1.3297777777777779e-05,
+      "loss": 1.6449,
+      "step": 25100
+    },
+    {
+      "epoch": 0.80352,
+      "grad_norm": 44.10059356689453,
+      "learning_rate": 1.3294222222222223e-05,
+      "loss": 1.6668,
+      "step": 25110
+    },
+    {
+      "epoch": 0.80384,
+      "grad_norm": 43.977718353271484,
+      "learning_rate": 1.3290666666666668e-05,
+      "loss": 1.6696,
+      "step": 25120
+    },
+    {
+      "epoch": 0.80416,
+      "grad_norm": 43.653648376464844,
+      "learning_rate": 1.3287111111111112e-05,
+      "loss": 1.6548,
+      "step": 25130
+    },
+    {
+      "epoch": 0.80448,
+      "grad_norm": 42.916500091552734,
+      "learning_rate": 1.3283555555555557e-05,
+      "loss": 1.6852,
+      "step": 25140
+    },
+    {
+      "epoch": 0.8048,
+      "grad_norm": 44.64195251464844,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 1.6425,
+      "step": 25150
+    },
+    {
+      "epoch": 0.80512,
+      "grad_norm": 44.12642288208008,
+      "learning_rate": 1.3276444444444444e-05,
+      "loss": 1.6456,
+      "step": 25160
+    },
+    {
+      "epoch": 0.80544,
+      "grad_norm": 44.04334259033203,
+      "learning_rate": 1.327288888888889e-05,
+      "loss": 1.6704,
+      "step": 25170
+    },
+    {
+      "epoch": 0.80576,
+      "grad_norm": 43.638301849365234,
+      "learning_rate": 1.3269333333333334e-05,
+      "loss": 1.6545,
+      "step": 25180
+    },
+    {
+      "epoch": 0.80608,
+      "grad_norm": 44.577510833740234,
+      "learning_rate": 1.326577777777778e-05,
+      "loss": 1.6479,
+      "step": 25190
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 43.7669563293457,
+      "learning_rate": 1.3262222222222223e-05,
+      "loss": 1.6511,
+      "step": 25200
+    },
+    {
+      "epoch": 0.80672,
+      "grad_norm": 44.01200485229492,
+      "learning_rate": 1.3258666666666669e-05,
+      "loss": 1.6738,
+      "step": 25210
+    },
+    {
+      "epoch": 0.80704,
+      "grad_norm": 44.7772216796875,
+      "learning_rate": 1.3255111111111112e-05,
+      "loss": 1.6402,
+      "step": 25220
+    },
+    {
+      "epoch": 0.80736,
+      "grad_norm": 43.836692810058594,
+      "learning_rate": 1.3251555555555558e-05,
+      "loss": 1.6447,
+      "step": 25230
+    },
+    {
+      "epoch": 0.80768,
+      "grad_norm": 43.029361724853516,
+      "learning_rate": 1.3248000000000001e-05,
+      "loss": 1.6474,
+      "step": 25240
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 44.712398529052734,
+      "learning_rate": 1.3244444444444447e-05,
+      "loss": 1.6796,
+      "step": 25250
+    },
+    {
+      "epoch": 0.80832,
+      "grad_norm": 44.133628845214844,
+      "learning_rate": 1.324088888888889e-05,
+      "loss": 1.6523,
+      "step": 25260
+    },
+    {
+      "epoch": 0.80864,
+      "grad_norm": 43.37736129760742,
+      "learning_rate": 1.3237333333333335e-05,
+      "loss": 1.659,
+      "step": 25270
+    },
+    {
+      "epoch": 0.80896,
+      "grad_norm": 45.66469955444336,
+      "learning_rate": 1.3233777777777779e-05,
+      "loss": 1.6422,
+      "step": 25280
+    },
+    {
+      "epoch": 0.80928,
+      "grad_norm": 44.68482208251953,
+      "learning_rate": 1.3230222222222224e-05,
+      "loss": 1.6417,
+      "step": 25290
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 45.07290267944336,
+      "learning_rate": 1.3226666666666668e-05,
+      "loss": 1.6818,
+      "step": 25300
+    },
+    {
+      "epoch": 0.80992,
+      "grad_norm": 44.1818962097168,
+      "learning_rate": 1.3223111111111113e-05,
+      "loss": 1.6762,
+      "step": 25310
+    },
+    {
+      "epoch": 0.81024,
+      "grad_norm": 44.05356216430664,
+      "learning_rate": 1.3219555555555557e-05,
+      "loss": 1.6383,
+      "step": 25320
+    },
+    {
+      "epoch": 0.81056,
+      "grad_norm": 43.593894958496094,
+      "learning_rate": 1.3216000000000002e-05,
+      "loss": 1.6475,
+      "step": 25330
+    },
+    {
+      "epoch": 0.81088,
+      "grad_norm": 44.107872009277344,
+      "learning_rate": 1.3212444444444445e-05,
+      "loss": 1.6635,
+      "step": 25340
+    },
+    {
+      "epoch": 0.8112,
+      "grad_norm": 45.551185607910156,
+      "learning_rate": 1.320888888888889e-05,
+      "loss": 1.66,
+      "step": 25350
+    },
+    {
+      "epoch": 0.81152,
+      "grad_norm": 44.834293365478516,
+      "learning_rate": 1.3205333333333334e-05,
+      "loss": 1.6653,
+      "step": 25360
+    },
+    {
+      "epoch": 0.81184,
+      "grad_norm": 44.376991271972656,
+      "learning_rate": 1.3201777777777778e-05,
+      "loss": 1.6607,
+      "step": 25370
+    },
+    {
+      "epoch": 0.81216,
+      "grad_norm": 46.16840744018555,
+      "learning_rate": 1.3198222222222223e-05,
+      "loss": 1.6611,
+      "step": 25380
+    },
+    {
+      "epoch": 0.81248,
+      "grad_norm": 43.237728118896484,
+      "learning_rate": 1.3194666666666668e-05,
+      "loss": 1.6563,
+      "step": 25390
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 46.14860153198242,
+      "learning_rate": 1.3191111111111112e-05,
+      "loss": 1.6526,
+      "step": 25400
+    },
+    {
+      "epoch": 0.81312,
+      "grad_norm": 43.493927001953125,
+      "learning_rate": 1.3187555555555555e-05,
+      "loss": 1.6353,
+      "step": 25410
+    },
+    {
+      "epoch": 0.81344,
+      "grad_norm": 43.593013763427734,
+      "learning_rate": 1.3184000000000001e-05,
+      "loss": 1.6373,
+      "step": 25420
+    },
+    {
+      "epoch": 0.81376,
+      "grad_norm": 44.05482864379883,
+      "learning_rate": 1.3180444444444444e-05,
+      "loss": 1.6472,
+      "step": 25430
+    },
+    {
+      "epoch": 0.81408,
+      "grad_norm": 45.95872116088867,
+      "learning_rate": 1.317688888888889e-05,
+      "loss": 1.6422,
+      "step": 25440
+    },
+    {
+      "epoch": 0.8144,
+      "grad_norm": 44.911746978759766,
+      "learning_rate": 1.3173333333333333e-05,
+      "loss": 1.635,
+      "step": 25450
+    },
+    {
+      "epoch": 0.81472,
+      "grad_norm": 44.14010238647461,
+      "learning_rate": 1.316977777777778e-05,
+      "loss": 1.6578,
+      "step": 25460
+    },
+    {
+      "epoch": 0.81504,
+      "grad_norm": 41.602718353271484,
+      "learning_rate": 1.3166222222222222e-05,
+      "loss": 1.6693,
+      "step": 25470
+    },
+    {
+      "epoch": 0.81536,
+      "grad_norm": 45.07976150512695,
+      "learning_rate": 1.3162666666666668e-05,
+      "loss": 1.6623,
+      "step": 25480
+    },
+    {
+      "epoch": 0.81568,
+      "grad_norm": 43.92379379272461,
+      "learning_rate": 1.3159111111111111e-05,
+      "loss": 1.6421,
+      "step": 25490
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 44.13690948486328,
+      "learning_rate": 1.3155555555555558e-05,
+      "loss": 1.6493,
+      "step": 25500
+    },
+    {
+      "epoch": 0.81632,
+      "grad_norm": 47.86391067504883,
+      "learning_rate": 1.3152e-05,
+      "loss": 1.6477,
+      "step": 25510
+    },
+    {
+      "epoch": 0.81664,
+      "grad_norm": 43.6804084777832,
+      "learning_rate": 1.3148444444444447e-05,
+      "loss": 1.6467,
+      "step": 25520
+    },
+    {
+      "epoch": 0.81696,
+      "grad_norm": 46.26144790649414,
+      "learning_rate": 1.314488888888889e-05,
+      "loss": 1.6511,
+      "step": 25530
+    },
+    {
+      "epoch": 0.81728,
+      "grad_norm": 42.43048095703125,
+      "learning_rate": 1.3141333333333334e-05,
+      "loss": 1.6476,
+      "step": 25540
+    },
+    {
+      "epoch": 0.8176,
+      "grad_norm": 45.539634704589844,
+      "learning_rate": 1.3137777777777779e-05,
+      "loss": 1.6565,
+      "step": 25550
+    },
+    {
+      "epoch": 0.81792,
+      "grad_norm": 46.257110595703125,
+      "learning_rate": 1.3134222222222223e-05,
+      "loss": 1.6453,
+      "step": 25560
+    },
+    {
+      "epoch": 0.81824,
+      "grad_norm": 45.317142486572266,
+      "learning_rate": 1.3130666666666668e-05,
+      "loss": 1.6515,
+      "step": 25570
+    },
+    {
+      "epoch": 0.81856,
+      "grad_norm": 44.00784683227539,
+      "learning_rate": 1.3127111111111112e-05,
+      "loss": 1.6858,
+      "step": 25580
+    },
+    {
+      "epoch": 0.81888,
+      "grad_norm": 45.211429595947266,
+      "learning_rate": 1.3123555555555557e-05,
+      "loss": 1.6481,
+      "step": 25590
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 44.39881896972656,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 1.6445,
+      "step": 25600
+    },
+    {
+      "epoch": 0.81952,
+      "grad_norm": 43.63643264770508,
+      "learning_rate": 1.3116444444444444e-05,
+      "loss": 1.6423,
+      "step": 25610
+    },
+    {
+      "epoch": 0.81984,
+      "grad_norm": 42.108158111572266,
+      "learning_rate": 1.311288888888889e-05,
+      "loss": 1.6704,
+      "step": 25620
+    },
+    {
+      "epoch": 0.82016,
+      "grad_norm": 43.87276840209961,
+      "learning_rate": 1.3109333333333333e-05,
+      "loss": 1.6401,
+      "step": 25630
+    },
+    {
+      "epoch": 0.82048,
+      "grad_norm": 43.294090270996094,
+      "learning_rate": 1.310577777777778e-05,
+      "loss": 1.688,
+      "step": 25640
+    },
+    {
+      "epoch": 0.8208,
+      "grad_norm": 42.391029357910156,
+      "learning_rate": 1.3102222222222223e-05,
+      "loss": 1.6719,
+      "step": 25650
+    },
+    {
+      "epoch": 0.82112,
+      "grad_norm": 44.66448211669922,
+      "learning_rate": 1.3098666666666669e-05,
+      "loss": 1.6376,
+      "step": 25660
+    },
+    {
+      "epoch": 0.82144,
+      "grad_norm": 42.991539001464844,
+      "learning_rate": 1.3095111111111112e-05,
+      "loss": 1.6522,
+      "step": 25670
+    },
+    {
+      "epoch": 0.82176,
+      "grad_norm": 46.60944366455078,
+      "learning_rate": 1.3091555555555558e-05,
+      "loss": 1.6269,
+      "step": 25680
+    },
+    {
+      "epoch": 0.82208,
+      "grad_norm": 44.123817443847656,
+      "learning_rate": 1.3088e-05,
+      "loss": 1.6325,
+      "step": 25690
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 44.38522720336914,
+      "learning_rate": 1.3084444444444447e-05,
+      "loss": 1.6804,
+      "step": 25700
+    },
+    {
+      "epoch": 0.82272,
+      "grad_norm": 44.40520477294922,
+      "learning_rate": 1.308088888888889e-05,
+      "loss": 1.6397,
+      "step": 25710
+    },
+    {
+      "epoch": 0.82304,
+      "grad_norm": 45.01036834716797,
+      "learning_rate": 1.3077333333333334e-05,
+      "loss": 1.6629,
+      "step": 25720
+    },
+    {
+      "epoch": 0.82336,
+      "grad_norm": 43.612449645996094,
+      "learning_rate": 1.3073777777777779e-05,
+      "loss": 1.6594,
+      "step": 25730
+    },
+    {
+      "epoch": 0.82368,
+      "grad_norm": 46.03797912597656,
+      "learning_rate": 1.3070222222222223e-05,
+      "loss": 1.6797,
+      "step": 25740
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 42.819480895996094,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 1.6234,
+      "step": 25750
+    },
+    {
+      "epoch": 0.82432,
+      "grad_norm": 45.527652740478516,
+      "learning_rate": 1.3063111111111113e-05,
+      "loss": 1.6523,
+      "step": 25760
+    },
+    {
+      "epoch": 0.82464,
+      "grad_norm": 45.140602111816406,
+      "learning_rate": 1.3059555555555557e-05,
+      "loss": 1.6684,
+      "step": 25770
+    },
+    {
+      "epoch": 0.82496,
+      "grad_norm": 43.334163665771484,
+      "learning_rate": 1.3056000000000002e-05,
+      "loss": 1.6393,
+      "step": 25780
+    },
+    {
+      "epoch": 0.82528,
+      "grad_norm": 44.33065414428711,
+      "learning_rate": 1.3052444444444445e-05,
+      "loss": 1.6711,
+      "step": 25790
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 43.01241683959961,
+      "learning_rate": 1.304888888888889e-05,
+      "loss": 1.6545,
+      "step": 25800
+    },
+    {
+      "epoch": 0.82592,
+      "grad_norm": 45.725337982177734,
+      "learning_rate": 1.3045333333333334e-05,
+      "loss": 1.6609,
+      "step": 25810
+    },
+    {
+      "epoch": 0.82624,
+      "grad_norm": 42.712215423583984,
+      "learning_rate": 1.304177777777778e-05,
+      "loss": 1.6755,
+      "step": 25820
+    },
+    {
+      "epoch": 0.82656,
+      "grad_norm": 45.29679489135742,
+      "learning_rate": 1.3038222222222223e-05,
+      "loss": 1.6409,
+      "step": 25830
+    },
+    {
+      "epoch": 0.82688,
+      "grad_norm": 44.52570724487305,
+      "learning_rate": 1.3034666666666669e-05,
+      "loss": 1.6313,
+      "step": 25840
+    },
+    {
+      "epoch": 0.8272,
+      "grad_norm": 44.2238883972168,
+      "learning_rate": 1.3031111111111112e-05,
+      "loss": 1.6631,
+      "step": 25850
+    },
+    {
+      "epoch": 0.82752,
+      "grad_norm": 42.84789276123047,
+      "learning_rate": 1.3027555555555558e-05,
+      "loss": 1.6233,
+      "step": 25860
+    },
+    {
+      "epoch": 0.82784,
+      "grad_norm": 43.984310150146484,
+      "learning_rate": 1.3024000000000001e-05,
+      "loss": 1.6362,
+      "step": 25870
+    },
+    {
+      "epoch": 0.82816,
+      "grad_norm": 42.9167594909668,
+      "learning_rate": 1.3020444444444444e-05,
+      "loss": 1.6166,
+      "step": 25880
+    },
+    {
+      "epoch": 0.82848,
+      "grad_norm": 44.87565612792969,
+      "learning_rate": 1.301688888888889e-05,
+      "loss": 1.6503,
+      "step": 25890
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 45.01673889160156,
+      "learning_rate": 1.3013333333333333e-05,
+      "loss": 1.6494,
+      "step": 25900
+    },
+    {
+      "epoch": 0.82912,
+      "grad_norm": 43.40441131591797,
+      "learning_rate": 1.300977777777778e-05,
+      "loss": 1.673,
+      "step": 25910
+    },
+    {
+      "epoch": 0.82944,
+      "grad_norm": 45.419349670410156,
+      "learning_rate": 1.3006222222222222e-05,
+      "loss": 1.6564,
+      "step": 25920
+    },
+    {
+      "epoch": 0.82976,
+      "grad_norm": 44.7979621887207,
+      "learning_rate": 1.3002666666666668e-05,
+      "loss": 1.6677,
+      "step": 25930
+    },
+    {
+      "epoch": 0.83008,
+      "grad_norm": 45.44395446777344,
+      "learning_rate": 1.2999111111111111e-05,
+      "loss": 1.6384,
+      "step": 25940
+    },
+    {
+      "epoch": 0.8304,
+      "grad_norm": 44.90794372558594,
+      "learning_rate": 1.2995555555555557e-05,
+      "loss": 1.6625,
+      "step": 25950
+    },
+    {
+      "epoch": 0.83072,
+      "grad_norm": 44.52241897583008,
+      "learning_rate": 1.2992e-05,
+      "loss": 1.6542,
+      "step": 25960
+    },
+    {
+      "epoch": 0.83104,
+      "grad_norm": 44.95444107055664,
+      "learning_rate": 1.2988444444444447e-05,
+      "loss": 1.6471,
+      "step": 25970
+    },
+    {
+      "epoch": 0.83136,
+      "grad_norm": 45.035301208496094,
+      "learning_rate": 1.298488888888889e-05,
+      "loss": 1.6533,
+      "step": 25980
+    },
+    {
+      "epoch": 0.83168,
+      "grad_norm": 45.0323371887207,
+      "learning_rate": 1.2981333333333334e-05,
+      "loss": 1.6419,
+      "step": 25990
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 46.45552444458008,
+      "learning_rate": 1.2977777777777779e-05,
+      "loss": 1.6587,
+      "step": 26000
+    },
+    {
+      "epoch": 0.83232,
+      "grad_norm": 46.57162094116211,
+      "learning_rate": 1.2974222222222223e-05,
+      "loss": 1.6624,
+      "step": 26010
+    },
+    {
+      "epoch": 0.83264,
+      "grad_norm": 45.604042053222656,
+      "learning_rate": 1.2970666666666668e-05,
+      "loss": 1.6699,
+      "step": 26020
+    },
+    {
+      "epoch": 0.83296,
+      "grad_norm": 45.68147277832031,
+      "learning_rate": 1.2967111111111112e-05,
+      "loss": 1.6642,
+      "step": 26030
+    },
+    {
+      "epoch": 0.83328,
+      "grad_norm": 44.60567092895508,
+      "learning_rate": 1.2963555555555557e-05,
+      "loss": 1.6366,
+      "step": 26040
+    },
+    {
+      "epoch": 0.8336,
+      "grad_norm": 42.047977447509766,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 1.6321,
+      "step": 26050
+    },
+    {
+      "epoch": 0.83392,
+      "grad_norm": 43.56084060668945,
+      "learning_rate": 1.2956444444444444e-05,
+      "loss": 1.66,
+      "step": 26060
+    },
+    {
+      "epoch": 0.83424,
+      "grad_norm": 45.93843460083008,
+      "learning_rate": 1.295288888888889e-05,
+      "loss": 1.6546,
+      "step": 26070
+    },
+    {
+      "epoch": 0.83456,
+      "grad_norm": 43.66746139526367,
+      "learning_rate": 1.2949333333333333e-05,
+      "loss": 1.631,
+      "step": 26080
+    },
+    {
+      "epoch": 0.83488,
+      "grad_norm": 44.373802185058594,
+      "learning_rate": 1.294577777777778e-05,
+      "loss": 1.6222,
+      "step": 26090
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 43.290977478027344,
+      "learning_rate": 1.2942222222222222e-05,
+      "loss": 1.6463,
+      "step": 26100
+    },
+    {
+      "epoch": 0.83552,
+      "grad_norm": 42.16154861450195,
+      "learning_rate": 1.2938666666666669e-05,
+      "loss": 1.6314,
+      "step": 26110
+    },
+    {
+      "epoch": 0.83584,
+      "grad_norm": 45.15300369262695,
+      "learning_rate": 1.2935111111111111e-05,
+      "loss": 1.6622,
+      "step": 26120
+    },
+    {
+      "epoch": 0.83616,
+      "grad_norm": 45.746482849121094,
+      "learning_rate": 1.2931555555555558e-05,
+      "loss": 1.6552,
+      "step": 26130
+    },
+    {
+      "epoch": 0.83648,
+      "grad_norm": 44.604976654052734,
+      "learning_rate": 1.2928e-05,
+      "loss": 1.6379,
+      "step": 26140
+    },
+    {
+      "epoch": 0.8368,
+      "grad_norm": 45.78082275390625,
+      "learning_rate": 1.2924444444444447e-05,
+      "loss": 1.656,
+      "step": 26150
+    },
+    {
+      "epoch": 0.83712,
+      "grad_norm": 42.38752746582031,
+      "learning_rate": 1.292088888888889e-05,
+      "loss": 1.6713,
+      "step": 26160
+    },
+    {
+      "epoch": 0.83744,
+      "grad_norm": 41.761966705322266,
+      "learning_rate": 1.2917333333333334e-05,
+      "loss": 1.6581,
+      "step": 26170
+    },
+    {
+      "epoch": 0.83776,
+      "grad_norm": 44.03702926635742,
+      "learning_rate": 1.2913777777777779e-05,
+      "loss": 1.6545,
+      "step": 26180
+    },
+    {
+      "epoch": 0.83808,
+      "grad_norm": 42.7269287109375,
+      "learning_rate": 1.2910222222222223e-05,
+      "loss": 1.6277,
+      "step": 26190
+    },
+    {
+      "epoch": 0.8384,
+      "grad_norm": 45.2806282043457,
+      "learning_rate": 1.2906666666666668e-05,
+      "loss": 1.6329,
+      "step": 26200
+    },
+    {
+      "epoch": 0.83872,
+      "grad_norm": 44.448692321777344,
+      "learning_rate": 1.2903111111111112e-05,
+      "loss": 1.6385,
+      "step": 26210
+    },
+    {
+      "epoch": 0.83904,
+      "grad_norm": 43.17620849609375,
+      "learning_rate": 1.2899555555555557e-05,
+      "loss": 1.6309,
+      "step": 26220
+    },
+    {
+      "epoch": 0.83936,
+      "grad_norm": 43.542423248291016,
+      "learning_rate": 1.2896000000000002e-05,
+      "loss": 1.6282,
+      "step": 26230
+    },
+    {
+      "epoch": 0.83968,
+      "grad_norm": 46.048954010009766,
+      "learning_rate": 1.2892444444444444e-05,
+      "loss": 1.6884,
+      "step": 26240
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 43.61574935913086,
+      "learning_rate": 1.288888888888889e-05,
+      "loss": 1.6829,
+      "step": 26250
+    },
+    {
+      "epoch": 0.84032,
+      "grad_norm": 42.595706939697266,
+      "learning_rate": 1.2885333333333334e-05,
+      "loss": 1.6519,
+      "step": 26260
+    },
+    {
+      "epoch": 0.84064,
+      "grad_norm": 42.97865676879883,
+      "learning_rate": 1.288177777777778e-05,
+      "loss": 1.6544,
+      "step": 26270
+    },
+    {
+      "epoch": 0.84096,
+      "grad_norm": 47.38331604003906,
+      "learning_rate": 1.2878222222222223e-05,
+      "loss": 1.6439,
+      "step": 26280
+    },
+    {
+      "epoch": 0.84128,
+      "grad_norm": 45.50075149536133,
+      "learning_rate": 1.2874666666666669e-05,
+      "loss": 1.6359,
+      "step": 26290
+    },
+    {
+      "epoch": 0.8416,
+      "grad_norm": 44.4566764831543,
+      "learning_rate": 1.2871111111111112e-05,
+      "loss": 1.6667,
+      "step": 26300
+    },
+    {
+      "epoch": 0.84192,
+      "grad_norm": 43.335777282714844,
+      "learning_rate": 1.2867555555555558e-05,
+      "loss": 1.6672,
+      "step": 26310
+    },
+    {
+      "epoch": 0.84224,
+      "grad_norm": 47.7451171875,
+      "learning_rate": 1.2864e-05,
+      "loss": 1.6576,
+      "step": 26320
+    },
+    {
+      "epoch": 0.84256,
+      "grad_norm": 46.96812057495117,
+      "learning_rate": 1.2860444444444447e-05,
+      "loss": 1.6441,
+      "step": 26330
+    },
+    {
+      "epoch": 0.84288,
+      "grad_norm": 45.21785354614258,
+      "learning_rate": 1.285688888888889e-05,
+      "loss": 1.6437,
+      "step": 26340
+    },
+    {
+      "epoch": 0.8432,
+      "grad_norm": 41.97297668457031,
+      "learning_rate": 1.2853333333333336e-05,
+      "loss": 1.6436,
+      "step": 26350
+    },
+    {
+      "epoch": 0.84352,
+      "grad_norm": 44.169273376464844,
+      "learning_rate": 1.2849777777777779e-05,
+      "loss": 1.6387,
+      "step": 26360
+    },
+    {
+      "epoch": 0.84384,
+      "grad_norm": 45.00596618652344,
+      "learning_rate": 1.2846222222222224e-05,
+      "loss": 1.6272,
+      "step": 26370
+    },
+    {
+      "epoch": 0.84416,
+      "grad_norm": 45.01812744140625,
+      "learning_rate": 1.2842666666666668e-05,
+      "loss": 1.6506,
+      "step": 26380
+    },
+    {
+      "epoch": 0.84448,
+      "grad_norm": 44.44124984741211,
+      "learning_rate": 1.2839111111111111e-05,
+      "loss": 1.6627,
+      "step": 26390
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 45.47779846191406,
+      "learning_rate": 1.2835555555555557e-05,
+      "loss": 1.6626,
+      "step": 26400
+    },
+    {
+      "epoch": 0.84512,
+      "grad_norm": 46.43983459472656,
+      "learning_rate": 1.2832e-05,
+      "loss": 1.6717,
+      "step": 26410
+    },
+    {
+      "epoch": 0.84544,
+      "grad_norm": 45.822357177734375,
+      "learning_rate": 1.2828444444444446e-05,
+      "loss": 1.6639,
+      "step": 26420
+    },
+    {
+      "epoch": 0.84576,
+      "grad_norm": 45.2042121887207,
+      "learning_rate": 1.282488888888889e-05,
+      "loss": 1.6461,
+      "step": 26430
+    },
+    {
+      "epoch": 0.84608,
+      "grad_norm": 44.78562927246094,
+      "learning_rate": 1.2821333333333334e-05,
+      "loss": 1.6853,
+      "step": 26440
+    },
+    {
+      "epoch": 0.8464,
+      "grad_norm": 43.18932342529297,
+      "learning_rate": 1.2817777777777778e-05,
+      "loss": 1.6753,
+      "step": 26450
+    },
+    {
+      "epoch": 0.84672,
+      "grad_norm": 43.83230209350586,
+      "learning_rate": 1.2814222222222223e-05,
+      "loss": 1.6618,
+      "step": 26460
+    },
+    {
+      "epoch": 0.84704,
+      "grad_norm": 45.577362060546875,
+      "learning_rate": 1.2810666666666667e-05,
+      "loss": 1.6194,
+      "step": 26470
+    },
+    {
+      "epoch": 0.84736,
+      "grad_norm": 46.04071807861328,
+      "learning_rate": 1.2807111111111112e-05,
+      "loss": 1.6443,
+      "step": 26480
+    },
+    {
+      "epoch": 0.84768,
+      "grad_norm": 44.70589828491211,
+      "learning_rate": 1.2803555555555557e-05,
+      "loss": 1.6251,
+      "step": 26490
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 44.009803771972656,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.6241,
+      "step": 26500
+    },
+    {
+      "epoch": 0.84832,
+      "grad_norm": 43.644805908203125,
+      "learning_rate": 1.2796444444444444e-05,
+      "loss": 1.6336,
+      "step": 26510
+    },
+    {
+      "epoch": 0.84864,
+      "grad_norm": 45.69171142578125,
+      "learning_rate": 1.279288888888889e-05,
+      "loss": 1.6607,
+      "step": 26520
+    },
+    {
+      "epoch": 0.84896,
+      "grad_norm": 43.19560241699219,
+      "learning_rate": 1.2789333333333333e-05,
+      "loss": 1.6603,
+      "step": 26530
+    },
+    {
+      "epoch": 0.84928,
+      "grad_norm": 44.87495422363281,
+      "learning_rate": 1.278577777777778e-05,
+      "loss": 1.6498,
+      "step": 26540
+    },
+    {
+      "epoch": 0.8496,
+      "grad_norm": 42.37331008911133,
+      "learning_rate": 1.2782222222222222e-05,
+      "loss": 1.6671,
+      "step": 26550
+    },
+    {
+      "epoch": 0.84992,
+      "grad_norm": 43.74976348876953,
+      "learning_rate": 1.2778666666666668e-05,
+      "loss": 1.6441,
+      "step": 26560
+    },
+    {
+      "epoch": 0.85024,
+      "grad_norm": 42.69112777709961,
+      "learning_rate": 1.2775111111111111e-05,
+      "loss": 1.6505,
+      "step": 26570
+    },
+    {
+      "epoch": 0.85056,
+      "grad_norm": 43.077308654785156,
+      "learning_rate": 1.2771555555555558e-05,
+      "loss": 1.6585,
+      "step": 26580
+    },
+    {
+      "epoch": 0.85088,
+      "grad_norm": 44.00758361816406,
+      "learning_rate": 1.2768e-05,
+      "loss": 1.6302,
+      "step": 26590
+    },
+    {
+      "epoch": 0.8512,
+      "grad_norm": 45.46588134765625,
+      "learning_rate": 1.2764444444444447e-05,
+      "loss": 1.6416,
+      "step": 26600
+    },
+    {
+      "epoch": 0.85152,
+      "grad_norm": 44.98797607421875,
+      "learning_rate": 1.276088888888889e-05,
+      "loss": 1.667,
+      "step": 26610
+    },
+    {
+      "epoch": 0.85184,
+      "grad_norm": 46.968017578125,
+      "learning_rate": 1.2757333333333334e-05,
+      "loss": 1.6459,
+      "step": 26620
+    },
+    {
+      "epoch": 0.85216,
+      "grad_norm": 43.3302001953125,
+      "learning_rate": 1.2753777777777779e-05,
+      "loss": 1.6834,
+      "step": 26630
+    },
+    {
+      "epoch": 0.85248,
+      "grad_norm": 45.77449417114258,
+      "learning_rate": 1.2750222222222223e-05,
+      "loss": 1.673,
+      "step": 26640
+    },
+    {
+      "epoch": 0.8528,
+      "grad_norm": 44.07681655883789,
+      "learning_rate": 1.2746666666666668e-05,
+      "loss": 1.6479,
+      "step": 26650
+    },
+    {
+      "epoch": 0.85312,
+      "grad_norm": 44.1185188293457,
+      "learning_rate": 1.2743111111111112e-05,
+      "loss": 1.6679,
+      "step": 26660
+    },
+    {
+      "epoch": 0.85344,
+      "grad_norm": 45.55058670043945,
+      "learning_rate": 1.2739555555555557e-05,
+      "loss": 1.6729,
+      "step": 26670
+    },
+    {
+      "epoch": 0.85376,
+      "grad_norm": 44.186737060546875,
+      "learning_rate": 1.2736000000000001e-05,
+      "loss": 1.659,
+      "step": 26680
+    },
+    {
+      "epoch": 0.85408,
+      "grad_norm": 43.83502197265625,
+      "learning_rate": 1.2732444444444444e-05,
+      "loss": 1.668,
+      "step": 26690
+    },
+    {
+      "epoch": 0.8544,
+      "grad_norm": 44.22084045410156,
+      "learning_rate": 1.272888888888889e-05,
+      "loss": 1.6527,
+      "step": 26700
+    },
+    {
+      "epoch": 0.85472,
+      "grad_norm": 46.551456451416016,
+      "learning_rate": 1.2725333333333333e-05,
+      "loss": 1.6546,
+      "step": 26710
+    },
+    {
+      "epoch": 0.85504,
+      "grad_norm": 45.00933074951172,
+      "learning_rate": 1.272177777777778e-05,
+      "loss": 1.624,
+      "step": 26720
+    },
+    {
+      "epoch": 0.85536,
+      "grad_norm": 43.585227966308594,
+      "learning_rate": 1.2718222222222222e-05,
+      "loss": 1.6377,
+      "step": 26730
+    },
+    {
+      "epoch": 0.85568,
+      "grad_norm": 44.3875846862793,
+      "learning_rate": 1.2714666666666669e-05,
+      "loss": 1.6469,
+      "step": 26740
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 42.42379379272461,
+      "learning_rate": 1.2711111111111112e-05,
+      "loss": 1.628,
+      "step": 26750
+    },
+    {
+      "epoch": 0.85632,
+      "grad_norm": 43.73078918457031,
+      "learning_rate": 1.2707555555555558e-05,
+      "loss": 1.6491,
+      "step": 26760
+    },
+    {
+      "epoch": 0.85664,
+      "grad_norm": 42.43165969848633,
+      "learning_rate": 1.2704e-05,
+      "loss": 1.6496,
+      "step": 26770
+    },
+    {
+      "epoch": 0.85696,
+      "grad_norm": 45.21994400024414,
+      "learning_rate": 1.2700444444444447e-05,
+      "loss": 1.662,
+      "step": 26780
+    },
+    {
+      "epoch": 0.85728,
+      "grad_norm": 43.44044876098633,
+      "learning_rate": 1.269688888888889e-05,
+      "loss": 1.6649,
+      "step": 26790
+    },
+    {
+      "epoch": 0.8576,
+      "grad_norm": 45.463809967041016,
+      "learning_rate": 1.2693333333333336e-05,
+      "loss": 1.6547,
+      "step": 26800
+    },
+    {
+      "epoch": 0.85792,
+      "grad_norm": 47.156829833984375,
+      "learning_rate": 1.2689777777777779e-05,
+      "loss": 1.6499,
+      "step": 26810
+    },
+    {
+      "epoch": 0.85824,
+      "grad_norm": 44.708038330078125,
+      "learning_rate": 1.2686222222222223e-05,
+      "loss": 1.6574,
+      "step": 26820
+    },
+    {
+      "epoch": 0.85856,
+      "grad_norm": 45.35110855102539,
+      "learning_rate": 1.2682666666666668e-05,
+      "loss": 1.6368,
+      "step": 26830
+    },
+    {
+      "epoch": 0.85888,
+      "grad_norm": 43.904296875,
+      "learning_rate": 1.2679111111111113e-05,
+      "loss": 1.6353,
+      "step": 26840
+    },
+    {
+      "epoch": 0.8592,
+      "grad_norm": 46.68516159057617,
+      "learning_rate": 1.2675555555555557e-05,
+      "loss": 1.6486,
+      "step": 26850
+    },
+    {
+      "epoch": 0.85952,
+      "grad_norm": 44.25490951538086,
+      "learning_rate": 1.2672000000000002e-05,
+      "loss": 1.6474,
+      "step": 26860
+    },
+    {
+      "epoch": 0.85984,
+      "grad_norm": 43.81814193725586,
+      "learning_rate": 1.2668444444444446e-05,
+      "loss": 1.6599,
+      "step": 26870
+    },
+    {
+      "epoch": 0.86016,
+      "grad_norm": 44.363643646240234,
+      "learning_rate": 1.266488888888889e-05,
+      "loss": 1.6494,
+      "step": 26880
+    },
+    {
+      "epoch": 0.86048,
+      "grad_norm": 45.49546432495117,
+      "learning_rate": 1.2661333333333334e-05,
+      "loss": 1.6594,
+      "step": 26890
+    },
+    {
+      "epoch": 0.8608,
+      "grad_norm": 41.69033432006836,
+      "learning_rate": 1.2657777777777778e-05,
+      "loss": 1.6266,
+      "step": 26900
+    },
+    {
+      "epoch": 0.86112,
+      "grad_norm": 43.465553283691406,
+      "learning_rate": 1.2654222222222223e-05,
+      "loss": 1.6648,
+      "step": 26910
+    },
+    {
+      "epoch": 0.86144,
+      "grad_norm": 44.84959030151367,
+      "learning_rate": 1.2650666666666667e-05,
+      "loss": 1.6632,
+      "step": 26920
+    },
+    {
+      "epoch": 0.86176,
+      "grad_norm": 43.108482360839844,
+      "learning_rate": 1.2647111111111112e-05,
+      "loss": 1.668,
+      "step": 26930
+    },
+    {
+      "epoch": 0.86208,
+      "grad_norm": 42.777320861816406,
+      "learning_rate": 1.2643555555555556e-05,
+      "loss": 1.656,
+      "step": 26940
+    },
+    {
+      "epoch": 0.8624,
+      "grad_norm": 44.51447296142578,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 1.6395,
+      "step": 26950
+    },
+    {
+      "epoch": 0.86272,
+      "grad_norm": 47.7524299621582,
+      "learning_rate": 1.2636444444444444e-05,
+      "loss": 1.6489,
+      "step": 26960
+    },
+    {
+      "epoch": 0.86304,
+      "grad_norm": 42.68575668334961,
+      "learning_rate": 1.263288888888889e-05,
+      "loss": 1.6581,
+      "step": 26970
+    },
+    {
+      "epoch": 0.86336,
+      "grad_norm": 44.79243087768555,
+      "learning_rate": 1.2629333333333333e-05,
+      "loss": 1.6275,
+      "step": 26980
+    },
+    {
+      "epoch": 0.86368,
+      "grad_norm": 45.371253967285156,
+      "learning_rate": 1.262577777777778e-05,
+      "loss": 1.6663,
+      "step": 26990
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 45.898128509521484,
+      "learning_rate": 1.2622222222222222e-05,
+      "loss": 1.6425,
+      "step": 27000
+    },
+    {
+      "epoch": 0.86432,
+      "grad_norm": 42.76716613769531,
+      "learning_rate": 1.2618666666666668e-05,
+      "loss": 1.6436,
+      "step": 27010
+    },
+    {
+      "epoch": 0.86464,
+      "grad_norm": 45.81767272949219,
+      "learning_rate": 1.2615111111111111e-05,
+      "loss": 1.6599,
+      "step": 27020
+    },
+    {
+      "epoch": 0.86496,
+      "grad_norm": 41.93647384643555,
+      "learning_rate": 1.2611555555555557e-05,
+      "loss": 1.6526,
+      "step": 27030
+    },
+    {
+      "epoch": 0.86528,
+      "grad_norm": 43.41840744018555,
+      "learning_rate": 1.2608e-05,
+      "loss": 1.6721,
+      "step": 27040
+    },
+    {
+      "epoch": 0.8656,
+      "grad_norm": 44.652591705322266,
+      "learning_rate": 1.2604444444444446e-05,
+      "loss": 1.6321,
+      "step": 27050
+    },
+    {
+      "epoch": 0.86592,
+      "grad_norm": 43.17992401123047,
+      "learning_rate": 1.260088888888889e-05,
+      "loss": 1.659,
+      "step": 27060
+    },
+    {
+      "epoch": 0.86624,
+      "grad_norm": 45.41444778442383,
+      "learning_rate": 1.2597333333333334e-05,
+      "loss": 1.6524,
+      "step": 27070
+    },
+    {
+      "epoch": 0.86656,
+      "grad_norm": 44.880409240722656,
+      "learning_rate": 1.2593777777777778e-05,
+      "loss": 1.6639,
+      "step": 27080
+    },
+    {
+      "epoch": 0.86688,
+      "grad_norm": 42.21232986450195,
+      "learning_rate": 1.2590222222222223e-05,
+      "loss": 1.6627,
+      "step": 27090
+    },
+    {
+      "epoch": 0.8672,
+      "grad_norm": 46.4969482421875,
+      "learning_rate": 1.2586666666666668e-05,
+      "loss": 1.6546,
+      "step": 27100
+    },
+    {
+      "epoch": 0.86752,
+      "grad_norm": 42.779335021972656,
+      "learning_rate": 1.2583111111111112e-05,
+      "loss": 1.6112,
+      "step": 27110
+    },
+    {
+      "epoch": 0.86784,
+      "grad_norm": 44.317108154296875,
+      "learning_rate": 1.2579555555555557e-05,
+      "loss": 1.6417,
+      "step": 27120
+    },
+    {
+      "epoch": 0.86816,
+      "grad_norm": 44.933109283447266,
+      "learning_rate": 1.2576000000000001e-05,
+      "loss": 1.6624,
+      "step": 27130
+    },
+    {
+      "epoch": 0.86848,
+      "grad_norm": 43.719783782958984,
+      "learning_rate": 1.2572444444444444e-05,
+      "loss": 1.6601,
+      "step": 27140
+    },
+    {
+      "epoch": 0.8688,
+      "grad_norm": 44.277103424072266,
+      "learning_rate": 1.256888888888889e-05,
+      "loss": 1.6492,
+      "step": 27150
+    },
+    {
+      "epoch": 0.86912,
+      "grad_norm": 43.31966781616211,
+      "learning_rate": 1.2565333333333333e-05,
+      "loss": 1.626,
+      "step": 27160
+    },
+    {
+      "epoch": 0.86944,
+      "grad_norm": 43.45155715942383,
+      "learning_rate": 1.256177777777778e-05,
+      "loss": 1.6536,
+      "step": 27170
+    },
+    {
+      "epoch": 0.86976,
+      "grad_norm": 43.08216094970703,
+      "learning_rate": 1.2558222222222222e-05,
+      "loss": 1.6415,
+      "step": 27180
+    },
+    {
+      "epoch": 0.87008,
+      "grad_norm": 44.816810607910156,
+      "learning_rate": 1.2554666666666669e-05,
+      "loss": 1.6711,
+      "step": 27190
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 44.163211822509766,
+      "learning_rate": 1.2551111111111111e-05,
+      "loss": 1.6593,
+      "step": 27200
+    },
+    {
+      "epoch": 0.87072,
+      "grad_norm": 47.842308044433594,
+      "learning_rate": 1.2547555555555558e-05,
+      "loss": 1.6535,
+      "step": 27210
+    },
+    {
+      "epoch": 0.87104,
+      "grad_norm": 43.351219177246094,
+      "learning_rate": 1.2544e-05,
+      "loss": 1.6299,
+      "step": 27220
+    },
+    {
+      "epoch": 0.87136,
+      "grad_norm": 44.729652404785156,
+      "learning_rate": 1.2540444444444447e-05,
+      "loss": 1.658,
+      "step": 27230
+    },
+    {
+      "epoch": 0.87168,
+      "grad_norm": 42.45663070678711,
+      "learning_rate": 1.253688888888889e-05,
+      "loss": 1.6354,
+      "step": 27240
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 44.611331939697266,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 1.6364,
+      "step": 27250
+    },
+    {
+      "epoch": 0.87232,
+      "grad_norm": 43.72536849975586,
+      "learning_rate": 1.2529777777777779e-05,
+      "loss": 1.6597,
+      "step": 27260
+    },
+    {
+      "epoch": 0.87264,
+      "grad_norm": 44.238258361816406,
+      "learning_rate": 1.2526222222222223e-05,
+      "loss": 1.6638,
+      "step": 27270
+    },
+    {
+      "epoch": 0.87296,
+      "grad_norm": 44.001956939697266,
+      "learning_rate": 1.2522666666666668e-05,
+      "loss": 1.6426,
+      "step": 27280
+    },
+    {
+      "epoch": 0.87328,
+      "grad_norm": 49.65640640258789,
+      "learning_rate": 1.2519111111111112e-05,
+      "loss": 1.6433,
+      "step": 27290
+    },
+    {
+      "epoch": 0.8736,
+      "grad_norm": 46.28944778442383,
+      "learning_rate": 1.2515555555555557e-05,
+      "loss": 1.662,
+      "step": 27300
+    },
+    {
+      "epoch": 0.87392,
+      "grad_norm": 44.71233367919922,
+      "learning_rate": 1.2512000000000002e-05,
+      "loss": 1.658,
+      "step": 27310
+    },
+    {
+      "epoch": 0.87424,
+      "grad_norm": 43.2108039855957,
+      "learning_rate": 1.2508444444444446e-05,
+      "loss": 1.6548,
+      "step": 27320
+    },
+    {
+      "epoch": 0.87456,
+      "grad_norm": 42.10308837890625,
+      "learning_rate": 1.250488888888889e-05,
+      "loss": 1.6557,
+      "step": 27330
+    },
+    {
+      "epoch": 0.87488,
+      "grad_norm": 46.05120849609375,
+      "learning_rate": 1.2501333333333333e-05,
+      "loss": 1.6751,
+      "step": 27340
+    },
+    {
+      "epoch": 0.8752,
+      "grad_norm": 42.84408950805664,
+      "learning_rate": 1.249777777777778e-05,
+      "loss": 1.6356,
+      "step": 27350
+    },
+    {
+      "epoch": 0.87552,
+      "grad_norm": 45.129425048828125,
+      "learning_rate": 1.2494222222222223e-05,
+      "loss": 1.6177,
+      "step": 27360
+    },
+    {
+      "epoch": 0.87584,
+      "grad_norm": 46.16420364379883,
+      "learning_rate": 1.2490666666666669e-05,
+      "loss": 1.6488,
+      "step": 27370
+    },
+    {
+      "epoch": 0.87616,
+      "grad_norm": 43.65485763549805,
+      "learning_rate": 1.2487111111111112e-05,
+      "loss": 1.6263,
+      "step": 27380
+    },
+    {
+      "epoch": 0.87648,
+      "grad_norm": 42.65005111694336,
+      "learning_rate": 1.2483555555555558e-05,
+      "loss": 1.6233,
+      "step": 27390
+    },
+    {
+      "epoch": 0.8768,
+      "grad_norm": 43.636959075927734,
+      "learning_rate": 1.248e-05,
+      "loss": 1.6207,
+      "step": 27400
+    },
+    {
+      "epoch": 0.87712,
+      "grad_norm": 43.3430290222168,
+      "learning_rate": 1.2476444444444444e-05,
+      "loss": 1.6531,
+      "step": 27410
+    },
+    {
+      "epoch": 0.87744,
+      "grad_norm": 44.20934295654297,
+      "learning_rate": 1.247288888888889e-05,
+      "loss": 1.6367,
+      "step": 27420
+    },
+    {
+      "epoch": 0.87776,
+      "grad_norm": 45.244789123535156,
+      "learning_rate": 1.2469333333333333e-05,
+      "loss": 1.6685,
+      "step": 27430
+    },
+    {
+      "epoch": 0.87808,
+      "grad_norm": 44.35482406616211,
+      "learning_rate": 1.2465777777777779e-05,
+      "loss": 1.645,
+      "step": 27440
+    },
+    {
+      "epoch": 0.8784,
+      "grad_norm": 46.52604675292969,
+      "learning_rate": 1.2462222222222222e-05,
+      "loss": 1.6398,
+      "step": 27450
+    },
+    {
+      "epoch": 0.87872,
+      "grad_norm": 44.6185302734375,
+      "learning_rate": 1.2458666666666668e-05,
+      "loss": 1.6597,
+      "step": 27460
+    },
+    {
+      "epoch": 0.87904,
+      "grad_norm": 42.396846771240234,
+      "learning_rate": 1.2455111111111111e-05,
+      "loss": 1.6788,
+      "step": 27470
+    },
+    {
+      "epoch": 0.87936,
+      "grad_norm": 45.505287170410156,
+      "learning_rate": 1.2451555555555557e-05,
+      "loss": 1.618,
+      "step": 27480
+    },
+    {
+      "epoch": 0.87968,
+      "grad_norm": 43.372291564941406,
+      "learning_rate": 1.2448e-05,
+      "loss": 1.6508,
+      "step": 27490
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 42.66139221191406,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 1.6607,
+      "step": 27500
+    },
+    {
+      "epoch": 0.88032,
+      "grad_norm": 44.73249435424805,
+      "learning_rate": 1.244088888888889e-05,
+      "loss": 1.6458,
+      "step": 27510
+    },
+    {
+      "epoch": 0.88064,
+      "grad_norm": 44.99702072143555,
+      "learning_rate": 1.2437333333333335e-05,
+      "loss": 1.6289,
+      "step": 27520
+    },
+    {
+      "epoch": 0.88096,
+      "grad_norm": 42.213645935058594,
+      "learning_rate": 1.2433777777777778e-05,
+      "loss": 1.6171,
+      "step": 27530
+    },
+    {
+      "epoch": 0.88128,
+      "grad_norm": 44.95713806152344,
+      "learning_rate": 1.2430222222222223e-05,
+      "loss": 1.6766,
+      "step": 27540
+    },
+    {
+      "epoch": 0.8816,
+      "grad_norm": 44.07640075683594,
+      "learning_rate": 1.2426666666666667e-05,
+      "loss": 1.6444,
+      "step": 27550
+    },
+    {
+      "epoch": 0.88192,
+      "grad_norm": 42.04390335083008,
+      "learning_rate": 1.2423111111111112e-05,
+      "loss": 1.6551,
+      "step": 27560
+    },
+    {
+      "epoch": 0.88224,
+      "grad_norm": 44.98395919799805,
+      "learning_rate": 1.2419555555555557e-05,
+      "loss": 1.6356,
+      "step": 27570
+    },
+    {
+      "epoch": 0.88256,
+      "grad_norm": 46.020111083984375,
+      "learning_rate": 1.2416000000000001e-05,
+      "loss": 1.6309,
+      "step": 27580
+    },
+    {
+      "epoch": 0.88288,
+      "grad_norm": 42.20390701293945,
+      "learning_rate": 1.2412444444444444e-05,
+      "loss": 1.6362,
+      "step": 27590
+    },
+    {
+      "epoch": 0.8832,
+      "grad_norm": 44.44026565551758,
+      "learning_rate": 1.240888888888889e-05,
+      "loss": 1.6688,
+      "step": 27600
+    },
+    {
+      "epoch": 0.88352,
+      "grad_norm": 45.78468322753906,
+      "learning_rate": 1.2405333333333333e-05,
+      "loss": 1.6402,
+      "step": 27610
+    },
+    {
+      "epoch": 0.88384,
+      "grad_norm": 43.677391052246094,
+      "learning_rate": 1.240177777777778e-05,
+      "loss": 1.6265,
+      "step": 27620
+    },
+    {
+      "epoch": 0.88416,
+      "grad_norm": 43.043800354003906,
+      "learning_rate": 1.2398222222222222e-05,
+      "loss": 1.6182,
+      "step": 27630
+    },
+    {
+      "epoch": 0.88448,
+      "grad_norm": 44.14324951171875,
+      "learning_rate": 1.2394666666666668e-05,
+      "loss": 1.6267,
+      "step": 27640
+    },
+    {
+      "epoch": 0.8848,
+      "grad_norm": 44.18344497680664,
+      "learning_rate": 1.2391111111111111e-05,
+      "loss": 1.6391,
+      "step": 27650
+    },
+    {
+      "epoch": 0.88512,
+      "grad_norm": 43.16364288330078,
+      "learning_rate": 1.2387555555555557e-05,
+      "loss": 1.6655,
+      "step": 27660
+    },
+    {
+      "epoch": 0.88544,
+      "grad_norm": 46.12438201904297,
+      "learning_rate": 1.2384e-05,
+      "loss": 1.6428,
+      "step": 27670
+    },
+    {
+      "epoch": 0.88576,
+      "grad_norm": 45.85373306274414,
+      "learning_rate": 1.2380444444444447e-05,
+      "loss": 1.6736,
+      "step": 27680
+    },
+    {
+      "epoch": 0.88608,
+      "grad_norm": 41.845664978027344,
+      "learning_rate": 1.237688888888889e-05,
+      "loss": 1.6536,
+      "step": 27690
+    },
+    {
+      "epoch": 0.8864,
+      "grad_norm": 43.84564208984375,
+      "learning_rate": 1.2373333333333336e-05,
+      "loss": 1.632,
+      "step": 27700
+    },
+    {
+      "epoch": 0.88672,
+      "grad_norm": 45.331031799316406,
+      "learning_rate": 1.2369777777777779e-05,
+      "loss": 1.6414,
+      "step": 27710
+    },
+    {
+      "epoch": 0.88704,
+      "grad_norm": 46.10980987548828,
+      "learning_rate": 1.2366222222222223e-05,
+      "loss": 1.6285,
+      "step": 27720
+    },
+    {
+      "epoch": 0.88736,
+      "grad_norm": 43.144920349121094,
+      "learning_rate": 1.2362666666666668e-05,
+      "loss": 1.6452,
+      "step": 27730
+    },
+    {
+      "epoch": 0.88768,
+      "grad_norm": 45.01664352416992,
+      "learning_rate": 1.2359111111111112e-05,
+      "loss": 1.652,
+      "step": 27740
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 44.60953140258789,
+      "learning_rate": 1.2355555555555557e-05,
+      "loss": 1.6461,
+      "step": 27750
+    },
+    {
+      "epoch": 0.88832,
+      "grad_norm": 43.47231674194336,
+      "learning_rate": 1.2352000000000001e-05,
+      "loss": 1.6385,
+      "step": 27760
+    },
+    {
+      "epoch": 0.88864,
+      "grad_norm": 43.14651107788086,
+      "learning_rate": 1.2348444444444446e-05,
+      "loss": 1.6667,
+      "step": 27770
+    },
+    {
+      "epoch": 0.88896,
+      "grad_norm": 44.596031188964844,
+      "learning_rate": 1.234488888888889e-05,
+      "loss": 1.6606,
+      "step": 27780
+    },
+    {
+      "epoch": 0.88928,
+      "grad_norm": 43.93633270263672,
+      "learning_rate": 1.2341333333333333e-05,
+      "loss": 1.6703,
+      "step": 27790
+    },
+    {
+      "epoch": 0.8896,
+      "grad_norm": 45.12252426147461,
+      "learning_rate": 1.233777777777778e-05,
+      "loss": 1.6845,
+      "step": 27800
+    },
+    {
+      "epoch": 0.88992,
+      "grad_norm": 42.742916107177734,
+      "learning_rate": 1.2334222222222222e-05,
+      "loss": 1.6519,
+      "step": 27810
+    },
+    {
+      "epoch": 0.89024,
+      "grad_norm": 42.931983947753906,
+      "learning_rate": 1.2330666666666669e-05,
+      "loss": 1.6157,
+      "step": 27820
+    },
+    {
+      "epoch": 0.89056,
+      "grad_norm": 45.96379852294922,
+      "learning_rate": 1.2327111111111112e-05,
+      "loss": 1.659,
+      "step": 27830
+    },
+    {
+      "epoch": 0.89088,
+      "grad_norm": 43.690372467041016,
+      "learning_rate": 1.2323555555555558e-05,
+      "loss": 1.6571,
+      "step": 27840
+    },
+    {
+      "epoch": 0.8912,
+      "grad_norm": 43.9526252746582,
+      "learning_rate": 1.232e-05,
+      "loss": 1.6165,
+      "step": 27850
+    },
+    {
+      "epoch": 0.89152,
+      "grad_norm": 44.779090881347656,
+      "learning_rate": 1.2316444444444447e-05,
+      "loss": 1.6393,
+      "step": 27860
+    },
+    {
+      "epoch": 0.89184,
+      "grad_norm": 44.47468185424805,
+      "learning_rate": 1.231288888888889e-05,
+      "loss": 1.6405,
+      "step": 27870
+    },
+    {
+      "epoch": 0.89216,
+      "grad_norm": 43.439083099365234,
+      "learning_rate": 1.2309333333333336e-05,
+      "loss": 1.6563,
+      "step": 27880
+    },
+    {
+      "epoch": 0.89248,
+      "grad_norm": 43.92070770263672,
+      "learning_rate": 1.2305777777777779e-05,
+      "loss": 1.6624,
+      "step": 27890
+    },
+    {
+      "epoch": 0.8928,
+      "grad_norm": 44.40715408325195,
+      "learning_rate": 1.2302222222222225e-05,
+      "loss": 1.6531,
+      "step": 27900
+    },
+    {
+      "epoch": 0.89312,
+      "grad_norm": 44.94832229614258,
+      "learning_rate": 1.2298666666666668e-05,
+      "loss": 1.6473,
+      "step": 27910
+    },
+    {
+      "epoch": 0.89344,
+      "grad_norm": 44.329341888427734,
+      "learning_rate": 1.2295111111111113e-05,
+      "loss": 1.6501,
+      "step": 27920
+    },
+    {
+      "epoch": 0.89376,
+      "grad_norm": 44.398345947265625,
+      "learning_rate": 1.2291555555555557e-05,
+      "loss": 1.6479,
+      "step": 27930
+    },
+    {
+      "epoch": 0.89408,
+      "grad_norm": 42.575931549072266,
+      "learning_rate": 1.2288e-05,
+      "loss": 1.6516,
+      "step": 27940
+    },
+    {
+      "epoch": 0.8944,
+      "grad_norm": 46.583187103271484,
+      "learning_rate": 1.2284444444444446e-05,
+      "loss": 1.6903,
+      "step": 27950
+    },
+    {
+      "epoch": 0.89472,
+      "grad_norm": 44.53683853149414,
+      "learning_rate": 1.2280888888888889e-05,
+      "loss": 1.6201,
+      "step": 27960
+    },
+    {
+      "epoch": 0.89504,
+      "grad_norm": 47.129783630371094,
+      "learning_rate": 1.2277333333333335e-05,
+      "loss": 1.6381,
+      "step": 27970
+    },
+    {
+      "epoch": 0.89536,
+      "grad_norm": 44.966064453125,
+      "learning_rate": 1.2273777777777778e-05,
+      "loss": 1.6346,
+      "step": 27980
+    },
+    {
+      "epoch": 0.89568,
+      "grad_norm": 43.610904693603516,
+      "learning_rate": 1.2270222222222223e-05,
+      "loss": 1.6691,
+      "step": 27990
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 44.72496032714844,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 1.6484,
+      "step": 28000
+    },
+    {
+      "epoch": 0.89632,
+      "grad_norm": 43.81786346435547,
+      "learning_rate": 1.2263111111111112e-05,
+      "loss": 1.6518,
+      "step": 28010
+    },
+    {
+      "epoch": 0.89664,
+      "grad_norm": 41.949825286865234,
+      "learning_rate": 1.2259555555555556e-05,
+      "loss": 1.6408,
+      "step": 28020
+    },
+    {
+      "epoch": 0.89696,
+      "grad_norm": 45.50816345214844,
+      "learning_rate": 1.2256000000000001e-05,
+      "loss": 1.6762,
+      "step": 28030
+    },
+    {
+      "epoch": 0.89728,
+      "grad_norm": 44.04999923706055,
+      "learning_rate": 1.2252444444444444e-05,
+      "loss": 1.6387,
+      "step": 28040
+    },
+    {
+      "epoch": 0.8976,
+      "grad_norm": 43.951969146728516,
+      "learning_rate": 1.224888888888889e-05,
+      "loss": 1.6345,
+      "step": 28050
+    },
+    {
+      "epoch": 0.89792,
+      "grad_norm": 44.5511360168457,
+      "learning_rate": 1.2245333333333333e-05,
+      "loss": 1.5975,
+      "step": 28060
+    },
+    {
+      "epoch": 0.89824,
+      "grad_norm": 43.500633239746094,
+      "learning_rate": 1.2241777777777779e-05,
+      "loss": 1.6463,
+      "step": 28070
+    },
+    {
+      "epoch": 0.89856,
+      "grad_norm": 44.43678665161133,
+      "learning_rate": 1.2238222222222222e-05,
+      "loss": 1.6813,
+      "step": 28080
+    },
+    {
+      "epoch": 0.89888,
+      "grad_norm": 43.428714752197266,
+      "learning_rate": 1.2234666666666668e-05,
+      "loss": 1.6547,
+      "step": 28090
+    },
+    {
+      "epoch": 0.8992,
+      "grad_norm": 43.72435760498047,
+      "learning_rate": 1.2231111111111111e-05,
+      "loss": 1.621,
+      "step": 28100
+    },
+    {
+      "epoch": 0.89952,
+      "grad_norm": 43.099552154541016,
+      "learning_rate": 1.2227555555555557e-05,
+      "loss": 1.6399,
+      "step": 28110
+    },
+    {
+      "epoch": 0.89984,
+      "grad_norm": 42.11380386352539,
+      "learning_rate": 1.2224e-05,
+      "loss": 1.6225,
+      "step": 28120
+    },
+    {
+      "epoch": 0.90016,
+      "grad_norm": 42.5230827331543,
+      "learning_rate": 1.2220444444444446e-05,
+      "loss": 1.6506,
+      "step": 28130
+    },
+    {
+      "epoch": 0.90048,
+      "grad_norm": 44.5308952331543,
+      "learning_rate": 1.221688888888889e-05,
+      "loss": 1.6425,
+      "step": 28140
+    },
+    {
+      "epoch": 0.9008,
+      "grad_norm": 47.679595947265625,
+      "learning_rate": 1.2213333333333336e-05,
+      "loss": 1.6505,
+      "step": 28150
+    },
+    {
+      "epoch": 0.90112,
+      "grad_norm": 42.81779861450195,
+      "learning_rate": 1.2209777777777778e-05,
+      "loss": 1.6382,
+      "step": 28160
+    },
+    {
+      "epoch": 0.90144,
+      "grad_norm": 43.218013763427734,
+      "learning_rate": 1.2206222222222223e-05,
+      "loss": 1.6254,
+      "step": 28170
+    },
+    {
+      "epoch": 0.90176,
+      "grad_norm": 44.04819869995117,
+      "learning_rate": 1.2202666666666668e-05,
+      "loss": 1.6533,
+      "step": 28180
+    },
+    {
+      "epoch": 0.90208,
+      "grad_norm": 45.29868698120117,
+      "learning_rate": 1.2199111111111112e-05,
+      "loss": 1.6509,
+      "step": 28190
+    },
+    {
+      "epoch": 0.9024,
+      "grad_norm": 42.42465591430664,
+      "learning_rate": 1.2195555555555557e-05,
+      "loss": 1.6312,
+      "step": 28200
+    },
+    {
+      "epoch": 0.90272,
+      "grad_norm": 44.77098846435547,
+      "learning_rate": 1.2192000000000001e-05,
+      "loss": 1.6445,
+      "step": 28210
+    },
+    {
+      "epoch": 0.90304,
+      "grad_norm": 43.591880798339844,
+      "learning_rate": 1.2188444444444446e-05,
+      "loss": 1.6475,
+      "step": 28220
+    },
+    {
+      "epoch": 0.90336,
+      "grad_norm": 42.93756103515625,
+      "learning_rate": 1.218488888888889e-05,
+      "loss": 1.6411,
+      "step": 28230
+    },
+    {
+      "epoch": 0.90368,
+      "grad_norm": 45.212730407714844,
+      "learning_rate": 1.2181333333333333e-05,
+      "loss": 1.6137,
+      "step": 28240
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 44.45610809326172,
+      "learning_rate": 1.217777777777778e-05,
+      "loss": 1.6571,
+      "step": 28250
+    },
+    {
+      "epoch": 0.90432,
+      "grad_norm": 43.78285217285156,
+      "learning_rate": 1.2174222222222222e-05,
+      "loss": 1.6473,
+      "step": 28260
+    },
+    {
+      "epoch": 0.90464,
+      "grad_norm": 44.92477035522461,
+      "learning_rate": 1.2170666666666668e-05,
+      "loss": 1.6356,
+      "step": 28270
+    },
+    {
+      "epoch": 0.90496,
+      "grad_norm": 44.4495849609375,
+      "learning_rate": 1.2167111111111111e-05,
+      "loss": 1.6697,
+      "step": 28280
+    },
+    {
+      "epoch": 0.90528,
+      "grad_norm": 43.52287292480469,
+      "learning_rate": 1.2163555555555558e-05,
+      "loss": 1.6756,
+      "step": 28290
+    },
+    {
+      "epoch": 0.9056,
+      "grad_norm": 45.75019454956055,
+      "learning_rate": 1.216e-05,
+      "loss": 1.6648,
+      "step": 28300
+    },
+    {
+      "epoch": 0.90592,
+      "grad_norm": 42.70742416381836,
+      "learning_rate": 1.2156444444444447e-05,
+      "loss": 1.6568,
+      "step": 28310
+    },
+    {
+      "epoch": 0.90624,
+      "grad_norm": 46.65618896484375,
+      "learning_rate": 1.215288888888889e-05,
+      "loss": 1.6157,
+      "step": 28320
+    },
+    {
+      "epoch": 0.90656,
+      "grad_norm": 44.10770797729492,
+      "learning_rate": 1.2149333333333336e-05,
+      "loss": 1.6449,
+      "step": 28330
+    },
+    {
+      "epoch": 0.90688,
+      "grad_norm": 44.807430267333984,
+      "learning_rate": 1.2145777777777779e-05,
+      "loss": 1.6285,
+      "step": 28340
+    },
+    {
+      "epoch": 0.9072,
+      "grad_norm": 46.65182113647461,
+      "learning_rate": 1.2142222222222225e-05,
+      "loss": 1.617,
+      "step": 28350
+    },
+    {
+      "epoch": 0.90752,
+      "grad_norm": 42.64336013793945,
+      "learning_rate": 1.2138666666666668e-05,
+      "loss": 1.6355,
+      "step": 28360
+    },
+    {
+      "epoch": 0.90784,
+      "grad_norm": 43.93117904663086,
+      "learning_rate": 1.2135111111111112e-05,
+      "loss": 1.648,
+      "step": 28370
+    },
+    {
+      "epoch": 0.90816,
+      "grad_norm": 45.63738250732422,
+      "learning_rate": 1.2131555555555557e-05,
+      "loss": 1.6653,
+      "step": 28380
+    },
+    {
+      "epoch": 0.90848,
+      "grad_norm": 45.124847412109375,
+      "learning_rate": 1.2128000000000001e-05,
+      "loss": 1.6288,
+      "step": 28390
+    },
+    {
+      "epoch": 0.9088,
+      "grad_norm": 43.86760330200195,
+      "learning_rate": 1.2124444444444446e-05,
+      "loss": 1.6478,
+      "step": 28400
+    },
+    {
+      "epoch": 0.90912,
+      "grad_norm": 43.364009857177734,
+      "learning_rate": 1.212088888888889e-05,
+      "loss": 1.6479,
+      "step": 28410
+    },
+    {
+      "epoch": 0.90944,
+      "grad_norm": 45.14009475708008,
+      "learning_rate": 1.2117333333333335e-05,
+      "loss": 1.6297,
+      "step": 28420
+    },
+    {
+      "epoch": 0.90976,
+      "grad_norm": 43.5945930480957,
+      "learning_rate": 1.211377777777778e-05,
+      "loss": 1.6466,
+      "step": 28430
+    },
+    {
+      "epoch": 0.91008,
+      "grad_norm": 43.39710998535156,
+      "learning_rate": 1.2110222222222223e-05,
+      "loss": 1.6827,
+      "step": 28440
+    },
+    {
+      "epoch": 0.9104,
+      "grad_norm": 43.625247955322266,
+      "learning_rate": 1.2106666666666667e-05,
+      "loss": 1.6457,
+      "step": 28450
+    },
+    {
+      "epoch": 0.91072,
+      "grad_norm": 43.71607971191406,
+      "learning_rate": 1.2103111111111112e-05,
+      "loss": 1.6719,
+      "step": 28460
+    },
+    {
+      "epoch": 0.91104,
+      "grad_norm": 43.85552215576172,
+      "learning_rate": 1.2099555555555556e-05,
+      "loss": 1.666,
+      "step": 28470
+    },
+    {
+      "epoch": 0.91136,
+      "grad_norm": 45.796871185302734,
+      "learning_rate": 1.2096e-05,
+      "loss": 1.6526,
+      "step": 28480
+    },
+    {
+      "epoch": 0.91168,
+      "grad_norm": 41.46237564086914,
+      "learning_rate": 1.2092444444444444e-05,
+      "loss": 1.6589,
+      "step": 28490
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 46.35646438598633,
+      "learning_rate": 1.208888888888889e-05,
+      "loss": 1.6578,
+      "step": 28500
+    },
+    {
+      "epoch": 0.91232,
+      "grad_norm": 44.154850006103516,
+      "learning_rate": 1.2085333333333333e-05,
+      "loss": 1.622,
+      "step": 28510
+    },
+    {
+      "epoch": 0.91264,
+      "grad_norm": 45.82959747314453,
+      "learning_rate": 1.2081777777777779e-05,
+      "loss": 1.6247,
+      "step": 28520
+    },
+    {
+      "epoch": 0.91296,
+      "grad_norm": 44.784549713134766,
+      "learning_rate": 1.2078222222222222e-05,
+      "loss": 1.6267,
+      "step": 28530
+    },
+    {
+      "epoch": 0.91328,
+      "grad_norm": 98.08938598632812,
+      "learning_rate": 1.2074666666666668e-05,
+      "loss": 1.6385,
+      "step": 28540
+    },
+    {
+      "epoch": 0.9136,
+      "grad_norm": 45.04857635498047,
+      "learning_rate": 1.2071111111111111e-05,
+      "loss": 1.6629,
+      "step": 28550
+    },
+    {
+      "epoch": 0.91392,
+      "grad_norm": 45.639442443847656,
+      "learning_rate": 1.2067555555555557e-05,
+      "loss": 1.6443,
+      "step": 28560
+    },
+    {
+      "epoch": 0.91424,
+      "grad_norm": 44.02001190185547,
+      "learning_rate": 1.2064e-05,
+      "loss": 1.6545,
+      "step": 28570
+    },
+    {
+      "epoch": 0.91456,
+      "grad_norm": 44.048583984375,
+      "learning_rate": 1.2060444444444446e-05,
+      "loss": 1.6316,
+      "step": 28580
+    },
+    {
+      "epoch": 0.91488,
+      "grad_norm": 43.934024810791016,
+      "learning_rate": 1.2056888888888889e-05,
+      "loss": 1.6516,
+      "step": 28590
+    },
+    {
+      "epoch": 0.9152,
+      "grad_norm": 43.88330841064453,
+      "learning_rate": 1.2053333333333335e-05,
+      "loss": 1.6364,
+      "step": 28600
+    },
+    {
+      "epoch": 0.91552,
+      "grad_norm": 44.25266647338867,
+      "learning_rate": 1.2049777777777778e-05,
+      "loss": 1.6422,
+      "step": 28610
+    },
+    {
+      "epoch": 0.91584,
+      "grad_norm": 43.72382736206055,
+      "learning_rate": 1.2046222222222223e-05,
+      "loss": 1.6472,
+      "step": 28620
+    },
+    {
+      "epoch": 0.91616,
+      "grad_norm": 44.77427291870117,
+      "learning_rate": 1.2042666666666667e-05,
+      "loss": 1.6509,
+      "step": 28630
+    },
+    {
+      "epoch": 0.91648,
+      "grad_norm": 47.40642547607422,
+      "learning_rate": 1.2039111111111112e-05,
+      "loss": 1.6523,
+      "step": 28640
+    },
+    {
+      "epoch": 0.9168,
+      "grad_norm": 43.53240966796875,
+      "learning_rate": 1.2035555555555556e-05,
+      "loss": 1.6204,
+      "step": 28650
+    },
+    {
+      "epoch": 0.91712,
+      "grad_norm": 42.13488006591797,
+      "learning_rate": 1.2032000000000001e-05,
+      "loss": 1.6457,
+      "step": 28660
+    },
+    {
+      "epoch": 0.91744,
+      "grad_norm": 46.60955047607422,
+      "learning_rate": 1.2028444444444446e-05,
+      "loss": 1.6273,
+      "step": 28670
+    },
+    {
+      "epoch": 0.91776,
+      "grad_norm": 45.47731399536133,
+      "learning_rate": 1.202488888888889e-05,
+      "loss": 1.6316,
+      "step": 28680
+    },
+    {
+      "epoch": 0.91808,
+      "grad_norm": 43.740535736083984,
+      "learning_rate": 1.2021333333333333e-05,
+      "loss": 1.6317,
+      "step": 28690
+    },
+    {
+      "epoch": 0.9184,
+      "grad_norm": 44.68436050415039,
+      "learning_rate": 1.201777777777778e-05,
+      "loss": 1.6506,
+      "step": 28700
+    },
+    {
+      "epoch": 0.91872,
+      "grad_norm": 44.82673263549805,
+      "learning_rate": 1.2014222222222222e-05,
+      "loss": 1.6467,
+      "step": 28710
+    },
+    {
+      "epoch": 0.91904,
+      "grad_norm": 41.92259979248047,
+      "learning_rate": 1.2010666666666668e-05,
+      "loss": 1.6017,
+      "step": 28720
+    },
+    {
+      "epoch": 0.91936,
+      "grad_norm": 44.171173095703125,
+      "learning_rate": 1.2007111111111111e-05,
+      "loss": 1.6213,
+      "step": 28730
+    },
+    {
+      "epoch": 0.91968,
+      "grad_norm": 45.21971130371094,
+      "learning_rate": 1.2003555555555557e-05,
+      "loss": 1.6214,
+      "step": 28740
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 44.320556640625,
+      "learning_rate": 1.2e-05,
+      "loss": 1.6314,
+      "step": 28750
+    },
+    {
+      "epoch": 0.92032,
+      "grad_norm": 42.63776779174805,
+      "learning_rate": 1.1996444444444447e-05,
+      "loss": 1.6227,
+      "step": 28760
+    },
+    {
+      "epoch": 0.92064,
+      "grad_norm": 43.34054183959961,
+      "learning_rate": 1.199288888888889e-05,
+      "loss": 1.6382,
+      "step": 28770
+    },
+    {
+      "epoch": 0.92096,
+      "grad_norm": 43.49864196777344,
+      "learning_rate": 1.1989333333333336e-05,
+      "loss": 1.6106,
+      "step": 28780
+    },
+    {
+      "epoch": 0.92128,
+      "grad_norm": 43.339412689208984,
+      "learning_rate": 1.1985777777777779e-05,
+      "loss": 1.6352,
+      "step": 28790
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 44.452125549316406,
+      "learning_rate": 1.1982222222222225e-05,
+      "loss": 1.6247,
+      "step": 28800
+    },
+    {
+      "epoch": 0.92192,
+      "grad_norm": 44.37428283691406,
+      "learning_rate": 1.1978666666666668e-05,
+      "loss": 1.6457,
+      "step": 28810
+    },
+    {
+      "epoch": 0.92224,
+      "grad_norm": 44.22825622558594,
+      "learning_rate": 1.1975111111111112e-05,
+      "loss": 1.6493,
+      "step": 28820
+    },
+    {
+      "epoch": 0.92256,
+      "grad_norm": 43.31019592285156,
+      "learning_rate": 1.1971555555555557e-05,
+      "loss": 1.6401,
+      "step": 28830
+    },
+    {
+      "epoch": 0.92288,
+      "grad_norm": 44.81446075439453,
+      "learning_rate": 1.1968000000000001e-05,
+      "loss": 1.6376,
+      "step": 28840
+    },
+    {
+      "epoch": 0.9232,
+      "grad_norm": 44.13412857055664,
+      "learning_rate": 1.1964444444444446e-05,
+      "loss": 1.6413,
+      "step": 28850
+    },
+    {
+      "epoch": 0.92352,
+      "grad_norm": 44.80860900878906,
+      "learning_rate": 1.196088888888889e-05,
+      "loss": 1.6442,
+      "step": 28860
+    },
+    {
+      "epoch": 0.92384,
+      "grad_norm": 43.5064582824707,
+      "learning_rate": 1.1957333333333335e-05,
+      "loss": 1.6279,
+      "step": 28870
+    },
+    {
+      "epoch": 0.92416,
+      "grad_norm": 42.37964630126953,
+      "learning_rate": 1.195377777777778e-05,
+      "loss": 1.6716,
+      "step": 28880
+    },
+    {
+      "epoch": 0.92448,
+      "grad_norm": 43.18762969970703,
+      "learning_rate": 1.1950222222222222e-05,
+      "loss": 1.6494,
+      "step": 28890
+    },
+    {
+      "epoch": 0.9248,
+      "grad_norm": 44.32823181152344,
+      "learning_rate": 1.1946666666666669e-05,
+      "loss": 1.6419,
+      "step": 28900
+    },
+    {
+      "epoch": 0.92512,
+      "grad_norm": 43.62746810913086,
+      "learning_rate": 1.1943111111111111e-05,
+      "loss": 1.6703,
+      "step": 28910
+    },
+    {
+      "epoch": 0.92544,
+      "grad_norm": 45.503726959228516,
+      "learning_rate": 1.1939555555555558e-05,
+      "loss": 1.6358,
+      "step": 28920
+    },
+    {
+      "epoch": 0.92576,
+      "grad_norm": 43.485008239746094,
+      "learning_rate": 1.1936e-05,
+      "loss": 1.6333,
+      "step": 28930
+    },
+    {
+      "epoch": 0.92608,
+      "grad_norm": 41.778106689453125,
+      "learning_rate": 1.1932444444444447e-05,
+      "loss": 1.6323,
+      "step": 28940
+    },
+    {
+      "epoch": 0.9264,
+      "grad_norm": 43.94820785522461,
+      "learning_rate": 1.192888888888889e-05,
+      "loss": 1.6447,
+      "step": 28950
+    },
+    {
+      "epoch": 0.92672,
+      "grad_norm": 43.780738830566406,
+      "learning_rate": 1.1925333333333333e-05,
+      "loss": 1.625,
+      "step": 28960
+    },
+    {
+      "epoch": 0.92704,
+      "grad_norm": 44.629703521728516,
+      "learning_rate": 1.1921777777777779e-05,
+      "loss": 1.6408,
+      "step": 28970
+    },
+    {
+      "epoch": 0.92736,
+      "grad_norm": 41.52702713012695,
+      "learning_rate": 1.1918222222222222e-05,
+      "loss": 1.6233,
+      "step": 28980
+    },
+    {
+      "epoch": 0.92768,
+      "grad_norm": 43.539798736572266,
+      "learning_rate": 1.1914666666666668e-05,
+      "loss": 1.625,
+      "step": 28990
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 43.91656494140625,
+      "learning_rate": 1.191111111111111e-05,
+      "loss": 1.6765,
+      "step": 29000
+    },
+    {
+      "epoch": 0.92832,
+      "grad_norm": 45.65538024902344,
+      "learning_rate": 1.1907555555555557e-05,
+      "loss": 1.6276,
+      "step": 29010
+    },
+    {
+      "epoch": 0.92864,
+      "grad_norm": 42.96657180786133,
+      "learning_rate": 1.1904e-05,
+      "loss": 1.6342,
+      "step": 29020
+    },
+    {
+      "epoch": 0.92896,
+      "grad_norm": 43.52238464355469,
+      "learning_rate": 1.1900444444444446e-05,
+      "loss": 1.6234,
+      "step": 29030
+    },
+    {
+      "epoch": 0.92928,
+      "grad_norm": 44.23517608642578,
+      "learning_rate": 1.1896888888888889e-05,
+      "loss": 1.6552,
+      "step": 29040
+    },
+    {
+      "epoch": 0.9296,
+      "grad_norm": 43.24574661254883,
+      "learning_rate": 1.1893333333333335e-05,
+      "loss": 1.6266,
+      "step": 29050
+    },
+    {
+      "epoch": 0.92992,
+      "grad_norm": 46.02854919433594,
+      "learning_rate": 1.1889777777777778e-05,
+      "loss": 1.6376,
+      "step": 29060
+    },
+    {
+      "epoch": 0.93024,
+      "grad_norm": 44.43922424316406,
+      "learning_rate": 1.1886222222222223e-05,
+      "loss": 1.6347,
+      "step": 29070
+    },
+    {
+      "epoch": 0.93056,
+      "grad_norm": 42.341453552246094,
+      "learning_rate": 1.1882666666666667e-05,
+      "loss": 1.6311,
+      "step": 29080
+    },
+    {
+      "epoch": 0.93088,
+      "grad_norm": 43.742149353027344,
+      "learning_rate": 1.1879111111111112e-05,
+      "loss": 1.5966,
+      "step": 29090
+    },
+    {
+      "epoch": 0.9312,
+      "grad_norm": 45.52425765991211,
+      "learning_rate": 1.1875555555555556e-05,
+      "loss": 1.6459,
+      "step": 29100
+    },
+    {
+      "epoch": 0.93152,
+      "grad_norm": 42.77261734008789,
+      "learning_rate": 1.1872000000000001e-05,
+      "loss": 1.5989,
+      "step": 29110
+    },
+    {
+      "epoch": 0.93184,
+      "grad_norm": 43.86947250366211,
+      "learning_rate": 1.1868444444444445e-05,
+      "loss": 1.6352,
+      "step": 29120
+    },
+    {
+      "epoch": 0.93216,
+      "grad_norm": 43.85417175292969,
+      "learning_rate": 1.186488888888889e-05,
+      "loss": 1.6306,
+      "step": 29130
+    },
+    {
+      "epoch": 0.93248,
+      "grad_norm": 46.29254150390625,
+      "learning_rate": 1.1861333333333333e-05,
+      "loss": 1.6471,
+      "step": 29140
+    },
+    {
+      "epoch": 0.9328,
+      "grad_norm": 45.65533447265625,
+      "learning_rate": 1.1857777777777779e-05,
+      "loss": 1.6235,
+      "step": 29150
+    },
+    {
+      "epoch": 0.93312,
+      "grad_norm": 42.154380798339844,
+      "learning_rate": 1.1854222222222222e-05,
+      "loss": 1.6275,
+      "step": 29160
+    },
+    {
+      "epoch": 0.93344,
+      "grad_norm": 43.51826858520508,
+      "learning_rate": 1.1850666666666668e-05,
+      "loss": 1.6388,
+      "step": 29170
+    },
+    {
+      "epoch": 0.93376,
+      "grad_norm": 45.132057189941406,
+      "learning_rate": 1.1847111111111111e-05,
+      "loss": 1.6193,
+      "step": 29180
+    },
+    {
+      "epoch": 0.93408,
+      "grad_norm": 45.140865325927734,
+      "learning_rate": 1.1843555555555557e-05,
+      "loss": 1.6266,
+      "step": 29190
+    },
+    {
+      "epoch": 0.9344,
+      "grad_norm": 44.52220153808594,
+      "learning_rate": 1.184e-05,
+      "loss": 1.6444,
+      "step": 29200
+    },
+    {
+      "epoch": 0.93472,
+      "grad_norm": 43.61417770385742,
+      "learning_rate": 1.1836444444444446e-05,
+      "loss": 1.6247,
+      "step": 29210
+    },
+    {
+      "epoch": 0.93504,
+      "grad_norm": 43.42637252807617,
+      "learning_rate": 1.183288888888889e-05,
+      "loss": 1.6392,
+      "step": 29220
+    },
+    {
+      "epoch": 0.93536,
+      "grad_norm": 42.174129486083984,
+      "learning_rate": 1.1829333333333335e-05,
+      "loss": 1.6534,
+      "step": 29230
+    },
+    {
+      "epoch": 0.93568,
+      "grad_norm": 42.143123626708984,
+      "learning_rate": 1.1825777777777778e-05,
+      "loss": 1.6345,
+      "step": 29240
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 46.554283142089844,
+      "learning_rate": 1.1822222222222225e-05,
+      "loss": 1.641,
+      "step": 29250
+    },
+    {
+      "epoch": 0.93632,
+      "grad_norm": 44.24030685424805,
+      "learning_rate": 1.1818666666666667e-05,
+      "loss": 1.6171,
+      "step": 29260
+    },
+    {
+      "epoch": 0.93664,
+      "grad_norm": 42.74769973754883,
+      "learning_rate": 1.1815111111111112e-05,
+      "loss": 1.6048,
+      "step": 29270
+    },
+    {
+      "epoch": 0.93696,
+      "grad_norm": 44.60746383666992,
+      "learning_rate": 1.1811555555555557e-05,
+      "loss": 1.6614,
+      "step": 29280
+    },
+    {
+      "epoch": 0.93728,
+      "grad_norm": 44.492679595947266,
+      "learning_rate": 1.1808000000000001e-05,
+      "loss": 1.6266,
+      "step": 29290
+    },
+    {
+      "epoch": 0.9376,
+      "grad_norm": 44.245113372802734,
+      "learning_rate": 1.1804444444444446e-05,
+      "loss": 1.637,
+      "step": 29300
+    },
+    {
+      "epoch": 0.93792,
+      "grad_norm": 43.29434585571289,
+      "learning_rate": 1.180088888888889e-05,
+      "loss": 1.6287,
+      "step": 29310
+    },
+    {
+      "epoch": 0.93824,
+      "grad_norm": 45.088443756103516,
+      "learning_rate": 1.1797333333333335e-05,
+      "loss": 1.6463,
+      "step": 29320
+    },
+    {
+      "epoch": 0.93856,
+      "grad_norm": 43.38103103637695,
+      "learning_rate": 1.179377777777778e-05,
+      "loss": 1.6177,
+      "step": 29330
+    },
+    {
+      "epoch": 0.93888,
+      "grad_norm": 45.160545349121094,
+      "learning_rate": 1.1790222222222222e-05,
+      "loss": 1.6182,
+      "step": 29340
+    },
+    {
+      "epoch": 0.9392,
+      "grad_norm": 43.422874450683594,
+      "learning_rate": 1.1786666666666668e-05,
+      "loss": 1.6388,
+      "step": 29350
+    },
+    {
+      "epoch": 0.93952,
+      "grad_norm": 45.420005798339844,
+      "learning_rate": 1.1783111111111111e-05,
+      "loss": 1.649,
+      "step": 29360
+    },
+    {
+      "epoch": 0.93984,
+      "grad_norm": 45.20408630371094,
+      "learning_rate": 1.1779555555555558e-05,
+      "loss": 1.6424,
+      "step": 29370
+    },
+    {
+      "epoch": 0.94016,
+      "grad_norm": 45.90959548950195,
+      "learning_rate": 1.1776e-05,
+      "loss": 1.629,
+      "step": 29380
+    },
+    {
+      "epoch": 0.94048,
+      "grad_norm": 43.6275520324707,
+      "learning_rate": 1.1772444444444447e-05,
+      "loss": 1.6303,
+      "step": 29390
+    },
+    {
+      "epoch": 0.9408,
+      "grad_norm": 44.0329704284668,
+      "learning_rate": 1.176888888888889e-05,
+      "loss": 1.6673,
+      "step": 29400
+    },
+    {
+      "epoch": 0.94112,
+      "grad_norm": 45.125205993652344,
+      "learning_rate": 1.1765333333333336e-05,
+      "loss": 1.6487,
+      "step": 29410
+    },
+    {
+      "epoch": 0.94144,
+      "grad_norm": 44.226348876953125,
+      "learning_rate": 1.1761777777777779e-05,
+      "loss": 1.6489,
+      "step": 29420
+    },
+    {
+      "epoch": 0.94176,
+      "grad_norm": 45.3155403137207,
+      "learning_rate": 1.1758222222222225e-05,
+      "loss": 1.6348,
+      "step": 29430
+    },
+    {
+      "epoch": 0.94208,
+      "grad_norm": 43.43427276611328,
+      "learning_rate": 1.1754666666666668e-05,
+      "loss": 1.6218,
+      "step": 29440
+    },
+    {
+      "epoch": 0.9424,
+      "grad_norm": 44.82199478149414,
+      "learning_rate": 1.1751111111111112e-05,
+      "loss": 1.6213,
+      "step": 29450
+    },
+    {
+      "epoch": 0.94272,
+      "grad_norm": 43.701515197753906,
+      "learning_rate": 1.1747555555555557e-05,
+      "loss": 1.675,
+      "step": 29460
+    },
+    {
+      "epoch": 0.94304,
+      "grad_norm": 44.70633316040039,
+      "learning_rate": 1.1744000000000001e-05,
+      "loss": 1.6333,
+      "step": 29470
+    },
+    {
+      "epoch": 0.94336,
+      "grad_norm": 45.40119552612305,
+      "learning_rate": 1.1740444444444446e-05,
+      "loss": 1.6605,
+      "step": 29480
+    },
+    {
+      "epoch": 0.94368,
+      "grad_norm": 42.81822967529297,
+      "learning_rate": 1.1736888888888889e-05,
+      "loss": 1.6182,
+      "step": 29490
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 43.596412658691406,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 1.6614,
+      "step": 29500
+    },
+    {
+      "epoch": 0.94432,
+      "grad_norm": 43.47703170776367,
+      "learning_rate": 1.1729777777777778e-05,
+      "loss": 1.6031,
+      "step": 29510
+    },
+    {
+      "epoch": 0.94464,
+      "grad_norm": 42.71162414550781,
+      "learning_rate": 1.1726222222222222e-05,
+      "loss": 1.6587,
+      "step": 29520
+    },
+    {
+      "epoch": 0.94496,
+      "grad_norm": 43.19927215576172,
+      "learning_rate": 1.1722666666666667e-05,
+      "loss": 1.6457,
+      "step": 29530
+    },
+    {
+      "epoch": 0.94528,
+      "grad_norm": 44.56126022338867,
+      "learning_rate": 1.1719111111111112e-05,
+      "loss": 1.6242,
+      "step": 29540
+    },
+    {
+      "epoch": 0.9456,
+      "grad_norm": 44.282997131347656,
+      "learning_rate": 1.1715555555555556e-05,
+      "loss": 1.6344,
+      "step": 29550
+    },
+    {
+      "epoch": 0.94592,
+      "grad_norm": 44.068756103515625,
+      "learning_rate": 1.1712e-05,
+      "loss": 1.6445,
+      "step": 29560
+    },
+    {
+      "epoch": 0.94624,
+      "grad_norm": 43.77211380004883,
+      "learning_rate": 1.1708444444444445e-05,
+      "loss": 1.6255,
+      "step": 29570
+    },
+    {
+      "epoch": 0.94656,
+      "grad_norm": 44.08302307128906,
+      "learning_rate": 1.170488888888889e-05,
+      "loss": 1.673,
+      "step": 29580
+    },
+    {
+      "epoch": 0.94688,
+      "grad_norm": 43.33510208129883,
+      "learning_rate": 1.1701333333333333e-05,
+      "loss": 1.5995,
+      "step": 29590
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 42.72581100463867,
+      "learning_rate": 1.1697777777777779e-05,
+      "loss": 1.6474,
+      "step": 29600
+    },
+    {
+      "epoch": 0.94752,
+      "grad_norm": 43.55382537841797,
+      "learning_rate": 1.1694222222222222e-05,
+      "loss": 1.64,
+      "step": 29610
+    },
+    {
+      "epoch": 0.94784,
+      "grad_norm": 42.47932815551758,
+      "learning_rate": 1.1690666666666668e-05,
+      "loss": 1.5957,
+      "step": 29620
+    },
+    {
+      "epoch": 0.94816,
+      "grad_norm": 44.38590621948242,
+      "learning_rate": 1.1687111111111111e-05,
+      "loss": 1.6377,
+      "step": 29630
+    },
+    {
+      "epoch": 0.94848,
+      "grad_norm": 42.693572998046875,
+      "learning_rate": 1.1683555555555557e-05,
+      "loss": 1.6525,
+      "step": 29640
+    },
+    {
+      "epoch": 0.9488,
+      "grad_norm": 46.156307220458984,
+      "learning_rate": 1.168e-05,
+      "loss": 1.6412,
+      "step": 29650
+    },
+    {
+      "epoch": 0.94912,
+      "grad_norm": 44.097286224365234,
+      "learning_rate": 1.1676444444444446e-05,
+      "loss": 1.6264,
+      "step": 29660
+    },
+    {
+      "epoch": 0.94944,
+      "grad_norm": 44.33402633666992,
+      "learning_rate": 1.1672888888888889e-05,
+      "loss": 1.6095,
+      "step": 29670
+    },
+    {
+      "epoch": 0.94976,
+      "grad_norm": 44.41627502441406,
+      "learning_rate": 1.1669333333333335e-05,
+      "loss": 1.6228,
+      "step": 29680
+    },
+    {
+      "epoch": 0.95008,
+      "grad_norm": 42.222347259521484,
+      "learning_rate": 1.1665777777777778e-05,
+      "loss": 1.6349,
+      "step": 29690
+    },
+    {
+      "epoch": 0.9504,
+      "grad_norm": 42.85820007324219,
+      "learning_rate": 1.1662222222222224e-05,
+      "loss": 1.6481,
+      "step": 29700
+    },
+    {
+      "epoch": 0.95072,
+      "grad_norm": 44.937313079833984,
+      "learning_rate": 1.1658666666666667e-05,
+      "loss": 1.6529,
+      "step": 29710
+    },
+    {
+      "epoch": 0.95104,
+      "grad_norm": 44.15386962890625,
+      "learning_rate": 1.1655111111111112e-05,
+      "loss": 1.6421,
+      "step": 29720
+    },
+    {
+      "epoch": 0.95136,
+      "grad_norm": 45.95855712890625,
+      "learning_rate": 1.1651555555555556e-05,
+      "loss": 1.6265,
+      "step": 29730
+    },
+    {
+      "epoch": 0.95168,
+      "grad_norm": 43.64844512939453,
+      "learning_rate": 1.1648000000000001e-05,
+      "loss": 1.6553,
+      "step": 29740
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 45.064292907714844,
+      "learning_rate": 1.1644444444444446e-05,
+      "loss": 1.6425,
+      "step": 29750
+    },
+    {
+      "epoch": 0.95232,
+      "grad_norm": 42.177894592285156,
+      "learning_rate": 1.164088888888889e-05,
+      "loss": 1.6312,
+      "step": 29760
+    },
+    {
+      "epoch": 0.95264,
+      "grad_norm": 43.947303771972656,
+      "learning_rate": 1.1637333333333335e-05,
+      "loss": 1.6438,
+      "step": 29770
+    },
+    {
+      "epoch": 0.95296,
+      "grad_norm": 43.64456558227539,
+      "learning_rate": 1.163377777777778e-05,
+      "loss": 1.6369,
+      "step": 29780
+    },
+    {
+      "epoch": 0.95328,
+      "grad_norm": 42.0169792175293,
+      "learning_rate": 1.1630222222222222e-05,
+      "loss": 1.6157,
+      "step": 29790
+    },
+    {
+      "epoch": 0.9536,
+      "grad_norm": 45.208892822265625,
+      "learning_rate": 1.1626666666666668e-05,
+      "loss": 1.6239,
+      "step": 29800
+    },
+    {
+      "epoch": 0.95392,
+      "grad_norm": 45.014469146728516,
+      "learning_rate": 1.1623111111111111e-05,
+      "loss": 1.6342,
+      "step": 29810
+    },
+    {
+      "epoch": 0.95424,
+      "grad_norm": 43.519657135009766,
+      "learning_rate": 1.1619555555555557e-05,
+      "loss": 1.662,
+      "step": 29820
+    },
+    {
+      "epoch": 0.95456,
+      "grad_norm": 42.91891098022461,
+      "learning_rate": 1.1616e-05,
+      "loss": 1.6255,
+      "step": 29830
+    },
+    {
+      "epoch": 0.95488,
+      "grad_norm": 43.61880111694336,
+      "learning_rate": 1.1612444444444447e-05,
+      "loss": 1.6335,
+      "step": 29840
+    },
+    {
+      "epoch": 0.9552,
+      "grad_norm": 43.08018493652344,
+      "learning_rate": 1.160888888888889e-05,
+      "loss": 1.6338,
+      "step": 29850
+    },
+    {
+      "epoch": 0.95552,
+      "grad_norm": 46.001365661621094,
+      "learning_rate": 1.1605333333333336e-05,
+      "loss": 1.6812,
+      "step": 29860
+    },
+    {
+      "epoch": 0.95584,
+      "grad_norm": 45.59417724609375,
+      "learning_rate": 1.1601777777777778e-05,
+      "loss": 1.6063,
+      "step": 29870
+    },
+    {
+      "epoch": 0.95616,
+      "grad_norm": 43.582889556884766,
+      "learning_rate": 1.1598222222222225e-05,
+      "loss": 1.6427,
+      "step": 29880
+    },
+    {
+      "epoch": 0.95648,
+      "grad_norm": 43.61624526977539,
+      "learning_rate": 1.1594666666666668e-05,
+      "loss": 1.6171,
+      "step": 29890
+    },
+    {
+      "epoch": 0.9568,
+      "grad_norm": 46.41237258911133,
+      "learning_rate": 1.1591111111111114e-05,
+      "loss": 1.6525,
+      "step": 29900
+    },
+    {
+      "epoch": 0.95712,
+      "grad_norm": 43.611732482910156,
+      "learning_rate": 1.1587555555555557e-05,
+      "loss": 1.6371,
+      "step": 29910
+    },
+    {
+      "epoch": 0.95744,
+      "grad_norm": 43.99297332763672,
+      "learning_rate": 1.1584000000000001e-05,
+      "loss": 1.6593,
+      "step": 29920
+    },
+    {
+      "epoch": 0.95776,
+      "grad_norm": 44.05895233154297,
+      "learning_rate": 1.1580444444444446e-05,
+      "loss": 1.6381,
+      "step": 29930
+    },
+    {
+      "epoch": 0.95808,
+      "grad_norm": 43.11012649536133,
+      "learning_rate": 1.157688888888889e-05,
+      "loss": 1.6465,
+      "step": 29940
+    },
+    {
+      "epoch": 0.9584,
+      "grad_norm": 46.63924026489258,
+      "learning_rate": 1.1573333333333335e-05,
+      "loss": 1.6507,
+      "step": 29950
+    },
+    {
+      "epoch": 0.95872,
+      "grad_norm": 44.59290313720703,
+      "learning_rate": 1.156977777777778e-05,
+      "loss": 1.6696,
+      "step": 29960
+    },
+    {
+      "epoch": 0.95904,
+      "grad_norm": 45.51963424682617,
+      "learning_rate": 1.1566222222222222e-05,
+      "loss": 1.6326,
+      "step": 29970
+    },
+    {
+      "epoch": 0.95936,
+      "grad_norm": 44.144065856933594,
+      "learning_rate": 1.1562666666666669e-05,
+      "loss": 1.6533,
+      "step": 29980
+    },
+    {
+      "epoch": 0.95968,
+      "grad_norm": 44.824337005615234,
+      "learning_rate": 1.1559111111111111e-05,
+      "loss": 1.6399,
+      "step": 29990
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 44.10457229614258,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 1.6111,
+      "step": 30000
+    },
+    {
+      "epoch": 0.96032,
+      "grad_norm": 44.56742858886719,
+      "learning_rate": 1.1552e-05,
+      "loss": 1.6558,
+      "step": 30010
+    },
+    {
+      "epoch": 0.96064,
+      "grad_norm": 43.97772216796875,
+      "learning_rate": 1.1548444444444445e-05,
+      "loss": 1.659,
+      "step": 30020
+    },
+    {
+      "epoch": 0.96096,
+      "grad_norm": 44.7454948425293,
+      "learning_rate": 1.154488888888889e-05,
+      "loss": 1.6539,
+      "step": 30030
+    },
+    {
+      "epoch": 0.96128,
+      "grad_norm": 45.64804458618164,
+      "learning_rate": 1.1541333333333332e-05,
+      "loss": 1.6259,
+      "step": 30040
+    },
+    {
+      "epoch": 0.9616,
+      "grad_norm": 48.72989273071289,
+      "learning_rate": 1.1537777777777779e-05,
+      "loss": 1.6336,
+      "step": 30050
+    },
+    {
+      "epoch": 0.96192,
+      "grad_norm": 46.82978439331055,
+      "learning_rate": 1.1534222222222222e-05,
+      "loss": 1.6409,
+      "step": 30060
+    },
+    {
+      "epoch": 0.96224,
+      "grad_norm": 43.353065490722656,
+      "learning_rate": 1.1530666666666668e-05,
+      "loss": 1.6433,
+      "step": 30070
+    },
+    {
+      "epoch": 0.96256,
+      "grad_norm": 42.573368072509766,
+      "learning_rate": 1.152711111111111e-05,
+      "loss": 1.6357,
+      "step": 30080
+    },
+    {
+      "epoch": 0.96288,
+      "grad_norm": 42.796142578125,
+      "learning_rate": 1.1523555555555557e-05,
+      "loss": 1.6457,
+      "step": 30090
+    },
+    {
+      "epoch": 0.9632,
+      "grad_norm": 45.9584846496582,
+      "learning_rate": 1.152e-05,
+      "loss": 1.6248,
+      "step": 30100
+    },
+    {
+      "epoch": 0.96352,
+      "grad_norm": 44.436767578125,
+      "learning_rate": 1.1516444444444446e-05,
+      "loss": 1.6573,
+      "step": 30110
+    },
+    {
+      "epoch": 0.96384,
+      "grad_norm": 44.212158203125,
+      "learning_rate": 1.1512888888888889e-05,
+      "loss": 1.6392,
+      "step": 30120
+    },
+    {
+      "epoch": 0.96416,
+      "grad_norm": 44.3819465637207,
+      "learning_rate": 1.1509333333333335e-05,
+      "loss": 1.6345,
+      "step": 30130
+    },
+    {
+      "epoch": 0.96448,
+      "grad_norm": 44.8933219909668,
+      "learning_rate": 1.1505777777777778e-05,
+      "loss": 1.6561,
+      "step": 30140
+    },
+    {
+      "epoch": 0.9648,
+      "grad_norm": 44.94137191772461,
+      "learning_rate": 1.1502222222222224e-05,
+      "loss": 1.6266,
+      "step": 30150
+    },
+    {
+      "epoch": 0.96512,
+      "grad_norm": 42.0122184753418,
+      "learning_rate": 1.1498666666666667e-05,
+      "loss": 1.6291,
+      "step": 30160
+    },
+    {
+      "epoch": 0.96544,
+      "grad_norm": 43.78432083129883,
+      "learning_rate": 1.1495111111111112e-05,
+      "loss": 1.676,
+      "step": 30170
+    },
+    {
+      "epoch": 0.96576,
+      "grad_norm": 43.573978424072266,
+      "learning_rate": 1.1491555555555556e-05,
+      "loss": 1.6278,
+      "step": 30180
+    },
+    {
+      "epoch": 0.96608,
+      "grad_norm": 44.11012268066406,
+      "learning_rate": 1.1488e-05,
+      "loss": 1.6484,
+      "step": 30190
+    },
+    {
+      "epoch": 0.9664,
+      "grad_norm": 43.47837448120117,
+      "learning_rate": 1.1484444444444445e-05,
+      "loss": 1.6247,
+      "step": 30200
+    },
+    {
+      "epoch": 0.96672,
+      "grad_norm": 44.80249786376953,
+      "learning_rate": 1.148088888888889e-05,
+      "loss": 1.6512,
+      "step": 30210
+    },
+    {
+      "epoch": 0.96704,
+      "grad_norm": 43.91743087768555,
+      "learning_rate": 1.1477333333333334e-05,
+      "loss": 1.644,
+      "step": 30220
+    },
+    {
+      "epoch": 0.96736,
+      "grad_norm": 42.66202926635742,
+      "learning_rate": 1.1473777777777779e-05,
+      "loss": 1.6557,
+      "step": 30230
+    },
+    {
+      "epoch": 0.96768,
+      "grad_norm": 43.54933547973633,
+      "learning_rate": 1.1470222222222222e-05,
+      "loss": 1.6268,
+      "step": 30240
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 43.00261306762695,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 1.6504,
+      "step": 30250
+    },
+    {
+      "epoch": 0.96832,
+      "grad_norm": 44.712650299072266,
+      "learning_rate": 1.1463111111111111e-05,
+      "loss": 1.6051,
+      "step": 30260
+    },
+    {
+      "epoch": 0.96864,
+      "grad_norm": 44.88301467895508,
+      "learning_rate": 1.1459555555555557e-05,
+      "loss": 1.6384,
+      "step": 30270
+    },
+    {
+      "epoch": 0.96896,
+      "grad_norm": 43.14665985107422,
+      "learning_rate": 1.1456e-05,
+      "loss": 1.6284,
+      "step": 30280
+    },
+    {
+      "epoch": 0.96928,
+      "grad_norm": 46.33767318725586,
+      "learning_rate": 1.1452444444444446e-05,
+      "loss": 1.649,
+      "step": 30290
+    },
+    {
+      "epoch": 0.9696,
+      "grad_norm": 43.687705993652344,
+      "learning_rate": 1.144888888888889e-05,
+      "loss": 1.6431,
+      "step": 30300
+    },
+    {
+      "epoch": 0.96992,
+      "grad_norm": 44.642269134521484,
+      "learning_rate": 1.1445333333333335e-05,
+      "loss": 1.6555,
+      "step": 30310
+    },
+    {
+      "epoch": 0.97024,
+      "grad_norm": 44.56669235229492,
+      "learning_rate": 1.1441777777777778e-05,
+      "loss": 1.643,
+      "step": 30320
+    },
+    {
+      "epoch": 0.97056,
+      "grad_norm": 45.14842224121094,
+      "learning_rate": 1.1438222222222225e-05,
+      "loss": 1.6202,
+      "step": 30330
+    },
+    {
+      "epoch": 0.97088,
+      "grad_norm": 47.29034423828125,
+      "learning_rate": 1.1434666666666667e-05,
+      "loss": 1.6382,
+      "step": 30340
+    },
+    {
+      "epoch": 0.9712,
+      "grad_norm": 44.174095153808594,
+      "learning_rate": 1.1431111111111114e-05,
+      "loss": 1.6335,
+      "step": 30350
+    },
+    {
+      "epoch": 0.97152,
+      "grad_norm": 44.7418098449707,
+      "learning_rate": 1.1427555555555557e-05,
+      "loss": 1.6421,
+      "step": 30360
+    },
+    {
+      "epoch": 0.97184,
+      "grad_norm": 42.68954849243164,
+      "learning_rate": 1.1424000000000001e-05,
+      "loss": 1.6339,
+      "step": 30370
+    },
+    {
+      "epoch": 0.97216,
+      "grad_norm": 43.91781997680664,
+      "learning_rate": 1.1420444444444446e-05,
+      "loss": 1.6163,
+      "step": 30380
+    },
+    {
+      "epoch": 0.97248,
+      "grad_norm": 42.75111770629883,
+      "learning_rate": 1.141688888888889e-05,
+      "loss": 1.63,
+      "step": 30390
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 44.555721282958984,
+      "learning_rate": 1.1413333333333335e-05,
+      "loss": 1.6256,
+      "step": 30400
+    },
+    {
+      "epoch": 0.97312,
+      "grad_norm": 46.852989196777344,
+      "learning_rate": 1.140977777777778e-05,
+      "loss": 1.6379,
+      "step": 30410
+    },
+    {
+      "epoch": 0.97344,
+      "grad_norm": 46.11513137817383,
+      "learning_rate": 1.1406222222222222e-05,
+      "loss": 1.6444,
+      "step": 30420
+    },
+    {
+      "epoch": 0.97376,
+      "grad_norm": 49.00882339477539,
+      "learning_rate": 1.1402666666666668e-05,
+      "loss": 1.6371,
+      "step": 30430
+    },
+    {
+      "epoch": 0.97408,
+      "grad_norm": 44.7541389465332,
+      "learning_rate": 1.1399111111111111e-05,
+      "loss": 1.6025,
+      "step": 30440
+    },
+    {
+      "epoch": 0.9744,
+      "grad_norm": 42.61855697631836,
+      "learning_rate": 1.1395555555555558e-05,
+      "loss": 1.6189,
+      "step": 30450
+    },
+    {
+      "epoch": 0.97472,
+      "grad_norm": 44.200260162353516,
+      "learning_rate": 1.1392e-05,
+      "loss": 1.6056,
+      "step": 30460
+    },
+    {
+      "epoch": 0.97504,
+      "grad_norm": 44.01932144165039,
+      "learning_rate": 1.1388444444444447e-05,
+      "loss": 1.6664,
+      "step": 30470
+    },
+    {
+      "epoch": 0.97536,
+      "grad_norm": 44.93379211425781,
+      "learning_rate": 1.138488888888889e-05,
+      "loss": 1.6603,
+      "step": 30480
+    },
+    {
+      "epoch": 0.97568,
+      "grad_norm": 42.41716766357422,
+      "learning_rate": 1.1381333333333336e-05,
+      "loss": 1.6235,
+      "step": 30490
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 46.20936584472656,
+      "learning_rate": 1.1377777777777779e-05,
+      "loss": 1.6323,
+      "step": 30500
+    },
+    {
+      "epoch": 0.97632,
+      "grad_norm": 41.390140533447266,
+      "learning_rate": 1.1374222222222221e-05,
+      "loss": 1.6059,
+      "step": 30510
+    },
+    {
+      "epoch": 0.97664,
+      "grad_norm": 44.24008560180664,
+      "learning_rate": 1.1370666666666668e-05,
+      "loss": 1.6441,
+      "step": 30520
+    },
+    {
+      "epoch": 0.97696,
+      "grad_norm": 44.89662170410156,
+      "learning_rate": 1.136711111111111e-05,
+      "loss": 1.6322,
+      "step": 30530
+    },
+    {
+      "epoch": 0.97728,
+      "grad_norm": 45.3996696472168,
+      "learning_rate": 1.1363555555555557e-05,
+      "loss": 1.6557,
+      "step": 30540
+    },
+    {
+      "epoch": 0.9776,
+      "grad_norm": 43.94699478149414,
+      "learning_rate": 1.136e-05,
+      "loss": 1.6289,
+      "step": 30550
+    },
+    {
+      "epoch": 0.97792,
+      "grad_norm": 44.350345611572266,
+      "learning_rate": 1.1356444444444446e-05,
+      "loss": 1.6466,
+      "step": 30560
+    },
+    {
+      "epoch": 0.97824,
+      "grad_norm": 45.596744537353516,
+      "learning_rate": 1.1352888888888889e-05,
+      "loss": 1.6226,
+      "step": 30570
+    },
+    {
+      "epoch": 0.97856,
+      "grad_norm": 43.384891510009766,
+      "learning_rate": 1.1349333333333335e-05,
+      "loss": 1.6647,
+      "step": 30580
+    },
+    {
+      "epoch": 0.97888,
+      "grad_norm": 44.46726608276367,
+      "learning_rate": 1.1345777777777778e-05,
+      "loss": 1.6085,
+      "step": 30590
+    },
+    {
+      "epoch": 0.9792,
+      "grad_norm": 42.867740631103516,
+      "learning_rate": 1.1342222222222224e-05,
+      "loss": 1.6466,
+      "step": 30600
+    },
+    {
+      "epoch": 0.97952,
+      "grad_norm": 43.6484260559082,
+      "learning_rate": 1.1338666666666667e-05,
+      "loss": 1.6434,
+      "step": 30610
+    },
+    {
+      "epoch": 0.97984,
+      "grad_norm": 42.653663635253906,
+      "learning_rate": 1.1335111111111112e-05,
+      "loss": 1.6357,
+      "step": 30620
+    },
+    {
+      "epoch": 0.98016,
+      "grad_norm": 44.374412536621094,
+      "learning_rate": 1.1331555555555556e-05,
+      "loss": 1.6397,
+      "step": 30630
+    },
+    {
+      "epoch": 0.98048,
+      "grad_norm": 44.84930419921875,
+      "learning_rate": 1.1328e-05,
+      "loss": 1.6097,
+      "step": 30640
+    },
+    {
+      "epoch": 0.9808,
+      "grad_norm": 43.1632194519043,
+      "learning_rate": 1.1324444444444445e-05,
+      "loss": 1.624,
+      "step": 30650
+    },
+    {
+      "epoch": 0.98112,
+      "grad_norm": 44.800209045410156,
+      "learning_rate": 1.132088888888889e-05,
+      "loss": 1.6803,
+      "step": 30660
+    },
+    {
+      "epoch": 0.98144,
+      "grad_norm": 44.4241943359375,
+      "learning_rate": 1.1317333333333334e-05,
+      "loss": 1.6252,
+      "step": 30670
+    },
+    {
+      "epoch": 0.98176,
+      "grad_norm": 43.48160934448242,
+      "learning_rate": 1.1313777777777779e-05,
+      "loss": 1.622,
+      "step": 30680
+    },
+    {
+      "epoch": 0.98208,
+      "grad_norm": 42.53988265991211,
+      "learning_rate": 1.1310222222222222e-05,
+      "loss": 1.6255,
+      "step": 30690
+    },
+    {
+      "epoch": 0.9824,
+      "grad_norm": 43.95156478881836,
+      "learning_rate": 1.1306666666666668e-05,
+      "loss": 1.6437,
+      "step": 30700
+    },
+    {
+      "epoch": 0.98272,
+      "grad_norm": 42.917320251464844,
+      "learning_rate": 1.130311111111111e-05,
+      "loss": 1.5937,
+      "step": 30710
+    },
+    {
+      "epoch": 0.98304,
+      "grad_norm": 44.980560302734375,
+      "learning_rate": 1.1299555555555557e-05,
+      "loss": 1.6841,
+      "step": 30720
+    },
+    {
+      "epoch": 0.98336,
+      "grad_norm": 45.12786102294922,
+      "learning_rate": 1.1296e-05,
+      "loss": 1.6192,
+      "step": 30730
+    },
+    {
+      "epoch": 0.98368,
+      "grad_norm": 43.9362678527832,
+      "learning_rate": 1.1292444444444446e-05,
+      "loss": 1.6137,
+      "step": 30740
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 43.08484649658203,
+      "learning_rate": 1.1288888888888889e-05,
+      "loss": 1.6136,
+      "step": 30750
+    },
+    {
+      "epoch": 0.98432,
+      "grad_norm": 44.23360061645508,
+      "learning_rate": 1.1285333333333335e-05,
+      "loss": 1.6281,
+      "step": 30760
+    },
+    {
+      "epoch": 0.98464,
+      "grad_norm": 43.02987289428711,
+      "learning_rate": 1.1281777777777778e-05,
+      "loss": 1.6106,
+      "step": 30770
+    },
+    {
+      "epoch": 0.98496,
+      "grad_norm": 42.354278564453125,
+      "learning_rate": 1.1278222222222224e-05,
+      "loss": 1.6278,
+      "step": 30780
+    },
+    {
+      "epoch": 0.98528,
+      "grad_norm": 43.63727951049805,
+      "learning_rate": 1.1274666666666667e-05,
+      "loss": 1.624,
+      "step": 30790
+    },
+    {
+      "epoch": 0.9856,
+      "grad_norm": 44.27710723876953,
+      "learning_rate": 1.1271111111111113e-05,
+      "loss": 1.6167,
+      "step": 30800
+    },
+    {
+      "epoch": 0.98592,
+      "grad_norm": 46.397186279296875,
+      "learning_rate": 1.1267555555555556e-05,
+      "loss": 1.6536,
+      "step": 30810
+    },
+    {
+      "epoch": 0.98624,
+      "grad_norm": 44.01591110229492,
+      "learning_rate": 1.1264000000000001e-05,
+      "loss": 1.5924,
+      "step": 30820
+    },
+    {
+      "epoch": 0.98656,
+      "grad_norm": 44.404296875,
+      "learning_rate": 1.1260444444444445e-05,
+      "loss": 1.6529,
+      "step": 30830
+    },
+    {
+      "epoch": 0.98688,
+      "grad_norm": 45.859962463378906,
+      "learning_rate": 1.125688888888889e-05,
+      "loss": 1.6152,
+      "step": 30840
+    },
+    {
+      "epoch": 0.9872,
+      "grad_norm": 45.23036193847656,
+      "learning_rate": 1.1253333333333335e-05,
+      "loss": 1.6502,
+      "step": 30850
+    },
+    {
+      "epoch": 0.98752,
+      "grad_norm": 44.69296646118164,
+      "learning_rate": 1.1249777777777779e-05,
+      "loss": 1.6376,
+      "step": 30860
+    },
+    {
+      "epoch": 0.98784,
+      "grad_norm": 45.085567474365234,
+      "learning_rate": 1.1246222222222222e-05,
+      "loss": 1.6334,
+      "step": 30870
+    },
+    {
+      "epoch": 0.98816,
+      "grad_norm": 44.13175582885742,
+      "learning_rate": 1.1242666666666668e-05,
+      "loss": 1.6392,
+      "step": 30880
+    },
+    {
+      "epoch": 0.98848,
+      "grad_norm": 45.575565338134766,
+      "learning_rate": 1.1239111111111111e-05,
+      "loss": 1.6456,
+      "step": 30890
+    },
+    {
+      "epoch": 0.9888,
+      "grad_norm": 461.7893371582031,
+      "learning_rate": 1.1235555555555557e-05,
+      "loss": 1.6544,
+      "step": 30900
+    },
+    {
+      "epoch": 0.98912,
+      "grad_norm": 44.190555572509766,
+      "learning_rate": 1.1232e-05,
+      "loss": 1.6594,
+      "step": 30910
+    },
+    {
+      "epoch": 0.98944,
+      "grad_norm": 46.13947677612305,
+      "learning_rate": 1.1228444444444446e-05,
+      "loss": 1.6127,
+      "step": 30920
+    },
+    {
+      "epoch": 0.98976,
+      "grad_norm": 44.94755554199219,
+      "learning_rate": 1.122488888888889e-05,
+      "loss": 1.6411,
+      "step": 30930
+    },
+    {
+      "epoch": 0.99008,
+      "grad_norm": 44.4129524230957,
+      "learning_rate": 1.1221333333333336e-05,
+      "loss": 1.6209,
+      "step": 30940
+    },
+    {
+      "epoch": 0.9904,
+      "grad_norm": 44.70603942871094,
+      "learning_rate": 1.1217777777777778e-05,
+      "loss": 1.6358,
+      "step": 30950
+    },
+    {
+      "epoch": 0.99072,
+      "grad_norm": 45.2507209777832,
+      "learning_rate": 1.1214222222222225e-05,
+      "loss": 1.651,
+      "step": 30960
+    },
+    {
+      "epoch": 0.99104,
+      "grad_norm": 45.008705139160156,
+      "learning_rate": 1.1210666666666668e-05,
+      "loss": 1.6225,
+      "step": 30970
+    },
+    {
+      "epoch": 0.99136,
+      "grad_norm": 45.01123046875,
+      "learning_rate": 1.1207111111111114e-05,
+      "loss": 1.6806,
+      "step": 30980
+    },
+    {
+      "epoch": 0.99168,
+      "grad_norm": 45.39712142944336,
+      "learning_rate": 1.1203555555555557e-05,
+      "loss": 1.6466,
+      "step": 30990
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 43.38059616088867,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.6306,
+      "step": 31000
+    },
+    {
+      "epoch": 0.99232,
+      "grad_norm": 44.428375244140625,
+      "learning_rate": 1.1196444444444446e-05,
+      "loss": 1.6364,
+      "step": 31010
+    },
+    {
+      "epoch": 0.99264,
+      "grad_norm": 44.255958557128906,
+      "learning_rate": 1.119288888888889e-05,
+      "loss": 1.6408,
+      "step": 31020
+    },
+    {
+      "epoch": 0.99296,
+      "grad_norm": 43.42549514770508,
+      "learning_rate": 1.1189333333333335e-05,
+      "loss": 1.6308,
+      "step": 31030
+    },
+    {
+      "epoch": 0.99328,
+      "grad_norm": 43.82905197143555,
+      "learning_rate": 1.1185777777777778e-05,
+      "loss": 1.6463,
+      "step": 31040
+    },
+    {
+      "epoch": 0.9936,
+      "grad_norm": 43.67567443847656,
+      "learning_rate": 1.1182222222222224e-05,
+      "loss": 1.651,
+      "step": 31050
+    },
+    {
+      "epoch": 0.99392,
+      "grad_norm": 43.85862350463867,
+      "learning_rate": 1.1178666666666667e-05,
+      "loss": 1.6369,
+      "step": 31060
+    },
+    {
+      "epoch": 0.99424,
+      "grad_norm": 45.972511291503906,
+      "learning_rate": 1.1175111111111111e-05,
+      "loss": 1.6217,
+      "step": 31070
+    },
+    {
+      "epoch": 0.99456,
+      "grad_norm": 44.29342269897461,
+      "learning_rate": 1.1171555555555556e-05,
+      "loss": 1.6585,
+      "step": 31080
+    },
+    {
+      "epoch": 0.99488,
+      "grad_norm": 45.08591079711914,
+      "learning_rate": 1.1168e-05,
+      "loss": 1.6769,
+      "step": 31090
+    },
+    {
+      "epoch": 0.9952,
+      "grad_norm": 45.8797721862793,
+      "learning_rate": 1.1164444444444445e-05,
+      "loss": 1.654,
+      "step": 31100
+    },
+    {
+      "epoch": 0.99552,
+      "grad_norm": 46.71592712402344,
+      "learning_rate": 1.116088888888889e-05,
+      "loss": 1.5952,
+      "step": 31110
+    },
+    {
+      "epoch": 0.99584,
+      "grad_norm": 42.47768020629883,
+      "learning_rate": 1.1157333333333334e-05,
+      "loss": 1.608,
+      "step": 31120
+    },
+    {
+      "epoch": 0.99616,
+      "grad_norm": 46.937469482421875,
+      "learning_rate": 1.1153777777777779e-05,
+      "loss": 1.6215,
+      "step": 31130
+    },
+    {
+      "epoch": 0.99648,
+      "grad_norm": 46.01224899291992,
+      "learning_rate": 1.1150222222222222e-05,
+      "loss": 1.6357,
+      "step": 31140
+    },
+    {
+      "epoch": 0.9968,
+      "grad_norm": 44.50090026855469,
+      "learning_rate": 1.1146666666666668e-05,
+      "loss": 1.6372,
+      "step": 31150
+    },
+    {
+      "epoch": 0.99712,
+      "grad_norm": 44.25981140136719,
+      "learning_rate": 1.114311111111111e-05,
+      "loss": 1.6213,
+      "step": 31160
+    },
+    {
+      "epoch": 0.99744,
+      "grad_norm": 47.06044006347656,
+      "learning_rate": 1.1139555555555557e-05,
+      "loss": 1.6116,
+      "step": 31170
+    },
+    {
+      "epoch": 0.99776,
+      "grad_norm": 45.36325454711914,
+      "learning_rate": 1.1136e-05,
+      "loss": 1.6098,
+      "step": 31180
+    },
+    {
+      "epoch": 0.99808,
+      "grad_norm": 43.87416076660156,
+      "learning_rate": 1.1132444444444446e-05,
+      "loss": 1.6587,
+      "step": 31190
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 45.90060806274414,
+      "learning_rate": 1.1128888888888889e-05,
+      "loss": 1.6715,
+      "step": 31200
+    },
+    {
+      "epoch": 0.99872,
+      "grad_norm": 44.73828125,
+      "learning_rate": 1.1125333333333335e-05,
+      "loss": 1.6297,
+      "step": 31210
+    },
+    {
+      "epoch": 0.99904,
+      "grad_norm": 45.52794647216797,
+      "learning_rate": 1.1121777777777778e-05,
+      "loss": 1.655,
+      "step": 31220
+    },
+    {
+      "epoch": 0.99936,
+      "grad_norm": 44.59988784790039,
+      "learning_rate": 1.1118222222222224e-05,
+      "loss": 1.6252,
+      "step": 31230
+    },
+    {
+      "epoch": 0.99968,
+      "grad_norm": 45.505130767822266,
+      "learning_rate": 1.1114666666666667e-05,
+      "loss": 1.6094,
+      "step": 31240
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 43.133392333984375,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.6227,
+      "step": 31250
+    },
+    {
+      "epoch": 1.00032,
+      "grad_norm": 46.064205169677734,
+      "learning_rate": 1.1107555555555556e-05,
+      "loss": 1.6462,
+      "step": 31260
+    },
+    {
+      "epoch": 1.00064,
+      "grad_norm": 43.47576904296875,
+      "learning_rate": 1.1104e-05,
+      "loss": 1.6394,
+      "step": 31270
+    },
+    {
+      "epoch": 1.00096,
+      "grad_norm": 44.37240219116211,
+      "learning_rate": 1.1100444444444445e-05,
+      "loss": 1.68,
+      "step": 31280
+    },
+    {
+      "epoch": 1.00128,
+      "grad_norm": 46.296592712402344,
+      "learning_rate": 1.109688888888889e-05,
+      "loss": 1.6122,
+      "step": 31290
+    },
+    {
+      "epoch": 1.0016,
+      "grad_norm": 45.33930206298828,
+      "learning_rate": 1.1093333333333334e-05,
+      "loss": 1.6252,
+      "step": 31300
+    },
+    {
+      "epoch": 1.00192,
+      "grad_norm": 45.15934753417969,
+      "learning_rate": 1.1089777777777779e-05,
+      "loss": 1.679,
+      "step": 31310
+    },
+    {
+      "epoch": 1.00224,
+      "grad_norm": 43.59319305419922,
+      "learning_rate": 1.1086222222222222e-05,
+      "loss": 1.6413,
+      "step": 31320
+    },
+    {
+      "epoch": 1.00256,
+      "grad_norm": 44.4049072265625,
+      "learning_rate": 1.1082666666666668e-05,
+      "loss": 1.6278,
+      "step": 31330
+    },
+    {
+      "epoch": 1.00288,
+      "grad_norm": 62.05976486206055,
+      "learning_rate": 1.1079111111111111e-05,
+      "loss": 1.6498,
+      "step": 31340
+    },
+    {
+      "epoch": 1.0032,
+      "grad_norm": 45.33911895751953,
+      "learning_rate": 1.1075555555555557e-05,
+      "loss": 1.6542,
+      "step": 31350
+    },
+    {
+      "epoch": 1.00352,
+      "grad_norm": 42.654972076416016,
+      "learning_rate": 1.1072e-05,
+      "loss": 1.6216,
+      "step": 31360
+    },
+    {
+      "epoch": 1.00384,
+      "grad_norm": 45.3847770690918,
+      "learning_rate": 1.1068444444444446e-05,
+      "loss": 1.6225,
+      "step": 31370
+    },
+    {
+      "epoch": 1.00416,
+      "grad_norm": 43.77812576293945,
+      "learning_rate": 1.1064888888888889e-05,
+      "loss": 1.6216,
+      "step": 31380
+    },
+    {
+      "epoch": 1.00448,
+      "grad_norm": 43.009376525878906,
+      "learning_rate": 1.1061333333333335e-05,
+      "loss": 1.6285,
+      "step": 31390
+    },
+    {
+      "epoch": 1.0048,
+      "grad_norm": 43.0247802734375,
+      "learning_rate": 1.1057777777777778e-05,
+      "loss": 1.6343,
+      "step": 31400
+    },
+    {
+      "epoch": 1.00512,
+      "grad_norm": 45.1480827331543,
+      "learning_rate": 1.1054222222222225e-05,
+      "loss": 1.6199,
+      "step": 31410
+    },
+    {
+      "epoch": 1.0054400000000001,
+      "grad_norm": 42.25900650024414,
+      "learning_rate": 1.1050666666666667e-05,
+      "loss": 1.6211,
+      "step": 31420
+    },
+    {
+      "epoch": 1.00576,
+      "grad_norm": 44.692604064941406,
+      "learning_rate": 1.1047111111111114e-05,
+      "loss": 1.6498,
+      "step": 31430
+    },
+    {
+      "epoch": 1.00608,
+      "grad_norm": 44.6058235168457,
+      "learning_rate": 1.1043555555555556e-05,
+      "loss": 1.6691,
+      "step": 31440
+    },
+    {
+      "epoch": 1.0064,
+      "grad_norm": 43.9179573059082,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 1.6219,
+      "step": 31450
+    },
+    {
+      "epoch": 1.00672,
+      "grad_norm": 42.16164779663086,
+      "learning_rate": 1.1036444444444446e-05,
+      "loss": 1.6215,
+      "step": 31460
+    },
+    {
+      "epoch": 1.00704,
+      "grad_norm": 44.84690856933594,
+      "learning_rate": 1.103288888888889e-05,
+      "loss": 1.6117,
+      "step": 31470
+    },
+    {
+      "epoch": 1.00736,
+      "grad_norm": 44.58945083618164,
+      "learning_rate": 1.1029333333333335e-05,
+      "loss": 1.6571,
+      "step": 31480
+    },
+    {
+      "epoch": 1.00768,
+      "grad_norm": 43.47794723510742,
+      "learning_rate": 1.102577777777778e-05,
+      "loss": 1.6731,
+      "step": 31490
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 44.99798583984375,
+      "learning_rate": 1.1022222222222224e-05,
+      "loss": 1.6104,
+      "step": 31500
+    },
+    {
+      "epoch": 1.00832,
+      "grad_norm": 44.819461822509766,
+      "learning_rate": 1.1018666666666668e-05,
+      "loss": 1.6118,
+      "step": 31510
+    },
+    {
+      "epoch": 1.00864,
+      "grad_norm": 44.44795608520508,
+      "learning_rate": 1.1015111111111111e-05,
+      "loss": 1.6252,
+      "step": 31520
+    },
+    {
+      "epoch": 1.00896,
+      "grad_norm": 45.462520599365234,
+      "learning_rate": 1.1011555555555557e-05,
+      "loss": 1.6215,
+      "step": 31530
+    },
+    {
+      "epoch": 1.00928,
+      "grad_norm": 44.72943878173828,
+      "learning_rate": 1.1008e-05,
+      "loss": 1.6085,
+      "step": 31540
+    },
+    {
+      "epoch": 1.0096,
+      "grad_norm": 42.78067398071289,
+      "learning_rate": 1.1004444444444445e-05,
+      "loss": 1.622,
+      "step": 31550
+    },
+    {
+      "epoch": 1.00992,
+      "grad_norm": 44.57320785522461,
+      "learning_rate": 1.100088888888889e-05,
+      "loss": 1.6739,
+      "step": 31560
+    },
+    {
+      "epoch": 1.01024,
+      "grad_norm": 46.82978439331055,
+      "learning_rate": 1.0997333333333334e-05,
+      "loss": 1.6334,
+      "step": 31570
+    },
+    {
+      "epoch": 1.01056,
+      "grad_norm": 44.21915817260742,
+      "learning_rate": 1.0993777777777779e-05,
+      "loss": 1.6436,
+      "step": 31580
+    },
+    {
+      "epoch": 1.01088,
+      "grad_norm": 44.011409759521484,
+      "learning_rate": 1.0990222222222221e-05,
+      "loss": 1.6548,
+      "step": 31590
+    },
+    {
+      "epoch": 1.0112,
+      "grad_norm": 46.1209716796875,
+      "learning_rate": 1.0986666666666668e-05,
+      "loss": 1.6127,
+      "step": 31600
+    },
+    {
+      "epoch": 1.01152,
+      "grad_norm": 43.016326904296875,
+      "learning_rate": 1.098311111111111e-05,
+      "loss": 1.6074,
+      "step": 31610
+    },
+    {
+      "epoch": 1.01184,
+      "grad_norm": 46.83198928833008,
+      "learning_rate": 1.0979555555555557e-05,
+      "loss": 1.6347,
+      "step": 31620
+    },
+    {
+      "epoch": 1.01216,
+      "grad_norm": 44.099212646484375,
+      "learning_rate": 1.0976e-05,
+      "loss": 1.6203,
+      "step": 31630
+    },
+    {
+      "epoch": 1.01248,
+      "grad_norm": 44.47719955444336,
+      "learning_rate": 1.0972444444444446e-05,
+      "loss": 1.6414,
+      "step": 31640
+    },
+    {
+      "epoch": 1.0128,
+      "grad_norm": 44.10932922363281,
+      "learning_rate": 1.0968888888888889e-05,
+      "loss": 1.6289,
+      "step": 31650
+    },
+    {
+      "epoch": 1.01312,
+      "grad_norm": 42.07052230834961,
+      "learning_rate": 1.0965333333333335e-05,
+      "loss": 1.6272,
+      "step": 31660
+    },
+    {
+      "epoch": 1.01344,
+      "grad_norm": 47.4359130859375,
+      "learning_rate": 1.0961777777777778e-05,
+      "loss": 1.6344,
+      "step": 31670
+    },
+    {
+      "epoch": 1.01376,
+      "grad_norm": 47.179473876953125,
+      "learning_rate": 1.0958222222222224e-05,
+      "loss": 1.6272,
+      "step": 31680
+    },
+    {
+      "epoch": 1.01408,
+      "grad_norm": 43.768314361572266,
+      "learning_rate": 1.0954666666666667e-05,
+      "loss": 1.6009,
+      "step": 31690
+    },
+    {
+      "epoch": 1.0144,
+      "grad_norm": 45.94318771362305,
+      "learning_rate": 1.0951111111111113e-05,
+      "loss": 1.6192,
+      "step": 31700
+    },
+    {
+      "epoch": 1.01472,
+      "grad_norm": 44.83135223388672,
+      "learning_rate": 1.0947555555555556e-05,
+      "loss": 1.6039,
+      "step": 31710
+    },
+    {
+      "epoch": 1.01504,
+      "grad_norm": 45.00038528442383,
+      "learning_rate": 1.0944e-05,
+      "loss": 1.6236,
+      "step": 31720
+    },
+    {
+      "epoch": 1.01536,
+      "grad_norm": 45.91115951538086,
+      "learning_rate": 1.0940444444444445e-05,
+      "loss": 1.6407,
+      "step": 31730
+    },
+    {
+      "epoch": 1.01568,
+      "grad_norm": 43.440773010253906,
+      "learning_rate": 1.093688888888889e-05,
+      "loss": 1.6389,
+      "step": 31740
+    },
+    {
+      "epoch": 1.016,
+      "grad_norm": 47.18890380859375,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 1.6178,
+      "step": 31750
+    },
+    {
+      "epoch": 1.01632,
+      "grad_norm": 44.28126525878906,
+      "learning_rate": 1.0929777777777779e-05,
+      "loss": 1.6278,
+      "step": 31760
+    },
+    {
+      "epoch": 1.01664,
+      "grad_norm": 43.794429779052734,
+      "learning_rate": 1.0926222222222222e-05,
+      "loss": 1.6526,
+      "step": 31770
+    },
+    {
+      "epoch": 1.01696,
+      "grad_norm": 44.310943603515625,
+      "learning_rate": 1.0922666666666668e-05,
+      "loss": 1.6271,
+      "step": 31780
+    },
+    {
+      "epoch": 1.01728,
+      "grad_norm": 42.87227249145508,
+      "learning_rate": 1.091911111111111e-05,
+      "loss": 1.629,
+      "step": 31790
+    },
+    {
+      "epoch": 1.0176,
+      "grad_norm": 42.844871520996094,
+      "learning_rate": 1.0915555555555557e-05,
+      "loss": 1.6304,
+      "step": 31800
+    },
+    {
+      "epoch": 1.01792,
+      "grad_norm": 45.118019104003906,
+      "learning_rate": 1.0912e-05,
+      "loss": 1.6531,
+      "step": 31810
+    },
+    {
+      "epoch": 1.01824,
+      "grad_norm": 43.47876739501953,
+      "learning_rate": 1.0908444444444446e-05,
+      "loss": 1.6381,
+      "step": 31820
+    },
+    {
+      "epoch": 1.01856,
+      "grad_norm": 43.193843841552734,
+      "learning_rate": 1.0904888888888889e-05,
+      "loss": 1.6264,
+      "step": 31830
+    },
+    {
+      "epoch": 1.01888,
+      "grad_norm": 42.943599700927734,
+      "learning_rate": 1.0901333333333335e-05,
+      "loss": 1.6012,
+      "step": 31840
+    },
+    {
+      "epoch": 1.0192,
+      "grad_norm": 45.247562408447266,
+      "learning_rate": 1.0897777777777778e-05,
+      "loss": 1.6206,
+      "step": 31850
+    },
+    {
+      "epoch": 1.01952,
+      "grad_norm": 44.1955680847168,
+      "learning_rate": 1.0894222222222224e-05,
+      "loss": 1.6449,
+      "step": 31860
+    },
+    {
+      "epoch": 1.01984,
+      "grad_norm": 48.465232849121094,
+      "learning_rate": 1.0890666666666667e-05,
+      "loss": 1.6447,
+      "step": 31870
+    },
+    {
+      "epoch": 1.02016,
+      "grad_norm": 43.41231918334961,
+      "learning_rate": 1.0887111111111113e-05,
+      "loss": 1.6501,
+      "step": 31880
+    },
+    {
+      "epoch": 1.02048,
+      "grad_norm": 46.5025520324707,
+      "learning_rate": 1.0883555555555556e-05,
+      "loss": 1.6259,
+      "step": 31890
+    },
+    {
+      "epoch": 1.0208,
+      "grad_norm": 45.25193786621094,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 1.5989,
+      "step": 31900
+    },
+    {
+      "epoch": 1.02112,
+      "grad_norm": 45.444549560546875,
+      "learning_rate": 1.0876444444444445e-05,
+      "loss": 1.6214,
+      "step": 31910
+    },
+    {
+      "epoch": 1.02144,
+      "grad_norm": 44.449981689453125,
+      "learning_rate": 1.087288888888889e-05,
+      "loss": 1.6275,
+      "step": 31920
+    },
+    {
+      "epoch": 1.02176,
+      "grad_norm": 43.29909896850586,
+      "learning_rate": 1.0869333333333335e-05,
+      "loss": 1.603,
+      "step": 31930
+    },
+    {
+      "epoch": 1.02208,
+      "grad_norm": 43.329078674316406,
+      "learning_rate": 1.0865777777777779e-05,
+      "loss": 1.646,
+      "step": 31940
+    },
+    {
+      "epoch": 1.0224,
+      "grad_norm": 43.6357421875,
+      "learning_rate": 1.0862222222222224e-05,
+      "loss": 1.654,
+      "step": 31950
+    },
+    {
+      "epoch": 1.02272,
+      "grad_norm": 44.786800384521484,
+      "learning_rate": 1.0858666666666668e-05,
+      "loss": 1.6129,
+      "step": 31960
+    },
+    {
+      "epoch": 1.02304,
+      "grad_norm": 44.97706604003906,
+      "learning_rate": 1.0855111111111111e-05,
+      "loss": 1.6147,
+      "step": 31970
+    },
+    {
+      "epoch": 1.02336,
+      "grad_norm": 44.8741340637207,
+      "learning_rate": 1.0851555555555557e-05,
+      "loss": 1.6381,
+      "step": 31980
+    },
+    {
+      "epoch": 1.02368,
+      "grad_norm": 45.01078796386719,
+      "learning_rate": 1.0848e-05,
+      "loss": 1.6143,
+      "step": 31990
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 44.32587814331055,
+      "learning_rate": 1.0844444444444446e-05,
+      "loss": 1.6145,
+      "step": 32000
+    },
+    {
+      "epoch": 1.02432,
+      "grad_norm": 44.67451477050781,
+      "learning_rate": 1.084088888888889e-05,
+      "loss": 1.6423,
+      "step": 32010
+    },
+    {
+      "epoch": 1.02464,
+      "grad_norm": 43.36486053466797,
+      "learning_rate": 1.0837333333333336e-05,
+      "loss": 1.6254,
+      "step": 32020
+    },
+    {
+      "epoch": 1.02496,
+      "grad_norm": 44.7286376953125,
+      "learning_rate": 1.0833777777777778e-05,
+      "loss": 1.6588,
+      "step": 32030
+    },
+    {
+      "epoch": 1.02528,
+      "grad_norm": 42.48271560668945,
+      "learning_rate": 1.0830222222222225e-05,
+      "loss": 1.6242,
+      "step": 32040
+    },
+    {
+      "epoch": 1.0256,
+      "grad_norm": 44.96989440917969,
+      "learning_rate": 1.0826666666666667e-05,
+      "loss": 1.6284,
+      "step": 32050
+    },
+    {
+      "epoch": 1.02592,
+      "grad_norm": 42.45598220825195,
+      "learning_rate": 1.082311111111111e-05,
+      "loss": 1.6255,
+      "step": 32060
+    },
+    {
+      "epoch": 1.02624,
+      "grad_norm": 44.257591247558594,
+      "learning_rate": 1.0819555555555557e-05,
+      "loss": 1.6006,
+      "step": 32070
+    },
+    {
+      "epoch": 1.02656,
+      "grad_norm": 43.706756591796875,
+      "learning_rate": 1.0816e-05,
+      "loss": 1.6175,
+      "step": 32080
+    },
+    {
+      "epoch": 1.02688,
+      "grad_norm": 43.40682601928711,
+      "learning_rate": 1.0812444444444446e-05,
+      "loss": 1.6196,
+      "step": 32090
+    },
+    {
+      "epoch": 1.0272,
+      "grad_norm": 46.219940185546875,
+      "learning_rate": 1.0808888888888889e-05,
+      "loss": 1.6259,
+      "step": 32100
+    },
+    {
+      "epoch": 1.02752,
+      "grad_norm": 43.60252380371094,
+      "learning_rate": 1.0805333333333335e-05,
+      "loss": 1.6402,
+      "step": 32110
+    },
+    {
+      "epoch": 1.02784,
+      "grad_norm": 43.27579879760742,
+      "learning_rate": 1.0801777777777778e-05,
+      "loss": 1.6249,
+      "step": 32120
+    },
+    {
+      "epoch": 1.02816,
+      "grad_norm": 46.17375946044922,
+      "learning_rate": 1.0798222222222224e-05,
+      "loss": 1.6575,
+      "step": 32130
+    },
+    {
+      "epoch": 1.02848,
+      "grad_norm": 44.743988037109375,
+      "learning_rate": 1.0794666666666667e-05,
+      "loss": 1.6429,
+      "step": 32140
+    },
+    {
+      "epoch": 1.0288,
+      "grad_norm": 44.23445510864258,
+      "learning_rate": 1.0791111111111113e-05,
+      "loss": 1.6322,
+      "step": 32150
+    },
+    {
+      "epoch": 1.02912,
+      "grad_norm": 41.932987213134766,
+      "learning_rate": 1.0787555555555556e-05,
+      "loss": 1.6479,
+      "step": 32160
+    },
+    {
+      "epoch": 1.02944,
+      "grad_norm": 42.20267105102539,
+      "learning_rate": 1.0784e-05,
+      "loss": 1.6139,
+      "step": 32170
+    },
+    {
+      "epoch": 1.02976,
+      "grad_norm": 44.15998077392578,
+      "learning_rate": 1.0780444444444445e-05,
+      "loss": 1.6416,
+      "step": 32180
+    },
+    {
+      "epoch": 1.03008,
+      "grad_norm": 45.692970275878906,
+      "learning_rate": 1.077688888888889e-05,
+      "loss": 1.637,
+      "step": 32190
+    },
+    {
+      "epoch": 1.0304,
+      "grad_norm": 45.96401596069336,
+      "learning_rate": 1.0773333333333334e-05,
+      "loss": 1.6245,
+      "step": 32200
+    },
+    {
+      "epoch": 1.03072,
+      "grad_norm": 43.48420333862305,
+      "learning_rate": 1.0769777777777779e-05,
+      "loss": 1.6248,
+      "step": 32210
+    },
+    {
+      "epoch": 1.03104,
+      "grad_norm": 44.58662414550781,
+      "learning_rate": 1.0766222222222223e-05,
+      "loss": 1.6376,
+      "step": 32220
+    },
+    {
+      "epoch": 1.03136,
+      "grad_norm": 44.91218948364258,
+      "learning_rate": 1.0762666666666668e-05,
+      "loss": 1.6193,
+      "step": 32230
+    },
+    {
+      "epoch": 1.03168,
+      "grad_norm": 43.115234375,
+      "learning_rate": 1.075911111111111e-05,
+      "loss": 1.6597,
+      "step": 32240
+    },
+    {
+      "epoch": 1.032,
+      "grad_norm": 46.89992141723633,
+      "learning_rate": 1.0755555555555557e-05,
+      "loss": 1.6303,
+      "step": 32250
+    },
+    {
+      "epoch": 1.03232,
+      "grad_norm": 42.8564453125,
+      "learning_rate": 1.0752e-05,
+      "loss": 1.604,
+      "step": 32260
+    },
+    {
+      "epoch": 1.03264,
+      "grad_norm": 44.03951644897461,
+      "learning_rate": 1.0748444444444446e-05,
+      "loss": 1.5788,
+      "step": 32270
+    },
+    {
+      "epoch": 1.03296,
+      "grad_norm": 45.001678466796875,
+      "learning_rate": 1.0744888888888889e-05,
+      "loss": 1.6211,
+      "step": 32280
+    },
+    {
+      "epoch": 1.03328,
+      "grad_norm": 44.68024444580078,
+      "learning_rate": 1.0741333333333335e-05,
+      "loss": 1.6518,
+      "step": 32290
+    },
+    {
+      "epoch": 1.0336,
+      "grad_norm": 44.97967529296875,
+      "learning_rate": 1.0737777777777778e-05,
+      "loss": 1.6354,
+      "step": 32300
+    },
+    {
+      "epoch": 1.03392,
+      "grad_norm": 45.28949737548828,
+      "learning_rate": 1.0734222222222224e-05,
+      "loss": 1.6284,
+      "step": 32310
+    },
+    {
+      "epoch": 1.03424,
+      "grad_norm": 48.3412971496582,
+      "learning_rate": 1.0730666666666667e-05,
+      "loss": 1.654,
+      "step": 32320
+    },
+    {
+      "epoch": 1.03456,
+      "grad_norm": 45.43934631347656,
+      "learning_rate": 1.0727111111111113e-05,
+      "loss": 1.6313,
+      "step": 32330
+    },
+    {
+      "epoch": 1.03488,
+      "grad_norm": 44.49418640136719,
+      "learning_rate": 1.0723555555555556e-05,
+      "loss": 1.6009,
+      "step": 32340
+    },
+    {
+      "epoch": 1.0352,
+      "grad_norm": 43.252567291259766,
+      "learning_rate": 1.072e-05,
+      "loss": 1.6191,
+      "step": 32350
+    },
+    {
+      "epoch": 1.03552,
+      "grad_norm": 43.64153289794922,
+      "learning_rate": 1.0716444444444445e-05,
+      "loss": 1.6482,
+      "step": 32360
+    },
+    {
+      "epoch": 1.03584,
+      "grad_norm": 42.04774856567383,
+      "learning_rate": 1.071288888888889e-05,
+      "loss": 1.6464,
+      "step": 32370
+    },
+    {
+      "epoch": 1.03616,
+      "grad_norm": 44.96035385131836,
+      "learning_rate": 1.0709333333333334e-05,
+      "loss": 1.6204,
+      "step": 32380
+    },
+    {
+      "epoch": 1.03648,
+      "grad_norm": 44.43547058105469,
+      "learning_rate": 1.0705777777777779e-05,
+      "loss": 1.6563,
+      "step": 32390
+    },
+    {
+      "epoch": 1.0368,
+      "grad_norm": 44.755558013916016,
+      "learning_rate": 1.0702222222222223e-05,
+      "loss": 1.6514,
+      "step": 32400
+    },
+    {
+      "epoch": 1.03712,
+      "grad_norm": 43.75396728515625,
+      "learning_rate": 1.0698666666666668e-05,
+      "loss": 1.6043,
+      "step": 32410
+    },
+    {
+      "epoch": 1.03744,
+      "grad_norm": 46.955711364746094,
+      "learning_rate": 1.0695111111111111e-05,
+      "loss": 1.6138,
+      "step": 32420
+    },
+    {
+      "epoch": 1.03776,
+      "grad_norm": 44.48612594604492,
+      "learning_rate": 1.0691555555555557e-05,
+      "loss": 1.6539,
+      "step": 32430
+    },
+    {
+      "epoch": 1.03808,
+      "grad_norm": 45.34321594238281,
+      "learning_rate": 1.0688e-05,
+      "loss": 1.6247,
+      "step": 32440
+    },
+    {
+      "epoch": 1.0384,
+      "grad_norm": 42.34244155883789,
+      "learning_rate": 1.0684444444444446e-05,
+      "loss": 1.6117,
+      "step": 32450
+    },
+    {
+      "epoch": 1.03872,
+      "grad_norm": 44.84130096435547,
+      "learning_rate": 1.0680888888888889e-05,
+      "loss": 1.6138,
+      "step": 32460
+    },
+    {
+      "epoch": 1.03904,
+      "grad_norm": 45.60774612426758,
+      "learning_rate": 1.0677333333333335e-05,
+      "loss": 1.6179,
+      "step": 32470
+    },
+    {
+      "epoch": 1.03936,
+      "grad_norm": 42.9847412109375,
+      "learning_rate": 1.0673777777777778e-05,
+      "loss": 1.6525,
+      "step": 32480
+    },
+    {
+      "epoch": 1.03968,
+      "grad_norm": 43.739463806152344,
+      "learning_rate": 1.0670222222222224e-05,
+      "loss": 1.6358,
+      "step": 32490
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 44.78418731689453,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.6449,
+      "step": 32500
+    },
+    {
+      "epoch": 1.04032,
+      "grad_norm": 42.9957275390625,
+      "learning_rate": 1.0663111111111114e-05,
+      "loss": 1.6481,
+      "step": 32510
+    },
+    {
+      "epoch": 1.04064,
+      "grad_norm": 42.810630798339844,
+      "learning_rate": 1.0659555555555556e-05,
+      "loss": 1.6158,
+      "step": 32520
+    },
+    {
+      "epoch": 1.04096,
+      "grad_norm": 44.490386962890625,
+      "learning_rate": 1.0656000000000003e-05,
+      "loss": 1.6086,
+      "step": 32530
+    },
+    {
+      "epoch": 1.04128,
+      "grad_norm": 44.92964172363281,
+      "learning_rate": 1.0652444444444446e-05,
+      "loss": 1.6429,
+      "step": 32540
+    },
+    {
+      "epoch": 1.0416,
+      "grad_norm": 44.007816314697266,
+      "learning_rate": 1.064888888888889e-05,
+      "loss": 1.648,
+      "step": 32550
+    },
+    {
+      "epoch": 1.04192,
+      "grad_norm": 44.374534606933594,
+      "learning_rate": 1.0645333333333335e-05,
+      "loss": 1.6251,
+      "step": 32560
+    },
+    {
+      "epoch": 1.04224,
+      "grad_norm": 44.10312271118164,
+      "learning_rate": 1.0641777777777777e-05,
+      "loss": 1.6235,
+      "step": 32570
+    },
+    {
+      "epoch": 1.04256,
+      "grad_norm": 43.174400329589844,
+      "learning_rate": 1.0638222222222224e-05,
+      "loss": 1.6138,
+      "step": 32580
+    },
+    {
+      "epoch": 1.04288,
+      "grad_norm": 43.963924407958984,
+      "learning_rate": 1.0634666666666667e-05,
+      "loss": 1.6331,
+      "step": 32590
+    },
+    {
+      "epoch": 1.0432,
+      "grad_norm": 45.72218322753906,
+      "learning_rate": 1.0631111111111113e-05,
+      "loss": 1.6469,
+      "step": 32600
+    },
+    {
+      "epoch": 1.04352,
+      "grad_norm": 44.228797912597656,
+      "learning_rate": 1.0627555555555556e-05,
+      "loss": 1.623,
+      "step": 32610
+    },
+    {
+      "epoch": 1.04384,
+      "grad_norm": 43.118465423583984,
+      "learning_rate": 1.0624e-05,
+      "loss": 1.6266,
+      "step": 32620
+    },
+    {
+      "epoch": 1.04416,
+      "grad_norm": 44.9131965637207,
+      "learning_rate": 1.0620444444444445e-05,
+      "loss": 1.6052,
+      "step": 32630
+    },
+    {
+      "epoch": 1.04448,
+      "grad_norm": 41.5771369934082,
+      "learning_rate": 1.061688888888889e-05,
+      "loss": 1.6338,
+      "step": 32640
+    },
+    {
+      "epoch": 1.0448,
+      "grad_norm": 43.707706451416016,
+      "learning_rate": 1.0613333333333334e-05,
+      "loss": 1.6444,
+      "step": 32650
+    },
+    {
+      "epoch": 1.04512,
+      "grad_norm": 43.63424301147461,
+      "learning_rate": 1.0609777777777778e-05,
+      "loss": 1.6122,
+      "step": 32660
+    },
+    {
+      "epoch": 1.04544,
+      "grad_norm": 43.490196228027344,
+      "learning_rate": 1.0606222222222223e-05,
+      "loss": 1.5986,
+      "step": 32670
+    },
+    {
+      "epoch": 1.04576,
+      "grad_norm": 44.29290771484375,
+      "learning_rate": 1.0602666666666668e-05,
+      "loss": 1.6355,
+      "step": 32680
+    },
+    {
+      "epoch": 1.04608,
+      "grad_norm": 46.40851974487305,
+      "learning_rate": 1.059911111111111e-05,
+      "loss": 1.6291,
+      "step": 32690
+    },
+    {
+      "epoch": 1.0464,
+      "grad_norm": 43.00209426879883,
+      "learning_rate": 1.0595555555555557e-05,
+      "loss": 1.6312,
+      "step": 32700
+    },
+    {
+      "epoch": 1.04672,
+      "grad_norm": 46.49836349487305,
+      "learning_rate": 1.0592e-05,
+      "loss": 1.6316,
+      "step": 32710
+    },
+    {
+      "epoch": 1.04704,
+      "grad_norm": 43.06816482543945,
+      "learning_rate": 1.0588444444444446e-05,
+      "loss": 1.6084,
+      "step": 32720
+    },
+    {
+      "epoch": 1.04736,
+      "grad_norm": 44.20458221435547,
+      "learning_rate": 1.0584888888888889e-05,
+      "loss": 1.6038,
+      "step": 32730
+    },
+    {
+      "epoch": 1.04768,
+      "grad_norm": 45.879676818847656,
+      "learning_rate": 1.0581333333333335e-05,
+      "loss": 1.6372,
+      "step": 32740
+    },
+    {
+      "epoch": 1.048,
+      "grad_norm": 44.341094970703125,
+      "learning_rate": 1.0577777777777778e-05,
+      "loss": 1.639,
+      "step": 32750
+    },
+    {
+      "epoch": 1.04832,
+      "grad_norm": 43.37816619873047,
+      "learning_rate": 1.0574222222222224e-05,
+      "loss": 1.654,
+      "step": 32760
+    },
+    {
+      "epoch": 1.04864,
+      "grad_norm": 43.5418586730957,
+      "learning_rate": 1.0570666666666667e-05,
+      "loss": 1.6361,
+      "step": 32770
+    },
+    {
+      "epoch": 1.04896,
+      "grad_norm": 46.33805847167969,
+      "learning_rate": 1.0567111111111113e-05,
+      "loss": 1.6388,
+      "step": 32780
+    },
+    {
+      "epoch": 1.04928,
+      "grad_norm": 42.2393913269043,
+      "learning_rate": 1.0563555555555556e-05,
+      "loss": 1.6152,
+      "step": 32790
+    },
+    {
+      "epoch": 1.0496,
+      "grad_norm": 45.337345123291016,
+      "learning_rate": 1.056e-05,
+      "loss": 1.6162,
+      "step": 32800
+    },
+    {
+      "epoch": 1.04992,
+      "grad_norm": 44.33757781982422,
+      "learning_rate": 1.0556444444444445e-05,
+      "loss": 1.6181,
+      "step": 32810
+    },
+    {
+      "epoch": 1.05024,
+      "grad_norm": 43.743751525878906,
+      "learning_rate": 1.055288888888889e-05,
+      "loss": 1.6245,
+      "step": 32820
+    },
+    {
+      "epoch": 1.05056,
+      "grad_norm": 44.40175247192383,
+      "learning_rate": 1.0549333333333334e-05,
+      "loss": 1.6206,
+      "step": 32830
+    },
+    {
+      "epoch": 1.05088,
+      "grad_norm": 45.282310485839844,
+      "learning_rate": 1.0545777777777779e-05,
+      "loss": 1.6371,
+      "step": 32840
+    },
+    {
+      "epoch": 1.0512,
+      "grad_norm": 44.07647705078125,
+      "learning_rate": 1.0542222222222223e-05,
+      "loss": 1.6001,
+      "step": 32850
+    },
+    {
+      "epoch": 1.05152,
+      "grad_norm": 42.87550735473633,
+      "learning_rate": 1.0538666666666668e-05,
+      "loss": 1.6511,
+      "step": 32860
+    },
+    {
+      "epoch": 1.0518399999999999,
+      "grad_norm": 44.68036651611328,
+      "learning_rate": 1.053511111111111e-05,
+      "loss": 1.6109,
+      "step": 32870
+    },
+    {
+      "epoch": 1.05216,
+      "grad_norm": 45.50474548339844,
+      "learning_rate": 1.0531555555555557e-05,
+      "loss": 1.6316,
+      "step": 32880
+    },
+    {
+      "epoch": 1.05248,
+      "grad_norm": 45.01093673706055,
+      "learning_rate": 1.0528e-05,
+      "loss": 1.6129,
+      "step": 32890
+    },
+    {
+      "epoch": 1.0528,
+      "grad_norm": 42.464080810546875,
+      "learning_rate": 1.0524444444444446e-05,
+      "loss": 1.6246,
+      "step": 32900
+    },
+    {
+      "epoch": 1.05312,
+      "grad_norm": 44.48116683959961,
+      "learning_rate": 1.0520888888888889e-05,
+      "loss": 1.6577,
+      "step": 32910
+    },
+    {
+      "epoch": 1.05344,
+      "grad_norm": 44.605037689208984,
+      "learning_rate": 1.0517333333333335e-05,
+      "loss": 1.654,
+      "step": 32920
+    },
+    {
+      "epoch": 1.05376,
+      "grad_norm": 43.643428802490234,
+      "learning_rate": 1.0513777777777778e-05,
+      "loss": 1.6288,
+      "step": 32930
+    },
+    {
+      "epoch": 1.05408,
+      "grad_norm": 42.4962272644043,
+      "learning_rate": 1.0510222222222224e-05,
+      "loss": 1.6439,
+      "step": 32940
+    },
+    {
+      "epoch": 1.0544,
+      "grad_norm": 44.38206481933594,
+      "learning_rate": 1.0506666666666667e-05,
+      "loss": 1.6318,
+      "step": 32950
+    },
+    {
+      "epoch": 1.05472,
+      "grad_norm": 42.56404113769531,
+      "learning_rate": 1.0503111111111113e-05,
+      "loss": 1.6332,
+      "step": 32960
+    },
+    {
+      "epoch": 1.05504,
+      "grad_norm": 47.60989761352539,
+      "learning_rate": 1.0499555555555556e-05,
+      "loss": 1.6258,
+      "step": 32970
+    },
+    {
+      "epoch": 1.05536,
+      "grad_norm": 47.522117614746094,
+      "learning_rate": 1.0496000000000003e-05,
+      "loss": 1.621,
+      "step": 32980
+    },
+    {
+      "epoch": 1.05568,
+      "grad_norm": 44.03647232055664,
+      "learning_rate": 1.0492444444444445e-05,
+      "loss": 1.6203,
+      "step": 32990
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 42.845619201660156,
+      "learning_rate": 1.048888888888889e-05,
+      "loss": 1.5946,
+      "step": 33000
+    },
+    {
+      "epoch": 1.05632,
+      "grad_norm": 46.014015197753906,
+      "learning_rate": 1.0485333333333334e-05,
+      "loss": 1.6354,
+      "step": 33010
+    },
+    {
+      "epoch": 1.05664,
+      "grad_norm": 45.520111083984375,
+      "learning_rate": 1.0481777777777779e-05,
+      "loss": 1.6023,
+      "step": 33020
+    },
+    {
+      "epoch": 1.05696,
+      "grad_norm": 43.73442840576172,
+      "learning_rate": 1.0478222222222224e-05,
+      "loss": 1.6684,
+      "step": 33030
+    },
+    {
+      "epoch": 1.05728,
+      "grad_norm": 46.191837310791016,
+      "learning_rate": 1.0474666666666668e-05,
+      "loss": 1.6118,
+      "step": 33040
+    },
+    {
+      "epoch": 1.0576,
+      "grad_norm": 46.10258483886719,
+      "learning_rate": 1.0471111111111113e-05,
+      "loss": 1.6706,
+      "step": 33050
+    },
+    {
+      "epoch": 1.05792,
+      "grad_norm": 42.99847412109375,
+      "learning_rate": 1.0467555555555557e-05,
+      "loss": 1.6016,
+      "step": 33060
+    },
+    {
+      "epoch": 1.05824,
+      "grad_norm": 46.8404655456543,
+      "learning_rate": 1.0464e-05,
+      "loss": 1.6362,
+      "step": 33070
+    },
+    {
+      "epoch": 1.05856,
+      "grad_norm": 44.80493927001953,
+      "learning_rate": 1.0460444444444446e-05,
+      "loss": 1.644,
+      "step": 33080
+    },
+    {
+      "epoch": 1.05888,
+      "grad_norm": 45.339271545410156,
+      "learning_rate": 1.045688888888889e-05,
+      "loss": 1.6411,
+      "step": 33090
+    },
+    {
+      "epoch": 1.0592,
+      "grad_norm": 44.27571105957031,
+      "learning_rate": 1.0453333333333334e-05,
+      "loss": 1.6183,
+      "step": 33100
+    },
+    {
+      "epoch": 1.05952,
+      "grad_norm": 45.33119583129883,
+      "learning_rate": 1.0449777777777778e-05,
+      "loss": 1.6413,
+      "step": 33110
+    },
+    {
+      "epoch": 1.05984,
+      "grad_norm": 45.91305160522461,
+      "learning_rate": 1.0446222222222223e-05,
+      "loss": 1.6397,
+      "step": 33120
+    },
+    {
+      "epoch": 1.06016,
+      "grad_norm": 43.498374938964844,
+      "learning_rate": 1.0442666666666667e-05,
+      "loss": 1.5971,
+      "step": 33130
+    },
+    {
+      "epoch": 1.06048,
+      "grad_norm": 44.306419372558594,
+      "learning_rate": 1.043911111111111e-05,
+      "loss": 1.6197,
+      "step": 33140
+    },
+    {
+      "epoch": 1.0608,
+      "grad_norm": 44.896297454833984,
+      "learning_rate": 1.0435555555555557e-05,
+      "loss": 1.6165,
+      "step": 33150
+    },
+    {
+      "epoch": 1.06112,
+      "grad_norm": 45.88273620605469,
+      "learning_rate": 1.0432e-05,
+      "loss": 1.6174,
+      "step": 33160
+    },
+    {
+      "epoch": 1.06144,
+      "grad_norm": 45.505531311035156,
+      "learning_rate": 1.0428444444444446e-05,
+      "loss": 1.6441,
+      "step": 33170
+    },
+    {
+      "epoch": 1.06176,
+      "grad_norm": 46.69767761230469,
+      "learning_rate": 1.0424888888888888e-05,
+      "loss": 1.6241,
+      "step": 33180
+    },
+    {
+      "epoch": 1.06208,
+      "grad_norm": 43.01450729370117,
+      "learning_rate": 1.0421333333333335e-05,
+      "loss": 1.6151,
+      "step": 33190
+    },
+    {
+      "epoch": 1.0624,
+      "grad_norm": 44.89165115356445,
+      "learning_rate": 1.0417777777777778e-05,
+      "loss": 1.6233,
+      "step": 33200
+    },
+    {
+      "epoch": 1.06272,
+      "grad_norm": 44.70990753173828,
+      "learning_rate": 1.0414222222222224e-05,
+      "loss": 1.6288,
+      "step": 33210
+    },
+    {
+      "epoch": 1.06304,
+      "grad_norm": 42.99964141845703,
+      "learning_rate": 1.0410666666666667e-05,
+      "loss": 1.655,
+      "step": 33220
+    },
+    {
+      "epoch": 1.06336,
+      "grad_norm": 42.71984100341797,
+      "learning_rate": 1.0407111111111113e-05,
+      "loss": 1.6225,
+      "step": 33230
+    },
+    {
+      "epoch": 1.06368,
+      "grad_norm": 43.9577751159668,
+      "learning_rate": 1.0403555555555556e-05,
+      "loss": 1.5939,
+      "step": 33240
+    },
+    {
+      "epoch": 1.064,
+      "grad_norm": 43.84825134277344,
+      "learning_rate": 1.04e-05,
+      "loss": 1.6543,
+      "step": 33250
+    },
+    {
+      "epoch": 1.06432,
+      "grad_norm": 44.09382247924805,
+      "learning_rate": 1.0396444444444445e-05,
+      "loss": 1.6217,
+      "step": 33260
+    },
+    {
+      "epoch": 1.06464,
+      "grad_norm": 47.45444869995117,
+      "learning_rate": 1.039288888888889e-05,
+      "loss": 1.6282,
+      "step": 33270
+    },
+    {
+      "epoch": 1.06496,
+      "grad_norm": 43.27195739746094,
+      "learning_rate": 1.0389333333333334e-05,
+      "loss": 1.6359,
+      "step": 33280
+    },
+    {
+      "epoch": 1.06528,
+      "grad_norm": 44.726478576660156,
+      "learning_rate": 1.0385777777777779e-05,
+      "loss": 1.6212,
+      "step": 33290
+    },
+    {
+      "epoch": 1.0656,
+      "grad_norm": 45.426143646240234,
+      "learning_rate": 1.0382222222222223e-05,
+      "loss": 1.6219,
+      "step": 33300
+    },
+    {
+      "epoch": 1.06592,
+      "grad_norm": 45.67955780029297,
+      "learning_rate": 1.0378666666666668e-05,
+      "loss": 1.6726,
+      "step": 33310
+    },
+    {
+      "epoch": 1.06624,
+      "grad_norm": 44.08571243286133,
+      "learning_rate": 1.037511111111111e-05,
+      "loss": 1.6149,
+      "step": 33320
+    },
+    {
+      "epoch": 1.06656,
+      "grad_norm": 45.839317321777344,
+      "learning_rate": 1.0371555555555557e-05,
+      "loss": 1.6193,
+      "step": 33330
+    },
+    {
+      "epoch": 1.06688,
+      "grad_norm": 43.84119415283203,
+      "learning_rate": 1.0368e-05,
+      "loss": 1.6141,
+      "step": 33340
+    },
+    {
+      "epoch": 1.0672,
+      "grad_norm": 43.55615997314453,
+      "learning_rate": 1.0364444444444446e-05,
+      "loss": 1.6403,
+      "step": 33350
+    },
+    {
+      "epoch": 1.06752,
+      "grad_norm": 43.03118133544922,
+      "learning_rate": 1.0360888888888889e-05,
+      "loss": 1.6034,
+      "step": 33360
+    },
+    {
+      "epoch": 1.06784,
+      "grad_norm": 43.8675422668457,
+      "learning_rate": 1.0357333333333335e-05,
+      "loss": 1.6393,
+      "step": 33370
+    },
+    {
+      "epoch": 1.06816,
+      "grad_norm": 43.28858947753906,
+      "learning_rate": 1.0353777777777778e-05,
+      "loss": 1.6148,
+      "step": 33380
+    },
+    {
+      "epoch": 1.06848,
+      "grad_norm": 44.096370697021484,
+      "learning_rate": 1.0350222222222224e-05,
+      "loss": 1.6145,
+      "step": 33390
+    },
+    {
+      "epoch": 1.0688,
+      "grad_norm": 42.880043029785156,
+      "learning_rate": 1.0346666666666667e-05,
+      "loss": 1.6401,
+      "step": 33400
+    },
+    {
+      "epoch": 1.06912,
+      "grad_norm": 42.06776428222656,
+      "learning_rate": 1.0343111111111113e-05,
+      "loss": 1.6416,
+      "step": 33410
+    },
+    {
+      "epoch": 1.06944,
+      "grad_norm": 44.670833587646484,
+      "learning_rate": 1.0339555555555556e-05,
+      "loss": 1.6126,
+      "step": 33420
+    },
+    {
+      "epoch": 1.06976,
+      "grad_norm": 44.110321044921875,
+      "learning_rate": 1.0336000000000002e-05,
+      "loss": 1.649,
+      "step": 33430
+    },
+    {
+      "epoch": 1.07008,
+      "grad_norm": 47.0807991027832,
+      "learning_rate": 1.0332444444444445e-05,
+      "loss": 1.637,
+      "step": 33440
+    },
+    {
+      "epoch": 1.0704,
+      "grad_norm": 45.72868728637695,
+      "learning_rate": 1.032888888888889e-05,
+      "loss": 1.6213,
+      "step": 33450
+    },
+    {
+      "epoch": 1.0707200000000001,
+      "grad_norm": 45.216796875,
+      "learning_rate": 1.0325333333333334e-05,
+      "loss": 1.6272,
+      "step": 33460
+    },
+    {
+      "epoch": 1.07104,
+      "grad_norm": 44.5551872253418,
+      "learning_rate": 1.0321777777777779e-05,
+      "loss": 1.6565,
+      "step": 33470
+    },
+    {
+      "epoch": 1.07136,
+      "grad_norm": 43.27811813354492,
+      "learning_rate": 1.0318222222222223e-05,
+      "loss": 1.6362,
+      "step": 33480
+    },
+    {
+      "epoch": 1.07168,
+      "grad_norm": 44.5125732421875,
+      "learning_rate": 1.0314666666666668e-05,
+      "loss": 1.6424,
+      "step": 33490
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 44.89219284057617,
+      "learning_rate": 1.0311111111111113e-05,
+      "loss": 1.6211,
+      "step": 33500
+    },
+    {
+      "epoch": 1.07232,
+      "grad_norm": 43.710697174072266,
+      "learning_rate": 1.0307555555555557e-05,
+      "loss": 1.6141,
+      "step": 33510
+    },
+    {
+      "epoch": 1.07264,
+      "grad_norm": 42.3712158203125,
+      "learning_rate": 1.0304e-05,
+      "loss": 1.5936,
+      "step": 33520
+    },
+    {
+      "epoch": 1.07296,
+      "grad_norm": 44.5188102722168,
+      "learning_rate": 1.0300444444444446e-05,
+      "loss": 1.6056,
+      "step": 33530
+    },
+    {
+      "epoch": 1.07328,
+      "grad_norm": 44.64246368408203,
+      "learning_rate": 1.0296888888888889e-05,
+      "loss": 1.6199,
+      "step": 33540
+    },
+    {
+      "epoch": 1.0735999999999999,
+      "grad_norm": 45.309329986572266,
+      "learning_rate": 1.0293333333333335e-05,
+      "loss": 1.6208,
+      "step": 33550
+    },
+    {
+      "epoch": 1.07392,
+      "grad_norm": 43.31216049194336,
+      "learning_rate": 1.0289777777777778e-05,
+      "loss": 1.6197,
+      "step": 33560
+    },
+    {
+      "epoch": 1.07424,
+      "grad_norm": 43.84577560424805,
+      "learning_rate": 1.0286222222222224e-05,
+      "loss": 1.6484,
+      "step": 33570
+    },
+    {
+      "epoch": 1.07456,
+      "grad_norm": 43.01450729370117,
+      "learning_rate": 1.0282666666666667e-05,
+      "loss": 1.6277,
+      "step": 33580
+    },
+    {
+      "epoch": 1.07488,
+      "grad_norm": 44.55772018432617,
+      "learning_rate": 1.0279111111111114e-05,
+      "loss": 1.6333,
+      "step": 33590
+    },
+    {
+      "epoch": 1.0752,
+      "grad_norm": 43.11758804321289,
+      "learning_rate": 1.0275555555555556e-05,
+      "loss": 1.63,
+      "step": 33600
+    },
+    {
+      "epoch": 1.07552,
+      "grad_norm": 42.40789794921875,
+      "learning_rate": 1.0272e-05,
+      "loss": 1.6005,
+      "step": 33610
+    },
+    {
+      "epoch": 1.07584,
+      "grad_norm": 43.52560043334961,
+      "learning_rate": 1.0268444444444445e-05,
+      "loss": 1.6346,
+      "step": 33620
+    },
+    {
+      "epoch": 1.07616,
+      "grad_norm": 42.63965606689453,
+      "learning_rate": 1.0264888888888888e-05,
+      "loss": 1.6006,
+      "step": 33630
+    },
+    {
+      "epoch": 1.07648,
+      "grad_norm": 44.4262809753418,
+      "learning_rate": 1.0261333333333335e-05,
+      "loss": 1.591,
+      "step": 33640
+    },
+    {
+      "epoch": 1.0768,
+      "grad_norm": 43.22575759887695,
+      "learning_rate": 1.0257777777777777e-05,
+      "loss": 1.6324,
+      "step": 33650
+    },
+    {
+      "epoch": 1.07712,
+      "grad_norm": 44.486385345458984,
+      "learning_rate": 1.0254222222222224e-05,
+      "loss": 1.5864,
+      "step": 33660
+    },
+    {
+      "epoch": 1.07744,
+      "grad_norm": 44.114166259765625,
+      "learning_rate": 1.0250666666666667e-05,
+      "loss": 1.6076,
+      "step": 33670
+    },
+    {
+      "epoch": 1.07776,
+      "grad_norm": 44.422061920166016,
+      "learning_rate": 1.0247111111111113e-05,
+      "loss": 1.6249,
+      "step": 33680
+    },
+    {
+      "epoch": 1.07808,
+      "grad_norm": 44.38449478149414,
+      "learning_rate": 1.0243555555555556e-05,
+      "loss": 1.5697,
+      "step": 33690
+    },
+    {
+      "epoch": 1.0784,
+      "grad_norm": 43.65522766113281,
+      "learning_rate": 1.024e-05,
+      "loss": 1.6384,
+      "step": 33700
+    },
+    {
+      "epoch": 1.07872,
+      "grad_norm": 45.55866241455078,
+      "learning_rate": 1.0236444444444445e-05,
+      "loss": 1.6395,
+      "step": 33710
+    },
+    {
+      "epoch": 1.07904,
+      "grad_norm": 44.74530792236328,
+      "learning_rate": 1.023288888888889e-05,
+      "loss": 1.6478,
+      "step": 33720
+    },
+    {
+      "epoch": 1.07936,
+      "grad_norm": 45.45027542114258,
+      "learning_rate": 1.0229333333333334e-05,
+      "loss": 1.6378,
+      "step": 33730
+    },
+    {
+      "epoch": 1.07968,
+      "grad_norm": 44.05388641357422,
+      "learning_rate": 1.0225777777777778e-05,
+      "loss": 1.6558,
+      "step": 33740
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 44.426246643066406,
+      "learning_rate": 1.0222222222222223e-05,
+      "loss": 1.6432,
+      "step": 33750
+    },
+    {
+      "epoch": 1.08032,
+      "grad_norm": 45.299373626708984,
+      "learning_rate": 1.0218666666666668e-05,
+      "loss": 1.6445,
+      "step": 33760
+    },
+    {
+      "epoch": 1.08064,
+      "grad_norm": 45.54659652709961,
+      "learning_rate": 1.021511111111111e-05,
+      "loss": 1.6199,
+      "step": 33770
+    },
+    {
+      "epoch": 1.08096,
+      "grad_norm": 43.4600830078125,
+      "learning_rate": 1.0211555555555557e-05,
+      "loss": 1.6035,
+      "step": 33780
+    },
+    {
+      "epoch": 1.08128,
+      "grad_norm": 45.97517395019531,
+      "learning_rate": 1.0208e-05,
+      "loss": 1.6512,
+      "step": 33790
+    },
+    {
+      "epoch": 1.0816,
+      "grad_norm": 42.87276077270508,
+      "learning_rate": 1.0204444444444446e-05,
+      "loss": 1.6324,
+      "step": 33800
+    },
+    {
+      "epoch": 1.08192,
+      "grad_norm": 45.31087112426758,
+      "learning_rate": 1.0200888888888889e-05,
+      "loss": 1.6516,
+      "step": 33810
+    },
+    {
+      "epoch": 1.08224,
+      "grad_norm": 43.140838623046875,
+      "learning_rate": 1.0197333333333335e-05,
+      "loss": 1.6259,
+      "step": 33820
+    },
+    {
+      "epoch": 1.08256,
+      "grad_norm": 44.9212646484375,
+      "learning_rate": 1.0193777777777778e-05,
+      "loss": 1.6139,
+      "step": 33830
+    },
+    {
+      "epoch": 1.08288,
+      "grad_norm": 45.999183654785156,
+      "learning_rate": 1.0190222222222224e-05,
+      "loss": 1.6218,
+      "step": 33840
+    },
+    {
+      "epoch": 1.0832,
+      "grad_norm": 42.98609161376953,
+      "learning_rate": 1.0186666666666667e-05,
+      "loss": 1.6599,
+      "step": 33850
+    },
+    {
+      "epoch": 1.08352,
+      "grad_norm": 45.387821197509766,
+      "learning_rate": 1.0183111111111113e-05,
+      "loss": 1.6251,
+      "step": 33860
+    },
+    {
+      "epoch": 1.08384,
+      "grad_norm": 44.79252624511719,
+      "learning_rate": 1.0179555555555556e-05,
+      "loss": 1.596,
+      "step": 33870
+    },
+    {
+      "epoch": 1.08416,
+      "grad_norm": 45.170902252197266,
+      "learning_rate": 1.0176000000000002e-05,
+      "loss": 1.6124,
+      "step": 33880
+    },
+    {
+      "epoch": 1.08448,
+      "grad_norm": 43.105857849121094,
+      "learning_rate": 1.0172444444444445e-05,
+      "loss": 1.6146,
+      "step": 33890
+    },
+    {
+      "epoch": 1.0848,
+      "grad_norm": 46.16950225830078,
+      "learning_rate": 1.016888888888889e-05,
+      "loss": 1.6163,
+      "step": 33900
+    },
+    {
+      "epoch": 1.08512,
+      "grad_norm": 43.8795051574707,
+      "learning_rate": 1.0165333333333334e-05,
+      "loss": 1.6246,
+      "step": 33910
+    },
+    {
+      "epoch": 1.08544,
+      "grad_norm": 44.4942626953125,
+      "learning_rate": 1.0161777777777779e-05,
+      "loss": 1.629,
+      "step": 33920
+    },
+    {
+      "epoch": 1.08576,
+      "grad_norm": 44.03514099121094,
+      "learning_rate": 1.0158222222222223e-05,
+      "loss": 1.6344,
+      "step": 33930
+    },
+    {
+      "epoch": 1.08608,
+      "grad_norm": 43.44086837768555,
+      "learning_rate": 1.0154666666666668e-05,
+      "loss": 1.6399,
+      "step": 33940
+    },
+    {
+      "epoch": 1.0864,
+      "grad_norm": 44.283836364746094,
+      "learning_rate": 1.0151111111111112e-05,
+      "loss": 1.6237,
+      "step": 33950
+    },
+    {
+      "epoch": 1.08672,
+      "grad_norm": 42.35454559326172,
+      "learning_rate": 1.0147555555555557e-05,
+      "loss": 1.6246,
+      "step": 33960
+    },
+    {
+      "epoch": 1.08704,
+      "grad_norm": 42.41131591796875,
+      "learning_rate": 1.0144e-05,
+      "loss": 1.6454,
+      "step": 33970
+    },
+    {
+      "epoch": 1.0873599999999999,
+      "grad_norm": 46.35525894165039,
+      "learning_rate": 1.0140444444444446e-05,
+      "loss": 1.6325,
+      "step": 33980
+    },
+    {
+      "epoch": 1.08768,
+      "grad_norm": 44.864994049072266,
+      "learning_rate": 1.0136888888888889e-05,
+      "loss": 1.6092,
+      "step": 33990
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 43.37247085571289,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 1.6148,
+      "step": 34000
+    },
+    {
+      "epoch": 1.08832,
+      "grad_norm": 43.918304443359375,
+      "learning_rate": 1.0129777777777778e-05,
+      "loss": 1.6151,
+      "step": 34010
+    },
+    {
+      "epoch": 1.08864,
+      "grad_norm": 46.054325103759766,
+      "learning_rate": 1.0126222222222224e-05,
+      "loss": 1.6226,
+      "step": 34020
+    },
+    {
+      "epoch": 1.08896,
+      "grad_norm": 45.775657653808594,
+      "learning_rate": 1.0122666666666667e-05,
+      "loss": 1.6358,
+      "step": 34030
+    },
+    {
+      "epoch": 1.08928,
+      "grad_norm": 43.96372985839844,
+      "learning_rate": 1.0119111111111113e-05,
+      "loss": 1.6379,
+      "step": 34040
+    },
+    {
+      "epoch": 1.0896,
+      "grad_norm": 48.16429138183594,
+      "learning_rate": 1.0115555555555556e-05,
+      "loss": 1.6075,
+      "step": 34050
+    },
+    {
+      "epoch": 1.08992,
+      "grad_norm": 47.51073455810547,
+      "learning_rate": 1.0112000000000002e-05,
+      "loss": 1.583,
+      "step": 34060
+    },
+    {
+      "epoch": 1.09024,
+      "grad_norm": 45.64347457885742,
+      "learning_rate": 1.0108444444444445e-05,
+      "loss": 1.6078,
+      "step": 34070
+    },
+    {
+      "epoch": 1.09056,
+      "grad_norm": 47.757633209228516,
+      "learning_rate": 1.0104888888888892e-05,
+      "loss": 1.6254,
+      "step": 34080
+    },
+    {
+      "epoch": 1.09088,
+      "grad_norm": 43.865596771240234,
+      "learning_rate": 1.0101333333333334e-05,
+      "loss": 1.6257,
+      "step": 34090
+    },
+    {
+      "epoch": 1.0912,
+      "grad_norm": 43.02202224731445,
+      "learning_rate": 1.0097777777777779e-05,
+      "loss": 1.6379,
+      "step": 34100
+    },
+    {
+      "epoch": 1.09152,
+      "grad_norm": 48.95008087158203,
+      "learning_rate": 1.0094222222222224e-05,
+      "loss": 1.6293,
+      "step": 34110
+    },
+    {
+      "epoch": 1.09184,
+      "grad_norm": 44.4239501953125,
+      "learning_rate": 1.0090666666666666e-05,
+      "loss": 1.6261,
+      "step": 34120
+    },
+    {
+      "epoch": 1.09216,
+      "grad_norm": 44.51701736450195,
+      "learning_rate": 1.0087111111111113e-05,
+      "loss": 1.6237,
+      "step": 34130
+    },
+    {
+      "epoch": 1.0924800000000001,
+      "grad_norm": 45.51099395751953,
+      "learning_rate": 1.0083555555555555e-05,
+      "loss": 1.598,
+      "step": 34140
+    },
+    {
+      "epoch": 1.0928,
+      "grad_norm": 44.413673400878906,
+      "learning_rate": 1.008e-05,
+      "loss": 1.6479,
+      "step": 34150
+    },
+    {
+      "epoch": 1.09312,
+      "grad_norm": 44.12715148925781,
+      "learning_rate": 1.0076444444444445e-05,
+      "loss": 1.6549,
+      "step": 34160
+    },
+    {
+      "epoch": 1.09344,
+      "grad_norm": 44.77954864501953,
+      "learning_rate": 1.007288888888889e-05,
+      "loss": 1.6036,
+      "step": 34170
+    },
+    {
+      "epoch": 1.09376,
+      "grad_norm": 43.165225982666016,
+      "learning_rate": 1.0069333333333334e-05,
+      "loss": 1.6159,
+      "step": 34180
+    },
+    {
+      "epoch": 1.09408,
+      "grad_norm": 43.7326545715332,
+      "learning_rate": 1.0065777777777778e-05,
+      "loss": 1.6138,
+      "step": 34190
+    },
+    {
+      "epoch": 1.0944,
+      "grad_norm": 45.99542999267578,
+      "learning_rate": 1.0062222222222223e-05,
+      "loss": 1.6264,
+      "step": 34200
+    },
+    {
+      "epoch": 1.09472,
+      "grad_norm": 45.08587646484375,
+      "learning_rate": 1.0058666666666667e-05,
+      "loss": 1.6064,
+      "step": 34210
+    },
+    {
+      "epoch": 1.09504,
+      "grad_norm": 44.10668182373047,
+      "learning_rate": 1.005511111111111e-05,
+      "loss": 1.6188,
+      "step": 34220
+    },
+    {
+      "epoch": 1.09536,
+      "grad_norm": 45.843345642089844,
+      "learning_rate": 1.0051555555555556e-05,
+      "loss": 1.6259,
+      "step": 34230
+    },
+    {
+      "epoch": 1.09568,
+      "grad_norm": 45.700645446777344,
+      "learning_rate": 1.0048e-05,
+      "loss": 1.628,
+      "step": 34240
+    },
+    {
+      "epoch": 1.096,
+      "grad_norm": 44.80293655395508,
+      "learning_rate": 1.0044444444444446e-05,
+      "loss": 1.5932,
+      "step": 34250
+    },
+    {
+      "epoch": 1.09632,
+      "grad_norm": 43.28338623046875,
+      "learning_rate": 1.0040888888888888e-05,
+      "loss": 1.6261,
+      "step": 34260
+    },
+    {
+      "epoch": 1.09664,
+      "grad_norm": 46.780269622802734,
+      "learning_rate": 1.0037333333333335e-05,
+      "loss": 1.6001,
+      "step": 34270
+    },
+    {
+      "epoch": 1.09696,
+      "grad_norm": 42.56535339355469,
+      "learning_rate": 1.0033777777777778e-05,
+      "loss": 1.6386,
+      "step": 34280
+    },
+    {
+      "epoch": 1.09728,
+      "grad_norm": 45.69676971435547,
+      "learning_rate": 1.0030222222222224e-05,
+      "loss": 1.6089,
+      "step": 34290
+    },
+    {
+      "epoch": 1.0976,
+      "grad_norm": 42.413429260253906,
+      "learning_rate": 1.0026666666666667e-05,
+      "loss": 1.6273,
+      "step": 34300
+    },
+    {
+      "epoch": 1.09792,
+      "grad_norm": 44.377784729003906,
+      "learning_rate": 1.0023111111111113e-05,
+      "loss": 1.6301,
+      "step": 34310
+    },
+    {
+      "epoch": 1.09824,
+      "grad_norm": 44.19623565673828,
+      "learning_rate": 1.0019555555555556e-05,
+      "loss": 1.6346,
+      "step": 34320
+    },
+    {
+      "epoch": 1.09856,
+      "grad_norm": 44.605125427246094,
+      "learning_rate": 1.0016000000000002e-05,
+      "loss": 1.6098,
+      "step": 34330
+    },
+    {
+      "epoch": 1.09888,
+      "grad_norm": 45.71287536621094,
+      "learning_rate": 1.0012444444444445e-05,
+      "loss": 1.6236,
+      "step": 34340
+    },
+    {
+      "epoch": 1.0992,
+      "grad_norm": 44.4366455078125,
+      "learning_rate": 1.000888888888889e-05,
+      "loss": 1.6575,
+      "step": 34350
+    },
+    {
+      "epoch": 1.09952,
+      "grad_norm": 46.797176361083984,
+      "learning_rate": 1.0005333333333334e-05,
+      "loss": 1.6285,
+      "step": 34360
+    },
+    {
+      "epoch": 1.09984,
+      "grad_norm": 42.749549865722656,
+      "learning_rate": 1.0001777777777779e-05,
+      "loss": 1.5832,
+      "step": 34370
+    },
+    {
+      "epoch": 1.10016,
+      "grad_norm": 44.592552185058594,
+      "learning_rate": 9.998222222222223e-06,
+      "loss": 1.6181,
+      "step": 34380
+    },
+    {
+      "epoch": 1.10048,
+      "grad_norm": 44.82821273803711,
+      "learning_rate": 9.994666666666668e-06,
+      "loss": 1.6198,
+      "step": 34390
+    },
+    {
+      "epoch": 1.1008,
+      "grad_norm": 45.68002700805664,
+      "learning_rate": 9.991111111111112e-06,
+      "loss": 1.6202,
+      "step": 34400
+    },
+    {
+      "epoch": 1.10112,
+      "grad_norm": 45.76319122314453,
+      "learning_rate": 9.987555555555555e-06,
+      "loss": 1.6217,
+      "step": 34410
+    },
+    {
+      "epoch": 1.10144,
+      "grad_norm": 45.374473571777344,
+      "learning_rate": 9.984e-06,
+      "loss": 1.6254,
+      "step": 34420
+    },
+    {
+      "epoch": 1.10176,
+      "grad_norm": 46.05216979980469,
+      "learning_rate": 9.980444444444444e-06,
+      "loss": 1.6517,
+      "step": 34430
+    },
+    {
+      "epoch": 1.10208,
+      "grad_norm": 45.405696868896484,
+      "learning_rate": 9.976888888888889e-06,
+      "loss": 1.5984,
+      "step": 34440
+    },
+    {
+      "epoch": 1.1024,
+      "grad_norm": 44.274539947509766,
+      "learning_rate": 9.973333333333333e-06,
+      "loss": 1.6108,
+      "step": 34450
+    },
+    {
+      "epoch": 1.10272,
+      "grad_norm": 43.765960693359375,
+      "learning_rate": 9.969777777777778e-06,
+      "loss": 1.6175,
+      "step": 34460
+    },
+    {
+      "epoch": 1.10304,
+      "grad_norm": 44.908416748046875,
+      "learning_rate": 9.966222222222222e-06,
+      "loss": 1.6131,
+      "step": 34470
+    },
+    {
+      "epoch": 1.10336,
+      "grad_norm": 45.51877975463867,
+      "learning_rate": 9.962666666666667e-06,
+      "loss": 1.6011,
+      "step": 34480
+    },
+    {
+      "epoch": 1.10368,
+      "grad_norm": 43.720340728759766,
+      "learning_rate": 9.959111111111111e-06,
+      "loss": 1.6129,
+      "step": 34490
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 44.68758773803711,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 1.6381,
+      "step": 34500
+    },
+    {
+      "epoch": 1.10432,
+      "grad_norm": 43.98762512207031,
+      "learning_rate": 9.952e-06,
+      "loss": 1.6444,
+      "step": 34510
+    },
+    {
+      "epoch": 1.10464,
+      "grad_norm": 44.053226470947266,
+      "learning_rate": 9.948444444444445e-06,
+      "loss": 1.6195,
+      "step": 34520
+    },
+    {
+      "epoch": 1.10496,
+      "grad_norm": 44.89874267578125,
+      "learning_rate": 9.94488888888889e-06,
+      "loss": 1.6158,
+      "step": 34530
+    },
+    {
+      "epoch": 1.10528,
+      "grad_norm": 45.770843505859375,
+      "learning_rate": 9.941333333333334e-06,
+      "loss": 1.6134,
+      "step": 34540
+    },
+    {
+      "epoch": 1.1056,
+      "grad_norm": 43.45571517944336,
+      "learning_rate": 9.937777777777779e-06,
+      "loss": 1.6221,
+      "step": 34550
+    },
+    {
+      "epoch": 1.10592,
+      "grad_norm": 44.796852111816406,
+      "learning_rate": 9.934222222222223e-06,
+      "loss": 1.6123,
+      "step": 34560
+    },
+    {
+      "epoch": 1.1062400000000001,
+      "grad_norm": 44.354251861572266,
+      "learning_rate": 9.930666666666668e-06,
+      "loss": 1.5849,
+      "step": 34570
+    },
+    {
+      "epoch": 1.10656,
+      "grad_norm": 44.47141647338867,
+      "learning_rate": 9.927111111111112e-06,
+      "loss": 1.6325,
+      "step": 34580
+    },
+    {
+      "epoch": 1.10688,
+      "grad_norm": 44.321067810058594,
+      "learning_rate": 9.923555555555557e-06,
+      "loss": 1.6389,
+      "step": 34590
+    },
+    {
+      "epoch": 1.1072,
+      "grad_norm": 44.76230239868164,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 1.6257,
+      "step": 34600
+    },
+    {
+      "epoch": 1.10752,
+      "grad_norm": 44.10141372680664,
+      "learning_rate": 9.916444444444444e-06,
+      "loss": 1.6173,
+      "step": 34610
+    },
+    {
+      "epoch": 1.10784,
+      "grad_norm": 45.80775833129883,
+      "learning_rate": 9.912888888888889e-06,
+      "loss": 1.6211,
+      "step": 34620
+    },
+    {
+      "epoch": 1.10816,
+      "grad_norm": 46.45954895019531,
+      "learning_rate": 9.909333333333334e-06,
+      "loss": 1.6053,
+      "step": 34630
+    },
+    {
+      "epoch": 1.10848,
+      "grad_norm": 42.96110916137695,
+      "learning_rate": 9.905777777777778e-06,
+      "loss": 1.6146,
+      "step": 34640
+    },
+    {
+      "epoch": 1.1088,
+      "grad_norm": 45.49494934082031,
+      "learning_rate": 9.902222222222223e-06,
+      "loss": 1.5967,
+      "step": 34650
+    },
+    {
+      "epoch": 1.1091199999999999,
+      "grad_norm": 43.099300384521484,
+      "learning_rate": 9.898666666666667e-06,
+      "loss": 1.6203,
+      "step": 34660
+    },
+    {
+      "epoch": 1.10944,
+      "grad_norm": 42.93831253051758,
+      "learning_rate": 9.895111111111112e-06,
+      "loss": 1.6078,
+      "step": 34670
+    },
+    {
+      "epoch": 1.10976,
+      "grad_norm": 43.761878967285156,
+      "learning_rate": 9.891555555555556e-06,
+      "loss": 1.6117,
+      "step": 34680
+    },
+    {
+      "epoch": 1.11008,
+      "grad_norm": 44.20676040649414,
+      "learning_rate": 9.888000000000001e-06,
+      "loss": 1.6246,
+      "step": 34690
+    },
+    {
+      "epoch": 1.1104,
+      "grad_norm": 43.08485412597656,
+      "learning_rate": 9.884444444444445e-06,
+      "loss": 1.6179,
+      "step": 34700
+    },
+    {
+      "epoch": 1.11072,
+      "grad_norm": 45.42795181274414,
+      "learning_rate": 9.88088888888889e-06,
+      "loss": 1.6055,
+      "step": 34710
+    },
+    {
+      "epoch": 1.11104,
+      "grad_norm": 46.919559478759766,
+      "learning_rate": 9.877333333333335e-06,
+      "loss": 1.6061,
+      "step": 34720
+    },
+    {
+      "epoch": 1.11136,
+      "grad_norm": 46.053768157958984,
+      "learning_rate": 9.873777777777779e-06,
+      "loss": 1.5963,
+      "step": 34730
+    },
+    {
+      "epoch": 1.11168,
+      "grad_norm": 44.76418685913086,
+      "learning_rate": 9.870222222222224e-06,
+      "loss": 1.6167,
+      "step": 34740
+    },
+    {
+      "epoch": 1.112,
+      "grad_norm": 44.72624969482422,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.6254,
+      "step": 34750
+    },
+    {
+      "epoch": 1.11232,
+      "grad_norm": 43.534324645996094,
+      "learning_rate": 9.863111111111113e-06,
+      "loss": 1.6042,
+      "step": 34760
+    },
+    {
+      "epoch": 1.11264,
+      "grad_norm": 45.52174377441406,
+      "learning_rate": 9.859555555555557e-06,
+      "loss": 1.6191,
+      "step": 34770
+    },
+    {
+      "epoch": 1.11296,
+      "grad_norm": 44.6978645324707,
+      "learning_rate": 9.856000000000002e-06,
+      "loss": 1.6425,
+      "step": 34780
+    },
+    {
+      "epoch": 1.11328,
+      "grad_norm": 44.247283935546875,
+      "learning_rate": 9.852444444444446e-06,
+      "loss": 1.6097,
+      "step": 34790
+    },
+    {
+      "epoch": 1.1136,
+      "grad_norm": 44.1110725402832,
+      "learning_rate": 9.84888888888889e-06,
+      "loss": 1.6406,
+      "step": 34800
+    },
+    {
+      "epoch": 1.11392,
+      "grad_norm": 44.1603889465332,
+      "learning_rate": 9.845333333333334e-06,
+      "loss": 1.6166,
+      "step": 34810
+    },
+    {
+      "epoch": 1.11424,
+      "grad_norm": 46.36598205566406,
+      "learning_rate": 9.841777777777778e-06,
+      "loss": 1.6222,
+      "step": 34820
+    },
+    {
+      "epoch": 1.11456,
+      "grad_norm": 42.831356048583984,
+      "learning_rate": 9.838222222222223e-06,
+      "loss": 1.6166,
+      "step": 34830
+    },
+    {
+      "epoch": 1.11488,
+      "grad_norm": 44.336055755615234,
+      "learning_rate": 9.834666666666667e-06,
+      "loss": 1.6025,
+      "step": 34840
+    },
+    {
+      "epoch": 1.1152,
+      "grad_norm": 45.25812911987305,
+      "learning_rate": 9.831111111111112e-06,
+      "loss": 1.6351,
+      "step": 34850
+    },
+    {
+      "epoch": 1.11552,
+      "grad_norm": 44.683589935302734,
+      "learning_rate": 9.827555555555557e-06,
+      "loss": 1.6015,
+      "step": 34860
+    },
+    {
+      "epoch": 1.11584,
+      "grad_norm": 47.49604797363281,
+      "learning_rate": 9.824000000000001e-06,
+      "loss": 1.6258,
+      "step": 34870
+    },
+    {
+      "epoch": 1.11616,
+      "grad_norm": 43.392303466796875,
+      "learning_rate": 9.820444444444446e-06,
+      "loss": 1.6146,
+      "step": 34880
+    },
+    {
+      "epoch": 1.11648,
+      "grad_norm": 44.247188568115234,
+      "learning_rate": 9.81688888888889e-06,
+      "loss": 1.6411,
+      "step": 34890
+    },
+    {
+      "epoch": 1.1168,
+      "grad_norm": 46.70269012451172,
+      "learning_rate": 9.813333333333333e-06,
+      "loss": 1.6291,
+      "step": 34900
+    },
+    {
+      "epoch": 1.11712,
+      "grad_norm": 44.9300537109375,
+      "learning_rate": 9.809777777777778e-06,
+      "loss": 1.6169,
+      "step": 34910
+    },
+    {
+      "epoch": 1.11744,
+      "grad_norm": 44.363807678222656,
+      "learning_rate": 9.806222222222222e-06,
+      "loss": 1.6618,
+      "step": 34920
+    },
+    {
+      "epoch": 1.11776,
+      "grad_norm": 46.45115661621094,
+      "learning_rate": 9.802666666666667e-06,
+      "loss": 1.6276,
+      "step": 34930
+    },
+    {
+      "epoch": 1.11808,
+      "grad_norm": 44.15941619873047,
+      "learning_rate": 9.799111111111111e-06,
+      "loss": 1.5878,
+      "step": 34940
+    },
+    {
+      "epoch": 1.1184,
+      "grad_norm": 46.215606689453125,
+      "learning_rate": 9.795555555555556e-06,
+      "loss": 1.642,
+      "step": 34950
+    },
+    {
+      "epoch": 1.11872,
+      "grad_norm": 42.38996887207031,
+      "learning_rate": 9.792e-06,
+      "loss": 1.5918,
+      "step": 34960
+    },
+    {
+      "epoch": 1.11904,
+      "grad_norm": 45.36106491088867,
+      "learning_rate": 9.788444444444445e-06,
+      "loss": 1.5951,
+      "step": 34970
+    },
+    {
+      "epoch": 1.11936,
+      "grad_norm": 42.64955520629883,
+      "learning_rate": 9.78488888888889e-06,
+      "loss": 1.5968,
+      "step": 34980
+    },
+    {
+      "epoch": 1.11968,
+      "grad_norm": 43.310054779052734,
+      "learning_rate": 9.781333333333334e-06,
+      "loss": 1.6062,
+      "step": 34990
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 45.69913101196289,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 1.6475,
+      "step": 35000
+    },
+    {
+      "epoch": 1.12032,
+      "grad_norm": 45.340972900390625,
+      "learning_rate": 9.774222222222223e-06,
+      "loss": 1.6389,
+      "step": 35010
+    },
+    {
+      "epoch": 1.12064,
+      "grad_norm": 43.69623947143555,
+      "learning_rate": 9.770666666666668e-06,
+      "loss": 1.6071,
+      "step": 35020
+    },
+    {
+      "epoch": 1.12096,
+      "grad_norm": 43.38682174682617,
+      "learning_rate": 9.767111111111112e-06,
+      "loss": 1.6332,
+      "step": 35030
+    },
+    {
+      "epoch": 1.12128,
+      "grad_norm": 44.497440338134766,
+      "learning_rate": 9.763555555555557e-06,
+      "loss": 1.6454,
+      "step": 35040
+    },
+    {
+      "epoch": 1.1216,
+      "grad_norm": 43.34943389892578,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 1.6254,
+      "step": 35050
+    },
+    {
+      "epoch": 1.12192,
+      "grad_norm": 45.34946060180664,
+      "learning_rate": 9.756444444444444e-06,
+      "loss": 1.638,
+      "step": 35060
+    },
+    {
+      "epoch": 1.12224,
+      "grad_norm": 46.76422882080078,
+      "learning_rate": 9.752888888888889e-06,
+      "loss": 1.6248,
+      "step": 35070
+    },
+    {
+      "epoch": 1.12256,
+      "grad_norm": 43.90962219238281,
+      "learning_rate": 9.749333333333333e-06,
+      "loss": 1.6249,
+      "step": 35080
+    },
+    {
+      "epoch": 1.12288,
+      "grad_norm": 45.085994720458984,
+      "learning_rate": 9.745777777777778e-06,
+      "loss": 1.6255,
+      "step": 35090
+    },
+    {
+      "epoch": 1.1232,
+      "grad_norm": 46.66875457763672,
+      "learning_rate": 9.742222222222222e-06,
+      "loss": 1.6219,
+      "step": 35100
+    },
+    {
+      "epoch": 1.12352,
+      "grad_norm": 46.58740234375,
+      "learning_rate": 9.738666666666667e-06,
+      "loss": 1.6079,
+      "step": 35110
+    },
+    {
+      "epoch": 1.12384,
+      "grad_norm": 45.570945739746094,
+      "learning_rate": 9.735111111111112e-06,
+      "loss": 1.629,
+      "step": 35120
+    },
+    {
+      "epoch": 1.12416,
+      "grad_norm": 43.049095153808594,
+      "learning_rate": 9.731555555555556e-06,
+      "loss": 1.608,
+      "step": 35130
+    },
+    {
+      "epoch": 1.12448,
+      "grad_norm": 45.4088020324707,
+      "learning_rate": 9.728e-06,
+      "loss": 1.5962,
+      "step": 35140
+    },
+    {
+      "epoch": 1.1248,
+      "grad_norm": 45.11723709106445,
+      "learning_rate": 9.724444444444445e-06,
+      "loss": 1.6141,
+      "step": 35150
+    },
+    {
+      "epoch": 1.12512,
+      "grad_norm": 44.503482818603516,
+      "learning_rate": 9.72088888888889e-06,
+      "loss": 1.643,
+      "step": 35160
+    },
+    {
+      "epoch": 1.12544,
+      "grad_norm": 42.30671310424805,
+      "learning_rate": 9.717333333333334e-06,
+      "loss": 1.6232,
+      "step": 35170
+    },
+    {
+      "epoch": 1.12576,
+      "grad_norm": 44.18793869018555,
+      "learning_rate": 9.713777777777779e-06,
+      "loss": 1.6206,
+      "step": 35180
+    },
+    {
+      "epoch": 1.12608,
+      "grad_norm": 44.02192687988281,
+      "learning_rate": 9.710222222222223e-06,
+      "loss": 1.6226,
+      "step": 35190
+    },
+    {
+      "epoch": 1.1264,
+      "grad_norm": 42.9251823425293,
+      "learning_rate": 9.706666666666668e-06,
+      "loss": 1.6467,
+      "step": 35200
+    },
+    {
+      "epoch": 1.12672,
+      "grad_norm": 43.18367385864258,
+      "learning_rate": 9.703111111111113e-06,
+      "loss": 1.6253,
+      "step": 35210
+    },
+    {
+      "epoch": 1.12704,
+      "grad_norm": 44.549320220947266,
+      "learning_rate": 9.699555555555557e-06,
+      "loss": 1.6039,
+      "step": 35220
+    },
+    {
+      "epoch": 1.12736,
+      "grad_norm": 45.207427978515625,
+      "learning_rate": 9.696000000000002e-06,
+      "loss": 1.6157,
+      "step": 35230
+    },
+    {
+      "epoch": 1.12768,
+      "grad_norm": 44.419551849365234,
+      "learning_rate": 9.692444444444446e-06,
+      "loss": 1.6288,
+      "step": 35240
+    },
+    {
+      "epoch": 1.1280000000000001,
+      "grad_norm": 47.223838806152344,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 1.6373,
+      "step": 35250
+    },
+    {
+      "epoch": 1.12832,
+      "grad_norm": 44.445526123046875,
+      "learning_rate": 9.685333333333334e-06,
+      "loss": 1.6128,
+      "step": 35260
+    },
+    {
+      "epoch": 1.12864,
+      "grad_norm": 42.873783111572266,
+      "learning_rate": 9.681777777777778e-06,
+      "loss": 1.624,
+      "step": 35270
+    },
+    {
+      "epoch": 1.12896,
+      "grad_norm": 43.351097106933594,
+      "learning_rate": 9.678222222222223e-06,
+      "loss": 1.606,
+      "step": 35280
+    },
+    {
+      "epoch": 1.12928,
+      "grad_norm": 42.88664245605469,
+      "learning_rate": 9.674666666666667e-06,
+      "loss": 1.5858,
+      "step": 35290
+    },
+    {
+      "epoch": 1.1296,
+      "grad_norm": 46.03646469116211,
+      "learning_rate": 9.671111111111112e-06,
+      "loss": 1.6352,
+      "step": 35300
+    },
+    {
+      "epoch": 1.12992,
+      "grad_norm": 43.69074630737305,
+      "learning_rate": 9.667555555555556e-06,
+      "loss": 1.6128,
+      "step": 35310
+    },
+    {
+      "epoch": 1.13024,
+      "grad_norm": 44.54518127441406,
+      "learning_rate": 9.664000000000001e-06,
+      "loss": 1.5902,
+      "step": 35320
+    },
+    {
+      "epoch": 1.13056,
+      "grad_norm": 43.998966217041016,
+      "learning_rate": 9.660444444444446e-06,
+      "loss": 1.5957,
+      "step": 35330
+    },
+    {
+      "epoch": 1.1308799999999999,
+      "grad_norm": 44.25476837158203,
+      "learning_rate": 9.65688888888889e-06,
+      "loss": 1.6201,
+      "step": 35340
+    },
+    {
+      "epoch": 1.1312,
+      "grad_norm": 43.57674026489258,
+      "learning_rate": 9.653333333333335e-06,
+      "loss": 1.6227,
+      "step": 35350
+    },
+    {
+      "epoch": 1.13152,
+      "grad_norm": 42.928321838378906,
+      "learning_rate": 9.64977777777778e-06,
+      "loss": 1.5962,
+      "step": 35360
+    },
+    {
+      "epoch": 1.13184,
+      "grad_norm": 46.871498107910156,
+      "learning_rate": 9.646222222222224e-06,
+      "loss": 1.5984,
+      "step": 35370
+    },
+    {
+      "epoch": 1.13216,
+      "grad_norm": 44.98406219482422,
+      "learning_rate": 9.642666666666668e-06,
+      "loss": 1.561,
+      "step": 35380
+    },
+    {
+      "epoch": 1.13248,
+      "grad_norm": 43.02797317504883,
+      "learning_rate": 9.639111111111113e-06,
+      "loss": 1.6324,
+      "step": 35390
+    },
+    {
+      "epoch": 1.1328,
+      "grad_norm": 47.9857292175293,
+      "learning_rate": 9.635555555555557e-06,
+      "loss": 1.6074,
+      "step": 35400
+    },
+    {
+      "epoch": 1.13312,
+      "grad_norm": 45.18562316894531,
+      "learning_rate": 9.632e-06,
+      "loss": 1.6103,
+      "step": 35410
+    },
+    {
+      "epoch": 1.13344,
+      "grad_norm": 45.36076736450195,
+      "learning_rate": 9.628444444444445e-06,
+      "loss": 1.6141,
+      "step": 35420
+    },
+    {
+      "epoch": 1.13376,
+      "grad_norm": 44.99217987060547,
+      "learning_rate": 9.62488888888889e-06,
+      "loss": 1.65,
+      "step": 35430
+    },
+    {
+      "epoch": 1.13408,
+      "grad_norm": 43.21048355102539,
+      "learning_rate": 9.621333333333334e-06,
+      "loss": 1.6333,
+      "step": 35440
+    },
+    {
+      "epoch": 1.1344,
+      "grad_norm": 46.94039535522461,
+      "learning_rate": 9.617777777777778e-06,
+      "loss": 1.6215,
+      "step": 35450
+    },
+    {
+      "epoch": 1.13472,
+      "grad_norm": 45.81171798706055,
+      "learning_rate": 9.614222222222223e-06,
+      "loss": 1.6278,
+      "step": 35460
+    },
+    {
+      "epoch": 1.13504,
+      "grad_norm": 42.52925109863281,
+      "learning_rate": 9.610666666666668e-06,
+      "loss": 1.6303,
+      "step": 35470
+    },
+    {
+      "epoch": 1.13536,
+      "grad_norm": 46.01049041748047,
+      "learning_rate": 9.607111111111112e-06,
+      "loss": 1.6218,
+      "step": 35480
+    },
+    {
+      "epoch": 1.13568,
+      "grad_norm": 43.44357681274414,
+      "learning_rate": 9.603555555555557e-06,
+      "loss": 1.5914,
+      "step": 35490
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 44.107540130615234,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.6313,
+      "step": 35500
+    },
+    {
+      "epoch": 1.13632,
+      "grad_norm": 47.5319709777832,
+      "learning_rate": 9.596444444444444e-06,
+      "loss": 1.6151,
+      "step": 35510
+    },
+    {
+      "epoch": 1.13664,
+      "grad_norm": 46.17417526245117,
+      "learning_rate": 9.592888888888889e-06,
+      "loss": 1.6174,
+      "step": 35520
+    },
+    {
+      "epoch": 1.13696,
+      "grad_norm": 47.07992172241211,
+      "learning_rate": 9.589333333333333e-06,
+      "loss": 1.6122,
+      "step": 35530
+    },
+    {
+      "epoch": 1.13728,
+      "grad_norm": 44.27360534667969,
+      "learning_rate": 9.585777777777778e-06,
+      "loss": 1.6238,
+      "step": 35540
+    },
+    {
+      "epoch": 1.1376,
+      "grad_norm": 45.99604797363281,
+      "learning_rate": 9.582222222222222e-06,
+      "loss": 1.619,
+      "step": 35550
+    },
+    {
+      "epoch": 1.13792,
+      "grad_norm": 47.591575622558594,
+      "learning_rate": 9.578666666666667e-06,
+      "loss": 1.6273,
+      "step": 35560
+    },
+    {
+      "epoch": 1.13824,
+      "grad_norm": 44.06424331665039,
+      "learning_rate": 9.575111111111111e-06,
+      "loss": 1.6236,
+      "step": 35570
+    },
+    {
+      "epoch": 1.13856,
+      "grad_norm": 43.544307708740234,
+      "learning_rate": 9.571555555555556e-06,
+      "loss": 1.6153,
+      "step": 35580
+    },
+    {
+      "epoch": 1.13888,
+      "grad_norm": 43.172245025634766,
+      "learning_rate": 9.568e-06,
+      "loss": 1.6304,
+      "step": 35590
+    },
+    {
+      "epoch": 1.1392,
+      "grad_norm": 43.66238021850586,
+      "learning_rate": 9.564444444444445e-06,
+      "loss": 1.5803,
+      "step": 35600
+    },
+    {
+      "epoch": 1.13952,
+      "grad_norm": 44.61842346191406,
+      "learning_rate": 9.56088888888889e-06,
+      "loss": 1.6163,
+      "step": 35610
+    },
+    {
+      "epoch": 1.13984,
+      "grad_norm": 43.94649887084961,
+      "learning_rate": 9.557333333333334e-06,
+      "loss": 1.6256,
+      "step": 35620
+    },
+    {
+      "epoch": 1.14016,
+      "grad_norm": 44.97169494628906,
+      "learning_rate": 9.553777777777779e-06,
+      "loss": 1.6265,
+      "step": 35630
+    },
+    {
+      "epoch": 1.14048,
+      "grad_norm": 42.63601303100586,
+      "learning_rate": 9.550222222222223e-06,
+      "loss": 1.6296,
+      "step": 35640
+    },
+    {
+      "epoch": 1.1408,
+      "grad_norm": 43.537925720214844,
+      "learning_rate": 9.546666666666668e-06,
+      "loss": 1.6142,
+      "step": 35650
+    },
+    {
+      "epoch": 1.14112,
+      "grad_norm": 43.652381896972656,
+      "learning_rate": 9.543111111111112e-06,
+      "loss": 1.6481,
+      "step": 35660
+    },
+    {
+      "epoch": 1.14144,
+      "grad_norm": 43.40591812133789,
+      "learning_rate": 9.539555555555557e-06,
+      "loss": 1.6477,
+      "step": 35670
+    },
+    {
+      "epoch": 1.14176,
+      "grad_norm": 44.605796813964844,
+      "learning_rate": 9.536000000000002e-06,
+      "loss": 1.6217,
+      "step": 35680
+    },
+    {
+      "epoch": 1.14208,
+      "grad_norm": 44.9923095703125,
+      "learning_rate": 9.532444444444446e-06,
+      "loss": 1.6236,
+      "step": 35690
+    },
+    {
+      "epoch": 1.1424,
+      "grad_norm": 45.73491668701172,
+      "learning_rate": 9.528888888888889e-06,
+      "loss": 1.6207,
+      "step": 35700
+    },
+    {
+      "epoch": 1.14272,
+      "grad_norm": 43.9902458190918,
+      "learning_rate": 9.525333333333334e-06,
+      "loss": 1.6332,
+      "step": 35710
+    },
+    {
+      "epoch": 1.14304,
+      "grad_norm": 44.50312805175781,
+      "learning_rate": 9.521777777777778e-06,
+      "loss": 1.6095,
+      "step": 35720
+    },
+    {
+      "epoch": 1.14336,
+      "grad_norm": 42.85441970825195,
+      "learning_rate": 9.518222222222223e-06,
+      "loss": 1.6086,
+      "step": 35730
+    },
+    {
+      "epoch": 1.14368,
+      "grad_norm": 43.633811950683594,
+      "learning_rate": 9.514666666666667e-06,
+      "loss": 1.6195,
+      "step": 35740
+    },
+    {
+      "epoch": 1.144,
+      "grad_norm": 44.819847106933594,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 1.6025,
+      "step": 35750
+    },
+    {
+      "epoch": 1.14432,
+      "grad_norm": 42.97698211669922,
+      "learning_rate": 9.507555555555556e-06,
+      "loss": 1.6075,
+      "step": 35760
+    },
+    {
+      "epoch": 1.1446399999999999,
+      "grad_norm": 42.89186477661133,
+      "learning_rate": 9.504e-06,
+      "loss": 1.604,
+      "step": 35770
+    },
+    {
+      "epoch": 1.14496,
+      "grad_norm": 44.783382415771484,
+      "learning_rate": 9.500444444444445e-06,
+      "loss": 1.642,
+      "step": 35780
+    },
+    {
+      "epoch": 1.14528,
+      "grad_norm": 46.417877197265625,
+      "learning_rate": 9.49688888888889e-06,
+      "loss": 1.6321,
+      "step": 35790
+    },
+    {
+      "epoch": 1.1456,
+      "grad_norm": 45.94801712036133,
+      "learning_rate": 9.493333333333334e-06,
+      "loss": 1.6116,
+      "step": 35800
+    },
+    {
+      "epoch": 1.14592,
+      "grad_norm": 45.10016632080078,
+      "learning_rate": 9.489777777777779e-06,
+      "loss": 1.6527,
+      "step": 35810
+    },
+    {
+      "epoch": 1.14624,
+      "grad_norm": 43.96493911743164,
+      "learning_rate": 9.486222222222224e-06,
+      "loss": 1.6229,
+      "step": 35820
+    },
+    {
+      "epoch": 1.14656,
+      "grad_norm": 46.303653717041016,
+      "learning_rate": 9.482666666666668e-06,
+      "loss": 1.6128,
+      "step": 35830
+    },
+    {
+      "epoch": 1.14688,
+      "grad_norm": 43.05647277832031,
+      "learning_rate": 9.479111111111113e-06,
+      "loss": 1.6246,
+      "step": 35840
+    },
+    {
+      "epoch": 1.1472,
+      "grad_norm": 43.26482391357422,
+      "learning_rate": 9.475555555555557e-06,
+      "loss": 1.617,
+      "step": 35850
+    },
+    {
+      "epoch": 1.14752,
+      "grad_norm": 46.7188720703125,
+      "learning_rate": 9.472000000000002e-06,
+      "loss": 1.6311,
+      "step": 35860
+    },
+    {
+      "epoch": 1.14784,
+      "grad_norm": 44.45231246948242,
+      "learning_rate": 9.468444444444446e-06,
+      "loss": 1.6216,
+      "step": 35870
+    },
+    {
+      "epoch": 1.14816,
+      "grad_norm": 44.6671028137207,
+      "learning_rate": 9.464888888888891e-06,
+      "loss": 1.6041,
+      "step": 35880
+    },
+    {
+      "epoch": 1.14848,
+      "grad_norm": 44.429935455322266,
+      "learning_rate": 9.461333333333334e-06,
+      "loss": 1.6353,
+      "step": 35890
+    },
+    {
+      "epoch": 1.1488,
+      "grad_norm": 45.848201751708984,
+      "learning_rate": 9.457777777777778e-06,
+      "loss": 1.6308,
+      "step": 35900
+    },
+    {
+      "epoch": 1.14912,
+      "grad_norm": 48.60066604614258,
+      "learning_rate": 9.454222222222223e-06,
+      "loss": 1.6377,
+      "step": 35910
+    },
+    {
+      "epoch": 1.14944,
+      "grad_norm": 43.88365173339844,
+      "learning_rate": 9.450666666666667e-06,
+      "loss": 1.6467,
+      "step": 35920
+    },
+    {
+      "epoch": 1.1497600000000001,
+      "grad_norm": 45.288902282714844,
+      "learning_rate": 9.447111111111112e-06,
+      "loss": 1.6111,
+      "step": 35930
+    },
+    {
+      "epoch": 1.15008,
+      "grad_norm": 44.7678337097168,
+      "learning_rate": 9.443555555555557e-06,
+      "loss": 1.624,
+      "step": 35940
+    },
+    {
+      "epoch": 1.1504,
+      "grad_norm": 42.77717971801758,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 1.6127,
+      "step": 35950
+    },
+    {
+      "epoch": 1.15072,
+      "grad_norm": 44.70463943481445,
+      "learning_rate": 9.436444444444444e-06,
+      "loss": 1.6006,
+      "step": 35960
+    },
+    {
+      "epoch": 1.15104,
+      "grad_norm": 45.02092742919922,
+      "learning_rate": 9.432888888888889e-06,
+      "loss": 1.6308,
+      "step": 35970
+    },
+    {
+      "epoch": 1.15136,
+      "grad_norm": 43.64936065673828,
+      "learning_rate": 9.429333333333333e-06,
+      "loss": 1.6191,
+      "step": 35980
+    },
+    {
+      "epoch": 1.15168,
+      "grad_norm": 46.581573486328125,
+      "learning_rate": 9.425777777777778e-06,
+      "loss": 1.6276,
+      "step": 35990
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 43.60871887207031,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 1.6308,
+      "step": 36000
+    },
+    {
+      "epoch": 1.15232,
+      "grad_norm": 46.670352935791016,
+      "learning_rate": 9.418666666666667e-06,
+      "loss": 1.6315,
+      "step": 36010
+    },
+    {
+      "epoch": 1.1526399999999999,
+      "grad_norm": 46.51318359375,
+      "learning_rate": 9.415111111111111e-06,
+      "loss": 1.6134,
+      "step": 36020
+    },
+    {
+      "epoch": 1.15296,
+      "grad_norm": 44.41347122192383,
+      "learning_rate": 9.411555555555556e-06,
+      "loss": 1.6124,
+      "step": 36030
+    },
+    {
+      "epoch": 1.15328,
+      "grad_norm": 45.49982833862305,
+      "learning_rate": 9.408e-06,
+      "loss": 1.6189,
+      "step": 36040
+    },
+    {
+      "epoch": 1.1536,
+      "grad_norm": 43.75909423828125,
+      "learning_rate": 9.404444444444445e-06,
+      "loss": 1.593,
+      "step": 36050
+    },
+    {
+      "epoch": 1.15392,
+      "grad_norm": 45.35482406616211,
+      "learning_rate": 9.40088888888889e-06,
+      "loss": 1.6289,
+      "step": 36060
+    },
+    {
+      "epoch": 1.15424,
+      "grad_norm": 44.05399703979492,
+      "learning_rate": 9.397333333333334e-06,
+      "loss": 1.6319,
+      "step": 36070
+    },
+    {
+      "epoch": 1.15456,
+      "grad_norm": 44.33708953857422,
+      "learning_rate": 9.393777777777779e-06,
+      "loss": 1.642,
+      "step": 36080
+    },
+    {
+      "epoch": 1.15488,
+      "grad_norm": 43.19767761230469,
+      "learning_rate": 9.390222222222223e-06,
+      "loss": 1.5925,
+      "step": 36090
+    },
+    {
+      "epoch": 1.1552,
+      "grad_norm": 55.544673919677734,
+      "learning_rate": 9.386666666666668e-06,
+      "loss": 1.5988,
+      "step": 36100
+    },
+    {
+      "epoch": 1.15552,
+      "grad_norm": 43.77720260620117,
+      "learning_rate": 9.383111111111112e-06,
+      "loss": 1.6369,
+      "step": 36110
+    },
+    {
+      "epoch": 1.15584,
+      "grad_norm": 44.2166748046875,
+      "learning_rate": 9.379555555555557e-06,
+      "loss": 1.6292,
+      "step": 36120
+    },
+    {
+      "epoch": 1.15616,
+      "grad_norm": 44.4605598449707,
+      "learning_rate": 9.376000000000001e-06,
+      "loss": 1.6638,
+      "step": 36130
+    },
+    {
+      "epoch": 1.15648,
+      "grad_norm": 43.97906494140625,
+      "learning_rate": 9.372444444444446e-06,
+      "loss": 1.6261,
+      "step": 36140
+    },
+    {
+      "epoch": 1.1568,
+      "grad_norm": 42.21242141723633,
+      "learning_rate": 9.368888888888889e-06,
+      "loss": 1.6303,
+      "step": 36150
+    },
+    {
+      "epoch": 1.15712,
+      "grad_norm": 45.19140625,
+      "learning_rate": 9.365333333333333e-06,
+      "loss": 1.6001,
+      "step": 36160
+    },
+    {
+      "epoch": 1.15744,
+      "grad_norm": 45.53489685058594,
+      "learning_rate": 9.361777777777778e-06,
+      "loss": 1.6404,
+      "step": 36170
+    },
+    {
+      "epoch": 1.1577600000000001,
+      "grad_norm": 48.85858917236328,
+      "learning_rate": 9.358222222222222e-06,
+      "loss": 1.6627,
+      "step": 36180
+    },
+    {
+      "epoch": 1.15808,
+      "grad_norm": 44.05198669433594,
+      "learning_rate": 9.354666666666667e-06,
+      "loss": 1.645,
+      "step": 36190
+    },
+    {
+      "epoch": 1.1584,
+      "grad_norm": 44.831424713134766,
+      "learning_rate": 9.351111111111112e-06,
+      "loss": 1.6328,
+      "step": 36200
+    },
+    {
+      "epoch": 1.15872,
+      "grad_norm": 43.89720153808594,
+      "learning_rate": 9.347555555555556e-06,
+      "loss": 1.6223,
+      "step": 36210
+    },
+    {
+      "epoch": 1.15904,
+      "grad_norm": 43.87096405029297,
+      "learning_rate": 9.344e-06,
+      "loss": 1.6305,
+      "step": 36220
+    },
+    {
+      "epoch": 1.15936,
+      "grad_norm": 45.65452194213867,
+      "learning_rate": 9.340444444444445e-06,
+      "loss": 1.6264,
+      "step": 36230
+    },
+    {
+      "epoch": 1.15968,
+      "grad_norm": 45.44213104248047,
+      "learning_rate": 9.33688888888889e-06,
+      "loss": 1.6148,
+      "step": 36240
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 43.852787017822266,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.6007,
+      "step": 36250
+    },
+    {
+      "epoch": 1.16032,
+      "grad_norm": 42.73979568481445,
+      "learning_rate": 9.329777777777779e-06,
+      "loss": 1.6034,
+      "step": 36260
+    },
+    {
+      "epoch": 1.16064,
+      "grad_norm": 44.16573715209961,
+      "learning_rate": 9.326222222222223e-06,
+      "loss": 1.5928,
+      "step": 36270
+    },
+    {
+      "epoch": 1.16096,
+      "grad_norm": 44.12287139892578,
+      "learning_rate": 9.322666666666668e-06,
+      "loss": 1.6128,
+      "step": 36280
+    },
+    {
+      "epoch": 1.16128,
+      "grad_norm": 46.16853713989258,
+      "learning_rate": 9.319111111111113e-06,
+      "loss": 1.6232,
+      "step": 36290
+    },
+    {
+      "epoch": 1.1616,
+      "grad_norm": 44.73285675048828,
+      "learning_rate": 9.315555555555557e-06,
+      "loss": 1.6311,
+      "step": 36300
+    },
+    {
+      "epoch": 1.16192,
+      "grad_norm": 44.60889434814453,
+      "learning_rate": 9.312000000000002e-06,
+      "loss": 1.5997,
+      "step": 36310
+    },
+    {
+      "epoch": 1.16224,
+      "grad_norm": 42.61589050292969,
+      "learning_rate": 9.308444444444446e-06,
+      "loss": 1.5993,
+      "step": 36320
+    },
+    {
+      "epoch": 1.16256,
+      "grad_norm": 43.2819938659668,
+      "learning_rate": 9.30488888888889e-06,
+      "loss": 1.6166,
+      "step": 36330
+    },
+    {
+      "epoch": 1.16288,
+      "grad_norm": 43.691932678222656,
+      "learning_rate": 9.301333333333334e-06,
+      "loss": 1.6158,
+      "step": 36340
+    },
+    {
+      "epoch": 1.1632,
+      "grad_norm": 45.57815170288086,
+      "learning_rate": 9.297777777777778e-06,
+      "loss": 1.6539,
+      "step": 36350
+    },
+    {
+      "epoch": 1.16352,
+      "grad_norm": 45.618343353271484,
+      "learning_rate": 9.294222222222223e-06,
+      "loss": 1.6236,
+      "step": 36360
+    },
+    {
+      "epoch": 1.16384,
+      "grad_norm": 43.39927291870117,
+      "learning_rate": 9.290666666666667e-06,
+      "loss": 1.5714,
+      "step": 36370
+    },
+    {
+      "epoch": 1.16416,
+      "grad_norm": 43.51185607910156,
+      "learning_rate": 9.287111111111112e-06,
+      "loss": 1.6148,
+      "step": 36380
+    },
+    {
+      "epoch": 1.16448,
+      "grad_norm": 44.672054290771484,
+      "learning_rate": 9.283555555555556e-06,
+      "loss": 1.6241,
+      "step": 36390
+    },
+    {
+      "epoch": 1.1648,
+      "grad_norm": 44.70090103149414,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 1.6138,
+      "step": 36400
+    },
+    {
+      "epoch": 1.16512,
+      "grad_norm": 46.62152099609375,
+      "learning_rate": 9.276444444444445e-06,
+      "loss": 1.6375,
+      "step": 36410
+    },
+    {
+      "epoch": 1.16544,
+      "grad_norm": 44.284019470214844,
+      "learning_rate": 9.27288888888889e-06,
+      "loss": 1.6125,
+      "step": 36420
+    },
+    {
+      "epoch": 1.16576,
+      "grad_norm": 43.17630386352539,
+      "learning_rate": 9.269333333333335e-06,
+      "loss": 1.6245,
+      "step": 36430
+    },
+    {
+      "epoch": 1.16608,
+      "grad_norm": 44.420249938964844,
+      "learning_rate": 9.265777777777777e-06,
+      "loss": 1.6003,
+      "step": 36440
+    },
+    {
+      "epoch": 1.1663999999999999,
+      "grad_norm": 46.07585525512695,
+      "learning_rate": 9.262222222222222e-06,
+      "loss": 1.6219,
+      "step": 36450
+    },
+    {
+      "epoch": 1.16672,
+      "grad_norm": 45.58720779418945,
+      "learning_rate": 9.258666666666667e-06,
+      "loss": 1.6262,
+      "step": 36460
+    },
+    {
+      "epoch": 1.16704,
+      "grad_norm": 45.04667282104492,
+      "learning_rate": 9.255111111111111e-06,
+      "loss": 1.6299,
+      "step": 36470
+    },
+    {
+      "epoch": 1.16736,
+      "grad_norm": 44.7858772277832,
+      "learning_rate": 9.251555555555556e-06,
+      "loss": 1.5922,
+      "step": 36480
+    },
+    {
+      "epoch": 1.16768,
+      "grad_norm": 45.5151252746582,
+      "learning_rate": 9.248e-06,
+      "loss": 1.6394,
+      "step": 36490
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 44.810909271240234,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 1.6145,
+      "step": 36500
+    },
+    {
+      "epoch": 1.16832,
+      "grad_norm": 44.83076858520508,
+      "learning_rate": 9.24088888888889e-06,
+      "loss": 1.6148,
+      "step": 36510
+    },
+    {
+      "epoch": 1.16864,
+      "grad_norm": 45.17132568359375,
+      "learning_rate": 9.237333333333334e-06,
+      "loss": 1.6096,
+      "step": 36520
+    },
+    {
+      "epoch": 1.16896,
+      "grad_norm": 45.50776290893555,
+      "learning_rate": 9.233777777777778e-06,
+      "loss": 1.596,
+      "step": 36530
+    },
+    {
+      "epoch": 1.16928,
+      "grad_norm": 45.95411682128906,
+      "learning_rate": 9.230222222222223e-06,
+      "loss": 1.5717,
+      "step": 36540
+    },
+    {
+      "epoch": 1.1696,
+      "grad_norm": 44.267269134521484,
+      "learning_rate": 9.226666666666668e-06,
+      "loss": 1.6102,
+      "step": 36550
+    },
+    {
+      "epoch": 1.16992,
+      "grad_norm": 43.271270751953125,
+      "learning_rate": 9.223111111111112e-06,
+      "loss": 1.6105,
+      "step": 36560
+    },
+    {
+      "epoch": 1.17024,
+      "grad_norm": 45.06768798828125,
+      "learning_rate": 9.219555555555557e-06,
+      "loss": 1.5963,
+      "step": 36570
+    },
+    {
+      "epoch": 1.17056,
+      "grad_norm": 43.08013153076172,
+      "learning_rate": 9.216000000000001e-06,
+      "loss": 1.634,
+      "step": 36580
+    },
+    {
+      "epoch": 1.17088,
+      "grad_norm": 44.94239044189453,
+      "learning_rate": 9.212444444444446e-06,
+      "loss": 1.6408,
+      "step": 36590
+    },
+    {
+      "epoch": 1.1712,
+      "grad_norm": 42.683597564697266,
+      "learning_rate": 9.208888888888889e-06,
+      "loss": 1.6013,
+      "step": 36600
+    },
+    {
+      "epoch": 1.1715200000000001,
+      "grad_norm": 43.48340606689453,
+      "learning_rate": 9.205333333333333e-06,
+      "loss": 1.6343,
+      "step": 36610
+    },
+    {
+      "epoch": 1.17184,
+      "grad_norm": 44.20817565917969,
+      "learning_rate": 9.201777777777778e-06,
+      "loss": 1.6184,
+      "step": 36620
+    },
+    {
+      "epoch": 1.17216,
+      "grad_norm": 44.68756103515625,
+      "learning_rate": 9.198222222222222e-06,
+      "loss": 1.5928,
+      "step": 36630
+    },
+    {
+      "epoch": 1.17248,
+      "grad_norm": 45.76731491088867,
+      "learning_rate": 9.194666666666667e-06,
+      "loss": 1.6248,
+      "step": 36640
+    },
+    {
+      "epoch": 1.1728,
+      "grad_norm": 45.566612243652344,
+      "learning_rate": 9.191111111111111e-06,
+      "loss": 1.6358,
+      "step": 36650
+    },
+    {
+      "epoch": 1.17312,
+      "grad_norm": 44.99404525756836,
+      "learning_rate": 9.187555555555556e-06,
+      "loss": 1.594,
+      "step": 36660
+    },
+    {
+      "epoch": 1.17344,
+      "grad_norm": 45.59415817260742,
+      "learning_rate": 9.184e-06,
+      "loss": 1.6198,
+      "step": 36670
+    },
+    {
+      "epoch": 1.17376,
+      "grad_norm": 44.73423767089844,
+      "learning_rate": 9.180444444444445e-06,
+      "loss": 1.6289,
+      "step": 36680
+    },
+    {
+      "epoch": 1.17408,
+      "grad_norm": 42.23807907104492,
+      "learning_rate": 9.17688888888889e-06,
+      "loss": 1.6122,
+      "step": 36690
+    },
+    {
+      "epoch": 1.1743999999999999,
+      "grad_norm": 46.371463775634766,
+      "learning_rate": 9.173333333333334e-06,
+      "loss": 1.593,
+      "step": 36700
+    },
+    {
+      "epoch": 1.17472,
+      "grad_norm": 42.88974380493164,
+      "learning_rate": 9.169777777777779e-06,
+      "loss": 1.5967,
+      "step": 36710
+    },
+    {
+      "epoch": 1.17504,
+      "grad_norm": 45.92631149291992,
+      "learning_rate": 9.166222222222223e-06,
+      "loss": 1.5994,
+      "step": 36720
+    },
+    {
+      "epoch": 1.17536,
+      "grad_norm": 42.9425048828125,
+      "learning_rate": 9.162666666666668e-06,
+      "loss": 1.6348,
+      "step": 36730
+    },
+    {
+      "epoch": 1.17568,
+      "grad_norm": 44.71460723876953,
+      "learning_rate": 9.159111111111112e-06,
+      "loss": 1.6139,
+      "step": 36740
+    },
+    {
+      "epoch": 1.176,
+      "grad_norm": 45.90800476074219,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 1.6034,
+      "step": 36750
+    },
+    {
+      "epoch": 1.17632,
+      "grad_norm": 45.773597717285156,
+      "learning_rate": 9.152000000000001e-06,
+      "loss": 1.6491,
+      "step": 36760
+    },
+    {
+      "epoch": 1.17664,
+      "grad_norm": 45.82653045654297,
+      "learning_rate": 9.148444444444446e-06,
+      "loss": 1.6415,
+      "step": 36770
+    },
+    {
+      "epoch": 1.17696,
+      "grad_norm": 43.5515022277832,
+      "learning_rate": 9.14488888888889e-06,
+      "loss": 1.6281,
+      "step": 36780
+    },
+    {
+      "epoch": 1.17728,
+      "grad_norm": 43.16986846923828,
+      "learning_rate": 9.141333333333333e-06,
+      "loss": 1.6479,
+      "step": 36790
+    },
+    {
+      "epoch": 1.1776,
+      "grad_norm": 46.149314880371094,
+      "learning_rate": 9.137777777777778e-06,
+      "loss": 1.6163,
+      "step": 36800
+    },
+    {
+      "epoch": 1.17792,
+      "grad_norm": 42.756710052490234,
+      "learning_rate": 9.134222222222223e-06,
+      "loss": 1.6276,
+      "step": 36810
+    },
+    {
+      "epoch": 1.17824,
+      "grad_norm": 45.28952407836914,
+      "learning_rate": 9.130666666666667e-06,
+      "loss": 1.6396,
+      "step": 36820
+    },
+    {
+      "epoch": 1.17856,
+      "grad_norm": 43.20062255859375,
+      "learning_rate": 9.127111111111112e-06,
+      "loss": 1.614,
+      "step": 36830
+    },
+    {
+      "epoch": 1.17888,
+      "grad_norm": 45.23955154418945,
+      "learning_rate": 9.123555555555556e-06,
+      "loss": 1.6036,
+      "step": 36840
+    },
+    {
+      "epoch": 1.1792,
+      "grad_norm": 44.820648193359375,
+      "learning_rate": 9.12e-06,
+      "loss": 1.6039,
+      "step": 36850
+    },
+    {
+      "epoch": 1.1795200000000001,
+      "grad_norm": 45.45117950439453,
+      "learning_rate": 9.116444444444445e-06,
+      "loss": 1.6328,
+      "step": 36860
+    },
+    {
+      "epoch": 1.17984,
+      "grad_norm": 44.188228607177734,
+      "learning_rate": 9.11288888888889e-06,
+      "loss": 1.611,
+      "step": 36870
+    },
+    {
+      "epoch": 1.1801599999999999,
+      "grad_norm": 43.97381591796875,
+      "learning_rate": 9.109333333333334e-06,
+      "loss": 1.6189,
+      "step": 36880
+    },
+    {
+      "epoch": 1.18048,
+      "grad_norm": 44.498863220214844,
+      "learning_rate": 9.105777777777779e-06,
+      "loss": 1.6056,
+      "step": 36890
+    },
+    {
+      "epoch": 1.1808,
+      "grad_norm": 45.77372360229492,
+      "learning_rate": 9.102222222222224e-06,
+      "loss": 1.6487,
+      "step": 36900
+    },
+    {
+      "epoch": 1.18112,
+      "grad_norm": 45.980873107910156,
+      "learning_rate": 9.098666666666668e-06,
+      "loss": 1.5998,
+      "step": 36910
+    },
+    {
+      "epoch": 1.18144,
+      "grad_norm": 46.08082962036133,
+      "learning_rate": 9.095111111111113e-06,
+      "loss": 1.6208,
+      "step": 36920
+    },
+    {
+      "epoch": 1.18176,
+      "grad_norm": 44.75496292114258,
+      "learning_rate": 9.091555555555557e-06,
+      "loss": 1.6204,
+      "step": 36930
+    },
+    {
+      "epoch": 1.18208,
+      "grad_norm": 43.503047943115234,
+      "learning_rate": 9.088000000000002e-06,
+      "loss": 1.5918,
+      "step": 36940
+    },
+    {
+      "epoch": 1.1824,
+      "grad_norm": 43.14474868774414,
+      "learning_rate": 9.084444444444446e-06,
+      "loss": 1.6343,
+      "step": 36950
+    },
+    {
+      "epoch": 1.18272,
+      "grad_norm": 44.471099853515625,
+      "learning_rate": 9.08088888888889e-06,
+      "loss": 1.6001,
+      "step": 36960
+    },
+    {
+      "epoch": 1.18304,
+      "grad_norm": 44.57168960571289,
+      "learning_rate": 9.077333333333334e-06,
+      "loss": 1.6016,
+      "step": 36970
+    },
+    {
+      "epoch": 1.18336,
+      "grad_norm": 45.980613708496094,
+      "learning_rate": 9.073777777777778e-06,
+      "loss": 1.6027,
+      "step": 36980
+    },
+    {
+      "epoch": 1.18368,
+      "grad_norm": 43.24493408203125,
+      "learning_rate": 9.070222222222223e-06,
+      "loss": 1.6249,
+      "step": 36990
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 42.255821228027344,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.6003,
+      "step": 37000
+    },
+    {
+      "epoch": 1.18432,
+      "grad_norm": 44.33125305175781,
+      "learning_rate": 9.063111111111112e-06,
+      "loss": 1.6218,
+      "step": 37010
+    },
+    {
+      "epoch": 1.18464,
+      "grad_norm": 43.5249137878418,
+      "learning_rate": 9.059555555555556e-06,
+      "loss": 1.633,
+      "step": 37020
+    },
+    {
+      "epoch": 1.18496,
+      "grad_norm": 45.45284652709961,
+      "learning_rate": 9.056000000000001e-06,
+      "loss": 1.6085,
+      "step": 37030
+    },
+    {
+      "epoch": 1.1852800000000001,
+      "grad_norm": 43.63113784790039,
+      "learning_rate": 9.052444444444446e-06,
+      "loss": 1.5775,
+      "step": 37040
+    },
+    {
+      "epoch": 1.1856,
+      "grad_norm": 44.04764175415039,
+      "learning_rate": 9.048888888888888e-06,
+      "loss": 1.5973,
+      "step": 37050
+    },
+    {
+      "epoch": 1.18592,
+      "grad_norm": 47.08494567871094,
+      "learning_rate": 9.045333333333333e-06,
+      "loss": 1.6013,
+      "step": 37060
+    },
+    {
+      "epoch": 1.18624,
+      "grad_norm": 44.20665740966797,
+      "learning_rate": 9.041777777777778e-06,
+      "loss": 1.6115,
+      "step": 37070
+    },
+    {
+      "epoch": 1.18656,
+      "grad_norm": 43.61125564575195,
+      "learning_rate": 9.038222222222222e-06,
+      "loss": 1.632,
+      "step": 37080
+    },
+    {
+      "epoch": 1.18688,
+      "grad_norm": 43.4201774597168,
+      "learning_rate": 9.034666666666667e-06,
+      "loss": 1.6255,
+      "step": 37090
+    },
+    {
+      "epoch": 1.1872,
+      "grad_norm": 45.814422607421875,
+      "learning_rate": 9.031111111111111e-06,
+      "loss": 1.6289,
+      "step": 37100
+    },
+    {
+      "epoch": 1.18752,
+      "grad_norm": 44.91145706176758,
+      "learning_rate": 9.027555555555556e-06,
+      "loss": 1.6335,
+      "step": 37110
+    },
+    {
+      "epoch": 1.18784,
+      "grad_norm": 45.26153564453125,
+      "learning_rate": 9.024e-06,
+      "loss": 1.5778,
+      "step": 37120
+    },
+    {
+      "epoch": 1.1881599999999999,
+      "grad_norm": 43.415077209472656,
+      "learning_rate": 9.020444444444445e-06,
+      "loss": 1.6352,
+      "step": 37130
+    },
+    {
+      "epoch": 1.18848,
+      "grad_norm": 46.07261276245117,
+      "learning_rate": 9.01688888888889e-06,
+      "loss": 1.6045,
+      "step": 37140
+    },
+    {
+      "epoch": 1.1888,
+      "grad_norm": 50.8298225402832,
+      "learning_rate": 9.013333333333334e-06,
+      "loss": 1.6127,
+      "step": 37150
+    },
+    {
+      "epoch": 1.18912,
+      "grad_norm": 45.64514923095703,
+      "learning_rate": 9.009777777777779e-06,
+      "loss": 1.6337,
+      "step": 37160
+    },
+    {
+      "epoch": 1.18944,
+      "grad_norm": 45.26530456542969,
+      "learning_rate": 9.006222222222223e-06,
+      "loss": 1.6126,
+      "step": 37170
+    },
+    {
+      "epoch": 1.18976,
+      "grad_norm": 44.7235107421875,
+      "learning_rate": 9.002666666666668e-06,
+      "loss": 1.6181,
+      "step": 37180
+    },
+    {
+      "epoch": 1.19008,
+      "grad_norm": 45.90180969238281,
+      "learning_rate": 8.999111111111112e-06,
+      "loss": 1.6258,
+      "step": 37190
+    },
+    {
+      "epoch": 1.1904,
+      "grad_norm": 42.16521072387695,
+      "learning_rate": 8.995555555555557e-06,
+      "loss": 1.6327,
+      "step": 37200
+    },
+    {
+      "epoch": 1.19072,
+      "grad_norm": 47.36225509643555,
+      "learning_rate": 8.992000000000001e-06,
+      "loss": 1.6321,
+      "step": 37210
+    },
+    {
+      "epoch": 1.19104,
+      "grad_norm": 45.1246337890625,
+      "learning_rate": 8.988444444444446e-06,
+      "loss": 1.6306,
+      "step": 37220
+    },
+    {
+      "epoch": 1.19136,
+      "grad_norm": 43.45327377319336,
+      "learning_rate": 8.98488888888889e-06,
+      "loss": 1.6261,
+      "step": 37230
+    },
+    {
+      "epoch": 1.19168,
+      "grad_norm": 47.724151611328125,
+      "learning_rate": 8.981333333333333e-06,
+      "loss": 1.6076,
+      "step": 37240
+    },
+    {
+      "epoch": 1.192,
+      "grad_norm": 43.875343322753906,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 1.6254,
+      "step": 37250
+    },
+    {
+      "epoch": 1.19232,
+      "grad_norm": 43.029109954833984,
+      "learning_rate": 8.974222222222222e-06,
+      "loss": 1.641,
+      "step": 37260
+    },
+    {
+      "epoch": 1.19264,
+      "grad_norm": 43.80043029785156,
+      "learning_rate": 8.970666666666667e-06,
+      "loss": 1.6183,
+      "step": 37270
+    },
+    {
+      "epoch": 1.19296,
+      "grad_norm": 45.554203033447266,
+      "learning_rate": 8.967111111111112e-06,
+      "loss": 1.6157,
+      "step": 37280
+    },
+    {
+      "epoch": 1.1932800000000001,
+      "grad_norm": 43.858116149902344,
+      "learning_rate": 8.963555555555556e-06,
+      "loss": 1.604,
+      "step": 37290
+    },
+    {
+      "epoch": 1.1936,
+      "grad_norm": 42.90398025512695,
+      "learning_rate": 8.96e-06,
+      "loss": 1.5865,
+      "step": 37300
+    },
+    {
+      "epoch": 1.19392,
+      "grad_norm": 45.20762634277344,
+      "learning_rate": 8.956444444444445e-06,
+      "loss": 1.6424,
+      "step": 37310
+    },
+    {
+      "epoch": 1.19424,
+      "grad_norm": 49.68057632446289,
+      "learning_rate": 8.95288888888889e-06,
+      "loss": 1.6262,
+      "step": 37320
+    },
+    {
+      "epoch": 1.19456,
+      "grad_norm": 45.9451904296875,
+      "learning_rate": 8.949333333333334e-06,
+      "loss": 1.6375,
+      "step": 37330
+    },
+    {
+      "epoch": 1.19488,
+      "grad_norm": 45.719268798828125,
+      "learning_rate": 8.945777777777779e-06,
+      "loss": 1.6278,
+      "step": 37340
+    },
+    {
+      "epoch": 1.1952,
+      "grad_norm": 48.056095123291016,
+      "learning_rate": 8.942222222222223e-06,
+      "loss": 1.6007,
+      "step": 37350
+    },
+    {
+      "epoch": 1.19552,
+      "grad_norm": 44.8497428894043,
+      "learning_rate": 8.938666666666668e-06,
+      "loss": 1.6129,
+      "step": 37360
+    },
+    {
+      "epoch": 1.19584,
+      "grad_norm": 44.91301727294922,
+      "learning_rate": 8.935111111111112e-06,
+      "loss": 1.5899,
+      "step": 37370
+    },
+    {
+      "epoch": 1.19616,
+      "grad_norm": 44.85026168823242,
+      "learning_rate": 8.931555555555557e-06,
+      "loss": 1.5988,
+      "step": 37380
+    },
+    {
+      "epoch": 1.19648,
+      "grad_norm": 44.962589263916016,
+      "learning_rate": 8.928000000000002e-06,
+      "loss": 1.6004,
+      "step": 37390
+    },
+    {
+      "epoch": 1.1968,
+      "grad_norm": 43.96834945678711,
+      "learning_rate": 8.924444444444446e-06,
+      "loss": 1.6294,
+      "step": 37400
+    },
+    {
+      "epoch": 1.19712,
+      "grad_norm": 45.65105056762695,
+      "learning_rate": 8.92088888888889e-06,
+      "loss": 1.6187,
+      "step": 37410
+    },
+    {
+      "epoch": 1.19744,
+      "grad_norm": 45.473331451416016,
+      "learning_rate": 8.917333333333335e-06,
+      "loss": 1.6166,
+      "step": 37420
+    },
+    {
+      "epoch": 1.19776,
+      "grad_norm": 46.43606185913086,
+      "learning_rate": 8.91377777777778e-06,
+      "loss": 1.6146,
+      "step": 37430
+    },
+    {
+      "epoch": 1.19808,
+      "grad_norm": 45.848690032958984,
+      "learning_rate": 8.910222222222223e-06,
+      "loss": 1.6137,
+      "step": 37440
+    },
+    {
+      "epoch": 1.1984,
+      "grad_norm": 44.7730827331543,
+      "learning_rate": 8.906666666666667e-06,
+      "loss": 1.6347,
+      "step": 37450
+    },
+    {
+      "epoch": 1.19872,
+      "grad_norm": 44.687313079833984,
+      "learning_rate": 8.903111111111112e-06,
+      "loss": 1.6101,
+      "step": 37460
+    },
+    {
+      "epoch": 1.19904,
+      "grad_norm": 43.56023025512695,
+      "learning_rate": 8.899555555555556e-06,
+      "loss": 1.6436,
+      "step": 37470
+    },
+    {
+      "epoch": 1.19936,
+      "grad_norm": 46.550960540771484,
+      "learning_rate": 8.896000000000001e-06,
+      "loss": 1.617,
+      "step": 37480
+    },
+    {
+      "epoch": 1.19968,
+      "grad_norm": 44.08442306518555,
+      "learning_rate": 8.892444444444445e-06,
+      "loss": 1.5949,
+      "step": 37490
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 47.61814880371094,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.6127,
+      "step": 37500
+    },
+    {
+      "epoch": 1.20032,
+      "grad_norm": 45.213722229003906,
+      "learning_rate": 8.885333333333333e-06,
+      "loss": 1.6172,
+      "step": 37510
+    },
+    {
+      "epoch": 1.20064,
+      "grad_norm": 43.18437194824219,
+      "learning_rate": 8.881777777777777e-06,
+      "loss": 1.5815,
+      "step": 37520
+    },
+    {
+      "epoch": 1.20096,
+      "grad_norm": 44.670265197753906,
+      "learning_rate": 8.878222222222222e-06,
+      "loss": 1.6497,
+      "step": 37530
+    },
+    {
+      "epoch": 1.20128,
+      "grad_norm": 43.545494079589844,
+      "learning_rate": 8.874666666666667e-06,
+      "loss": 1.6373,
+      "step": 37540
+    },
+    {
+      "epoch": 1.2016,
+      "grad_norm": 46.32993698120117,
+      "learning_rate": 8.871111111111111e-06,
+      "loss": 1.6065,
+      "step": 37550
+    },
+    {
+      "epoch": 1.2019199999999999,
+      "grad_norm": 46.05718994140625,
+      "learning_rate": 8.867555555555556e-06,
+      "loss": 1.6116,
+      "step": 37560
+    },
+    {
+      "epoch": 1.20224,
+      "grad_norm": 43.07898712158203,
+      "learning_rate": 8.864e-06,
+      "loss": 1.6107,
+      "step": 37570
+    },
+    {
+      "epoch": 1.20256,
+      "grad_norm": 45.01841354370117,
+      "learning_rate": 8.860444444444445e-06,
+      "loss": 1.611,
+      "step": 37580
+    },
+    {
+      "epoch": 1.20288,
+      "grad_norm": 44.30756759643555,
+      "learning_rate": 8.85688888888889e-06,
+      "loss": 1.6408,
+      "step": 37590
+    },
+    {
+      "epoch": 1.2032,
+      "grad_norm": 44.6888427734375,
+      "learning_rate": 8.853333333333334e-06,
+      "loss": 1.5964,
+      "step": 37600
+    },
+    {
+      "epoch": 1.20352,
+      "grad_norm": 43.844486236572266,
+      "learning_rate": 8.849777777777778e-06,
+      "loss": 1.6205,
+      "step": 37610
+    },
+    {
+      "epoch": 1.20384,
+      "grad_norm": 44.9466552734375,
+      "learning_rate": 8.846222222222223e-06,
+      "loss": 1.6416,
+      "step": 37620
+    },
+    {
+      "epoch": 1.20416,
+      "grad_norm": 44.492984771728516,
+      "learning_rate": 8.842666666666667e-06,
+      "loss": 1.6136,
+      "step": 37630
+    },
+    {
+      "epoch": 1.20448,
+      "grad_norm": 45.26192855834961,
+      "learning_rate": 8.839111111111112e-06,
+      "loss": 1.5964,
+      "step": 37640
+    },
+    {
+      "epoch": 1.2048,
+      "grad_norm": 41.83114242553711,
+      "learning_rate": 8.835555555555557e-06,
+      "loss": 1.6058,
+      "step": 37650
+    },
+    {
+      "epoch": 1.20512,
+      "grad_norm": 45.18032455444336,
+      "learning_rate": 8.832000000000001e-06,
+      "loss": 1.6301,
+      "step": 37660
+    },
+    {
+      "epoch": 1.20544,
+      "grad_norm": 43.96016311645508,
+      "learning_rate": 8.828444444444446e-06,
+      "loss": 1.5992,
+      "step": 37670
+    },
+    {
+      "epoch": 1.20576,
+      "grad_norm": 44.86780548095703,
+      "learning_rate": 8.82488888888889e-06,
+      "loss": 1.6098,
+      "step": 37680
+    },
+    {
+      "epoch": 1.20608,
+      "grad_norm": 45.063751220703125,
+      "learning_rate": 8.821333333333333e-06,
+      "loss": 1.6084,
+      "step": 37690
+    },
+    {
+      "epoch": 1.2064,
+      "grad_norm": 44.57073211669922,
+      "learning_rate": 8.817777777777778e-06,
+      "loss": 1.6194,
+      "step": 37700
+    },
+    {
+      "epoch": 1.20672,
+      "grad_norm": 45.337684631347656,
+      "learning_rate": 8.814222222222222e-06,
+      "loss": 1.617,
+      "step": 37710
+    },
+    {
+      "epoch": 1.2070400000000001,
+      "grad_norm": 43.16669845581055,
+      "learning_rate": 8.810666666666667e-06,
+      "loss": 1.649,
+      "step": 37720
+    },
+    {
+      "epoch": 1.20736,
+      "grad_norm": 43.83868408203125,
+      "learning_rate": 8.807111111111111e-06,
+      "loss": 1.627,
+      "step": 37730
+    },
+    {
+      "epoch": 1.20768,
+      "grad_norm": 46.1161003112793,
+      "learning_rate": 8.803555555555556e-06,
+      "loss": 1.6164,
+      "step": 37740
+    },
+    {
+      "epoch": 1.208,
+      "grad_norm": 43.969825744628906,
+      "learning_rate": 8.8e-06,
+      "loss": 1.5992,
+      "step": 37750
+    },
+    {
+      "epoch": 1.20832,
+      "grad_norm": 45.1153450012207,
+      "learning_rate": 8.796444444444445e-06,
+      "loss": 1.601,
+      "step": 37760
+    },
+    {
+      "epoch": 1.20864,
+      "grad_norm": 45.29838180541992,
+      "learning_rate": 8.79288888888889e-06,
+      "loss": 1.6346,
+      "step": 37770
+    },
+    {
+      "epoch": 1.20896,
+      "grad_norm": 43.7877082824707,
+      "learning_rate": 8.789333333333334e-06,
+      "loss": 1.59,
+      "step": 37780
+    },
+    {
+      "epoch": 1.20928,
+      "grad_norm": 44.93843460083008,
+      "learning_rate": 8.785777777777779e-06,
+      "loss": 1.5945,
+      "step": 37790
+    },
+    {
+      "epoch": 1.2096,
+      "grad_norm": 46.926719665527344,
+      "learning_rate": 8.782222222222223e-06,
+      "loss": 1.622,
+      "step": 37800
+    },
+    {
+      "epoch": 1.2099199999999999,
+      "grad_norm": 46.099117279052734,
+      "learning_rate": 8.778666666666668e-06,
+      "loss": 1.6455,
+      "step": 37810
+    },
+    {
+      "epoch": 1.21024,
+      "grad_norm": 46.05337142944336,
+      "learning_rate": 8.775111111111112e-06,
+      "loss": 1.5997,
+      "step": 37820
+    },
+    {
+      "epoch": 1.21056,
+      "grad_norm": 43.47398376464844,
+      "learning_rate": 8.771555555555557e-06,
+      "loss": 1.6133,
+      "step": 37830
+    },
+    {
+      "epoch": 1.21088,
+      "grad_norm": 46.43062973022461,
+      "learning_rate": 8.768000000000001e-06,
+      "loss": 1.6512,
+      "step": 37840
+    },
+    {
+      "epoch": 1.2112,
+      "grad_norm": 49.61890411376953,
+      "learning_rate": 8.764444444444446e-06,
+      "loss": 1.5892,
+      "step": 37850
+    },
+    {
+      "epoch": 1.21152,
+      "grad_norm": 43.67453384399414,
+      "learning_rate": 8.76088888888889e-06,
+      "loss": 1.5901,
+      "step": 37860
+    },
+    {
+      "epoch": 1.21184,
+      "grad_norm": 42.86765670776367,
+      "learning_rate": 8.757333333333335e-06,
+      "loss": 1.6123,
+      "step": 37870
+    },
+    {
+      "epoch": 1.21216,
+      "grad_norm": 44.75029754638672,
+      "learning_rate": 8.75377777777778e-06,
+      "loss": 1.5822,
+      "step": 37880
+    },
+    {
+      "epoch": 1.21248,
+      "grad_norm": 45.341773986816406,
+      "learning_rate": 8.750222222222223e-06,
+      "loss": 1.6031,
+      "step": 37890
+    },
+    {
+      "epoch": 1.2128,
+      "grad_norm": 42.46044921875,
+      "learning_rate": 8.746666666666667e-06,
+      "loss": 1.6322,
+      "step": 37900
+    },
+    {
+      "epoch": 1.21312,
+      "grad_norm": 46.101863861083984,
+      "learning_rate": 8.743111111111112e-06,
+      "loss": 1.6122,
+      "step": 37910
+    },
+    {
+      "epoch": 1.21344,
+      "grad_norm": 45.06989669799805,
+      "learning_rate": 8.739555555555556e-06,
+      "loss": 1.6133,
+      "step": 37920
+    },
+    {
+      "epoch": 1.21376,
+      "grad_norm": 47.32184600830078,
+      "learning_rate": 8.736e-06,
+      "loss": 1.6447,
+      "step": 37930
+    },
+    {
+      "epoch": 1.21408,
+      "grad_norm": 43.49120330810547,
+      "learning_rate": 8.732444444444445e-06,
+      "loss": 1.6104,
+      "step": 37940
+    },
+    {
+      "epoch": 1.2144,
+      "grad_norm": 41.85742950439453,
+      "learning_rate": 8.72888888888889e-06,
+      "loss": 1.6254,
+      "step": 37950
+    },
+    {
+      "epoch": 1.21472,
+      "grad_norm": 44.84233093261719,
+      "learning_rate": 8.725333333333334e-06,
+      "loss": 1.6282,
+      "step": 37960
+    },
+    {
+      "epoch": 1.2150400000000001,
+      "grad_norm": 45.8273811340332,
+      "learning_rate": 8.721777777777779e-06,
+      "loss": 1.6182,
+      "step": 37970
+    },
+    {
+      "epoch": 1.21536,
+      "grad_norm": 44.68730163574219,
+      "learning_rate": 8.718222222222223e-06,
+      "loss": 1.6062,
+      "step": 37980
+    },
+    {
+      "epoch": 1.21568,
+      "grad_norm": 47.33452224731445,
+      "learning_rate": 8.714666666666666e-06,
+      "loss": 1.6,
+      "step": 37990
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 43.193973541259766,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 1.6034,
+      "step": 38000
+    },
+    {
+      "epoch": 1.21632,
+      "grad_norm": 45.360347747802734,
+      "learning_rate": 8.707555555555555e-06,
+      "loss": 1.6528,
+      "step": 38010
+    },
+    {
+      "epoch": 1.21664,
+      "grad_norm": 43.036781311035156,
+      "learning_rate": 8.704e-06,
+      "loss": 1.6224,
+      "step": 38020
+    },
+    {
+      "epoch": 1.21696,
+      "grad_norm": 45.99064636230469,
+      "learning_rate": 8.700444444444445e-06,
+      "loss": 1.6257,
+      "step": 38030
+    },
+    {
+      "epoch": 1.21728,
+      "grad_norm": 44.47370147705078,
+      "learning_rate": 8.696888888888889e-06,
+      "loss": 1.6098,
+      "step": 38040
+    },
+    {
+      "epoch": 1.2176,
+      "grad_norm": 44.06339645385742,
+      "learning_rate": 8.693333333333334e-06,
+      "loss": 1.6181,
+      "step": 38050
+    },
+    {
+      "epoch": 1.21792,
+      "grad_norm": 43.332637786865234,
+      "learning_rate": 8.689777777777778e-06,
+      "loss": 1.6168,
+      "step": 38060
+    },
+    {
+      "epoch": 1.21824,
+      "grad_norm": 44.66054153442383,
+      "learning_rate": 8.686222222222223e-06,
+      "loss": 1.6235,
+      "step": 38070
+    },
+    {
+      "epoch": 1.21856,
+      "grad_norm": 43.89647674560547,
+      "learning_rate": 8.682666666666667e-06,
+      "loss": 1.6104,
+      "step": 38080
+    },
+    {
+      "epoch": 1.21888,
+      "grad_norm": 45.6687126159668,
+      "learning_rate": 8.679111111111112e-06,
+      "loss": 1.6235,
+      "step": 38090
+    },
+    {
+      "epoch": 1.2192,
+      "grad_norm": 46.06743621826172,
+      "learning_rate": 8.675555555555556e-06,
+      "loss": 1.615,
+      "step": 38100
+    },
+    {
+      "epoch": 1.21952,
+      "grad_norm": 43.16071701049805,
+      "learning_rate": 8.672000000000001e-06,
+      "loss": 1.6079,
+      "step": 38110
+    },
+    {
+      "epoch": 1.21984,
+      "grad_norm": 43.79576873779297,
+      "learning_rate": 8.668444444444446e-06,
+      "loss": 1.6133,
+      "step": 38120
+    },
+    {
+      "epoch": 1.22016,
+      "grad_norm": 43.21852111816406,
+      "learning_rate": 8.66488888888889e-06,
+      "loss": 1.6361,
+      "step": 38130
+    },
+    {
+      "epoch": 1.22048,
+      "grad_norm": 44.7907829284668,
+      "learning_rate": 8.661333333333335e-06,
+      "loss": 1.6229,
+      "step": 38140
+    },
+    {
+      "epoch": 1.2208,
+      "grad_norm": 45.25461196899414,
+      "learning_rate": 8.657777777777778e-06,
+      "loss": 1.6424,
+      "step": 38150
+    },
+    {
+      "epoch": 1.22112,
+      "grad_norm": 46.63352966308594,
+      "learning_rate": 8.654222222222222e-06,
+      "loss": 1.6004,
+      "step": 38160
+    },
+    {
+      "epoch": 1.22144,
+      "grad_norm": 44.5078125,
+      "learning_rate": 8.650666666666667e-06,
+      "loss": 1.6004,
+      "step": 38170
+    },
+    {
+      "epoch": 1.22176,
+      "grad_norm": 46.22889709472656,
+      "learning_rate": 8.647111111111111e-06,
+      "loss": 1.6127,
+      "step": 38180
+    },
+    {
+      "epoch": 1.22208,
+      "grad_norm": 44.953792572021484,
+      "learning_rate": 8.643555555555556e-06,
+      "loss": 1.5963,
+      "step": 38190
+    },
+    {
+      "epoch": 1.2224,
+      "grad_norm": 45.0565071105957,
+      "learning_rate": 8.64e-06,
+      "loss": 1.6602,
+      "step": 38200
+    },
+    {
+      "epoch": 1.22272,
+      "grad_norm": 44.015262603759766,
+      "learning_rate": 8.636444444444445e-06,
+      "loss": 1.5775,
+      "step": 38210
+    },
+    {
+      "epoch": 1.22304,
+      "grad_norm": 44.347557067871094,
+      "learning_rate": 8.63288888888889e-06,
+      "loss": 1.5998,
+      "step": 38220
+    },
+    {
+      "epoch": 1.22336,
+      "grad_norm": 43.69554901123047,
+      "learning_rate": 8.629333333333334e-06,
+      "loss": 1.6048,
+      "step": 38230
+    },
+    {
+      "epoch": 1.2236799999999999,
+      "grad_norm": 45.76776123046875,
+      "learning_rate": 8.625777777777779e-06,
+      "loss": 1.6488,
+      "step": 38240
+    },
+    {
+      "epoch": 1.224,
+      "grad_norm": 46.74126052856445,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 1.6095,
+      "step": 38250
+    },
+    {
+      "epoch": 1.22432,
+      "grad_norm": 45.625762939453125,
+      "learning_rate": 8.618666666666668e-06,
+      "loss": 1.645,
+      "step": 38260
+    },
+    {
+      "epoch": 1.22464,
+      "grad_norm": 46.620445251464844,
+      "learning_rate": 8.615111111111112e-06,
+      "loss": 1.6095,
+      "step": 38270
+    },
+    {
+      "epoch": 1.22496,
+      "grad_norm": 45.9921875,
+      "learning_rate": 8.611555555555557e-06,
+      "loss": 1.5838,
+      "step": 38280
+    },
+    {
+      "epoch": 1.22528,
+      "grad_norm": 46.95634078979492,
+      "learning_rate": 8.608000000000001e-06,
+      "loss": 1.6117,
+      "step": 38290
+    },
+    {
+      "epoch": 1.2256,
+      "grad_norm": 44.37270736694336,
+      "learning_rate": 8.604444444444446e-06,
+      "loss": 1.6037,
+      "step": 38300
+    },
+    {
+      "epoch": 1.22592,
+      "grad_norm": 44.26304626464844,
+      "learning_rate": 8.60088888888889e-06,
+      "loss": 1.6024,
+      "step": 38310
+    },
+    {
+      "epoch": 1.22624,
+      "grad_norm": 44.726661682128906,
+      "learning_rate": 8.597333333333335e-06,
+      "loss": 1.6381,
+      "step": 38320
+    },
+    {
+      "epoch": 1.22656,
+      "grad_norm": 45.11967849731445,
+      "learning_rate": 8.59377777777778e-06,
+      "loss": 1.602,
+      "step": 38330
+    },
+    {
+      "epoch": 1.22688,
+      "grad_norm": 43.5937385559082,
+      "learning_rate": 8.590222222222222e-06,
+      "loss": 1.6049,
+      "step": 38340
+    },
+    {
+      "epoch": 1.2272,
+      "grad_norm": 45.37797164916992,
+      "learning_rate": 8.586666666666667e-06,
+      "loss": 1.6027,
+      "step": 38350
+    },
+    {
+      "epoch": 1.22752,
+      "grad_norm": 45.6694221496582,
+      "learning_rate": 8.583111111111111e-06,
+      "loss": 1.6565,
+      "step": 38360
+    },
+    {
+      "epoch": 1.22784,
+      "grad_norm": 44.0611686706543,
+      "learning_rate": 8.579555555555556e-06,
+      "loss": 1.6051,
+      "step": 38370
+    },
+    {
+      "epoch": 1.22816,
+      "grad_norm": 46.050472259521484,
+      "learning_rate": 8.576e-06,
+      "loss": 1.6169,
+      "step": 38380
+    },
+    {
+      "epoch": 1.22848,
+      "grad_norm": 44.67304229736328,
+      "learning_rate": 8.572444444444445e-06,
+      "loss": 1.5995,
+      "step": 38390
+    },
+    {
+      "epoch": 1.2288000000000001,
+      "grad_norm": 52.205501556396484,
+      "learning_rate": 8.56888888888889e-06,
+      "loss": 1.6195,
+      "step": 38400
+    },
+    {
+      "epoch": 1.22912,
+      "grad_norm": 44.350059509277344,
+      "learning_rate": 8.565333333333334e-06,
+      "loss": 1.628,
+      "step": 38410
+    },
+    {
+      "epoch": 1.22944,
+      "grad_norm": 44.488616943359375,
+      "learning_rate": 8.561777777777779e-06,
+      "loss": 1.6148,
+      "step": 38420
+    },
+    {
+      "epoch": 1.22976,
+      "grad_norm": 44.28691482543945,
+      "learning_rate": 8.558222222222223e-06,
+      "loss": 1.623,
+      "step": 38430
+    },
+    {
+      "epoch": 1.23008,
+      "grad_norm": 45.6378288269043,
+      "learning_rate": 8.554666666666668e-06,
+      "loss": 1.6289,
+      "step": 38440
+    },
+    {
+      "epoch": 1.2304,
+      "grad_norm": 43.49380874633789,
+      "learning_rate": 8.551111111111112e-06,
+      "loss": 1.6431,
+      "step": 38450
+    },
+    {
+      "epoch": 1.23072,
+      "grad_norm": 44.094791412353516,
+      "learning_rate": 8.547555555555557e-06,
+      "loss": 1.615,
+      "step": 38460
+    },
+    {
+      "epoch": 1.23104,
+      "grad_norm": 43.89348602294922,
+      "learning_rate": 8.544000000000002e-06,
+      "loss": 1.6217,
+      "step": 38470
+    },
+    {
+      "epoch": 1.23136,
+      "grad_norm": 44.966827392578125,
+      "learning_rate": 8.540444444444446e-06,
+      "loss": 1.5995,
+      "step": 38480
+    },
+    {
+      "epoch": 1.2316799999999999,
+      "grad_norm": 46.915889739990234,
+      "learning_rate": 8.53688888888889e-06,
+      "loss": 1.6289,
+      "step": 38490
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 44.71723937988281,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.603,
+      "step": 38500
+    },
+    {
+      "epoch": 1.23232,
+      "grad_norm": 43.65388870239258,
+      "learning_rate": 8.529777777777778e-06,
+      "loss": 1.5869,
+      "step": 38510
+    },
+    {
+      "epoch": 1.23264,
+      "grad_norm": 45.82659149169922,
+      "learning_rate": 8.526222222222223e-06,
+      "loss": 1.6324,
+      "step": 38520
+    },
+    {
+      "epoch": 1.23296,
+      "grad_norm": 46.039920806884766,
+      "learning_rate": 8.522666666666667e-06,
+      "loss": 1.5892,
+      "step": 38530
+    },
+    {
+      "epoch": 1.23328,
+      "grad_norm": 46.55792236328125,
+      "learning_rate": 8.519111111111112e-06,
+      "loss": 1.6527,
+      "step": 38540
+    },
+    {
+      "epoch": 1.2336,
+      "grad_norm": 45.67931365966797,
+      "learning_rate": 8.515555555555556e-06,
+      "loss": 1.6159,
+      "step": 38550
+    },
+    {
+      "epoch": 1.23392,
+      "grad_norm": 44.48359298706055,
+      "learning_rate": 8.512e-06,
+      "loss": 1.6629,
+      "step": 38560
+    },
+    {
+      "epoch": 1.23424,
+      "grad_norm": 45.672264099121094,
+      "learning_rate": 8.508444444444445e-06,
+      "loss": 1.6139,
+      "step": 38570
+    },
+    {
+      "epoch": 1.23456,
+      "grad_norm": 46.08909606933594,
+      "learning_rate": 8.50488888888889e-06,
+      "loss": 1.6346,
+      "step": 38580
+    },
+    {
+      "epoch": 1.23488,
+      "grad_norm": 43.823631286621094,
+      "learning_rate": 8.501333333333334e-06,
+      "loss": 1.6083,
+      "step": 38590
+    },
+    {
+      "epoch": 1.2352,
+      "grad_norm": 43.065364837646484,
+      "learning_rate": 8.497777777777777e-06,
+      "loss": 1.6003,
+      "step": 38600
+    },
+    {
+      "epoch": 1.23552,
+      "grad_norm": 46.16888427734375,
+      "learning_rate": 8.494222222222222e-06,
+      "loss": 1.6133,
+      "step": 38610
+    },
+    {
+      "epoch": 1.23584,
+      "grad_norm": 42.988792419433594,
+      "learning_rate": 8.490666666666666e-06,
+      "loss": 1.6111,
+      "step": 38620
+    },
+    {
+      "epoch": 1.23616,
+      "grad_norm": 44.04892349243164,
+      "learning_rate": 8.487111111111111e-06,
+      "loss": 1.5899,
+      "step": 38630
+    },
+    {
+      "epoch": 1.23648,
+      "grad_norm": 44.41130065917969,
+      "learning_rate": 8.483555555555556e-06,
+      "loss": 1.6107,
+      "step": 38640
+    },
+    {
+      "epoch": 1.2368000000000001,
+      "grad_norm": 45.48244857788086,
+      "learning_rate": 8.48e-06,
+      "loss": 1.6073,
+      "step": 38650
+    },
+    {
+      "epoch": 1.23712,
+      "grad_norm": 45.22336959838867,
+      "learning_rate": 8.476444444444445e-06,
+      "loss": 1.6161,
+      "step": 38660
+    },
+    {
+      "epoch": 1.23744,
+      "grad_norm": 42.87507247924805,
+      "learning_rate": 8.47288888888889e-06,
+      "loss": 1.6071,
+      "step": 38670
+    },
+    {
+      "epoch": 1.23776,
+      "grad_norm": 45.26953125,
+      "learning_rate": 8.469333333333334e-06,
+      "loss": 1.6351,
+      "step": 38680
+    },
+    {
+      "epoch": 1.23808,
+      "grad_norm": 45.478519439697266,
+      "learning_rate": 8.465777777777778e-06,
+      "loss": 1.6162,
+      "step": 38690
+    },
+    {
+      "epoch": 1.2384,
+      "grad_norm": 41.99089813232422,
+      "learning_rate": 8.462222222222223e-06,
+      "loss": 1.6087,
+      "step": 38700
+    },
+    {
+      "epoch": 1.23872,
+      "grad_norm": 45.846282958984375,
+      "learning_rate": 8.458666666666667e-06,
+      "loss": 1.6067,
+      "step": 38710
+    },
+    {
+      "epoch": 1.23904,
+      "grad_norm": 43.07404327392578,
+      "learning_rate": 8.455111111111112e-06,
+      "loss": 1.6027,
+      "step": 38720
+    },
+    {
+      "epoch": 1.23936,
+      "grad_norm": 44.055213928222656,
+      "learning_rate": 8.451555555555557e-06,
+      "loss": 1.6201,
+      "step": 38730
+    },
+    {
+      "epoch": 1.23968,
+      "grad_norm": 43.944400787353516,
+      "learning_rate": 8.448000000000001e-06,
+      "loss": 1.6016,
+      "step": 38740
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 44.895599365234375,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 1.6301,
+      "step": 38750
+    },
+    {
+      "epoch": 1.24032,
+      "grad_norm": 46.72675323486328,
+      "learning_rate": 8.44088888888889e-06,
+      "loss": 1.6448,
+      "step": 38760
+    },
+    {
+      "epoch": 1.24064,
+      "grad_norm": 46.3919792175293,
+      "learning_rate": 8.437333333333335e-06,
+      "loss": 1.6113,
+      "step": 38770
+    },
+    {
+      "epoch": 1.24096,
+      "grad_norm": 46.99666976928711,
+      "learning_rate": 8.43377777777778e-06,
+      "loss": 1.6063,
+      "step": 38780
+    },
+    {
+      "epoch": 1.24128,
+      "grad_norm": 44.31560516357422,
+      "learning_rate": 8.430222222222222e-06,
+      "loss": 1.6138,
+      "step": 38790
+    },
+    {
+      "epoch": 1.2416,
+      "grad_norm": 44.34001922607422,
+      "learning_rate": 8.426666666666667e-06,
+      "loss": 1.622,
+      "step": 38800
+    },
+    {
+      "epoch": 1.24192,
+      "grad_norm": 45.2856559753418,
+      "learning_rate": 8.423111111111111e-06,
+      "loss": 1.6052,
+      "step": 38810
+    },
+    {
+      "epoch": 1.24224,
+      "grad_norm": 45.01423263549805,
+      "learning_rate": 8.419555555555556e-06,
+      "loss": 1.6357,
+      "step": 38820
+    },
+    {
+      "epoch": 1.24256,
+      "grad_norm": 43.551082611083984,
+      "learning_rate": 8.416e-06,
+      "loss": 1.5987,
+      "step": 38830
+    },
+    {
+      "epoch": 1.24288,
+      "grad_norm": 42.74105453491211,
+      "learning_rate": 8.412444444444445e-06,
+      "loss": 1.5585,
+      "step": 38840
+    },
+    {
+      "epoch": 1.2432,
+      "grad_norm": 44.34646224975586,
+      "learning_rate": 8.40888888888889e-06,
+      "loss": 1.6188,
+      "step": 38850
+    },
+    {
+      "epoch": 1.24352,
+      "grad_norm": 44.549407958984375,
+      "learning_rate": 8.405333333333334e-06,
+      "loss": 1.608,
+      "step": 38860
+    },
+    {
+      "epoch": 1.24384,
+      "grad_norm": 48.80892562866211,
+      "learning_rate": 8.401777777777779e-06,
+      "loss": 1.6355,
+      "step": 38870
+    },
+    {
+      "epoch": 1.24416,
+      "grad_norm": 45.6882209777832,
+      "learning_rate": 8.398222222222223e-06,
+      "loss": 1.6417,
+      "step": 38880
+    },
+    {
+      "epoch": 1.24448,
+      "grad_norm": 43.90581512451172,
+      "learning_rate": 8.394666666666668e-06,
+      "loss": 1.6485,
+      "step": 38890
+    },
+    {
+      "epoch": 1.2448,
+      "grad_norm": 44.31962966918945,
+      "learning_rate": 8.391111111111112e-06,
+      "loss": 1.638,
+      "step": 38900
+    },
+    {
+      "epoch": 1.24512,
+      "grad_norm": 45.7739372253418,
+      "learning_rate": 8.387555555555557e-06,
+      "loss": 1.6122,
+      "step": 38910
+    },
+    {
+      "epoch": 1.2454399999999999,
+      "grad_norm": 45.06602096557617,
+      "learning_rate": 8.384000000000001e-06,
+      "loss": 1.6004,
+      "step": 38920
+    },
+    {
+      "epoch": 1.24576,
+      "grad_norm": 47.67781066894531,
+      "learning_rate": 8.380444444444446e-06,
+      "loss": 1.6465,
+      "step": 38930
+    },
+    {
+      "epoch": 1.24608,
+      "grad_norm": 45.568111419677734,
+      "learning_rate": 8.37688888888889e-06,
+      "loss": 1.5947,
+      "step": 38940
+    },
+    {
+      "epoch": 1.2464,
+      "grad_norm": 45.62173080444336,
+      "learning_rate": 8.373333333333335e-06,
+      "loss": 1.6228,
+      "step": 38950
+    },
+    {
+      "epoch": 1.24672,
+      "grad_norm": 43.738197326660156,
+      "learning_rate": 8.36977777777778e-06,
+      "loss": 1.6312,
+      "step": 38960
+    },
+    {
+      "epoch": 1.24704,
+      "grad_norm": 43.58649444580078,
+      "learning_rate": 8.366222222222224e-06,
+      "loss": 1.6031,
+      "step": 38970
+    },
+    {
+      "epoch": 1.24736,
+      "grad_norm": 43.96977615356445,
+      "learning_rate": 8.362666666666667e-06,
+      "loss": 1.6238,
+      "step": 38980
+    },
+    {
+      "epoch": 1.24768,
+      "grad_norm": 42.86772155761719,
+      "learning_rate": 8.359111111111112e-06,
+      "loss": 1.6035,
+      "step": 38990
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 45.149566650390625,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 1.6048,
+      "step": 39000
+    },
+    {
+      "epoch": 1.24832,
+      "grad_norm": 50.66899871826172,
+      "learning_rate": 8.352e-06,
+      "loss": 1.6034,
+      "step": 39010
+    },
+    {
+      "epoch": 1.24864,
+      "grad_norm": 44.265892028808594,
+      "learning_rate": 8.348444444444445e-06,
+      "loss": 1.6194,
+      "step": 39020
+    },
+    {
+      "epoch": 1.24896,
+      "grad_norm": 43.48944854736328,
+      "learning_rate": 8.34488888888889e-06,
+      "loss": 1.6136,
+      "step": 39030
+    },
+    {
+      "epoch": 1.24928,
+      "grad_norm": 45.849212646484375,
+      "learning_rate": 8.341333333333334e-06,
+      "loss": 1.6193,
+      "step": 39040
+    },
+    {
+      "epoch": 1.2496,
+      "grad_norm": 45.44252395629883,
+      "learning_rate": 8.337777777777777e-06,
+      "loss": 1.6313,
+      "step": 39050
+    },
+    {
+      "epoch": 1.24992,
+      "grad_norm": 43.57227325439453,
+      "learning_rate": 8.334222222222222e-06,
+      "loss": 1.6059,
+      "step": 39060
+    },
+    {
+      "epoch": 1.25024,
+      "grad_norm": 46.95555114746094,
+      "learning_rate": 8.330666666666666e-06,
+      "loss": 1.6226,
+      "step": 39070
+    },
+    {
+      "epoch": 1.2505600000000001,
+      "grad_norm": 44.509010314941406,
+      "learning_rate": 8.327111111111111e-06,
+      "loss": 1.5949,
+      "step": 39080
+    },
+    {
+      "epoch": 1.25088,
+      "grad_norm": 46.533512115478516,
+      "learning_rate": 8.323555555555555e-06,
+      "loss": 1.6491,
+      "step": 39090
+    },
+    {
+      "epoch": 1.2511999999999999,
+      "grad_norm": 45.26840591430664,
+      "learning_rate": 8.32e-06,
+      "loss": 1.609,
+      "step": 39100
+    },
+    {
+      "epoch": 1.25152,
+      "grad_norm": 43.79453659057617,
+      "learning_rate": 8.316444444444445e-06,
+      "loss": 1.6116,
+      "step": 39110
+    },
+    {
+      "epoch": 1.25184,
+      "grad_norm": 45.80735778808594,
+      "learning_rate": 8.312888888888889e-06,
+      "loss": 1.5951,
+      "step": 39120
+    },
+    {
+      "epoch": 1.25216,
+      "grad_norm": 44.02614212036133,
+      "learning_rate": 8.309333333333334e-06,
+      "loss": 1.6093,
+      "step": 39130
+    },
+    {
+      "epoch": 1.25248,
+      "grad_norm": 44.34676742553711,
+      "learning_rate": 8.305777777777778e-06,
+      "loss": 1.5938,
+      "step": 39140
+    },
+    {
+      "epoch": 1.2528000000000001,
+      "grad_norm": 44.790584564208984,
+      "learning_rate": 8.302222222222223e-06,
+      "loss": 1.5792,
+      "step": 39150
+    },
+    {
+      "epoch": 1.25312,
+      "grad_norm": 44.13827133178711,
+      "learning_rate": 8.298666666666667e-06,
+      "loss": 1.606,
+      "step": 39160
+    },
+    {
+      "epoch": 1.2534399999999999,
+      "grad_norm": 44.26396179199219,
+      "learning_rate": 8.295111111111112e-06,
+      "loss": 1.6278,
+      "step": 39170
+    },
+    {
+      "epoch": 1.25376,
+      "grad_norm": 44.31110763549805,
+      "learning_rate": 8.291555555555556e-06,
+      "loss": 1.6014,
+      "step": 39180
+    },
+    {
+      "epoch": 1.25408,
+      "grad_norm": 46.95048522949219,
+      "learning_rate": 8.288000000000001e-06,
+      "loss": 1.6161,
+      "step": 39190
+    },
+    {
+      "epoch": 1.2544,
+      "grad_norm": 44.23671340942383,
+      "learning_rate": 8.284444444444446e-06,
+      "loss": 1.5999,
+      "step": 39200
+    },
+    {
+      "epoch": 1.25472,
+      "grad_norm": 44.4106330871582,
+      "learning_rate": 8.28088888888889e-06,
+      "loss": 1.5989,
+      "step": 39210
+    },
+    {
+      "epoch": 1.25504,
+      "grad_norm": 46.071533203125,
+      "learning_rate": 8.277333333333335e-06,
+      "loss": 1.6218,
+      "step": 39220
+    },
+    {
+      "epoch": 1.25536,
+      "grad_norm": 45.08268737792969,
+      "learning_rate": 8.27377777777778e-06,
+      "loss": 1.6325,
+      "step": 39230
+    },
+    {
+      "epoch": 1.25568,
+      "grad_norm": 44.903846740722656,
+      "learning_rate": 8.270222222222222e-06,
+      "loss": 1.5971,
+      "step": 39240
+    },
+    {
+      "epoch": 1.256,
+      "grad_norm": 43.16526412963867,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.6335,
+      "step": 39250
+    },
+    {
+      "epoch": 1.25632,
+      "grad_norm": 45.51958084106445,
+      "learning_rate": 8.263111111111111e-06,
+      "loss": 1.6138,
+      "step": 39260
+    },
+    {
+      "epoch": 1.25664,
+      "grad_norm": 44.03669738769531,
+      "learning_rate": 8.259555555555556e-06,
+      "loss": 1.5963,
+      "step": 39270
+    },
+    {
+      "epoch": 1.25696,
+      "grad_norm": 43.991878509521484,
+      "learning_rate": 8.256e-06,
+      "loss": 1.5921,
+      "step": 39280
+    },
+    {
+      "epoch": 1.25728,
+      "grad_norm": 45.92048263549805,
+      "learning_rate": 8.252444444444445e-06,
+      "loss": 1.6665,
+      "step": 39290
+    },
+    {
+      "epoch": 1.2576,
+      "grad_norm": 44.205162048339844,
+      "learning_rate": 8.24888888888889e-06,
+      "loss": 1.5928,
+      "step": 39300
+    },
+    {
+      "epoch": 1.25792,
+      "grad_norm": 45.7902946472168,
+      "learning_rate": 8.245333333333334e-06,
+      "loss": 1.5982,
+      "step": 39310
+    },
+    {
+      "epoch": 1.25824,
+      "grad_norm": 45.62728500366211,
+      "learning_rate": 8.241777777777778e-06,
+      "loss": 1.6445,
+      "step": 39320
+    },
+    {
+      "epoch": 1.2585600000000001,
+      "grad_norm": 46.222023010253906,
+      "learning_rate": 8.238222222222223e-06,
+      "loss": 1.5962,
+      "step": 39330
+    },
+    {
+      "epoch": 1.25888,
+      "grad_norm": 45.977638244628906,
+      "learning_rate": 8.234666666666668e-06,
+      "loss": 1.612,
+      "step": 39340
+    },
+    {
+      "epoch": 1.2591999999999999,
+      "grad_norm": 46.27615737915039,
+      "learning_rate": 8.231111111111112e-06,
+      "loss": 1.6264,
+      "step": 39350
+    },
+    {
+      "epoch": 1.25952,
+      "grad_norm": 45.03913497924805,
+      "learning_rate": 8.227555555555557e-06,
+      "loss": 1.6275,
+      "step": 39360
+    },
+    {
+      "epoch": 1.25984,
+      "grad_norm": 44.86140060424805,
+      "learning_rate": 8.224000000000001e-06,
+      "loss": 1.6316,
+      "step": 39370
+    },
+    {
+      "epoch": 1.26016,
+      "grad_norm": 44.86174392700195,
+      "learning_rate": 8.220444444444446e-06,
+      "loss": 1.6095,
+      "step": 39380
+    },
+    {
+      "epoch": 1.26048,
+      "grad_norm": 43.89189910888672,
+      "learning_rate": 8.21688888888889e-06,
+      "loss": 1.6291,
+      "step": 39390
+    },
+    {
+      "epoch": 1.2608,
+      "grad_norm": 45.40138626098633,
+      "learning_rate": 8.213333333333335e-06,
+      "loss": 1.6167,
+      "step": 39400
+    },
+    {
+      "epoch": 1.26112,
+      "grad_norm": 43.08382797241211,
+      "learning_rate": 8.20977777777778e-06,
+      "loss": 1.6063,
+      "step": 39410
+    },
+    {
+      "epoch": 1.26144,
+      "grad_norm": 45.12299728393555,
+      "learning_rate": 8.206222222222224e-06,
+      "loss": 1.5989,
+      "step": 39420
+    },
+    {
+      "epoch": 1.26176,
+      "grad_norm": 42.76207733154297,
+      "learning_rate": 8.202666666666667e-06,
+      "loss": 1.5841,
+      "step": 39430
+    },
+    {
+      "epoch": 1.26208,
+      "grad_norm": 43.17430114746094,
+      "learning_rate": 8.199111111111111e-06,
+      "loss": 1.5885,
+      "step": 39440
+    },
+    {
+      "epoch": 1.2624,
+      "grad_norm": 45.012516021728516,
+      "learning_rate": 8.195555555555556e-06,
+      "loss": 1.5991,
+      "step": 39450
+    },
+    {
+      "epoch": 1.26272,
+      "grad_norm": 44.140464782714844,
+      "learning_rate": 8.192e-06,
+      "loss": 1.5906,
+      "step": 39460
+    },
+    {
+      "epoch": 1.26304,
+      "grad_norm": 44.850975036621094,
+      "learning_rate": 8.188444444444445e-06,
+      "loss": 1.609,
+      "step": 39470
+    },
+    {
+      "epoch": 1.26336,
+      "grad_norm": 46.2093620300293,
+      "learning_rate": 8.18488888888889e-06,
+      "loss": 1.6166,
+      "step": 39480
+    },
+    {
+      "epoch": 1.26368,
+      "grad_norm": 44.715633392333984,
+      "learning_rate": 8.181333333333334e-06,
+      "loss": 1.5741,
+      "step": 39490
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 43.745506286621094,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 1.607,
+      "step": 39500
+    },
+    {
+      "epoch": 1.26432,
+      "grad_norm": 44.87568283081055,
+      "learning_rate": 8.174222222222223e-06,
+      "loss": 1.623,
+      "step": 39510
+    },
+    {
+      "epoch": 1.26464,
+      "grad_norm": 44.68089294433594,
+      "learning_rate": 8.170666666666668e-06,
+      "loss": 1.5934,
+      "step": 39520
+    },
+    {
+      "epoch": 1.2649599999999999,
+      "grad_norm": 42.680606842041016,
+      "learning_rate": 8.167111111111112e-06,
+      "loss": 1.602,
+      "step": 39530
+    },
+    {
+      "epoch": 1.26528,
+      "grad_norm": 48.34457015991211,
+      "learning_rate": 8.163555555555555e-06,
+      "loss": 1.6289,
+      "step": 39540
+    },
+    {
+      "epoch": 1.2656,
+      "grad_norm": 43.446624755859375,
+      "learning_rate": 8.16e-06,
+      "loss": 1.6195,
+      "step": 39550
+    },
+    {
+      "epoch": 1.26592,
+      "grad_norm": 46.034645080566406,
+      "learning_rate": 8.156444444444444e-06,
+      "loss": 1.6067,
+      "step": 39560
+    },
+    {
+      "epoch": 1.26624,
+      "grad_norm": 41.680397033691406,
+      "learning_rate": 8.152888888888889e-06,
+      "loss": 1.6123,
+      "step": 39570
+    },
+    {
+      "epoch": 1.2665600000000001,
+      "grad_norm": 46.30161666870117,
+      "learning_rate": 8.149333333333333e-06,
+      "loss": 1.6132,
+      "step": 39580
+    },
+    {
+      "epoch": 1.26688,
+      "grad_norm": 44.9428596496582,
+      "learning_rate": 8.145777777777778e-06,
+      "loss": 1.6079,
+      "step": 39590
+    },
+    {
+      "epoch": 1.2671999999999999,
+      "grad_norm": 45.387638092041016,
+      "learning_rate": 8.142222222222223e-06,
+      "loss": 1.5962,
+      "step": 39600
+    },
+    {
+      "epoch": 1.26752,
+      "grad_norm": 44.43503189086914,
+      "learning_rate": 8.138666666666667e-06,
+      "loss": 1.6002,
+      "step": 39610
+    },
+    {
+      "epoch": 1.26784,
+      "grad_norm": 46.8868522644043,
+      "learning_rate": 8.135111111111112e-06,
+      "loss": 1.6457,
+      "step": 39620
+    },
+    {
+      "epoch": 1.26816,
+      "grad_norm": 44.164154052734375,
+      "learning_rate": 8.131555555555556e-06,
+      "loss": 1.6158,
+      "step": 39630
+    },
+    {
+      "epoch": 1.26848,
+      "grad_norm": 45.72709274291992,
+      "learning_rate": 8.128e-06,
+      "loss": 1.6293,
+      "step": 39640
+    },
+    {
+      "epoch": 1.2688,
+      "grad_norm": 44.85696029663086,
+      "learning_rate": 8.124444444444445e-06,
+      "loss": 1.6238,
+      "step": 39650
+    },
+    {
+      "epoch": 1.26912,
+      "grad_norm": 43.19333267211914,
+      "learning_rate": 8.12088888888889e-06,
+      "loss": 1.5928,
+      "step": 39660
+    },
+    {
+      "epoch": 1.26944,
+      "grad_norm": 46.190467834472656,
+      "learning_rate": 8.117333333333334e-06,
+      "loss": 1.6359,
+      "step": 39670
+    },
+    {
+      "epoch": 1.26976,
+      "grad_norm": 44.76688003540039,
+      "learning_rate": 8.113777777777779e-06,
+      "loss": 1.6088,
+      "step": 39680
+    },
+    {
+      "epoch": 1.27008,
+      "grad_norm": 43.66759490966797,
+      "learning_rate": 8.110222222222222e-06,
+      "loss": 1.6151,
+      "step": 39690
+    },
+    {
+      "epoch": 1.2704,
+      "grad_norm": 44.64482879638672,
+      "learning_rate": 8.106666666666666e-06,
+      "loss": 1.6206,
+      "step": 39700
+    },
+    {
+      "epoch": 1.27072,
+      "grad_norm": 44.56056594848633,
+      "learning_rate": 8.103111111111111e-06,
+      "loss": 1.6027,
+      "step": 39710
+    },
+    {
+      "epoch": 1.27104,
+      "grad_norm": 45.265018463134766,
+      "learning_rate": 8.099555555555556e-06,
+      "loss": 1.5946,
+      "step": 39720
+    },
+    {
+      "epoch": 1.27136,
+      "grad_norm": 45.138187408447266,
+      "learning_rate": 8.096e-06,
+      "loss": 1.6091,
+      "step": 39730
+    },
+    {
+      "epoch": 1.27168,
+      "grad_norm": 45.547271728515625,
+      "learning_rate": 8.092444444444445e-06,
+      "loss": 1.581,
+      "step": 39740
+    },
+    {
+      "epoch": 1.272,
+      "grad_norm": 43.008277893066406,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 1.6314,
+      "step": 39750
+    },
+    {
+      "epoch": 1.2723200000000001,
+      "grad_norm": 45.12823486328125,
+      "learning_rate": 8.085333333333334e-06,
+      "loss": 1.5787,
+      "step": 39760
+    },
+    {
+      "epoch": 1.27264,
+      "grad_norm": 46.13259506225586,
+      "learning_rate": 8.081777777777778e-06,
+      "loss": 1.5724,
+      "step": 39770
+    },
+    {
+      "epoch": 1.2729599999999999,
+      "grad_norm": 44.257450103759766,
+      "learning_rate": 8.078222222222223e-06,
+      "loss": 1.6312,
+      "step": 39780
+    },
+    {
+      "epoch": 1.27328,
+      "grad_norm": 44.41106414794922,
+      "learning_rate": 8.074666666666667e-06,
+      "loss": 1.6285,
+      "step": 39790
+    },
+    {
+      "epoch": 1.2736,
+      "grad_norm": 43.90785217285156,
+      "learning_rate": 8.071111111111112e-06,
+      "loss": 1.5971,
+      "step": 39800
+    },
+    {
+      "epoch": 1.27392,
+      "grad_norm": 43.35899353027344,
+      "learning_rate": 8.067555555555557e-06,
+      "loss": 1.6186,
+      "step": 39810
+    },
+    {
+      "epoch": 1.27424,
+      "grad_norm": 47.12132263183594,
+      "learning_rate": 8.064000000000001e-06,
+      "loss": 1.6181,
+      "step": 39820
+    },
+    {
+      "epoch": 1.2745600000000001,
+      "grad_norm": 43.785736083984375,
+      "learning_rate": 8.060444444444446e-06,
+      "loss": 1.5915,
+      "step": 39830
+    },
+    {
+      "epoch": 1.27488,
+      "grad_norm": 44.73807907104492,
+      "learning_rate": 8.05688888888889e-06,
+      "loss": 1.6149,
+      "step": 39840
+    },
+    {
+      "epoch": 1.2752,
+      "grad_norm": 44.06450271606445,
+      "learning_rate": 8.053333333333335e-06,
+      "loss": 1.6082,
+      "step": 39850
+    },
+    {
+      "epoch": 1.27552,
+      "grad_norm": 44.62391662597656,
+      "learning_rate": 8.04977777777778e-06,
+      "loss": 1.6064,
+      "step": 39860
+    },
+    {
+      "epoch": 1.27584,
+      "grad_norm": 48.22390365600586,
+      "learning_rate": 8.046222222222224e-06,
+      "loss": 1.6159,
+      "step": 39870
+    },
+    {
+      "epoch": 1.27616,
+      "grad_norm": 45.42574691772461,
+      "learning_rate": 8.042666666666667e-06,
+      "loss": 1.625,
+      "step": 39880
+    },
+    {
+      "epoch": 1.27648,
+      "grad_norm": 45.23716735839844,
+      "learning_rate": 8.039111111111111e-06,
+      "loss": 1.6274,
+      "step": 39890
+    },
+    {
+      "epoch": 1.2768,
+      "grad_norm": 44.92580795288086,
+      "learning_rate": 8.035555555555556e-06,
+      "loss": 1.596,
+      "step": 39900
+    },
+    {
+      "epoch": 1.27712,
+      "grad_norm": 45.10259246826172,
+      "learning_rate": 8.032e-06,
+      "loss": 1.5999,
+      "step": 39910
+    },
+    {
+      "epoch": 1.27744,
+      "grad_norm": 44.05598068237305,
+      "learning_rate": 8.028444444444445e-06,
+      "loss": 1.616,
+      "step": 39920
+    },
+    {
+      "epoch": 1.27776,
+      "grad_norm": 43.27324676513672,
+      "learning_rate": 8.02488888888889e-06,
+      "loss": 1.613,
+      "step": 39930
+    },
+    {
+      "epoch": 1.27808,
+      "grad_norm": 45.73674392700195,
+      "learning_rate": 8.021333333333334e-06,
+      "loss": 1.612,
+      "step": 39940
+    },
+    {
+      "epoch": 1.2784,
+      "grad_norm": 44.60195541381836,
+      "learning_rate": 8.017777777777779e-06,
+      "loss": 1.6288,
+      "step": 39950
+    },
+    {
+      "epoch": 1.27872,
+      "grad_norm": 45.1451416015625,
+      "learning_rate": 8.014222222222223e-06,
+      "loss": 1.577,
+      "step": 39960
+    },
+    {
+      "epoch": 1.27904,
+      "grad_norm": 44.034568786621094,
+      "learning_rate": 8.010666666666668e-06,
+      "loss": 1.632,
+      "step": 39970
+    },
+    {
+      "epoch": 1.27936,
+      "grad_norm": 46.17295455932617,
+      "learning_rate": 8.007111111111112e-06,
+      "loss": 1.6267,
+      "step": 39980
+    },
+    {
+      "epoch": 1.27968,
+      "grad_norm": 45.9102668762207,
+      "learning_rate": 8.003555555555557e-06,
+      "loss": 1.6077,
+      "step": 39990
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 43.993289947509766,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.5885,
+      "step": 40000
+    },
+    {
+      "epoch": 1.2803200000000001,
+      "grad_norm": 45.628448486328125,
+      "learning_rate": 7.996444444444446e-06,
+      "loss": 1.6315,
+      "step": 40010
+    },
+    {
+      "epoch": 1.28064,
+      "grad_norm": 44.6242561340332,
+      "learning_rate": 7.99288888888889e-06,
+      "loss": 1.6086,
+      "step": 40020
+    },
+    {
+      "epoch": 1.2809599999999999,
+      "grad_norm": 47.21349334716797,
+      "learning_rate": 7.989333333333335e-06,
+      "loss": 1.6256,
+      "step": 40030
+    },
+    {
+      "epoch": 1.28128,
+      "grad_norm": 46.001495361328125,
+      "learning_rate": 7.98577777777778e-06,
+      "loss": 1.6245,
+      "step": 40040
+    },
+    {
+      "epoch": 1.2816,
+      "grad_norm": 43.2928466796875,
+      "learning_rate": 7.982222222222224e-06,
+      "loss": 1.6257,
+      "step": 40050
+    },
+    {
+      "epoch": 1.28192,
+      "grad_norm": 44.0401496887207,
+      "learning_rate": 7.978666666666667e-06,
+      "loss": 1.6124,
+      "step": 40060
+    },
+    {
+      "epoch": 1.28224,
+      "grad_norm": 45.14518737792969,
+      "learning_rate": 7.975111111111112e-06,
+      "loss": 1.6006,
+      "step": 40070
+    },
+    {
+      "epoch": 1.28256,
+      "grad_norm": 43.09754180908203,
+      "learning_rate": 7.971555555555556e-06,
+      "loss": 1.5802,
+      "step": 40080
+    },
+    {
+      "epoch": 1.28288,
+      "grad_norm": 45.571258544921875,
+      "learning_rate": 7.968e-06,
+      "loss": 1.6197,
+      "step": 40090
+    },
+    {
+      "epoch": 1.2832,
+      "grad_norm": 45.45378112792969,
+      "learning_rate": 7.964444444444445e-06,
+      "loss": 1.623,
+      "step": 40100
+    },
+    {
+      "epoch": 1.28352,
+      "grad_norm": 44.5523796081543,
+      "learning_rate": 7.96088888888889e-06,
+      "loss": 1.6224,
+      "step": 40110
+    },
+    {
+      "epoch": 1.28384,
+      "grad_norm": 44.05735397338867,
+      "learning_rate": 7.957333333333334e-06,
+      "loss": 1.6143,
+      "step": 40120
+    },
+    {
+      "epoch": 1.28416,
+      "grad_norm": 44.72103500366211,
+      "learning_rate": 7.953777777777779e-06,
+      "loss": 1.6412,
+      "step": 40130
+    },
+    {
+      "epoch": 1.28448,
+      "grad_norm": 47.94450378417969,
+      "learning_rate": 7.950222222222222e-06,
+      "loss": 1.6276,
+      "step": 40140
+    },
+    {
+      "epoch": 1.2848,
+      "grad_norm": 46.01536560058594,
+      "learning_rate": 7.946666666666666e-06,
+      "loss": 1.6162,
+      "step": 40150
+    },
+    {
+      "epoch": 1.28512,
+      "grad_norm": 44.166263580322266,
+      "learning_rate": 7.94311111111111e-06,
+      "loss": 1.5908,
+      "step": 40160
+    },
+    {
+      "epoch": 1.28544,
+      "grad_norm": 45.8519287109375,
+      "learning_rate": 7.939555555555555e-06,
+      "loss": 1.6292,
+      "step": 40170
+    },
+    {
+      "epoch": 1.28576,
+      "grad_norm": 45.5851936340332,
+      "learning_rate": 7.936e-06,
+      "loss": 1.6074,
+      "step": 40180
+    },
+    {
+      "epoch": 1.2860800000000001,
+      "grad_norm": 45.92979431152344,
+      "learning_rate": 7.932444444444444e-06,
+      "loss": 1.5674,
+      "step": 40190
+    },
+    {
+      "epoch": 1.2864,
+      "grad_norm": 43.50894546508789,
+      "learning_rate": 7.928888888888889e-06,
+      "loss": 1.5989,
+      "step": 40200
+    },
+    {
+      "epoch": 1.2867199999999999,
+      "grad_norm": 46.01730728149414,
+      "learning_rate": 7.925333333333334e-06,
+      "loss": 1.6008,
+      "step": 40210
+    },
+    {
+      "epoch": 1.28704,
+      "grad_norm": 45.77237319946289,
+      "learning_rate": 7.921777777777778e-06,
+      "loss": 1.6168,
+      "step": 40220
+    },
+    {
+      "epoch": 1.28736,
+      "grad_norm": 44.72514343261719,
+      "learning_rate": 7.918222222222223e-06,
+      "loss": 1.6021,
+      "step": 40230
+    },
+    {
+      "epoch": 1.28768,
+      "grad_norm": 44.8151969909668,
+      "learning_rate": 7.914666666666667e-06,
+      "loss": 1.5952,
+      "step": 40240
+    },
+    {
+      "epoch": 1.288,
+      "grad_norm": 44.08286666870117,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 1.6163,
+      "step": 40250
+    },
+    {
+      "epoch": 1.2883200000000001,
+      "grad_norm": 44.34801483154297,
+      "learning_rate": 7.907555555555556e-06,
+      "loss": 1.6181,
+      "step": 40260
+    },
+    {
+      "epoch": 1.28864,
+      "grad_norm": 43.255027770996094,
+      "learning_rate": 7.904000000000001e-06,
+      "loss": 1.6056,
+      "step": 40270
+    },
+    {
+      "epoch": 1.2889599999999999,
+      "grad_norm": 45.57148361206055,
+      "learning_rate": 7.900444444444445e-06,
+      "loss": 1.6077,
+      "step": 40280
+    },
+    {
+      "epoch": 1.28928,
+      "grad_norm": 46.56278991699219,
+      "learning_rate": 7.89688888888889e-06,
+      "loss": 1.6375,
+      "step": 40290
+    },
+    {
+      "epoch": 1.2896,
+      "grad_norm": 46.41355514526367,
+      "learning_rate": 7.893333333333335e-06,
+      "loss": 1.6111,
+      "step": 40300
+    },
+    {
+      "epoch": 1.28992,
+      "grad_norm": 43.413761138916016,
+      "learning_rate": 7.889777777777779e-06,
+      "loss": 1.6313,
+      "step": 40310
+    },
+    {
+      "epoch": 1.29024,
+      "grad_norm": 43.898345947265625,
+      "learning_rate": 7.886222222222224e-06,
+      "loss": 1.6062,
+      "step": 40320
+    },
+    {
+      "epoch": 1.29056,
+      "grad_norm": 43.248104095458984,
+      "learning_rate": 7.882666666666667e-06,
+      "loss": 1.61,
+      "step": 40330
+    },
+    {
+      "epoch": 1.29088,
+      "grad_norm": 42.96113967895508,
+      "learning_rate": 7.879111111111111e-06,
+      "loss": 1.5724,
+      "step": 40340
+    },
+    {
+      "epoch": 1.2912,
+      "grad_norm": 43.738590240478516,
+      "learning_rate": 7.875555555555556e-06,
+      "loss": 1.6262,
+      "step": 40350
+    },
+    {
+      "epoch": 1.29152,
+      "grad_norm": 45.08527755737305,
+      "learning_rate": 7.872e-06,
+      "loss": 1.6212,
+      "step": 40360
+    },
+    {
+      "epoch": 1.29184,
+      "grad_norm": 44.45360565185547,
+      "learning_rate": 7.868444444444445e-06,
+      "loss": 1.625,
+      "step": 40370
+    },
+    {
+      "epoch": 1.29216,
+      "grad_norm": 47.08750534057617,
+      "learning_rate": 7.86488888888889e-06,
+      "loss": 1.6254,
+      "step": 40380
+    },
+    {
+      "epoch": 1.29248,
+      "grad_norm": 42.82807540893555,
+      "learning_rate": 7.861333333333334e-06,
+      "loss": 1.6114,
+      "step": 40390
+    },
+    {
+      "epoch": 1.2928,
+      "grad_norm": 43.76680374145508,
+      "learning_rate": 7.857777777777778e-06,
+      "loss": 1.6219,
+      "step": 40400
+    },
+    {
+      "epoch": 1.29312,
+      "grad_norm": 43.96428680419922,
+      "learning_rate": 7.854222222222223e-06,
+      "loss": 1.5789,
+      "step": 40410
+    },
+    {
+      "epoch": 1.29344,
+      "grad_norm": 46.33516311645508,
+      "learning_rate": 7.850666666666668e-06,
+      "loss": 1.6083,
+      "step": 40420
+    },
+    {
+      "epoch": 1.29376,
+      "grad_norm": 44.434627532958984,
+      "learning_rate": 7.847111111111112e-06,
+      "loss": 1.6226,
+      "step": 40430
+    },
+    {
+      "epoch": 1.2940800000000001,
+      "grad_norm": 43.88167953491211,
+      "learning_rate": 7.843555555555557e-06,
+      "loss": 1.5743,
+      "step": 40440
+    },
+    {
+      "epoch": 1.2944,
+      "grad_norm": 46.007728576660156,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 1.6166,
+      "step": 40450
+    },
+    {
+      "epoch": 1.2947199999999999,
+      "grad_norm": 44.59865951538086,
+      "learning_rate": 7.836444444444446e-06,
+      "loss": 1.6035,
+      "step": 40460
+    },
+    {
+      "epoch": 1.29504,
+      "grad_norm": 44.90102005004883,
+      "learning_rate": 7.83288888888889e-06,
+      "loss": 1.6403,
+      "step": 40470
+    },
+    {
+      "epoch": 1.29536,
+      "grad_norm": 46.19745635986328,
+      "learning_rate": 7.829333333333335e-06,
+      "loss": 1.6255,
+      "step": 40480
+    },
+    {
+      "epoch": 1.29568,
+      "grad_norm": 43.79157638549805,
+      "learning_rate": 7.82577777777778e-06,
+      "loss": 1.6345,
+      "step": 40490
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 44.32958221435547,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 1.5844,
+      "step": 40500
+    },
+    {
+      "epoch": 1.29632,
+      "grad_norm": 46.034942626953125,
+      "learning_rate": 7.818666666666668e-06,
+      "loss": 1.6114,
+      "step": 40510
+    },
+    {
+      "epoch": 1.29664,
+      "grad_norm": 46.09269714355469,
+      "learning_rate": 7.815111111111113e-06,
+      "loss": 1.6044,
+      "step": 40520
+    },
+    {
+      "epoch": 1.29696,
+      "grad_norm": 45.11857986450195,
+      "learning_rate": 7.811555555555556e-06,
+      "loss": 1.6074,
+      "step": 40530
+    },
+    {
+      "epoch": 1.29728,
+      "grad_norm": 42.98341369628906,
+      "learning_rate": 7.808e-06,
+      "loss": 1.597,
+      "step": 40540
+    },
+    {
+      "epoch": 1.2976,
+      "grad_norm": 45.964637756347656,
+      "learning_rate": 7.804444444444445e-06,
+      "loss": 1.6156,
+      "step": 40550
+    },
+    {
+      "epoch": 1.29792,
+      "grad_norm": 43.90168762207031,
+      "learning_rate": 7.80088888888889e-06,
+      "loss": 1.5891,
+      "step": 40560
+    },
+    {
+      "epoch": 1.29824,
+      "grad_norm": 45.83740997314453,
+      "learning_rate": 7.797333333333334e-06,
+      "loss": 1.592,
+      "step": 40570
+    },
+    {
+      "epoch": 1.29856,
+      "grad_norm": 44.05854797363281,
+      "learning_rate": 7.793777777777779e-06,
+      "loss": 1.5761,
+      "step": 40580
+    },
+    {
+      "epoch": 1.29888,
+      "grad_norm": 47.54555130004883,
+      "learning_rate": 7.790222222222222e-06,
+      "loss": 1.6345,
+      "step": 40590
+    },
+    {
+      "epoch": 1.2992,
+      "grad_norm": 44.667266845703125,
+      "learning_rate": 7.786666666666666e-06,
+      "loss": 1.5964,
+      "step": 40600
+    },
+    {
+      "epoch": 1.29952,
+      "grad_norm": 46.65251922607422,
+      "learning_rate": 7.78311111111111e-06,
+      "loss": 1.6193,
+      "step": 40610
+    },
+    {
+      "epoch": 1.29984,
+      "grad_norm": 43.384761810302734,
+      "learning_rate": 7.779555555555555e-06,
+      "loss": 1.6043,
+      "step": 40620
+    },
+    {
+      "epoch": 1.30016,
+      "grad_norm": 44.91738510131836,
+      "learning_rate": 7.776e-06,
+      "loss": 1.5873,
+      "step": 40630
+    },
+    {
+      "epoch": 1.30048,
+      "grad_norm": 45.923622131347656,
+      "learning_rate": 7.772444444444444e-06,
+      "loss": 1.6069,
+      "step": 40640
+    },
+    {
+      "epoch": 1.3008,
+      "grad_norm": 45.634765625,
+      "learning_rate": 7.768888888888889e-06,
+      "loss": 1.6204,
+      "step": 40650
+    },
+    {
+      "epoch": 1.30112,
+      "grad_norm": 46.069881439208984,
+      "learning_rate": 7.765333333333333e-06,
+      "loss": 1.6164,
+      "step": 40660
+    },
+    {
+      "epoch": 1.30144,
+      "grad_norm": 43.63170623779297,
+      "learning_rate": 7.761777777777778e-06,
+      "loss": 1.6255,
+      "step": 40670
+    },
+    {
+      "epoch": 1.30176,
+      "grad_norm": 43.349246978759766,
+      "learning_rate": 7.758222222222223e-06,
+      "loss": 1.6029,
+      "step": 40680
+    },
+    {
+      "epoch": 1.3020800000000001,
+      "grad_norm": 45.24386215209961,
+      "learning_rate": 7.754666666666667e-06,
+      "loss": 1.6056,
+      "step": 40690
+    },
+    {
+      "epoch": 1.3024,
+      "grad_norm": 43.81452560424805,
+      "learning_rate": 7.751111111111112e-06,
+      "loss": 1.6055,
+      "step": 40700
+    },
+    {
+      "epoch": 1.3027199999999999,
+      "grad_norm": 46.62950134277344,
+      "learning_rate": 7.747555555555556e-06,
+      "loss": 1.624,
+      "step": 40710
+    },
+    {
+      "epoch": 1.30304,
+      "grad_norm": 46.991554260253906,
+      "learning_rate": 7.744e-06,
+      "loss": 1.642,
+      "step": 40720
+    },
+    {
+      "epoch": 1.30336,
+      "grad_norm": 46.24224853515625,
+      "learning_rate": 7.740444444444445e-06,
+      "loss": 1.6157,
+      "step": 40730
+    },
+    {
+      "epoch": 1.30368,
+      "grad_norm": 44.421443939208984,
+      "learning_rate": 7.73688888888889e-06,
+      "loss": 1.603,
+      "step": 40740
+    },
+    {
+      "epoch": 1.304,
+      "grad_norm": 45.14869689941406,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.6015,
+      "step": 40750
+    },
+    {
+      "epoch": 1.30432,
+      "grad_norm": 43.42634582519531,
+      "learning_rate": 7.729777777777779e-06,
+      "loss": 1.6017,
+      "step": 40760
+    },
+    {
+      "epoch": 1.30464,
+      "grad_norm": 45.673404693603516,
+      "learning_rate": 7.726222222222224e-06,
+      "loss": 1.6036,
+      "step": 40770
+    },
+    {
+      "epoch": 1.30496,
+      "grad_norm": 45.315757751464844,
+      "learning_rate": 7.722666666666666e-06,
+      "loss": 1.6234,
+      "step": 40780
+    },
+    {
+      "epoch": 1.30528,
+      "grad_norm": 45.67897415161133,
+      "learning_rate": 7.719111111111111e-06,
+      "loss": 1.5891,
+      "step": 40790
+    },
+    {
+      "epoch": 1.3056,
+      "grad_norm": 45.099693298339844,
+      "learning_rate": 7.715555555555555e-06,
+      "loss": 1.6294,
+      "step": 40800
+    },
+    {
+      "epoch": 1.30592,
+      "grad_norm": 44.96028518676758,
+      "learning_rate": 7.712e-06,
+      "loss": 1.626,
+      "step": 40810
+    },
+    {
+      "epoch": 1.30624,
+      "grad_norm": 45.61831283569336,
+      "learning_rate": 7.708444444444445e-06,
+      "loss": 1.6303,
+      "step": 40820
+    },
+    {
+      "epoch": 1.30656,
+      "grad_norm": 44.800445556640625,
+      "learning_rate": 7.704888888888889e-06,
+      "loss": 1.5906,
+      "step": 40830
+    },
+    {
+      "epoch": 1.30688,
+      "grad_norm": 45.365909576416016,
+      "learning_rate": 7.701333333333334e-06,
+      "loss": 1.6078,
+      "step": 40840
+    },
+    {
+      "epoch": 1.3072,
+      "grad_norm": 47.193538665771484,
+      "learning_rate": 7.697777777777778e-06,
+      "loss": 1.6195,
+      "step": 40850
+    },
+    {
+      "epoch": 1.30752,
+      "grad_norm": 44.04536437988281,
+      "learning_rate": 7.694222222222223e-06,
+      "loss": 1.6339,
+      "step": 40860
+    },
+    {
+      "epoch": 1.3078400000000001,
+      "grad_norm": 44.930973052978516,
+      "learning_rate": 7.690666666666667e-06,
+      "loss": 1.6223,
+      "step": 40870
+    },
+    {
+      "epoch": 1.30816,
+      "grad_norm": 46.143306732177734,
+      "learning_rate": 7.687111111111112e-06,
+      "loss": 1.5855,
+      "step": 40880
+    },
+    {
+      "epoch": 1.3084799999999999,
+      "grad_norm": 44.251708984375,
+      "learning_rate": 7.683555555555556e-06,
+      "loss": 1.5935,
+      "step": 40890
+    },
+    {
+      "epoch": 1.3088,
+      "grad_norm": 44.97640609741211,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 1.6104,
+      "step": 40900
+    },
+    {
+      "epoch": 1.30912,
+      "grad_norm": 46.09316635131836,
+      "learning_rate": 7.676444444444446e-06,
+      "loss": 1.6225,
+      "step": 40910
+    },
+    {
+      "epoch": 1.30944,
+      "grad_norm": 44.81732940673828,
+      "learning_rate": 7.67288888888889e-06,
+      "loss": 1.6374,
+      "step": 40920
+    },
+    {
+      "epoch": 1.30976,
+      "grad_norm": 43.48009490966797,
+      "learning_rate": 7.669333333333335e-06,
+      "loss": 1.5855,
+      "step": 40930
+    },
+    {
+      "epoch": 1.3100800000000001,
+      "grad_norm": 42.132633209228516,
+      "learning_rate": 7.66577777777778e-06,
+      "loss": 1.6129,
+      "step": 40940
+    },
+    {
+      "epoch": 1.3104,
+      "grad_norm": 43.00667190551758,
+      "learning_rate": 7.662222222222224e-06,
+      "loss": 1.5724,
+      "step": 40950
+    },
+    {
+      "epoch": 1.3107199999999999,
+      "grad_norm": 47.47547149658203,
+      "learning_rate": 7.658666666666668e-06,
+      "loss": 1.6237,
+      "step": 40960
+    },
+    {
+      "epoch": 1.31104,
+      "grad_norm": 44.40707778930664,
+      "learning_rate": 7.655111111111113e-06,
+      "loss": 1.6235,
+      "step": 40970
+    },
+    {
+      "epoch": 1.31136,
+      "grad_norm": 45.595489501953125,
+      "learning_rate": 7.651555555555556e-06,
+      "loss": 1.5843,
+      "step": 40980
+    },
+    {
+      "epoch": 1.31168,
+      "grad_norm": 45.07762908935547,
+      "learning_rate": 7.648e-06,
+      "loss": 1.6275,
+      "step": 40990
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 44.50121307373047,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 1.6012,
+      "step": 41000
+    },
+    {
+      "epoch": 1.31232,
+      "grad_norm": 44.96818923950195,
+      "learning_rate": 7.64088888888889e-06,
+      "loss": 1.5821,
+      "step": 41010
+    },
+    {
+      "epoch": 1.31264,
+      "grad_norm": 44.83430862426758,
+      "learning_rate": 7.637333333333334e-06,
+      "loss": 1.6233,
+      "step": 41020
+    },
+    {
+      "epoch": 1.31296,
+      "grad_norm": 45.33064270019531,
+      "learning_rate": 7.633777777777779e-06,
+      "loss": 1.582,
+      "step": 41030
+    },
+    {
+      "epoch": 1.31328,
+      "grad_norm": 44.263946533203125,
+      "learning_rate": 7.630222222222223e-06,
+      "loss": 1.5741,
+      "step": 41040
+    },
+    {
+      "epoch": 1.3136,
+      "grad_norm": 44.7042350769043,
+      "learning_rate": 7.626666666666668e-06,
+      "loss": 1.6065,
+      "step": 41050
+    },
+    {
+      "epoch": 1.31392,
+      "grad_norm": 45.056270599365234,
+      "learning_rate": 7.623111111111112e-06,
+      "loss": 1.6121,
+      "step": 41060
+    },
+    {
+      "epoch": 1.31424,
+      "grad_norm": 46.58217239379883,
+      "learning_rate": 7.619555555555557e-06,
+      "loss": 1.6193,
+      "step": 41070
+    },
+    {
+      "epoch": 1.31456,
+      "grad_norm": 44.109134674072266,
+      "learning_rate": 7.616000000000001e-06,
+      "loss": 1.5887,
+      "step": 41080
+    },
+    {
+      "epoch": 1.31488,
+      "grad_norm": 45.42112350463867,
+      "learning_rate": 7.612444444444444e-06,
+      "loss": 1.616,
+      "step": 41090
+    },
+    {
+      "epoch": 1.3152,
+      "grad_norm": 45.975318908691406,
+      "learning_rate": 7.608888888888889e-06,
+      "loss": 1.5935,
+      "step": 41100
+    },
+    {
+      "epoch": 1.31552,
+      "grad_norm": 45.71128845214844,
+      "learning_rate": 7.605333333333333e-06,
+      "loss": 1.6239,
+      "step": 41110
+    },
+    {
+      "epoch": 1.3158400000000001,
+      "grad_norm": 45.99323272705078,
+      "learning_rate": 7.601777777777778e-06,
+      "loss": 1.6085,
+      "step": 41120
+    },
+    {
+      "epoch": 1.31616,
+      "grad_norm": 44.56714630126953,
+      "learning_rate": 7.598222222222222e-06,
+      "loss": 1.5948,
+      "step": 41130
+    },
+    {
+      "epoch": 1.3164799999999999,
+      "grad_norm": 44.28768539428711,
+      "learning_rate": 7.594666666666667e-06,
+      "loss": 1.6283,
+      "step": 41140
+    },
+    {
+      "epoch": 1.3168,
+      "grad_norm": 46.32910919189453,
+      "learning_rate": 7.5911111111111115e-06,
+      "loss": 1.6269,
+      "step": 41150
+    },
+    {
+      "epoch": 1.31712,
+      "grad_norm": 44.03658676147461,
+      "learning_rate": 7.587555555555556e-06,
+      "loss": 1.5864,
+      "step": 41160
+    },
+    {
+      "epoch": 1.31744,
+      "grad_norm": 46.9725227355957,
+      "learning_rate": 7.5840000000000006e-06,
+      "loss": 1.5915,
+      "step": 41170
+    },
+    {
+      "epoch": 1.31776,
+      "grad_norm": 44.8245849609375,
+      "learning_rate": 7.580444444444445e-06,
+      "loss": 1.6049,
+      "step": 41180
+    },
+    {
+      "epoch": 1.31808,
+      "grad_norm": 44.639652252197266,
+      "learning_rate": 7.576888888888889e-06,
+      "loss": 1.6207,
+      "step": 41190
+    },
+    {
+      "epoch": 1.3184,
+      "grad_norm": 44.07933044433594,
+      "learning_rate": 7.573333333333333e-06,
+      "loss": 1.6292,
+      "step": 41200
+    },
+    {
+      "epoch": 1.31872,
+      "grad_norm": 46.18787384033203,
+      "learning_rate": 7.569777777777778e-06,
+      "loss": 1.6415,
+      "step": 41210
+    },
+    {
+      "epoch": 1.31904,
+      "grad_norm": 47.621158599853516,
+      "learning_rate": 7.5662222222222225e-06,
+      "loss": 1.616,
+      "step": 41220
+    },
+    {
+      "epoch": 1.31936,
+      "grad_norm": 46.72417449951172,
+      "learning_rate": 7.562666666666667e-06,
+      "loss": 1.6256,
+      "step": 41230
+    },
+    {
+      "epoch": 1.31968,
+      "grad_norm": 45.38915252685547,
+      "learning_rate": 7.559111111111112e-06,
+      "loss": 1.6002,
+      "step": 41240
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 45.14508819580078,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 1.6152,
+      "step": 41250
+    },
+    {
+      "epoch": 1.32032,
+      "grad_norm": 44.36618423461914,
+      "learning_rate": 7.552000000000001e-06,
+      "loss": 1.6239,
+      "step": 41260
+    },
+    {
+      "epoch": 1.32064,
+      "grad_norm": 46.560482025146484,
+      "learning_rate": 7.548444444444445e-06,
+      "loss": 1.6267,
+      "step": 41270
+    },
+    {
+      "epoch": 1.32096,
+      "grad_norm": 44.61228561401367,
+      "learning_rate": 7.544888888888889e-06,
+      "loss": 1.6021,
+      "step": 41280
+    },
+    {
+      "epoch": 1.32128,
+      "grad_norm": 45.6788444519043,
+      "learning_rate": 7.5413333333333335e-06,
+      "loss": 1.5753,
+      "step": 41290
+    },
+    {
+      "epoch": 1.3216,
+      "grad_norm": 44.486167907714844,
+      "learning_rate": 7.537777777777778e-06,
+      "loss": 1.5784,
+      "step": 41300
+    },
+    {
+      "epoch": 1.32192,
+      "grad_norm": 46.06887435913086,
+      "learning_rate": 7.534222222222223e-06,
+      "loss": 1.6227,
+      "step": 41310
+    },
+    {
+      "epoch": 1.32224,
+      "grad_norm": 42.2825813293457,
+      "learning_rate": 7.530666666666667e-06,
+      "loss": 1.6255,
+      "step": 41320
+    },
+    {
+      "epoch": 1.32256,
+      "grad_norm": 44.45964813232422,
+      "learning_rate": 7.527111111111112e-06,
+      "loss": 1.5935,
+      "step": 41330
+    },
+    {
+      "epoch": 1.32288,
+      "grad_norm": 43.408695220947266,
+      "learning_rate": 7.523555555555556e-06,
+      "loss": 1.6149,
+      "step": 41340
+    },
+    {
+      "epoch": 1.3232,
+      "grad_norm": 44.885047912597656,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 1.6165,
+      "step": 41350
+    },
+    {
+      "epoch": 1.32352,
+      "grad_norm": 45.36357498168945,
+      "learning_rate": 7.516444444444445e-06,
+      "loss": 1.5888,
+      "step": 41360
+    },
+    {
+      "epoch": 1.3238400000000001,
+      "grad_norm": 43.71674346923828,
+      "learning_rate": 7.51288888888889e-06,
+      "loss": 1.6012,
+      "step": 41370
+    },
+    {
+      "epoch": 1.32416,
+      "grad_norm": 44.332122802734375,
+      "learning_rate": 7.509333333333334e-06,
+      "loss": 1.614,
+      "step": 41380
+    },
+    {
+      "epoch": 1.3244799999999999,
+      "grad_norm": 44.52497100830078,
+      "learning_rate": 7.505777777777778e-06,
+      "loss": 1.6144,
+      "step": 41390
+    },
+    {
+      "epoch": 1.3248,
+      "grad_norm": 45.074867248535156,
+      "learning_rate": 7.502222222222223e-06,
+      "loss": 1.6412,
+      "step": 41400
+    },
+    {
+      "epoch": 1.32512,
+      "grad_norm": 44.016639709472656,
+      "learning_rate": 7.498666666666667e-06,
+      "loss": 1.6299,
+      "step": 41410
+    },
+    {
+      "epoch": 1.32544,
+      "grad_norm": 44.419776916503906,
+      "learning_rate": 7.495111111111112e-06,
+      "loss": 1.6163,
+      "step": 41420
+    },
+    {
+      "epoch": 1.32576,
+      "grad_norm": 46.50283432006836,
+      "learning_rate": 7.4915555555555564e-06,
+      "loss": 1.573,
+      "step": 41430
+    },
+    {
+      "epoch": 1.32608,
+      "grad_norm": 46.12255859375,
+      "learning_rate": 7.488000000000001e-06,
+      "loss": 1.6028,
+      "step": 41440
+    },
+    {
+      "epoch": 1.3264,
+      "grad_norm": 45.1977653503418,
+      "learning_rate": 7.4844444444444455e-06,
+      "loss": 1.6403,
+      "step": 41450
+    },
+    {
+      "epoch": 1.32672,
+      "grad_norm": 46.167457580566406,
+      "learning_rate": 7.48088888888889e-06,
+      "loss": 1.5972,
+      "step": 41460
+    },
+    {
+      "epoch": 1.32704,
+      "grad_norm": 46.73789978027344,
+      "learning_rate": 7.477333333333335e-06,
+      "loss": 1.6071,
+      "step": 41470
+    },
+    {
+      "epoch": 1.32736,
+      "grad_norm": 45.93767547607422,
+      "learning_rate": 7.473777777777778e-06,
+      "loss": 1.6254,
+      "step": 41480
+    },
+    {
+      "epoch": 1.32768,
+      "grad_norm": 44.71085739135742,
+      "learning_rate": 7.470222222222223e-06,
+      "loss": 1.5951,
+      "step": 41490
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 45.44325256347656,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.6183,
+      "step": 41500
+    },
+    {
+      "epoch": 1.32832,
+      "grad_norm": 45.813087463378906,
+      "learning_rate": 7.463111111111112e-06,
+      "loss": 1.6019,
+      "step": 41510
+    },
+    {
+      "epoch": 1.32864,
+      "grad_norm": 43.0788459777832,
+      "learning_rate": 7.4595555555555566e-06,
+      "loss": 1.6332,
+      "step": 41520
+    },
+    {
+      "epoch": 1.32896,
+      "grad_norm": 50.90683364868164,
+      "learning_rate": 7.456000000000001e-06,
+      "loss": 1.6036,
+      "step": 41530
+    },
+    {
+      "epoch": 1.32928,
+      "grad_norm": 44.96916198730469,
+      "learning_rate": 7.452444444444446e-06,
+      "loss": 1.6276,
+      "step": 41540
+    },
+    {
+      "epoch": 1.3296000000000001,
+      "grad_norm": 45.77922439575195,
+      "learning_rate": 7.44888888888889e-06,
+      "loss": 1.6187,
+      "step": 41550
+    },
+    {
+      "epoch": 1.32992,
+      "grad_norm": 46.79462432861328,
+      "learning_rate": 7.445333333333335e-06,
+      "loss": 1.6277,
+      "step": 41560
+    },
+    {
+      "epoch": 1.3302399999999999,
+      "grad_norm": 45.878135681152344,
+      "learning_rate": 7.4417777777777785e-06,
+      "loss": 1.605,
+      "step": 41570
+    },
+    {
+      "epoch": 1.33056,
+      "grad_norm": 45.22720718383789,
+      "learning_rate": 7.438222222222223e-06,
+      "loss": 1.601,
+      "step": 41580
+    },
+    {
+      "epoch": 1.33088,
+      "grad_norm": 43.85465621948242,
+      "learning_rate": 7.434666666666668e-06,
+      "loss": 1.6047,
+      "step": 41590
+    },
+    {
+      "epoch": 1.3312,
+      "grad_norm": 43.823081970214844,
+      "learning_rate": 7.431111111111111e-06,
+      "loss": 1.6103,
+      "step": 41600
+    },
+    {
+      "epoch": 1.33152,
+      "grad_norm": 43.73310089111328,
+      "learning_rate": 7.427555555555556e-06,
+      "loss": 1.6175,
+      "step": 41610
+    },
+    {
+      "epoch": 1.3318400000000001,
+      "grad_norm": 45.74242401123047,
+      "learning_rate": 7.424e-06,
+      "loss": 1.6308,
+      "step": 41620
+    },
+    {
+      "epoch": 1.33216,
+      "grad_norm": 49.32040023803711,
+      "learning_rate": 7.420444444444445e-06,
+      "loss": 1.6227,
+      "step": 41630
+    },
+    {
+      "epoch": 1.3324799999999999,
+      "grad_norm": 44.08414840698242,
+      "learning_rate": 7.416888888888889e-06,
+      "loss": 1.618,
+      "step": 41640
+    },
+    {
+      "epoch": 1.3328,
+      "grad_norm": 42.85681915283203,
+      "learning_rate": 7.413333333333333e-06,
+      "loss": 1.6222,
+      "step": 41650
+    },
+    {
+      "epoch": 1.33312,
+      "grad_norm": 44.588043212890625,
+      "learning_rate": 7.409777777777778e-06,
+      "loss": 1.6339,
+      "step": 41660
+    },
+    {
+      "epoch": 1.33344,
+      "grad_norm": 46.16299057006836,
+      "learning_rate": 7.406222222222222e-06,
+      "loss": 1.6058,
+      "step": 41670
+    },
+    {
+      "epoch": 1.33376,
+      "grad_norm": 42.70748519897461,
+      "learning_rate": 7.402666666666667e-06,
+      "loss": 1.6168,
+      "step": 41680
+    },
+    {
+      "epoch": 1.33408,
+      "grad_norm": 45.55185317993164,
+      "learning_rate": 7.3991111111111114e-06,
+      "loss": 1.6146,
+      "step": 41690
+    },
+    {
+      "epoch": 1.3344,
+      "grad_norm": 44.13607406616211,
+      "learning_rate": 7.395555555555556e-06,
+      "loss": 1.5932,
+      "step": 41700
+    },
+    {
+      "epoch": 1.33472,
+      "grad_norm": 44.90840530395508,
+      "learning_rate": 7.3920000000000005e-06,
+      "loss": 1.6389,
+      "step": 41710
+    },
+    {
+      "epoch": 1.33504,
+      "grad_norm": 45.11433792114258,
+      "learning_rate": 7.388444444444445e-06,
+      "loss": 1.6186,
+      "step": 41720
+    },
+    {
+      "epoch": 1.33536,
+      "grad_norm": 44.75900650024414,
+      "learning_rate": 7.384888888888889e-06,
+      "loss": 1.617,
+      "step": 41730
+    },
+    {
+      "epoch": 1.33568,
+      "grad_norm": 45.705841064453125,
+      "learning_rate": 7.381333333333333e-06,
+      "loss": 1.6117,
+      "step": 41740
+    },
+    {
+      "epoch": 1.336,
+      "grad_norm": 46.18943405151367,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 1.6025,
+      "step": 41750
+    },
+    {
+      "epoch": 1.33632,
+      "grad_norm": 46.8182258605957,
+      "learning_rate": 7.3742222222222225e-06,
+      "loss": 1.5902,
+      "step": 41760
+    },
+    {
+      "epoch": 1.33664,
+      "grad_norm": 46.080604553222656,
+      "learning_rate": 7.370666666666667e-06,
+      "loss": 1.6143,
+      "step": 41770
+    },
+    {
+      "epoch": 1.33696,
+      "grad_norm": 45.11826705932617,
+      "learning_rate": 7.3671111111111116e-06,
+      "loss": 1.5883,
+      "step": 41780
+    },
+    {
+      "epoch": 1.33728,
+      "grad_norm": 44.33161926269531,
+      "learning_rate": 7.363555555555556e-06,
+      "loss": 1.6075,
+      "step": 41790
+    },
+    {
+      "epoch": 1.3376000000000001,
+      "grad_norm": 45.10640335083008,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 1.623,
+      "step": 41800
+    },
+    {
+      "epoch": 1.33792,
+      "grad_norm": 43.860267639160156,
+      "learning_rate": 7.356444444444445e-06,
+      "loss": 1.5983,
+      "step": 41810
+    },
+    {
+      "epoch": 1.3382399999999999,
+      "grad_norm": 45.181495666503906,
+      "learning_rate": 7.35288888888889e-06,
+      "loss": 1.6088,
+      "step": 41820
+    },
+    {
+      "epoch": 1.33856,
+      "grad_norm": 46.86771774291992,
+      "learning_rate": 7.3493333333333335e-06,
+      "loss": 1.5976,
+      "step": 41830
+    },
+    {
+      "epoch": 1.33888,
+      "grad_norm": 43.91908264160156,
+      "learning_rate": 7.345777777777778e-06,
+      "loss": 1.6186,
+      "step": 41840
+    },
+    {
+      "epoch": 1.3392,
+      "grad_norm": 45.94272994995117,
+      "learning_rate": 7.342222222222223e-06,
+      "loss": 1.6071,
+      "step": 41850
+    },
+    {
+      "epoch": 1.33952,
+      "grad_norm": 44.59323501586914,
+      "learning_rate": 7.338666666666667e-06,
+      "loss": 1.6292,
+      "step": 41860
+    },
+    {
+      "epoch": 1.33984,
+      "grad_norm": 46.60374450683594,
+      "learning_rate": 7.335111111111112e-06,
+      "loss": 1.6075,
+      "step": 41870
+    },
+    {
+      "epoch": 1.34016,
+      "grad_norm": 44.81357192993164,
+      "learning_rate": 7.331555555555556e-06,
+      "loss": 1.5965,
+      "step": 41880
+    },
+    {
+      "epoch": 1.34048,
+      "grad_norm": 44.506500244140625,
+      "learning_rate": 7.328000000000001e-06,
+      "loss": 1.589,
+      "step": 41890
+    },
+    {
+      "epoch": 1.3408,
+      "grad_norm": 45.52299880981445,
+      "learning_rate": 7.324444444444445e-06,
+      "loss": 1.622,
+      "step": 41900
+    },
+    {
+      "epoch": 1.34112,
+      "grad_norm": 45.362396240234375,
+      "learning_rate": 7.32088888888889e-06,
+      "loss": 1.5961,
+      "step": 41910
+    },
+    {
+      "epoch": 1.34144,
+      "grad_norm": 48.476627349853516,
+      "learning_rate": 7.3173333333333345e-06,
+      "loss": 1.6273,
+      "step": 41920
+    },
+    {
+      "epoch": 1.34176,
+      "grad_norm": 46.45351791381836,
+      "learning_rate": 7.313777777777778e-06,
+      "loss": 1.6076,
+      "step": 41930
+    },
+    {
+      "epoch": 1.34208,
+      "grad_norm": 44.8245849609375,
+      "learning_rate": 7.310222222222223e-06,
+      "loss": 1.6257,
+      "step": 41940
+    },
+    {
+      "epoch": 1.3424,
+      "grad_norm": 46.39732360839844,
+      "learning_rate": 7.306666666666667e-06,
+      "loss": 1.617,
+      "step": 41950
+    },
+    {
+      "epoch": 1.34272,
+      "grad_norm": 44.4480094909668,
+      "learning_rate": 7.303111111111112e-06,
+      "loss": 1.6128,
+      "step": 41960
+    },
+    {
+      "epoch": 1.34304,
+      "grad_norm": 46.01774597167969,
+      "learning_rate": 7.299555555555556e-06,
+      "loss": 1.6377,
+      "step": 41970
+    },
+    {
+      "epoch": 1.34336,
+      "grad_norm": 47.32066345214844,
+      "learning_rate": 7.296000000000001e-06,
+      "loss": 1.5866,
+      "step": 41980
+    },
+    {
+      "epoch": 1.34368,
+      "grad_norm": 44.58960723876953,
+      "learning_rate": 7.2924444444444455e-06,
+      "loss": 1.6074,
+      "step": 41990
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 44.3505973815918,
+      "learning_rate": 7.28888888888889e-06,
+      "loss": 1.6045,
+      "step": 42000
+    },
+    {
+      "epoch": 1.34432,
+      "grad_norm": 44.56719207763672,
+      "learning_rate": 7.285333333333335e-06,
+      "loss": 1.6252,
+      "step": 42010
+    },
+    {
+      "epoch": 1.34464,
+      "grad_norm": 45.77865219116211,
+      "learning_rate": 7.281777777777778e-06,
+      "loss": 1.6433,
+      "step": 42020
+    },
+    {
+      "epoch": 1.34496,
+      "grad_norm": 46.384979248046875,
+      "learning_rate": 7.278222222222223e-06,
+      "loss": 1.6384,
+      "step": 42030
+    },
+    {
+      "epoch": 1.34528,
+      "grad_norm": 45.07937240600586,
+      "learning_rate": 7.2746666666666674e-06,
+      "loss": 1.5896,
+      "step": 42040
+    },
+    {
+      "epoch": 1.3456000000000001,
+      "grad_norm": 43.41228485107422,
+      "learning_rate": 7.271111111111112e-06,
+      "loss": 1.5997,
+      "step": 42050
+    },
+    {
+      "epoch": 1.34592,
+      "grad_norm": 45.3345947265625,
+      "learning_rate": 7.2675555555555565e-06,
+      "loss": 1.5932,
+      "step": 42060
+    },
+    {
+      "epoch": 1.3462399999999999,
+      "grad_norm": 52.81439208984375,
+      "learning_rate": 7.264000000000001e-06,
+      "loss": 1.6182,
+      "step": 42070
+    },
+    {
+      "epoch": 1.34656,
+      "grad_norm": 77.6665267944336,
+      "learning_rate": 7.260444444444446e-06,
+      "loss": 1.5891,
+      "step": 42080
+    },
+    {
+      "epoch": 1.34688,
+      "grad_norm": 43.742164611816406,
+      "learning_rate": 7.25688888888889e-06,
+      "loss": 1.5789,
+      "step": 42090
+    },
+    {
+      "epoch": 1.3472,
+      "grad_norm": 46.076324462890625,
+      "learning_rate": 7.253333333333335e-06,
+      "loss": 1.6129,
+      "step": 42100
+    },
+    {
+      "epoch": 1.34752,
+      "grad_norm": 45.37517166137695,
+      "learning_rate": 7.249777777777779e-06,
+      "loss": 1.5823,
+      "step": 42110
+    },
+    {
+      "epoch": 1.34784,
+      "grad_norm": 45.582679748535156,
+      "learning_rate": 7.246222222222222e-06,
+      "loss": 1.6309,
+      "step": 42120
+    },
+    {
+      "epoch": 1.34816,
+      "grad_norm": 45.533573150634766,
+      "learning_rate": 7.242666666666667e-06,
+      "loss": 1.6029,
+      "step": 42130
+    },
+    {
+      "epoch": 1.34848,
+      "grad_norm": 46.39393615722656,
+      "learning_rate": 7.239111111111111e-06,
+      "loss": 1.5989,
+      "step": 42140
+    },
+    {
+      "epoch": 1.3488,
+      "grad_norm": 41.84919738769531,
+      "learning_rate": 7.235555555555556e-06,
+      "loss": 1.6176,
+      "step": 42150
+    },
+    {
+      "epoch": 1.34912,
+      "grad_norm": 45.00162887573242,
+      "learning_rate": 7.232e-06,
+      "loss": 1.6299,
+      "step": 42160
+    },
+    {
+      "epoch": 1.34944,
+      "grad_norm": 47.52516174316406,
+      "learning_rate": 7.228444444444445e-06,
+      "loss": 1.5941,
+      "step": 42170
+    },
+    {
+      "epoch": 1.34976,
+      "grad_norm": 43.75261306762695,
+      "learning_rate": 7.224888888888889e-06,
+      "loss": 1.6165,
+      "step": 42180
+    },
+    {
+      "epoch": 1.35008,
+      "grad_norm": 45.649635314941406,
+      "learning_rate": 7.221333333333333e-06,
+      "loss": 1.606,
+      "step": 42190
+    },
+    {
+      "epoch": 1.3504,
+      "grad_norm": 43.680885314941406,
+      "learning_rate": 7.217777777777778e-06,
+      "loss": 1.6042,
+      "step": 42200
+    },
+    {
+      "epoch": 1.35072,
+      "grad_norm": 45.03094482421875,
+      "learning_rate": 7.214222222222222e-06,
+      "loss": 1.5914,
+      "step": 42210
+    },
+    {
+      "epoch": 1.35104,
+      "grad_norm": 44.120704650878906,
+      "learning_rate": 7.210666666666667e-06,
+      "loss": 1.622,
+      "step": 42220
+    },
+    {
+      "epoch": 1.3513600000000001,
+      "grad_norm": 44.486473083496094,
+      "learning_rate": 7.207111111111111e-06,
+      "loss": 1.6104,
+      "step": 42230
+    },
+    {
+      "epoch": 1.35168,
+      "grad_norm": 44.795597076416016,
+      "learning_rate": 7.203555555555556e-06,
+      "loss": 1.5942,
+      "step": 42240
+    },
+    {
+      "epoch": 1.3519999999999999,
+      "grad_norm": 45.60823059082031,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.5909,
+      "step": 42250
+    },
+    {
+      "epoch": 1.35232,
+      "grad_norm": 45.53849792480469,
+      "learning_rate": 7.196444444444445e-06,
+      "loss": 1.6015,
+      "step": 42260
+    },
+    {
+      "epoch": 1.35264,
+      "grad_norm": 45.30690002441406,
+      "learning_rate": 7.19288888888889e-06,
+      "loss": 1.6008,
+      "step": 42270
+    },
+    {
+      "epoch": 1.35296,
+      "grad_norm": 45.725833892822266,
+      "learning_rate": 7.189333333333333e-06,
+      "loss": 1.6238,
+      "step": 42280
+    },
+    {
+      "epoch": 1.35328,
+      "grad_norm": 45.394039154052734,
+      "learning_rate": 7.185777777777778e-06,
+      "loss": 1.6399,
+      "step": 42290
+    },
+    {
+      "epoch": 1.3536000000000001,
+      "grad_norm": 43.31956481933594,
+      "learning_rate": 7.1822222222222224e-06,
+      "loss": 1.5954,
+      "step": 42300
+    },
+    {
+      "epoch": 1.35392,
+      "grad_norm": 45.152774810791016,
+      "learning_rate": 7.178666666666667e-06,
+      "loss": 1.6,
+      "step": 42310
+    },
+    {
+      "epoch": 1.3542399999999999,
+      "grad_norm": 45.98184585571289,
+      "learning_rate": 7.1751111111111116e-06,
+      "loss": 1.6067,
+      "step": 42320
+    },
+    {
+      "epoch": 1.35456,
+      "grad_norm": 44.53315734863281,
+      "learning_rate": 7.171555555555556e-06,
+      "loss": 1.6073,
+      "step": 42330
+    },
+    {
+      "epoch": 1.35488,
+      "grad_norm": 50.22261047363281,
+      "learning_rate": 7.168000000000001e-06,
+      "loss": 1.6198,
+      "step": 42340
+    },
+    {
+      "epoch": 1.3552,
+      "grad_norm": 44.62266159057617,
+      "learning_rate": 7.164444444444445e-06,
+      "loss": 1.623,
+      "step": 42350
+    },
+    {
+      "epoch": 1.35552,
+      "grad_norm": 44.818748474121094,
+      "learning_rate": 7.16088888888889e-06,
+      "loss": 1.5935,
+      "step": 42360
+    },
+    {
+      "epoch": 1.35584,
+      "grad_norm": 43.746421813964844,
+      "learning_rate": 7.157333333333334e-06,
+      "loss": 1.5749,
+      "step": 42370
+    },
+    {
+      "epoch": 1.35616,
+      "grad_norm": 46.356353759765625,
+      "learning_rate": 7.153777777777778e-06,
+      "loss": 1.6146,
+      "step": 42380
+    },
+    {
+      "epoch": 1.35648,
+      "grad_norm": 47.375152587890625,
+      "learning_rate": 7.150222222222223e-06,
+      "loss": 1.6118,
+      "step": 42390
+    },
+    {
+      "epoch": 1.3568,
+      "grad_norm": 46.35893630981445,
+      "learning_rate": 7.146666666666667e-06,
+      "loss": 1.6158,
+      "step": 42400
+    },
+    {
+      "epoch": 1.35712,
+      "grad_norm": 46.352848052978516,
+      "learning_rate": 7.143111111111112e-06,
+      "loss": 1.5878,
+      "step": 42410
+    },
+    {
+      "epoch": 1.35744,
+      "grad_norm": 43.320281982421875,
+      "learning_rate": 7.139555555555556e-06,
+      "loss": 1.6167,
+      "step": 42420
+    },
+    {
+      "epoch": 1.35776,
+      "grad_norm": 42.796791076660156,
+      "learning_rate": 7.136000000000001e-06,
+      "loss": 1.6257,
+      "step": 42430
+    },
+    {
+      "epoch": 1.35808,
+      "grad_norm": 45.10505294799805,
+      "learning_rate": 7.132444444444445e-06,
+      "loss": 1.5949,
+      "step": 42440
+    },
+    {
+      "epoch": 1.3584,
+      "grad_norm": 45.902618408203125,
+      "learning_rate": 7.12888888888889e-06,
+      "loss": 1.635,
+      "step": 42450
+    },
+    {
+      "epoch": 1.35872,
+      "grad_norm": 46.71749496459961,
+      "learning_rate": 7.1253333333333345e-06,
+      "loss": 1.6115,
+      "step": 42460
+    },
+    {
+      "epoch": 1.35904,
+      "grad_norm": 43.46259689331055,
+      "learning_rate": 7.121777777777778e-06,
+      "loss": 1.5951,
+      "step": 42470
+    },
+    {
+      "epoch": 1.3593600000000001,
+      "grad_norm": 47.766231536865234,
+      "learning_rate": 7.118222222222223e-06,
+      "loss": 1.6121,
+      "step": 42480
+    },
+    {
+      "epoch": 1.35968,
+      "grad_norm": 44.797122955322266,
+      "learning_rate": 7.114666666666667e-06,
+      "loss": 1.5854,
+      "step": 42490
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 45.8564567565918,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 1.6086,
+      "step": 42500
+    },
+    {
+      "epoch": 1.36032,
+      "grad_norm": 45.51469802856445,
+      "learning_rate": 7.107555555555556e-06,
+      "loss": 1.6422,
+      "step": 42510
+    },
+    {
+      "epoch": 1.36064,
+      "grad_norm": 45.435096740722656,
+      "learning_rate": 7.104000000000001e-06,
+      "loss": 1.6109,
+      "step": 42520
+    },
+    {
+      "epoch": 1.36096,
+      "grad_norm": 46.81600570678711,
+      "learning_rate": 7.1004444444444455e-06,
+      "loss": 1.6098,
+      "step": 42530
+    },
+    {
+      "epoch": 1.36128,
+      "grad_norm": 44.625938415527344,
+      "learning_rate": 7.09688888888889e-06,
+      "loss": 1.615,
+      "step": 42540
+    },
+    {
+      "epoch": 1.3616,
+      "grad_norm": 44.89677429199219,
+      "learning_rate": 7.093333333333335e-06,
+      "loss": 1.613,
+      "step": 42550
+    },
+    {
+      "epoch": 1.36192,
+      "grad_norm": 44.43601989746094,
+      "learning_rate": 7.089777777777779e-06,
+      "loss": 1.5775,
+      "step": 42560
+    },
+    {
+      "epoch": 1.36224,
+      "grad_norm": 45.43404006958008,
+      "learning_rate": 7.086222222222223e-06,
+      "loss": 1.6413,
+      "step": 42570
+    },
+    {
+      "epoch": 1.36256,
+      "grad_norm": 45.00292205810547,
+      "learning_rate": 7.082666666666667e-06,
+      "loss": 1.6518,
+      "step": 42580
+    },
+    {
+      "epoch": 1.36288,
+      "grad_norm": 46.85793685913086,
+      "learning_rate": 7.079111111111112e-06,
+      "loss": 1.5907,
+      "step": 42590
+    },
+    {
+      "epoch": 1.3632,
+      "grad_norm": 42.96330642700195,
+      "learning_rate": 7.0755555555555565e-06,
+      "loss": 1.6054,
+      "step": 42600
+    },
+    {
+      "epoch": 1.36352,
+      "grad_norm": 44.131099700927734,
+      "learning_rate": 7.072000000000001e-06,
+      "loss": 1.6187,
+      "step": 42610
+    },
+    {
+      "epoch": 1.36384,
+      "grad_norm": 44.26656723022461,
+      "learning_rate": 7.068444444444446e-06,
+      "loss": 1.6102,
+      "step": 42620
+    },
+    {
+      "epoch": 1.36416,
+      "grad_norm": 44.00492858886719,
+      "learning_rate": 7.06488888888889e-06,
+      "loss": 1.6255,
+      "step": 42630
+    },
+    {
+      "epoch": 1.36448,
+      "grad_norm": 45.8413200378418,
+      "learning_rate": 7.061333333333333e-06,
+      "loss": 1.5964,
+      "step": 42640
+    },
+    {
+      "epoch": 1.3648,
+      "grad_norm": 46.41115951538086,
+      "learning_rate": 7.057777777777778e-06,
+      "loss": 1.5851,
+      "step": 42650
+    },
+    {
+      "epoch": 1.3651200000000001,
+      "grad_norm": 45.85788345336914,
+      "learning_rate": 7.054222222222222e-06,
+      "loss": 1.6009,
+      "step": 42660
+    },
+    {
+      "epoch": 1.36544,
+      "grad_norm": 44.54059600830078,
+      "learning_rate": 7.050666666666667e-06,
+      "loss": 1.5774,
+      "step": 42670
+    },
+    {
+      "epoch": 1.3657599999999999,
+      "grad_norm": 43.62693786621094,
+      "learning_rate": 7.047111111111111e-06,
+      "loss": 1.6298,
+      "step": 42680
+    },
+    {
+      "epoch": 1.36608,
+      "grad_norm": 44.06049346923828,
+      "learning_rate": 7.043555555555556e-06,
+      "loss": 1.6006,
+      "step": 42690
+    },
+    {
+      "epoch": 1.3664,
+      "grad_norm": 45.65952682495117,
+      "learning_rate": 7.04e-06,
+      "loss": 1.6428,
+      "step": 42700
+    },
+    {
+      "epoch": 1.36672,
+      "grad_norm": 45.37552261352539,
+      "learning_rate": 7.036444444444445e-06,
+      "loss": 1.6304,
+      "step": 42710
+    },
+    {
+      "epoch": 1.36704,
+      "grad_norm": 45.344573974609375,
+      "learning_rate": 7.0328888888888895e-06,
+      "loss": 1.5978,
+      "step": 42720
+    },
+    {
+      "epoch": 1.3673600000000001,
+      "grad_norm": 45.009918212890625,
+      "learning_rate": 7.029333333333333e-06,
+      "loss": 1.6036,
+      "step": 42730
+    },
+    {
+      "epoch": 1.36768,
+      "grad_norm": 47.78025436401367,
+      "learning_rate": 7.025777777777778e-06,
+      "loss": 1.6182,
+      "step": 42740
+    },
+    {
+      "epoch": 1.3679999999999999,
+      "grad_norm": 43.18815612792969,
+      "learning_rate": 7.022222222222222e-06,
+      "loss": 1.6248,
+      "step": 42750
+    },
+    {
+      "epoch": 1.36832,
+      "grad_norm": 44.154388427734375,
+      "learning_rate": 7.018666666666667e-06,
+      "loss": 1.5975,
+      "step": 42760
+    },
+    {
+      "epoch": 1.36864,
+      "grad_norm": 44.73991775512695,
+      "learning_rate": 7.015111111111111e-06,
+      "loss": 1.5802,
+      "step": 42770
+    },
+    {
+      "epoch": 1.36896,
+      "grad_norm": 45.32402801513672,
+      "learning_rate": 7.011555555555556e-06,
+      "loss": 1.6301,
+      "step": 42780
+    },
+    {
+      "epoch": 1.36928,
+      "grad_norm": 45.557464599609375,
+      "learning_rate": 7.0080000000000005e-06,
+      "loss": 1.6012,
+      "step": 42790
+    },
+    {
+      "epoch": 1.3696,
+      "grad_norm": 44.68947982788086,
+      "learning_rate": 7.004444444444445e-06,
+      "loss": 1.5771,
+      "step": 42800
+    },
+    {
+      "epoch": 1.36992,
+      "grad_norm": 47.947853088378906,
+      "learning_rate": 7.00088888888889e-06,
+      "loss": 1.6116,
+      "step": 42810
+    },
+    {
+      "epoch": 1.37024,
+      "grad_norm": 46.1669807434082,
+      "learning_rate": 6.997333333333334e-06,
+      "loss": 1.6344,
+      "step": 42820
+    },
+    {
+      "epoch": 1.37056,
+      "grad_norm": 42.729087829589844,
+      "learning_rate": 6.993777777777778e-06,
+      "loss": 1.6033,
+      "step": 42830
+    },
+    {
+      "epoch": 1.37088,
+      "grad_norm": 43.983116149902344,
+      "learning_rate": 6.990222222222222e-06,
+      "loss": 1.5901,
+      "step": 42840
+    },
+    {
+      "epoch": 1.3712,
+      "grad_norm": 47.0743408203125,
+      "learning_rate": 6.986666666666667e-06,
+      "loss": 1.5971,
+      "step": 42850
+    },
+    {
+      "epoch": 1.37152,
+      "grad_norm": 45.89794158935547,
+      "learning_rate": 6.9831111111111115e-06,
+      "loss": 1.5891,
+      "step": 42860
+    },
+    {
+      "epoch": 1.37184,
+      "grad_norm": 47.009464263916016,
+      "learning_rate": 6.979555555555556e-06,
+      "loss": 1.624,
+      "step": 42870
+    },
+    {
+      "epoch": 1.37216,
+      "grad_norm": 44.28515625,
+      "learning_rate": 6.976000000000001e-06,
+      "loss": 1.6088,
+      "step": 42880
+    },
+    {
+      "epoch": 1.37248,
+      "grad_norm": 46.215370178222656,
+      "learning_rate": 6.972444444444445e-06,
+      "loss": 1.5858,
+      "step": 42890
+    },
+    {
+      "epoch": 1.3728,
+      "grad_norm": 43.991416931152344,
+      "learning_rate": 6.96888888888889e-06,
+      "loss": 1.5814,
+      "step": 42900
+    },
+    {
+      "epoch": 1.3731200000000001,
+      "grad_norm": 46.066654205322266,
+      "learning_rate": 6.965333333333334e-06,
+      "loss": 1.6109,
+      "step": 42910
+    },
+    {
+      "epoch": 1.37344,
+      "grad_norm": 44.713748931884766,
+      "learning_rate": 6.961777777777778e-06,
+      "loss": 1.5892,
+      "step": 42920
+    },
+    {
+      "epoch": 1.3737599999999999,
+      "grad_norm": 45.68426513671875,
+      "learning_rate": 6.9582222222222226e-06,
+      "loss": 1.5985,
+      "step": 42930
+    },
+    {
+      "epoch": 1.37408,
+      "grad_norm": 44.62791442871094,
+      "learning_rate": 6.954666666666667e-06,
+      "loss": 1.5908,
+      "step": 42940
+    },
+    {
+      "epoch": 1.3744,
+      "grad_norm": 44.590370178222656,
+      "learning_rate": 6.951111111111112e-06,
+      "loss": 1.5942,
+      "step": 42950
+    },
+    {
+      "epoch": 1.37472,
+      "grad_norm": 46.614261627197266,
+      "learning_rate": 6.947555555555556e-06,
+      "loss": 1.6259,
+      "step": 42960
+    },
+    {
+      "epoch": 1.37504,
+      "grad_norm": 45.88407897949219,
+      "learning_rate": 6.944000000000001e-06,
+      "loss": 1.6235,
+      "step": 42970
+    },
+    {
+      "epoch": 1.3753600000000001,
+      "grad_norm": 46.90294647216797,
+      "learning_rate": 6.940444444444445e-06,
+      "loss": 1.5938,
+      "step": 42980
+    },
+    {
+      "epoch": 1.37568,
+      "grad_norm": 44.70612716674805,
+      "learning_rate": 6.93688888888889e-06,
+      "loss": 1.605,
+      "step": 42990
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 43.3810920715332,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.6153,
+      "step": 43000
+    },
+    {
+      "epoch": 1.37632,
+      "grad_norm": 43.62282180786133,
+      "learning_rate": 6.929777777777779e-06,
+      "loss": 1.5724,
+      "step": 43010
+    },
+    {
+      "epoch": 1.37664,
+      "grad_norm": 44.38874816894531,
+      "learning_rate": 6.926222222222223e-06,
+      "loss": 1.637,
+      "step": 43020
+    },
+    {
+      "epoch": 1.37696,
+      "grad_norm": 44.056522369384766,
+      "learning_rate": 6.922666666666667e-06,
+      "loss": 1.6198,
+      "step": 43030
+    },
+    {
+      "epoch": 1.37728,
+      "grad_norm": 44.81947326660156,
+      "learning_rate": 6.919111111111112e-06,
+      "loss": 1.6117,
+      "step": 43040
+    },
+    {
+      "epoch": 1.3776,
+      "grad_norm": 43.83502197265625,
+      "learning_rate": 6.915555555555556e-06,
+      "loss": 1.5649,
+      "step": 43050
+    },
+    {
+      "epoch": 1.37792,
+      "grad_norm": 44.356773376464844,
+      "learning_rate": 6.912000000000001e-06,
+      "loss": 1.6297,
+      "step": 43060
+    },
+    {
+      "epoch": 1.37824,
+      "grad_norm": 48.54316711425781,
+      "learning_rate": 6.9084444444444455e-06,
+      "loss": 1.6002,
+      "step": 43070
+    },
+    {
+      "epoch": 1.37856,
+      "grad_norm": 45.56771469116211,
+      "learning_rate": 6.90488888888889e-06,
+      "loss": 1.6219,
+      "step": 43080
+    },
+    {
+      "epoch": 1.37888,
+      "grad_norm": 43.97990036010742,
+      "learning_rate": 6.9013333333333346e-06,
+      "loss": 1.6474,
+      "step": 43090
+    },
+    {
+      "epoch": 1.3792,
+      "grad_norm": 44.80003356933594,
+      "learning_rate": 6.897777777777779e-06,
+      "loss": 1.5931,
+      "step": 43100
+    },
+    {
+      "epoch": 1.37952,
+      "grad_norm": 43.22229766845703,
+      "learning_rate": 6.894222222222224e-06,
+      "loss": 1.5845,
+      "step": 43110
+    },
+    {
+      "epoch": 1.37984,
+      "grad_norm": 45.73298263549805,
+      "learning_rate": 6.890666666666667e-06,
+      "loss": 1.6028,
+      "step": 43120
+    },
+    {
+      "epoch": 1.38016,
+      "grad_norm": 44.59761047363281,
+      "learning_rate": 6.887111111111112e-06,
+      "loss": 1.5751,
+      "step": 43130
+    },
+    {
+      "epoch": 1.38048,
+      "grad_norm": 45.15092849731445,
+      "learning_rate": 6.8835555555555565e-06,
+      "loss": 1.6216,
+      "step": 43140
+    },
+    {
+      "epoch": 1.3808,
+      "grad_norm": 45.5977897644043,
+      "learning_rate": 6.88e-06,
+      "loss": 1.6063,
+      "step": 43150
+    },
+    {
+      "epoch": 1.3811200000000001,
+      "grad_norm": 43.23210525512695,
+      "learning_rate": 6.876444444444445e-06,
+      "loss": 1.6057,
+      "step": 43160
+    },
+    {
+      "epoch": 1.38144,
+      "grad_norm": 44.85715866088867,
+      "learning_rate": 6.872888888888889e-06,
+      "loss": 1.6179,
+      "step": 43170
+    },
+    {
+      "epoch": 1.3817599999999999,
+      "grad_norm": 45.367855072021484,
+      "learning_rate": 6.869333333333333e-06,
+      "loss": 1.5685,
+      "step": 43180
+    },
+    {
+      "epoch": 1.38208,
+      "grad_norm": 43.84122085571289,
+      "learning_rate": 6.8657777777777776e-06,
+      "loss": 1.6033,
+      "step": 43190
+    },
+    {
+      "epoch": 1.3824,
+      "grad_norm": 44.58828353881836,
+      "learning_rate": 6.862222222222222e-06,
+      "loss": 1.6334,
+      "step": 43200
+    },
+    {
+      "epoch": 1.38272,
+      "grad_norm": 45.26247787475586,
+      "learning_rate": 6.858666666666667e-06,
+      "loss": 1.6057,
+      "step": 43210
+    },
+    {
+      "epoch": 1.38304,
+      "grad_norm": 42.196754455566406,
+      "learning_rate": 6.855111111111111e-06,
+      "loss": 1.6067,
+      "step": 43220
+    },
+    {
+      "epoch": 1.38336,
+      "grad_norm": 47.0052375793457,
+      "learning_rate": 6.851555555555556e-06,
+      "loss": 1.6125,
+      "step": 43230
+    },
+    {
+      "epoch": 1.38368,
+      "grad_norm": 44.26320266723633,
+      "learning_rate": 6.848e-06,
+      "loss": 1.6237,
+      "step": 43240
+    },
+    {
+      "epoch": 1.384,
+      "grad_norm": 44.31113052368164,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 1.6111,
+      "step": 43250
+    },
+    {
+      "epoch": 1.38432,
+      "grad_norm": 45.720027923583984,
+      "learning_rate": 6.8408888888888894e-06,
+      "loss": 1.5989,
+      "step": 43260
+    },
+    {
+      "epoch": 1.38464,
+      "grad_norm": 46.495758056640625,
+      "learning_rate": 6.837333333333334e-06,
+      "loss": 1.605,
+      "step": 43270
+    },
+    {
+      "epoch": 1.38496,
+      "grad_norm": 45.887367248535156,
+      "learning_rate": 6.833777777777778e-06,
+      "loss": 1.5901,
+      "step": 43280
+    },
+    {
+      "epoch": 1.38528,
+      "grad_norm": 44.445804595947266,
+      "learning_rate": 6.830222222222222e-06,
+      "loss": 1.5793,
+      "step": 43290
+    },
+    {
+      "epoch": 1.3856,
+      "grad_norm": 46.46500778198242,
+      "learning_rate": 6.826666666666667e-06,
+      "loss": 1.6001,
+      "step": 43300
+    },
+    {
+      "epoch": 1.38592,
+      "grad_norm": 44.54768753051758,
+      "learning_rate": 6.823111111111111e-06,
+      "loss": 1.6045,
+      "step": 43310
+    },
+    {
+      "epoch": 1.38624,
+      "grad_norm": 45.025169372558594,
+      "learning_rate": 6.819555555555556e-06,
+      "loss": 1.6225,
+      "step": 43320
+    },
+    {
+      "epoch": 1.38656,
+      "grad_norm": 44.17384719848633,
+      "learning_rate": 6.8160000000000005e-06,
+      "loss": 1.576,
+      "step": 43330
+    },
+    {
+      "epoch": 1.3868800000000001,
+      "grad_norm": 44.44326400756836,
+      "learning_rate": 6.812444444444445e-06,
+      "loss": 1.5932,
+      "step": 43340
+    },
+    {
+      "epoch": 1.3872,
+      "grad_norm": 45.172157287597656,
+      "learning_rate": 6.80888888888889e-06,
+      "loss": 1.5878,
+      "step": 43350
+    },
+    {
+      "epoch": 1.3875199999999999,
+      "grad_norm": 45.87239074707031,
+      "learning_rate": 6.805333333333334e-06,
+      "loss": 1.6009,
+      "step": 43360
+    },
+    {
+      "epoch": 1.38784,
+      "grad_norm": 44.63097381591797,
+      "learning_rate": 6.801777777777778e-06,
+      "loss": 1.6016,
+      "step": 43370
+    },
+    {
+      "epoch": 1.38816,
+      "grad_norm": 43.94593811035156,
+      "learning_rate": 6.798222222222222e-06,
+      "loss": 1.5834,
+      "step": 43380
+    },
+    {
+      "epoch": 1.38848,
+      "grad_norm": 46.12236785888672,
+      "learning_rate": 6.794666666666667e-06,
+      "loss": 1.5865,
+      "step": 43390
+    },
+    {
+      "epoch": 1.3888,
+      "grad_norm": 45.60383605957031,
+      "learning_rate": 6.7911111111111115e-06,
+      "loss": 1.5848,
+      "step": 43400
+    },
+    {
+      "epoch": 1.3891200000000001,
+      "grad_norm": 45.66219711303711,
+      "learning_rate": 6.787555555555556e-06,
+      "loss": 1.6455,
+      "step": 43410
+    },
+    {
+      "epoch": 1.38944,
+      "grad_norm": 43.15108108520508,
+      "learning_rate": 6.784000000000001e-06,
+      "loss": 1.6232,
+      "step": 43420
+    },
+    {
+      "epoch": 1.3897599999999999,
+      "grad_norm": 44.42317581176758,
+      "learning_rate": 6.780444444444445e-06,
+      "loss": 1.6398,
+      "step": 43430
+    },
+    {
+      "epoch": 1.39008,
+      "grad_norm": 46.42605972290039,
+      "learning_rate": 6.77688888888889e-06,
+      "loss": 1.6111,
+      "step": 43440
+    },
+    {
+      "epoch": 1.3904,
+      "grad_norm": 44.38256072998047,
+      "learning_rate": 6.773333333333334e-06,
+      "loss": 1.5726,
+      "step": 43450
+    },
+    {
+      "epoch": 1.39072,
+      "grad_norm": 45.85690689086914,
+      "learning_rate": 6.769777777777779e-06,
+      "loss": 1.6,
+      "step": 43460
+    },
+    {
+      "epoch": 1.39104,
+      "grad_norm": 44.399715423583984,
+      "learning_rate": 6.7662222222222225e-06,
+      "loss": 1.6328,
+      "step": 43470
+    },
+    {
+      "epoch": 1.39136,
+      "grad_norm": 45.213165283203125,
+      "learning_rate": 6.762666666666667e-06,
+      "loss": 1.6013,
+      "step": 43480
+    },
+    {
+      "epoch": 1.39168,
+      "grad_norm": 44.65245819091797,
+      "learning_rate": 6.759111111111112e-06,
+      "loss": 1.6072,
+      "step": 43490
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 47.13508987426758,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 1.6562,
+      "step": 43500
+    },
+    {
+      "epoch": 1.39232,
+      "grad_norm": 46.34877395629883,
+      "learning_rate": 6.752000000000001e-06,
+      "loss": 1.6082,
+      "step": 43510
+    },
+    {
+      "epoch": 1.39264,
+      "grad_norm": 48.1592903137207,
+      "learning_rate": 6.748444444444445e-06,
+      "loss": 1.6268,
+      "step": 43520
+    },
+    {
+      "epoch": 1.39296,
+      "grad_norm": 43.81608581542969,
+      "learning_rate": 6.74488888888889e-06,
+      "loss": 1.5885,
+      "step": 43530
+    },
+    {
+      "epoch": 1.39328,
+      "grad_norm": 44.738468170166016,
+      "learning_rate": 6.741333333333334e-06,
+      "loss": 1.6022,
+      "step": 43540
+    },
+    {
+      "epoch": 1.3936,
+      "grad_norm": 45.94084930419922,
+      "learning_rate": 6.737777777777779e-06,
+      "loss": 1.6218,
+      "step": 43550
+    },
+    {
+      "epoch": 1.39392,
+      "grad_norm": 43.44023132324219,
+      "learning_rate": 6.7342222222222235e-06,
+      "loss": 1.6135,
+      "step": 43560
+    },
+    {
+      "epoch": 1.39424,
+      "grad_norm": 45.58948516845703,
+      "learning_rate": 6.730666666666667e-06,
+      "loss": 1.598,
+      "step": 43570
+    },
+    {
+      "epoch": 1.39456,
+      "grad_norm": 48.460453033447266,
+      "learning_rate": 6.727111111111112e-06,
+      "loss": 1.6107,
+      "step": 43580
+    },
+    {
+      "epoch": 1.3948800000000001,
+      "grad_norm": 43.78839111328125,
+      "learning_rate": 6.723555555555556e-06,
+      "loss": 1.5936,
+      "step": 43590
+    },
+    {
+      "epoch": 1.3952,
+      "grad_norm": 43.97420883178711,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 1.5876,
+      "step": 43600
+    },
+    {
+      "epoch": 1.3955199999999999,
+      "grad_norm": 46.42491149902344,
+      "learning_rate": 6.7164444444444454e-06,
+      "loss": 1.5972,
+      "step": 43610
+    },
+    {
+      "epoch": 1.39584,
+      "grad_norm": 45.202362060546875,
+      "learning_rate": 6.71288888888889e-06,
+      "loss": 1.6245,
+      "step": 43620
+    },
+    {
+      "epoch": 1.39616,
+      "grad_norm": 44.71420669555664,
+      "learning_rate": 6.7093333333333345e-06,
+      "loss": 1.6262,
+      "step": 43630
+    },
+    {
+      "epoch": 1.39648,
+      "grad_norm": 45.896907806396484,
+      "learning_rate": 6.705777777777779e-06,
+      "loss": 1.5995,
+      "step": 43640
+    },
+    {
+      "epoch": 1.3968,
+      "grad_norm": 44.3741455078125,
+      "learning_rate": 6.702222222222224e-06,
+      "loss": 1.6059,
+      "step": 43650
+    },
+    {
+      "epoch": 1.39712,
+      "grad_norm": 43.47319412231445,
+      "learning_rate": 6.698666666666667e-06,
+      "loss": 1.588,
+      "step": 43660
+    },
+    {
+      "epoch": 1.39744,
+      "grad_norm": 45.12184143066406,
+      "learning_rate": 6.695111111111111e-06,
+      "loss": 1.5827,
+      "step": 43670
+    },
+    {
+      "epoch": 1.39776,
+      "grad_norm": 46.409576416015625,
+      "learning_rate": 6.691555555555556e-06,
+      "loss": 1.617,
+      "step": 43680
+    },
+    {
+      "epoch": 1.39808,
+      "grad_norm": 45.21237564086914,
+      "learning_rate": 6.688e-06,
+      "loss": 1.627,
+      "step": 43690
+    },
+    {
+      "epoch": 1.3984,
+      "grad_norm": 45.58821105957031,
+      "learning_rate": 6.684444444444445e-06,
+      "loss": 1.6018,
+      "step": 43700
+    },
+    {
+      "epoch": 1.39872,
+      "grad_norm": 45.74357604980469,
+      "learning_rate": 6.680888888888889e-06,
+      "loss": 1.6084,
+      "step": 43710
+    },
+    {
+      "epoch": 1.39904,
+      "grad_norm": 44.05308151245117,
+      "learning_rate": 6.677333333333334e-06,
+      "loss": 1.5947,
+      "step": 43720
+    },
+    {
+      "epoch": 1.39936,
+      "grad_norm": 45.306331634521484,
+      "learning_rate": 6.6737777777777775e-06,
+      "loss": 1.6247,
+      "step": 43730
+    },
+    {
+      "epoch": 1.39968,
+      "grad_norm": 44.71645736694336,
+      "learning_rate": 6.670222222222222e-06,
+      "loss": 1.6086,
+      "step": 43740
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 43.80806350708008,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.6023,
+      "step": 43750
+    },
+    {
+      "epoch": 1.40032,
+      "grad_norm": 44.42367172241211,
+      "learning_rate": 6.663111111111111e-06,
+      "loss": 1.557,
+      "step": 43760
+    },
+    {
+      "epoch": 1.40064,
+      "grad_norm": 44.97092056274414,
+      "learning_rate": 6.659555555555556e-06,
+      "loss": 1.6089,
+      "step": 43770
+    },
+    {
+      "epoch": 1.40096,
+      "grad_norm": 43.39394760131836,
+      "learning_rate": 6.656e-06,
+      "loss": 1.6084,
+      "step": 43780
+    },
+    {
+      "epoch": 1.40128,
+      "grad_norm": 47.56688690185547,
+      "learning_rate": 6.652444444444445e-06,
+      "loss": 1.5989,
+      "step": 43790
+    },
+    {
+      "epoch": 1.4016,
+      "grad_norm": 44.7179069519043,
+      "learning_rate": 6.648888888888889e-06,
+      "loss": 1.5982,
+      "step": 43800
+    },
+    {
+      "epoch": 1.40192,
+      "grad_norm": 44.70976257324219,
+      "learning_rate": 6.645333333333334e-06,
+      "loss": 1.6146,
+      "step": 43810
+    },
+    {
+      "epoch": 1.40224,
+      "grad_norm": 43.4413948059082,
+      "learning_rate": 6.6417777777777785e-06,
+      "loss": 1.6083,
+      "step": 43820
+    },
+    {
+      "epoch": 1.40256,
+      "grad_norm": 44.52791976928711,
+      "learning_rate": 6.638222222222222e-06,
+      "loss": 1.5993,
+      "step": 43830
+    },
+    {
+      "epoch": 1.4028800000000001,
+      "grad_norm": 44.93545150756836,
+      "learning_rate": 6.634666666666667e-06,
+      "loss": 1.5721,
+      "step": 43840
+    },
+    {
+      "epoch": 1.4032,
+      "grad_norm": 46.78096389770508,
+      "learning_rate": 6.631111111111111e-06,
+      "loss": 1.6,
+      "step": 43850
+    },
+    {
+      "epoch": 1.4035199999999999,
+      "grad_norm": 49.197288513183594,
+      "learning_rate": 6.627555555555556e-06,
+      "loss": 1.6161,
+      "step": 43860
+    },
+    {
+      "epoch": 1.40384,
+      "grad_norm": 45.466251373291016,
+      "learning_rate": 6.6240000000000004e-06,
+      "loss": 1.6268,
+      "step": 43870
+    },
+    {
+      "epoch": 1.40416,
+      "grad_norm": 45.11639404296875,
+      "learning_rate": 6.620444444444445e-06,
+      "loss": 1.5927,
+      "step": 43880
+    },
+    {
+      "epoch": 1.40448,
+      "grad_norm": 43.47707748413086,
+      "learning_rate": 6.6168888888888896e-06,
+      "loss": 1.6192,
+      "step": 43890
+    },
+    {
+      "epoch": 1.4048,
+      "grad_norm": 43.626075744628906,
+      "learning_rate": 6.613333333333334e-06,
+      "loss": 1.588,
+      "step": 43900
+    },
+    {
+      "epoch": 1.40512,
+      "grad_norm": 43.15353775024414,
+      "learning_rate": 6.609777777777779e-06,
+      "loss": 1.6191,
+      "step": 43910
+    },
+    {
+      "epoch": 1.40544,
+      "grad_norm": 47.36770248413086,
+      "learning_rate": 6.606222222222222e-06,
+      "loss": 1.6061,
+      "step": 43920
+    },
+    {
+      "epoch": 1.40576,
+      "grad_norm": 44.976531982421875,
+      "learning_rate": 6.602666666666667e-06,
+      "loss": 1.6152,
+      "step": 43930
+    },
+    {
+      "epoch": 1.40608,
+      "grad_norm": 44.371978759765625,
+      "learning_rate": 6.5991111111111115e-06,
+      "loss": 1.5464,
+      "step": 43940
+    },
+    {
+      "epoch": 1.4064,
+      "grad_norm": 43.97093200683594,
+      "learning_rate": 6.595555555555556e-06,
+      "loss": 1.5851,
+      "step": 43950
+    },
+    {
+      "epoch": 1.40672,
+      "grad_norm": 43.370418548583984,
+      "learning_rate": 6.592000000000001e-06,
+      "loss": 1.5712,
+      "step": 43960
+    },
+    {
+      "epoch": 1.40704,
+      "grad_norm": 45.69548797607422,
+      "learning_rate": 6.588444444444445e-06,
+      "loss": 1.5918,
+      "step": 43970
+    },
+    {
+      "epoch": 1.40736,
+      "grad_norm": 44.62795639038086,
+      "learning_rate": 6.58488888888889e-06,
+      "loss": 1.569,
+      "step": 43980
+    },
+    {
+      "epoch": 1.40768,
+      "grad_norm": 45.75224304199219,
+      "learning_rate": 6.581333333333334e-06,
+      "loss": 1.6054,
+      "step": 43990
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 45.715789794921875,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 1.6067,
+      "step": 44000
+    },
+    {
+      "epoch": 1.40832,
+      "grad_norm": 44.8160285949707,
+      "learning_rate": 6.574222222222223e-06,
+      "loss": 1.6198,
+      "step": 44010
+    },
+    {
+      "epoch": 1.4086400000000001,
+      "grad_norm": 46.13142395019531,
+      "learning_rate": 6.570666666666667e-06,
+      "loss": 1.6129,
+      "step": 44020
+    },
+    {
+      "epoch": 1.40896,
+      "grad_norm": 45.89512634277344,
+      "learning_rate": 6.567111111111112e-06,
+      "loss": 1.6048,
+      "step": 44030
+    },
+    {
+      "epoch": 1.4092799999999999,
+      "grad_norm": 48.633846282958984,
+      "learning_rate": 6.563555555555556e-06,
+      "loss": 1.6084,
+      "step": 44040
+    },
+    {
+      "epoch": 1.4096,
+      "grad_norm": 45.565189361572266,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 1.5784,
+      "step": 44050
+    },
+    {
+      "epoch": 1.40992,
+      "grad_norm": 44.69120407104492,
+      "learning_rate": 6.556444444444445e-06,
+      "loss": 1.5901,
+      "step": 44060
+    },
+    {
+      "epoch": 1.41024,
+      "grad_norm": 44.379817962646484,
+      "learning_rate": 6.55288888888889e-06,
+      "loss": 1.5949,
+      "step": 44070
+    },
+    {
+      "epoch": 1.41056,
+      "grad_norm": 47.116477966308594,
+      "learning_rate": 6.549333333333334e-06,
+      "loss": 1.6227,
+      "step": 44080
+    },
+    {
+      "epoch": 1.4108800000000001,
+      "grad_norm": 44.753299713134766,
+      "learning_rate": 6.545777777777779e-06,
+      "loss": 1.5981,
+      "step": 44090
+    },
+    {
+      "epoch": 1.4112,
+      "grad_norm": 44.2072639465332,
+      "learning_rate": 6.5422222222222235e-06,
+      "loss": 1.6162,
+      "step": 44100
+    },
+    {
+      "epoch": 1.4115199999999999,
+      "grad_norm": 47.168739318847656,
+      "learning_rate": 6.538666666666667e-06,
+      "loss": 1.5781,
+      "step": 44110
+    },
+    {
+      "epoch": 1.41184,
+      "grad_norm": 47.949363708496094,
+      "learning_rate": 6.535111111111112e-06,
+      "loss": 1.6039,
+      "step": 44120
+    },
+    {
+      "epoch": 1.41216,
+      "grad_norm": 45.053653717041016,
+      "learning_rate": 6.531555555555556e-06,
+      "loss": 1.6081,
+      "step": 44130
+    },
+    {
+      "epoch": 1.41248,
+      "grad_norm": 44.90990447998047,
+      "learning_rate": 6.528000000000001e-06,
+      "loss": 1.5493,
+      "step": 44140
+    },
+    {
+      "epoch": 1.4128,
+      "grad_norm": 46.25327682495117,
+      "learning_rate": 6.524444444444445e-06,
+      "loss": 1.6253,
+      "step": 44150
+    },
+    {
+      "epoch": 1.41312,
+      "grad_norm": 43.844017028808594,
+      "learning_rate": 6.52088888888889e-06,
+      "loss": 1.5917,
+      "step": 44160
+    },
+    {
+      "epoch": 1.41344,
+      "grad_norm": 44.53428268432617,
+      "learning_rate": 6.5173333333333345e-06,
+      "loss": 1.5935,
+      "step": 44170
+    },
+    {
+      "epoch": 1.41376,
+      "grad_norm": 46.895931243896484,
+      "learning_rate": 6.513777777777779e-06,
+      "loss": 1.6101,
+      "step": 44180
+    },
+    {
+      "epoch": 1.41408,
+      "grad_norm": 45.996299743652344,
+      "learning_rate": 6.510222222222222e-06,
+      "loss": 1.6097,
+      "step": 44190
+    },
+    {
+      "epoch": 1.4144,
+      "grad_norm": 45.133392333984375,
+      "learning_rate": 6.5066666666666665e-06,
+      "loss": 1.615,
+      "step": 44200
+    },
+    {
+      "epoch": 1.41472,
+      "grad_norm": 44.23000717163086,
+      "learning_rate": 6.503111111111111e-06,
+      "loss": 1.5988,
+      "step": 44210
+    },
+    {
+      "epoch": 1.41504,
+      "grad_norm": 44.77185821533203,
+      "learning_rate": 6.499555555555556e-06,
+      "loss": 1.5761,
+      "step": 44220
+    },
+    {
+      "epoch": 1.41536,
+      "grad_norm": 43.65971374511719,
+      "learning_rate": 6.496e-06,
+      "loss": 1.5773,
+      "step": 44230
+    },
+    {
+      "epoch": 1.41568,
+      "grad_norm": 44.112850189208984,
+      "learning_rate": 6.492444444444445e-06,
+      "loss": 1.6143,
+      "step": 44240
+    },
+    {
+      "epoch": 1.416,
+      "grad_norm": 44.179534912109375,
+      "learning_rate": 6.488888888888889e-06,
+      "loss": 1.6028,
+      "step": 44250
+    },
+    {
+      "epoch": 1.41632,
+      "grad_norm": 44.71501922607422,
+      "learning_rate": 6.485333333333334e-06,
+      "loss": 1.6185,
+      "step": 44260
+    },
+    {
+      "epoch": 1.4166400000000001,
+      "grad_norm": 44.295318603515625,
+      "learning_rate": 6.481777777777778e-06,
+      "loss": 1.575,
+      "step": 44270
+    },
+    {
+      "epoch": 1.41696,
+      "grad_norm": 43.224884033203125,
+      "learning_rate": 6.478222222222222e-06,
+      "loss": 1.5957,
+      "step": 44280
+    },
+    {
+      "epoch": 1.4172799999999999,
+      "grad_norm": 43.866905212402344,
+      "learning_rate": 6.474666666666667e-06,
+      "loss": 1.6108,
+      "step": 44290
+    },
+    {
+      "epoch": 1.4176,
+      "grad_norm": 46.90814208984375,
+      "learning_rate": 6.471111111111111e-06,
+      "loss": 1.609,
+      "step": 44300
+    },
+    {
+      "epoch": 1.41792,
+      "grad_norm": 45.48236083984375,
+      "learning_rate": 6.467555555555556e-06,
+      "loss": 1.5964,
+      "step": 44310
+    },
+    {
+      "epoch": 1.41824,
+      "grad_norm": 46.96796417236328,
+      "learning_rate": 6.464e-06,
+      "loss": 1.6068,
+      "step": 44320
+    },
+    {
+      "epoch": 1.41856,
+      "grad_norm": 43.96994400024414,
+      "learning_rate": 6.460444444444445e-06,
+      "loss": 1.5838,
+      "step": 44330
+    },
+    {
+      "epoch": 1.41888,
+      "grad_norm": 45.90093231201172,
+      "learning_rate": 6.456888888888889e-06,
+      "loss": 1.611,
+      "step": 44340
+    },
+    {
+      "epoch": 1.4192,
+      "grad_norm": 45.67726516723633,
+      "learning_rate": 6.453333333333334e-06,
+      "loss": 1.6252,
+      "step": 44350
+    },
+    {
+      "epoch": 1.41952,
+      "grad_norm": 46.475929260253906,
+      "learning_rate": 6.4497777777777785e-06,
+      "loss": 1.5988,
+      "step": 44360
+    },
+    {
+      "epoch": 1.41984,
+      "grad_norm": 46.8716926574707,
+      "learning_rate": 6.446222222222222e-06,
+      "loss": 1.6138,
+      "step": 44370
+    },
+    {
+      "epoch": 1.42016,
+      "grad_norm": 44.12192916870117,
+      "learning_rate": 6.442666666666667e-06,
+      "loss": 1.6223,
+      "step": 44380
+    },
+    {
+      "epoch": 1.42048,
+      "grad_norm": 43.43100357055664,
+      "learning_rate": 6.439111111111111e-06,
+      "loss": 1.6221,
+      "step": 44390
+    },
+    {
+      "epoch": 1.4208,
+      "grad_norm": 45.15028381347656,
+      "learning_rate": 6.435555555555556e-06,
+      "loss": 1.6043,
+      "step": 44400
+    },
+    {
+      "epoch": 1.42112,
+      "grad_norm": 46.52452087402344,
+      "learning_rate": 6.432e-06,
+      "loss": 1.6164,
+      "step": 44410
+    },
+    {
+      "epoch": 1.42144,
+      "grad_norm": 45.797088623046875,
+      "learning_rate": 6.428444444444445e-06,
+      "loss": 1.5884,
+      "step": 44420
+    },
+    {
+      "epoch": 1.42176,
+      "grad_norm": 46.37058639526367,
+      "learning_rate": 6.4248888888888895e-06,
+      "loss": 1.567,
+      "step": 44430
+    },
+    {
+      "epoch": 1.42208,
+      "grad_norm": 43.526065826416016,
+      "learning_rate": 6.421333333333334e-06,
+      "loss": 1.6117,
+      "step": 44440
+    },
+    {
+      "epoch": 1.4224,
+      "grad_norm": 43.67570114135742,
+      "learning_rate": 6.417777777777779e-06,
+      "loss": 1.6155,
+      "step": 44450
+    },
+    {
+      "epoch": 1.42272,
+      "grad_norm": 44.37187576293945,
+      "learning_rate": 6.414222222222223e-06,
+      "loss": 1.594,
+      "step": 44460
+    },
+    {
+      "epoch": 1.42304,
+      "grad_norm": 45.66731262207031,
+      "learning_rate": 6.410666666666667e-06,
+      "loss": 1.6081,
+      "step": 44470
+    },
+    {
+      "epoch": 1.42336,
+      "grad_norm": 46.02083206176758,
+      "learning_rate": 6.4071111111111114e-06,
+      "loss": 1.6136,
+      "step": 44480
+    },
+    {
+      "epoch": 1.42368,
+      "grad_norm": 44.25553894042969,
+      "learning_rate": 6.403555555555556e-06,
+      "loss": 1.6016,
+      "step": 44490
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 44.574031829833984,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.6203,
+      "step": 44500
+    },
+    {
+      "epoch": 1.42432,
+      "grad_norm": 45.1065788269043,
+      "learning_rate": 6.396444444444445e-06,
+      "loss": 1.6158,
+      "step": 44510
+    },
+    {
+      "epoch": 1.4246400000000001,
+      "grad_norm": 45.92455291748047,
+      "learning_rate": 6.39288888888889e-06,
+      "loss": 1.6105,
+      "step": 44520
+    },
+    {
+      "epoch": 1.42496,
+      "grad_norm": 45.41129684448242,
+      "learning_rate": 6.389333333333334e-06,
+      "loss": 1.6069,
+      "step": 44530
+    },
+    {
+      "epoch": 1.4252799999999999,
+      "grad_norm": 45.73555374145508,
+      "learning_rate": 6.385777777777779e-06,
+      "loss": 1.6026,
+      "step": 44540
+    },
+    {
+      "epoch": 1.4256,
+      "grad_norm": 44.59724426269531,
+      "learning_rate": 6.382222222222223e-06,
+      "loss": 1.6014,
+      "step": 44550
+    },
+    {
+      "epoch": 1.42592,
+      "grad_norm": 46.82771301269531,
+      "learning_rate": 6.378666666666667e-06,
+      "loss": 1.6057,
+      "step": 44560
+    },
+    {
+      "epoch": 1.42624,
+      "grad_norm": 43.83695602416992,
+      "learning_rate": 6.375111111111112e-06,
+      "loss": 1.6141,
+      "step": 44570
+    },
+    {
+      "epoch": 1.42656,
+      "grad_norm": 42.86970138549805,
+      "learning_rate": 6.371555555555556e-06,
+      "loss": 1.6237,
+      "step": 44580
+    },
+    {
+      "epoch": 1.42688,
+      "grad_norm": 43.6492919921875,
+      "learning_rate": 6.368000000000001e-06,
+      "loss": 1.6163,
+      "step": 44590
+    },
+    {
+      "epoch": 1.4272,
+      "grad_norm": 42.75118637084961,
+      "learning_rate": 6.364444444444445e-06,
+      "loss": 1.5957,
+      "step": 44600
+    },
+    {
+      "epoch": 1.42752,
+      "grad_norm": 45.4508056640625,
+      "learning_rate": 6.36088888888889e-06,
+      "loss": 1.5855,
+      "step": 44610
+    },
+    {
+      "epoch": 1.42784,
+      "grad_norm": 45.386810302734375,
+      "learning_rate": 6.357333333333334e-06,
+      "loss": 1.5955,
+      "step": 44620
+    },
+    {
+      "epoch": 1.42816,
+      "grad_norm": 47.42978286743164,
+      "learning_rate": 6.353777777777779e-06,
+      "loss": 1.6036,
+      "step": 44630
+    },
+    {
+      "epoch": 1.42848,
+      "grad_norm": 44.261627197265625,
+      "learning_rate": 6.3502222222222235e-06,
+      "loss": 1.589,
+      "step": 44640
+    },
+    {
+      "epoch": 1.4288,
+      "grad_norm": 42.385276794433594,
+      "learning_rate": 6.346666666666668e-06,
+      "loss": 1.6108,
+      "step": 44650
+    },
+    {
+      "epoch": 1.42912,
+      "grad_norm": 47.136287689208984,
+      "learning_rate": 6.343111111111112e-06,
+      "loss": 1.6159,
+      "step": 44660
+    },
+    {
+      "epoch": 1.42944,
+      "grad_norm": 47.89680480957031,
+      "learning_rate": 6.339555555555556e-06,
+      "loss": 1.6096,
+      "step": 44670
+    },
+    {
+      "epoch": 1.42976,
+      "grad_norm": 43.50481414794922,
+      "learning_rate": 6.336000000000001e-06,
+      "loss": 1.603,
+      "step": 44680
+    },
+    {
+      "epoch": 1.43008,
+      "grad_norm": 44.940738677978516,
+      "learning_rate": 6.332444444444445e-06,
+      "loss": 1.605,
+      "step": 44690
+    },
+    {
+      "epoch": 1.4304000000000001,
+      "grad_norm": 45.36077117919922,
+      "learning_rate": 6.328888888888889e-06,
+      "loss": 1.5932,
+      "step": 44700
+    },
+    {
+      "epoch": 1.43072,
+      "grad_norm": 44.730690002441406,
+      "learning_rate": 6.325333333333334e-06,
+      "loss": 1.6017,
+      "step": 44710
+    },
+    {
+      "epoch": 1.4310399999999999,
+      "grad_norm": 43.576080322265625,
+      "learning_rate": 6.321777777777778e-06,
+      "loss": 1.6213,
+      "step": 44720
+    },
+    {
+      "epoch": 1.43136,
+      "grad_norm": 44.608856201171875,
+      "learning_rate": 6.318222222222222e-06,
+      "loss": 1.5938,
+      "step": 44730
+    },
+    {
+      "epoch": 1.43168,
+      "grad_norm": 47.434425354003906,
+      "learning_rate": 6.3146666666666665e-06,
+      "loss": 1.6161,
+      "step": 44740
+    },
+    {
+      "epoch": 1.432,
+      "grad_norm": 44.625858306884766,
+      "learning_rate": 6.311111111111111e-06,
+      "loss": 1.6059,
+      "step": 44750
+    },
+    {
+      "epoch": 1.43232,
+      "grad_norm": 44.30402374267578,
+      "learning_rate": 6.3075555555555556e-06,
+      "loss": 1.5754,
+      "step": 44760
+    },
+    {
+      "epoch": 1.4326400000000001,
+      "grad_norm": 44.90681838989258,
+      "learning_rate": 6.304e-06,
+      "loss": 1.5993,
+      "step": 44770
+    },
+    {
+      "epoch": 1.43296,
+      "grad_norm": 42.78556442260742,
+      "learning_rate": 6.300444444444445e-06,
+      "loss": 1.5762,
+      "step": 44780
+    },
+    {
+      "epoch": 1.4332799999999999,
+      "grad_norm": 44.31805419921875,
+      "learning_rate": 6.296888888888889e-06,
+      "loss": 1.5601,
+      "step": 44790
+    },
+    {
+      "epoch": 1.4336,
+      "grad_norm": 46.09400939941406,
+      "learning_rate": 6.293333333333334e-06,
+      "loss": 1.5807,
+      "step": 44800
+    },
+    {
+      "epoch": 1.43392,
+      "grad_norm": 43.824424743652344,
+      "learning_rate": 6.289777777777778e-06,
+      "loss": 1.5881,
+      "step": 44810
+    },
+    {
+      "epoch": 1.43424,
+      "grad_norm": 44.83427047729492,
+      "learning_rate": 6.286222222222222e-06,
+      "loss": 1.6233,
+      "step": 44820
+    },
+    {
+      "epoch": 1.43456,
+      "grad_norm": 45.876094818115234,
+      "learning_rate": 6.282666666666667e-06,
+      "loss": 1.6087,
+      "step": 44830
+    },
+    {
+      "epoch": 1.43488,
+      "grad_norm": 48.019325256347656,
+      "learning_rate": 6.279111111111111e-06,
+      "loss": 1.6231,
+      "step": 44840
+    },
+    {
+      "epoch": 1.4352,
+      "grad_norm": 44.39468765258789,
+      "learning_rate": 6.275555555555556e-06,
+      "loss": 1.6105,
+      "step": 44850
+    },
+    {
+      "epoch": 1.43552,
+      "grad_norm": 43.1924934387207,
+      "learning_rate": 6.272e-06,
+      "loss": 1.6201,
+      "step": 44860
+    },
+    {
+      "epoch": 1.43584,
+      "grad_norm": 45.5162239074707,
+      "learning_rate": 6.268444444444445e-06,
+      "loss": 1.6036,
+      "step": 44870
+    },
+    {
+      "epoch": 1.43616,
+      "grad_norm": 47.22616958618164,
+      "learning_rate": 6.264888888888889e-06,
+      "loss": 1.6227,
+      "step": 44880
+    },
+    {
+      "epoch": 1.43648,
+      "grad_norm": 45.4078483581543,
+      "learning_rate": 6.261333333333334e-06,
+      "loss": 1.5907,
+      "step": 44890
+    },
+    {
+      "epoch": 1.4368,
+      "grad_norm": 48.54788589477539,
+      "learning_rate": 6.2577777777777785e-06,
+      "loss": 1.5929,
+      "step": 44900
+    },
+    {
+      "epoch": 1.43712,
+      "grad_norm": 46.870609283447266,
+      "learning_rate": 6.254222222222223e-06,
+      "loss": 1.5986,
+      "step": 44910
+    },
+    {
+      "epoch": 1.43744,
+      "grad_norm": 46.138065338134766,
+      "learning_rate": 6.250666666666667e-06,
+      "loss": 1.6462,
+      "step": 44920
+    },
+    {
+      "epoch": 1.43776,
+      "grad_norm": 49.69923400878906,
+      "learning_rate": 6.247111111111111e-06,
+      "loss": 1.63,
+      "step": 44930
+    },
+    {
+      "epoch": 1.43808,
+      "grad_norm": 44.56926345825195,
+      "learning_rate": 6.243555555555556e-06,
+      "loss": 1.6069,
+      "step": 44940
+    },
+    {
+      "epoch": 1.4384000000000001,
+      "grad_norm": 46.90085220336914,
+      "learning_rate": 6.24e-06,
+      "loss": 1.6143,
+      "step": 44950
+    },
+    {
+      "epoch": 1.43872,
+      "grad_norm": 45.12166213989258,
+      "learning_rate": 6.236444444444445e-06,
+      "loss": 1.6029,
+      "step": 44960
+    },
+    {
+      "epoch": 1.4390399999999999,
+      "grad_norm": 45.84695053100586,
+      "learning_rate": 6.2328888888888895e-06,
+      "loss": 1.5996,
+      "step": 44970
+    },
+    {
+      "epoch": 1.43936,
+      "grad_norm": 46.82448196411133,
+      "learning_rate": 6.229333333333334e-06,
+      "loss": 1.609,
+      "step": 44980
+    },
+    {
+      "epoch": 1.43968,
+      "grad_norm": 44.80758285522461,
+      "learning_rate": 6.225777777777779e-06,
+      "loss": 1.5981,
+      "step": 44990
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 44.90853500366211,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 1.5888,
+      "step": 45000
+    },
+    {
+      "epoch": 1.44032,
+      "grad_norm": 45.14985275268555,
+      "learning_rate": 6.218666666666668e-06,
+      "loss": 1.6153,
+      "step": 45010
+    },
+    {
+      "epoch": 1.44064,
+      "grad_norm": 46.903263092041016,
+      "learning_rate": 6.2151111111111114e-06,
+      "loss": 1.6259,
+      "step": 45020
+    },
+    {
+      "epoch": 1.44096,
+      "grad_norm": 46.76694869995117,
+      "learning_rate": 6.211555555555556e-06,
+      "loss": 1.6259,
+      "step": 45030
+    },
+    {
+      "epoch": 1.44128,
+      "grad_norm": 45.14419174194336,
+      "learning_rate": 6.2080000000000005e-06,
+      "loss": 1.6069,
+      "step": 45040
+    },
+    {
+      "epoch": 1.4416,
+      "grad_norm": 43.43990707397461,
+      "learning_rate": 6.204444444444445e-06,
+      "loss": 1.5894,
+      "step": 45050
+    },
+    {
+      "epoch": 1.44192,
+      "grad_norm": 46.003562927246094,
+      "learning_rate": 6.20088888888889e-06,
+      "loss": 1.6213,
+      "step": 45060
+    },
+    {
+      "epoch": 1.44224,
+      "grad_norm": 45.48784637451172,
+      "learning_rate": 6.197333333333334e-06,
+      "loss": 1.5838,
+      "step": 45070
+    },
+    {
+      "epoch": 1.44256,
+      "grad_norm": 45.909122467041016,
+      "learning_rate": 6.193777777777779e-06,
+      "loss": 1.579,
+      "step": 45080
+    },
+    {
+      "epoch": 1.44288,
+      "grad_norm": 43.655574798583984,
+      "learning_rate": 6.190222222222223e-06,
+      "loss": 1.6093,
+      "step": 45090
+    },
+    {
+      "epoch": 1.4432,
+      "grad_norm": 45.166168212890625,
+      "learning_rate": 6.186666666666668e-06,
+      "loss": 1.6342,
+      "step": 45100
+    },
+    {
+      "epoch": 1.44352,
+      "grad_norm": 45.9638671875,
+      "learning_rate": 6.1831111111111116e-06,
+      "loss": 1.6118,
+      "step": 45110
+    },
+    {
+      "epoch": 1.44384,
+      "grad_norm": 45.01591873168945,
+      "learning_rate": 6.179555555555556e-06,
+      "loss": 1.6021,
+      "step": 45120
+    },
+    {
+      "epoch": 1.44416,
+      "grad_norm": 46.7148323059082,
+      "learning_rate": 6.176000000000001e-06,
+      "loss": 1.5778,
+      "step": 45130
+    },
+    {
+      "epoch": 1.44448,
+      "grad_norm": 47.17043685913086,
+      "learning_rate": 6.172444444444445e-06,
+      "loss": 1.5896,
+      "step": 45140
+    },
+    {
+      "epoch": 1.4447999999999999,
+      "grad_norm": 42.31821823120117,
+      "learning_rate": 6.16888888888889e-06,
+      "loss": 1.5922,
+      "step": 45150
+    },
+    {
+      "epoch": 1.44512,
+      "grad_norm": 47.12434387207031,
+      "learning_rate": 6.165333333333334e-06,
+      "loss": 1.5817,
+      "step": 45160
+    },
+    {
+      "epoch": 1.44544,
+      "grad_norm": 43.81653594970703,
+      "learning_rate": 6.161777777777779e-06,
+      "loss": 1.6213,
+      "step": 45170
+    },
+    {
+      "epoch": 1.44576,
+      "grad_norm": 47.70396041870117,
+      "learning_rate": 6.1582222222222234e-06,
+      "loss": 1.6112,
+      "step": 45180
+    },
+    {
+      "epoch": 1.44608,
+      "grad_norm": 46.75013732910156,
+      "learning_rate": 6.154666666666668e-06,
+      "loss": 1.6432,
+      "step": 45190
+    },
+    {
+      "epoch": 1.4464000000000001,
+      "grad_norm": 43.111427307128906,
+      "learning_rate": 6.1511111111111125e-06,
+      "loss": 1.5941,
+      "step": 45200
+    },
+    {
+      "epoch": 1.44672,
+      "grad_norm": 43.45669174194336,
+      "learning_rate": 6.147555555555556e-06,
+      "loss": 1.6131,
+      "step": 45210
+    },
+    {
+      "epoch": 1.4470399999999999,
+      "grad_norm": 44.20582962036133,
+      "learning_rate": 6.144e-06,
+      "loss": 1.6047,
+      "step": 45220
+    },
+    {
+      "epoch": 1.44736,
+      "grad_norm": 47.43238830566406,
+      "learning_rate": 6.1404444444444445e-06,
+      "loss": 1.5997,
+      "step": 45230
+    },
+    {
+      "epoch": 1.44768,
+      "grad_norm": 45.39116668701172,
+      "learning_rate": 6.136888888888889e-06,
+      "loss": 1.5715,
+      "step": 45240
+    },
+    {
+      "epoch": 1.448,
+      "grad_norm": 47.27134704589844,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.6172,
+      "step": 45250
+    },
+    {
+      "epoch": 1.44832,
+      "grad_norm": 45.037811279296875,
+      "learning_rate": 6.129777777777778e-06,
+      "loss": 1.6083,
+      "step": 45260
+    },
+    {
+      "epoch": 1.44864,
+      "grad_norm": 46.992244720458984,
+      "learning_rate": 6.126222222222222e-06,
+      "loss": 1.6296,
+      "step": 45270
+    },
+    {
+      "epoch": 1.44896,
+      "grad_norm": 44.33993911743164,
+      "learning_rate": 6.1226666666666664e-06,
+      "loss": 1.6257,
+      "step": 45280
+    },
+    {
+      "epoch": 1.44928,
+      "grad_norm": 45.206356048583984,
+      "learning_rate": 6.119111111111111e-06,
+      "loss": 1.6328,
+      "step": 45290
+    },
+    {
+      "epoch": 1.4496,
+      "grad_norm": 43.93304443359375,
+      "learning_rate": 6.1155555555555555e-06,
+      "loss": 1.6025,
+      "step": 45300
+    },
+    {
+      "epoch": 1.44992,
+      "grad_norm": 44.74345397949219,
+      "learning_rate": 6.112e-06,
+      "loss": 1.5956,
+      "step": 45310
+    },
+    {
+      "epoch": 1.45024,
+      "grad_norm": 44.77165603637695,
+      "learning_rate": 6.108444444444445e-06,
+      "loss": 1.612,
+      "step": 45320
+    },
+    {
+      "epoch": 1.45056,
+      "grad_norm": 43.939292907714844,
+      "learning_rate": 6.104888888888889e-06,
+      "loss": 1.6043,
+      "step": 45330
+    },
+    {
+      "epoch": 1.45088,
+      "grad_norm": 45.1342658996582,
+      "learning_rate": 6.101333333333334e-06,
+      "loss": 1.6139,
+      "step": 45340
+    },
+    {
+      "epoch": 1.4512,
+      "grad_norm": 43.455413818359375,
+      "learning_rate": 6.097777777777778e-06,
+      "loss": 1.6196,
+      "step": 45350
+    },
+    {
+      "epoch": 1.45152,
+      "grad_norm": 45.01537322998047,
+      "learning_rate": 6.094222222222223e-06,
+      "loss": 1.6553,
+      "step": 45360
+    },
+    {
+      "epoch": 1.45184,
+      "grad_norm": 45.8966178894043,
+      "learning_rate": 6.0906666666666666e-06,
+      "loss": 1.6301,
+      "step": 45370
+    },
+    {
+      "epoch": 1.4521600000000001,
+      "grad_norm": 44.561424255371094,
+      "learning_rate": 6.087111111111111e-06,
+      "loss": 1.6058,
+      "step": 45380
+    },
+    {
+      "epoch": 1.45248,
+      "grad_norm": 43.578922271728516,
+      "learning_rate": 6.083555555555556e-06,
+      "loss": 1.5702,
+      "step": 45390
+    },
+    {
+      "epoch": 1.4527999999999999,
+      "grad_norm": 44.27152633666992,
+      "learning_rate": 6.08e-06,
+      "loss": 1.6097,
+      "step": 45400
+    },
+    {
+      "epoch": 1.45312,
+      "grad_norm": 44.5775032043457,
+      "learning_rate": 6.076444444444445e-06,
+      "loss": 1.5524,
+      "step": 45410
+    },
+    {
+      "epoch": 1.45344,
+      "grad_norm": 47.667057037353516,
+      "learning_rate": 6.072888888888889e-06,
+      "loss": 1.6013,
+      "step": 45420
+    },
+    {
+      "epoch": 1.45376,
+      "grad_norm": 45.95570373535156,
+      "learning_rate": 6.069333333333334e-06,
+      "loss": 1.5831,
+      "step": 45430
+    },
+    {
+      "epoch": 1.45408,
+      "grad_norm": 47.346534729003906,
+      "learning_rate": 6.0657777777777784e-06,
+      "loss": 1.588,
+      "step": 45440
+    },
+    {
+      "epoch": 1.4544000000000001,
+      "grad_norm": 43.89128875732422,
+      "learning_rate": 6.062222222222223e-06,
+      "loss": 1.5948,
+      "step": 45450
+    },
+    {
+      "epoch": 1.45472,
+      "grad_norm": 45.23402786254883,
+      "learning_rate": 6.0586666666666676e-06,
+      "loss": 1.6128,
+      "step": 45460
+    },
+    {
+      "epoch": 1.45504,
+      "grad_norm": 44.87834930419922,
+      "learning_rate": 6.055111111111111e-06,
+      "loss": 1.5972,
+      "step": 45470
+    },
+    {
+      "epoch": 1.45536,
+      "grad_norm": 44.89365768432617,
+      "learning_rate": 6.051555555555556e-06,
+      "loss": 1.5834,
+      "step": 45480
+    },
+    {
+      "epoch": 1.45568,
+      "grad_norm": 47.11971664428711,
+      "learning_rate": 6.048e-06,
+      "loss": 1.5941,
+      "step": 45490
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 44.76010513305664,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 1.6244,
+      "step": 45500
+    },
+    {
+      "epoch": 1.45632,
+      "grad_norm": 42.880149841308594,
+      "learning_rate": 6.0408888888888895e-06,
+      "loss": 1.5966,
+      "step": 45510
+    },
+    {
+      "epoch": 1.45664,
+      "grad_norm": 46.598670959472656,
+      "learning_rate": 6.037333333333334e-06,
+      "loss": 1.5942,
+      "step": 45520
+    },
+    {
+      "epoch": 1.45696,
+      "grad_norm": 45.492271423339844,
+      "learning_rate": 6.033777777777779e-06,
+      "loss": 1.5833,
+      "step": 45530
+    },
+    {
+      "epoch": 1.45728,
+      "grad_norm": 47.63541793823242,
+      "learning_rate": 6.030222222222223e-06,
+      "loss": 1.5963,
+      "step": 45540
+    },
+    {
+      "epoch": 1.4576,
+      "grad_norm": 45.30315017700195,
+      "learning_rate": 6.026666666666668e-06,
+      "loss": 1.5825,
+      "step": 45550
+    },
+    {
+      "epoch": 1.45792,
+      "grad_norm": 45.10110855102539,
+      "learning_rate": 6.023111111111111e-06,
+      "loss": 1.5897,
+      "step": 45560
+    },
+    {
+      "epoch": 1.45824,
+      "grad_norm": 45.75407028198242,
+      "learning_rate": 6.019555555555556e-06,
+      "loss": 1.6144,
+      "step": 45570
+    },
+    {
+      "epoch": 1.45856,
+      "grad_norm": 46.492149353027344,
+      "learning_rate": 6.0160000000000005e-06,
+      "loss": 1.6031,
+      "step": 45580
+    },
+    {
+      "epoch": 1.45888,
+      "grad_norm": 44.73735809326172,
+      "learning_rate": 6.012444444444445e-06,
+      "loss": 1.5914,
+      "step": 45590
+    },
+    {
+      "epoch": 1.4592,
+      "grad_norm": 45.741764068603516,
+      "learning_rate": 6.00888888888889e-06,
+      "loss": 1.6092,
+      "step": 45600
+    },
+    {
+      "epoch": 1.45952,
+      "grad_norm": 46.8515510559082,
+      "learning_rate": 6.005333333333334e-06,
+      "loss": 1.6302,
+      "step": 45610
+    },
+    {
+      "epoch": 1.45984,
+      "grad_norm": 44.54631805419922,
+      "learning_rate": 6.001777777777779e-06,
+      "loss": 1.6044,
+      "step": 45620
+    },
+    {
+      "epoch": 1.4601600000000001,
+      "grad_norm": 45.480140686035156,
+      "learning_rate": 5.998222222222223e-06,
+      "loss": 1.6075,
+      "step": 45630
+    },
+    {
+      "epoch": 1.46048,
+      "grad_norm": 45.182369232177734,
+      "learning_rate": 5.994666666666668e-06,
+      "loss": 1.5839,
+      "step": 45640
+    },
+    {
+      "epoch": 1.4607999999999999,
+      "grad_norm": 46.069602966308594,
+      "learning_rate": 5.991111111111112e-06,
+      "loss": 1.6244,
+      "step": 45650
+    },
+    {
+      "epoch": 1.46112,
+      "grad_norm": 45.65216827392578,
+      "learning_rate": 5.987555555555556e-06,
+      "loss": 1.6179,
+      "step": 45660
+    },
+    {
+      "epoch": 1.46144,
+      "grad_norm": 44.958251953125,
+      "learning_rate": 5.984000000000001e-06,
+      "loss": 1.6392,
+      "step": 45670
+    },
+    {
+      "epoch": 1.46176,
+      "grad_norm": 46.477901458740234,
+      "learning_rate": 5.980444444444445e-06,
+      "loss": 1.6115,
+      "step": 45680
+    },
+    {
+      "epoch": 1.46208,
+      "grad_norm": 47.5823974609375,
+      "learning_rate": 5.97688888888889e-06,
+      "loss": 1.6286,
+      "step": 45690
+    },
+    {
+      "epoch": 1.4624,
+      "grad_norm": 46.62644958496094,
+      "learning_rate": 5.973333333333334e-06,
+      "loss": 1.5928,
+      "step": 45700
+    },
+    {
+      "epoch": 1.46272,
+      "grad_norm": 45.06728744506836,
+      "learning_rate": 5.969777777777779e-06,
+      "loss": 1.6042,
+      "step": 45710
+    },
+    {
+      "epoch": 1.46304,
+      "grad_norm": 43.426483154296875,
+      "learning_rate": 5.966222222222223e-06,
+      "loss": 1.5828,
+      "step": 45720
+    },
+    {
+      "epoch": 1.46336,
+      "grad_norm": 47.46270751953125,
+      "learning_rate": 5.962666666666666e-06,
+      "loss": 1.5998,
+      "step": 45730
+    },
+    {
+      "epoch": 1.46368,
+      "grad_norm": 45.45689392089844,
+      "learning_rate": 5.959111111111111e-06,
+      "loss": 1.5961,
+      "step": 45740
+    },
+    {
+      "epoch": 1.464,
+      "grad_norm": 45.77447509765625,
+      "learning_rate": 5.955555555555555e-06,
+      "loss": 1.612,
+      "step": 45750
+    },
+    {
+      "epoch": 1.46432,
+      "grad_norm": 44.932464599609375,
+      "learning_rate": 5.952e-06,
+      "loss": 1.6301,
+      "step": 45760
+    },
+    {
+      "epoch": 1.46464,
+      "grad_norm": 43.912498474121094,
+      "learning_rate": 5.9484444444444445e-06,
+      "loss": 1.6006,
+      "step": 45770
+    },
+    {
+      "epoch": 1.46496,
+      "grad_norm": 44.95744705200195,
+      "learning_rate": 5.944888888888889e-06,
+      "loss": 1.5926,
+      "step": 45780
+    },
+    {
+      "epoch": 1.46528,
+      "grad_norm": 45.56084060668945,
+      "learning_rate": 5.941333333333334e-06,
+      "loss": 1.5965,
+      "step": 45790
+    },
+    {
+      "epoch": 1.4656,
+      "grad_norm": 45.82440185546875,
+      "learning_rate": 5.937777777777778e-06,
+      "loss": 1.6084,
+      "step": 45800
+    },
+    {
+      "epoch": 1.4659200000000001,
+      "grad_norm": 45.97123718261719,
+      "learning_rate": 5.934222222222223e-06,
+      "loss": 1.6145,
+      "step": 45810
+    },
+    {
+      "epoch": 1.46624,
+      "grad_norm": 42.941612243652344,
+      "learning_rate": 5.930666666666666e-06,
+      "loss": 1.5818,
+      "step": 45820
+    },
+    {
+      "epoch": 1.4665599999999999,
+      "grad_norm": 47.44602966308594,
+      "learning_rate": 5.927111111111111e-06,
+      "loss": 1.5927,
+      "step": 45830
+    },
+    {
+      "epoch": 1.46688,
+      "grad_norm": 44.84784698486328,
+      "learning_rate": 5.9235555555555555e-06,
+      "loss": 1.5655,
+      "step": 45840
+    },
+    {
+      "epoch": 1.4672,
+      "grad_norm": 44.033390045166016,
+      "learning_rate": 5.92e-06,
+      "loss": 1.5958,
+      "step": 45850
+    },
+    {
+      "epoch": 1.46752,
+      "grad_norm": 42.78225326538086,
+      "learning_rate": 5.916444444444445e-06,
+      "loss": 1.6028,
+      "step": 45860
+    },
+    {
+      "epoch": 1.46784,
+      "grad_norm": 42.93524932861328,
+      "learning_rate": 5.912888888888889e-06,
+      "loss": 1.5768,
+      "step": 45870
+    },
+    {
+      "epoch": 1.4681600000000001,
+      "grad_norm": 47.09729766845703,
+      "learning_rate": 5.909333333333334e-06,
+      "loss": 1.6107,
+      "step": 45880
+    },
+    {
+      "epoch": 1.46848,
+      "grad_norm": 45.86977005004883,
+      "learning_rate": 5.905777777777778e-06,
+      "loss": 1.6132,
+      "step": 45890
+    },
+    {
+      "epoch": 1.4687999999999999,
+      "grad_norm": 46.18256759643555,
+      "learning_rate": 5.902222222222223e-06,
+      "loss": 1.6015,
+      "step": 45900
+    },
+    {
+      "epoch": 1.46912,
+      "grad_norm": 44.45927810668945,
+      "learning_rate": 5.898666666666667e-06,
+      "loss": 1.604,
+      "step": 45910
+    },
+    {
+      "epoch": 1.46944,
+      "grad_norm": 46.4095344543457,
+      "learning_rate": 5.895111111111111e-06,
+      "loss": 1.6144,
+      "step": 45920
+    },
+    {
+      "epoch": 1.46976,
+      "grad_norm": 47.12724304199219,
+      "learning_rate": 5.891555555555556e-06,
+      "loss": 1.606,
+      "step": 45930
+    },
+    {
+      "epoch": 1.47008,
+      "grad_norm": 45.679046630859375,
+      "learning_rate": 5.888e-06,
+      "loss": 1.5945,
+      "step": 45940
+    },
+    {
+      "epoch": 1.4704,
+      "grad_norm": 46.56907272338867,
+      "learning_rate": 5.884444444444445e-06,
+      "loss": 1.5931,
+      "step": 45950
+    },
+    {
+      "epoch": 1.47072,
+      "grad_norm": 44.32508850097656,
+      "learning_rate": 5.880888888888889e-06,
+      "loss": 1.5924,
+      "step": 45960
+    },
+    {
+      "epoch": 1.47104,
+      "grad_norm": 45.68732452392578,
+      "learning_rate": 5.877333333333334e-06,
+      "loss": 1.6142,
+      "step": 45970
+    },
+    {
+      "epoch": 1.47136,
+      "grad_norm": 46.45209503173828,
+      "learning_rate": 5.873777777777778e-06,
+      "loss": 1.6032,
+      "step": 45980
+    },
+    {
+      "epoch": 1.47168,
+      "grad_norm": 45.467323303222656,
+      "learning_rate": 5.870222222222223e-06,
+      "loss": 1.5986,
+      "step": 45990
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 46.137996673583984,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.6074,
+      "step": 46000
+    },
+    {
+      "epoch": 1.47232,
+      "grad_norm": 44.18669128417969,
+      "learning_rate": 5.863111111111111e-06,
+      "loss": 1.5913,
+      "step": 46010
+    },
+    {
+      "epoch": 1.47264,
+      "grad_norm": 44.52817153930664,
+      "learning_rate": 5.859555555555556e-06,
+      "loss": 1.6008,
+      "step": 46020
+    },
+    {
+      "epoch": 1.47296,
+      "grad_norm": 44.14411163330078,
+      "learning_rate": 5.856e-06,
+      "loss": 1.593,
+      "step": 46030
+    },
+    {
+      "epoch": 1.47328,
+      "grad_norm": 46.95811462402344,
+      "learning_rate": 5.852444444444445e-06,
+      "loss": 1.6368,
+      "step": 46040
+    },
+    {
+      "epoch": 1.4736,
+      "grad_norm": 46.106109619140625,
+      "learning_rate": 5.8488888888888895e-06,
+      "loss": 1.6116,
+      "step": 46050
+    },
+    {
+      "epoch": 1.4739200000000001,
+      "grad_norm": 45.18093490600586,
+      "learning_rate": 5.845333333333334e-06,
+      "loss": 1.5901,
+      "step": 46060
+    },
+    {
+      "epoch": 1.47424,
+      "grad_norm": 44.52831268310547,
+      "learning_rate": 5.8417777777777786e-06,
+      "loss": 1.6002,
+      "step": 46070
+    },
+    {
+      "epoch": 1.4745599999999999,
+      "grad_norm": 46.35838317871094,
+      "learning_rate": 5.838222222222223e-06,
+      "loss": 1.604,
+      "step": 46080
+    },
+    {
+      "epoch": 1.47488,
+      "grad_norm": 47.06918716430664,
+      "learning_rate": 5.834666666666668e-06,
+      "loss": 1.6245,
+      "step": 46090
+    },
+    {
+      "epoch": 1.4752,
+      "grad_norm": 47.2197151184082,
+      "learning_rate": 5.831111111111112e-06,
+      "loss": 1.5986,
+      "step": 46100
+    },
+    {
+      "epoch": 1.47552,
+      "grad_norm": 45.36792755126953,
+      "learning_rate": 5.827555555555556e-06,
+      "loss": 1.61,
+      "step": 46110
+    },
+    {
+      "epoch": 1.47584,
+      "grad_norm": 48.5582275390625,
+      "learning_rate": 5.8240000000000005e-06,
+      "loss": 1.6092,
+      "step": 46120
+    },
+    {
+      "epoch": 1.4761600000000001,
+      "grad_norm": 46.83791732788086,
+      "learning_rate": 5.820444444444445e-06,
+      "loss": 1.6158,
+      "step": 46130
+    },
+    {
+      "epoch": 1.47648,
+      "grad_norm": 45.138389587402344,
+      "learning_rate": 5.81688888888889e-06,
+      "loss": 1.6268,
+      "step": 46140
+    },
+    {
+      "epoch": 1.4768,
+      "grad_norm": 45.886837005615234,
+      "learning_rate": 5.813333333333334e-06,
+      "loss": 1.6131,
+      "step": 46150
+    },
+    {
+      "epoch": 1.47712,
+      "grad_norm": 45.56637191772461,
+      "learning_rate": 5.809777777777779e-06,
+      "loss": 1.6202,
+      "step": 46160
+    },
+    {
+      "epoch": 1.47744,
+      "grad_norm": 44.13711929321289,
+      "learning_rate": 5.806222222222223e-06,
+      "loss": 1.5829,
+      "step": 46170
+    },
+    {
+      "epoch": 1.47776,
+      "grad_norm": 44.50336456298828,
+      "learning_rate": 5.802666666666668e-06,
+      "loss": 1.5959,
+      "step": 46180
+    },
+    {
+      "epoch": 1.47808,
+      "grad_norm": 45.071685791015625,
+      "learning_rate": 5.799111111111112e-06,
+      "loss": 1.6065,
+      "step": 46190
+    },
+    {
+      "epoch": 1.4784,
+      "grad_norm": 42.92729568481445,
+      "learning_rate": 5.795555555555557e-06,
+      "loss": 1.6215,
+      "step": 46200
+    },
+    {
+      "epoch": 1.47872,
+      "grad_norm": 43.77521896362305,
+      "learning_rate": 5.792000000000001e-06,
+      "loss": 1.5754,
+      "step": 46210
+    },
+    {
+      "epoch": 1.47904,
+      "grad_norm": 46.85535430908203,
+      "learning_rate": 5.788444444444445e-06,
+      "loss": 1.6214,
+      "step": 46220
+    },
+    {
+      "epoch": 1.47936,
+      "grad_norm": 44.22446823120117,
+      "learning_rate": 5.78488888888889e-06,
+      "loss": 1.5873,
+      "step": 46230
+    },
+    {
+      "epoch": 1.47968,
+      "grad_norm": 44.680049896240234,
+      "learning_rate": 5.781333333333334e-06,
+      "loss": 1.6033,
+      "step": 46240
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 44.47894287109375,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 1.5551,
+      "step": 46250
+    },
+    {
+      "epoch": 1.48032,
+      "grad_norm": 45.85274887084961,
+      "learning_rate": 5.7742222222222225e-06,
+      "loss": 1.5752,
+      "step": 46260
+    },
+    {
+      "epoch": 1.48064,
+      "grad_norm": 43.49520492553711,
+      "learning_rate": 5.770666666666666e-06,
+      "loss": 1.603,
+      "step": 46270
+    },
+    {
+      "epoch": 1.48096,
+      "grad_norm": 46.607933044433594,
+      "learning_rate": 5.767111111111111e-06,
+      "loss": 1.6183,
+      "step": 46280
+    },
+    {
+      "epoch": 1.48128,
+      "grad_norm": 43.7589225769043,
+      "learning_rate": 5.763555555555555e-06,
+      "loss": 1.5868,
+      "step": 46290
+    },
+    {
+      "epoch": 1.4816,
+      "grad_norm": 45.31899642944336,
+      "learning_rate": 5.76e-06,
+      "loss": 1.5847,
+      "step": 46300
+    },
+    {
+      "epoch": 1.4819200000000001,
+      "grad_norm": 44.8742790222168,
+      "learning_rate": 5.7564444444444445e-06,
+      "loss": 1.6009,
+      "step": 46310
+    },
+    {
+      "epoch": 1.48224,
+      "grad_norm": 44.30742645263672,
+      "learning_rate": 5.752888888888889e-06,
+      "loss": 1.618,
+      "step": 46320
+    },
+    {
+      "epoch": 1.4825599999999999,
+      "grad_norm": 48.276710510253906,
+      "learning_rate": 5.7493333333333336e-06,
+      "loss": 1.6021,
+      "step": 46330
+    },
+    {
+      "epoch": 1.48288,
+      "grad_norm": 45.02067947387695,
+      "learning_rate": 5.745777777777778e-06,
+      "loss": 1.6088,
+      "step": 46340
+    },
+    {
+      "epoch": 1.4832,
+      "grad_norm": 43.991207122802734,
+      "learning_rate": 5.742222222222223e-06,
+      "loss": 1.6089,
+      "step": 46350
+    },
+    {
+      "epoch": 1.48352,
+      "grad_norm": 44.636775970458984,
+      "learning_rate": 5.738666666666667e-06,
+      "loss": 1.5817,
+      "step": 46360
+    },
+    {
+      "epoch": 1.48384,
+      "grad_norm": 45.58132553100586,
+      "learning_rate": 5.735111111111111e-06,
+      "loss": 1.6172,
+      "step": 46370
+    },
+    {
+      "epoch": 1.48416,
+      "grad_norm": 46.23253631591797,
+      "learning_rate": 5.7315555555555555e-06,
+      "loss": 1.597,
+      "step": 46380
+    },
+    {
+      "epoch": 1.48448,
+      "grad_norm": 45.27110290527344,
+      "learning_rate": 5.728e-06,
+      "loss": 1.5837,
+      "step": 46390
+    },
+    {
+      "epoch": 1.4848,
+      "grad_norm": 44.85041046142578,
+      "learning_rate": 5.724444444444445e-06,
+      "loss": 1.6112,
+      "step": 46400
+    },
+    {
+      "epoch": 1.48512,
+      "grad_norm": 44.14605712890625,
+      "learning_rate": 5.720888888888889e-06,
+      "loss": 1.6024,
+      "step": 46410
+    },
+    {
+      "epoch": 1.48544,
+      "grad_norm": 44.91823196411133,
+      "learning_rate": 5.717333333333334e-06,
+      "loss": 1.6021,
+      "step": 46420
+    },
+    {
+      "epoch": 1.48576,
+      "grad_norm": 47.42062759399414,
+      "learning_rate": 5.713777777777778e-06,
+      "loss": 1.5879,
+      "step": 46430
+    },
+    {
+      "epoch": 1.48608,
+      "grad_norm": 43.79571533203125,
+      "learning_rate": 5.710222222222223e-06,
+      "loss": 1.5729,
+      "step": 46440
+    },
+    {
+      "epoch": 1.4864,
+      "grad_norm": 43.80337905883789,
+      "learning_rate": 5.706666666666667e-06,
+      "loss": 1.6017,
+      "step": 46450
+    },
+    {
+      "epoch": 1.48672,
+      "grad_norm": 44.557674407958984,
+      "learning_rate": 5.703111111111111e-06,
+      "loss": 1.6278,
+      "step": 46460
+    },
+    {
+      "epoch": 1.48704,
+      "grad_norm": 43.0338134765625,
+      "learning_rate": 5.699555555555556e-06,
+      "loss": 1.593,
+      "step": 46470
+    },
+    {
+      "epoch": 1.48736,
+      "grad_norm": 44.92878341674805,
+      "learning_rate": 5.696e-06,
+      "loss": 1.6161,
+      "step": 46480
+    },
+    {
+      "epoch": 1.4876800000000001,
+      "grad_norm": 46.56928253173828,
+      "learning_rate": 5.692444444444445e-06,
+      "loss": 1.5964,
+      "step": 46490
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 44.472347259521484,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 1.6109,
+      "step": 46500
+    },
+    {
+      "epoch": 1.4883199999999999,
+      "grad_norm": 47.81435012817383,
+      "learning_rate": 5.685333333333334e-06,
+      "loss": 1.6079,
+      "step": 46510
+    },
+    {
+      "epoch": 1.48864,
+      "grad_norm": 45.43522644042969,
+      "learning_rate": 5.681777777777778e-06,
+      "loss": 1.5983,
+      "step": 46520
+    },
+    {
+      "epoch": 1.48896,
+      "grad_norm": 43.59268569946289,
+      "learning_rate": 5.678222222222223e-06,
+      "loss": 1.6173,
+      "step": 46530
+    },
+    {
+      "epoch": 1.48928,
+      "grad_norm": 46.434661865234375,
+      "learning_rate": 5.6746666666666675e-06,
+      "loss": 1.637,
+      "step": 46540
+    },
+    {
+      "epoch": 1.4896,
+      "grad_norm": 45.176727294921875,
+      "learning_rate": 5.671111111111112e-06,
+      "loss": 1.5801,
+      "step": 46550
+    },
+    {
+      "epoch": 1.4899200000000001,
+      "grad_norm": 44.13080596923828,
+      "learning_rate": 5.667555555555556e-06,
+      "loss": 1.606,
+      "step": 46560
+    },
+    {
+      "epoch": 1.49024,
+      "grad_norm": 45.61445617675781,
+      "learning_rate": 5.664e-06,
+      "loss": 1.6441,
+      "step": 46570
+    },
+    {
+      "epoch": 1.4905599999999999,
+      "grad_norm": 45.91599655151367,
+      "learning_rate": 5.660444444444445e-06,
+      "loss": 1.5853,
+      "step": 46580
+    },
+    {
+      "epoch": 1.49088,
+      "grad_norm": 48.271461486816406,
+      "learning_rate": 5.6568888888888894e-06,
+      "loss": 1.6122,
+      "step": 46590
+    },
+    {
+      "epoch": 1.4912,
+      "grad_norm": 44.87429428100586,
+      "learning_rate": 5.653333333333334e-06,
+      "loss": 1.6104,
+      "step": 46600
+    },
+    {
+      "epoch": 1.49152,
+      "grad_norm": 44.66387939453125,
+      "learning_rate": 5.6497777777777785e-06,
+      "loss": 1.6249,
+      "step": 46610
+    },
+    {
+      "epoch": 1.49184,
+      "grad_norm": 44.479129791259766,
+      "learning_rate": 5.646222222222223e-06,
+      "loss": 1.6212,
+      "step": 46620
+    },
+    {
+      "epoch": 1.49216,
+      "grad_norm": 45.07963180541992,
+      "learning_rate": 5.642666666666668e-06,
+      "loss": 1.6018,
+      "step": 46630
+    },
+    {
+      "epoch": 1.49248,
+      "grad_norm": 44.61531448364258,
+      "learning_rate": 5.639111111111112e-06,
+      "loss": 1.6329,
+      "step": 46640
+    },
+    {
+      "epoch": 1.4928,
+      "grad_norm": 44.30540084838867,
+      "learning_rate": 5.635555555555557e-06,
+      "loss": 1.5831,
+      "step": 46650
+    },
+    {
+      "epoch": 1.49312,
+      "grad_norm": 46.067989349365234,
+      "learning_rate": 5.6320000000000005e-06,
+      "loss": 1.5975,
+      "step": 46660
+    },
+    {
+      "epoch": 1.49344,
+      "grad_norm": 42.87896728515625,
+      "learning_rate": 5.628444444444445e-06,
+      "loss": 1.5897,
+      "step": 46670
+    },
+    {
+      "epoch": 1.49376,
+      "grad_norm": 44.7100830078125,
+      "learning_rate": 5.6248888888888896e-06,
+      "loss": 1.6028,
+      "step": 46680
+    },
+    {
+      "epoch": 1.49408,
+      "grad_norm": 45.40373229980469,
+      "learning_rate": 5.621333333333334e-06,
+      "loss": 1.603,
+      "step": 46690
+    },
+    {
+      "epoch": 1.4944,
+      "grad_norm": 45.81529235839844,
+      "learning_rate": 5.617777777777779e-06,
+      "loss": 1.5886,
+      "step": 46700
+    },
+    {
+      "epoch": 1.49472,
+      "grad_norm": 45.32423782348633,
+      "learning_rate": 5.614222222222223e-06,
+      "loss": 1.6118,
+      "step": 46710
+    },
+    {
+      "epoch": 1.49504,
+      "grad_norm": 53.411312103271484,
+      "learning_rate": 5.610666666666668e-06,
+      "loss": 1.6016,
+      "step": 46720
+    },
+    {
+      "epoch": 1.49536,
+      "grad_norm": 44.250362396240234,
+      "learning_rate": 5.607111111111112e-06,
+      "loss": 1.5979,
+      "step": 46730
+    },
+    {
+      "epoch": 1.4956800000000001,
+      "grad_norm": 43.215576171875,
+      "learning_rate": 5.603555555555557e-06,
+      "loss": 1.5766,
+      "step": 46740
+    },
+    {
+      "epoch": 1.496,
+      "grad_norm": 45.43357849121094,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.6177,
+      "step": 46750
+    },
+    {
+      "epoch": 1.4963199999999999,
+      "grad_norm": 43.38033676147461,
+      "learning_rate": 5.596444444444445e-06,
+      "loss": 1.6082,
+      "step": 46760
+    },
+    {
+      "epoch": 1.49664,
+      "grad_norm": 45.17076873779297,
+      "learning_rate": 5.592888888888889e-06,
+      "loss": 1.5966,
+      "step": 46770
+    },
+    {
+      "epoch": 1.49696,
+      "grad_norm": 44.70008850097656,
+      "learning_rate": 5.589333333333333e-06,
+      "loss": 1.5992,
+      "step": 46780
+    },
+    {
+      "epoch": 1.49728,
+      "grad_norm": 44.854427337646484,
+      "learning_rate": 5.585777777777778e-06,
+      "loss": 1.6058,
+      "step": 46790
+    },
+    {
+      "epoch": 1.4976,
+      "grad_norm": 45.27376937866211,
+      "learning_rate": 5.5822222222222225e-06,
+      "loss": 1.5978,
+      "step": 46800
+    },
+    {
+      "epoch": 1.49792,
+      "grad_norm": 45.82780838012695,
+      "learning_rate": 5.578666666666667e-06,
+      "loss": 1.6005,
+      "step": 46810
+    },
+    {
+      "epoch": 1.49824,
+      "grad_norm": 45.11985397338867,
+      "learning_rate": 5.575111111111111e-06,
+      "loss": 1.6075,
+      "step": 46820
+    },
+    {
+      "epoch": 1.49856,
+      "grad_norm": 45.486507415771484,
+      "learning_rate": 5.571555555555555e-06,
+      "loss": 1.58,
+      "step": 46830
+    },
+    {
+      "epoch": 1.49888,
+      "grad_norm": 46.610107421875,
+      "learning_rate": 5.568e-06,
+      "loss": 1.5867,
+      "step": 46840
+    },
+    {
+      "epoch": 1.4992,
+      "grad_norm": 45.96479415893555,
+      "learning_rate": 5.5644444444444444e-06,
+      "loss": 1.6277,
+      "step": 46850
+    },
+    {
+      "epoch": 1.49952,
+      "grad_norm": 44.71353530883789,
+      "learning_rate": 5.560888888888889e-06,
+      "loss": 1.5942,
+      "step": 46860
+    },
+    {
+      "epoch": 1.49984,
+      "grad_norm": 44.52565002441406,
+      "learning_rate": 5.5573333333333335e-06,
+      "loss": 1.5795,
+      "step": 46870
+    },
+    {
+      "epoch": 1.5001600000000002,
+      "grad_norm": 44.66649627685547,
+      "learning_rate": 5.553777777777778e-06,
+      "loss": 1.5923,
+      "step": 46880
+    },
+    {
+      "epoch": 1.50048,
+      "grad_norm": 44.97325134277344,
+      "learning_rate": 5.550222222222223e-06,
+      "loss": 1.5878,
+      "step": 46890
+    },
+    {
+      "epoch": 1.5008,
+      "grad_norm": 45.9375114440918,
+      "learning_rate": 5.546666666666667e-06,
+      "loss": 1.5763,
+      "step": 46900
+    },
+    {
+      "epoch": 1.50112,
+      "grad_norm": 45.02996063232422,
+      "learning_rate": 5.543111111111111e-06,
+      "loss": 1.6027,
+      "step": 46910
+    },
+    {
+      "epoch": 1.50144,
+      "grad_norm": 46.09992980957031,
+      "learning_rate": 5.5395555555555555e-06,
+      "loss": 1.5972,
+      "step": 46920
+    },
+    {
+      "epoch": 1.50176,
+      "grad_norm": 43.5955696105957,
+      "learning_rate": 5.536e-06,
+      "loss": 1.6164,
+      "step": 46930
+    },
+    {
+      "epoch": 1.5020799999999999,
+      "grad_norm": 43.62666320800781,
+      "learning_rate": 5.5324444444444446e-06,
+      "loss": 1.5872,
+      "step": 46940
+    },
+    {
+      "epoch": 1.5024,
+      "grad_norm": 45.09193801879883,
+      "learning_rate": 5.528888888888889e-06,
+      "loss": 1.5942,
+      "step": 46950
+    },
+    {
+      "epoch": 1.50272,
+      "grad_norm": 43.42613220214844,
+      "learning_rate": 5.525333333333334e-06,
+      "loss": 1.5889,
+      "step": 46960
+    },
+    {
+      "epoch": 1.50304,
+      "grad_norm": 45.32449722290039,
+      "learning_rate": 5.521777777777778e-06,
+      "loss": 1.5731,
+      "step": 46970
+    },
+    {
+      "epoch": 1.50336,
+      "grad_norm": 47.36893081665039,
+      "learning_rate": 5.518222222222223e-06,
+      "loss": 1.6103,
+      "step": 46980
+    },
+    {
+      "epoch": 1.5036800000000001,
+      "grad_norm": 45.34735870361328,
+      "learning_rate": 5.514666666666667e-06,
+      "loss": 1.6225,
+      "step": 46990
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 46.7711296081543,
+      "learning_rate": 5.511111111111112e-06,
+      "loss": 1.6042,
+      "step": 47000
+    },
+    {
+      "epoch": 1.5043199999999999,
+      "grad_norm": 44.58100891113281,
+      "learning_rate": 5.507555555555556e-06,
+      "loss": 1.6325,
+      "step": 47010
+    },
+    {
+      "epoch": 1.50464,
+      "grad_norm": 45.050392150878906,
+      "learning_rate": 5.504e-06,
+      "loss": 1.5773,
+      "step": 47020
+    },
+    {
+      "epoch": 1.50496,
+      "grad_norm": 46.156028747558594,
+      "learning_rate": 5.500444444444445e-06,
+      "loss": 1.5925,
+      "step": 47030
+    },
+    {
+      "epoch": 1.50528,
+      "grad_norm": 45.153507232666016,
+      "learning_rate": 5.496888888888889e-06,
+      "loss": 1.5523,
+      "step": 47040
+    },
+    {
+      "epoch": 1.5056,
+      "grad_norm": 44.15436935424805,
+      "learning_rate": 5.493333333333334e-06,
+      "loss": 1.6102,
+      "step": 47050
+    },
+    {
+      "epoch": 1.5059200000000001,
+      "grad_norm": 44.35607147216797,
+      "learning_rate": 5.489777777777778e-06,
+      "loss": 1.6043,
+      "step": 47060
+    },
+    {
+      "epoch": 1.50624,
+      "grad_norm": 47.05322265625,
+      "learning_rate": 5.486222222222223e-06,
+      "loss": 1.6261,
+      "step": 47070
+    },
+    {
+      "epoch": 1.50656,
+      "grad_norm": 47.88312911987305,
+      "learning_rate": 5.4826666666666675e-06,
+      "loss": 1.6146,
+      "step": 47080
+    },
+    {
+      "epoch": 1.50688,
+      "grad_norm": 44.9384651184082,
+      "learning_rate": 5.479111111111112e-06,
+      "loss": 1.6216,
+      "step": 47090
+    },
+    {
+      "epoch": 1.5072,
+      "grad_norm": 45.72306823730469,
+      "learning_rate": 5.475555555555557e-06,
+      "loss": 1.6122,
+      "step": 47100
+    },
+    {
+      "epoch": 1.50752,
+      "grad_norm": 44.87019729614258,
+      "learning_rate": 5.472e-06,
+      "loss": 1.6009,
+      "step": 47110
+    },
+    {
+      "epoch": 1.5078399999999998,
+      "grad_norm": 43.695926666259766,
+      "learning_rate": 5.468444444444445e-06,
+      "loss": 1.6109,
+      "step": 47120
+    },
+    {
+      "epoch": 1.50816,
+      "grad_norm": 44.68513107299805,
+      "learning_rate": 5.464888888888889e-06,
+      "loss": 1.5772,
+      "step": 47130
+    },
+    {
+      "epoch": 1.50848,
+      "grad_norm": 44.061100006103516,
+      "learning_rate": 5.461333333333334e-06,
+      "loss": 1.6069,
+      "step": 47140
+    },
+    {
+      "epoch": 1.5088,
+      "grad_norm": 45.412845611572266,
+      "learning_rate": 5.4577777777777785e-06,
+      "loss": 1.5927,
+      "step": 47150
+    },
+    {
+      "epoch": 1.50912,
+      "grad_norm": 46.47201919555664,
+      "learning_rate": 5.454222222222223e-06,
+      "loss": 1.6105,
+      "step": 47160
+    },
+    {
+      "epoch": 1.5094400000000001,
+      "grad_norm": 42.82103729248047,
+      "learning_rate": 5.450666666666668e-06,
+      "loss": 1.5919,
+      "step": 47170
+    },
+    {
+      "epoch": 1.50976,
+      "grad_norm": 46.60007858276367,
+      "learning_rate": 5.447111111111112e-06,
+      "loss": 1.5819,
+      "step": 47180
+    },
+    {
+      "epoch": 1.5100799999999999,
+      "grad_norm": 45.34518051147461,
+      "learning_rate": 5.443555555555557e-06,
+      "loss": 1.5926,
+      "step": 47190
+    },
+    {
+      "epoch": 1.5104,
+      "grad_norm": 44.76665496826172,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 1.6135,
+      "step": 47200
+    },
+    {
+      "epoch": 1.51072,
+      "grad_norm": 44.275394439697266,
+      "learning_rate": 5.436444444444445e-06,
+      "loss": 1.6198,
+      "step": 47210
+    },
+    {
+      "epoch": 1.51104,
+      "grad_norm": 45.03862380981445,
+      "learning_rate": 5.4328888888888895e-06,
+      "loss": 1.5881,
+      "step": 47220
+    },
+    {
+      "epoch": 1.51136,
+      "grad_norm": 46.152610778808594,
+      "learning_rate": 5.429333333333334e-06,
+      "loss": 1.6101,
+      "step": 47230
+    },
+    {
+      "epoch": 1.5116800000000001,
+      "grad_norm": 46.49415969848633,
+      "learning_rate": 5.425777777777779e-06,
+      "loss": 1.5826,
+      "step": 47240
+    },
+    {
+      "epoch": 1.512,
+      "grad_norm": 44.52889633178711,
+      "learning_rate": 5.422222222222223e-06,
+      "loss": 1.5879,
+      "step": 47250
+    },
+    {
+      "epoch": 1.5123199999999999,
+      "grad_norm": 50.50562286376953,
+      "learning_rate": 5.418666666666668e-06,
+      "loss": 1.5655,
+      "step": 47260
+    },
+    {
+      "epoch": 1.51264,
+      "grad_norm": 46.33927536010742,
+      "learning_rate": 5.415111111111112e-06,
+      "loss": 1.5988,
+      "step": 47270
+    },
+    {
+      "epoch": 1.51296,
+      "grad_norm": 43.79774856567383,
+      "learning_rate": 5.411555555555555e-06,
+      "loss": 1.614,
+      "step": 47280
+    },
+    {
+      "epoch": 1.51328,
+      "grad_norm": 46.49938201904297,
+      "learning_rate": 5.408e-06,
+      "loss": 1.5785,
+      "step": 47290
+    },
+    {
+      "epoch": 1.5135999999999998,
+      "grad_norm": 47.394378662109375,
+      "learning_rate": 5.404444444444444e-06,
+      "loss": 1.5937,
+      "step": 47300
+    },
+    {
+      "epoch": 1.5139200000000002,
+      "grad_norm": 44.6570930480957,
+      "learning_rate": 5.400888888888889e-06,
+      "loss": 1.6198,
+      "step": 47310
+    },
+    {
+      "epoch": 1.51424,
+      "grad_norm": 45.7407341003418,
+      "learning_rate": 5.397333333333333e-06,
+      "loss": 1.6085,
+      "step": 47320
+    },
+    {
+      "epoch": 1.51456,
+      "grad_norm": 44.883331298828125,
+      "learning_rate": 5.393777777777778e-06,
+      "loss": 1.5917,
+      "step": 47330
+    },
+    {
+      "epoch": 1.51488,
+      "grad_norm": 45.57594680786133,
+      "learning_rate": 5.3902222222222225e-06,
+      "loss": 1.5957,
+      "step": 47340
+    },
+    {
+      "epoch": 1.5152,
+      "grad_norm": 45.7113151550293,
+      "learning_rate": 5.386666666666667e-06,
+      "loss": 1.6068,
+      "step": 47350
+    },
+    {
+      "epoch": 1.51552,
+      "grad_norm": 47.294921875,
+      "learning_rate": 5.383111111111112e-06,
+      "loss": 1.5901,
+      "step": 47360
+    },
+    {
+      "epoch": 1.5158399999999999,
+      "grad_norm": 46.925968170166016,
+      "learning_rate": 5.379555555555555e-06,
+      "loss": 1.6108,
+      "step": 47370
+    },
+    {
+      "epoch": 1.51616,
+      "grad_norm": 46.3890495300293,
+      "learning_rate": 5.376e-06,
+      "loss": 1.5755,
+      "step": 47380
+    },
+    {
+      "epoch": 1.51648,
+      "grad_norm": 46.12717056274414,
+      "learning_rate": 5.372444444444444e-06,
+      "loss": 1.6021,
+      "step": 47390
+    },
+    {
+      "epoch": 1.5168,
+      "grad_norm": 47.06908416748047,
+      "learning_rate": 5.368888888888889e-06,
+      "loss": 1.5884,
+      "step": 47400
+    },
+    {
+      "epoch": 1.51712,
+      "grad_norm": 46.1263427734375,
+      "learning_rate": 5.3653333333333335e-06,
+      "loss": 1.5993,
+      "step": 47410
+    },
+    {
+      "epoch": 1.5174400000000001,
+      "grad_norm": 44.17634582519531,
+      "learning_rate": 5.361777777777778e-06,
+      "loss": 1.5801,
+      "step": 47420
+    },
+    {
+      "epoch": 1.51776,
+      "grad_norm": 44.7050895690918,
+      "learning_rate": 5.358222222222223e-06,
+      "loss": 1.6294,
+      "step": 47430
+    },
+    {
+      "epoch": 1.5180799999999999,
+      "grad_norm": 45.57835388183594,
+      "learning_rate": 5.354666666666667e-06,
+      "loss": 1.6072,
+      "step": 47440
+    },
+    {
+      "epoch": 1.5184,
+      "grad_norm": 43.80586242675781,
+      "learning_rate": 5.351111111111112e-06,
+      "loss": 1.5884,
+      "step": 47450
+    },
+    {
+      "epoch": 1.51872,
+      "grad_norm": 47.311214447021484,
+      "learning_rate": 5.3475555555555554e-06,
+      "loss": 1.5966,
+      "step": 47460
+    },
+    {
+      "epoch": 1.51904,
+      "grad_norm": 44.86566162109375,
+      "learning_rate": 5.344e-06,
+      "loss": 1.5913,
+      "step": 47470
+    },
+    {
+      "epoch": 1.51936,
+      "grad_norm": 46.597415924072266,
+      "learning_rate": 5.3404444444444445e-06,
+      "loss": 1.6077,
+      "step": 47480
+    },
+    {
+      "epoch": 1.5196800000000001,
+      "grad_norm": 46.57792663574219,
+      "learning_rate": 5.336888888888889e-06,
+      "loss": 1.6201,
+      "step": 47490
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 42.707462310791016,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.6004,
+      "step": 47500
+    },
+    {
+      "epoch": 1.52032,
+      "grad_norm": 46.76858139038086,
+      "learning_rate": 5.329777777777778e-06,
+      "loss": 1.5991,
+      "step": 47510
+    },
+    {
+      "epoch": 1.52064,
+      "grad_norm": 46.61631393432617,
+      "learning_rate": 5.326222222222223e-06,
+      "loss": 1.6124,
+      "step": 47520
+    },
+    {
+      "epoch": 1.52096,
+      "grad_norm": 43.41951370239258,
+      "learning_rate": 5.322666666666667e-06,
+      "loss": 1.5968,
+      "step": 47530
+    },
+    {
+      "epoch": 1.52128,
+      "grad_norm": 47.41061782836914,
+      "learning_rate": 5.319111111111112e-06,
+      "loss": 1.6163,
+      "step": 47540
+    },
+    {
+      "epoch": 1.5215999999999998,
+      "grad_norm": 44.8211669921875,
+      "learning_rate": 5.3155555555555564e-06,
+      "loss": 1.6268,
+      "step": 47550
+    },
+    {
+      "epoch": 1.5219200000000002,
+      "grad_norm": 47.11516571044922,
+      "learning_rate": 5.312e-06,
+      "loss": 1.6076,
+      "step": 47560
+    },
+    {
+      "epoch": 1.52224,
+      "grad_norm": 46.79738998413086,
+      "learning_rate": 5.308444444444445e-06,
+      "loss": 1.6105,
+      "step": 47570
+    },
+    {
+      "epoch": 1.52256,
+      "grad_norm": 46.0588493347168,
+      "learning_rate": 5.304888888888889e-06,
+      "loss": 1.596,
+      "step": 47580
+    },
+    {
+      "epoch": 1.52288,
+      "grad_norm": 45.7782096862793,
+      "learning_rate": 5.301333333333334e-06,
+      "loss": 1.5706,
+      "step": 47590
+    },
+    {
+      "epoch": 1.5232,
+      "grad_norm": 45.6116828918457,
+      "learning_rate": 5.297777777777778e-06,
+      "loss": 1.5931,
+      "step": 47600
+    },
+    {
+      "epoch": 1.52352,
+      "grad_norm": 46.35176467895508,
+      "learning_rate": 5.294222222222223e-06,
+      "loss": 1.6091,
+      "step": 47610
+    },
+    {
+      "epoch": 1.5238399999999999,
+      "grad_norm": 46.78037643432617,
+      "learning_rate": 5.2906666666666675e-06,
+      "loss": 1.5805,
+      "step": 47620
+    },
+    {
+      "epoch": 1.52416,
+      "grad_norm": 45.619407653808594,
+      "learning_rate": 5.287111111111112e-06,
+      "loss": 1.6012,
+      "step": 47630
+    },
+    {
+      "epoch": 1.52448,
+      "grad_norm": 45.890953063964844,
+      "learning_rate": 5.2835555555555566e-06,
+      "loss": 1.607,
+      "step": 47640
+    },
+    {
+      "epoch": 1.5248,
+      "grad_norm": 47.279937744140625,
+      "learning_rate": 5.28e-06,
+      "loss": 1.5893,
+      "step": 47650
+    },
+    {
+      "epoch": 1.52512,
+      "grad_norm": 44.826416015625,
+      "learning_rate": 5.276444444444445e-06,
+      "loss": 1.6026,
+      "step": 47660
+    },
+    {
+      "epoch": 1.5254400000000001,
+      "grad_norm": 47.3812255859375,
+      "learning_rate": 5.272888888888889e-06,
+      "loss": 1.6193,
+      "step": 47670
+    },
+    {
+      "epoch": 1.52576,
+      "grad_norm": 44.42546081542969,
+      "learning_rate": 5.269333333333334e-06,
+      "loss": 1.5867,
+      "step": 47680
+    },
+    {
+      "epoch": 1.5260799999999999,
+      "grad_norm": 45.50373840332031,
+      "learning_rate": 5.2657777777777785e-06,
+      "loss": 1.5956,
+      "step": 47690
+    },
+    {
+      "epoch": 1.5264,
+      "grad_norm": 47.383426666259766,
+      "learning_rate": 5.262222222222223e-06,
+      "loss": 1.5812,
+      "step": 47700
+    },
+    {
+      "epoch": 1.52672,
+      "grad_norm": 43.08425521850586,
+      "learning_rate": 5.258666666666668e-06,
+      "loss": 1.5956,
+      "step": 47710
+    },
+    {
+      "epoch": 1.52704,
+      "grad_norm": 45.51042556762695,
+      "learning_rate": 5.255111111111112e-06,
+      "loss": 1.5905,
+      "step": 47720
+    },
+    {
+      "epoch": 1.52736,
+      "grad_norm": 44.803035736083984,
+      "learning_rate": 5.251555555555557e-06,
+      "loss": 1.6111,
+      "step": 47730
+    },
+    {
+      "epoch": 1.5276800000000001,
+      "grad_norm": 45.717918395996094,
+      "learning_rate": 5.248000000000001e-06,
+      "loss": 1.6296,
+      "step": 47740
+    },
+    {
+      "epoch": 1.528,
+      "grad_norm": 46.95458984375,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 1.5974,
+      "step": 47750
+    },
+    {
+      "epoch": 1.52832,
+      "grad_norm": 46.433319091796875,
+      "learning_rate": 5.2408888888888895e-06,
+      "loss": 1.6004,
+      "step": 47760
+    },
+    {
+      "epoch": 1.52864,
+      "grad_norm": 45.6478271484375,
+      "learning_rate": 5.237333333333334e-06,
+      "loss": 1.6422,
+      "step": 47770
+    },
+    {
+      "epoch": 1.52896,
+      "grad_norm": 45.677242279052734,
+      "learning_rate": 5.233777777777779e-06,
+      "loss": 1.5906,
+      "step": 47780
+    },
+    {
+      "epoch": 1.52928,
+      "grad_norm": 44.30915069580078,
+      "learning_rate": 5.230222222222223e-06,
+      "loss": 1.5966,
+      "step": 47790
+    },
+    {
+      "epoch": 1.5295999999999998,
+      "grad_norm": 43.836761474609375,
+      "learning_rate": 5.226666666666667e-06,
+      "loss": 1.5933,
+      "step": 47800
+    },
+    {
+      "epoch": 1.52992,
+      "grad_norm": 46.55387878417969,
+      "learning_rate": 5.2231111111111114e-06,
+      "loss": 1.6344,
+      "step": 47810
+    },
+    {
+      "epoch": 1.53024,
+      "grad_norm": 44.398651123046875,
+      "learning_rate": 5.219555555555555e-06,
+      "loss": 1.5997,
+      "step": 47820
+    },
+    {
+      "epoch": 1.53056,
+      "grad_norm": 47.1364860534668,
+      "learning_rate": 5.216e-06,
+      "loss": 1.5878,
+      "step": 47830
+    },
+    {
+      "epoch": 1.53088,
+      "grad_norm": 45.02302932739258,
+      "learning_rate": 5.212444444444444e-06,
+      "loss": 1.599,
+      "step": 47840
+    },
+    {
+      "epoch": 1.5312000000000001,
+      "grad_norm": 47.26573181152344,
+      "learning_rate": 5.208888888888889e-06,
+      "loss": 1.5955,
+      "step": 47850
+    },
+    {
+      "epoch": 1.53152,
+      "grad_norm": 43.69180679321289,
+      "learning_rate": 5.205333333333333e-06,
+      "loss": 1.5938,
+      "step": 47860
+    },
+    {
+      "epoch": 1.5318399999999999,
+      "grad_norm": 43.49293899536133,
+      "learning_rate": 5.201777777777778e-06,
+      "loss": 1.5799,
+      "step": 47870
+    },
+    {
+      "epoch": 1.53216,
+      "grad_norm": 46.23622131347656,
+      "learning_rate": 5.1982222222222225e-06,
+      "loss": 1.6039,
+      "step": 47880
+    },
+    {
+      "epoch": 1.53248,
+      "grad_norm": 47.350921630859375,
+      "learning_rate": 5.194666666666667e-06,
+      "loss": 1.615,
+      "step": 47890
+    },
+    {
+      "epoch": 1.5328,
+      "grad_norm": 45.750858306884766,
+      "learning_rate": 5.1911111111111116e-06,
+      "loss": 1.5888,
+      "step": 47900
+    },
+    {
+      "epoch": 1.53312,
+      "grad_norm": 45.1241455078125,
+      "learning_rate": 5.187555555555555e-06,
+      "loss": 1.5794,
+      "step": 47910
+    },
+    {
+      "epoch": 1.5334400000000001,
+      "grad_norm": 41.951683044433594,
+      "learning_rate": 5.184e-06,
+      "loss": 1.5688,
+      "step": 47920
+    },
+    {
+      "epoch": 1.53376,
+      "grad_norm": 45.95042419433594,
+      "learning_rate": 5.180444444444444e-06,
+      "loss": 1.6301,
+      "step": 47930
+    },
+    {
+      "epoch": 1.5340799999999999,
+      "grad_norm": 46.773109436035156,
+      "learning_rate": 5.176888888888889e-06,
+      "loss": 1.5846,
+      "step": 47940
+    },
+    {
+      "epoch": 1.5344,
+      "grad_norm": 46.49460983276367,
+      "learning_rate": 5.1733333333333335e-06,
+      "loss": 1.5899,
+      "step": 47950
+    },
+    {
+      "epoch": 1.53472,
+      "grad_norm": 43.58417510986328,
+      "learning_rate": 5.169777777777778e-06,
+      "loss": 1.5804,
+      "step": 47960
+    },
+    {
+      "epoch": 1.53504,
+      "grad_norm": 47.46805953979492,
+      "learning_rate": 5.166222222222223e-06,
+      "loss": 1.6281,
+      "step": 47970
+    },
+    {
+      "epoch": 1.5353599999999998,
+      "grad_norm": 45.62743377685547,
+      "learning_rate": 5.162666666666667e-06,
+      "loss": 1.6028,
+      "step": 47980
+    },
+    {
+      "epoch": 1.5356800000000002,
+      "grad_norm": 45.07927703857422,
+      "learning_rate": 5.159111111111112e-06,
+      "loss": 1.5915,
+      "step": 47990
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 45.753665924072266,
+      "learning_rate": 5.155555555555556e-06,
+      "loss": 1.6215,
+      "step": 48000
+    },
+    {
+      "epoch": 1.53632,
+      "grad_norm": 44.055084228515625,
+      "learning_rate": 5.152e-06,
+      "loss": 1.5835,
+      "step": 48010
+    },
+    {
+      "epoch": 1.53664,
+      "grad_norm": 44.468421936035156,
+      "learning_rate": 5.1484444444444445e-06,
+      "loss": 1.5915,
+      "step": 48020
+    },
+    {
+      "epoch": 1.53696,
+      "grad_norm": 45.8538932800293,
+      "learning_rate": 5.144888888888889e-06,
+      "loss": 1.605,
+      "step": 48030
+    },
+    {
+      "epoch": 1.53728,
+      "grad_norm": 45.57746505737305,
+      "learning_rate": 5.141333333333334e-06,
+      "loss": 1.6134,
+      "step": 48040
+    },
+    {
+      "epoch": 1.5375999999999999,
+      "grad_norm": 46.702674865722656,
+      "learning_rate": 5.137777777777778e-06,
+      "loss": 1.5768,
+      "step": 48050
+    },
+    {
+      "epoch": 1.53792,
+      "grad_norm": 45.325279235839844,
+      "learning_rate": 5.134222222222223e-06,
+      "loss": 1.6043,
+      "step": 48060
+    },
+    {
+      "epoch": 1.53824,
+      "grad_norm": 44.36808776855469,
+      "learning_rate": 5.130666666666667e-06,
+      "loss": 1.6235,
+      "step": 48070
+    },
+    {
+      "epoch": 1.53856,
+      "grad_norm": 45.456642150878906,
+      "learning_rate": 5.127111111111112e-06,
+      "loss": 1.5894,
+      "step": 48080
+    },
+    {
+      "epoch": 1.53888,
+      "grad_norm": 45.39940643310547,
+      "learning_rate": 5.123555555555556e-06,
+      "loss": 1.5723,
+      "step": 48090
+    },
+    {
+      "epoch": 1.5392000000000001,
+      "grad_norm": 44.75956344604492,
+      "learning_rate": 5.12e-06,
+      "loss": 1.6198,
+      "step": 48100
+    },
+    {
+      "epoch": 1.53952,
+      "grad_norm": 45.92814254760742,
+      "learning_rate": 5.116444444444445e-06,
+      "loss": 1.6059,
+      "step": 48110
+    },
+    {
+      "epoch": 1.5398399999999999,
+      "grad_norm": 45.60244369506836,
+      "learning_rate": 5.112888888888889e-06,
+      "loss": 1.6104,
+      "step": 48120
+    },
+    {
+      "epoch": 1.54016,
+      "grad_norm": 44.82307052612305,
+      "learning_rate": 5.109333333333334e-06,
+      "loss": 1.6071,
+      "step": 48130
+    },
+    {
+      "epoch": 1.54048,
+      "grad_norm": 45.584922790527344,
+      "learning_rate": 5.105777777777778e-06,
+      "loss": 1.5904,
+      "step": 48140
+    },
+    {
+      "epoch": 1.5408,
+      "grad_norm": 46.74277114868164,
+      "learning_rate": 5.102222222222223e-06,
+      "loss": 1.5904,
+      "step": 48150
+    },
+    {
+      "epoch": 1.54112,
+      "grad_norm": 43.10594177246094,
+      "learning_rate": 5.0986666666666674e-06,
+      "loss": 1.565,
+      "step": 48160
+    },
+    {
+      "epoch": 1.5414400000000001,
+      "grad_norm": 45.78068161010742,
+      "learning_rate": 5.095111111111112e-06,
+      "loss": 1.6054,
+      "step": 48170
+    },
+    {
+      "epoch": 1.54176,
+      "grad_norm": 47.33755111694336,
+      "learning_rate": 5.0915555555555565e-06,
+      "loss": 1.5906,
+      "step": 48180
+    },
+    {
+      "epoch": 1.54208,
+      "grad_norm": 46.029823303222656,
+      "learning_rate": 5.088000000000001e-06,
+      "loss": 1.5908,
+      "step": 48190
+    },
+    {
+      "epoch": 1.5424,
+      "grad_norm": 45.3862419128418,
+      "learning_rate": 5.084444444444445e-06,
+      "loss": 1.6119,
+      "step": 48200
+    },
+    {
+      "epoch": 1.54272,
+      "grad_norm": 45.942222595214844,
+      "learning_rate": 5.080888888888889e-06,
+      "loss": 1.6183,
+      "step": 48210
+    },
+    {
+      "epoch": 1.54304,
+      "grad_norm": 46.743202209472656,
+      "learning_rate": 5.077333333333334e-06,
+      "loss": 1.5942,
+      "step": 48220
+    },
+    {
+      "epoch": 1.5433599999999998,
+      "grad_norm": 45.592220306396484,
+      "learning_rate": 5.0737777777777785e-06,
+      "loss": 1.579,
+      "step": 48230
+    },
+    {
+      "epoch": 1.5436800000000002,
+      "grad_norm": 43.967987060546875,
+      "learning_rate": 5.070222222222223e-06,
+      "loss": 1.5893,
+      "step": 48240
+    },
+    {
+      "epoch": 1.544,
+      "grad_norm": 45.874881744384766,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.5631,
+      "step": 48250
+    },
+    {
+      "epoch": 1.54432,
+      "grad_norm": 44.52368927001953,
+      "learning_rate": 5.063111111111112e-06,
+      "loss": 1.6084,
+      "step": 48260
+    },
+    {
+      "epoch": 1.54464,
+      "grad_norm": 44.105743408203125,
+      "learning_rate": 5.059555555555557e-06,
+      "loss": 1.6041,
+      "step": 48270
+    },
+    {
+      "epoch": 1.54496,
+      "grad_norm": 45.011207580566406,
+      "learning_rate": 5.056000000000001e-06,
+      "loss": 1.6005,
+      "step": 48280
+    },
+    {
+      "epoch": 1.54528,
+      "grad_norm": 49.53215026855469,
+      "learning_rate": 5.052444444444446e-06,
+      "loss": 1.6248,
+      "step": 48290
+    },
+    {
+      "epoch": 1.5455999999999999,
+      "grad_norm": 48.222412109375,
+      "learning_rate": 5.0488888888888895e-06,
+      "loss": 1.5842,
+      "step": 48300
+    },
+    {
+      "epoch": 1.54592,
+      "grad_norm": 43.69586944580078,
+      "learning_rate": 5.045333333333333e-06,
+      "loss": 1.6032,
+      "step": 48310
+    },
+    {
+      "epoch": 1.54624,
+      "grad_norm": 46.159278869628906,
+      "learning_rate": 5.041777777777778e-06,
+      "loss": 1.6157,
+      "step": 48320
+    },
+    {
+      "epoch": 1.54656,
+      "grad_norm": 43.497520446777344,
+      "learning_rate": 5.038222222222222e-06,
+      "loss": 1.584,
+      "step": 48330
+    },
+    {
+      "epoch": 1.54688,
+      "grad_norm": 45.150970458984375,
+      "learning_rate": 5.034666666666667e-06,
+      "loss": 1.6224,
+      "step": 48340
+    },
+    {
+      "epoch": 1.5472000000000001,
+      "grad_norm": 45.95301818847656,
+      "learning_rate": 5.031111111111111e-06,
+      "loss": 1.584,
+      "step": 48350
+    },
+    {
+      "epoch": 1.54752,
+      "grad_norm": 46.46894836425781,
+      "learning_rate": 5.027555555555555e-06,
+      "loss": 1.6119,
+      "step": 48360
+    },
+    {
+      "epoch": 1.5478399999999999,
+      "grad_norm": 49.09371566772461,
+      "learning_rate": 5.024e-06,
+      "loss": 1.6229,
+      "step": 48370
+    },
+    {
+      "epoch": 1.54816,
+      "grad_norm": 47.55086135864258,
+      "learning_rate": 5.020444444444444e-06,
+      "loss": 1.6188,
+      "step": 48380
+    },
+    {
+      "epoch": 1.54848,
+      "grad_norm": 44.514366149902344,
+      "learning_rate": 5.016888888888889e-06,
+      "loss": 1.6023,
+      "step": 48390
+    },
+    {
+      "epoch": 1.5488,
+      "grad_norm": 45.68317794799805,
+      "learning_rate": 5.013333333333333e-06,
+      "loss": 1.5956,
+      "step": 48400
+    },
+    {
+      "epoch": 1.54912,
+      "grad_norm": 47.13374328613281,
+      "learning_rate": 5.009777777777778e-06,
+      "loss": 1.5704,
+      "step": 48410
+    },
+    {
+      "epoch": 1.5494400000000002,
+      "grad_norm": 44.4520149230957,
+      "learning_rate": 5.0062222222222224e-06,
+      "loss": 1.5915,
+      "step": 48420
+    },
+    {
+      "epoch": 1.54976,
+      "grad_norm": 43.78715896606445,
+      "learning_rate": 5.002666666666667e-06,
+      "loss": 1.5887,
+      "step": 48430
+    },
+    {
+      "epoch": 1.55008,
+      "grad_norm": 48.16891098022461,
+      "learning_rate": 4.9991111111111115e-06,
+      "loss": 1.6097,
+      "step": 48440
+    },
+    {
+      "epoch": 1.5504,
+      "grad_norm": 45.72539138793945,
+      "learning_rate": 4.995555555555556e-06,
+      "loss": 1.5952,
+      "step": 48450
+    },
+    {
+      "epoch": 1.55072,
+      "grad_norm": 49.00065231323242,
+      "learning_rate": 4.992e-06,
+      "loss": 1.5906,
+      "step": 48460
+    },
+    {
+      "epoch": 1.55104,
+      "grad_norm": 45.903076171875,
+      "learning_rate": 4.988444444444444e-06,
+      "loss": 1.6303,
+      "step": 48470
+    },
+    {
+      "epoch": 1.5513599999999999,
+      "grad_norm": 45.5451774597168,
+      "learning_rate": 4.984888888888889e-06,
+      "loss": 1.5971,
+      "step": 48480
+    },
+    {
+      "epoch": 1.55168,
+      "grad_norm": 47.048728942871094,
+      "learning_rate": 4.9813333333333335e-06,
+      "loss": 1.5911,
+      "step": 48490
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 47.47088623046875,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 1.5933,
+      "step": 48500
+    },
+    {
+      "epoch": 1.55232,
+      "grad_norm": 46.04580307006836,
+      "learning_rate": 4.974222222222223e-06,
+      "loss": 1.576,
+      "step": 48510
+    },
+    {
+      "epoch": 1.55264,
+      "grad_norm": 47.11751937866211,
+      "learning_rate": 4.970666666666667e-06,
+      "loss": 1.5979,
+      "step": 48520
+    },
+    {
+      "epoch": 1.5529600000000001,
+      "grad_norm": 46.44295883178711,
+      "learning_rate": 4.967111111111112e-06,
+      "loss": 1.6133,
+      "step": 48530
+    },
+    {
+      "epoch": 1.55328,
+      "grad_norm": 47.402000427246094,
+      "learning_rate": 4.963555555555556e-06,
+      "loss": 1.6198,
+      "step": 48540
+    },
+    {
+      "epoch": 1.5535999999999999,
+      "grad_norm": 45.78688430786133,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 1.5773,
+      "step": 48550
+    },
+    {
+      "epoch": 1.55392,
+      "grad_norm": 45.35833740234375,
+      "learning_rate": 4.9564444444444445e-06,
+      "loss": 1.5802,
+      "step": 48560
+    },
+    {
+      "epoch": 1.55424,
+      "grad_norm": 46.027183532714844,
+      "learning_rate": 4.952888888888889e-06,
+      "loss": 1.6121,
+      "step": 48570
+    },
+    {
+      "epoch": 1.55456,
+      "grad_norm": 46.775455474853516,
+      "learning_rate": 4.949333333333334e-06,
+      "loss": 1.6137,
+      "step": 48580
+    },
+    {
+      "epoch": 1.55488,
+      "grad_norm": 42.35320281982422,
+      "learning_rate": 4.945777777777778e-06,
+      "loss": 1.6006,
+      "step": 48590
+    },
+    {
+      "epoch": 1.5552000000000001,
+      "grad_norm": 44.261661529541016,
+      "learning_rate": 4.942222222222223e-06,
+      "loss": 1.6055,
+      "step": 48600
+    },
+    {
+      "epoch": 1.55552,
+      "grad_norm": 43.88352966308594,
+      "learning_rate": 4.938666666666667e-06,
+      "loss": 1.5861,
+      "step": 48610
+    },
+    {
+      "epoch": 1.55584,
+      "grad_norm": 46.54319381713867,
+      "learning_rate": 4.935111111111112e-06,
+      "loss": 1.587,
+      "step": 48620
+    },
+    {
+      "epoch": 1.55616,
+      "grad_norm": 43.899959564208984,
+      "learning_rate": 4.931555555555556e-06,
+      "loss": 1.5613,
+      "step": 48630
+    },
+    {
+      "epoch": 1.55648,
+      "grad_norm": 44.880645751953125,
+      "learning_rate": 4.928000000000001e-06,
+      "loss": 1.6211,
+      "step": 48640
+    },
+    {
+      "epoch": 1.5568,
+      "grad_norm": 46.75686264038086,
+      "learning_rate": 4.924444444444445e-06,
+      "loss": 1.6005,
+      "step": 48650
+    },
+    {
+      "epoch": 1.5571199999999998,
+      "grad_norm": 45.599708557128906,
+      "learning_rate": 4.920888888888889e-06,
+      "loss": 1.5848,
+      "step": 48660
+    },
+    {
+      "epoch": 1.5574400000000002,
+      "grad_norm": 47.24040222167969,
+      "learning_rate": 4.917333333333334e-06,
+      "loss": 1.5788,
+      "step": 48670
+    },
+    {
+      "epoch": 1.55776,
+      "grad_norm": 45.050682067871094,
+      "learning_rate": 4.913777777777778e-06,
+      "loss": 1.6101,
+      "step": 48680
+    },
+    {
+      "epoch": 1.55808,
+      "grad_norm": 46.12038803100586,
+      "learning_rate": 4.910222222222223e-06,
+      "loss": 1.5966,
+      "step": 48690
+    },
+    {
+      "epoch": 1.5584,
+      "grad_norm": 45.54237365722656,
+      "learning_rate": 4.9066666666666666e-06,
+      "loss": 1.5553,
+      "step": 48700
+    },
+    {
+      "epoch": 1.55872,
+      "grad_norm": 46.248050689697266,
+      "learning_rate": 4.903111111111111e-06,
+      "loss": 1.5962,
+      "step": 48710
+    },
+    {
+      "epoch": 1.55904,
+      "grad_norm": 45.845863342285156,
+      "learning_rate": 4.899555555555556e-06,
+      "loss": 1.5888,
+      "step": 48720
+    },
+    {
+      "epoch": 1.5593599999999999,
+      "grad_norm": 47.461910247802734,
+      "learning_rate": 4.896e-06,
+      "loss": 1.6009,
+      "step": 48730
+    },
+    {
+      "epoch": 1.55968,
+      "grad_norm": 46.03202819824219,
+      "learning_rate": 4.892444444444445e-06,
+      "loss": 1.5805,
+      "step": 48740
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 48.9558219909668,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 1.5842,
+      "step": 48750
+    },
+    {
+      "epoch": 1.56032,
+      "grad_norm": 45.61824035644531,
+      "learning_rate": 4.885333333333334e-06,
+      "loss": 1.569,
+      "step": 48760
+    },
+    {
+      "epoch": 1.56064,
+      "grad_norm": 43.683650970458984,
+      "learning_rate": 4.8817777777777784e-06,
+      "loss": 1.5892,
+      "step": 48770
+    },
+    {
+      "epoch": 1.5609600000000001,
+      "grad_norm": 46.38489532470703,
+      "learning_rate": 4.878222222222222e-06,
+      "loss": 1.6174,
+      "step": 48780
+    },
+    {
+      "epoch": 1.56128,
+      "grad_norm": 45.02937316894531,
+      "learning_rate": 4.874666666666667e-06,
+      "loss": 1.5899,
+      "step": 48790
+    },
+    {
+      "epoch": 1.5615999999999999,
+      "grad_norm": 46.43217468261719,
+      "learning_rate": 4.871111111111111e-06,
+      "loss": 1.592,
+      "step": 48800
+    },
+    {
+      "epoch": 1.56192,
+      "grad_norm": 44.164634704589844,
+      "learning_rate": 4.867555555555556e-06,
+      "loss": 1.5769,
+      "step": 48810
+    },
+    {
+      "epoch": 1.56224,
+      "grad_norm": 46.32484817504883,
+      "learning_rate": 4.864e-06,
+      "loss": 1.5899,
+      "step": 48820
+    },
+    {
+      "epoch": 1.56256,
+      "grad_norm": 47.25856018066406,
+      "learning_rate": 4.860444444444445e-06,
+      "loss": 1.5823,
+      "step": 48830
+    },
+    {
+      "epoch": 1.56288,
+      "grad_norm": 47.051937103271484,
+      "learning_rate": 4.8568888888888895e-06,
+      "loss": 1.5783,
+      "step": 48840
+    },
+    {
+      "epoch": 1.5632000000000001,
+      "grad_norm": 43.570640563964844,
+      "learning_rate": 4.853333333333334e-06,
+      "loss": 1.5962,
+      "step": 48850
+    },
+    {
+      "epoch": 1.56352,
+      "grad_norm": 47.42235565185547,
+      "learning_rate": 4.8497777777777786e-06,
+      "loss": 1.5896,
+      "step": 48860
+    },
+    {
+      "epoch": 1.56384,
+      "grad_norm": 45.89921951293945,
+      "learning_rate": 4.846222222222223e-06,
+      "loss": 1.5705,
+      "step": 48870
+    },
+    {
+      "epoch": 1.56416,
+      "grad_norm": 44.323524475097656,
+      "learning_rate": 4.842666666666667e-06,
+      "loss": 1.5953,
+      "step": 48880
+    },
+    {
+      "epoch": 1.56448,
+      "grad_norm": 45.725799560546875,
+      "learning_rate": 4.839111111111111e-06,
+      "loss": 1.5988,
+      "step": 48890
+    },
+    {
+      "epoch": 1.5648,
+      "grad_norm": 44.503028869628906,
+      "learning_rate": 4.835555555555556e-06,
+      "loss": 1.5833,
+      "step": 48900
+    },
+    {
+      "epoch": 1.5651199999999998,
+      "grad_norm": 44.74811553955078,
+      "learning_rate": 4.8320000000000005e-06,
+      "loss": 1.6001,
+      "step": 48910
+    },
+    {
+      "epoch": 1.5654400000000002,
+      "grad_norm": 46.46488952636719,
+      "learning_rate": 4.828444444444445e-06,
+      "loss": 1.5868,
+      "step": 48920
+    },
+    {
+      "epoch": 1.56576,
+      "grad_norm": 44.30849838256836,
+      "learning_rate": 4.82488888888889e-06,
+      "loss": 1.5965,
+      "step": 48930
+    },
+    {
+      "epoch": 1.56608,
+      "grad_norm": 45.116268157958984,
+      "learning_rate": 4.821333333333334e-06,
+      "loss": 1.6137,
+      "step": 48940
+    },
+    {
+      "epoch": 1.5664,
+      "grad_norm": 44.71920394897461,
+      "learning_rate": 4.817777777777779e-06,
+      "loss": 1.5803,
+      "step": 48950
+    },
+    {
+      "epoch": 1.5667200000000001,
+      "grad_norm": 45.48208999633789,
+      "learning_rate": 4.814222222222222e-06,
+      "loss": 1.5986,
+      "step": 48960
+    },
+    {
+      "epoch": 1.56704,
+      "grad_norm": 47.023948669433594,
+      "learning_rate": 4.810666666666667e-06,
+      "loss": 1.6032,
+      "step": 48970
+    },
+    {
+      "epoch": 1.5673599999999999,
+      "grad_norm": 46.040283203125,
+      "learning_rate": 4.8071111111111115e-06,
+      "loss": 1.5844,
+      "step": 48980
+    },
+    {
+      "epoch": 1.56768,
+      "grad_norm": 44.496978759765625,
+      "learning_rate": 4.803555555555556e-06,
+      "loss": 1.5661,
+      "step": 48990
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 45.307952880859375,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.5877,
+      "step": 49000
+    },
+    {
+      "epoch": 1.56832,
+      "grad_norm": 45.56713104248047,
+      "learning_rate": 4.796444444444444e-06,
+      "loss": 1.5885,
+      "step": 49010
+    },
+    {
+      "epoch": 1.56864,
+      "grad_norm": 44.79756546020508,
+      "learning_rate": 4.792888888888889e-06,
+      "loss": 1.5636,
+      "step": 49020
+    },
+    {
+      "epoch": 1.5689600000000001,
+      "grad_norm": 45.48712921142578,
+      "learning_rate": 4.7893333333333334e-06,
+      "loss": 1.6004,
+      "step": 49030
+    },
+    {
+      "epoch": 1.56928,
+      "grad_norm": 46.09973907470703,
+      "learning_rate": 4.785777777777778e-06,
+      "loss": 1.5947,
+      "step": 49040
+    },
+    {
+      "epoch": 1.5695999999999999,
+      "grad_norm": 45.13676834106445,
+      "learning_rate": 4.7822222222222226e-06,
+      "loss": 1.5717,
+      "step": 49050
+    },
+    {
+      "epoch": 1.56992,
+      "grad_norm": 47.738285064697266,
+      "learning_rate": 4.778666666666667e-06,
+      "loss": 1.6115,
+      "step": 49060
+    },
+    {
+      "epoch": 1.57024,
+      "grad_norm": 44.058349609375,
+      "learning_rate": 4.775111111111112e-06,
+      "loss": 1.5722,
+      "step": 49070
+    },
+    {
+      "epoch": 1.57056,
+      "grad_norm": 46.16585159301758,
+      "learning_rate": 4.771555555555556e-06,
+      "loss": 1.6029,
+      "step": 49080
+    },
+    {
+      "epoch": 1.57088,
+      "grad_norm": 46.44235610961914,
+      "learning_rate": 4.768000000000001e-06,
+      "loss": 1.5949,
+      "step": 49090
+    },
+    {
+      "epoch": 1.5712000000000002,
+      "grad_norm": 45.075191497802734,
+      "learning_rate": 4.7644444444444445e-06,
+      "loss": 1.6002,
+      "step": 49100
+    },
+    {
+      "epoch": 1.57152,
+      "grad_norm": 46.433570861816406,
+      "learning_rate": 4.760888888888889e-06,
+      "loss": 1.5938,
+      "step": 49110
+    },
+    {
+      "epoch": 1.57184,
+      "grad_norm": 43.26811981201172,
+      "learning_rate": 4.757333333333334e-06,
+      "loss": 1.5908,
+      "step": 49120
+    },
+    {
+      "epoch": 1.57216,
+      "grad_norm": 45.4081916809082,
+      "learning_rate": 4.753777777777778e-06,
+      "loss": 1.5939,
+      "step": 49130
+    },
+    {
+      "epoch": 1.57248,
+      "grad_norm": 47.71382141113281,
+      "learning_rate": 4.750222222222223e-06,
+      "loss": 1.5864,
+      "step": 49140
+    },
+    {
+      "epoch": 1.5728,
+      "grad_norm": 44.84182357788086,
+      "learning_rate": 4.746666666666667e-06,
+      "loss": 1.6356,
+      "step": 49150
+    },
+    {
+      "epoch": 1.5731199999999999,
+      "grad_norm": 46.003990173339844,
+      "learning_rate": 4.743111111111112e-06,
+      "loss": 1.6466,
+      "step": 49160
+    },
+    {
+      "epoch": 1.57344,
+      "grad_norm": 44.50692367553711,
+      "learning_rate": 4.739555555555556e-06,
+      "loss": 1.6076,
+      "step": 49170
+    },
+    {
+      "epoch": 1.57376,
+      "grad_norm": 47.03111267089844,
+      "learning_rate": 4.736000000000001e-06,
+      "loss": 1.5869,
+      "step": 49180
+    },
+    {
+      "epoch": 1.57408,
+      "grad_norm": 46.598873138427734,
+      "learning_rate": 4.7324444444444455e-06,
+      "loss": 1.5847,
+      "step": 49190
+    },
+    {
+      "epoch": 1.5744,
+      "grad_norm": 46.089786529541016,
+      "learning_rate": 4.728888888888889e-06,
+      "loss": 1.5767,
+      "step": 49200
+    },
+    {
+      "epoch": 1.5747200000000001,
+      "grad_norm": 45.16957092285156,
+      "learning_rate": 4.725333333333334e-06,
+      "loss": 1.5933,
+      "step": 49210
+    },
+    {
+      "epoch": 1.57504,
+      "grad_norm": 45.54878234863281,
+      "learning_rate": 4.721777777777778e-06,
+      "loss": 1.5773,
+      "step": 49220
+    },
+    {
+      "epoch": 1.5753599999999999,
+      "grad_norm": 46.04212188720703,
+      "learning_rate": 4.718222222222222e-06,
+      "loss": 1.5895,
+      "step": 49230
+    },
+    {
+      "epoch": 1.57568,
+      "grad_norm": 46.73760986328125,
+      "learning_rate": 4.7146666666666665e-06,
+      "loss": 1.6408,
+      "step": 49240
+    },
+    {
+      "epoch": 1.576,
+      "grad_norm": 47.090755462646484,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 1.623,
+      "step": 49250
+    },
+    {
+      "epoch": 1.57632,
+      "grad_norm": 46.1207160949707,
+      "learning_rate": 4.707555555555556e-06,
+      "loss": 1.5815,
+      "step": 49260
+    },
+    {
+      "epoch": 1.57664,
+      "grad_norm": 47.64818572998047,
+      "learning_rate": 4.704e-06,
+      "loss": 1.597,
+      "step": 49270
+    },
+    {
+      "epoch": 1.5769600000000001,
+      "grad_norm": 46.952392578125,
+      "learning_rate": 4.700444444444445e-06,
+      "loss": 1.5954,
+      "step": 49280
+    },
+    {
+      "epoch": 1.57728,
+      "grad_norm": 46.08356475830078,
+      "learning_rate": 4.696888888888889e-06,
+      "loss": 1.5949,
+      "step": 49290
+    },
+    {
+      "epoch": 1.5776,
+      "grad_norm": 45.83845901489258,
+      "learning_rate": 4.693333333333334e-06,
+      "loss": 1.6158,
+      "step": 49300
+    },
+    {
+      "epoch": 1.57792,
+      "grad_norm": 45.37620162963867,
+      "learning_rate": 4.689777777777778e-06,
+      "loss": 1.5881,
+      "step": 49310
+    },
+    {
+      "epoch": 1.57824,
+      "grad_norm": 44.12497329711914,
+      "learning_rate": 4.686222222222223e-06,
+      "loss": 1.6028,
+      "step": 49320
+    },
+    {
+      "epoch": 1.57856,
+      "grad_norm": 46.23371124267578,
+      "learning_rate": 4.682666666666667e-06,
+      "loss": 1.6129,
+      "step": 49330
+    },
+    {
+      "epoch": 1.5788799999999998,
+      "grad_norm": 46.70281982421875,
+      "learning_rate": 4.679111111111111e-06,
+      "loss": 1.6013,
+      "step": 49340
+    },
+    {
+      "epoch": 1.5792000000000002,
+      "grad_norm": 46.354454040527344,
+      "learning_rate": 4.675555555555556e-06,
+      "loss": 1.5801,
+      "step": 49350
+    },
+    {
+      "epoch": 1.57952,
+      "grad_norm": 44.693077087402344,
+      "learning_rate": 4.672e-06,
+      "loss": 1.6157,
+      "step": 49360
+    },
+    {
+      "epoch": 1.57984,
+      "grad_norm": 45.19998550415039,
+      "learning_rate": 4.668444444444445e-06,
+      "loss": 1.5855,
+      "step": 49370
+    },
+    {
+      "epoch": 1.58016,
+      "grad_norm": 43.8433837890625,
+      "learning_rate": 4.6648888888888894e-06,
+      "loss": 1.6182,
+      "step": 49380
+    },
+    {
+      "epoch": 1.58048,
+      "grad_norm": 44.49872589111328,
+      "learning_rate": 4.661333333333334e-06,
+      "loss": 1.6011,
+      "step": 49390
+    },
+    {
+      "epoch": 1.5808,
+      "grad_norm": 45.483192443847656,
+      "learning_rate": 4.6577777777777785e-06,
+      "loss": 1.5845,
+      "step": 49400
+    },
+    {
+      "epoch": 1.5811199999999999,
+      "grad_norm": 44.29515838623047,
+      "learning_rate": 4.654222222222223e-06,
+      "loss": 1.6335,
+      "step": 49410
+    },
+    {
+      "epoch": 1.58144,
+      "grad_norm": 43.83500671386719,
+      "learning_rate": 4.650666666666667e-06,
+      "loss": 1.6099,
+      "step": 49420
+    },
+    {
+      "epoch": 1.58176,
+      "grad_norm": 46.206172943115234,
+      "learning_rate": 4.647111111111111e-06,
+      "loss": 1.6153,
+      "step": 49430
+    },
+    {
+      "epoch": 1.58208,
+      "grad_norm": 47.253238677978516,
+      "learning_rate": 4.643555555555556e-06,
+      "loss": 1.6045,
+      "step": 49440
+    },
+    {
+      "epoch": 1.5824,
+      "grad_norm": 46.3438835144043,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 1.5862,
+      "step": 49450
+    },
+    {
+      "epoch": 1.5827200000000001,
+      "grad_norm": 46.675289154052734,
+      "learning_rate": 4.636444444444445e-06,
+      "loss": 1.5973,
+      "step": 49460
+    },
+    {
+      "epoch": 1.58304,
+      "grad_norm": 46.53513717651367,
+      "learning_rate": 4.632888888888889e-06,
+      "loss": 1.6053,
+      "step": 49470
+    },
+    {
+      "epoch": 1.5833599999999999,
+      "grad_norm": 44.555091857910156,
+      "learning_rate": 4.629333333333333e-06,
+      "loss": 1.6145,
+      "step": 49480
+    },
+    {
+      "epoch": 1.58368,
+      "grad_norm": 47.14988327026367,
+      "learning_rate": 4.625777777777778e-06,
+      "loss": 1.5914,
+      "step": 49490
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 47.16956329345703,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 1.582,
+      "step": 49500
+    },
+    {
+      "epoch": 1.58432,
+      "grad_norm": 45.73616409301758,
+      "learning_rate": 4.618666666666667e-06,
+      "loss": 1.5876,
+      "step": 49510
+    },
+    {
+      "epoch": 1.58464,
+      "grad_norm": 47.80562210083008,
+      "learning_rate": 4.6151111111111115e-06,
+      "loss": 1.5855,
+      "step": 49520
+    },
+    {
+      "epoch": 1.5849600000000001,
+      "grad_norm": 45.63853454589844,
+      "learning_rate": 4.611555555555556e-06,
+      "loss": 1.5892,
+      "step": 49530
+    },
+    {
+      "epoch": 1.58528,
+      "grad_norm": 46.739627838134766,
+      "learning_rate": 4.608000000000001e-06,
+      "loss": 1.5997,
+      "step": 49540
+    },
+    {
+      "epoch": 1.5856,
+      "grad_norm": 46.625587463378906,
+      "learning_rate": 4.604444444444444e-06,
+      "loss": 1.586,
+      "step": 49550
+    },
+    {
+      "epoch": 1.58592,
+      "grad_norm": 47.605796813964844,
+      "learning_rate": 4.600888888888889e-06,
+      "loss": 1.6017,
+      "step": 49560
+    },
+    {
+      "epoch": 1.58624,
+      "grad_norm": 45.937286376953125,
+      "learning_rate": 4.597333333333333e-06,
+      "loss": 1.5918,
+      "step": 49570
+    },
+    {
+      "epoch": 1.58656,
+      "grad_norm": 44.786460876464844,
+      "learning_rate": 4.593777777777778e-06,
+      "loss": 1.6187,
+      "step": 49580
+    },
+    {
+      "epoch": 1.5868799999999998,
+      "grad_norm": 44.00181579589844,
+      "learning_rate": 4.5902222222222225e-06,
+      "loss": 1.597,
+      "step": 49590
+    },
+    {
+      "epoch": 1.5872000000000002,
+      "grad_norm": 45.921409606933594,
+      "learning_rate": 4.586666666666667e-06,
+      "loss": 1.6047,
+      "step": 49600
+    },
+    {
+      "epoch": 1.58752,
+      "grad_norm": 44.386146545410156,
+      "learning_rate": 4.583111111111112e-06,
+      "loss": 1.5841,
+      "step": 49610
+    },
+    {
+      "epoch": 1.58784,
+      "grad_norm": 44.642608642578125,
+      "learning_rate": 4.579555555555556e-06,
+      "loss": 1.5969,
+      "step": 49620
+    },
+    {
+      "epoch": 1.58816,
+      "grad_norm": 45.06999588012695,
+      "learning_rate": 4.576000000000001e-06,
+      "loss": 1.6094,
+      "step": 49630
+    },
+    {
+      "epoch": 1.5884800000000001,
+      "grad_norm": 45.32560729980469,
+      "learning_rate": 4.572444444444445e-06,
+      "loss": 1.5804,
+      "step": 49640
+    },
+    {
+      "epoch": 1.5888,
+      "grad_norm": 47.215904235839844,
+      "learning_rate": 4.568888888888889e-06,
+      "loss": 1.622,
+      "step": 49650
+    },
+    {
+      "epoch": 1.5891199999999999,
+      "grad_norm": 45.09397888183594,
+      "learning_rate": 4.5653333333333336e-06,
+      "loss": 1.6058,
+      "step": 49660
+    },
+    {
+      "epoch": 1.58944,
+      "grad_norm": 47.4116325378418,
+      "learning_rate": 4.561777777777778e-06,
+      "loss": 1.6086,
+      "step": 49670
+    },
+    {
+      "epoch": 1.58976,
+      "grad_norm": 44.43334197998047,
+      "learning_rate": 4.558222222222223e-06,
+      "loss": 1.63,
+      "step": 49680
+    },
+    {
+      "epoch": 1.59008,
+      "grad_norm": 44.518882751464844,
+      "learning_rate": 4.554666666666667e-06,
+      "loss": 1.6114,
+      "step": 49690
+    },
+    {
+      "epoch": 1.5904,
+      "grad_norm": 45.74620056152344,
+      "learning_rate": 4.551111111111112e-06,
+      "loss": 1.5886,
+      "step": 49700
+    },
+    {
+      "epoch": 1.5907200000000001,
+      "grad_norm": 45.82404327392578,
+      "learning_rate": 4.547555555555556e-06,
+      "loss": 1.6022,
+      "step": 49710
+    },
+    {
+      "epoch": 1.59104,
+      "grad_norm": 47.253089904785156,
+      "learning_rate": 4.544000000000001e-06,
+      "loss": 1.5895,
+      "step": 49720
+    },
+    {
+      "epoch": 1.5913599999999999,
+      "grad_norm": 46.178253173828125,
+      "learning_rate": 4.540444444444445e-06,
+      "loss": 1.5902,
+      "step": 49730
+    },
+    {
+      "epoch": 1.59168,
+      "grad_norm": 46.93673324584961,
+      "learning_rate": 4.536888888888889e-06,
+      "loss": 1.6254,
+      "step": 49740
+    },
+    {
+      "epoch": 1.592,
+      "grad_norm": 46.61408996582031,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.6315,
+      "step": 49750
+    },
+    {
+      "epoch": 1.59232,
+      "grad_norm": 46.882930755615234,
+      "learning_rate": 4.529777777777778e-06,
+      "loss": 1.6253,
+      "step": 49760
+    },
+    {
+      "epoch": 1.5926399999999998,
+      "grad_norm": 47.14992141723633,
+      "learning_rate": 4.526222222222223e-06,
+      "loss": 1.5998,
+      "step": 49770
+    },
+    {
+      "epoch": 1.5929600000000002,
+      "grad_norm": 46.04390335083008,
+      "learning_rate": 4.5226666666666665e-06,
+      "loss": 1.5912,
+      "step": 49780
+    },
+    {
+      "epoch": 1.59328,
+      "grad_norm": 43.933555603027344,
+      "learning_rate": 4.519111111111111e-06,
+      "loss": 1.5897,
+      "step": 49790
+    },
+    {
+      "epoch": 1.5936,
+      "grad_norm": 48.31332015991211,
+      "learning_rate": 4.515555555555556e-06,
+      "loss": 1.5821,
+      "step": 49800
+    },
+    {
+      "epoch": 1.59392,
+      "grad_norm": 43.75424575805664,
+      "learning_rate": 4.512e-06,
+      "loss": 1.6014,
+      "step": 49810
+    },
+    {
+      "epoch": 1.59424,
+      "grad_norm": 45.77436447143555,
+      "learning_rate": 4.508444444444445e-06,
+      "loss": 1.6042,
+      "step": 49820
+    },
+    {
+      "epoch": 1.59456,
+      "grad_norm": 45.634220123291016,
+      "learning_rate": 4.504888888888889e-06,
+      "loss": 1.6119,
+      "step": 49830
+    },
+    {
+      "epoch": 1.5948799999999999,
+      "grad_norm": 45.671112060546875,
+      "learning_rate": 4.501333333333334e-06,
+      "loss": 1.6364,
+      "step": 49840
+    },
+    {
+      "epoch": 1.5952,
+      "grad_norm": 45.11393356323242,
+      "learning_rate": 4.497777777777778e-06,
+      "loss": 1.5874,
+      "step": 49850
+    },
+    {
+      "epoch": 1.59552,
+      "grad_norm": 46.57709884643555,
+      "learning_rate": 4.494222222222223e-06,
+      "loss": 1.5769,
+      "step": 49860
+    },
+    {
+      "epoch": 1.59584,
+      "grad_norm": 45.363006591796875,
+      "learning_rate": 4.490666666666667e-06,
+      "loss": 1.5677,
+      "step": 49870
+    },
+    {
+      "epoch": 1.59616,
+      "grad_norm": 45.681785583496094,
+      "learning_rate": 4.487111111111111e-06,
+      "loss": 1.5862,
+      "step": 49880
+    },
+    {
+      "epoch": 1.5964800000000001,
+      "grad_norm": 47.11852264404297,
+      "learning_rate": 4.483555555555556e-06,
+      "loss": 1.585,
+      "step": 49890
+    },
+    {
+      "epoch": 1.5968,
+      "grad_norm": 45.29941940307617,
+      "learning_rate": 4.48e-06,
+      "loss": 1.585,
+      "step": 49900
+    },
+    {
+      "epoch": 1.5971199999999999,
+      "grad_norm": 43.873050689697266,
+      "learning_rate": 4.476444444444445e-06,
+      "loss": 1.6232,
+      "step": 49910
+    },
+    {
+      "epoch": 1.59744,
+      "grad_norm": 44.94419860839844,
+      "learning_rate": 4.472888888888889e-06,
+      "loss": 1.6314,
+      "step": 49920
+    },
+    {
+      "epoch": 1.59776,
+      "grad_norm": 44.24449920654297,
+      "learning_rate": 4.469333333333334e-06,
+      "loss": 1.6162,
+      "step": 49930
+    },
+    {
+      "epoch": 1.59808,
+      "grad_norm": 44.8932991027832,
+      "learning_rate": 4.4657777777777785e-06,
+      "loss": 1.5827,
+      "step": 49940
+    },
+    {
+      "epoch": 1.5984,
+      "grad_norm": 43.619686126708984,
+      "learning_rate": 4.462222222222223e-06,
+      "loss": 1.5784,
+      "step": 49950
+    },
+    {
+      "epoch": 1.5987200000000001,
+      "grad_norm": 46.18088150024414,
+      "learning_rate": 4.458666666666668e-06,
+      "loss": 1.6026,
+      "step": 49960
+    },
+    {
+      "epoch": 1.59904,
+      "grad_norm": 59.165828704833984,
+      "learning_rate": 4.455111111111111e-06,
+      "loss": 1.6044,
+      "step": 49970
+    },
+    {
+      "epoch": 1.59936,
+      "grad_norm": 44.62675094604492,
+      "learning_rate": 4.451555555555556e-06,
+      "loss": 1.6104,
+      "step": 49980
+    },
+    {
+      "epoch": 1.59968,
+      "grad_norm": 45.41344451904297,
+      "learning_rate": 4.4480000000000004e-06,
+      "loss": 1.5886,
+      "step": 49990
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 45.50170135498047,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.5892,
+      "step": 50000
+    },
+    {
+      "epoch": 1.60032,
+      "grad_norm": 45.556915283203125,
+      "learning_rate": 4.440888888888889e-06,
+      "loss": 1.5747,
+      "step": 50010
+    },
+    {
+      "epoch": 1.6006399999999998,
+      "grad_norm": 43.671234130859375,
+      "learning_rate": 4.437333333333333e-06,
+      "loss": 1.6092,
+      "step": 50020
+    },
+    {
+      "epoch": 1.6009600000000002,
+      "grad_norm": 45.89202117919922,
+      "learning_rate": 4.433777777777778e-06,
+      "loss": 1.585,
+      "step": 50030
+    },
+    {
+      "epoch": 1.60128,
+      "grad_norm": 44.5806770324707,
+      "learning_rate": 4.430222222222222e-06,
+      "loss": 1.6141,
+      "step": 50040
+    },
+    {
+      "epoch": 1.6016,
+      "grad_norm": 47.617652893066406,
+      "learning_rate": 4.426666666666667e-06,
+      "loss": 1.5834,
+      "step": 50050
+    },
+    {
+      "epoch": 1.60192,
+      "grad_norm": 45.94891357421875,
+      "learning_rate": 4.4231111111111115e-06,
+      "loss": 1.5604,
+      "step": 50060
+    },
+    {
+      "epoch": 1.60224,
+      "grad_norm": 46.36294937133789,
+      "learning_rate": 4.419555555555556e-06,
+      "loss": 1.581,
+      "step": 50070
+    },
+    {
+      "epoch": 1.60256,
+      "grad_norm": 44.878177642822266,
+      "learning_rate": 4.416000000000001e-06,
+      "loss": 1.5884,
+      "step": 50080
+    },
+    {
+      "epoch": 1.6028799999999999,
+      "grad_norm": 46.650146484375,
+      "learning_rate": 4.412444444444445e-06,
+      "loss": 1.59,
+      "step": 50090
+    },
+    {
+      "epoch": 1.6032,
+      "grad_norm": 44.59064483642578,
+      "learning_rate": 4.408888888888889e-06,
+      "loss": 1.5989,
+      "step": 50100
+    },
+    {
+      "epoch": 1.60352,
+      "grad_norm": 45.031578063964844,
+      "learning_rate": 4.405333333333333e-06,
+      "loss": 1.5923,
+      "step": 50110
+    },
+    {
+      "epoch": 1.60384,
+      "grad_norm": 44.28569793701172,
+      "learning_rate": 4.401777777777778e-06,
+      "loss": 1.6092,
+      "step": 50120
+    },
+    {
+      "epoch": 1.60416,
+      "grad_norm": 44.284339904785156,
+      "learning_rate": 4.3982222222222225e-06,
+      "loss": 1.6016,
+      "step": 50130
+    },
+    {
+      "epoch": 1.6044800000000001,
+      "grad_norm": 44.18607711791992,
+      "learning_rate": 4.394666666666667e-06,
+      "loss": 1.6137,
+      "step": 50140
+    },
+    {
+      "epoch": 1.6048,
+      "grad_norm": 45.102230072021484,
+      "learning_rate": 4.391111111111112e-06,
+      "loss": 1.5995,
+      "step": 50150
+    },
+    {
+      "epoch": 1.6051199999999999,
+      "grad_norm": 44.339927673339844,
+      "learning_rate": 4.387555555555556e-06,
+      "loss": 1.6157,
+      "step": 50160
+    },
+    {
+      "epoch": 1.60544,
+      "grad_norm": 45.931880950927734,
+      "learning_rate": 4.384000000000001e-06,
+      "loss": 1.5997,
+      "step": 50170
+    },
+    {
+      "epoch": 1.60576,
+      "grad_norm": 44.674312591552734,
+      "learning_rate": 4.380444444444445e-06,
+      "loss": 1.6045,
+      "step": 50180
+    },
+    {
+      "epoch": 1.60608,
+      "grad_norm": 47.0339241027832,
+      "learning_rate": 4.37688888888889e-06,
+      "loss": 1.6022,
+      "step": 50190
+    },
+    {
+      "epoch": 1.6064,
+      "grad_norm": 49.61772155761719,
+      "learning_rate": 4.3733333333333335e-06,
+      "loss": 1.583,
+      "step": 50200
+    },
+    {
+      "epoch": 1.6067200000000001,
+      "grad_norm": 46.44553756713867,
+      "learning_rate": 4.369777777777778e-06,
+      "loss": 1.5689,
+      "step": 50210
+    },
+    {
+      "epoch": 1.60704,
+      "grad_norm": 44.75165939331055,
+      "learning_rate": 4.366222222222223e-06,
+      "loss": 1.5962,
+      "step": 50220
+    },
+    {
+      "epoch": 1.60736,
+      "grad_norm": 46.96379470825195,
+      "learning_rate": 4.362666666666667e-06,
+      "loss": 1.5637,
+      "step": 50230
+    },
+    {
+      "epoch": 1.60768,
+      "grad_norm": 44.98953628540039,
+      "learning_rate": 4.359111111111112e-06,
+      "loss": 1.5939,
+      "step": 50240
+    },
+    {
+      "epoch": 1.608,
+      "grad_norm": 46.01952362060547,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 1.6029,
+      "step": 50250
+    },
+    {
+      "epoch": 1.60832,
+      "grad_norm": 44.32866287231445,
+      "learning_rate": 4.352e-06,
+      "loss": 1.5855,
+      "step": 50260
+    },
+    {
+      "epoch": 1.6086399999999998,
+      "grad_norm": 47.755096435546875,
+      "learning_rate": 4.3484444444444446e-06,
+      "loss": 1.6094,
+      "step": 50270
+    },
+    {
+      "epoch": 1.60896,
+      "grad_norm": 44.24766159057617,
+      "learning_rate": 4.344888888888889e-06,
+      "loss": 1.5795,
+      "step": 50280
+    },
+    {
+      "epoch": 1.60928,
+      "grad_norm": 45.763004302978516,
+      "learning_rate": 4.341333333333334e-06,
+      "loss": 1.6071,
+      "step": 50290
+    },
+    {
+      "epoch": 1.6096,
+      "grad_norm": 45.684059143066406,
+      "learning_rate": 4.337777777777778e-06,
+      "loss": 1.619,
+      "step": 50300
+    },
+    {
+      "epoch": 1.60992,
+      "grad_norm": 44.89235305786133,
+      "learning_rate": 4.334222222222223e-06,
+      "loss": 1.563,
+      "step": 50310
+    },
+    {
+      "epoch": 1.6102400000000001,
+      "grad_norm": 47.123172760009766,
+      "learning_rate": 4.330666666666667e-06,
+      "loss": 1.6095,
+      "step": 50320
+    },
+    {
+      "epoch": 1.61056,
+      "grad_norm": 47.982330322265625,
+      "learning_rate": 4.327111111111111e-06,
+      "loss": 1.5968,
+      "step": 50330
+    },
+    {
+      "epoch": 1.6108799999999999,
+      "grad_norm": 44.019073486328125,
+      "learning_rate": 4.323555555555556e-06,
+      "loss": 1.5747,
+      "step": 50340
+    },
+    {
+      "epoch": 1.6112,
+      "grad_norm": 44.900474548339844,
+      "learning_rate": 4.32e-06,
+      "loss": 1.5865,
+      "step": 50350
+    },
+    {
+      "epoch": 1.61152,
+      "grad_norm": 46.74665451049805,
+      "learning_rate": 4.316444444444445e-06,
+      "loss": 1.5976,
+      "step": 50360
+    },
+    {
+      "epoch": 1.61184,
+      "grad_norm": 45.36431884765625,
+      "learning_rate": 4.312888888888889e-06,
+      "loss": 1.5888,
+      "step": 50370
+    },
+    {
+      "epoch": 1.61216,
+      "grad_norm": 44.3800163269043,
+      "learning_rate": 4.309333333333334e-06,
+      "loss": 1.6082,
+      "step": 50380
+    },
+    {
+      "epoch": 1.6124800000000001,
+      "grad_norm": 45.327423095703125,
+      "learning_rate": 4.305777777777778e-06,
+      "loss": 1.5848,
+      "step": 50390
+    },
+    {
+      "epoch": 1.6128,
+      "grad_norm": 46.212669372558594,
+      "learning_rate": 4.302222222222223e-06,
+      "loss": 1.6142,
+      "step": 50400
+    },
+    {
+      "epoch": 1.6131199999999999,
+      "grad_norm": 47.48764419555664,
+      "learning_rate": 4.2986666666666675e-06,
+      "loss": 1.6003,
+      "step": 50410
+    },
+    {
+      "epoch": 1.61344,
+      "grad_norm": 47.0029182434082,
+      "learning_rate": 4.295111111111111e-06,
+      "loss": 1.581,
+      "step": 50420
+    },
+    {
+      "epoch": 1.61376,
+      "grad_norm": 44.517669677734375,
+      "learning_rate": 4.291555555555556e-06,
+      "loss": 1.5755,
+      "step": 50430
+    },
+    {
+      "epoch": 1.61408,
+      "grad_norm": 44.641807556152344,
+      "learning_rate": 4.288e-06,
+      "loss": 1.6121,
+      "step": 50440
+    },
+    {
+      "epoch": 1.6143999999999998,
+      "grad_norm": 46.77535629272461,
+      "learning_rate": 4.284444444444445e-06,
+      "loss": 1.6039,
+      "step": 50450
+    },
+    {
+      "epoch": 1.6147200000000002,
+      "grad_norm": 47.83342742919922,
+      "learning_rate": 4.280888888888889e-06,
+      "loss": 1.5909,
+      "step": 50460
+    },
+    {
+      "epoch": 1.61504,
+      "grad_norm": 46.20832061767578,
+      "learning_rate": 4.277333333333334e-06,
+      "loss": 1.6097,
+      "step": 50470
+    },
+    {
+      "epoch": 1.61536,
+      "grad_norm": 46.99225616455078,
+      "learning_rate": 4.2737777777777785e-06,
+      "loss": 1.5748,
+      "step": 50480
+    },
+    {
+      "epoch": 1.61568,
+      "grad_norm": 48.268760681152344,
+      "learning_rate": 4.270222222222223e-06,
+      "loss": 1.5669,
+      "step": 50490
+    },
+    {
+      "epoch": 1.616,
+      "grad_norm": 45.4713249206543,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.6042,
+      "step": 50500
+    },
+    {
+      "epoch": 1.61632,
+      "grad_norm": 46.854278564453125,
+      "learning_rate": 4.263111111111111e-06,
+      "loss": 1.6017,
+      "step": 50510
+    },
+    {
+      "epoch": 1.6166399999999999,
+      "grad_norm": 45.30042266845703,
+      "learning_rate": 4.259555555555556e-06,
+      "loss": 1.6155,
+      "step": 50520
+    },
+    {
+      "epoch": 1.61696,
+      "grad_norm": 45.22309112548828,
+      "learning_rate": 4.256e-06,
+      "loss": 1.5923,
+      "step": 50530
+    },
+    {
+      "epoch": 1.61728,
+      "grad_norm": 45.24235916137695,
+      "learning_rate": 4.252444444444445e-06,
+      "loss": 1.5976,
+      "step": 50540
+    },
+    {
+      "epoch": 1.6176,
+      "grad_norm": 44.514007568359375,
+      "learning_rate": 4.248888888888889e-06,
+      "loss": 1.6064,
+      "step": 50550
+    },
+    {
+      "epoch": 1.61792,
+      "grad_norm": 44.3435173034668,
+      "learning_rate": 4.245333333333333e-06,
+      "loss": 1.6048,
+      "step": 50560
+    },
+    {
+      "epoch": 1.6182400000000001,
+      "grad_norm": 44.056148529052734,
+      "learning_rate": 4.241777777777778e-06,
+      "loss": 1.5859,
+      "step": 50570
+    },
+    {
+      "epoch": 1.61856,
+      "grad_norm": 46.236839294433594,
+      "learning_rate": 4.238222222222222e-06,
+      "loss": 1.5929,
+      "step": 50580
+    },
+    {
+      "epoch": 1.6188799999999999,
+      "grad_norm": 43.43770980834961,
+      "learning_rate": 4.234666666666667e-06,
+      "loss": 1.5758,
+      "step": 50590
+    },
+    {
+      "epoch": 1.6192,
+      "grad_norm": 46.04836654663086,
+      "learning_rate": 4.2311111111111114e-06,
+      "loss": 1.6049,
+      "step": 50600
+    },
+    {
+      "epoch": 1.61952,
+      "grad_norm": 45.94651794433594,
+      "learning_rate": 4.227555555555556e-06,
+      "loss": 1.5727,
+      "step": 50610
+    },
+    {
+      "epoch": 1.61984,
+      "grad_norm": 46.23525619506836,
+      "learning_rate": 4.2240000000000006e-06,
+      "loss": 1.5732,
+      "step": 50620
+    },
+    {
+      "epoch": 1.62016,
+      "grad_norm": 46.8028678894043,
+      "learning_rate": 4.220444444444445e-06,
+      "loss": 1.5841,
+      "step": 50630
+    },
+    {
+      "epoch": 1.6204800000000001,
+      "grad_norm": 45.00838851928711,
+      "learning_rate": 4.21688888888889e-06,
+      "loss": 1.5685,
+      "step": 50640
+    },
+    {
+      "epoch": 1.6208,
+      "grad_norm": 44.92591857910156,
+      "learning_rate": 4.213333333333333e-06,
+      "loss": 1.5978,
+      "step": 50650
+    },
+    {
+      "epoch": 1.62112,
+      "grad_norm": 45.20899963378906,
+      "learning_rate": 4.209777777777778e-06,
+      "loss": 1.5992,
+      "step": 50660
+    },
+    {
+      "epoch": 1.62144,
+      "grad_norm": 46.38852310180664,
+      "learning_rate": 4.2062222222222225e-06,
+      "loss": 1.5855,
+      "step": 50670
+    },
+    {
+      "epoch": 1.62176,
+      "grad_norm": 42.88059997558594,
+      "learning_rate": 4.202666666666667e-06,
+      "loss": 1.5763,
+      "step": 50680
+    },
+    {
+      "epoch": 1.62208,
+      "grad_norm": 46.491859436035156,
+      "learning_rate": 4.199111111111112e-06,
+      "loss": 1.5542,
+      "step": 50690
+    },
+    {
+      "epoch": 1.6223999999999998,
+      "grad_norm": 46.772457122802734,
+      "learning_rate": 4.195555555555556e-06,
+      "loss": 1.5973,
+      "step": 50700
+    },
+    {
+      "epoch": 1.6227200000000002,
+      "grad_norm": 45.93109893798828,
+      "learning_rate": 4.192000000000001e-06,
+      "loss": 1.6171,
+      "step": 50710
+    },
+    {
+      "epoch": 1.62304,
+      "grad_norm": 46.431365966796875,
+      "learning_rate": 4.188444444444445e-06,
+      "loss": 1.5954,
+      "step": 50720
+    },
+    {
+      "epoch": 1.62336,
+      "grad_norm": 47.1487922668457,
+      "learning_rate": 4.18488888888889e-06,
+      "loss": 1.6097,
+      "step": 50730
+    },
+    {
+      "epoch": 1.62368,
+      "grad_norm": 44.51396179199219,
+      "learning_rate": 4.1813333333333335e-06,
+      "loss": 1.6403,
+      "step": 50740
+    },
+    {
+      "epoch": 1.624,
+      "grad_norm": 45.483638763427734,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 1.622,
+      "step": 50750
+    },
+    {
+      "epoch": 1.62432,
+      "grad_norm": 46.74037551879883,
+      "learning_rate": 4.174222222222223e-06,
+      "loss": 1.5975,
+      "step": 50760
+    },
+    {
+      "epoch": 1.6246399999999999,
+      "grad_norm": 45.22315979003906,
+      "learning_rate": 4.170666666666667e-06,
+      "loss": 1.5898,
+      "step": 50770
+    },
+    {
+      "epoch": 1.62496,
+      "grad_norm": 47.88893508911133,
+      "learning_rate": 4.167111111111111e-06,
+      "loss": 1.5861,
+      "step": 50780
+    },
+    {
+      "epoch": 1.62528,
+      "grad_norm": 47.07771682739258,
+      "learning_rate": 4.1635555555555554e-06,
+      "loss": 1.5988,
+      "step": 50790
+    },
+    {
+      "epoch": 1.6256,
+      "grad_norm": 46.006805419921875,
+      "learning_rate": 4.16e-06,
+      "loss": 1.6015,
+      "step": 50800
+    },
+    {
+      "epoch": 1.62592,
+      "grad_norm": 45.821189880371094,
+      "learning_rate": 4.1564444444444445e-06,
+      "loss": 1.5748,
+      "step": 50810
+    },
+    {
+      "epoch": 1.6262400000000001,
+      "grad_norm": 44.78703308105469,
+      "learning_rate": 4.152888888888889e-06,
+      "loss": 1.6025,
+      "step": 50820
+    },
+    {
+      "epoch": 1.62656,
+      "grad_norm": 46.905548095703125,
+      "learning_rate": 4.149333333333334e-06,
+      "loss": 1.5806,
+      "step": 50830
+    },
+    {
+      "epoch": 1.6268799999999999,
+      "grad_norm": 45.533382415771484,
+      "learning_rate": 4.145777777777778e-06,
+      "loss": 1.5724,
+      "step": 50840
+    },
+    {
+      "epoch": 1.6272,
+      "grad_norm": 46.332847595214844,
+      "learning_rate": 4.142222222222223e-06,
+      "loss": 1.6117,
+      "step": 50850
+    },
+    {
+      "epoch": 1.62752,
+      "grad_norm": 47.08037567138672,
+      "learning_rate": 4.138666666666667e-06,
+      "loss": 1.6313,
+      "step": 50860
+    },
+    {
+      "epoch": 1.62784,
+      "grad_norm": 45.778968811035156,
+      "learning_rate": 4.135111111111111e-06,
+      "loss": 1.5783,
+      "step": 50870
+    },
+    {
+      "epoch": 1.62816,
+      "grad_norm": 46.12378692626953,
+      "learning_rate": 4.1315555555555556e-06,
+      "loss": 1.5866,
+      "step": 50880
+    },
+    {
+      "epoch": 1.6284800000000001,
+      "grad_norm": 46.11397933959961,
+      "learning_rate": 4.128e-06,
+      "loss": 1.5727,
+      "step": 50890
+    },
+    {
+      "epoch": 1.6288,
+      "grad_norm": 45.41218185424805,
+      "learning_rate": 4.124444444444445e-06,
+      "loss": 1.6054,
+      "step": 50900
+    },
+    {
+      "epoch": 1.62912,
+      "grad_norm": 44.79296875,
+      "learning_rate": 4.120888888888889e-06,
+      "loss": 1.5776,
+      "step": 50910
+    },
+    {
+      "epoch": 1.62944,
+      "grad_norm": 46.6500129699707,
+      "learning_rate": 4.117333333333334e-06,
+      "loss": 1.5955,
+      "step": 50920
+    },
+    {
+      "epoch": 1.62976,
+      "grad_norm": 44.73964309692383,
+      "learning_rate": 4.113777777777778e-06,
+      "loss": 1.6007,
+      "step": 50930
+    },
+    {
+      "epoch": 1.63008,
+      "grad_norm": 44.71598815917969,
+      "learning_rate": 4.110222222222223e-06,
+      "loss": 1.6068,
+      "step": 50940
+    },
+    {
+      "epoch": 1.6303999999999998,
+      "grad_norm": 45.50582504272461,
+      "learning_rate": 4.1066666666666674e-06,
+      "loss": 1.5937,
+      "step": 50950
+    },
+    {
+      "epoch": 1.63072,
+      "grad_norm": 47.319496154785156,
+      "learning_rate": 4.103111111111112e-06,
+      "loss": 1.5752,
+      "step": 50960
+    },
+    {
+      "epoch": 1.63104,
+      "grad_norm": 42.19738006591797,
+      "learning_rate": 4.099555555555556e-06,
+      "loss": 1.607,
+      "step": 50970
+    },
+    {
+      "epoch": 1.63136,
+      "grad_norm": 48.765106201171875,
+      "learning_rate": 4.096e-06,
+      "loss": 1.59,
+      "step": 50980
+    },
+    {
+      "epoch": 1.63168,
+      "grad_norm": 43.94348907470703,
+      "learning_rate": 4.092444444444445e-06,
+      "loss": 1.5813,
+      "step": 50990
+    },
+    {
+      "epoch": 1.6320000000000001,
+      "grad_norm": 46.9215087890625,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 1.6027,
+      "step": 51000
+    },
+    {
+      "epoch": 1.63232,
+      "grad_norm": 45.82568359375,
+      "learning_rate": 4.085333333333334e-06,
+      "loss": 1.5715,
+      "step": 51010
+    },
+    {
+      "epoch": 1.6326399999999999,
+      "grad_norm": 44.695281982421875,
+      "learning_rate": 4.081777777777778e-06,
+      "loss": 1.6268,
+      "step": 51020
+    },
+    {
+      "epoch": 1.63296,
+      "grad_norm": 47.35905456542969,
+      "learning_rate": 4.078222222222222e-06,
+      "loss": 1.5944,
+      "step": 51030
+    },
+    {
+      "epoch": 1.63328,
+      "grad_norm": 46.44203567504883,
+      "learning_rate": 4.074666666666667e-06,
+      "loss": 1.5942,
+      "step": 51040
+    },
+    {
+      "epoch": 1.6336,
+      "grad_norm": 46.74188995361328,
+      "learning_rate": 4.071111111111111e-06,
+      "loss": 1.6059,
+      "step": 51050
+    },
+    {
+      "epoch": 1.63392,
+      "grad_norm": 45.445526123046875,
+      "learning_rate": 4.067555555555556e-06,
+      "loss": 1.6191,
+      "step": 51060
+    },
+    {
+      "epoch": 1.6342400000000001,
+      "grad_norm": 43.05592346191406,
+      "learning_rate": 4.064e-06,
+      "loss": 1.5822,
+      "step": 51070
+    },
+    {
+      "epoch": 1.63456,
+      "grad_norm": 47.14398193359375,
+      "learning_rate": 4.060444444444445e-06,
+      "loss": 1.571,
+      "step": 51080
+    },
+    {
+      "epoch": 1.6348799999999999,
+      "grad_norm": 44.45650100708008,
+      "learning_rate": 4.0568888888888895e-06,
+      "loss": 1.5836,
+      "step": 51090
+    },
+    {
+      "epoch": 1.6352,
+      "grad_norm": 46.076725006103516,
+      "learning_rate": 4.053333333333333e-06,
+      "loss": 1.5917,
+      "step": 51100
+    },
+    {
+      "epoch": 1.63552,
+      "grad_norm": 48.551517486572266,
+      "learning_rate": 4.049777777777778e-06,
+      "loss": 1.5996,
+      "step": 51110
+    },
+    {
+      "epoch": 1.63584,
+      "grad_norm": 46.424808502197266,
+      "learning_rate": 4.046222222222222e-06,
+      "loss": 1.6065,
+      "step": 51120
+    },
+    {
+      "epoch": 1.6361599999999998,
+      "grad_norm": 44.893001556396484,
+      "learning_rate": 4.042666666666667e-06,
+      "loss": 1.5853,
+      "step": 51130
+    },
+    {
+      "epoch": 1.6364800000000002,
+      "grad_norm": 45.07635498046875,
+      "learning_rate": 4.039111111111111e-06,
+      "loss": 1.5802,
+      "step": 51140
+    },
+    {
+      "epoch": 1.6368,
+      "grad_norm": 45.24571990966797,
+      "learning_rate": 4.035555555555556e-06,
+      "loss": 1.6256,
+      "step": 51150
+    },
+    {
+      "epoch": 1.63712,
+      "grad_norm": 47.38965606689453,
+      "learning_rate": 4.0320000000000005e-06,
+      "loss": 1.6005,
+      "step": 51160
+    },
+    {
+      "epoch": 1.63744,
+      "grad_norm": 44.441802978515625,
+      "learning_rate": 4.028444444444445e-06,
+      "loss": 1.595,
+      "step": 51170
+    },
+    {
+      "epoch": 1.63776,
+      "grad_norm": 45.896514892578125,
+      "learning_rate": 4.02488888888889e-06,
+      "loss": 1.5747,
+      "step": 51180
+    },
+    {
+      "epoch": 1.63808,
+      "grad_norm": 45.57615280151367,
+      "learning_rate": 4.021333333333333e-06,
+      "loss": 1.5843,
+      "step": 51190
+    },
+    {
+      "epoch": 1.6383999999999999,
+      "grad_norm": 44.5887336730957,
+      "learning_rate": 4.017777777777778e-06,
+      "loss": 1.5938,
+      "step": 51200
+    },
+    {
+      "epoch": 1.63872,
+      "grad_norm": 44.419898986816406,
+      "learning_rate": 4.0142222222222225e-06,
+      "loss": 1.5938,
+      "step": 51210
+    },
+    {
+      "epoch": 1.63904,
+      "grad_norm": 46.093406677246094,
+      "learning_rate": 4.010666666666667e-06,
+      "loss": 1.5766,
+      "step": 51220
+    },
+    {
+      "epoch": 1.63936,
+      "grad_norm": 44.555179595947266,
+      "learning_rate": 4.0071111111111116e-06,
+      "loss": 1.6233,
+      "step": 51230
+    },
+    {
+      "epoch": 1.63968,
+      "grad_norm": 45.38337707519531,
+      "learning_rate": 4.003555555555556e-06,
+      "loss": 1.5829,
+      "step": 51240
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 47.5709114074707,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5673,
+      "step": 51250
+    },
+    {
+      "epoch": 1.64032,
+      "grad_norm": 46.36149597167969,
+      "learning_rate": 3.996444444444445e-06,
+      "loss": 1.5784,
+      "step": 51260
+    },
+    {
+      "epoch": 1.6406399999999999,
+      "grad_norm": 43.161376953125,
+      "learning_rate": 3.99288888888889e-06,
+      "loss": 1.5826,
+      "step": 51270
+    },
+    {
+      "epoch": 1.64096,
+      "grad_norm": 44.26531219482422,
+      "learning_rate": 3.9893333333333335e-06,
+      "loss": 1.5945,
+      "step": 51280
+    },
+    {
+      "epoch": 1.64128,
+      "grad_norm": 46.19755172729492,
+      "learning_rate": 3.985777777777778e-06,
+      "loss": 1.6204,
+      "step": 51290
+    },
+    {
+      "epoch": 1.6416,
+      "grad_norm": 45.3082275390625,
+      "learning_rate": 3.982222222222223e-06,
+      "loss": 1.6296,
+      "step": 51300
+    },
+    {
+      "epoch": 1.64192,
+      "grad_norm": 46.39171600341797,
+      "learning_rate": 3.978666666666667e-06,
+      "loss": 1.5551,
+      "step": 51310
+    },
+    {
+      "epoch": 1.6422400000000001,
+      "grad_norm": 46.921443939208984,
+      "learning_rate": 3.975111111111111e-06,
+      "loss": 1.5899,
+      "step": 51320
+    },
+    {
+      "epoch": 1.64256,
+      "grad_norm": 45.100067138671875,
+      "learning_rate": 3.971555555555555e-06,
+      "loss": 1.5889,
+      "step": 51330
+    },
+    {
+      "epoch": 1.64288,
+      "grad_norm": 46.697940826416016,
+      "learning_rate": 3.968e-06,
+      "loss": 1.5864,
+      "step": 51340
+    },
+    {
+      "epoch": 1.6432,
+      "grad_norm": 45.882144927978516,
+      "learning_rate": 3.9644444444444445e-06,
+      "loss": 1.6122,
+      "step": 51350
+    },
+    {
+      "epoch": 1.64352,
+      "grad_norm": 43.447383880615234,
+      "learning_rate": 3.960888888888889e-06,
+      "loss": 1.5687,
+      "step": 51360
+    },
+    {
+      "epoch": 1.64384,
+      "grad_norm": 44.602203369140625,
+      "learning_rate": 3.957333333333334e-06,
+      "loss": 1.5972,
+      "step": 51370
+    },
+    {
+      "epoch": 1.6441599999999998,
+      "grad_norm": 46.61246109008789,
+      "learning_rate": 3.953777777777778e-06,
+      "loss": 1.6087,
+      "step": 51380
+    },
+    {
+      "epoch": 1.6444800000000002,
+      "grad_norm": 43.60810852050781,
+      "learning_rate": 3.950222222222223e-06,
+      "loss": 1.5684,
+      "step": 51390
+    },
+    {
+      "epoch": 1.6448,
+      "grad_norm": 46.67555236816406,
+      "learning_rate": 3.946666666666667e-06,
+      "loss": 1.593,
+      "step": 51400
+    },
+    {
+      "epoch": 1.64512,
+      "grad_norm": 46.29722213745117,
+      "learning_rate": 3.943111111111112e-06,
+      "loss": 1.5815,
+      "step": 51410
+    },
+    {
+      "epoch": 1.64544,
+      "grad_norm": 45.6417121887207,
+      "learning_rate": 3.9395555555555555e-06,
+      "loss": 1.5646,
+      "step": 51420
+    },
+    {
+      "epoch": 1.6457600000000001,
+      "grad_norm": 45.40904235839844,
+      "learning_rate": 3.936e-06,
+      "loss": 1.5829,
+      "step": 51430
+    },
+    {
+      "epoch": 1.64608,
+      "grad_norm": 45.21997833251953,
+      "learning_rate": 3.932444444444445e-06,
+      "loss": 1.5967,
+      "step": 51440
+    },
+    {
+      "epoch": 1.6463999999999999,
+      "grad_norm": 44.75946044921875,
+      "learning_rate": 3.928888888888889e-06,
+      "loss": 1.5784,
+      "step": 51450
+    },
+    {
+      "epoch": 1.64672,
+      "grad_norm": 45.29618835449219,
+      "learning_rate": 3.925333333333334e-06,
+      "loss": 1.5942,
+      "step": 51460
+    },
+    {
+      "epoch": 1.64704,
+      "grad_norm": 45.929935455322266,
+      "learning_rate": 3.921777777777778e-06,
+      "loss": 1.6064,
+      "step": 51470
+    },
+    {
+      "epoch": 1.64736,
+      "grad_norm": 45.438934326171875,
+      "learning_rate": 3.918222222222223e-06,
+      "loss": 1.5798,
+      "step": 51480
+    },
+    {
+      "epoch": 1.64768,
+      "grad_norm": 47.577274322509766,
+      "learning_rate": 3.914666666666667e-06,
+      "loss": 1.6149,
+      "step": 51490
+    },
+    {
+      "epoch": 1.6480000000000001,
+      "grad_norm": 45.842472076416016,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 1.6094,
+      "step": 51500
+    },
+    {
+      "epoch": 1.64832,
+      "grad_norm": 45.71033477783203,
+      "learning_rate": 3.9075555555555565e-06,
+      "loss": 1.5777,
+      "step": 51510
+    },
+    {
+      "epoch": 1.6486399999999999,
+      "grad_norm": 47.43943405151367,
+      "learning_rate": 3.904e-06,
+      "loss": 1.605,
+      "step": 51520
+    },
+    {
+      "epoch": 1.64896,
+      "grad_norm": 45.22209167480469,
+      "learning_rate": 3.900444444444445e-06,
+      "loss": 1.6268,
+      "step": 51530
+    },
+    {
+      "epoch": 1.64928,
+      "grad_norm": 47.26148223876953,
+      "learning_rate": 3.896888888888889e-06,
+      "loss": 1.5841,
+      "step": 51540
+    },
+    {
+      "epoch": 1.6496,
+      "grad_norm": 45.25746536254883,
+      "learning_rate": 3.893333333333333e-06,
+      "loss": 1.584,
+      "step": 51550
+    },
+    {
+      "epoch": 1.64992,
+      "grad_norm": 45.87767791748047,
+      "learning_rate": 3.889777777777778e-06,
+      "loss": 1.5847,
+      "step": 51560
+    },
+    {
+      "epoch": 1.6502400000000002,
+      "grad_norm": 44.278568267822266,
+      "learning_rate": 3.886222222222222e-06,
+      "loss": 1.6103,
+      "step": 51570
+    },
+    {
+      "epoch": 1.65056,
+      "grad_norm": 45.72529983520508,
+      "learning_rate": 3.882666666666667e-06,
+      "loss": 1.6127,
+      "step": 51580
+    },
+    {
+      "epoch": 1.65088,
+      "grad_norm": 44.123146057128906,
+      "learning_rate": 3.879111111111111e-06,
+      "loss": 1.5901,
+      "step": 51590
+    },
+    {
+      "epoch": 1.6512,
+      "grad_norm": 45.891361236572266,
+      "learning_rate": 3.875555555555556e-06,
+      "loss": 1.5728,
+      "step": 51600
+    },
+    {
+      "epoch": 1.65152,
+      "grad_norm": 46.983604431152344,
+      "learning_rate": 3.872e-06,
+      "loss": 1.6013,
+      "step": 51610
+    },
+    {
+      "epoch": 1.65184,
+      "grad_norm": 45.0738525390625,
+      "learning_rate": 3.868444444444445e-06,
+      "loss": 1.6019,
+      "step": 51620
+    },
+    {
+      "epoch": 1.6521599999999999,
+      "grad_norm": 45.97278594970703,
+      "learning_rate": 3.8648888888888895e-06,
+      "loss": 1.5751,
+      "step": 51630
+    },
+    {
+      "epoch": 1.65248,
+      "grad_norm": 44.256500244140625,
+      "learning_rate": 3.861333333333333e-06,
+      "loss": 1.5589,
+      "step": 51640
+    },
+    {
+      "epoch": 1.6528,
+      "grad_norm": 45.32353973388672,
+      "learning_rate": 3.857777777777778e-06,
+      "loss": 1.5763,
+      "step": 51650
+    },
+    {
+      "epoch": 1.65312,
+      "grad_norm": 45.695960998535156,
+      "learning_rate": 3.854222222222222e-06,
+      "loss": 1.567,
+      "step": 51660
+    },
+    {
+      "epoch": 1.65344,
+      "grad_norm": 43.78841781616211,
+      "learning_rate": 3.850666666666667e-06,
+      "loss": 1.5825,
+      "step": 51670
+    },
+    {
+      "epoch": 1.6537600000000001,
+      "grad_norm": 45.8950080871582,
+      "learning_rate": 3.847111111111111e-06,
+      "loss": 1.6075,
+      "step": 51680
+    },
+    {
+      "epoch": 1.65408,
+      "grad_norm": 46.352874755859375,
+      "learning_rate": 3.843555555555556e-06,
+      "loss": 1.62,
+      "step": 51690
+    },
+    {
+      "epoch": 1.6543999999999999,
+      "grad_norm": 43.862728118896484,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.6047,
+      "step": 51700
+    },
+    {
+      "epoch": 1.65472,
+      "grad_norm": 46.20279312133789,
+      "learning_rate": 3.836444444444445e-06,
+      "loss": 1.5958,
+      "step": 51710
+    },
+    {
+      "epoch": 1.65504,
+      "grad_norm": 47.41118621826172,
+      "learning_rate": 3.83288888888889e-06,
+      "loss": 1.6116,
+      "step": 51720
+    },
+    {
+      "epoch": 1.65536,
+      "grad_norm": 44.2399787902832,
+      "learning_rate": 3.829333333333334e-06,
+      "loss": 1.6,
+      "step": 51730
+    },
+    {
+      "epoch": 1.65568,
+      "grad_norm": 45.41532897949219,
+      "learning_rate": 3.825777777777778e-06,
+      "loss": 1.5894,
+      "step": 51740
+    },
+    {
+      "epoch": 1.6560000000000001,
+      "grad_norm": 49.13901901245117,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 1.5931,
+      "step": 51750
+    },
+    {
+      "epoch": 1.65632,
+      "grad_norm": 46.375240325927734,
+      "learning_rate": 3.818666666666667e-06,
+      "loss": 1.5792,
+      "step": 51760
+    },
+    {
+      "epoch": 1.65664,
+      "grad_norm": 43.16218185424805,
+      "learning_rate": 3.8151111111111115e-06,
+      "loss": 1.6248,
+      "step": 51770
+    },
+    {
+      "epoch": 1.65696,
+      "grad_norm": 47.00339889526367,
+      "learning_rate": 3.811555555555556e-06,
+      "loss": 1.5988,
+      "step": 51780
+    },
+    {
+      "epoch": 1.65728,
+      "grad_norm": 43.82436752319336,
+      "learning_rate": 3.8080000000000006e-06,
+      "loss": 1.592,
+      "step": 51790
+    },
+    {
+      "epoch": 1.6576,
+      "grad_norm": 47.84590530395508,
+      "learning_rate": 3.8044444444444443e-06,
+      "loss": 1.5862,
+      "step": 51800
+    },
+    {
+      "epoch": 1.6579199999999998,
+      "grad_norm": 45.108680725097656,
+      "learning_rate": 3.800888888888889e-06,
+      "loss": 1.5799,
+      "step": 51810
+    },
+    {
+      "epoch": 1.6582400000000002,
+      "grad_norm": 45.71805953979492,
+      "learning_rate": 3.7973333333333335e-06,
+      "loss": 1.6113,
+      "step": 51820
+    },
+    {
+      "epoch": 1.65856,
+      "grad_norm": 46.040164947509766,
+      "learning_rate": 3.793777777777778e-06,
+      "loss": 1.5847,
+      "step": 51830
+    },
+    {
+      "epoch": 1.65888,
+      "grad_norm": 45.32584762573242,
+      "learning_rate": 3.7902222222222226e-06,
+      "loss": 1.5995,
+      "step": 51840
+    },
+    {
+      "epoch": 1.6592,
+      "grad_norm": 44.63491439819336,
+      "learning_rate": 3.7866666666666667e-06,
+      "loss": 1.6002,
+      "step": 51850
+    },
+    {
+      "epoch": 1.65952,
+      "grad_norm": 46.6467399597168,
+      "learning_rate": 3.7831111111111112e-06,
+      "loss": 1.599,
+      "step": 51860
+    },
+    {
+      "epoch": 1.65984,
+      "grad_norm": 44.993499755859375,
+      "learning_rate": 3.779555555555556e-06,
+      "loss": 1.6072,
+      "step": 51870
+    },
+    {
+      "epoch": 1.6601599999999999,
+      "grad_norm": 45.52958679199219,
+      "learning_rate": 3.7760000000000004e-06,
+      "loss": 1.5949,
+      "step": 51880
+    },
+    {
+      "epoch": 1.66048,
+      "grad_norm": 45.84259033203125,
+      "learning_rate": 3.7724444444444445e-06,
+      "loss": 1.5956,
+      "step": 51890
+    },
+    {
+      "epoch": 1.6608,
+      "grad_norm": 46.66897964477539,
+      "learning_rate": 3.768888888888889e-06,
+      "loss": 1.6052,
+      "step": 51900
+    },
+    {
+      "epoch": 1.66112,
+      "grad_norm": 45.33606719970703,
+      "learning_rate": 3.7653333333333336e-06,
+      "loss": 1.6023,
+      "step": 51910
+    },
+    {
+      "epoch": 1.66144,
+      "grad_norm": 45.42198944091797,
+      "learning_rate": 3.761777777777778e-06,
+      "loss": 1.5922,
+      "step": 51920
+    },
+    {
+      "epoch": 1.6617600000000001,
+      "grad_norm": 46.46091842651367,
+      "learning_rate": 3.7582222222222227e-06,
+      "loss": 1.6021,
+      "step": 51930
+    },
+    {
+      "epoch": 1.66208,
+      "grad_norm": 44.122169494628906,
+      "learning_rate": 3.754666666666667e-06,
+      "loss": 1.6073,
+      "step": 51940
+    },
+    {
+      "epoch": 1.6623999999999999,
+      "grad_norm": 44.32771682739258,
+      "learning_rate": 3.7511111111111114e-06,
+      "loss": 1.5977,
+      "step": 51950
+    },
+    {
+      "epoch": 1.66272,
+      "grad_norm": 44.800567626953125,
+      "learning_rate": 3.747555555555556e-06,
+      "loss": 1.5832,
+      "step": 51960
+    },
+    {
+      "epoch": 1.66304,
+      "grad_norm": 45.98767852783203,
+      "learning_rate": 3.7440000000000005e-06,
+      "loss": 1.5969,
+      "step": 51970
+    },
+    {
+      "epoch": 1.66336,
+      "grad_norm": 45.25199890136719,
+      "learning_rate": 3.740444444444445e-06,
+      "loss": 1.5853,
+      "step": 51980
+    },
+    {
+      "epoch": 1.66368,
+      "grad_norm": 44.7115478515625,
+      "learning_rate": 3.736888888888889e-06,
+      "loss": 1.5913,
+      "step": 51990
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 46.41364288330078,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.5784,
+      "step": 52000
+    },
+    {
+      "epoch": 1.66432,
+      "grad_norm": 45.47352981567383,
+      "learning_rate": 3.7297777777777783e-06,
+      "loss": 1.5906,
+      "step": 52010
+    },
+    {
+      "epoch": 1.66464,
+      "grad_norm": 44.713462829589844,
+      "learning_rate": 3.726222222222223e-06,
+      "loss": 1.5641,
+      "step": 52020
+    },
+    {
+      "epoch": 1.66496,
+      "grad_norm": 43.92695999145508,
+      "learning_rate": 3.7226666666666674e-06,
+      "loss": 1.5806,
+      "step": 52030
+    },
+    {
+      "epoch": 1.66528,
+      "grad_norm": 44.497310638427734,
+      "learning_rate": 3.7191111111111115e-06,
+      "loss": 1.593,
+      "step": 52040
+    },
+    {
+      "epoch": 1.6656,
+      "grad_norm": 44.12403106689453,
+      "learning_rate": 3.7155555555555557e-06,
+      "loss": 1.5966,
+      "step": 52050
+    },
+    {
+      "epoch": 1.6659199999999998,
+      "grad_norm": 46.343746185302734,
+      "learning_rate": 3.712e-06,
+      "loss": 1.5958,
+      "step": 52060
+    },
+    {
+      "epoch": 1.6662400000000002,
+      "grad_norm": 44.707275390625,
+      "learning_rate": 3.7084444444444443e-06,
+      "loss": 1.5921,
+      "step": 52070
+    },
+    {
+      "epoch": 1.66656,
+      "grad_norm": 44.39040756225586,
+      "learning_rate": 3.704888888888889e-06,
+      "loss": 1.5704,
+      "step": 52080
+    },
+    {
+      "epoch": 1.66688,
+      "grad_norm": 45.53627395629883,
+      "learning_rate": 3.7013333333333334e-06,
+      "loss": 1.5879,
+      "step": 52090
+    },
+    {
+      "epoch": 1.6672,
+      "grad_norm": 46.279083251953125,
+      "learning_rate": 3.697777777777778e-06,
+      "loss": 1.6016,
+      "step": 52100
+    },
+    {
+      "epoch": 1.6675200000000001,
+      "grad_norm": 44.5371208190918,
+      "learning_rate": 3.6942222222222226e-06,
+      "loss": 1.5677,
+      "step": 52110
+    },
+    {
+      "epoch": 1.66784,
+      "grad_norm": 42.929622650146484,
+      "learning_rate": 3.6906666666666667e-06,
+      "loss": 1.5781,
+      "step": 52120
+    },
+    {
+      "epoch": 1.6681599999999999,
+      "grad_norm": 44.74525833129883,
+      "learning_rate": 3.6871111111111112e-06,
+      "loss": 1.5949,
+      "step": 52130
+    },
+    {
+      "epoch": 1.66848,
+      "grad_norm": 43.849388122558594,
+      "learning_rate": 3.6835555555555558e-06,
+      "loss": 1.5873,
+      "step": 52140
+    },
+    {
+      "epoch": 1.6688,
+      "grad_norm": 45.275665283203125,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 1.5767,
+      "step": 52150
+    },
+    {
+      "epoch": 1.66912,
+      "grad_norm": 45.785057067871094,
+      "learning_rate": 3.676444444444445e-06,
+      "loss": 1.5831,
+      "step": 52160
+    },
+    {
+      "epoch": 1.66944,
+      "grad_norm": 46.84950256347656,
+      "learning_rate": 3.672888888888889e-06,
+      "loss": 1.5949,
+      "step": 52170
+    },
+    {
+      "epoch": 1.6697600000000001,
+      "grad_norm": 43.389034271240234,
+      "learning_rate": 3.6693333333333336e-06,
+      "loss": 1.5767,
+      "step": 52180
+    },
+    {
+      "epoch": 1.67008,
+      "grad_norm": 43.85615921020508,
+      "learning_rate": 3.665777777777778e-06,
+      "loss": 1.601,
+      "step": 52190
+    },
+    {
+      "epoch": 1.6703999999999999,
+      "grad_norm": 44.621917724609375,
+      "learning_rate": 3.6622222222222227e-06,
+      "loss": 1.5791,
+      "step": 52200
+    },
+    {
+      "epoch": 1.67072,
+      "grad_norm": 45.9107551574707,
+      "learning_rate": 3.6586666666666672e-06,
+      "loss": 1.5946,
+      "step": 52210
+    },
+    {
+      "epoch": 1.67104,
+      "grad_norm": 46.91019058227539,
+      "learning_rate": 3.6551111111111114e-06,
+      "loss": 1.5811,
+      "step": 52220
+    },
+    {
+      "epoch": 1.67136,
+      "grad_norm": 43.92897033691406,
+      "learning_rate": 3.651555555555556e-06,
+      "loss": 1.5966,
+      "step": 52230
+    },
+    {
+      "epoch": 1.67168,
+      "grad_norm": 48.71943283081055,
+      "learning_rate": 3.6480000000000005e-06,
+      "loss": 1.591,
+      "step": 52240
+    },
+    {
+      "epoch": 1.6720000000000002,
+      "grad_norm": 44.170555114746094,
+      "learning_rate": 3.644444444444445e-06,
+      "loss": 1.5861,
+      "step": 52250
+    },
+    {
+      "epoch": 1.67232,
+      "grad_norm": 44.75503158569336,
+      "learning_rate": 3.640888888888889e-06,
+      "loss": 1.5603,
+      "step": 52260
+    },
+    {
+      "epoch": 1.67264,
+      "grad_norm": 45.242271423339844,
+      "learning_rate": 3.6373333333333337e-06,
+      "loss": 1.585,
+      "step": 52270
+    },
+    {
+      "epoch": 1.67296,
+      "grad_norm": 45.12602233886719,
+      "learning_rate": 3.6337777777777783e-06,
+      "loss": 1.6057,
+      "step": 52280
+    },
+    {
+      "epoch": 1.67328,
+      "grad_norm": 45.389461517333984,
+      "learning_rate": 3.630222222222223e-06,
+      "loss": 1.5982,
+      "step": 52290
+    },
+    {
+      "epoch": 1.6736,
+      "grad_norm": 44.52558898925781,
+      "learning_rate": 3.6266666666666674e-06,
+      "loss": 1.5794,
+      "step": 52300
+    },
+    {
+      "epoch": 1.6739199999999999,
+      "grad_norm": 44.75640869140625,
+      "learning_rate": 3.623111111111111e-06,
+      "loss": 1.5652,
+      "step": 52310
+    },
+    {
+      "epoch": 1.67424,
+      "grad_norm": 46.04345703125,
+      "learning_rate": 3.6195555555555556e-06,
+      "loss": 1.5881,
+      "step": 52320
+    },
+    {
+      "epoch": 1.67456,
+      "grad_norm": 44.973365783691406,
+      "learning_rate": 3.616e-06,
+      "loss": 1.5792,
+      "step": 52330
+    },
+    {
+      "epoch": 1.67488,
+      "grad_norm": 47.327857971191406,
+      "learning_rate": 3.6124444444444443e-06,
+      "loss": 1.586,
+      "step": 52340
+    },
+    {
+      "epoch": 1.6752,
+      "grad_norm": 44.85529708862305,
+      "learning_rate": 3.608888888888889e-06,
+      "loss": 1.5715,
+      "step": 52350
+    },
+    {
+      "epoch": 1.6755200000000001,
+      "grad_norm": 44.864768981933594,
+      "learning_rate": 3.6053333333333334e-06,
+      "loss": 1.6091,
+      "step": 52360
+    },
+    {
+      "epoch": 1.67584,
+      "grad_norm": 46.082679748535156,
+      "learning_rate": 3.601777777777778e-06,
+      "loss": 1.5772,
+      "step": 52370
+    },
+    {
+      "epoch": 1.6761599999999999,
+      "grad_norm": 45.27592086791992,
+      "learning_rate": 3.5982222222222225e-06,
+      "loss": 1.6049,
+      "step": 52380
+    },
+    {
+      "epoch": 1.67648,
+      "grad_norm": 45.963401794433594,
+      "learning_rate": 3.5946666666666667e-06,
+      "loss": 1.6017,
+      "step": 52390
+    },
+    {
+      "epoch": 1.6768,
+      "grad_norm": 46.299800872802734,
+      "learning_rate": 3.5911111111111112e-06,
+      "loss": 1.5578,
+      "step": 52400
+    },
+    {
+      "epoch": 1.67712,
+      "grad_norm": 46.58485794067383,
+      "learning_rate": 3.5875555555555558e-06,
+      "loss": 1.5812,
+      "step": 52410
+    },
+    {
+      "epoch": 1.67744,
+      "grad_norm": 44.165306091308594,
+      "learning_rate": 3.5840000000000003e-06,
+      "loss": 1.5728,
+      "step": 52420
+    },
+    {
+      "epoch": 1.6777600000000001,
+      "grad_norm": 45.76871109008789,
+      "learning_rate": 3.580444444444445e-06,
+      "loss": 1.5943,
+      "step": 52430
+    },
+    {
+      "epoch": 1.67808,
+      "grad_norm": 46.34144592285156,
+      "learning_rate": 3.576888888888889e-06,
+      "loss": 1.6141,
+      "step": 52440
+    },
+    {
+      "epoch": 1.6784,
+      "grad_norm": 44.653202056884766,
+      "learning_rate": 3.5733333333333336e-06,
+      "loss": 1.5804,
+      "step": 52450
+    },
+    {
+      "epoch": 1.67872,
+      "grad_norm": 44.92979431152344,
+      "learning_rate": 3.569777777777778e-06,
+      "loss": 1.6015,
+      "step": 52460
+    },
+    {
+      "epoch": 1.67904,
+      "grad_norm": 47.18275451660156,
+      "learning_rate": 3.5662222222222227e-06,
+      "loss": 1.6068,
+      "step": 52470
+    },
+    {
+      "epoch": 1.67936,
+      "grad_norm": 44.27334976196289,
+      "learning_rate": 3.5626666666666672e-06,
+      "loss": 1.5736,
+      "step": 52480
+    },
+    {
+      "epoch": 1.6796799999999998,
+      "grad_norm": 44.15816116333008,
+      "learning_rate": 3.5591111111111114e-06,
+      "loss": 1.6013,
+      "step": 52490
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 46.573646545410156,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 1.6026,
+      "step": 52500
+    },
+    {
+      "epoch": 1.68032,
+      "grad_norm": 46.70231628417969,
+      "learning_rate": 3.5520000000000005e-06,
+      "loss": 1.6064,
+      "step": 52510
+    },
+    {
+      "epoch": 1.68064,
+      "grad_norm": 47.547264099121094,
+      "learning_rate": 3.548444444444445e-06,
+      "loss": 1.5905,
+      "step": 52520
+    },
+    {
+      "epoch": 1.68096,
+      "grad_norm": 46.42538070678711,
+      "learning_rate": 3.5448888888888896e-06,
+      "loss": 1.5743,
+      "step": 52530
+    },
+    {
+      "epoch": 1.68128,
+      "grad_norm": 47.20460891723633,
+      "learning_rate": 3.5413333333333337e-06,
+      "loss": 1.6122,
+      "step": 52540
+    },
+    {
+      "epoch": 1.6816,
+      "grad_norm": 46.976173400878906,
+      "learning_rate": 3.5377777777777783e-06,
+      "loss": 1.6072,
+      "step": 52550
+    },
+    {
+      "epoch": 1.6819199999999999,
+      "grad_norm": 45.38555145263672,
+      "learning_rate": 3.534222222222223e-06,
+      "loss": 1.6041,
+      "step": 52560
+    },
+    {
+      "epoch": 1.68224,
+      "grad_norm": 44.77622985839844,
+      "learning_rate": 3.5306666666666665e-06,
+      "loss": 1.5783,
+      "step": 52570
+    },
+    {
+      "epoch": 1.68256,
+      "grad_norm": 46.87044906616211,
+      "learning_rate": 3.527111111111111e-06,
+      "loss": 1.5894,
+      "step": 52580
+    },
+    {
+      "epoch": 1.68288,
+      "grad_norm": 42.50729751586914,
+      "learning_rate": 3.5235555555555556e-06,
+      "loss": 1.5774,
+      "step": 52590
+    },
+    {
+      "epoch": 1.6832,
+      "grad_norm": 45.89781951904297,
+      "learning_rate": 3.52e-06,
+      "loss": 1.5917,
+      "step": 52600
+    },
+    {
+      "epoch": 1.6835200000000001,
+      "grad_norm": 45.55253601074219,
+      "learning_rate": 3.5164444444444447e-06,
+      "loss": 1.6051,
+      "step": 52610
+    },
+    {
+      "epoch": 1.68384,
+      "grad_norm": 45.637691497802734,
+      "learning_rate": 3.512888888888889e-06,
+      "loss": 1.5723,
+      "step": 52620
+    },
+    {
+      "epoch": 1.6841599999999999,
+      "grad_norm": 46.54056167602539,
+      "learning_rate": 3.5093333333333334e-06,
+      "loss": 1.6018,
+      "step": 52630
+    },
+    {
+      "epoch": 1.68448,
+      "grad_norm": 45.923065185546875,
+      "learning_rate": 3.505777777777778e-06,
+      "loss": 1.5924,
+      "step": 52640
+    },
+    {
+      "epoch": 1.6848,
+      "grad_norm": 45.01921844482422,
+      "learning_rate": 3.5022222222222225e-06,
+      "loss": 1.6038,
+      "step": 52650
+    },
+    {
+      "epoch": 1.68512,
+      "grad_norm": 46.71082305908203,
+      "learning_rate": 3.498666666666667e-06,
+      "loss": 1.6226,
+      "step": 52660
+    },
+    {
+      "epoch": 1.68544,
+      "grad_norm": 43.84824752807617,
+      "learning_rate": 3.495111111111111e-06,
+      "loss": 1.5696,
+      "step": 52670
+    },
+    {
+      "epoch": 1.6857600000000001,
+      "grad_norm": 47.342376708984375,
+      "learning_rate": 3.4915555555555558e-06,
+      "loss": 1.5872,
+      "step": 52680
+    },
+    {
+      "epoch": 1.68608,
+      "grad_norm": 43.14659118652344,
+      "learning_rate": 3.4880000000000003e-06,
+      "loss": 1.5725,
+      "step": 52690
+    },
+    {
+      "epoch": 1.6864,
+      "grad_norm": 47.94723129272461,
+      "learning_rate": 3.484444444444445e-06,
+      "loss": 1.576,
+      "step": 52700
+    },
+    {
+      "epoch": 1.68672,
+      "grad_norm": 48.267154693603516,
+      "learning_rate": 3.480888888888889e-06,
+      "loss": 1.611,
+      "step": 52710
+    },
+    {
+      "epoch": 1.68704,
+      "grad_norm": 46.32544708251953,
+      "learning_rate": 3.4773333333333336e-06,
+      "loss": 1.5865,
+      "step": 52720
+    },
+    {
+      "epoch": 1.68736,
+      "grad_norm": 46.21999740600586,
+      "learning_rate": 3.473777777777778e-06,
+      "loss": 1.6158,
+      "step": 52730
+    },
+    {
+      "epoch": 1.6876799999999998,
+      "grad_norm": 46.073204040527344,
+      "learning_rate": 3.4702222222222227e-06,
+      "loss": 1.6152,
+      "step": 52740
+    },
+    {
+      "epoch": 1.688,
+      "grad_norm": 47.04712677001953,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.6032,
+      "step": 52750
+    },
+    {
+      "epoch": 1.68832,
+      "grad_norm": 46.21564483642578,
+      "learning_rate": 3.4631111111111113e-06,
+      "loss": 1.589,
+      "step": 52760
+    },
+    {
+      "epoch": 1.68864,
+      "grad_norm": 44.94911193847656,
+      "learning_rate": 3.459555555555556e-06,
+      "loss": 1.5702,
+      "step": 52770
+    },
+    {
+      "epoch": 1.68896,
+      "grad_norm": 46.054893493652344,
+      "learning_rate": 3.4560000000000005e-06,
+      "loss": 1.5829,
+      "step": 52780
+    },
+    {
+      "epoch": 1.6892800000000001,
+      "grad_norm": 45.82691955566406,
+      "learning_rate": 3.452444444444445e-06,
+      "loss": 1.6029,
+      "step": 52790
+    },
+    {
+      "epoch": 1.6896,
+      "grad_norm": 45.559146881103516,
+      "learning_rate": 3.4488888888888896e-06,
+      "loss": 1.5699,
+      "step": 52800
+    },
+    {
+      "epoch": 1.6899199999999999,
+      "grad_norm": 44.324974060058594,
+      "learning_rate": 3.4453333333333337e-06,
+      "loss": 1.5822,
+      "step": 52810
+    },
+    {
+      "epoch": 1.69024,
+      "grad_norm": 49.79753875732422,
+      "learning_rate": 3.4417777777777782e-06,
+      "loss": 1.591,
+      "step": 52820
+    },
+    {
+      "epoch": 1.69056,
+      "grad_norm": 46.32487487792969,
+      "learning_rate": 3.4382222222222224e-06,
+      "loss": 1.5929,
+      "step": 52830
+    },
+    {
+      "epoch": 1.69088,
+      "grad_norm": 47.12722396850586,
+      "learning_rate": 3.4346666666666665e-06,
+      "loss": 1.5979,
+      "step": 52840
+    },
+    {
+      "epoch": 1.6912,
+      "grad_norm": 45.32570266723633,
+      "learning_rate": 3.431111111111111e-06,
+      "loss": 1.5673,
+      "step": 52850
+    },
+    {
+      "epoch": 1.6915200000000001,
+      "grad_norm": 46.18455505371094,
+      "learning_rate": 3.4275555555555556e-06,
+      "loss": 1.6429,
+      "step": 52860
+    },
+    {
+      "epoch": 1.69184,
+      "grad_norm": 45.9107780456543,
+      "learning_rate": 3.424e-06,
+      "loss": 1.6182,
+      "step": 52870
+    },
+    {
+      "epoch": 1.6921599999999999,
+      "grad_norm": 45.0168342590332,
+      "learning_rate": 3.4204444444444447e-06,
+      "loss": 1.5625,
+      "step": 52880
+    },
+    {
+      "epoch": 1.69248,
+      "grad_norm": 45.60793685913086,
+      "learning_rate": 3.416888888888889e-06,
+      "loss": 1.5905,
+      "step": 52890
+    },
+    {
+      "epoch": 1.6928,
+      "grad_norm": 45.69491195678711,
+      "learning_rate": 3.4133333333333334e-06,
+      "loss": 1.5774,
+      "step": 52900
+    },
+    {
+      "epoch": 1.69312,
+      "grad_norm": 45.09488296508789,
+      "learning_rate": 3.409777777777778e-06,
+      "loss": 1.5902,
+      "step": 52910
+    },
+    {
+      "epoch": 1.6934399999999998,
+      "grad_norm": 47.69397735595703,
+      "learning_rate": 3.4062222222222225e-06,
+      "loss": 1.5888,
+      "step": 52920
+    },
+    {
+      "epoch": 1.6937600000000002,
+      "grad_norm": 46.744075775146484,
+      "learning_rate": 3.402666666666667e-06,
+      "loss": 1.577,
+      "step": 52930
+    },
+    {
+      "epoch": 1.69408,
+      "grad_norm": 45.290775299072266,
+      "learning_rate": 3.399111111111111e-06,
+      "loss": 1.5799,
+      "step": 52940
+    },
+    {
+      "epoch": 1.6944,
+      "grad_norm": 47.012481689453125,
+      "learning_rate": 3.3955555555555558e-06,
+      "loss": 1.6094,
+      "step": 52950
+    },
+    {
+      "epoch": 1.69472,
+      "grad_norm": 47.2081298828125,
+      "learning_rate": 3.3920000000000003e-06,
+      "loss": 1.572,
+      "step": 52960
+    },
+    {
+      "epoch": 1.69504,
+      "grad_norm": 47.48590850830078,
+      "learning_rate": 3.388444444444445e-06,
+      "loss": 1.6143,
+      "step": 52970
+    },
+    {
+      "epoch": 1.69536,
+      "grad_norm": 47.304962158203125,
+      "learning_rate": 3.3848888888888894e-06,
+      "loss": 1.5967,
+      "step": 52980
+    },
+    {
+      "epoch": 1.6956799999999999,
+      "grad_norm": 46.138160705566406,
+      "learning_rate": 3.3813333333333335e-06,
+      "loss": 1.623,
+      "step": 52990
+    },
+    {
+      "epoch": 1.696,
+      "grad_norm": 45.87535095214844,
+      "learning_rate": 3.377777777777778e-06,
+      "loss": 1.5781,
+      "step": 53000
+    },
+    {
+      "epoch": 1.69632,
+      "grad_norm": 45.48112106323242,
+      "learning_rate": 3.3742222222222226e-06,
+      "loss": 1.575,
+      "step": 53010
+    },
+    {
+      "epoch": 1.69664,
+      "grad_norm": 45.52760314941406,
+      "learning_rate": 3.370666666666667e-06,
+      "loss": 1.5674,
+      "step": 53020
+    },
+    {
+      "epoch": 1.69696,
+      "grad_norm": 45.82063674926758,
+      "learning_rate": 3.3671111111111118e-06,
+      "loss": 1.6147,
+      "step": 53030
+    },
+    {
+      "epoch": 1.6972800000000001,
+      "grad_norm": 46.35969924926758,
+      "learning_rate": 3.363555555555556e-06,
+      "loss": 1.5475,
+      "step": 53040
+    },
+    {
+      "epoch": 1.6976,
+      "grad_norm": 45.41106033325195,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.6013,
+      "step": 53050
+    },
+    {
+      "epoch": 1.6979199999999999,
+      "grad_norm": 44.992618560791016,
+      "learning_rate": 3.356444444444445e-06,
+      "loss": 1.5914,
+      "step": 53060
+    },
+    {
+      "epoch": 1.69824,
+      "grad_norm": 48.927650451660156,
+      "learning_rate": 3.3528888888888895e-06,
+      "loss": 1.5979,
+      "step": 53070
+    },
+    {
+      "epoch": 1.69856,
+      "grad_norm": 45.46303176879883,
+      "learning_rate": 3.3493333333333337e-06,
+      "loss": 1.5867,
+      "step": 53080
+    },
+    {
+      "epoch": 1.69888,
+      "grad_norm": 46.1902961730957,
+      "learning_rate": 3.345777777777778e-06,
+      "loss": 1.6074,
+      "step": 53090
+    },
+    {
+      "epoch": 1.6992,
+      "grad_norm": 44.09128952026367,
+      "learning_rate": 3.3422222222222224e-06,
+      "loss": 1.6105,
+      "step": 53100
+    },
+    {
+      "epoch": 1.6995200000000001,
+      "grad_norm": 44.388973236083984,
+      "learning_rate": 3.338666666666667e-06,
+      "loss": 1.5736,
+      "step": 53110
+    },
+    {
+      "epoch": 1.69984,
+      "grad_norm": 44.886451721191406,
+      "learning_rate": 3.335111111111111e-06,
+      "loss": 1.6158,
+      "step": 53120
+    },
+    {
+      "epoch": 1.70016,
+      "grad_norm": 45.723670959472656,
+      "learning_rate": 3.3315555555555556e-06,
+      "loss": 1.599,
+      "step": 53130
+    },
+    {
+      "epoch": 1.70048,
+      "grad_norm": 47.3483772277832,
+      "learning_rate": 3.328e-06,
+      "loss": 1.613,
+      "step": 53140
+    },
+    {
+      "epoch": 1.7008,
+      "grad_norm": 45.45103073120117,
+      "learning_rate": 3.3244444444444447e-06,
+      "loss": 1.5638,
+      "step": 53150
+    },
+    {
+      "epoch": 1.70112,
+      "grad_norm": 45.69742202758789,
+      "learning_rate": 3.3208888888888893e-06,
+      "loss": 1.5924,
+      "step": 53160
+    },
+    {
+      "epoch": 1.7014399999999998,
+      "grad_norm": 45.6076545715332,
+      "learning_rate": 3.3173333333333334e-06,
+      "loss": 1.601,
+      "step": 53170
+    },
+    {
+      "epoch": 1.7017600000000002,
+      "grad_norm": 44.57272720336914,
+      "learning_rate": 3.313777777777778e-06,
+      "loss": 1.6027,
+      "step": 53180
+    },
+    {
+      "epoch": 1.70208,
+      "grad_norm": 46.239906311035156,
+      "learning_rate": 3.3102222222222225e-06,
+      "loss": 1.5891,
+      "step": 53190
+    },
+    {
+      "epoch": 1.7024,
+      "grad_norm": 45.24344253540039,
+      "learning_rate": 3.306666666666667e-06,
+      "loss": 1.569,
+      "step": 53200
+    },
+    {
+      "epoch": 1.70272,
+      "grad_norm": 44.35649871826172,
+      "learning_rate": 3.303111111111111e-06,
+      "loss": 1.5861,
+      "step": 53210
+    },
+    {
+      "epoch": 1.70304,
+      "grad_norm": 46.33483123779297,
+      "learning_rate": 3.2995555555555557e-06,
+      "loss": 1.5841,
+      "step": 53220
+    },
+    {
+      "epoch": 1.70336,
+      "grad_norm": 45.243778228759766,
+      "learning_rate": 3.2960000000000003e-06,
+      "loss": 1.5776,
+      "step": 53230
+    },
+    {
+      "epoch": 1.7036799999999999,
+      "grad_norm": 48.42351531982422,
+      "learning_rate": 3.292444444444445e-06,
+      "loss": 1.5709,
+      "step": 53240
+    },
+    {
+      "epoch": 1.704,
+      "grad_norm": 46.71955871582031,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 1.6112,
+      "step": 53250
+    },
+    {
+      "epoch": 1.70432,
+      "grad_norm": 43.32274627685547,
+      "learning_rate": 3.2853333333333335e-06,
+      "loss": 1.553,
+      "step": 53260
+    },
+    {
+      "epoch": 1.70464,
+      "grad_norm": 47.185638427734375,
+      "learning_rate": 3.281777777777778e-06,
+      "loss": 1.6212,
+      "step": 53270
+    },
+    {
+      "epoch": 1.70496,
+      "grad_norm": 45.04874038696289,
+      "learning_rate": 3.2782222222222226e-06,
+      "loss": 1.5856,
+      "step": 53280
+    },
+    {
+      "epoch": 1.7052800000000001,
+      "grad_norm": 48.05552291870117,
+      "learning_rate": 3.274666666666667e-06,
+      "loss": 1.6068,
+      "step": 53290
+    },
+    {
+      "epoch": 1.7056,
+      "grad_norm": 45.691795349121094,
+      "learning_rate": 3.2711111111111117e-06,
+      "loss": 1.5867,
+      "step": 53300
+    },
+    {
+      "epoch": 1.7059199999999999,
+      "grad_norm": 46.98811721801758,
+      "learning_rate": 3.267555555555556e-06,
+      "loss": 1.6024,
+      "step": 53310
+    },
+    {
+      "epoch": 1.70624,
+      "grad_norm": 46.40603256225586,
+      "learning_rate": 3.2640000000000004e-06,
+      "loss": 1.5938,
+      "step": 53320
+    },
+    {
+      "epoch": 1.70656,
+      "grad_norm": 46.9987678527832,
+      "learning_rate": 3.260444444444445e-06,
+      "loss": 1.5696,
+      "step": 53330
+    },
+    {
+      "epoch": 1.70688,
+      "grad_norm": 45.82811737060547,
+      "learning_rate": 3.2568888888888895e-06,
+      "loss": 1.5863,
+      "step": 53340
+    },
+    {
+      "epoch": 1.7072,
+      "grad_norm": 44.975677490234375,
+      "learning_rate": 3.2533333333333332e-06,
+      "loss": 1.5977,
+      "step": 53350
+    },
+    {
+      "epoch": 1.7075200000000001,
+      "grad_norm": 45.661842346191406,
+      "learning_rate": 3.249777777777778e-06,
+      "loss": 1.5828,
+      "step": 53360
+    },
+    {
+      "epoch": 1.70784,
+      "grad_norm": 47.03694534301758,
+      "learning_rate": 3.2462222222222224e-06,
+      "loss": 1.6006,
+      "step": 53370
+    },
+    {
+      "epoch": 1.70816,
+      "grad_norm": 48.213199615478516,
+      "learning_rate": 3.242666666666667e-06,
+      "loss": 1.6157,
+      "step": 53380
+    },
+    {
+      "epoch": 1.70848,
+      "grad_norm": 48.341346740722656,
+      "learning_rate": 3.239111111111111e-06,
+      "loss": 1.6048,
+      "step": 53390
+    },
+    {
+      "epoch": 1.7088,
+      "grad_norm": 44.660884857177734,
+      "learning_rate": 3.2355555555555556e-06,
+      "loss": 1.5845,
+      "step": 53400
+    },
+    {
+      "epoch": 1.70912,
+      "grad_norm": 47.39234924316406,
+      "learning_rate": 3.232e-06,
+      "loss": 1.5914,
+      "step": 53410
+    },
+    {
+      "epoch": 1.7094399999999998,
+      "grad_norm": 45.54050064086914,
+      "learning_rate": 3.2284444444444447e-06,
+      "loss": 1.5943,
+      "step": 53420
+    },
+    {
+      "epoch": 1.70976,
+      "grad_norm": 46.718017578125,
+      "learning_rate": 3.2248888888888892e-06,
+      "loss": 1.5672,
+      "step": 53430
+    },
+    {
+      "epoch": 1.71008,
+      "grad_norm": 45.91179656982422,
+      "learning_rate": 3.2213333333333334e-06,
+      "loss": 1.5901,
+      "step": 53440
+    },
+    {
+      "epoch": 1.7104,
+      "grad_norm": 46.24087905883789,
+      "learning_rate": 3.217777777777778e-06,
+      "loss": 1.6141,
+      "step": 53450
+    },
+    {
+      "epoch": 1.71072,
+      "grad_norm": 43.00766372680664,
+      "learning_rate": 3.2142222222222225e-06,
+      "loss": 1.6029,
+      "step": 53460
+    },
+    {
+      "epoch": 1.7110400000000001,
+      "grad_norm": 44.39076232910156,
+      "learning_rate": 3.210666666666667e-06,
+      "loss": 1.546,
+      "step": 53470
+    },
+    {
+      "epoch": 1.71136,
+      "grad_norm": 46.70647048950195,
+      "learning_rate": 3.2071111111111116e-06,
+      "loss": 1.5961,
+      "step": 53480
+    },
+    {
+      "epoch": 1.7116799999999999,
+      "grad_norm": 48.059104919433594,
+      "learning_rate": 3.2035555555555557e-06,
+      "loss": 1.6026,
+      "step": 53490
+    },
+    {
+      "epoch": 1.712,
+      "grad_norm": 44.19499206542969,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.5858,
+      "step": 53500
+    },
+    {
+      "epoch": 1.71232,
+      "grad_norm": 46.01230239868164,
+      "learning_rate": 3.196444444444445e-06,
+      "loss": 1.5919,
+      "step": 53510
+    },
+    {
+      "epoch": 1.71264,
+      "grad_norm": 45.61040496826172,
+      "learning_rate": 3.1928888888888894e-06,
+      "loss": 1.5565,
+      "step": 53520
+    },
+    {
+      "epoch": 1.71296,
+      "grad_norm": 46.07856369018555,
+      "learning_rate": 3.1893333333333335e-06,
+      "loss": 1.5847,
+      "step": 53530
+    },
+    {
+      "epoch": 1.7132800000000001,
+      "grad_norm": 43.263511657714844,
+      "learning_rate": 3.185777777777778e-06,
+      "loss": 1.5748,
+      "step": 53540
+    },
+    {
+      "epoch": 1.7136,
+      "grad_norm": 47.8712272644043,
+      "learning_rate": 3.1822222222222226e-06,
+      "loss": 1.6012,
+      "step": 53550
+    },
+    {
+      "epoch": 1.7139199999999999,
+      "grad_norm": 46.34992218017578,
+      "learning_rate": 3.178666666666667e-06,
+      "loss": 1.6002,
+      "step": 53560
+    },
+    {
+      "epoch": 1.71424,
+      "grad_norm": 43.75440979003906,
+      "learning_rate": 3.1751111111111117e-06,
+      "loss": 1.5925,
+      "step": 53570
+    },
+    {
+      "epoch": 1.71456,
+      "grad_norm": 44.676578521728516,
+      "learning_rate": 3.171555555555556e-06,
+      "loss": 1.5746,
+      "step": 53580
+    },
+    {
+      "epoch": 1.71488,
+      "grad_norm": 45.60486602783203,
+      "learning_rate": 3.1680000000000004e-06,
+      "loss": 1.5715,
+      "step": 53590
+    },
+    {
+      "epoch": 1.7151999999999998,
+      "grad_norm": 44.54808044433594,
+      "learning_rate": 3.1644444444444445e-06,
+      "loss": 1.5822,
+      "step": 53600
+    },
+    {
+      "epoch": 1.7155200000000002,
+      "grad_norm": 46.047264099121094,
+      "learning_rate": 3.160888888888889e-06,
+      "loss": 1.6015,
+      "step": 53610
+    },
+    {
+      "epoch": 1.71584,
+      "grad_norm": 45.982940673828125,
+      "learning_rate": 3.1573333333333332e-06,
+      "loss": 1.5928,
+      "step": 53620
+    },
+    {
+      "epoch": 1.71616,
+      "grad_norm": 44.72659683227539,
+      "learning_rate": 3.1537777777777778e-06,
+      "loss": 1.5941,
+      "step": 53630
+    },
+    {
+      "epoch": 1.71648,
+      "grad_norm": 46.68376541137695,
+      "learning_rate": 3.1502222222222223e-06,
+      "loss": 1.585,
+      "step": 53640
+    },
+    {
+      "epoch": 1.7168,
+      "grad_norm": 46.007652282714844,
+      "learning_rate": 3.146666666666667e-06,
+      "loss": 1.6095,
+      "step": 53650
+    },
+    {
+      "epoch": 1.71712,
+      "grad_norm": 48.402584075927734,
+      "learning_rate": 3.143111111111111e-06,
+      "loss": 1.5706,
+      "step": 53660
+    },
+    {
+      "epoch": 1.7174399999999999,
+      "grad_norm": 46.359947204589844,
+      "learning_rate": 3.1395555555555556e-06,
+      "loss": 1.5995,
+      "step": 53670
+    },
+    {
+      "epoch": 1.71776,
+      "grad_norm": 45.970184326171875,
+      "learning_rate": 3.136e-06,
+      "loss": 1.5891,
+      "step": 53680
+    },
+    {
+      "epoch": 1.71808,
+      "grad_norm": 46.32889175415039,
+      "learning_rate": 3.1324444444444447e-06,
+      "loss": 1.593,
+      "step": 53690
+    },
+    {
+      "epoch": 1.7184,
+      "grad_norm": 44.1278190612793,
+      "learning_rate": 3.1288888888888892e-06,
+      "loss": 1.6027,
+      "step": 53700
+    },
+    {
+      "epoch": 1.71872,
+      "grad_norm": 46.170387268066406,
+      "learning_rate": 3.1253333333333334e-06,
+      "loss": 1.6016,
+      "step": 53710
+    },
+    {
+      "epoch": 1.7190400000000001,
+      "grad_norm": 48.95341491699219,
+      "learning_rate": 3.121777777777778e-06,
+      "loss": 1.5959,
+      "step": 53720
+    },
+    {
+      "epoch": 1.71936,
+      "grad_norm": 46.69742965698242,
+      "learning_rate": 3.1182222222222225e-06,
+      "loss": 1.58,
+      "step": 53730
+    },
+    {
+      "epoch": 1.7196799999999999,
+      "grad_norm": 44.949180603027344,
+      "learning_rate": 3.114666666666667e-06,
+      "loss": 1.5845,
+      "step": 53740
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 47.0362434387207,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 1.6146,
+      "step": 53750
+    },
+    {
+      "epoch": 1.72032,
+      "grad_norm": 47.64339828491211,
+      "learning_rate": 3.1075555555555557e-06,
+      "loss": 1.6238,
+      "step": 53760
+    },
+    {
+      "epoch": 1.72064,
+      "grad_norm": 47.11689758300781,
+      "learning_rate": 3.1040000000000003e-06,
+      "loss": 1.5878,
+      "step": 53770
+    },
+    {
+      "epoch": 1.72096,
+      "grad_norm": 46.65193557739258,
+      "learning_rate": 3.100444444444445e-06,
+      "loss": 1.5894,
+      "step": 53780
+    },
+    {
+      "epoch": 1.7212800000000001,
+      "grad_norm": 47.493778228759766,
+      "learning_rate": 3.0968888888888894e-06,
+      "loss": 1.5807,
+      "step": 53790
+    },
+    {
+      "epoch": 1.7216,
+      "grad_norm": 45.632720947265625,
+      "learning_rate": 3.093333333333334e-06,
+      "loss": 1.6096,
+      "step": 53800
+    },
+    {
+      "epoch": 1.72192,
+      "grad_norm": 46.928646087646484,
+      "learning_rate": 3.089777777777778e-06,
+      "loss": 1.5793,
+      "step": 53810
+    },
+    {
+      "epoch": 1.72224,
+      "grad_norm": 48.068153381347656,
+      "learning_rate": 3.0862222222222226e-06,
+      "loss": 1.6036,
+      "step": 53820
+    },
+    {
+      "epoch": 1.72256,
+      "grad_norm": 45.10832977294922,
+      "learning_rate": 3.082666666666667e-06,
+      "loss": 1.5993,
+      "step": 53830
+    },
+    {
+      "epoch": 1.72288,
+      "grad_norm": 47.08902359008789,
+      "learning_rate": 3.0791111111111117e-06,
+      "loss": 1.5951,
+      "step": 53840
+    },
+    {
+      "epoch": 1.7231999999999998,
+      "grad_norm": 46.66806411743164,
+      "learning_rate": 3.0755555555555563e-06,
+      "loss": 1.5956,
+      "step": 53850
+    },
+    {
+      "epoch": 1.7235200000000002,
+      "grad_norm": 44.86949157714844,
+      "learning_rate": 3.072e-06,
+      "loss": 1.5838,
+      "step": 53860
+    },
+    {
+      "epoch": 1.72384,
+      "grad_norm": 44.62445831298828,
+      "learning_rate": 3.0684444444444445e-06,
+      "loss": 1.6053,
+      "step": 53870
+    },
+    {
+      "epoch": 1.72416,
+      "grad_norm": 58.58441162109375,
+      "learning_rate": 3.064888888888889e-06,
+      "loss": 1.6101,
+      "step": 53880
+    },
+    {
+      "epoch": 1.72448,
+      "grad_norm": 47.462833404541016,
+      "learning_rate": 3.0613333333333332e-06,
+      "loss": 1.5816,
+      "step": 53890
+    },
+    {
+      "epoch": 1.7248,
+      "grad_norm": 44.10990905761719,
+      "learning_rate": 3.0577777777777778e-06,
+      "loss": 1.5876,
+      "step": 53900
+    },
+    {
+      "epoch": 1.72512,
+      "grad_norm": 46.68988037109375,
+      "learning_rate": 3.0542222222222223e-06,
+      "loss": 1.5852,
+      "step": 53910
+    },
+    {
+      "epoch": 1.7254399999999999,
+      "grad_norm": 45.82401657104492,
+      "learning_rate": 3.050666666666667e-06,
+      "loss": 1.5877,
+      "step": 53920
+    },
+    {
+      "epoch": 1.72576,
+      "grad_norm": 45.19087219238281,
+      "learning_rate": 3.0471111111111114e-06,
+      "loss": 1.5625,
+      "step": 53930
+    },
+    {
+      "epoch": 1.72608,
+      "grad_norm": 47.10851287841797,
+      "learning_rate": 3.0435555555555556e-06,
+      "loss": 1.5695,
+      "step": 53940
+    },
+    {
+      "epoch": 1.7264,
+      "grad_norm": 46.2303581237793,
+      "learning_rate": 3.04e-06,
+      "loss": 1.5625,
+      "step": 53950
+    },
+    {
+      "epoch": 1.72672,
+      "grad_norm": 45.25043487548828,
+      "learning_rate": 3.0364444444444447e-06,
+      "loss": 1.5951,
+      "step": 53960
+    },
+    {
+      "epoch": 1.7270400000000001,
+      "grad_norm": 46.983970642089844,
+      "learning_rate": 3.0328888888888892e-06,
+      "loss": 1.5889,
+      "step": 53970
+    },
+    {
+      "epoch": 1.72736,
+      "grad_norm": 45.9723014831543,
+      "learning_rate": 3.0293333333333338e-06,
+      "loss": 1.5972,
+      "step": 53980
+    },
+    {
+      "epoch": 1.7276799999999999,
+      "grad_norm": 46.250850677490234,
+      "learning_rate": 3.025777777777778e-06,
+      "loss": 1.5962,
+      "step": 53990
+    },
+    {
+      "epoch": 1.728,
+      "grad_norm": 45.149532318115234,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 1.5752,
+      "step": 54000
+    },
+    {
+      "epoch": 1.72832,
+      "grad_norm": 44.299720764160156,
+      "learning_rate": 3.018666666666667e-06,
+      "loss": 1.5883,
+      "step": 54010
+    },
+    {
+      "epoch": 1.72864,
+      "grad_norm": 45.48408508300781,
+      "learning_rate": 3.0151111111111116e-06,
+      "loss": 1.5829,
+      "step": 54020
+    },
+    {
+      "epoch": 1.72896,
+      "grad_norm": 45.040367126464844,
+      "learning_rate": 3.0115555555555557e-06,
+      "loss": 1.5962,
+      "step": 54030
+    },
+    {
+      "epoch": 1.7292800000000002,
+      "grad_norm": 46.36260986328125,
+      "learning_rate": 3.0080000000000003e-06,
+      "loss": 1.5923,
+      "step": 54040
+    },
+    {
+      "epoch": 1.7296,
+      "grad_norm": 44.90981674194336,
+      "learning_rate": 3.004444444444445e-06,
+      "loss": 1.6213,
+      "step": 54050
+    },
+    {
+      "epoch": 1.72992,
+      "grad_norm": 47.245059967041016,
+      "learning_rate": 3.0008888888888894e-06,
+      "loss": 1.578,
+      "step": 54060
+    },
+    {
+      "epoch": 1.73024,
+      "grad_norm": 44.638729095458984,
+      "learning_rate": 2.997333333333334e-06,
+      "loss": 1.5771,
+      "step": 54070
+    },
+    {
+      "epoch": 1.73056,
+      "grad_norm": 45.962371826171875,
+      "learning_rate": 2.993777777777778e-06,
+      "loss": 1.5925,
+      "step": 54080
+    },
+    {
+      "epoch": 1.73088,
+      "grad_norm": 47.59477996826172,
+      "learning_rate": 2.9902222222222226e-06,
+      "loss": 1.6139,
+      "step": 54090
+    },
+    {
+      "epoch": 1.7311999999999999,
+      "grad_norm": 44.2559928894043,
+      "learning_rate": 2.986666666666667e-06,
+      "loss": 1.5937,
+      "step": 54100
+    },
+    {
+      "epoch": 1.73152,
+      "grad_norm": 45.09414291381836,
+      "learning_rate": 2.9831111111111117e-06,
+      "loss": 1.5887,
+      "step": 54110
+    },
+    {
+      "epoch": 1.73184,
+      "grad_norm": 45.167659759521484,
+      "learning_rate": 2.9795555555555554e-06,
+      "loss": 1.5933,
+      "step": 54120
+    },
+    {
+      "epoch": 1.73216,
+      "grad_norm": 46.760459899902344,
+      "learning_rate": 2.976e-06,
+      "loss": 1.5563,
+      "step": 54130
+    },
+    {
+      "epoch": 1.73248,
+      "grad_norm": 46.089717864990234,
+      "learning_rate": 2.9724444444444445e-06,
+      "loss": 1.6039,
+      "step": 54140
+    },
+    {
+      "epoch": 1.7328000000000001,
+      "grad_norm": 44.89067077636719,
+      "learning_rate": 2.968888888888889e-06,
+      "loss": 1.598,
+      "step": 54150
+    },
+    {
+      "epoch": 1.73312,
+      "grad_norm": 46.289737701416016,
+      "learning_rate": 2.965333333333333e-06,
+      "loss": 1.5787,
+      "step": 54160
+    },
+    {
+      "epoch": 1.7334399999999999,
+      "grad_norm": 54.458438873291016,
+      "learning_rate": 2.9617777777777778e-06,
+      "loss": 1.6064,
+      "step": 54170
+    },
+    {
+      "epoch": 1.73376,
+      "grad_norm": 46.293182373046875,
+      "learning_rate": 2.9582222222222223e-06,
+      "loss": 1.5778,
+      "step": 54180
+    },
+    {
+      "epoch": 1.73408,
+      "grad_norm": 44.743141174316406,
+      "learning_rate": 2.954666666666667e-06,
+      "loss": 1.5935,
+      "step": 54190
+    },
+    {
+      "epoch": 1.7344,
+      "grad_norm": 45.82085418701172,
+      "learning_rate": 2.9511111111111114e-06,
+      "loss": 1.565,
+      "step": 54200
+    },
+    {
+      "epoch": 1.73472,
+      "grad_norm": 45.540897369384766,
+      "learning_rate": 2.9475555555555556e-06,
+      "loss": 1.575,
+      "step": 54210
+    },
+    {
+      "epoch": 1.7350400000000001,
+      "grad_norm": 43.974327087402344,
+      "learning_rate": 2.944e-06,
+      "loss": 1.5982,
+      "step": 54220
+    },
+    {
+      "epoch": 1.73536,
+      "grad_norm": 45.236446380615234,
+      "learning_rate": 2.9404444444444447e-06,
+      "loss": 1.5448,
+      "step": 54230
+    },
+    {
+      "epoch": 1.73568,
+      "grad_norm": 45.35393142700195,
+      "learning_rate": 2.936888888888889e-06,
+      "loss": 1.5719,
+      "step": 54240
+    },
+    {
+      "epoch": 1.736,
+      "grad_norm": 45.527549743652344,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 1.6116,
+      "step": 54250
+    },
+    {
+      "epoch": 1.73632,
+      "grad_norm": 46.48612594604492,
+      "learning_rate": 2.929777777777778e-06,
+      "loss": 1.5727,
+      "step": 54260
+    },
+    {
+      "epoch": 1.73664,
+      "grad_norm": 46.852294921875,
+      "learning_rate": 2.9262222222222224e-06,
+      "loss": 1.6184,
+      "step": 54270
+    },
+    {
+      "epoch": 1.7369599999999998,
+      "grad_norm": 45.39048767089844,
+      "learning_rate": 2.922666666666667e-06,
+      "loss": 1.5797,
+      "step": 54280
+    },
+    {
+      "epoch": 1.7372800000000002,
+      "grad_norm": 46.60129928588867,
+      "learning_rate": 2.9191111111111116e-06,
+      "loss": 1.615,
+      "step": 54290
+    },
+    {
+      "epoch": 1.7376,
+      "grad_norm": 45.626441955566406,
+      "learning_rate": 2.915555555555556e-06,
+      "loss": 1.5825,
+      "step": 54300
+    },
+    {
+      "epoch": 1.73792,
+      "grad_norm": 48.38727569580078,
+      "learning_rate": 2.9120000000000002e-06,
+      "loss": 1.6134,
+      "step": 54310
+    },
+    {
+      "epoch": 1.73824,
+      "grad_norm": 45.35850143432617,
+      "learning_rate": 2.908444444444445e-06,
+      "loss": 1.5666,
+      "step": 54320
+    },
+    {
+      "epoch": 1.73856,
+      "grad_norm": 45.78057861328125,
+      "learning_rate": 2.9048888888888893e-06,
+      "loss": 1.5809,
+      "step": 54330
+    },
+    {
+      "epoch": 1.73888,
+      "grad_norm": 45.44707489013672,
+      "learning_rate": 2.901333333333334e-06,
+      "loss": 1.5908,
+      "step": 54340
+    },
+    {
+      "epoch": 1.7391999999999999,
+      "grad_norm": 44.71982955932617,
+      "learning_rate": 2.8977777777777785e-06,
+      "loss": 1.5843,
+      "step": 54350
+    },
+    {
+      "epoch": 1.73952,
+      "grad_norm": 45.987815856933594,
+      "learning_rate": 2.8942222222222226e-06,
+      "loss": 1.585,
+      "step": 54360
+    },
+    {
+      "epoch": 1.73984,
+      "grad_norm": 47.58973693847656,
+      "learning_rate": 2.890666666666667e-06,
+      "loss": 1.5962,
+      "step": 54370
+    },
+    {
+      "epoch": 1.74016,
+      "grad_norm": 47.01103973388672,
+      "learning_rate": 2.8871111111111113e-06,
+      "loss": 1.5784,
+      "step": 54380
+    },
+    {
+      "epoch": 1.74048,
+      "grad_norm": 45.42546463012695,
+      "learning_rate": 2.8835555555555554e-06,
+      "loss": 1.6064,
+      "step": 54390
+    },
+    {
+      "epoch": 1.7408000000000001,
+      "grad_norm": 45.88031768798828,
+      "learning_rate": 2.88e-06,
+      "loss": 1.5828,
+      "step": 54400
+    },
+    {
+      "epoch": 1.74112,
+      "grad_norm": 43.90048599243164,
+      "learning_rate": 2.8764444444444445e-06,
+      "loss": 1.5748,
+      "step": 54410
+    },
+    {
+      "epoch": 1.7414399999999999,
+      "grad_norm": 45.916751861572266,
+      "learning_rate": 2.872888888888889e-06,
+      "loss": 1.6149,
+      "step": 54420
+    },
+    {
+      "epoch": 1.74176,
+      "grad_norm": 44.58837890625,
+      "learning_rate": 2.8693333333333336e-06,
+      "loss": 1.6083,
+      "step": 54430
+    },
+    {
+      "epoch": 1.74208,
+      "grad_norm": 45.08499526977539,
+      "learning_rate": 2.8657777777777777e-06,
+      "loss": 1.5823,
+      "step": 54440
+    },
+    {
+      "epoch": 1.7424,
+      "grad_norm": 45.388545989990234,
+      "learning_rate": 2.8622222222222223e-06,
+      "loss": 1.5837,
+      "step": 54450
+    },
+    {
+      "epoch": 1.74272,
+      "grad_norm": 45.021602630615234,
+      "learning_rate": 2.858666666666667e-06,
+      "loss": 1.5933,
+      "step": 54460
+    },
+    {
+      "epoch": 1.7430400000000001,
+      "grad_norm": 45.340152740478516,
+      "learning_rate": 2.8551111111111114e-06,
+      "loss": 1.6089,
+      "step": 54470
+    },
+    {
+      "epoch": 1.74336,
+      "grad_norm": 48.71046829223633,
+      "learning_rate": 2.8515555555555555e-06,
+      "loss": 1.5735,
+      "step": 54480
+    },
+    {
+      "epoch": 1.74368,
+      "grad_norm": 46.213844299316406,
+      "learning_rate": 2.848e-06,
+      "loss": 1.5872,
+      "step": 54490
+    },
+    {
+      "epoch": 1.744,
+      "grad_norm": 44.88610076904297,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 1.5619,
+      "step": 54500
+    },
+    {
+      "epoch": 1.74432,
+      "grad_norm": 48.63814163208008,
+      "learning_rate": 2.840888888888889e-06,
+      "loss": 1.5965,
+      "step": 54510
+    },
+    {
+      "epoch": 1.74464,
+      "grad_norm": 45.273155212402344,
+      "learning_rate": 2.8373333333333338e-06,
+      "loss": 1.587,
+      "step": 54520
+    },
+    {
+      "epoch": 1.7449599999999998,
+      "grad_norm": 47.4951057434082,
+      "learning_rate": 2.833777777777778e-06,
+      "loss": 1.5887,
+      "step": 54530
+    },
+    {
+      "epoch": 1.7452800000000002,
+      "grad_norm": 46.31680679321289,
+      "learning_rate": 2.8302222222222224e-06,
+      "loss": 1.6039,
+      "step": 54540
+    },
+    {
+      "epoch": 1.7456,
+      "grad_norm": 45.749420166015625,
+      "learning_rate": 2.826666666666667e-06,
+      "loss": 1.5674,
+      "step": 54550
+    },
+    {
+      "epoch": 1.74592,
+      "grad_norm": 47.789306640625,
+      "learning_rate": 2.8231111111111115e-06,
+      "loss": 1.5951,
+      "step": 54560
+    },
+    {
+      "epoch": 1.74624,
+      "grad_norm": 48.388572692871094,
+      "learning_rate": 2.819555555555556e-06,
+      "loss": 1.605,
+      "step": 54570
+    },
+    {
+      "epoch": 1.7465600000000001,
+      "grad_norm": 46.346153259277344,
+      "learning_rate": 2.8160000000000002e-06,
+      "loss": 1.5703,
+      "step": 54580
+    },
+    {
+      "epoch": 1.74688,
+      "grad_norm": 44.61640548706055,
+      "learning_rate": 2.8124444444444448e-06,
+      "loss": 1.6008,
+      "step": 54590
+    },
+    {
+      "epoch": 1.7471999999999999,
+      "grad_norm": 45.16740417480469,
+      "learning_rate": 2.8088888888888893e-06,
+      "loss": 1.5427,
+      "step": 54600
+    },
+    {
+      "epoch": 1.74752,
+      "grad_norm": 47.40975570678711,
+      "learning_rate": 2.805333333333334e-06,
+      "loss": 1.5796,
+      "step": 54610
+    },
+    {
+      "epoch": 1.74784,
+      "grad_norm": 44.373897552490234,
+      "learning_rate": 2.8017777777777784e-06,
+      "loss": 1.5912,
+      "step": 54620
+    },
+    {
+      "epoch": 1.74816,
+      "grad_norm": 46.67420196533203,
+      "learning_rate": 2.7982222222222226e-06,
+      "loss": 1.5763,
+      "step": 54630
+    },
+    {
+      "epoch": 1.74848,
+      "grad_norm": 44.51902770996094,
+      "learning_rate": 2.7946666666666667e-06,
+      "loss": 1.5819,
+      "step": 54640
+    },
+    {
+      "epoch": 1.7488000000000001,
+      "grad_norm": 48.27885437011719,
+      "learning_rate": 2.7911111111111113e-06,
+      "loss": 1.5972,
+      "step": 54650
+    },
+    {
+      "epoch": 1.74912,
+      "grad_norm": 45.42656707763672,
+      "learning_rate": 2.7875555555555554e-06,
+      "loss": 1.5925,
+      "step": 54660
+    },
+    {
+      "epoch": 1.7494399999999999,
+      "grad_norm": 44.766998291015625,
+      "learning_rate": 2.784e-06,
+      "loss": 1.6199,
+      "step": 54670
+    },
+    {
+      "epoch": 1.74976,
+      "grad_norm": 44.84164047241211,
+      "learning_rate": 2.7804444444444445e-06,
+      "loss": 1.5559,
+      "step": 54680
+    },
+    {
+      "epoch": 1.75008,
+      "grad_norm": 45.92261505126953,
+      "learning_rate": 2.776888888888889e-06,
+      "loss": 1.5874,
+      "step": 54690
+    },
+    {
+      "epoch": 1.7504,
+      "grad_norm": 45.73942565917969,
+      "learning_rate": 2.7733333333333336e-06,
+      "loss": 1.5739,
+      "step": 54700
+    },
+    {
+      "epoch": 1.75072,
+      "grad_norm": 47.31524658203125,
+      "learning_rate": 2.7697777777777777e-06,
+      "loss": 1.5882,
+      "step": 54710
+    },
+    {
+      "epoch": 1.7510400000000002,
+      "grad_norm": 46.772430419921875,
+      "learning_rate": 2.7662222222222223e-06,
+      "loss": 1.587,
+      "step": 54720
+    },
+    {
+      "epoch": 1.75136,
+      "grad_norm": 45.29179763793945,
+      "learning_rate": 2.762666666666667e-06,
+      "loss": 1.5967,
+      "step": 54730
+    },
+    {
+      "epoch": 1.75168,
+      "grad_norm": 46.218624114990234,
+      "learning_rate": 2.7591111111111114e-06,
+      "loss": 1.5756,
+      "step": 54740
+    },
+    {
+      "epoch": 1.752,
+      "grad_norm": 46.63479995727539,
+      "learning_rate": 2.755555555555556e-06,
+      "loss": 1.5881,
+      "step": 54750
+    },
+    {
+      "epoch": 1.75232,
+      "grad_norm": 46.75434112548828,
+      "learning_rate": 2.752e-06,
+      "loss": 1.5659,
+      "step": 54760
+    },
+    {
+      "epoch": 1.75264,
+      "grad_norm": 45.202186584472656,
+      "learning_rate": 2.7484444444444446e-06,
+      "loss": 1.605,
+      "step": 54770
+    },
+    {
+      "epoch": 1.7529599999999999,
+      "grad_norm": 44.8332633972168,
+      "learning_rate": 2.744888888888889e-06,
+      "loss": 1.5793,
+      "step": 54780
+    },
+    {
+      "epoch": 1.75328,
+      "grad_norm": 44.32758331298828,
+      "learning_rate": 2.7413333333333337e-06,
+      "loss": 1.5865,
+      "step": 54790
+    },
+    {
+      "epoch": 1.7536,
+      "grad_norm": 45.434085845947266,
+      "learning_rate": 2.7377777777777783e-06,
+      "loss": 1.5872,
+      "step": 54800
+    },
+    {
+      "epoch": 1.75392,
+      "grad_norm": 45.59672546386719,
+      "learning_rate": 2.7342222222222224e-06,
+      "loss": 1.5765,
+      "step": 54810
+    },
+    {
+      "epoch": 1.75424,
+      "grad_norm": 47.09999084472656,
+      "learning_rate": 2.730666666666667e-06,
+      "loss": 1.5781,
+      "step": 54820
+    },
+    {
+      "epoch": 1.7545600000000001,
+      "grad_norm": 46.603023529052734,
+      "learning_rate": 2.7271111111111115e-06,
+      "loss": 1.5639,
+      "step": 54830
+    },
+    {
+      "epoch": 1.75488,
+      "grad_norm": 43.42913055419922,
+      "learning_rate": 2.723555555555556e-06,
+      "loss": 1.5731,
+      "step": 54840
+    },
+    {
+      "epoch": 1.7551999999999999,
+      "grad_norm": 44.83296585083008,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 1.5562,
+      "step": 54850
+    },
+    {
+      "epoch": 1.75552,
+      "grad_norm": 47.64613723754883,
+      "learning_rate": 2.7164444444444448e-06,
+      "loss": 1.57,
+      "step": 54860
+    },
+    {
+      "epoch": 1.75584,
+      "grad_norm": 47.252655029296875,
+      "learning_rate": 2.7128888888888893e-06,
+      "loss": 1.6299,
+      "step": 54870
+    },
+    {
+      "epoch": 1.75616,
+      "grad_norm": 48.46607208251953,
+      "learning_rate": 2.709333333333334e-06,
+      "loss": 1.5777,
+      "step": 54880
+    },
+    {
+      "epoch": 1.75648,
+      "grad_norm": 47.050479888916016,
+      "learning_rate": 2.7057777777777776e-06,
+      "loss": 1.5911,
+      "step": 54890
+    },
+    {
+      "epoch": 1.7568000000000001,
+      "grad_norm": 45.11897659301758,
+      "learning_rate": 2.702222222222222e-06,
+      "loss": 1.5852,
+      "step": 54900
+    },
+    {
+      "epoch": 1.75712,
+      "grad_norm": 44.798404693603516,
+      "learning_rate": 2.6986666666666667e-06,
+      "loss": 1.5997,
+      "step": 54910
+    },
+    {
+      "epoch": 1.75744,
+      "grad_norm": 45.404212951660156,
+      "learning_rate": 2.6951111111111112e-06,
+      "loss": 1.5861,
+      "step": 54920
+    },
+    {
+      "epoch": 1.75776,
+      "grad_norm": 45.11406326293945,
+      "learning_rate": 2.691555555555556e-06,
+      "loss": 1.5691,
+      "step": 54930
+    },
+    {
+      "epoch": 1.75808,
+      "grad_norm": 47.126953125,
+      "learning_rate": 2.688e-06,
+      "loss": 1.5937,
+      "step": 54940
+    },
+    {
+      "epoch": 1.7584,
+      "grad_norm": 49.89951705932617,
+      "learning_rate": 2.6844444444444445e-06,
+      "loss": 1.5766,
+      "step": 54950
+    },
+    {
+      "epoch": 1.7587199999999998,
+      "grad_norm": 45.07528305053711,
+      "learning_rate": 2.680888888888889e-06,
+      "loss": 1.5754,
+      "step": 54960
+    },
+    {
+      "epoch": 1.7590400000000002,
+      "grad_norm": 44.071266174316406,
+      "learning_rate": 2.6773333333333336e-06,
+      "loss": 1.6159,
+      "step": 54970
+    },
+    {
+      "epoch": 1.75936,
+      "grad_norm": 47.069217681884766,
+      "learning_rate": 2.6737777777777777e-06,
+      "loss": 1.6047,
+      "step": 54980
+    },
+    {
+      "epoch": 1.75968,
+      "grad_norm": 44.97282409667969,
+      "learning_rate": 2.6702222222222223e-06,
+      "loss": 1.5753,
+      "step": 54990
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 46.30501937866211,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.6084,
+      "step": 55000
+    },
+    {
+      "epoch": 1.76032,
+      "grad_norm": 46.91132736206055,
+      "learning_rate": 2.6631111111111114e-06,
+      "loss": 1.601,
+      "step": 55010
+    },
+    {
+      "epoch": 1.76064,
+      "grad_norm": 47.901580810546875,
+      "learning_rate": 2.659555555555556e-06,
+      "loss": 1.5803,
+      "step": 55020
+    },
+    {
+      "epoch": 1.7609599999999999,
+      "grad_norm": 46.67458724975586,
+      "learning_rate": 2.656e-06,
+      "loss": 1.5893,
+      "step": 55030
+    },
+    {
+      "epoch": 1.76128,
+      "grad_norm": 45.30000686645508,
+      "learning_rate": 2.6524444444444446e-06,
+      "loss": 1.5668,
+      "step": 55040
+    },
+    {
+      "epoch": 1.7616,
+      "grad_norm": 47.15391159057617,
+      "learning_rate": 2.648888888888889e-06,
+      "loss": 1.5733,
+      "step": 55050
+    },
+    {
+      "epoch": 1.76192,
+      "grad_norm": 47.432796478271484,
+      "learning_rate": 2.6453333333333337e-06,
+      "loss": 1.6094,
+      "step": 55060
+    },
+    {
+      "epoch": 1.76224,
+      "grad_norm": 45.1340446472168,
+      "learning_rate": 2.6417777777777783e-06,
+      "loss": 1.5999,
+      "step": 55070
+    },
+    {
+      "epoch": 1.7625600000000001,
+      "grad_norm": 45.92671203613281,
+      "learning_rate": 2.6382222222222224e-06,
+      "loss": 1.6099,
+      "step": 55080
+    },
+    {
+      "epoch": 1.76288,
+      "grad_norm": 46.85237503051758,
+      "learning_rate": 2.634666666666667e-06,
+      "loss": 1.6062,
+      "step": 55090
+    },
+    {
+      "epoch": 1.7631999999999999,
+      "grad_norm": 44.259727478027344,
+      "learning_rate": 2.6311111111111115e-06,
+      "loss": 1.5687,
+      "step": 55100
+    },
+    {
+      "epoch": 1.76352,
+      "grad_norm": 47.20001983642578,
+      "learning_rate": 2.627555555555556e-06,
+      "loss": 1.5658,
+      "step": 55110
+    },
+    {
+      "epoch": 1.76384,
+      "grad_norm": 46.446468353271484,
+      "learning_rate": 2.6240000000000006e-06,
+      "loss": 1.6,
+      "step": 55120
+    },
+    {
+      "epoch": 1.76416,
+      "grad_norm": 48.24945831298828,
+      "learning_rate": 2.6204444444444448e-06,
+      "loss": 1.5677,
+      "step": 55130
+    },
+    {
+      "epoch": 1.76448,
+      "grad_norm": 45.008094787597656,
+      "learning_rate": 2.6168888888888893e-06,
+      "loss": 1.5802,
+      "step": 55140
+    },
+    {
+      "epoch": 1.7648000000000001,
+      "grad_norm": 44.15761947631836,
+      "learning_rate": 2.6133333333333334e-06,
+      "loss": 1.5379,
+      "step": 55150
+    },
+    {
+      "epoch": 1.76512,
+      "grad_norm": 45.80890655517578,
+      "learning_rate": 2.6097777777777776e-06,
+      "loss": 1.5746,
+      "step": 55160
+    },
+    {
+      "epoch": 1.76544,
+      "grad_norm": 44.99577713012695,
+      "learning_rate": 2.606222222222222e-06,
+      "loss": 1.5863,
+      "step": 55170
+    },
+    {
+      "epoch": 1.76576,
+      "grad_norm": 45.428462982177734,
+      "learning_rate": 2.6026666666666667e-06,
+      "loss": 1.5704,
+      "step": 55180
+    },
+    {
+      "epoch": 1.76608,
+      "grad_norm": 45.20389938354492,
+      "learning_rate": 2.5991111111111112e-06,
+      "loss": 1.5804,
+      "step": 55190
+    },
+    {
+      "epoch": 1.7664,
+      "grad_norm": 49.05274200439453,
+      "learning_rate": 2.5955555555555558e-06,
+      "loss": 1.5703,
+      "step": 55200
+    },
+    {
+      "epoch": 1.7667199999999998,
+      "grad_norm": 44.522544860839844,
+      "learning_rate": 2.592e-06,
+      "loss": 1.6099,
+      "step": 55210
+    },
+    {
+      "epoch": 1.7670400000000002,
+      "grad_norm": 46.331634521484375,
+      "learning_rate": 2.5884444444444445e-06,
+      "loss": 1.6057,
+      "step": 55220
+    },
+    {
+      "epoch": 1.76736,
+      "grad_norm": 46.33926773071289,
+      "learning_rate": 2.584888888888889e-06,
+      "loss": 1.591,
+      "step": 55230
+    },
+    {
+      "epoch": 1.76768,
+      "grad_norm": 46.363548278808594,
+      "learning_rate": 2.5813333333333336e-06,
+      "loss": 1.5776,
+      "step": 55240
+    },
+    {
+      "epoch": 1.768,
+      "grad_norm": 44.61825942993164,
+      "learning_rate": 2.577777777777778e-06,
+      "loss": 1.5853,
+      "step": 55250
+    },
+    {
+      "epoch": 1.7683200000000001,
+      "grad_norm": 45.92292404174805,
+      "learning_rate": 2.5742222222222223e-06,
+      "loss": 1.6115,
+      "step": 55260
+    },
+    {
+      "epoch": 1.76864,
+      "grad_norm": 46.37104415893555,
+      "learning_rate": 2.570666666666667e-06,
+      "loss": 1.6099,
+      "step": 55270
+    },
+    {
+      "epoch": 1.7689599999999999,
+      "grad_norm": 47.2564697265625,
+      "learning_rate": 2.5671111111111114e-06,
+      "loss": 1.5999,
+      "step": 55280
+    },
+    {
+      "epoch": 1.76928,
+      "grad_norm": 45.823089599609375,
+      "learning_rate": 2.563555555555556e-06,
+      "loss": 1.5867,
+      "step": 55290
+    },
+    {
+      "epoch": 1.7696,
+      "grad_norm": 45.80973815917969,
+      "learning_rate": 2.56e-06,
+      "loss": 1.5808,
+      "step": 55300
+    },
+    {
+      "epoch": 1.76992,
+      "grad_norm": 45.57069778442383,
+      "learning_rate": 2.5564444444444446e-06,
+      "loss": 1.6039,
+      "step": 55310
+    },
+    {
+      "epoch": 1.77024,
+      "grad_norm": 48.02677917480469,
+      "learning_rate": 2.552888888888889e-06,
+      "loss": 1.5714,
+      "step": 55320
+    },
+    {
+      "epoch": 1.7705600000000001,
+      "grad_norm": 46.72894287109375,
+      "learning_rate": 2.5493333333333337e-06,
+      "loss": 1.6051,
+      "step": 55330
+    },
+    {
+      "epoch": 1.77088,
+      "grad_norm": 47.3582763671875,
+      "learning_rate": 2.5457777777777783e-06,
+      "loss": 1.5619,
+      "step": 55340
+    },
+    {
+      "epoch": 1.7711999999999999,
+      "grad_norm": 47.09748077392578,
+      "learning_rate": 2.5422222222222224e-06,
+      "loss": 1.5879,
+      "step": 55350
+    },
+    {
+      "epoch": 1.77152,
+      "grad_norm": 45.226200103759766,
+      "learning_rate": 2.538666666666667e-06,
+      "loss": 1.5996,
+      "step": 55360
+    },
+    {
+      "epoch": 1.77184,
+      "grad_norm": 46.5212287902832,
+      "learning_rate": 2.5351111111111115e-06,
+      "loss": 1.5795,
+      "step": 55370
+    },
+    {
+      "epoch": 1.77216,
+      "grad_norm": 45.95960235595703,
+      "learning_rate": 2.531555555555556e-06,
+      "loss": 1.5892,
+      "step": 55380
+    },
+    {
+      "epoch": 1.77248,
+      "grad_norm": 47.54762649536133,
+      "learning_rate": 2.5280000000000006e-06,
+      "loss": 1.5966,
+      "step": 55390
+    },
+    {
+      "epoch": 1.7728000000000002,
+      "grad_norm": 46.98671340942383,
+      "learning_rate": 2.5244444444444447e-06,
+      "loss": 1.5964,
+      "step": 55400
+    },
+    {
+      "epoch": 1.77312,
+      "grad_norm": 44.890193939208984,
+      "learning_rate": 2.520888888888889e-06,
+      "loss": 1.5641,
+      "step": 55410
+    },
+    {
+      "epoch": 1.77344,
+      "grad_norm": 46.220298767089844,
+      "learning_rate": 2.5173333333333334e-06,
+      "loss": 1.5601,
+      "step": 55420
+    },
+    {
+      "epoch": 1.77376,
+      "grad_norm": 46.14614486694336,
+      "learning_rate": 2.5137777777777776e-06,
+      "loss": 1.5798,
+      "step": 55430
+    },
+    {
+      "epoch": 1.77408,
+      "grad_norm": 46.30167770385742,
+      "learning_rate": 2.510222222222222e-06,
+      "loss": 1.6126,
+      "step": 55440
+    },
+    {
+      "epoch": 1.7744,
+      "grad_norm": 46.571044921875,
+      "learning_rate": 2.5066666666666667e-06,
+      "loss": 1.6008,
+      "step": 55450
+    },
+    {
+      "epoch": 1.7747199999999999,
+      "grad_norm": 46.22939682006836,
+      "learning_rate": 2.5031111111111112e-06,
+      "loss": 1.59,
+      "step": 55460
+    },
+    {
+      "epoch": 1.77504,
+      "grad_norm": 46.0821533203125,
+      "learning_rate": 2.4995555555555558e-06,
+      "loss": 1.604,
+      "step": 55470
+    },
+    {
+      "epoch": 1.77536,
+      "grad_norm": 44.83059310913086,
+      "learning_rate": 2.496e-06,
+      "loss": 1.5583,
+      "step": 55480
+    },
+    {
+      "epoch": 1.77568,
+      "grad_norm": 45.116336822509766,
+      "learning_rate": 2.4924444444444445e-06,
+      "loss": 1.5915,
+      "step": 55490
+    },
+    {
+      "epoch": 1.776,
+      "grad_norm": 45.40968704223633,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 1.5715,
+      "step": 55500
+    },
+    {
+      "epoch": 1.7763200000000001,
+      "grad_norm": 47.14876937866211,
+      "learning_rate": 2.4853333333333336e-06,
+      "loss": 1.5946,
+      "step": 55510
+    },
+    {
+      "epoch": 1.77664,
+      "grad_norm": 45.58320617675781,
+      "learning_rate": 2.481777777777778e-06,
+      "loss": 1.5863,
+      "step": 55520
+    },
+    {
+      "epoch": 1.7769599999999999,
+      "grad_norm": 44.24687576293945,
+      "learning_rate": 2.4782222222222222e-06,
+      "loss": 1.6072,
+      "step": 55530
+    },
+    {
+      "epoch": 1.77728,
+      "grad_norm": 44.717613220214844,
+      "learning_rate": 2.474666666666667e-06,
+      "loss": 1.5911,
+      "step": 55540
+    },
+    {
+      "epoch": 1.7776,
+      "grad_norm": 45.04279327392578,
+      "learning_rate": 2.4711111111111114e-06,
+      "loss": 1.5845,
+      "step": 55550
+    },
+    {
+      "epoch": 1.77792,
+      "grad_norm": 46.93659591674805,
+      "learning_rate": 2.467555555555556e-06,
+      "loss": 1.578,
+      "step": 55560
+    },
+    {
+      "epoch": 1.77824,
+      "grad_norm": 44.31327819824219,
+      "learning_rate": 2.4640000000000005e-06,
+      "loss": 1.5721,
+      "step": 55570
+    },
+    {
+      "epoch": 1.7785600000000001,
+      "grad_norm": 45.77568817138672,
+      "learning_rate": 2.4604444444444446e-06,
+      "loss": 1.5834,
+      "step": 55580
+    },
+    {
+      "epoch": 1.77888,
+      "grad_norm": 44.93718338012695,
+      "learning_rate": 2.456888888888889e-06,
+      "loss": 1.5598,
+      "step": 55590
+    },
+    {
+      "epoch": 1.7792,
+      "grad_norm": 47.456947326660156,
+      "learning_rate": 2.4533333333333333e-06,
+      "loss": 1.5909,
+      "step": 55600
+    },
+    {
+      "epoch": 1.77952,
+      "grad_norm": 46.9789924621582,
+      "learning_rate": 2.449777777777778e-06,
+      "loss": 1.5777,
+      "step": 55610
+    },
+    {
+      "epoch": 1.77984,
+      "grad_norm": 53.498592376708984,
+      "learning_rate": 2.4462222222222224e-06,
+      "loss": 1.5873,
+      "step": 55620
+    },
+    {
+      "epoch": 1.78016,
+      "grad_norm": 47.94544982910156,
+      "learning_rate": 2.442666666666667e-06,
+      "loss": 1.5902,
+      "step": 55630
+    },
+    {
+      "epoch": 1.7804799999999998,
+      "grad_norm": 44.35692596435547,
+      "learning_rate": 2.439111111111111e-06,
+      "loss": 1.5891,
+      "step": 55640
+    },
+    {
+      "epoch": 1.7808000000000002,
+      "grad_norm": 45.31683349609375,
+      "learning_rate": 2.4355555555555556e-06,
+      "loss": 1.5967,
+      "step": 55650
+    },
+    {
+      "epoch": 1.78112,
+      "grad_norm": 46.0125732421875,
+      "learning_rate": 2.432e-06,
+      "loss": 1.5759,
+      "step": 55660
+    },
+    {
+      "epoch": 1.78144,
+      "grad_norm": 45.48175811767578,
+      "learning_rate": 2.4284444444444447e-06,
+      "loss": 1.5829,
+      "step": 55670
+    },
+    {
+      "epoch": 1.78176,
+      "grad_norm": 45.606834411621094,
+      "learning_rate": 2.4248888888888893e-06,
+      "loss": 1.5964,
+      "step": 55680
+    },
+    {
+      "epoch": 1.78208,
+      "grad_norm": 61.35505294799805,
+      "learning_rate": 2.4213333333333334e-06,
+      "loss": 1.596,
+      "step": 55690
+    },
+    {
+      "epoch": 1.7824,
+      "grad_norm": 46.245399475097656,
+      "learning_rate": 2.417777777777778e-06,
+      "loss": 1.5756,
+      "step": 55700
+    },
+    {
+      "epoch": 1.7827199999999999,
+      "grad_norm": 45.621490478515625,
+      "learning_rate": 2.4142222222222225e-06,
+      "loss": 1.6155,
+      "step": 55710
+    },
+    {
+      "epoch": 1.78304,
+      "grad_norm": 46.63582992553711,
+      "learning_rate": 2.410666666666667e-06,
+      "loss": 1.61,
+      "step": 55720
+    },
+    {
+      "epoch": 1.78336,
+      "grad_norm": 46.067195892333984,
+      "learning_rate": 2.407111111111111e-06,
+      "loss": 1.6075,
+      "step": 55730
+    },
+    {
+      "epoch": 1.78368,
+      "grad_norm": 47.03789520263672,
+      "learning_rate": 2.4035555555555558e-06,
+      "loss": 1.6083,
+      "step": 55740
+    },
+    {
+      "epoch": 1.784,
+      "grad_norm": 43.586143493652344,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.5847,
+      "step": 55750
+    },
+    {
+      "epoch": 1.7843200000000001,
+      "grad_norm": 46.635459899902344,
+      "learning_rate": 2.3964444444444444e-06,
+      "loss": 1.5636,
+      "step": 55760
+    },
+    {
+      "epoch": 1.78464,
+      "grad_norm": 47.186370849609375,
+      "learning_rate": 2.392888888888889e-06,
+      "loss": 1.5964,
+      "step": 55770
+    },
+    {
+      "epoch": 1.7849599999999999,
+      "grad_norm": 45.05913543701172,
+      "learning_rate": 2.3893333333333336e-06,
+      "loss": 1.589,
+      "step": 55780
+    },
+    {
+      "epoch": 1.78528,
+      "grad_norm": 45.96673583984375,
+      "learning_rate": 2.385777777777778e-06,
+      "loss": 1.5975,
+      "step": 55790
+    },
+    {
+      "epoch": 1.7856,
+      "grad_norm": 49.0675163269043,
+      "learning_rate": 2.3822222222222222e-06,
+      "loss": 1.6115,
+      "step": 55800
+    },
+    {
+      "epoch": 1.78592,
+      "grad_norm": 45.164005279541016,
+      "learning_rate": 2.378666666666667e-06,
+      "loss": 1.573,
+      "step": 55810
+    },
+    {
+      "epoch": 1.78624,
+      "grad_norm": 47.418373107910156,
+      "learning_rate": 2.3751111111111113e-06,
+      "loss": 1.5898,
+      "step": 55820
+    },
+    {
+      "epoch": 1.7865600000000001,
+      "grad_norm": 47.338897705078125,
+      "learning_rate": 2.371555555555556e-06,
+      "loss": 1.5831,
+      "step": 55830
+    },
+    {
+      "epoch": 1.78688,
+      "grad_norm": 46.223323822021484,
+      "learning_rate": 2.3680000000000005e-06,
+      "loss": 1.5614,
+      "step": 55840
+    },
+    {
+      "epoch": 1.7872,
+      "grad_norm": 46.24585723876953,
+      "learning_rate": 2.3644444444444446e-06,
+      "loss": 1.5753,
+      "step": 55850
+    },
+    {
+      "epoch": 1.78752,
+      "grad_norm": 45.412471771240234,
+      "learning_rate": 2.360888888888889e-06,
+      "loss": 1.5642,
+      "step": 55860
+    },
+    {
+      "epoch": 1.78784,
+      "grad_norm": 46.19806671142578,
+      "learning_rate": 2.3573333333333333e-06,
+      "loss": 1.6342,
+      "step": 55870
+    },
+    {
+      "epoch": 1.78816,
+      "grad_norm": 45.83122634887695,
+      "learning_rate": 2.353777777777778e-06,
+      "loss": 1.5747,
+      "step": 55880
+    },
+    {
+      "epoch": 1.7884799999999998,
+      "grad_norm": 46.43621826171875,
+      "learning_rate": 2.3502222222222224e-06,
+      "loss": 1.6143,
+      "step": 55890
+    },
+    {
+      "epoch": 1.7888,
+      "grad_norm": 46.50686264038086,
+      "learning_rate": 2.346666666666667e-06,
+      "loss": 1.6208,
+      "step": 55900
+    },
+    {
+      "epoch": 1.78912,
+      "grad_norm": 46.26643371582031,
+      "learning_rate": 2.3431111111111115e-06,
+      "loss": 1.6048,
+      "step": 55910
+    },
+    {
+      "epoch": 1.78944,
+      "grad_norm": 46.18266677856445,
+      "learning_rate": 2.3395555555555556e-06,
+      "loss": 1.5608,
+      "step": 55920
+    },
+    {
+      "epoch": 1.78976,
+      "grad_norm": 44.978973388671875,
+      "learning_rate": 2.336e-06,
+      "loss": 1.5868,
+      "step": 55930
+    },
+    {
+      "epoch": 1.7900800000000001,
+      "grad_norm": 46.06013107299805,
+      "learning_rate": 2.3324444444444447e-06,
+      "loss": 1.6047,
+      "step": 55940
+    },
+    {
+      "epoch": 1.7904,
+      "grad_norm": 45.7539176940918,
+      "learning_rate": 2.3288888888888893e-06,
+      "loss": 1.5809,
+      "step": 55950
+    },
+    {
+      "epoch": 1.7907199999999999,
+      "grad_norm": 44.96236801147461,
+      "learning_rate": 2.3253333333333334e-06,
+      "loss": 1.5617,
+      "step": 55960
+    },
+    {
+      "epoch": 1.79104,
+      "grad_norm": 46.3323974609375,
+      "learning_rate": 2.321777777777778e-06,
+      "loss": 1.5557,
+      "step": 55970
+    },
+    {
+      "epoch": 1.79136,
+      "grad_norm": 45.12177276611328,
+      "learning_rate": 2.3182222222222225e-06,
+      "loss": 1.6195,
+      "step": 55980
+    },
+    {
+      "epoch": 1.79168,
+      "grad_norm": 46.304664611816406,
+      "learning_rate": 2.3146666666666666e-06,
+      "loss": 1.5838,
+      "step": 55990
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 46.32132339477539,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 1.5968,
+      "step": 56000
+    },
+    {
+      "epoch": 1.7923200000000001,
+      "grad_norm": 45.86129379272461,
+      "learning_rate": 2.3075555555555557e-06,
+      "loss": 1.5952,
+      "step": 56010
+    },
+    {
+      "epoch": 1.79264,
+      "grad_norm": 46.15711212158203,
+      "learning_rate": 2.3040000000000003e-06,
+      "loss": 1.568,
+      "step": 56020
+    },
+    {
+      "epoch": 1.7929599999999999,
+      "grad_norm": 46.497249603271484,
+      "learning_rate": 2.3004444444444444e-06,
+      "loss": 1.6112,
+      "step": 56030
+    },
+    {
+      "epoch": 1.79328,
+      "grad_norm": 45.105045318603516,
+      "learning_rate": 2.296888888888889e-06,
+      "loss": 1.6041,
+      "step": 56040
+    },
+    {
+      "epoch": 1.7936,
+      "grad_norm": 44.266658782958984,
+      "learning_rate": 2.2933333333333335e-06,
+      "loss": 1.5485,
+      "step": 56050
+    },
+    {
+      "epoch": 1.79392,
+      "grad_norm": 45.53001022338867,
+      "learning_rate": 2.289777777777778e-06,
+      "loss": 1.606,
+      "step": 56060
+    },
+    {
+      "epoch": 1.7942399999999998,
+      "grad_norm": 47.24198913574219,
+      "learning_rate": 2.2862222222222226e-06,
+      "loss": 1.5648,
+      "step": 56070
+    },
+    {
+      "epoch": 1.7945600000000002,
+      "grad_norm": 46.86454391479492,
+      "learning_rate": 2.2826666666666668e-06,
+      "loss": 1.5816,
+      "step": 56080
+    },
+    {
+      "epoch": 1.79488,
+      "grad_norm": 47.774723052978516,
+      "learning_rate": 2.2791111111111113e-06,
+      "loss": 1.6021,
+      "step": 56090
+    },
+    {
+      "epoch": 1.7952,
+      "grad_norm": 45.98683166503906,
+      "learning_rate": 2.275555555555556e-06,
+      "loss": 1.5843,
+      "step": 56100
+    },
+    {
+      "epoch": 1.79552,
+      "grad_norm": 45.332523345947266,
+      "learning_rate": 2.2720000000000004e-06,
+      "loss": 1.6126,
+      "step": 56110
+    },
+    {
+      "epoch": 1.79584,
+      "grad_norm": 46.97977066040039,
+      "learning_rate": 2.2684444444444446e-06,
+      "loss": 1.5895,
+      "step": 56120
+    },
+    {
+      "epoch": 1.79616,
+      "grad_norm": 45.19950866699219,
+      "learning_rate": 2.264888888888889e-06,
+      "loss": 1.6076,
+      "step": 56130
+    },
+    {
+      "epoch": 1.7964799999999999,
+      "grad_norm": 46.60654830932617,
+      "learning_rate": 2.2613333333333333e-06,
+      "loss": 1.5804,
+      "step": 56140
+    },
+    {
+      "epoch": 1.7968,
+      "grad_norm": 43.94088363647461,
+      "learning_rate": 2.257777777777778e-06,
+      "loss": 1.5936,
+      "step": 56150
+    },
+    {
+      "epoch": 1.79712,
+      "grad_norm": 46.990787506103516,
+      "learning_rate": 2.2542222222222224e-06,
+      "loss": 1.617,
+      "step": 56160
+    },
+    {
+      "epoch": 1.79744,
+      "grad_norm": 45.92596435546875,
+      "learning_rate": 2.250666666666667e-06,
+      "loss": 1.59,
+      "step": 56170
+    },
+    {
+      "epoch": 1.79776,
+      "grad_norm": 45.383148193359375,
+      "learning_rate": 2.2471111111111115e-06,
+      "loss": 1.5845,
+      "step": 56180
+    },
+    {
+      "epoch": 1.7980800000000001,
+      "grad_norm": 45.40883255004883,
+      "learning_rate": 2.2435555555555556e-06,
+      "loss": 1.5849,
+      "step": 56190
+    },
+    {
+      "epoch": 1.7984,
+      "grad_norm": 45.579917907714844,
+      "learning_rate": 2.24e-06,
+      "loss": 1.5701,
+      "step": 56200
+    },
+    {
+      "epoch": 1.7987199999999999,
+      "grad_norm": 44.44224548339844,
+      "learning_rate": 2.2364444444444447e-06,
+      "loss": 1.5925,
+      "step": 56210
+    },
+    {
+      "epoch": 1.79904,
+      "grad_norm": 45.03302764892578,
+      "learning_rate": 2.2328888888888893e-06,
+      "loss": 1.5808,
+      "step": 56220
+    },
+    {
+      "epoch": 1.79936,
+      "grad_norm": 46.868614196777344,
+      "learning_rate": 2.229333333333334e-06,
+      "loss": 1.5644,
+      "step": 56230
+    },
+    {
+      "epoch": 1.79968,
+      "grad_norm": 45.28606033325195,
+      "learning_rate": 2.225777777777778e-06,
+      "loss": 1.5832,
+      "step": 56240
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 45.76778793334961,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.587,
+      "step": 56250
+    },
+    {
+      "epoch": 1.8003200000000001,
+      "grad_norm": 44.555789947509766,
+      "learning_rate": 2.2186666666666666e-06,
+      "loss": 1.5767,
+      "step": 56260
+    },
+    {
+      "epoch": 1.80064,
+      "grad_norm": 45.19745635986328,
+      "learning_rate": 2.215111111111111e-06,
+      "loss": 1.6335,
+      "step": 56270
+    },
+    {
+      "epoch": 1.80096,
+      "grad_norm": 44.41781997680664,
+      "learning_rate": 2.2115555555555557e-06,
+      "loss": 1.6064,
+      "step": 56280
+    },
+    {
+      "epoch": 1.80128,
+      "grad_norm": 45.24692916870117,
+      "learning_rate": 2.2080000000000003e-06,
+      "loss": 1.5886,
+      "step": 56290
+    },
+    {
+      "epoch": 1.8016,
+      "grad_norm": 44.298160552978516,
+      "learning_rate": 2.2044444444444444e-06,
+      "loss": 1.5731,
+      "step": 56300
+    },
+    {
+      "epoch": 1.80192,
+      "grad_norm": 48.903236389160156,
+      "learning_rate": 2.200888888888889e-06,
+      "loss": 1.5567,
+      "step": 56310
+    },
+    {
+      "epoch": 1.8022399999999998,
+      "grad_norm": 44.9493293762207,
+      "learning_rate": 2.1973333333333335e-06,
+      "loss": 1.5892,
+      "step": 56320
+    },
+    {
+      "epoch": 1.8025600000000002,
+      "grad_norm": 46.3359489440918,
+      "learning_rate": 2.193777777777778e-06,
+      "loss": 1.5631,
+      "step": 56330
+    },
+    {
+      "epoch": 1.80288,
+      "grad_norm": 46.73094940185547,
+      "learning_rate": 2.1902222222222226e-06,
+      "loss": 1.5904,
+      "step": 56340
+    },
+    {
+      "epoch": 1.8032,
+      "grad_norm": 47.31883239746094,
+      "learning_rate": 2.1866666666666668e-06,
+      "loss": 1.5641,
+      "step": 56350
+    },
+    {
+      "epoch": 1.80352,
+      "grad_norm": 49.820865631103516,
+      "learning_rate": 2.1831111111111113e-06,
+      "loss": 1.6101,
+      "step": 56360
+    },
+    {
+      "epoch": 1.80384,
+      "grad_norm": 46.607112884521484,
+      "learning_rate": 2.179555555555556e-06,
+      "loss": 1.5932,
+      "step": 56370
+    },
+    {
+      "epoch": 1.80416,
+      "grad_norm": 45.860992431640625,
+      "learning_rate": 2.176e-06,
+      "loss": 1.5899,
+      "step": 56380
+    },
+    {
+      "epoch": 1.8044799999999999,
+      "grad_norm": 44.20576477050781,
+      "learning_rate": 2.1724444444444446e-06,
+      "loss": 1.5934,
+      "step": 56390
+    },
+    {
+      "epoch": 1.8048,
+      "grad_norm": 45.662288665771484,
+      "learning_rate": 2.168888888888889e-06,
+      "loss": 1.5845,
+      "step": 56400
+    },
+    {
+      "epoch": 1.80512,
+      "grad_norm": 46.737857818603516,
+      "learning_rate": 2.1653333333333337e-06,
+      "loss": 1.5815,
+      "step": 56410
+    },
+    {
+      "epoch": 1.80544,
+      "grad_norm": 46.261051177978516,
+      "learning_rate": 2.161777777777778e-06,
+      "loss": 1.5693,
+      "step": 56420
+    },
+    {
+      "epoch": 1.80576,
+      "grad_norm": 45.66044235229492,
+      "learning_rate": 2.1582222222222223e-06,
+      "loss": 1.6025,
+      "step": 56430
+    },
+    {
+      "epoch": 1.8060800000000001,
+      "grad_norm": 45.01734924316406,
+      "learning_rate": 2.154666666666667e-06,
+      "loss": 1.5983,
+      "step": 56440
+    },
+    {
+      "epoch": 1.8064,
+      "grad_norm": 45.47689437866211,
+      "learning_rate": 2.1511111111111115e-06,
+      "loss": 1.6084,
+      "step": 56450
+    },
+    {
+      "epoch": 1.8067199999999999,
+      "grad_norm": 45.470767974853516,
+      "learning_rate": 2.1475555555555556e-06,
+      "loss": 1.584,
+      "step": 56460
+    },
+    {
+      "epoch": 1.80704,
+      "grad_norm": 45.65459060668945,
+      "learning_rate": 2.144e-06,
+      "loss": 1.5848,
+      "step": 56470
+    },
+    {
+      "epoch": 1.80736,
+      "grad_norm": 46.29679489135742,
+      "learning_rate": 2.1404444444444447e-06,
+      "loss": 1.5596,
+      "step": 56480
+    },
+    {
+      "epoch": 1.80768,
+      "grad_norm": 43.89360046386719,
+      "learning_rate": 2.1368888888888892e-06,
+      "loss": 1.5724,
+      "step": 56490
+    },
+    {
+      "epoch": 1.808,
+      "grad_norm": 44.68407440185547,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.5904,
+      "step": 56500
+    },
+    {
+      "epoch": 1.8083200000000001,
+      "grad_norm": 44.551185607910156,
+      "learning_rate": 2.129777777777778e-06,
+      "loss": 1.5921,
+      "step": 56510
+    },
+    {
+      "epoch": 1.80864,
+      "grad_norm": 47.263126373291016,
+      "learning_rate": 2.1262222222222225e-06,
+      "loss": 1.5847,
+      "step": 56520
+    },
+    {
+      "epoch": 1.80896,
+      "grad_norm": 45.34291458129883,
+      "learning_rate": 2.1226666666666666e-06,
+      "loss": 1.6079,
+      "step": 56530
+    },
+    {
+      "epoch": 1.80928,
+      "grad_norm": 46.09779739379883,
+      "learning_rate": 2.119111111111111e-06,
+      "loss": 1.6251,
+      "step": 56540
+    },
+    {
+      "epoch": 1.8096,
+      "grad_norm": 46.95382308959961,
+      "learning_rate": 2.1155555555555557e-06,
+      "loss": 1.5672,
+      "step": 56550
+    },
+    {
+      "epoch": 1.80992,
+      "grad_norm": 45.27779006958008,
+      "learning_rate": 2.1120000000000003e-06,
+      "loss": 1.5892,
+      "step": 56560
+    },
+    {
+      "epoch": 1.8102399999999998,
+      "grad_norm": 44.9322624206543,
+      "learning_rate": 2.108444444444445e-06,
+      "loss": 1.5847,
+      "step": 56570
+    },
+    {
+      "epoch": 1.81056,
+      "grad_norm": 48.82658004760742,
+      "learning_rate": 2.104888888888889e-06,
+      "loss": 1.5899,
+      "step": 56580
+    },
+    {
+      "epoch": 1.81088,
+      "grad_norm": 48.06019973754883,
+      "learning_rate": 2.1013333333333335e-06,
+      "loss": 1.5513,
+      "step": 56590
+    },
+    {
+      "epoch": 1.8112,
+      "grad_norm": 45.88014221191406,
+      "learning_rate": 2.097777777777778e-06,
+      "loss": 1.5952,
+      "step": 56600
+    },
+    {
+      "epoch": 1.81152,
+      "grad_norm": 42.34318161010742,
+      "learning_rate": 2.0942222222222226e-06,
+      "loss": 1.5755,
+      "step": 56610
+    },
+    {
+      "epoch": 1.8118400000000001,
+      "grad_norm": 45.93275451660156,
+      "learning_rate": 2.0906666666666668e-06,
+      "loss": 1.5776,
+      "step": 56620
+    },
+    {
+      "epoch": 1.81216,
+      "grad_norm": 43.69744873046875,
+      "learning_rate": 2.0871111111111113e-06,
+      "loss": 1.5893,
+      "step": 56630
+    },
+    {
+      "epoch": 1.8124799999999999,
+      "grad_norm": 44.21855926513672,
+      "learning_rate": 2.0835555555555554e-06,
+      "loss": 1.5874,
+      "step": 56640
+    },
+    {
+      "epoch": 1.8128,
+      "grad_norm": 45.70142364501953,
+      "learning_rate": 2.08e-06,
+      "loss": 1.5684,
+      "step": 56650
+    },
+    {
+      "epoch": 1.81312,
+      "grad_norm": 46.35956573486328,
+      "learning_rate": 2.0764444444444445e-06,
+      "loss": 1.6093,
+      "step": 56660
+    },
+    {
+      "epoch": 1.81344,
+      "grad_norm": 47.33995056152344,
+      "learning_rate": 2.072888888888889e-06,
+      "loss": 1.5873,
+      "step": 56670
+    },
+    {
+      "epoch": 1.81376,
+      "grad_norm": 45.1085205078125,
+      "learning_rate": 2.0693333333333337e-06,
+      "loss": 1.5867,
+      "step": 56680
+    },
+    {
+      "epoch": 1.8140800000000001,
+      "grad_norm": 46.70127868652344,
+      "learning_rate": 2.0657777777777778e-06,
+      "loss": 1.6007,
+      "step": 56690
+    },
+    {
+      "epoch": 1.8144,
+      "grad_norm": 45.99809646606445,
+      "learning_rate": 2.0622222222222223e-06,
+      "loss": 1.6079,
+      "step": 56700
+    },
+    {
+      "epoch": 1.8147199999999999,
+      "grad_norm": 46.1832275390625,
+      "learning_rate": 2.058666666666667e-06,
+      "loss": 1.6042,
+      "step": 56710
+    },
+    {
+      "epoch": 1.81504,
+      "grad_norm": 47.55323791503906,
+      "learning_rate": 2.0551111111111114e-06,
+      "loss": 1.5965,
+      "step": 56720
+    },
+    {
+      "epoch": 1.81536,
+      "grad_norm": 45.24617004394531,
+      "learning_rate": 2.051555555555556e-06,
+      "loss": 1.5797,
+      "step": 56730
+    },
+    {
+      "epoch": 1.81568,
+      "grad_norm": 44.48154067993164,
+      "learning_rate": 2.048e-06,
+      "loss": 1.568,
+      "step": 56740
+    },
+    {
+      "epoch": 1.8159999999999998,
+      "grad_norm": 47.31721878051758,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 1.5819,
+      "step": 56750
+    },
+    {
+      "epoch": 1.8163200000000002,
+      "grad_norm": 58.64398193359375,
+      "learning_rate": 2.040888888888889e-06,
+      "loss": 1.5612,
+      "step": 56760
+    },
+    {
+      "epoch": 1.81664,
+      "grad_norm": 48.63230514526367,
+      "learning_rate": 2.0373333333333334e-06,
+      "loss": 1.6083,
+      "step": 56770
+    },
+    {
+      "epoch": 1.81696,
+      "grad_norm": 47.880393981933594,
+      "learning_rate": 2.033777777777778e-06,
+      "loss": 1.5927,
+      "step": 56780
+    },
+    {
+      "epoch": 1.81728,
+      "grad_norm": 45.82305145263672,
+      "learning_rate": 2.0302222222222225e-06,
+      "loss": 1.5959,
+      "step": 56790
+    },
+    {
+      "epoch": 1.8176,
+      "grad_norm": 44.46937561035156,
+      "learning_rate": 2.0266666666666666e-06,
+      "loss": 1.6333,
+      "step": 56800
+    },
+    {
+      "epoch": 1.81792,
+      "grad_norm": 45.742794036865234,
+      "learning_rate": 2.023111111111111e-06,
+      "loss": 1.5872,
+      "step": 56810
+    },
+    {
+      "epoch": 1.8182399999999999,
+      "grad_norm": 45.73196029663086,
+      "learning_rate": 2.0195555555555557e-06,
+      "loss": 1.5682,
+      "step": 56820
+    },
+    {
+      "epoch": 1.81856,
+      "grad_norm": 45.85786437988281,
+      "learning_rate": 2.0160000000000003e-06,
+      "loss": 1.5906,
+      "step": 56830
+    },
+    {
+      "epoch": 1.81888,
+      "grad_norm": 45.84141159057617,
+      "learning_rate": 2.012444444444445e-06,
+      "loss": 1.5962,
+      "step": 56840
+    },
+    {
+      "epoch": 1.8192,
+      "grad_norm": 45.378883361816406,
+      "learning_rate": 2.008888888888889e-06,
+      "loss": 1.5749,
+      "step": 56850
+    },
+    {
+      "epoch": 1.81952,
+      "grad_norm": 46.715023040771484,
+      "learning_rate": 2.0053333333333335e-06,
+      "loss": 1.5586,
+      "step": 56860
+    },
+    {
+      "epoch": 1.8198400000000001,
+      "grad_norm": 45.816993713378906,
+      "learning_rate": 2.001777777777778e-06,
+      "loss": 1.5638,
+      "step": 56870
+    },
+    {
+      "epoch": 1.82016,
+      "grad_norm": 44.63705062866211,
+      "learning_rate": 1.9982222222222226e-06,
+      "loss": 1.5921,
+      "step": 56880
+    },
+    {
+      "epoch": 1.8204799999999999,
+      "grad_norm": 45.8122444152832,
+      "learning_rate": 1.9946666666666667e-06,
+      "loss": 1.6025,
+      "step": 56890
+    },
+    {
+      "epoch": 1.8208,
+      "grad_norm": 45.895965576171875,
+      "learning_rate": 1.9911111111111113e-06,
+      "loss": 1.5821,
+      "step": 56900
+    },
+    {
+      "epoch": 1.82112,
+      "grad_norm": 45.55632019042969,
+      "learning_rate": 1.9875555555555554e-06,
+      "loss": 1.6013,
+      "step": 56910
+    },
+    {
+      "epoch": 1.82144,
+      "grad_norm": 46.17961883544922,
+      "learning_rate": 1.984e-06,
+      "loss": 1.5889,
+      "step": 56920
+    },
+    {
+      "epoch": 1.82176,
+      "grad_norm": 46.277687072753906,
+      "learning_rate": 1.9804444444444445e-06,
+      "loss": 1.5787,
+      "step": 56930
+    },
+    {
+      "epoch": 1.8220800000000001,
+      "grad_norm": 45.59478759765625,
+      "learning_rate": 1.976888888888889e-06,
+      "loss": 1.588,
+      "step": 56940
+    },
+    {
+      "epoch": 1.8224,
+      "grad_norm": 45.2066764831543,
+      "learning_rate": 1.9733333333333336e-06,
+      "loss": 1.5655,
+      "step": 56950
+    },
+    {
+      "epoch": 1.82272,
+      "grad_norm": 44.63209533691406,
+      "learning_rate": 1.9697777777777778e-06,
+      "loss": 1.574,
+      "step": 56960
+    },
+    {
+      "epoch": 1.82304,
+      "grad_norm": 47.10441589355469,
+      "learning_rate": 1.9662222222222223e-06,
+      "loss": 1.5783,
+      "step": 56970
+    },
+    {
+      "epoch": 1.82336,
+      "grad_norm": 45.74570083618164,
+      "learning_rate": 1.962666666666667e-06,
+      "loss": 1.5658,
+      "step": 56980
+    },
+    {
+      "epoch": 1.82368,
+      "grad_norm": 44.77720642089844,
+      "learning_rate": 1.9591111111111114e-06,
+      "loss": 1.5996,
+      "step": 56990
+    },
+    {
+      "epoch": 1.8239999999999998,
+      "grad_norm": 46.88585662841797,
+      "learning_rate": 1.955555555555556e-06,
+      "loss": 1.6126,
+      "step": 57000
+    },
+    {
+      "epoch": 1.8243200000000002,
+      "grad_norm": 47.91762161254883,
+      "learning_rate": 1.952e-06,
+      "loss": 1.5885,
+      "step": 57010
+    },
+    {
+      "epoch": 1.82464,
+      "grad_norm": 45.013240814208984,
+      "learning_rate": 1.9484444444444447e-06,
+      "loss": 1.5731,
+      "step": 57020
+    },
+    {
+      "epoch": 1.82496,
+      "grad_norm": 44.162113189697266,
+      "learning_rate": 1.944888888888889e-06,
+      "loss": 1.6139,
+      "step": 57030
+    },
+    {
+      "epoch": 1.82528,
+      "grad_norm": 46.55525588989258,
+      "learning_rate": 1.9413333333333334e-06,
+      "loss": 1.5635,
+      "step": 57040
+    },
+    {
+      "epoch": 1.8256000000000001,
+      "grad_norm": 46.97242736816406,
+      "learning_rate": 1.937777777777778e-06,
+      "loss": 1.5949,
+      "step": 57050
+    },
+    {
+      "epoch": 1.82592,
+      "grad_norm": 44.33784484863281,
+      "learning_rate": 1.9342222222222225e-06,
+      "loss": 1.5636,
+      "step": 57060
+    },
+    {
+      "epoch": 1.8262399999999999,
+      "grad_norm": 44.85808181762695,
+      "learning_rate": 1.9306666666666666e-06,
+      "loss": 1.599,
+      "step": 57070
+    },
+    {
+      "epoch": 1.82656,
+      "grad_norm": 48.044742584228516,
+      "learning_rate": 1.927111111111111e-06,
+      "loss": 1.5846,
+      "step": 57080
+    },
+    {
+      "epoch": 1.82688,
+      "grad_norm": 48.76578140258789,
+      "learning_rate": 1.9235555555555557e-06,
+      "loss": 1.5993,
+      "step": 57090
+    },
+    {
+      "epoch": 1.8272,
+      "grad_norm": 46.712890625,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.5997,
+      "step": 57100
+    },
+    {
+      "epoch": 1.82752,
+      "grad_norm": 44.625770568847656,
+      "learning_rate": 1.916444444444445e-06,
+      "loss": 1.5811,
+      "step": 57110
+    },
+    {
+      "epoch": 1.8278400000000001,
+      "grad_norm": 47.16094970703125,
+      "learning_rate": 1.912888888888889e-06,
+      "loss": 1.5732,
+      "step": 57120
+    },
+    {
+      "epoch": 1.82816,
+      "grad_norm": 46.85769271850586,
+      "learning_rate": 1.9093333333333335e-06,
+      "loss": 1.5915,
+      "step": 57130
+    },
+    {
+      "epoch": 1.8284799999999999,
+      "grad_norm": 46.92313766479492,
+      "learning_rate": 1.905777777777778e-06,
+      "loss": 1.5702,
+      "step": 57140
+    },
+    {
+      "epoch": 1.8288,
+      "grad_norm": 46.07356643676758,
+      "learning_rate": 1.9022222222222222e-06,
+      "loss": 1.6097,
+      "step": 57150
+    },
+    {
+      "epoch": 1.82912,
+      "grad_norm": 46.94747543334961,
+      "learning_rate": 1.8986666666666667e-06,
+      "loss": 1.588,
+      "step": 57160
+    },
+    {
+      "epoch": 1.82944,
+      "grad_norm": 46.5937614440918,
+      "learning_rate": 1.8951111111111113e-06,
+      "loss": 1.6113,
+      "step": 57170
+    },
+    {
+      "epoch": 1.82976,
+      "grad_norm": 47.07693862915039,
+      "learning_rate": 1.8915555555555556e-06,
+      "loss": 1.5907,
+      "step": 57180
+    },
+    {
+      "epoch": 1.8300800000000002,
+      "grad_norm": 46.30846405029297,
+      "learning_rate": 1.8880000000000002e-06,
+      "loss": 1.591,
+      "step": 57190
+    },
+    {
+      "epoch": 1.8304,
+      "grad_norm": 48.294960021972656,
+      "learning_rate": 1.8844444444444445e-06,
+      "loss": 1.5888,
+      "step": 57200
+    },
+    {
+      "epoch": 1.83072,
+      "grad_norm": 45.11846160888672,
+      "learning_rate": 1.880888888888889e-06,
+      "loss": 1.579,
+      "step": 57210
+    },
+    {
+      "epoch": 1.83104,
+      "grad_norm": 44.790489196777344,
+      "learning_rate": 1.8773333333333334e-06,
+      "loss": 1.591,
+      "step": 57220
+    },
+    {
+      "epoch": 1.83136,
+      "grad_norm": 45.15334701538086,
+      "learning_rate": 1.873777777777778e-06,
+      "loss": 1.61,
+      "step": 57230
+    },
+    {
+      "epoch": 1.83168,
+      "grad_norm": 46.32352828979492,
+      "learning_rate": 1.8702222222222225e-06,
+      "loss": 1.61,
+      "step": 57240
+    },
+    {
+      "epoch": 1.8319999999999999,
+      "grad_norm": 46.18299865722656,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.5517,
+      "step": 57250
+    },
+    {
+      "epoch": 1.83232,
+      "grad_norm": 44.044254302978516,
+      "learning_rate": 1.8631111111111114e-06,
+      "loss": 1.5758,
+      "step": 57260
+    },
+    {
+      "epoch": 1.83264,
+      "grad_norm": 46.72759246826172,
+      "learning_rate": 1.8595555555555558e-06,
+      "loss": 1.6195,
+      "step": 57270
+    },
+    {
+      "epoch": 1.83296,
+      "grad_norm": 45.010780334472656,
+      "learning_rate": 1.856e-06,
+      "loss": 1.5719,
+      "step": 57280
+    },
+    {
+      "epoch": 1.83328,
+      "grad_norm": 44.996883392333984,
+      "learning_rate": 1.8524444444444444e-06,
+      "loss": 1.5723,
+      "step": 57290
+    },
+    {
+      "epoch": 1.8336000000000001,
+      "grad_norm": 46.22783660888672,
+      "learning_rate": 1.848888888888889e-06,
+      "loss": 1.6152,
+      "step": 57300
+    },
+    {
+      "epoch": 1.83392,
+      "grad_norm": 45.228973388671875,
+      "learning_rate": 1.8453333333333333e-06,
+      "loss": 1.6088,
+      "step": 57310
+    },
+    {
+      "epoch": 1.8342399999999999,
+      "grad_norm": 46.14060974121094,
+      "learning_rate": 1.8417777777777779e-06,
+      "loss": 1.5891,
+      "step": 57320
+    },
+    {
+      "epoch": 1.83456,
+      "grad_norm": 46.00922775268555,
+      "learning_rate": 1.8382222222222224e-06,
+      "loss": 1.5811,
+      "step": 57330
+    },
+    {
+      "epoch": 1.83488,
+      "grad_norm": 45.049034118652344,
+      "learning_rate": 1.8346666666666668e-06,
+      "loss": 1.5736,
+      "step": 57340
+    },
+    {
+      "epoch": 1.8352,
+      "grad_norm": 45.5625,
+      "learning_rate": 1.8311111111111113e-06,
+      "loss": 1.5981,
+      "step": 57350
+    },
+    {
+      "epoch": 1.83552,
+      "grad_norm": 46.143245697021484,
+      "learning_rate": 1.8275555555555557e-06,
+      "loss": 1.59,
+      "step": 57360
+    },
+    {
+      "epoch": 1.8358400000000001,
+      "grad_norm": 44.78925323486328,
+      "learning_rate": 1.8240000000000002e-06,
+      "loss": 1.5689,
+      "step": 57370
+    },
+    {
+      "epoch": 1.83616,
+      "grad_norm": 46.51466369628906,
+      "learning_rate": 1.8204444444444446e-06,
+      "loss": 1.5868,
+      "step": 57380
+    },
+    {
+      "epoch": 1.83648,
+      "grad_norm": 46.2537956237793,
+      "learning_rate": 1.8168888888888891e-06,
+      "loss": 1.5799,
+      "step": 57390
+    },
+    {
+      "epoch": 1.8368,
+      "grad_norm": 47.15256881713867,
+      "learning_rate": 1.8133333333333337e-06,
+      "loss": 1.5817,
+      "step": 57400
+    },
+    {
+      "epoch": 1.83712,
+      "grad_norm": 43.75941467285156,
+      "learning_rate": 1.8097777777777778e-06,
+      "loss": 1.5915,
+      "step": 57410
+    },
+    {
+      "epoch": 1.83744,
+      "grad_norm": 45.291229248046875,
+      "learning_rate": 1.8062222222222222e-06,
+      "loss": 1.584,
+      "step": 57420
+    },
+    {
+      "epoch": 1.8377599999999998,
+      "grad_norm": 46.50223159790039,
+      "learning_rate": 1.8026666666666667e-06,
+      "loss": 1.6023,
+      "step": 57430
+    },
+    {
+      "epoch": 1.8380800000000002,
+      "grad_norm": 45.744083404541016,
+      "learning_rate": 1.7991111111111113e-06,
+      "loss": 1.5789,
+      "step": 57440
+    },
+    {
+      "epoch": 1.8384,
+      "grad_norm": 46.86274337768555,
+      "learning_rate": 1.7955555555555556e-06,
+      "loss": 1.5859,
+      "step": 57450
+    },
+    {
+      "epoch": 1.83872,
+      "grad_norm": 44.3387336730957,
+      "learning_rate": 1.7920000000000002e-06,
+      "loss": 1.5576,
+      "step": 57460
+    },
+    {
+      "epoch": 1.83904,
+      "grad_norm": 48.54364013671875,
+      "learning_rate": 1.7884444444444445e-06,
+      "loss": 1.5805,
+      "step": 57470
+    },
+    {
+      "epoch": 1.83936,
+      "grad_norm": 45.73419952392578,
+      "learning_rate": 1.784888888888889e-06,
+      "loss": 1.5986,
+      "step": 57480
+    },
+    {
+      "epoch": 1.83968,
+      "grad_norm": 46.14574432373047,
+      "learning_rate": 1.7813333333333336e-06,
+      "loss": 1.5944,
+      "step": 57490
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 45.9505615234375,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 1.5767,
+      "step": 57500
+    },
+    {
+      "epoch": 1.84032,
+      "grad_norm": 47.18000411987305,
+      "learning_rate": 1.7742222222222225e-06,
+      "loss": 1.5771,
+      "step": 57510
+    },
+    {
+      "epoch": 1.84064,
+      "grad_norm": 47.036415100097656,
+      "learning_rate": 1.7706666666666669e-06,
+      "loss": 1.5874,
+      "step": 57520
+    },
+    {
+      "epoch": 1.84096,
+      "grad_norm": 45.88491439819336,
+      "learning_rate": 1.7671111111111114e-06,
+      "loss": 1.5872,
+      "step": 57530
+    },
+    {
+      "epoch": 1.84128,
+      "grad_norm": 45.427555084228516,
+      "learning_rate": 1.7635555555555555e-06,
+      "loss": 1.5916,
+      "step": 57540
+    },
+    {
+      "epoch": 1.8416000000000001,
+      "grad_norm": 47.513031005859375,
+      "learning_rate": 1.76e-06,
+      "loss": 1.58,
+      "step": 57550
+    },
+    {
+      "epoch": 1.84192,
+      "grad_norm": 46.50860595703125,
+      "learning_rate": 1.7564444444444444e-06,
+      "loss": 1.6057,
+      "step": 57560
+    },
+    {
+      "epoch": 1.8422399999999999,
+      "grad_norm": 44.6576042175293,
+      "learning_rate": 1.752888888888889e-06,
+      "loss": 1.577,
+      "step": 57570
+    },
+    {
+      "epoch": 1.84256,
+      "grad_norm": 45.63453674316406,
+      "learning_rate": 1.7493333333333335e-06,
+      "loss": 1.6053,
+      "step": 57580
+    },
+    {
+      "epoch": 1.84288,
+      "grad_norm": 45.769283294677734,
+      "learning_rate": 1.7457777777777779e-06,
+      "loss": 1.5621,
+      "step": 57590
+    },
+    {
+      "epoch": 1.8432,
+      "grad_norm": 44.64745330810547,
+      "learning_rate": 1.7422222222222224e-06,
+      "loss": 1.595,
+      "step": 57600
+    },
+    {
+      "epoch": 1.84352,
+      "grad_norm": 46.16799545288086,
+      "learning_rate": 1.7386666666666668e-06,
+      "loss": 1.5653,
+      "step": 57610
+    },
+    {
+      "epoch": 1.8438400000000001,
+      "grad_norm": 46.11171340942383,
+      "learning_rate": 1.7351111111111113e-06,
+      "loss": 1.5946,
+      "step": 57620
+    },
+    {
+      "epoch": 1.84416,
+      "grad_norm": 46.63682174682617,
+      "learning_rate": 1.7315555555555557e-06,
+      "loss": 1.622,
+      "step": 57630
+    },
+    {
+      "epoch": 1.84448,
+      "grad_norm": 45.37915802001953,
+      "learning_rate": 1.7280000000000002e-06,
+      "loss": 1.5712,
+      "step": 57640
+    },
+    {
+      "epoch": 1.8448,
+      "grad_norm": 42.86368179321289,
+      "learning_rate": 1.7244444444444448e-06,
+      "loss": 1.5723,
+      "step": 57650
+    },
+    {
+      "epoch": 1.84512,
+      "grad_norm": 45.46616744995117,
+      "learning_rate": 1.7208888888888891e-06,
+      "loss": 1.5876,
+      "step": 57660
+    },
+    {
+      "epoch": 1.84544,
+      "grad_norm": 44.86207962036133,
+      "learning_rate": 1.7173333333333333e-06,
+      "loss": 1.5989,
+      "step": 57670
+    },
+    {
+      "epoch": 1.8457599999999998,
+      "grad_norm": 44.445980072021484,
+      "learning_rate": 1.7137777777777778e-06,
+      "loss": 1.589,
+      "step": 57680
+    },
+    {
+      "epoch": 1.8460800000000002,
+      "grad_norm": 46.44892501831055,
+      "learning_rate": 1.7102222222222224e-06,
+      "loss": 1.5794,
+      "step": 57690
+    },
+    {
+      "epoch": 1.8464,
+      "grad_norm": 45.75631332397461,
+      "learning_rate": 1.7066666666666667e-06,
+      "loss": 1.5837,
+      "step": 57700
+    },
+    {
+      "epoch": 1.84672,
+      "grad_norm": 45.64183044433594,
+      "learning_rate": 1.7031111111111113e-06,
+      "loss": 1.5808,
+      "step": 57710
+    },
+    {
+      "epoch": 1.84704,
+      "grad_norm": 45.0387077331543,
+      "learning_rate": 1.6995555555555556e-06,
+      "loss": 1.6077,
+      "step": 57720
+    },
+    {
+      "epoch": 1.8473600000000001,
+      "grad_norm": 48.08765411376953,
+      "learning_rate": 1.6960000000000002e-06,
+      "loss": 1.5649,
+      "step": 57730
+    },
+    {
+      "epoch": 1.84768,
+      "grad_norm": 47.8968505859375,
+      "learning_rate": 1.6924444444444447e-06,
+      "loss": 1.6016,
+      "step": 57740
+    },
+    {
+      "epoch": 1.8479999999999999,
+      "grad_norm": 43.858253479003906,
+      "learning_rate": 1.688888888888889e-06,
+      "loss": 1.5971,
+      "step": 57750
+    },
+    {
+      "epoch": 1.84832,
+      "grad_norm": 46.54716491699219,
+      "learning_rate": 1.6853333333333336e-06,
+      "loss": 1.6025,
+      "step": 57760
+    },
+    {
+      "epoch": 1.84864,
+      "grad_norm": 46.3195915222168,
+      "learning_rate": 1.681777777777778e-06,
+      "loss": 1.5747,
+      "step": 57770
+    },
+    {
+      "epoch": 1.84896,
+      "grad_norm": 46.96797561645508,
+      "learning_rate": 1.6782222222222225e-06,
+      "loss": 1.5735,
+      "step": 57780
+    },
+    {
+      "epoch": 1.84928,
+      "grad_norm": 46.94729995727539,
+      "learning_rate": 1.6746666666666668e-06,
+      "loss": 1.5827,
+      "step": 57790
+    },
+    {
+      "epoch": 1.8496000000000001,
+      "grad_norm": 45.07993698120117,
+      "learning_rate": 1.6711111111111112e-06,
+      "loss": 1.5997,
+      "step": 57800
+    },
+    {
+      "epoch": 1.84992,
+      "grad_norm": 44.66395950317383,
+      "learning_rate": 1.6675555555555555e-06,
+      "loss": 1.6015,
+      "step": 57810
+    },
+    {
+      "epoch": 1.8502399999999999,
+      "grad_norm": 45.70362091064453,
+      "learning_rate": 1.664e-06,
+      "loss": 1.5976,
+      "step": 57820
+    },
+    {
+      "epoch": 1.85056,
+      "grad_norm": 43.54787063598633,
+      "learning_rate": 1.6604444444444446e-06,
+      "loss": 1.5739,
+      "step": 57830
+    },
+    {
+      "epoch": 1.85088,
+      "grad_norm": 46.17134475708008,
+      "learning_rate": 1.656888888888889e-06,
+      "loss": 1.572,
+      "step": 57840
+    },
+    {
+      "epoch": 1.8512,
+      "grad_norm": 45.06362533569336,
+      "learning_rate": 1.6533333333333335e-06,
+      "loss": 1.5897,
+      "step": 57850
+    },
+    {
+      "epoch": 1.85152,
+      "grad_norm": 45.86150360107422,
+      "learning_rate": 1.6497777777777779e-06,
+      "loss": 1.5891,
+      "step": 57860
+    },
+    {
+      "epoch": 1.8518400000000002,
+      "grad_norm": 46.10201644897461,
+      "learning_rate": 1.6462222222222224e-06,
+      "loss": 1.5785,
+      "step": 57870
+    },
+    {
+      "epoch": 1.85216,
+      "grad_norm": 45.461830139160156,
+      "learning_rate": 1.6426666666666668e-06,
+      "loss": 1.5795,
+      "step": 57880
+    },
+    {
+      "epoch": 1.85248,
+      "grad_norm": 44.85392761230469,
+      "learning_rate": 1.6391111111111113e-06,
+      "loss": 1.619,
+      "step": 57890
+    },
+    {
+      "epoch": 1.8528,
+      "grad_norm": 49.077579498291016,
+      "learning_rate": 1.6355555555555559e-06,
+      "loss": 1.5659,
+      "step": 57900
+    },
+    {
+      "epoch": 1.85312,
+      "grad_norm": 47.31373596191406,
+      "learning_rate": 1.6320000000000002e-06,
+      "loss": 1.5953,
+      "step": 57910
+    },
+    {
+      "epoch": 1.85344,
+      "grad_norm": 45.67685317993164,
+      "learning_rate": 1.6284444444444448e-06,
+      "loss": 1.5822,
+      "step": 57920
+    },
+    {
+      "epoch": 1.8537599999999999,
+      "grad_norm": 45.24131393432617,
+      "learning_rate": 1.624888888888889e-06,
+      "loss": 1.5587,
+      "step": 57930
+    },
+    {
+      "epoch": 1.85408,
+      "grad_norm": 44.38886642456055,
+      "learning_rate": 1.6213333333333335e-06,
+      "loss": 1.5701,
+      "step": 57940
     },
     {
-      "epoch": 0.12864,
-      "grad_norm": 3.4011449813842773,
-      "learning_rate": 9.682122031005548e-06,
-      "loss": 1.9667,
-      "step": 2010
+      "epoch": 1.8544,
+      "grad_norm": 48.03721237182617,
+      "learning_rate": 1.6177777777777778e-06,
+      "loss": 1.5777,
+      "step": 57950
     },
     {
-      "epoch": 0.13056,
-      "grad_norm": 3.3228976726531982,
-      "learning_rate": 9.660787939126725e-06,
-      "loss": 1.9512,
-      "step": 2040
+      "epoch": 1.85472,
+      "grad_norm": 47.05076599121094,
+      "learning_rate": 1.6142222222222223e-06,
+      "loss": 1.569,
+      "step": 57960
     },
     {
-      "epoch": 0.13248,
-      "grad_norm": 3.4176251888275146,
-      "learning_rate": 9.639453847247903e-06,
-      "loss": 1.9678,
-      "step": 2070
+      "epoch": 1.85504,
+      "grad_norm": 45.041385650634766,
+      "learning_rate": 1.6106666666666667e-06,
+      "loss": 1.5948,
+      "step": 57970
     },
     {
-      "epoch": 0.1344,
-      "grad_norm": 3.648175001144409,
-      "learning_rate": 9.61811975536908e-06,
-      "loss": 1.9506,
-      "step": 2100
+      "epoch": 1.8553600000000001,
+      "grad_norm": 46.04569625854492,
+      "learning_rate": 1.6071111111111112e-06,
+      "loss": 1.5724,
+      "step": 57980
     },
     {
-      "epoch": 0.13632,
-      "grad_norm": 3.431642770767212,
-      "learning_rate": 9.596785663490257e-06,
-      "loss": 1.9724,
-      "step": 2130
+      "epoch": 1.85568,
+      "grad_norm": 46.36091995239258,
+      "learning_rate": 1.6035555555555558e-06,
+      "loss": 1.5818,
+      "step": 57990
     },
     {
-      "epoch": 0.13824,
-      "grad_norm": 3.643929958343506,
-      "learning_rate": 9.575451571611436e-06,
-      "loss": 1.9546,
-      "step": 2160
+      "epoch": 1.8559999999999999,
+      "grad_norm": 45.36357116699219,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.5672,
+      "step": 58000
     },
     {
-      "epoch": 0.14016,
-      "grad_norm": 3.302874803543091,
-      "learning_rate": 9.554117479732612e-06,
-      "loss": 1.9537,
-      "step": 2190
+      "epoch": 1.85632,
+      "grad_norm": 45.477821350097656,
+      "learning_rate": 1.5964444444444447e-06,
+      "loss": 1.5666,
+      "step": 58010
     },
     {
-      "epoch": 0.14208,
-      "grad_norm": 3.366065740585327,
-      "learning_rate": 9.532783387853791e-06,
-      "loss": 1.9653,
-      "step": 2220
+      "epoch": 1.85664,
+      "grad_norm": 47.501224517822266,
+      "learning_rate": 1.592888888888889e-06,
+      "loss": 1.5782,
+      "step": 58020
     },
     {
-      "epoch": 0.144,
-      "grad_norm": 3.475569248199463,
-      "learning_rate": 9.511449295974968e-06,
-      "loss": 1.9629,
-      "step": 2250
+      "epoch": 1.85696,
+      "grad_norm": 46.90428161621094,
+      "learning_rate": 1.5893333333333336e-06,
+      "loss": 1.5975,
+      "step": 58030
     },
     {
-      "epoch": 0.14592,
-      "grad_norm": 3.403857946395874,
-      "learning_rate": 9.490115204096146e-06,
-      "loss": 1.9633,
-      "step": 2280
+      "epoch": 1.85728,
+      "grad_norm": 47.66033935546875,
+      "learning_rate": 1.585777777777778e-06,
+      "loss": 1.5813,
+      "step": 58040
     },
     {
-      "epoch": 0.14784,
-      "grad_norm": 3.483151912689209,
-      "learning_rate": 9.468781112217325e-06,
-      "loss": 1.9257,
-      "step": 2310
+      "epoch": 1.8576000000000001,
+      "grad_norm": 48.22922134399414,
+      "learning_rate": 1.5822222222222223e-06,
+      "loss": 1.6164,
+      "step": 58050
     },
     {
-      "epoch": 0.14976,
-      "grad_norm": 3.370187282562256,
-      "learning_rate": 9.447447020338502e-06,
-      "loss": 1.9634,
-      "step": 2340
+      "epoch": 1.85792,
+      "grad_norm": 46.36241149902344,
+      "learning_rate": 1.5786666666666666e-06,
+      "loss": 1.6169,
+      "step": 58060
     },
     {
-      "epoch": 0.15168,
-      "grad_norm": 3.5452382564544678,
-      "learning_rate": 9.42611292845968e-06,
-      "loss": 1.9846,
-      "step": 2370
+      "epoch": 1.85824,
+      "grad_norm": 46.98133850097656,
+      "learning_rate": 1.5751111111111112e-06,
+      "loss": 1.5872,
+      "step": 58070
     },
     {
-      "epoch": 0.1536,
-      "grad_norm": 3.664879322052002,
-      "learning_rate": 9.404778836580857e-06,
-      "loss": 1.9589,
-      "step": 2400
+      "epoch": 1.85856,
+      "grad_norm": 46.50725555419922,
+      "learning_rate": 1.5715555555555555e-06,
+      "loss": 1.5744,
+      "step": 58080
     },
     {
-      "epoch": 0.15552,
-      "grad_norm": 3.553185224533081,
-      "learning_rate": 9.383444744702036e-06,
-      "loss": 1.9754,
-      "step": 2430
+      "epoch": 1.85888,
+      "grad_norm": 46.72871398925781,
+      "learning_rate": 1.568e-06,
+      "loss": 1.5727,
+      "step": 58090
     },
     {
-      "epoch": 0.15744,
-      "grad_norm": 3.3952085971832275,
-      "learning_rate": 9.362110652823212e-06,
-      "loss": 1.964,
-      "step": 2460
+      "epoch": 1.8592,
+      "grad_norm": 44.92698287963867,
+      "learning_rate": 1.5644444444444446e-06,
+      "loss": 1.5775,
+      "step": 58100
     },
     {
-      "epoch": 0.15936,
-      "grad_norm": 3.511155366897583,
-      "learning_rate": 9.34077656094439e-06,
-      "loss": 1.9578,
-      "step": 2490
+      "epoch": 1.8595199999999998,
+      "grad_norm": 44.662654876708984,
+      "learning_rate": 1.560888888888889e-06,
+      "loss": 1.5792,
+      "step": 58110
     },
     {
-      "epoch": 0.16128,
-      "grad_norm": 3.512826919555664,
-      "learning_rate": 9.319442469065568e-06,
-      "loss": 1.9517,
-      "step": 2520
+      "epoch": 1.8598400000000002,
+      "grad_norm": 46.40205764770508,
+      "learning_rate": 1.5573333333333335e-06,
+      "loss": 1.5961,
+      "step": 58120
     },
     {
-      "epoch": 0.1632,
-      "grad_norm": 3.4710845947265625,
-      "learning_rate": 9.298108377186745e-06,
-      "loss": 1.9644,
-      "step": 2550
+      "epoch": 1.86016,
+      "grad_norm": 46.643638610839844,
+      "learning_rate": 1.5537777777777779e-06,
+      "loss": 1.5731,
+      "step": 58130
     },
     {
-      "epoch": 0.16512,
-      "grad_norm": 3.2872207164764404,
-      "learning_rate": 9.276774285307922e-06,
-      "loss": 1.9652,
-      "step": 2580
+      "epoch": 1.86048,
+      "grad_norm": 45.57893753051758,
+      "learning_rate": 1.5502222222222224e-06,
+      "loss": 1.5809,
+      "step": 58140
     },
     {
-      "epoch": 0.16704,
-      "grad_norm": 3.413088083267212,
-      "learning_rate": 9.2554401934291e-06,
-      "loss": 1.9614,
-      "step": 2610
+      "epoch": 1.8608,
+      "grad_norm": 47.35958480834961,
+      "learning_rate": 1.546666666666667e-06,
+      "loss": 1.6098,
+      "step": 58150
     },
     {
-      "epoch": 0.16896,
-      "grad_norm": 3.3675224781036377,
-      "learning_rate": 9.234106101550279e-06,
-      "loss": 1.9335,
-      "step": 2640
+      "epoch": 1.86112,
+      "grad_norm": 46.90447235107422,
+      "learning_rate": 1.5431111111111113e-06,
+      "loss": 1.5867,
+      "step": 58160
     },
     {
-      "epoch": 0.17088,
-      "grad_norm": 3.41202449798584,
-      "learning_rate": 9.212772009671455e-06,
-      "loss": 1.9531,
-      "step": 2670
+      "epoch": 1.86144,
+      "grad_norm": 46.82596206665039,
+      "learning_rate": 1.5395555555555559e-06,
+      "loss": 1.595,
+      "step": 58170
     },
     {
-      "epoch": 0.1728,
-      "grad_norm": 3.3609135150909424,
-      "learning_rate": 9.191437917792634e-06,
-      "loss": 1.9252,
-      "step": 2700
+      "epoch": 1.8617599999999999,
+      "grad_norm": 45.697059631347656,
+      "learning_rate": 1.536e-06,
+      "loss": 1.6127,
+      "step": 58180
     },
     {
-      "epoch": 0.17472,
-      "grad_norm": 3.23222279548645,
-      "learning_rate": 9.17010382591381e-06,
-      "loss": 1.9611,
-      "step": 2730
+      "epoch": 1.86208,
+      "grad_norm": 44.96653747558594,
+      "learning_rate": 1.5324444444444445e-06,
+      "loss": 1.5948,
+      "step": 58190
     },
     {
-      "epoch": 0.17664,
-      "grad_norm": 3.502366304397583,
-      "learning_rate": 9.14876973403499e-06,
-      "loss": 1.9613,
-      "step": 2760
+      "epoch": 1.8624,
+      "grad_norm": 46.40520095825195,
+      "learning_rate": 1.5288888888888889e-06,
+      "loss": 1.5979,
+      "step": 58200
     },
     {
-      "epoch": 0.17856,
-      "grad_norm": 3.340182304382324,
-      "learning_rate": 9.127435642156166e-06,
-      "loss": 1.9646,
-      "step": 2790
+      "epoch": 1.86272,
+      "grad_norm": 46.27067184448242,
+      "learning_rate": 1.5253333333333334e-06,
+      "loss": 1.5443,
+      "step": 58210
     },
     {
-      "epoch": 0.18048,
-      "grad_norm": 3.416520118713379,
-      "learning_rate": 9.106101550277345e-06,
-      "loss": 1.9591,
-      "step": 2820
+      "epoch": 1.86304,
+      "grad_norm": 46.467071533203125,
+      "learning_rate": 1.5217777777777778e-06,
+      "loss": 1.6093,
+      "step": 58220
     },
     {
-      "epoch": 0.1824,
-      "grad_norm": 3.671225070953369,
-      "learning_rate": 9.084767458398522e-06,
-      "loss": 1.9691,
-      "step": 2850
+      "epoch": 1.8633600000000001,
+      "grad_norm": 47.1767463684082,
+      "learning_rate": 1.5182222222222223e-06,
+      "loss": 1.588,
+      "step": 58230
     },
     {
-      "epoch": 0.18432,
-      "grad_norm": 3.367443323135376,
-      "learning_rate": 9.0634333665197e-06,
-      "loss": 1.9667,
-      "step": 2880
+      "epoch": 1.86368,
+      "grad_norm": 46.3292121887207,
+      "learning_rate": 1.5146666666666669e-06,
+      "loss": 1.5683,
+      "step": 58240
     },
     {
-      "epoch": 0.18624,
-      "grad_norm": 3.3736813068389893,
-      "learning_rate": 9.042099274640877e-06,
-      "loss": 1.9577,
-      "step": 2910
+      "epoch": 1.8639999999999999,
+      "grad_norm": 45.56246566772461,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 1.5762,
+      "step": 58250
     },
     {
-      "epoch": 0.18816,
-      "grad_norm": 3.200767993927002,
-      "learning_rate": 9.020765182762055e-06,
-      "loss": 1.9656,
-      "step": 2940
+      "epoch": 1.86432,
+      "grad_norm": 45.426937103271484,
+      "learning_rate": 1.5075555555555558e-06,
+      "loss": 1.5917,
+      "step": 58260
     },
     {
-      "epoch": 0.19008,
-      "grad_norm": 3.3908658027648926,
-      "learning_rate": 8.999431090883232e-06,
-      "loss": 1.9722,
-      "step": 2970
+      "epoch": 1.86464,
+      "grad_norm": 48.11009979248047,
+      "learning_rate": 1.5040000000000001e-06,
+      "loss": 1.5851,
+      "step": 58270
     },
     {
-      "epoch": 0.192,
-      "grad_norm": 3.582303524017334,
-      "learning_rate": 8.978096999004409e-06,
-      "loss": 1.9472,
-      "step": 3000
+      "epoch": 1.86496,
+      "grad_norm": 45.19404220581055,
+      "learning_rate": 1.5004444444444447e-06,
+      "loss": 1.5762,
+      "step": 58280
     },
     {
-      "epoch": 0.19392,
-      "grad_norm": 3.3229641914367676,
-      "learning_rate": 8.956762907125588e-06,
-      "loss": 1.9521,
-      "step": 3030
+      "epoch": 1.86528,
+      "grad_norm": 45.58405303955078,
+      "learning_rate": 1.496888888888889e-06,
+      "loss": 1.5829,
+      "step": 58290
     },
     {
-      "epoch": 0.19584,
-      "grad_norm": 3.4357879161834717,
-      "learning_rate": 8.935428815246764e-06,
-      "loss": 1.9547,
-      "step": 3060
+      "epoch": 1.8656000000000001,
+      "grad_norm": 46.4255485534668,
+      "learning_rate": 1.4933333333333336e-06,
+      "loss": 1.5929,
+      "step": 58300
     },
     {
-      "epoch": 0.19776,
-      "grad_norm": 3.452303886413574,
-      "learning_rate": 8.914094723367943e-06,
-      "loss": 1.98,
-      "step": 3090
+      "epoch": 1.86592,
+      "grad_norm": 49.003501892089844,
+      "learning_rate": 1.4897777777777777e-06,
+      "loss": 1.5872,
+      "step": 58310
     },
     {
-      "epoch": 0.19968,
-      "grad_norm": 3.4541122913360596,
-      "learning_rate": 8.89276063148912e-06,
-      "loss": 1.9553,
-      "step": 3120
+      "epoch": 1.86624,
+      "grad_norm": 45.09196853637695,
+      "learning_rate": 1.4862222222222223e-06,
+      "loss": 1.5773,
+      "step": 58320
     },
     {
-      "epoch": 0.2016,
-      "grad_norm": 3.385713577270508,
-      "learning_rate": 8.871426539610298e-06,
-      "loss": 1.9433,
-      "step": 3150
+      "epoch": 1.86656,
+      "grad_norm": 45.00556945800781,
+      "learning_rate": 1.4826666666666666e-06,
+      "loss": 1.5949,
+      "step": 58330
     },
     {
-      "epoch": 0.20352,
-      "grad_norm": 3.2894997596740723,
-      "learning_rate": 8.850092447731475e-06,
-      "loss": 1.9559,
-      "step": 3180
+      "epoch": 1.86688,
+      "grad_norm": 47.26096725463867,
+      "learning_rate": 1.4791111111111112e-06,
+      "loss": 1.6113,
+      "step": 58340
     },
     {
-      "epoch": 0.20544,
-      "grad_norm": 3.3867385387420654,
-      "learning_rate": 8.828758355852654e-06,
-      "loss": 1.9716,
-      "step": 3210
+      "epoch": 1.8672,
+      "grad_norm": 45.97952651977539,
+      "learning_rate": 1.4755555555555557e-06,
+      "loss": 1.572,
+      "step": 58350
     },
     {
-      "epoch": 0.20736,
-      "grad_norm": 3.348278284072876,
-      "learning_rate": 8.80742426397383e-06,
-      "loss": 1.9412,
-      "step": 3240
+      "epoch": 1.8675199999999998,
+      "grad_norm": 44.738224029541016,
+      "learning_rate": 1.472e-06,
+      "loss": 1.5745,
+      "step": 58360
     },
     {
-      "epoch": 0.20928,
-      "grad_norm": 3.502211809158325,
-      "learning_rate": 8.786090172095009e-06,
-      "loss": 1.9578,
-      "step": 3270
+      "epoch": 1.86784,
+      "grad_norm": 43.97645950317383,
+      "learning_rate": 1.4684444444444446e-06,
+      "loss": 1.6096,
+      "step": 58370
     },
     {
-      "epoch": 0.2112,
-      "grad_norm": 3.339075803756714,
-      "learning_rate": 8.764756080216186e-06,
-      "loss": 1.9591,
-      "step": 3300
+      "epoch": 1.86816,
+      "grad_norm": 47.645835876464844,
+      "learning_rate": 1.464888888888889e-06,
+      "loss": 1.5828,
+      "step": 58380
     },
     {
-      "epoch": 0.21312,
-      "grad_norm": 3.1661159992218018,
-      "learning_rate": 8.743421988337364e-06,
-      "loss": 1.9556,
-      "step": 3330
+      "epoch": 1.86848,
+      "grad_norm": 46.627506256103516,
+      "learning_rate": 1.4613333333333335e-06,
+      "loss": 1.5731,
+      "step": 58390
     },
     {
-      "epoch": 0.21504,
-      "grad_norm": 3.2936370372772217,
-      "learning_rate": 8.722087896458541e-06,
-      "loss": 1.9366,
-      "step": 3360
+      "epoch": 1.8688,
+      "grad_norm": 48.27983474731445,
+      "learning_rate": 1.457777777777778e-06,
+      "loss": 1.5859,
+      "step": 58400
     },
     {
-      "epoch": 0.21696,
-      "grad_norm": 3.3629443645477295,
-      "learning_rate": 8.70075380457972e-06,
-      "loss": 1.9566,
-      "step": 3390
+      "epoch": 1.8691200000000001,
+      "grad_norm": 45.809268951416016,
+      "learning_rate": 1.4542222222222224e-06,
+      "loss": 1.5843,
+      "step": 58410
     },
     {
-      "epoch": 0.21888,
-      "grad_norm": 3.289152145385742,
-      "learning_rate": 8.679419712700897e-06,
-      "loss": 1.9596,
-      "step": 3420
+      "epoch": 1.86944,
+      "grad_norm": 46.93641662597656,
+      "learning_rate": 1.450666666666667e-06,
+      "loss": 1.5581,
+      "step": 58420
     },
     {
-      "epoch": 0.2208,
-      "grad_norm": 3.277226209640503,
-      "learning_rate": 8.658085620822074e-06,
-      "loss": 1.9391,
-      "step": 3450
+      "epoch": 1.8697599999999999,
+      "grad_norm": 45.70205307006836,
+      "learning_rate": 1.4471111111111113e-06,
+      "loss": 1.5773,
+      "step": 58430
     },
     {
-      "epoch": 0.22272,
-      "grad_norm": 3.4808197021484375,
-      "learning_rate": 8.636751528943252e-06,
-      "loss": 1.9443,
-      "step": 3480
+      "epoch": 1.87008,
+      "grad_norm": 47.058006286621094,
+      "learning_rate": 1.4435555555555556e-06,
+      "loss": 1.5623,
+      "step": 58440
     },
     {
-      "epoch": 0.22464,
-      "grad_norm": 3.3526194095611572,
-      "learning_rate": 8.615417437064429e-06,
-      "loss": 1.954,
-      "step": 3510
+      "epoch": 1.8704,
+      "grad_norm": 43.49956512451172,
+      "learning_rate": 1.44e-06,
+      "loss": 1.5601,
+      "step": 58450
     },
     {
-      "epoch": 0.22656,
-      "grad_norm": 3.3914992809295654,
-      "learning_rate": 8.594083345185607e-06,
-      "loss": 1.953,
-      "step": 3540
+      "epoch": 1.87072,
+      "grad_norm": 47.000003814697266,
+      "learning_rate": 1.4364444444444445e-06,
+      "loss": 1.565,
+      "step": 58460
     },
     {
-      "epoch": 0.22848,
-      "grad_norm": 3.3112599849700928,
-      "learning_rate": 8.572749253306784e-06,
-      "loss": 1.9367,
-      "step": 3570
+      "epoch": 1.87104,
+      "grad_norm": 43.984580993652344,
+      "learning_rate": 1.4328888888888889e-06,
+      "loss": 1.5795,
+      "step": 58470
     },
     {
-      "epoch": 0.2304,
-      "grad_norm": 3.4000375270843506,
-      "learning_rate": 8.551415161427963e-06,
-      "loss": 1.9632,
-      "step": 3600
+      "epoch": 1.8713600000000001,
+      "grad_norm": 44.211822509765625,
+      "learning_rate": 1.4293333333333334e-06,
+      "loss": 1.5957,
+      "step": 58480
     },
     {
-      "epoch": 0.23232,
-      "grad_norm": 3.3150362968444824,
-      "learning_rate": 8.53008106954914e-06,
-      "loss": 1.9397,
-      "step": 3630
+      "epoch": 1.87168,
+      "grad_norm": 47.10588073730469,
+      "learning_rate": 1.4257777777777778e-06,
+      "loss": 1.5852,
+      "step": 58490
     },
     {
-      "epoch": 0.23424,
-      "grad_norm": 3.308851718902588,
-      "learning_rate": 8.508746977670318e-06,
-      "loss": 1.9323,
-      "step": 3660
+      "epoch": 1.8719999999999999,
+      "grad_norm": 45.84103775024414,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 1.5888,
+      "step": 58500
     },
     {
-      "epoch": 0.23616,
-      "grad_norm": 5.065930366516113,
-      "learning_rate": 8.487412885791495e-06,
-      "loss": 1.9494,
-      "step": 3690
+      "epoch": 1.87232,
+      "grad_norm": 47.888851165771484,
+      "learning_rate": 1.4186666666666669e-06,
+      "loss": 1.6286,
+      "step": 58510
     },
     {
-      "epoch": 0.23808,
-      "grad_norm": 3.528893232345581,
-      "learning_rate": 8.466078793912674e-06,
-      "loss": 1.9501,
-      "step": 3720
+      "epoch": 1.87264,
+      "grad_norm": 46.03474807739258,
+      "learning_rate": 1.4151111111111112e-06,
+      "loss": 1.6226,
+      "step": 58520
     },
     {
-      "epoch": 0.24,
-      "grad_norm": 3.4435763359069824,
-      "learning_rate": 8.444744702033852e-06,
-      "loss": 1.9529,
-      "step": 3750
+      "epoch": 1.87296,
+      "grad_norm": 45.63921356201172,
+      "learning_rate": 1.4115555555555558e-06,
+      "loss": 1.5904,
+      "step": 58530
     },
     {
-      "epoch": 0.24192,
-      "grad_norm": 3.47472882270813,
-      "learning_rate": 8.423410610155029e-06,
-      "loss": 1.9519,
-      "step": 3780
+      "epoch": 1.8732799999999998,
+      "grad_norm": 46.81883239746094,
+      "learning_rate": 1.4080000000000001e-06,
+      "loss": 1.5945,
+      "step": 58540
     },
     {
-      "epoch": 0.24384,
-      "grad_norm": 3.35807466506958,
-      "learning_rate": 8.402076518276206e-06,
-      "loss": 1.9348,
-      "step": 3810
+      "epoch": 1.8736000000000002,
+      "grad_norm": 45.993690490722656,
+      "learning_rate": 1.4044444444444447e-06,
+      "loss": 1.5804,
+      "step": 58550
     },
     {
-      "epoch": 0.24576,
-      "grad_norm": 3.4333138465881348,
-      "learning_rate": 8.380742426397384e-06,
-      "loss": 1.9423,
-      "step": 3840
+      "epoch": 1.87392,
+      "grad_norm": 47.72322082519531,
+      "learning_rate": 1.4008888888888892e-06,
+      "loss": 1.5495,
+      "step": 58560
     },
     {
-      "epoch": 0.24768,
-      "grad_norm": 3.6836507320404053,
-      "learning_rate": 8.359408334518561e-06,
-      "loss": 1.9574,
-      "step": 3870
+      "epoch": 1.87424,
+      "grad_norm": 46.83625411987305,
+      "learning_rate": 1.3973333333333334e-06,
+      "loss": 1.6094,
+      "step": 58570
     },
     {
-      "epoch": 0.2496,
-      "grad_norm": 3.507800817489624,
-      "learning_rate": 8.338074242639738e-06,
-      "loss": 1.9491,
-      "step": 3900
+      "epoch": 1.87456,
+      "grad_norm": 46.28458786010742,
+      "learning_rate": 1.3937777777777777e-06,
+      "loss": 1.6091,
+      "step": 58580
     },
     {
-      "epoch": 0.25152,
-      "grad_norm": 3.330439329147339,
-      "learning_rate": 8.316740150760916e-06,
-      "loss": 1.9573,
-      "step": 3930
+      "epoch": 1.87488,
+      "grad_norm": 45.972381591796875,
+      "learning_rate": 1.3902222222222222e-06,
+      "loss": 1.6046,
+      "step": 58590
     },
     {
-      "epoch": 0.25344,
-      "grad_norm": 3.437272548675537,
-      "learning_rate": 8.295406058882093e-06,
-      "loss": 1.961,
-      "step": 3960
+      "epoch": 1.8752,
+      "grad_norm": 44.876861572265625,
+      "learning_rate": 1.3866666666666668e-06,
+      "loss": 1.5909,
+      "step": 58600
     },
     {
-      "epoch": 0.25536,
-      "grad_norm": 3.485556125640869,
-      "learning_rate": 8.274071967003272e-06,
-      "loss": 1.9524,
-      "step": 3990
+      "epoch": 1.8755199999999999,
+      "grad_norm": 45.53847122192383,
+      "learning_rate": 1.3831111111111111e-06,
+      "loss": 1.5822,
+      "step": 58610
     },
     {
-      "epoch": 0.25728,
-      "grad_norm": 3.3399152755737305,
-      "learning_rate": 8.252737875124449e-06,
-      "loss": 1.956,
-      "step": 4020
+      "epoch": 1.87584,
+      "grad_norm": 45.47560501098633,
+      "learning_rate": 1.3795555555555557e-06,
+      "loss": 1.5897,
+      "step": 58620
     },
     {
-      "epoch": 0.2592,
-      "grad_norm": 3.38154935836792,
-      "learning_rate": 8.231403783245627e-06,
-      "loss": 1.9537,
-      "step": 4050
+      "epoch": 1.87616,
+      "grad_norm": 46.871673583984375,
+      "learning_rate": 1.376e-06,
+      "loss": 1.6033,
+      "step": 58630
     },
     {
-      "epoch": 0.26112,
-      "grad_norm": 3.303474187850952,
-      "learning_rate": 8.210069691366804e-06,
-      "loss": 1.9293,
-      "step": 4080
+      "epoch": 1.87648,
+      "grad_norm": 45.24031448364258,
+      "learning_rate": 1.3724444444444446e-06,
+      "loss": 1.6099,
+      "step": 58640
     },
     {
-      "epoch": 0.26304,
-      "grad_norm": 3.3951666355133057,
-      "learning_rate": 8.188735599487983e-06,
-      "loss": 1.9331,
-      "step": 4110
+      "epoch": 1.8768,
+      "grad_norm": 44.30140686035156,
+      "learning_rate": 1.3688888888888891e-06,
+      "loss": 1.6058,
+      "step": 58650
     },
     {
-      "epoch": 0.26496,
-      "grad_norm": 3.317218542098999,
-      "learning_rate": 8.167401507609161e-06,
-      "loss": 1.9484,
-      "step": 4140
+      "epoch": 1.8771200000000001,
+      "grad_norm": 47.0979118347168,
+      "learning_rate": 1.3653333333333335e-06,
+      "loss": 1.6106,
+      "step": 58660
     },
     {
-      "epoch": 0.26688,
-      "grad_norm": 3.5020060539245605,
-      "learning_rate": 8.146067415730338e-06,
-      "loss": 1.9371,
-      "step": 4170
+      "epoch": 1.87744,
+      "grad_norm": 50.710105895996094,
+      "learning_rate": 1.361777777777778e-06,
+      "loss": 1.5798,
+      "step": 58670
     },
     {
-      "epoch": 0.2688,
-      "grad_norm": 3.640859365463257,
-      "learning_rate": 8.124733323851516e-06,
-      "loss": 1.93,
-      "step": 4200
+      "epoch": 1.8777599999999999,
+      "grad_norm": 47.16532897949219,
+      "learning_rate": 1.3582222222222224e-06,
+      "loss": 1.5975,
+      "step": 58680
     },
     {
-      "epoch": 0.27072,
-      "grad_norm": 3.3728644847869873,
-      "learning_rate": 8.103399231972693e-06,
-      "loss": 1.9367,
-      "step": 4230
+      "epoch": 1.87808,
+      "grad_norm": 47.154144287109375,
+      "learning_rate": 1.354666666666667e-06,
+      "loss": 1.6065,
+      "step": 58690
     },
     {
-      "epoch": 0.27264,
-      "grad_norm": 3.414140462875366,
-      "learning_rate": 8.082065140093872e-06,
-      "loss": 1.9351,
-      "step": 4260
+      "epoch": 1.8784,
+      "grad_norm": 44.24601745605469,
+      "learning_rate": 1.351111111111111e-06,
+      "loss": 1.5771,
+      "step": 58700
     },
     {
-      "epoch": 0.27456,
-      "grad_norm": 3.5059468746185303,
-      "learning_rate": 8.060731048215049e-06,
-      "loss": 1.9421,
-      "step": 4290
+      "epoch": 1.87872,
+      "grad_norm": 45.991661071777344,
+      "learning_rate": 1.3475555555555556e-06,
+      "loss": 1.5962,
+      "step": 58710
     },
     {
-      "epoch": 0.27648,
-      "grad_norm": 3.249708652496338,
-      "learning_rate": 8.039396956336226e-06,
-      "loss": 1.9451,
-      "step": 4320
+      "epoch": 1.87904,
+      "grad_norm": 45.59969711303711,
+      "learning_rate": 1.344e-06,
+      "loss": 1.6073,
+      "step": 58720
     },
     {
-      "epoch": 0.2784,
-      "grad_norm": 3.4159162044525146,
-      "learning_rate": 8.018062864457404e-06,
-      "loss": 1.9307,
-      "step": 4350
+      "epoch": 1.8793600000000001,
+      "grad_norm": 46.94809341430664,
+      "learning_rate": 1.3404444444444445e-06,
+      "loss": 1.6476,
+      "step": 58730
     },
     {
-      "epoch": 0.28032,
-      "grad_norm": 3.460524082183838,
-      "learning_rate": 7.996728772578581e-06,
-      "loss": 1.9434,
-      "step": 4380
+      "epoch": 1.87968,
+      "grad_norm": 45.00490951538086,
+      "learning_rate": 1.3368888888888889e-06,
+      "loss": 1.5637,
+      "step": 58740
     },
     {
-      "epoch": 0.28224,
-      "grad_norm": 3.423017978668213,
-      "learning_rate": 7.975394680699758e-06,
-      "loss": 1.9429,
-      "step": 4410
+      "epoch": 1.88,
+      "grad_norm": 46.691856384277344,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.5814,
+      "step": 58750
     },
     {
-      "epoch": 0.28416,
-      "grad_norm": 3.277519464492798,
-      "learning_rate": 7.954060588820936e-06,
-      "loss": 1.9591,
-      "step": 4440
+      "epoch": 1.88032,
+      "grad_norm": 45.54383850097656,
+      "learning_rate": 1.329777777777778e-06,
+      "loss": 1.5851,
+      "step": 58760
     },
     {
-      "epoch": 0.28608,
-      "grad_norm": 3.335141181945801,
-      "learning_rate": 7.932726496942113e-06,
-      "loss": 1.9632,
-      "step": 4470
+      "epoch": 1.88064,
+      "grad_norm": 47.79774856567383,
+      "learning_rate": 1.3262222222222223e-06,
+      "loss": 1.582,
+      "step": 58770
     },
     {
-      "epoch": 0.288,
-      "grad_norm": 3.3671059608459473,
-      "learning_rate": 7.911392405063292e-06,
-      "loss": 1.9381,
-      "step": 4500
+      "epoch": 1.88096,
+      "grad_norm": 44.72974395751953,
+      "learning_rate": 1.3226666666666669e-06,
+      "loss": 1.5903,
+      "step": 58780
     },
     {
-      "epoch": 0.28992,
-      "grad_norm": 3.3104846477508545,
-      "learning_rate": 7.89005831318447e-06,
-      "loss": 1.9378,
-      "step": 4530
+      "epoch": 1.8812799999999998,
+      "grad_norm": 46.53852081298828,
+      "learning_rate": 1.3191111111111112e-06,
+      "loss": 1.5552,
+      "step": 58790
     },
     {
-      "epoch": 0.29184,
-      "grad_norm": 3.1727168560028076,
-      "learning_rate": 7.868724221305647e-06,
-      "loss": 1.9533,
-      "step": 4560
+      "epoch": 1.8816000000000002,
+      "grad_norm": 47.67750930786133,
+      "learning_rate": 1.3155555555555558e-06,
+      "loss": 1.5877,
+      "step": 58800
     },
     {
-      "epoch": 0.29376,
-      "grad_norm": 3.4788355827331543,
-      "learning_rate": 7.847390129426826e-06,
-      "loss": 1.9365,
-      "step": 4590
+      "epoch": 1.88192,
+      "grad_norm": 46.062530517578125,
+      "learning_rate": 1.3120000000000003e-06,
+      "loss": 1.6022,
+      "step": 58810
     },
     {
-      "epoch": 0.29568,
-      "grad_norm": 3.195474147796631,
-      "learning_rate": 7.826056037548002e-06,
-      "loss": 1.9378,
-      "step": 4620
+      "epoch": 1.88224,
+      "grad_norm": 44.168548583984375,
+      "learning_rate": 1.3084444444444447e-06,
+      "loss": 1.5655,
+      "step": 58820
     },
     {
-      "epoch": 0.2976,
-      "grad_norm": 3.4869565963745117,
-      "learning_rate": 7.804721945669181e-06,
-      "loss": 1.9373,
-      "step": 4650
+      "epoch": 1.88256,
+      "grad_norm": 47.523563385009766,
+      "learning_rate": 1.3048888888888888e-06,
+      "loss": 1.6147,
+      "step": 58830
     },
     {
-      "epoch": 0.29952,
-      "grad_norm": 3.308915376663208,
-      "learning_rate": 7.783387853790358e-06,
-      "loss": 1.9351,
-      "step": 4680
+      "epoch": 1.88288,
+      "grad_norm": 47.97063446044922,
+      "learning_rate": 1.3013333333333333e-06,
+      "loss": 1.5899,
+      "step": 58840
     },
     {
-      "epoch": 0.30144,
-      "grad_norm": 3.258909225463867,
-      "learning_rate": 7.762053761911536e-06,
-      "loss": 1.9255,
-      "step": 4710
+      "epoch": 1.8832,
+      "grad_norm": 45.672088623046875,
+      "learning_rate": 1.2977777777777779e-06,
+      "loss": 1.5966,
+      "step": 58850
     },
     {
-      "epoch": 0.30336,
-      "grad_norm": 3.321763038635254,
-      "learning_rate": 7.740719670032713e-06,
-      "loss": 1.9355,
-      "step": 4740
+      "epoch": 1.8835199999999999,
+      "grad_norm": 47.125675201416016,
+      "learning_rate": 1.2942222222222222e-06,
+      "loss": 1.5992,
+      "step": 58860
     },
     {
-      "epoch": 0.30528,
-      "grad_norm": 3.3691372871398926,
-      "learning_rate": 7.71938557815389e-06,
-      "loss": 1.9369,
-      "step": 4770
+      "epoch": 1.88384,
+      "grad_norm": 46.30353927612305,
+      "learning_rate": 1.2906666666666668e-06,
+      "loss": 1.5933,
+      "step": 58870
     },
     {
-      "epoch": 0.3072,
-      "grad_norm": 3.3551723957061768,
-      "learning_rate": 7.698051486275068e-06,
-      "loss": 1.9328,
-      "step": 4800
+      "epoch": 1.88416,
+      "grad_norm": 45.49856185913086,
+      "learning_rate": 1.2871111111111111e-06,
+      "loss": 1.5756,
+      "step": 58880
     },
     {
-      "epoch": 0.30912,
-      "grad_norm": 3.4278182983398438,
-      "learning_rate": 7.676717394396245e-06,
-      "loss": 1.9428,
-      "step": 4830
+      "epoch": 1.88448,
+      "grad_norm": 44.48989486694336,
+      "learning_rate": 1.2835555555555557e-06,
+      "loss": 1.5633,
+      "step": 58890
     },
     {
-      "epoch": 0.31104,
-      "grad_norm": 3.3448307514190674,
-      "learning_rate": 7.655383302517424e-06,
-      "loss": 1.9267,
-      "step": 4860
+      "epoch": 1.8848,
+      "grad_norm": 45.93333435058594,
+      "learning_rate": 1.28e-06,
+      "loss": 1.5816,
+      "step": 58900
     },
     {
-      "epoch": 0.31296,
-      "grad_norm": 3.485520124435425,
-      "learning_rate": 7.6340492106386e-06,
-      "loss": 1.9495,
-      "step": 4890
+      "epoch": 1.8851200000000001,
+      "grad_norm": 45.116729736328125,
+      "learning_rate": 1.2764444444444446e-06,
+      "loss": 1.6162,
+      "step": 58910
     },
     {
-      "epoch": 0.31488,
-      "grad_norm": 3.436784029006958,
-      "learning_rate": 7.612715118759779e-06,
-      "loss": 1.9376,
-      "step": 4920
+      "epoch": 1.88544,
+      "grad_norm": 44.85403823852539,
+      "learning_rate": 1.2728888888888891e-06,
+      "loss": 1.5737,
+      "step": 58920
     },
     {
-      "epoch": 0.3168,
-      "grad_norm": 3.1847822666168213,
-      "learning_rate": 7.591381026880956e-06,
-      "loss": 1.945,
-      "step": 4950
+      "epoch": 1.8857599999999999,
+      "grad_norm": 45.997314453125,
+      "learning_rate": 1.2693333333333335e-06,
+      "loss": 1.5953,
+      "step": 58930
     },
     {
-      "epoch": 0.31872,
-      "grad_norm": 3.325716972351074,
-      "learning_rate": 7.5700469350021346e-06,
-      "loss": 1.9426,
-      "step": 4980
+      "epoch": 1.88608,
+      "grad_norm": 45.64460372924805,
+      "learning_rate": 1.265777777777778e-06,
+      "loss": 1.593,
+      "step": 58940
     },
     {
-      "epoch": 0.32064,
-      "grad_norm": 3.433797836303711,
-      "learning_rate": 7.548712843123311e-06,
-      "loss": 1.9476,
-      "step": 5010
+      "epoch": 1.8864,
+      "grad_norm": 51.03400421142578,
+      "learning_rate": 1.2622222222222224e-06,
+      "loss": 1.5836,
+      "step": 58950
     },
     {
-      "epoch": 0.32256,
-      "grad_norm": 3.3621113300323486,
-      "learning_rate": 7.52737875124449e-06,
-      "loss": 1.9447,
-      "step": 5040
+      "epoch": 1.88672,
+      "grad_norm": 45.448829650878906,
+      "learning_rate": 1.2586666666666667e-06,
+      "loss": 1.5865,
+      "step": 58960
     },
     {
-      "epoch": 0.32448,
-      "grad_norm": 3.3403565883636475,
-      "learning_rate": 7.506044659365667e-06,
-      "loss": 1.9425,
-      "step": 5070
+      "epoch": 1.88704,
+      "grad_norm": 43.558738708496094,
+      "learning_rate": 1.255111111111111e-06,
+      "loss": 1.5797,
+      "step": 58970
     },
     {
-      "epoch": 0.3264,
-      "grad_norm": 3.3888230323791504,
-      "learning_rate": 7.4847105674868445e-06,
-      "loss": 1.9296,
-      "step": 5100
+      "epoch": 1.8873600000000001,
+      "grad_norm": 46.4416389465332,
+      "learning_rate": 1.2515555555555556e-06,
+      "loss": 1.6047,
+      "step": 58980
     },
     {
-      "epoch": 0.32832,
-      "grad_norm": 3.263082504272461,
-      "learning_rate": 7.463376475608022e-06,
-      "loss": 1.9231,
-      "step": 5130
+      "epoch": 1.88768,
+      "grad_norm": 43.49354553222656,
+      "learning_rate": 1.248e-06,
+      "loss": 1.5868,
+      "step": 58990
     },
     {
-      "epoch": 0.33024,
-      "grad_norm": 3.2603297233581543,
-      "learning_rate": 7.4420423837292e-06,
-      "loss": 1.9397,
-      "step": 5160
+      "epoch": 1.888,
+      "grad_norm": 46.484886169433594,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 1.6069,
+      "step": 59000
     },
     {
-      "epoch": 0.33216,
-      "grad_norm": 3.3704400062561035,
-      "learning_rate": 7.420708291850377e-06,
-      "loss": 1.9323,
-      "step": 5190
+      "epoch": 1.88832,
+      "grad_norm": 46.54124069213867,
+      "learning_rate": 1.240888888888889e-06,
+      "loss": 1.6042,
+      "step": 59010
     },
     {
-      "epoch": 0.33408,
-      "grad_norm": 3.437042713165283,
-      "learning_rate": 7.399374199971555e-06,
-      "loss": 1.9353,
-      "step": 5220
+      "epoch": 1.88864,
+      "grad_norm": 44.74590301513672,
+      "learning_rate": 1.2373333333333334e-06,
+      "loss": 1.6299,
+      "step": 59020
     },
     {
-      "epoch": 0.336,
-      "grad_norm": 3.3123514652252197,
-      "learning_rate": 7.378040108092733e-06,
-      "loss": 1.9328,
-      "step": 5250
+      "epoch": 1.88896,
+      "grad_norm": 47.56993865966797,
+      "learning_rate": 1.233777777777778e-06,
+      "loss": 1.6019,
+      "step": 59030
     },
     {
-      "epoch": 0.33792,
-      "grad_norm": 3.3057658672332764,
-      "learning_rate": 7.3567060162139106e-06,
-      "loss": 1.9293,
-      "step": 5280
+      "epoch": 1.8892799999999998,
+      "grad_norm": 47.559814453125,
+      "learning_rate": 1.2302222222222223e-06,
+      "loss": 1.5463,
+      "step": 59040
     },
     {
-      "epoch": 0.33984,
-      "grad_norm": 3.2438759803771973,
-      "learning_rate": 7.335371924335088e-06,
-      "loss": 1.9336,
-      "step": 5310
+      "epoch": 1.8896,
+      "grad_norm": 46.389854431152344,
+      "learning_rate": 1.2266666666666666e-06,
+      "loss": 1.5817,
+      "step": 59050
     },
     {
-      "epoch": 0.34176,
-      "grad_norm": 3.431074857711792,
-      "learning_rate": 7.314037832456266e-06,
-      "loss": 1.9257,
-      "step": 5340
+      "epoch": 1.88992,
+      "grad_norm": 46.8560791015625,
+      "learning_rate": 1.2231111111111112e-06,
+      "loss": 1.5512,
+      "step": 59060
     },
     {
-      "epoch": 0.34368,
-      "grad_norm": 3.3365626335144043,
-      "learning_rate": 7.292703740577444e-06,
-      "loss": 1.9446,
-      "step": 5370
+      "epoch": 1.89024,
+      "grad_norm": 48.501808166503906,
+      "learning_rate": 1.2195555555555555e-06,
+      "loss": 1.6029,
+      "step": 59070
     },
     {
-      "epoch": 0.3456,
-      "grad_norm": 3.36806321144104,
-      "learning_rate": 7.2713696486986205e-06,
-      "loss": 1.929,
-      "step": 5400
+      "epoch": 1.89056,
+      "grad_norm": 44.118343353271484,
+      "learning_rate": 1.216e-06,
+      "loss": 1.576,
+      "step": 59080
     },
     {
-      "epoch": 0.34752,
-      "grad_norm": 3.343510389328003,
-      "learning_rate": 7.250035556819799e-06,
-      "loss": 1.9261,
-      "step": 5430
+      "epoch": 1.8908800000000001,
+      "grad_norm": 45.0555305480957,
+      "learning_rate": 1.2124444444444446e-06,
+      "loss": 1.5718,
+      "step": 59090
     },
     {
-      "epoch": 0.34944,
-      "grad_norm": 3.451723337173462,
-      "learning_rate": 7.228701464940976e-06,
-      "loss": 1.9248,
-      "step": 5460
+      "epoch": 1.8912,
+      "grad_norm": 46.08884048461914,
+      "learning_rate": 1.208888888888889e-06,
+      "loss": 1.5912,
+      "step": 59100
     },
     {
-      "epoch": 0.35136,
-      "grad_norm": 3.3111345767974854,
-      "learning_rate": 7.207367373062154e-06,
-      "loss": 1.9325,
-      "step": 5490
+      "epoch": 1.8915199999999999,
+      "grad_norm": 45.68895721435547,
+      "learning_rate": 1.2053333333333335e-06,
+      "loss": 1.5773,
+      "step": 59110
     },
     {
-      "epoch": 0.35328,
-      "grad_norm": 3.3748598098754883,
-      "learning_rate": 7.186033281183331e-06,
-      "loss": 1.9118,
-      "step": 5520
+      "epoch": 1.89184,
+      "grad_norm": 47.57466506958008,
+      "learning_rate": 1.2017777777777779e-06,
+      "loss": 1.5952,
+      "step": 59120
     },
     {
-      "epoch": 0.3552,
-      "grad_norm": 3.364699363708496,
-      "learning_rate": 7.164699189304509e-06,
-      "loss": 1.9296,
-      "step": 5550
+      "epoch": 1.89216,
+      "grad_norm": 46.38992691040039,
+      "learning_rate": 1.1982222222222222e-06,
+      "loss": 1.5856,
+      "step": 59130
     },
     {
-      "epoch": 0.35712,
-      "grad_norm": 3.486098527908325,
-      "learning_rate": 7.1433650974256866e-06,
-      "loss": 1.9294,
-      "step": 5580
+      "epoch": 1.89248,
+      "grad_norm": 45.40290069580078,
+      "learning_rate": 1.1946666666666668e-06,
+      "loss": 1.5807,
+      "step": 59140
     },
     {
-      "epoch": 0.35904,
-      "grad_norm": 3.38917875289917,
-      "learning_rate": 7.122031005546864e-06,
-      "loss": 1.9539,
-      "step": 5610
+      "epoch": 1.8928,
+      "grad_norm": 44.96638107299805,
+      "learning_rate": 1.1911111111111111e-06,
+      "loss": 1.567,
+      "step": 59150
     },
     {
-      "epoch": 0.36096,
-      "grad_norm": 3.5631625652313232,
-      "learning_rate": 7.100696913668043e-06,
-      "loss": 1.9355,
-      "step": 5640
+      "epoch": 1.8931200000000001,
+      "grad_norm": 45.714420318603516,
+      "learning_rate": 1.1875555555555557e-06,
+      "loss": 1.5797,
+      "step": 59160
     },
     {
-      "epoch": 0.36288,
-      "grad_norm": 3.2892417907714844,
-      "learning_rate": 7.07936282178922e-06,
-      "loss": 1.9422,
-      "step": 5670
+      "epoch": 1.89344,
+      "grad_norm": 44.78868865966797,
+      "learning_rate": 1.1840000000000002e-06,
+      "loss": 1.5492,
+      "step": 59170
     },
     {
-      "epoch": 0.3648,
-      "grad_norm": 3.281036853790283,
-      "learning_rate": 7.058028729910398e-06,
-      "loss": 1.9251,
-      "step": 5700
+      "epoch": 1.8937599999999999,
+      "grad_norm": 48.12905502319336,
+      "learning_rate": 1.1804444444444446e-06,
+      "loss": 1.568,
+      "step": 59180
     },
     {
-      "epoch": 0.36672,
-      "grad_norm": 3.466916799545288,
-      "learning_rate": 7.036694638031575e-06,
-      "loss": 1.9129,
-      "step": 5730
+      "epoch": 1.89408,
+      "grad_norm": 46.8428955078125,
+      "learning_rate": 1.176888888888889e-06,
+      "loss": 1.5864,
+      "step": 59190
     },
     {
-      "epoch": 0.36864,
-      "grad_norm": 3.4291369915008545,
-      "learning_rate": 7.015360546152753e-06,
-      "loss": 1.9362,
-      "step": 5760
+      "epoch": 1.8944,
+      "grad_norm": 47.230228424072266,
+      "learning_rate": 1.1733333333333335e-06,
+      "loss": 1.5794,
+      "step": 59200
     },
     {
-      "epoch": 0.37056,
-      "grad_norm": 3.2582342624664307,
-      "learning_rate": 6.99402645427393e-06,
-      "loss": 1.9394,
-      "step": 5790
+      "epoch": 1.89472,
+      "grad_norm": 46.057979583740234,
+      "learning_rate": 1.1697777777777778e-06,
+      "loss": 1.5976,
+      "step": 59210
     },
     {
-      "epoch": 0.37248,
-      "grad_norm": 3.343424081802368,
-      "learning_rate": 6.972692362395108e-06,
-      "loss": 1.9372,
-      "step": 5820
+      "epoch": 1.8950399999999998,
+      "grad_norm": 49.09218215942383,
+      "learning_rate": 1.1662222222222224e-06,
+      "loss": 1.6047,
+      "step": 59220
     },
     {
-      "epoch": 0.3744,
-      "grad_norm": 3.2941505908966064,
-      "learning_rate": 6.951358270516285e-06,
-      "loss": 1.9523,
-      "step": 5850
+      "epoch": 1.8953600000000002,
+      "grad_norm": 45.197288513183594,
+      "learning_rate": 1.1626666666666667e-06,
+      "loss": 1.5651,
+      "step": 59230
     },
     {
-      "epoch": 0.37632,
-      "grad_norm": 3.4718210697174072,
-      "learning_rate": 6.930024178637463e-06,
-      "loss": 1.9276,
-      "step": 5880
+      "epoch": 1.89568,
+      "grad_norm": 46.362613677978516,
+      "learning_rate": 1.1591111111111113e-06,
+      "loss": 1.6009,
+      "step": 59240
     },
     {
-      "epoch": 0.37824,
-      "grad_norm": 3.448021173477173,
-      "learning_rate": 6.90869008675864e-06,
-      "loss": 1.9293,
-      "step": 5910
+      "epoch": 1.896,
+      "grad_norm": 45.984867095947266,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 1.5618,
+      "step": 59250
     },
     {
-      "epoch": 0.38016,
-      "grad_norm": 3.3645715713500977,
-      "learning_rate": 6.887355994879819e-06,
-      "loss": 1.9349,
-      "step": 5940
+      "epoch": 1.89632,
+      "grad_norm": 46.774959564208984,
+      "learning_rate": 1.1520000000000002e-06,
+      "loss": 1.5684,
+      "step": 59260
     },
     {
-      "epoch": 0.38208,
-      "grad_norm": 3.295274496078491,
-      "learning_rate": 6.8660219030009965e-06,
-      "loss": 1.9366,
-      "step": 5970
+      "epoch": 1.89664,
+      "grad_norm": 45.798126220703125,
+      "learning_rate": 1.1484444444444445e-06,
+      "loss": 1.5963,
+      "step": 59270
     },
     {
-      "epoch": 0.384,
-      "grad_norm": 3.1582870483398438,
-      "learning_rate": 6.844687811122174e-06,
-      "loss": 1.9205,
-      "step": 6000
+      "epoch": 1.89696,
+      "grad_norm": 46.795806884765625,
+      "learning_rate": 1.144888888888889e-06,
+      "loss": 1.564,
+      "step": 59280
     },
     {
-      "epoch": 0.38592,
-      "grad_norm": 3.530031442642212,
-      "learning_rate": 6.823353719243352e-06,
-      "loss": 1.9353,
-      "step": 6030
+      "epoch": 1.8972799999999999,
+      "grad_norm": 46.9173469543457,
+      "learning_rate": 1.1413333333333334e-06,
+      "loss": 1.5965,
+      "step": 59290
     },
     {
-      "epoch": 0.38784,
-      "grad_norm": 3.395073890686035,
-      "learning_rate": 6.802019627364529e-06,
-      "loss": 1.927,
-      "step": 6060
+      "epoch": 1.8976,
+      "grad_norm": 45.2878532409668,
+      "learning_rate": 1.137777777777778e-06,
+      "loss": 1.5755,
+      "step": 59300
     },
     {
-      "epoch": 0.38976,
-      "grad_norm": 3.2854127883911133,
-      "learning_rate": 6.780685535485707e-06,
-      "loss": 1.9308,
-      "step": 6090
+      "epoch": 1.89792,
+      "grad_norm": 46.399452209472656,
+      "learning_rate": 1.1342222222222223e-06,
+      "loss": 1.5575,
+      "step": 59310
     },
     {
-      "epoch": 0.39168,
-      "grad_norm": 3.1009058952331543,
-      "learning_rate": 6.759351443606884e-06,
-      "loss": 1.9434,
-      "step": 6120
+      "epoch": 1.89824,
+      "grad_norm": 45.01802062988281,
+      "learning_rate": 1.1306666666666666e-06,
+      "loss": 1.5821,
+      "step": 59320
     },
     {
-      "epoch": 0.3936,
-      "grad_norm": 3.40997052192688,
-      "learning_rate": 6.7380173517280626e-06,
-      "loss": 1.9258,
-      "step": 6150
+      "epoch": 1.89856,
+      "grad_norm": 46.38847351074219,
+      "learning_rate": 1.1271111111111112e-06,
+      "loss": 1.587,
+      "step": 59330
     },
     {
-      "epoch": 0.39552,
-      "grad_norm": 3.2788240909576416,
-      "learning_rate": 6.716683259849239e-06,
-      "loss": 1.9403,
-      "step": 6180
+      "epoch": 1.8988800000000001,
+      "grad_norm": 47.17742156982422,
+      "learning_rate": 1.1235555555555557e-06,
+      "loss": 1.6044,
+      "step": 59340
     },
     {
-      "epoch": 0.39744,
-      "grad_norm": 3.2732865810394287,
-      "learning_rate": 6.695349167970417e-06,
-      "loss": 1.9319,
-      "step": 6210
+      "epoch": 1.8992,
+      "grad_norm": 47.04226303100586,
+      "learning_rate": 1.12e-06,
+      "loss": 1.5855,
+      "step": 59350
     },
     {
-      "epoch": 0.39936,
-      "grad_norm": 3.2851779460906982,
-      "learning_rate": 6.674015076091595e-06,
-      "loss": 1.9252,
-      "step": 6240
+      "epoch": 1.8995199999999999,
+      "grad_norm": 46.88846206665039,
+      "learning_rate": 1.1164444444444446e-06,
+      "loss": 1.5676,
+      "step": 59360
     },
     {
-      "epoch": 0.40128,
-      "grad_norm": 3.380159854888916,
-      "learning_rate": 6.6526809842127725e-06,
-      "loss": 1.9365,
-      "step": 6270
+      "epoch": 1.89984,
+      "grad_norm": 44.450767517089844,
+      "learning_rate": 1.112888888888889e-06,
+      "loss": 1.6056,
+      "step": 59370
     },
     {
-      "epoch": 0.4032,
-      "grad_norm": 3.4840269088745117,
-      "learning_rate": 6.631346892333949e-06,
-      "loss": 1.9312,
-      "step": 6300
+      "epoch": 1.90016,
+      "grad_norm": 43.49285888671875,
+      "learning_rate": 1.1093333333333333e-06,
+      "loss": 1.5647,
+      "step": 59380
     },
     {
-      "epoch": 0.40512,
-      "grad_norm": 3.5580124855041504,
-      "learning_rate": 6.610012800455128e-06,
-      "loss": 1.9311,
-      "step": 6330
+      "epoch": 1.90048,
+      "grad_norm": 55.5472526550293,
+      "learning_rate": 1.1057777777777779e-06,
+      "loss": 1.5936,
+      "step": 59390
     },
     {
-      "epoch": 0.40704,
-      "grad_norm": 3.346583604812622,
-      "learning_rate": 6.588678708576306e-06,
-      "loss": 1.9493,
-      "step": 6360
+      "epoch": 1.9008,
+      "grad_norm": 44.852840423583984,
+      "learning_rate": 1.1022222222222222e-06,
+      "loss": 1.5658,
+      "step": 59400
     },
     {
-      "epoch": 0.40896,
-      "grad_norm": 3.549868583679199,
-      "learning_rate": 6.567344616697483e-06,
-      "loss": 1.9423,
-      "step": 6390
+      "epoch": 1.9011200000000001,
+      "grad_norm": 45.111331939697266,
+      "learning_rate": 1.0986666666666668e-06,
+      "loss": 1.6017,
+      "step": 59410
     },
     {
-      "epoch": 0.41088,
-      "grad_norm": 3.2973897457122803,
-      "learning_rate": 6.546010524818661e-06,
-      "loss": 1.9433,
-      "step": 6420
+      "epoch": 1.90144,
+      "grad_norm": 45.814456939697266,
+      "learning_rate": 1.0951111111111113e-06,
+      "loss": 1.5871,
+      "step": 59420
     },
     {
-      "epoch": 0.4128,
-      "grad_norm": 3.4193708896636963,
-      "learning_rate": 6.5246764329398386e-06,
-      "loss": 1.9399,
-      "step": 6450
+      "epoch": 1.90176,
+      "grad_norm": 46.770057678222656,
+      "learning_rate": 1.0915555555555557e-06,
+      "loss": 1.5695,
+      "step": 59430
     },
     {
-      "epoch": 0.41472,
-      "grad_norm": 3.323042392730713,
-      "learning_rate": 6.503342341061016e-06,
-      "loss": 1.9202,
-      "step": 6480
+      "epoch": 1.90208,
+      "grad_norm": 45.28121566772461,
+      "learning_rate": 1.088e-06,
+      "loss": 1.59,
+      "step": 59440
     },
     {
-      "epoch": 0.41664,
-      "grad_norm": 3.3325157165527344,
-      "learning_rate": 6.482008249182193e-06,
-      "loss": 1.919,
-      "step": 6510
+      "epoch": 1.9024,
+      "grad_norm": 46.66707229614258,
+      "learning_rate": 1.0844444444444446e-06,
+      "loss": 1.588,
+      "step": 59450
     },
     {
-      "epoch": 0.41856,
-      "grad_norm": 3.3677148818969727,
-      "learning_rate": 6.460674157303372e-06,
-      "loss": 1.9451,
-      "step": 6540
+      "epoch": 1.90272,
+      "grad_norm": 45.3797721862793,
+      "learning_rate": 1.080888888888889e-06,
+      "loss": 1.5848,
+      "step": 59460
     },
     {
-      "epoch": 0.42048,
-      "grad_norm": 3.4558961391448975,
-      "learning_rate": 6.4393400654245485e-06,
-      "loss": 1.9372,
-      "step": 6570
+      "epoch": 1.9030399999999998,
+      "grad_norm": 45.462615966796875,
+      "learning_rate": 1.0773333333333335e-06,
+      "loss": 1.5868,
+      "step": 59470
     },
     {
-      "epoch": 0.4224,
-      "grad_norm": 3.3911585807800293,
-      "learning_rate": 6.418005973545727e-06,
-      "loss": 1.9239,
-      "step": 6600
+      "epoch": 1.9033600000000002,
+      "grad_norm": 46.903961181640625,
+      "learning_rate": 1.0737777777777778e-06,
+      "loss": 1.5728,
+      "step": 59480
     },
     {
-      "epoch": 0.42432,
-      "grad_norm": 3.1638882160186768,
-      "learning_rate": 6.396671881666904e-06,
-      "loss": 1.9153,
-      "step": 6630
+      "epoch": 1.90368,
+      "grad_norm": 45.53373336791992,
+      "learning_rate": 1.0702222222222223e-06,
+      "loss": 1.5891,
+      "step": 59490
     },
     {
-      "epoch": 0.42624,
-      "grad_norm": 3.3997435569763184,
-      "learning_rate": 6.3753377897880815e-06,
-      "loss": 1.919,
-      "step": 6660
+      "epoch": 1.904,
+      "grad_norm": 46.55686950683594,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.6067,
+      "step": 59500
     },
     {
-      "epoch": 0.42816,
-      "grad_norm": 3.224567413330078,
-      "learning_rate": 6.354003697909259e-06,
-      "loss": 1.933,
-      "step": 6690
+      "epoch": 1.90432,
+      "grad_norm": 43.63496780395508,
+      "learning_rate": 1.0631111111111112e-06,
+      "loss": 1.5753,
+      "step": 59510
     },
     {
-      "epoch": 0.43008,
-      "grad_norm": 3.4127113819122314,
-      "learning_rate": 6.332669606030437e-06,
-      "loss": 1.9345,
-      "step": 6720
+      "epoch": 1.90464,
+      "grad_norm": 45.74709701538086,
+      "learning_rate": 1.0595555555555556e-06,
+      "loss": 1.5759,
+      "step": 59520
     },
     {
-      "epoch": 0.432,
-      "grad_norm": 3.4132776260375977,
-      "learning_rate": 6.311335514151615e-06,
-      "loss": 1.9302,
-      "step": 6750
+      "epoch": 1.90496,
+      "grad_norm": 46.07301330566406,
+      "learning_rate": 1.0560000000000001e-06,
+      "loss": 1.5863,
+      "step": 59530
     },
     {
-      "epoch": 0.43392,
-      "grad_norm": 3.2404375076293945,
-      "learning_rate": 6.290001422272792e-06,
-      "loss": 1.929,
-      "step": 6780
+      "epoch": 1.9052799999999999,
+      "grad_norm": 46.41236877441406,
+      "learning_rate": 1.0524444444444445e-06,
+      "loss": 1.61,
+      "step": 59540
     },
     {
-      "epoch": 0.43584,
-      "grad_norm": 3.236316442489624,
-      "learning_rate": 6.268667330393971e-06,
-      "loss": 1.9115,
-      "step": 6810
+      "epoch": 1.9056,
+      "grad_norm": 47.11589813232422,
+      "learning_rate": 1.048888888888889e-06,
+      "loss": 1.579,
+      "step": 59550
     },
     {
-      "epoch": 0.43776,
-      "grad_norm": 3.3159878253936768,
-      "learning_rate": 6.247333238515148e-06,
-      "loss": 1.9319,
-      "step": 6840
+      "epoch": 1.90592,
+      "grad_norm": 46.7946662902832,
+      "learning_rate": 1.0453333333333334e-06,
+      "loss": 1.5736,
+      "step": 59560
     },
     {
-      "epoch": 0.43968,
-      "grad_norm": 3.282456874847412,
-      "learning_rate": 6.225999146636325e-06,
-      "loss": 1.9309,
-      "step": 6870
+      "epoch": 1.90624,
+      "grad_norm": 46.487003326416016,
+      "learning_rate": 1.0417777777777777e-06,
+      "loss": 1.6078,
+      "step": 59570
     },
     {
-      "epoch": 0.4416,
-      "grad_norm": 3.4098093509674072,
-      "learning_rate": 6.204665054757503e-06,
-      "loss": 1.9251,
-      "step": 6900
+      "epoch": 1.90656,
+      "grad_norm": 45.817604064941406,
+      "learning_rate": 1.0382222222222223e-06,
+      "loss": 1.6001,
+      "step": 59580
     },
     {
-      "epoch": 0.44352,
-      "grad_norm": 3.403411865234375,
-      "learning_rate": 6.183330962878681e-06,
-      "loss": 1.9364,
-      "step": 6930
+      "epoch": 1.9068800000000001,
+      "grad_norm": 46.2359733581543,
+      "learning_rate": 1.0346666666666668e-06,
+      "loss": 1.5903,
+      "step": 59590
     },
     {
-      "epoch": 0.44544,
-      "grad_norm": 3.1580772399902344,
-      "learning_rate": 6.1619968709998575e-06,
-      "loss": 1.9204,
-      "step": 6960
+      "epoch": 1.9072,
+      "grad_norm": 46.53152084350586,
+      "learning_rate": 1.0311111111111112e-06,
+      "loss": 1.5816,
+      "step": 59600
     },
     {
-      "epoch": 0.44736,
-      "grad_norm": 3.356703519821167,
-      "learning_rate": 6.140662779121036e-06,
-      "loss": 1.9276,
-      "step": 6990
+      "epoch": 1.9075199999999999,
+      "grad_norm": 45.750003814697266,
+      "learning_rate": 1.0275555555555557e-06,
+      "loss": 1.5715,
+      "step": 59610
     },
     {
-      "epoch": 0.44928,
-      "grad_norm": 3.3060195446014404,
-      "learning_rate": 6.119328687242213e-06,
-      "loss": 1.9266,
-      "step": 7020
+      "epoch": 1.90784,
+      "grad_norm": 45.677181243896484,
+      "learning_rate": 1.024e-06,
+      "loss": 1.5714,
+      "step": 59620
     },
     {
-      "epoch": 0.4512,
-      "grad_norm": 3.3636345863342285,
-      "learning_rate": 6.097994595363391e-06,
-      "loss": 1.9288,
-      "step": 7050
+      "epoch": 1.90816,
+      "grad_norm": 45.89830780029297,
+      "learning_rate": 1.0204444444444444e-06,
+      "loss": 1.5841,
+      "step": 59630
     },
     {
-      "epoch": 0.45312,
-      "grad_norm": 3.2763824462890625,
-      "learning_rate": 6.076660503484569e-06,
-      "loss": 1.9202,
-      "step": 7080
+      "epoch": 1.90848,
+      "grad_norm": 44.421539306640625,
+      "learning_rate": 1.016888888888889e-06,
+      "loss": 1.5739,
+      "step": 59640
     },
     {
-      "epoch": 0.45504,
-      "grad_norm": 3.283390760421753,
-      "learning_rate": 6.055326411605747e-06,
-      "loss": 1.9298,
-      "step": 7110
+      "epoch": 1.9088,
+      "grad_norm": 44.68858337402344,
+      "learning_rate": 1.0133333333333333e-06,
+      "loss": 1.5422,
+      "step": 59650
     },
     {
-      "epoch": 0.45696,
-      "grad_norm": 3.4001858234405518,
-      "learning_rate": 6.0339923197269245e-06,
-      "loss": 1.9273,
-      "step": 7140
+      "epoch": 1.9091200000000002,
+      "grad_norm": 45.24201202392578,
+      "learning_rate": 1.0097777777777779e-06,
+      "loss": 1.5637,
+      "step": 59660
     },
     {
-      "epoch": 0.45888,
-      "grad_norm": 3.2864480018615723,
-      "learning_rate": 6.012658227848101e-06,
-      "loss": 1.9285,
-      "step": 7170
+      "epoch": 1.90944,
+      "grad_norm": 45.254791259765625,
+      "learning_rate": 1.0062222222222224e-06,
+      "loss": 1.5583,
+      "step": 59670
     },
     {
-      "epoch": 0.4608,
-      "grad_norm": 3.623194694519043,
-      "learning_rate": 5.99132413596928e-06,
-      "loss": 1.9335,
-      "step": 7200
+      "epoch": 1.90976,
+      "grad_norm": 46.583251953125,
+      "learning_rate": 1.0026666666666668e-06,
+      "loss": 1.5624,
+      "step": 59680
     },
     {
-      "epoch": 0.46272,
-      "grad_norm": 3.319883108139038,
-      "learning_rate": 5.969990044090457e-06,
-      "loss": 1.9377,
-      "step": 7230
+      "epoch": 1.91008,
+      "grad_norm": 47.10293960571289,
+      "learning_rate": 9.991111111111113e-07,
+      "loss": 1.6102,
+      "step": 59690
     },
     {
-      "epoch": 0.46464,
-      "grad_norm": 3.5105578899383545,
-      "learning_rate": 5.948655952211635e-06,
-      "loss": 1.9082,
-      "step": 7260
+      "epoch": 1.9104,
+      "grad_norm": 47.41055679321289,
+      "learning_rate": 9.955555555555556e-07,
+      "loss": 1.5775,
+      "step": 59700
     },
     {
-      "epoch": 0.46656,
-      "grad_norm": 3.364349126815796,
-      "learning_rate": 5.927321860332812e-06,
-      "loss": 1.9163,
-      "step": 7290
+      "epoch": 1.91072,
+      "grad_norm": 47.239410400390625,
+      "learning_rate": 9.92e-07,
+      "loss": 1.6053,
+      "step": 59710
     },
     {
-      "epoch": 0.46848,
-      "grad_norm": 3.3421549797058105,
-      "learning_rate": 5.90598776845399e-06,
-      "loss": 1.904,
-      "step": 7320
+      "epoch": 1.9110399999999998,
+      "grad_norm": 45.12579345703125,
+      "learning_rate": 9.884444444444445e-07,
+      "loss": 1.5884,
+      "step": 59720
     },
     {
-      "epoch": 0.4704,
-      "grad_norm": 3.2868313789367676,
-      "learning_rate": 5.884653676575167e-06,
-      "loss": 1.9238,
-      "step": 7350
+      "epoch": 1.91136,
+      "grad_norm": 43.950286865234375,
+      "learning_rate": 9.848888888888889e-07,
+      "loss": 1.5695,
+      "step": 59730
     },
     {
-      "epoch": 0.47232,
-      "grad_norm": 3.396576404571533,
-      "learning_rate": 5.863319584696345e-06,
-      "loss": 1.9338,
-      "step": 7380
+      "epoch": 1.91168,
+      "grad_norm": 45.75297927856445,
+      "learning_rate": 9.813333333333334e-07,
+      "loss": 1.5729,
+      "step": 59740
     },
     {
-      "epoch": 0.47424,
-      "grad_norm": 3.4226326942443848,
-      "learning_rate": 5.841985492817523e-06,
-      "loss": 1.9122,
-      "step": 7410
+      "epoch": 1.912,
+      "grad_norm": 46.05004119873047,
+      "learning_rate": 9.77777777777778e-07,
+      "loss": 1.5979,
+      "step": 59750
     },
     {
-      "epoch": 0.47616,
-      "grad_norm": 3.475670576095581,
-      "learning_rate": 5.8206514009387005e-06,
-      "loss": 1.9147,
-      "step": 7440
+      "epoch": 1.91232,
+      "grad_norm": 45.577293395996094,
+      "learning_rate": 9.742222222222223e-07,
+      "loss": 1.5887,
+      "step": 59760
     },
     {
-      "epoch": 0.47808,
-      "grad_norm": 3.4537410736083984,
-      "learning_rate": 5.799317309059879e-06,
-      "loss": 1.9266,
-      "step": 7470
+      "epoch": 1.9126400000000001,
+      "grad_norm": 47.2728157043457,
+      "learning_rate": 9.706666666666667e-07,
+      "loss": 1.5899,
+      "step": 59770
     },
     {
-      "epoch": 0.48,
-      "grad_norm": 3.360386610031128,
-      "learning_rate": 5.777983217181056e-06,
-      "loss": 1.929,
-      "step": 7500
+      "epoch": 1.91296,
+      "grad_norm": 46.47046661376953,
+      "learning_rate": 9.671111111111112e-07,
+      "loss": 1.5795,
+      "step": 59780
     },
     {
-      "epoch": 0.48192,
-      "grad_norm": 3.339397668838501,
-      "learning_rate": 5.7566491253022335e-06,
-      "loss": 1.9159,
-      "step": 7530
+      "epoch": 1.9132799999999999,
+      "grad_norm": 47.05642318725586,
+      "learning_rate": 9.635555555555556e-07,
+      "loss": 1.5726,
+      "step": 59790
     },
     {
-      "epoch": 0.48384,
-      "grad_norm": 3.3917977809906006,
-      "learning_rate": 5.735315033423411e-06,
-      "loss": 1.901,
-      "step": 7560
+      "epoch": 1.9136,
+      "grad_norm": 55.471946716308594,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.5624,
+      "step": 59800
     },
     {
-      "epoch": 0.48576,
-      "grad_norm": 3.376100540161133,
-      "learning_rate": 5.713980941544589e-06,
-      "loss": 1.9158,
-      "step": 7590
+      "epoch": 1.91392,
+      "grad_norm": 46.46736145019531,
+      "learning_rate": 9.564444444444445e-07,
+      "loss": 1.567,
+      "step": 59810
     },
     {
-      "epoch": 0.48768,
-      "grad_norm": 3.324289560317993,
-      "learning_rate": 5.692646849665766e-06,
-      "loss": 1.9322,
-      "step": 7620
+      "epoch": 1.91424,
+      "grad_norm": 45.56180953979492,
+      "learning_rate": 9.52888888888889e-07,
+      "loss": 1.604,
+      "step": 59820
     },
     {
-      "epoch": 0.4896,
-      "grad_norm": 3.2312843799591064,
-      "learning_rate": 5.671312757786944e-06,
-      "loss": 1.927,
-      "step": 7650
+      "epoch": 1.91456,
+      "grad_norm": 46.7037353515625,
+      "learning_rate": 9.493333333333334e-07,
+      "loss": 1.5803,
+      "step": 59830
     },
     {
-      "epoch": 0.49152,
-      "grad_norm": 3.2674617767333984,
-      "learning_rate": 5.649978665908121e-06,
-      "loss": 1.8982,
-      "step": 7680
+      "epoch": 1.9148800000000001,
+      "grad_norm": 43.98044204711914,
+      "learning_rate": 9.457777777777778e-07,
+      "loss": 1.5684,
+      "step": 59840
     },
     {
-      "epoch": 0.49344,
-      "grad_norm": 3.2147467136383057,
-      "learning_rate": 5.6286445740293e-06,
-      "loss": 1.9119,
-      "step": 7710
+      "epoch": 1.9152,
+      "grad_norm": 46.93412780761719,
+      "learning_rate": 9.422222222222223e-07,
+      "loss": 1.5731,
+      "step": 59850
     },
     {
-      "epoch": 0.49536,
-      "grad_norm": 3.28013277053833,
-      "learning_rate": 5.6073104821504765e-06,
-      "loss": 1.912,
-      "step": 7740
+      "epoch": 1.91552,
+      "grad_norm": 48.32170486450195,
+      "learning_rate": 9.386666666666667e-07,
+      "loss": 1.6005,
+      "step": 59860
     },
     {
-      "epoch": 0.49728,
-      "grad_norm": 3.373405694961548,
-      "learning_rate": 5.585976390271655e-06,
-      "loss": 1.9026,
-      "step": 7770
+      "epoch": 1.91584,
+      "grad_norm": 48.01280975341797,
+      "learning_rate": 9.351111111111113e-07,
+      "loss": 1.565,
+      "step": 59870
     },
     {
-      "epoch": 0.4992,
-      "grad_norm": 3.404052972793579,
-      "learning_rate": 5.564642298392832e-06,
-      "loss": 1.9253,
-      "step": 7800
+      "epoch": 1.91616,
+      "grad_norm": 45.45317077636719,
+      "learning_rate": 9.315555555555557e-07,
+      "loss": 1.5921,
+      "step": 59880
     },
     {
-      "epoch": 0.50112,
-      "grad_norm": 3.240499973297119,
-      "learning_rate": 5.5433082065140095e-06,
-      "loss": 1.9104,
-      "step": 7830
+      "epoch": 1.91648,
+      "grad_norm": 45.727210998535156,
+      "learning_rate": 9.28e-07,
+      "loss": 1.5935,
+      "step": 59890
     },
     {
-      "epoch": 0.50304,
-      "grad_norm": 3.4477837085723877,
-      "learning_rate": 5.521974114635188e-06,
-      "loss": 1.9418,
-      "step": 7860
+      "epoch": 1.9167999999999998,
+      "grad_norm": 46.390628814697266,
+      "learning_rate": 9.244444444444445e-07,
+      "loss": 1.5885,
+      "step": 59900
     },
     {
-      "epoch": 0.50496,
-      "grad_norm": 3.3998944759368896,
-      "learning_rate": 5.500640022756365e-06,
-      "loss": 1.9135,
-      "step": 7890
+      "epoch": 1.9171200000000002,
+      "grad_norm": 47.422386169433594,
+      "learning_rate": 9.208888888888889e-07,
+      "loss": 1.5562,
+      "step": 59910
     },
     {
-      "epoch": 0.50688,
-      "grad_norm": 3.579172134399414,
-      "learning_rate": 5.479305930877543e-06,
-      "loss": 1.9287,
-      "step": 7920
+      "epoch": 1.91744,
+      "grad_norm": 44.31741714477539,
+      "learning_rate": 9.173333333333334e-07,
+      "loss": 1.5926,
+      "step": 59920
     },
     {
-      "epoch": 0.5088,
-      "grad_norm": 3.487696886062622,
-      "learning_rate": 5.45797183899872e-06,
-      "loss": 1.9312,
-      "step": 7950
+      "epoch": 1.91776,
+      "grad_norm": 48.28932571411133,
+      "learning_rate": 9.137777777777778e-07,
+      "loss": 1.5839,
+      "step": 59930
     },
     {
-      "epoch": 0.51072,
-      "grad_norm": 3.320296287536621,
-      "learning_rate": 5.436637747119898e-06,
-      "loss": 1.9176,
-      "step": 7980
+      "epoch": 1.91808,
+      "grad_norm": 46.55033493041992,
+      "learning_rate": 9.102222222222223e-07,
+      "loss": 1.5974,
+      "step": 59940
     },
     {
-      "epoch": 0.51264,
-      "grad_norm": 3.2531321048736572,
-      "learning_rate": 5.415303655241076e-06,
-      "loss": 1.9157,
-      "step": 8010
+      "epoch": 1.9184,
+      "grad_norm": 48.564693450927734,
+      "learning_rate": 9.066666666666668e-07,
+      "loss": 1.5815,
+      "step": 59950
     },
     {
-      "epoch": 0.51456,
-      "grad_norm": 3.4431114196777344,
-      "learning_rate": 5.393969563362253e-06,
-      "loss": 1.932,
-      "step": 8040
+      "epoch": 1.91872,
+      "grad_norm": 47.17411804199219,
+      "learning_rate": 9.031111111111111e-07,
+      "loss": 1.5669,
+      "step": 59960
     },
     {
-      "epoch": 0.51648,
-      "grad_norm": 3.403808832168579,
-      "learning_rate": 5.372635471483431e-06,
-      "loss": 1.9265,
-      "step": 8070
+      "epoch": 1.9190399999999999,
+      "grad_norm": 45.52403259277344,
+      "learning_rate": 8.995555555555556e-07,
+      "loss": 1.5781,
+      "step": 59970
     },
     {
-      "epoch": 0.5184,
-      "grad_norm": 3.2141788005828857,
-      "learning_rate": 5.351301379604609e-06,
-      "loss": 1.922,
-      "step": 8100
+      "epoch": 1.91936,
+      "grad_norm": 47.34508514404297,
+      "learning_rate": 8.960000000000001e-07,
+      "loss": 1.5662,
+      "step": 59980
     },
     {
-      "epoch": 0.52032,
-      "grad_norm": 3.3635435104370117,
-      "learning_rate": 5.3299672877257855e-06,
-      "loss": 1.9284,
-      "step": 8130
+      "epoch": 1.91968,
+      "grad_norm": 45.80388641357422,
+      "learning_rate": 8.924444444444445e-07,
+      "loss": 1.561,
+      "step": 59990
     },
     {
-      "epoch": 0.52224,
-      "grad_norm": 3.2985947132110596,
-      "learning_rate": 5.308633195846964e-06,
-      "loss": 1.9125,
-      "step": 8160
+      "epoch": 1.92,
+      "grad_norm": 43.46025848388672,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 1.5916,
+      "step": 60000
     },
     {
-      "epoch": 0.52416,
-      "grad_norm": 3.318425416946411,
-      "learning_rate": 5.287299103968142e-06,
-      "loss": 1.9239,
-      "step": 8190
+      "epoch": 1.92032,
+      "grad_norm": 47.30317306518555,
+      "learning_rate": 8.853333333333334e-07,
+      "loss": 1.5744,
+      "step": 60010
     },
     {
-      "epoch": 0.52608,
-      "grad_norm": 3.254631280899048,
-      "learning_rate": 5.265965012089319e-06,
-      "loss": 1.9053,
-      "step": 8220
+      "epoch": 1.9206400000000001,
+      "grad_norm": 47.16181945800781,
+      "learning_rate": 8.817777777777778e-07,
+      "loss": 1.6053,
+      "step": 60020
     },
     {
-      "epoch": 0.528,
-      "grad_norm": 3.3900418281555176,
-      "learning_rate": 5.244630920210497e-06,
-      "loss": 1.9195,
-      "step": 8250
+      "epoch": 1.92096,
+      "grad_norm": 46.15432357788086,
+      "learning_rate": 8.782222222222222e-07,
+      "loss": 1.5566,
+      "step": 60030
     },
     {
-      "epoch": 0.52992,
-      "grad_norm": 3.3403663635253906,
-      "learning_rate": 5.223296828331674e-06,
-      "loss": 1.9193,
-      "step": 8280
+      "epoch": 1.9212799999999999,
+      "grad_norm": 46.11896896362305,
+      "learning_rate": 8.746666666666668e-07,
+      "loss": 1.589,
+      "step": 60040
     },
     {
-      "epoch": 0.53184,
-      "grad_norm": 3.3622353076934814,
-      "learning_rate": 5.2019627364528525e-06,
-      "loss": 1.9065,
-      "step": 8310
+      "epoch": 1.9216,
+      "grad_norm": 46.01621627807617,
+      "learning_rate": 8.711111111111112e-07,
+      "loss": 1.5917,
+      "step": 60050
     },
     {
-      "epoch": 0.53376,
-      "grad_norm": 3.498197555541992,
-      "learning_rate": 5.180628644574029e-06,
-      "loss": 1.926,
-      "step": 8340
+      "epoch": 1.92192,
+      "grad_norm": 46.327884674072266,
+      "learning_rate": 8.675555555555557e-07,
+      "loss": 1.6255,
+      "step": 60060
     },
     {
-      "epoch": 0.53568,
-      "grad_norm": 3.394606590270996,
-      "learning_rate": 5.159294552695208e-06,
-      "loss": 1.933,
-      "step": 8370
+      "epoch": 1.92224,
+      "grad_norm": 48.15599060058594,
+      "learning_rate": 8.640000000000001e-07,
+      "loss": 1.6075,
+      "step": 60070
     },
     {
-      "epoch": 0.5376,
-      "grad_norm": 3.237445116043091,
-      "learning_rate": 5.137960460816385e-06,
-      "loss": 1.9311,
-      "step": 8400
+      "epoch": 1.92256,
+      "grad_norm": 45.618263244628906,
+      "learning_rate": 8.604444444444446e-07,
+      "loss": 1.5768,
+      "step": 60080
     },
     {
-      "epoch": 0.53952,
-      "grad_norm": 3.6076912879943848,
-      "learning_rate": 5.116626368937563e-06,
-      "loss": 1.9071,
-      "step": 8430
+      "epoch": 1.9228800000000001,
+      "grad_norm": 49.41157531738281,
+      "learning_rate": 8.568888888888889e-07,
+      "loss": 1.5467,
+      "step": 60090
     },
     {
-      "epoch": 0.54144,
-      "grad_norm": 3.30111026763916,
-      "learning_rate": 5.09529227705874e-06,
-      "loss": 1.9301,
-      "step": 8460
+      "epoch": 1.9232,
+      "grad_norm": 46.044471740722656,
+      "learning_rate": 8.533333333333334e-07,
+      "loss": 1.5702,
+      "step": 60100
     },
     {
-      "epoch": 0.54336,
-      "grad_norm": 3.3714241981506348,
-      "learning_rate": 5.073958185179918e-06,
-      "loss": 1.9179,
-      "step": 8490
+      "epoch": 1.92352,
+      "grad_norm": 45.666664123535156,
+      "learning_rate": 8.497777777777778e-07,
+      "loss": 1.5838,
+      "step": 60110
     },
     {
-      "epoch": 0.54528,
-      "grad_norm": 3.257412910461426,
-      "learning_rate": 5.052624093301095e-06,
-      "loss": 1.9202,
-      "step": 8520
+      "epoch": 1.92384,
+      "grad_norm": 46.842872619628906,
+      "learning_rate": 8.462222222222224e-07,
+      "loss": 1.5762,
+      "step": 60120
     },
     {
-      "epoch": 0.5472,
-      "grad_norm": 3.565938949584961,
-      "learning_rate": 5.031290001422273e-06,
-      "loss": 1.9059,
-      "step": 8550
+      "epoch": 1.92416,
+      "grad_norm": 44.966983795166016,
+      "learning_rate": 8.426666666666668e-07,
+      "loss": 1.598,
+      "step": 60130
     },
     {
-      "epoch": 0.54912,
-      "grad_norm": 3.2780563831329346,
-      "learning_rate": 5.009955909543452e-06,
-      "loss": 1.9171,
-      "step": 8580
+      "epoch": 1.92448,
+      "grad_norm": 45.7222785949707,
+      "learning_rate": 8.391111111111112e-07,
+      "loss": 1.5715,
+      "step": 60140
     },
     {
-      "epoch": 0.55104,
-      "grad_norm": 3.514734983444214,
-      "learning_rate": 4.9886218176646285e-06,
-      "loss": 1.9073,
-      "step": 8610
+      "epoch": 1.9247999999999998,
+      "grad_norm": 46.058494567871094,
+      "learning_rate": 8.355555555555556e-07,
+      "loss": 1.5818,
+      "step": 60150
     },
     {
-      "epoch": 0.55296,
-      "grad_norm": 3.3951613903045654,
-      "learning_rate": 4.967287725785806e-06,
-      "loss": 1.9065,
-      "step": 8640
+      "epoch": 1.9251200000000002,
+      "grad_norm": 45.96712112426758,
+      "learning_rate": 8.32e-07,
+      "loss": 1.5925,
+      "step": 60160
     },
     {
-      "epoch": 0.55488,
-      "grad_norm": 3.360833168029785,
-      "learning_rate": 4.945953633906984e-06,
-      "loss": 1.9116,
-      "step": 8670
+      "epoch": 1.92544,
+      "grad_norm": 46.22428512573242,
+      "learning_rate": 8.284444444444445e-07,
+      "loss": 1.5582,
+      "step": 60170
     },
     {
-      "epoch": 0.5568,
-      "grad_norm": 3.3218815326690674,
-      "learning_rate": 4.9246195420281615e-06,
-      "loss": 1.9025,
-      "step": 8700
+      "epoch": 1.92576,
+      "grad_norm": 47.24472427368164,
+      "learning_rate": 8.248888888888889e-07,
+      "loss": 1.588,
+      "step": 60180
     },
     {
-      "epoch": 0.55872,
-      "grad_norm": 3.248537063598633,
-      "learning_rate": 4.903285450149339e-06,
-      "loss": 1.9255,
-      "step": 8730
+      "epoch": 1.92608,
+      "grad_norm": 46.91118240356445,
+      "learning_rate": 8.213333333333334e-07,
+      "loss": 1.5993,
+      "step": 60190
     },
     {
-      "epoch": 0.56064,
-      "grad_norm": 3.369391679763794,
-      "learning_rate": 4.881951358270517e-06,
-      "loss": 1.9248,
-      "step": 8760
+      "epoch": 1.9264000000000001,
+      "grad_norm": 47.56013488769531,
+      "learning_rate": 8.177777777777779e-07,
+      "loss": 1.5783,
+      "step": 60200
     },
     {
-      "epoch": 0.56256,
-      "grad_norm": 3.3814923763275146,
-      "learning_rate": 4.8606172663916946e-06,
-      "loss": 1.9282,
-      "step": 8790
+      "epoch": 1.92672,
+      "grad_norm": 45.22146987915039,
+      "learning_rate": 8.142222222222224e-07,
+      "loss": 1.5904,
+      "step": 60210
+    },
+    {
+      "epoch": 1.9270399999999999,
+      "grad_norm": 45.85641860961914,
+      "learning_rate": 8.106666666666667e-07,
+      "loss": 1.5745,
+      "step": 60220
     },
     {
-      "epoch": 0.56448,
-      "grad_norm": 3.248319625854492,
-      "learning_rate": 4.839283174512872e-06,
-      "loss": 1.9227,
-      "step": 8820
+      "epoch": 1.92736,
+      "grad_norm": 43.82615280151367,
+      "learning_rate": 8.071111111111112e-07,
+      "loss": 1.5985,
+      "step": 60230
     },
     {
-      "epoch": 0.5664,
-      "grad_norm": 3.400087594985962,
-      "learning_rate": 4.81794908263405e-06,
-      "loss": 1.9261,
-      "step": 8850
+      "epoch": 1.92768,
+      "grad_norm": 46.80747604370117,
+      "learning_rate": 8.035555555555556e-07,
+      "loss": 1.6022,
+      "step": 60240
     },
     {
-      "epoch": 0.56832,
-      "grad_norm": 3.3626043796539307,
-      "learning_rate": 4.796614990755228e-06,
-      "loss": 1.9166,
-      "step": 8880
+      "epoch": 1.928,
+      "grad_norm": 45.95657730102539,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.5953,
+      "step": 60250
     },
     {
-      "epoch": 0.57024,
-      "grad_norm": 3.4026668071746826,
-      "learning_rate": 4.775280898876405e-06,
-      "loss": 1.9189,
-      "step": 8910
+      "epoch": 1.92832,
+      "grad_norm": 46.12740707397461,
+      "learning_rate": 7.964444444444445e-07,
+      "loss": 1.5758,
+      "step": 60260
     },
     {
-      "epoch": 0.57216,
-      "grad_norm": 3.4355111122131348,
-      "learning_rate": 4.753946806997582e-06,
-      "loss": 1.9323,
-      "step": 8940
+      "epoch": 1.9286400000000001,
+      "grad_norm": 45.15321731567383,
+      "learning_rate": 7.92888888888889e-07,
+      "loss": 1.5914,
+      "step": 60270
     },
     {
-      "epoch": 0.57408,
-      "grad_norm": 3.3372621536254883,
-      "learning_rate": 4.73261271511876e-06,
-      "loss": 1.9028,
-      "step": 8970
+      "epoch": 1.92896,
+      "grad_norm": 46.08040237426758,
+      "learning_rate": 7.893333333333333e-07,
+      "loss": 1.6012,
+      "step": 60280
     },
     {
-      "epoch": 0.576,
-      "grad_norm": 3.334395408630371,
-      "learning_rate": 4.7112786232399375e-06,
-      "loss": 1.9203,
-      "step": 9000
+      "epoch": 1.9292799999999999,
+      "grad_norm": 44.98908996582031,
+      "learning_rate": 7.857777777777778e-07,
+      "loss": 1.5835,
+      "step": 60290
     },
     {
-      "epoch": 0.57792,
-      "grad_norm": 3.4541335105895996,
-      "learning_rate": 4.689944531361115e-06,
-      "loss": 1.8992,
-      "step": 9030
+      "epoch": 1.9296,
+      "grad_norm": 45.012203216552734,
+      "learning_rate": 7.822222222222223e-07,
+      "loss": 1.5841,
+      "step": 60300
     },
     {
-      "epoch": 0.57984,
-      "grad_norm": 3.3860585689544678,
-      "learning_rate": 4.668610439482293e-06,
-      "loss": 1.9173,
-      "step": 9060
+      "epoch": 1.92992,
+      "grad_norm": 44.48496627807617,
+      "learning_rate": 7.786666666666668e-07,
+      "loss": 1.5726,
+      "step": 60310
     },
     {
-      "epoch": 0.58176,
-      "grad_norm": 3.390652894973755,
-      "learning_rate": 4.6472763476034706e-06,
-      "loss": 1.9259,
-      "step": 9090
+      "epoch": 1.93024,
+      "grad_norm": 43.63051986694336,
+      "learning_rate": 7.751111111111112e-07,
+      "loss": 1.5879,
+      "step": 60320
     },
     {
-      "epoch": 0.58368,
-      "grad_norm": 3.3855576515197754,
-      "learning_rate": 4.625942255724648e-06,
-      "loss": 1.9133,
-      "step": 9120
+      "epoch": 1.93056,
+      "grad_norm": 45.21602249145508,
+      "learning_rate": 7.715555555555557e-07,
+      "loss": 1.5784,
+      "step": 60330
     },
     {
-      "epoch": 0.5856,
-      "grad_norm": 3.2181034088134766,
-      "learning_rate": 4.604608163845826e-06,
-      "loss": 1.9173,
-      "step": 9150
+      "epoch": 1.9308800000000002,
+      "grad_norm": 46.25171661376953,
+      "learning_rate": 7.68e-07,
+      "loss": 1.6037,
+      "step": 60340
     },
     {
-      "epoch": 0.58752,
-      "grad_norm": 3.317294120788574,
-      "learning_rate": 4.583274071967004e-06,
-      "loss": 1.9092,
-      "step": 9180
+      "epoch": 1.9312,
+      "grad_norm": 44.56587600708008,
+      "learning_rate": 7.644444444444444e-07,
+      "loss": 1.5812,
+      "step": 60350
     },
     {
-      "epoch": 0.58944,
-      "grad_norm": 3.251826763153076,
-      "learning_rate": 4.561939980088181e-06,
-      "loss": 1.9208,
-      "step": 9210
+      "epoch": 1.93152,
+      "grad_norm": 45.03590393066406,
+      "learning_rate": 7.608888888888889e-07,
+      "loss": 1.6225,
+      "step": 60360
     },
     {
-      "epoch": 0.59136,
-      "grad_norm": 3.2404837608337402,
-      "learning_rate": 4.540605888209359e-06,
-      "loss": 1.9207,
-      "step": 9240
+      "epoch": 1.93184,
+      "grad_norm": 44.908931732177734,
+      "learning_rate": 7.573333333333334e-07,
+      "loss": 1.5856,
+      "step": 60370
     },
     {
-      "epoch": 0.59328,
-      "grad_norm": 3.400644540786743,
-      "learning_rate": 4.519271796330537e-06,
-      "loss": 1.9108,
-      "step": 9270
+      "epoch": 1.93216,
+      "grad_norm": 43.26665496826172,
+      "learning_rate": 7.537777777777779e-07,
+      "loss": 1.5954,
+      "step": 60380
     },
     {
-      "epoch": 0.5952,
-      "grad_norm": 3.343181848526001,
-      "learning_rate": 4.497937704451714e-06,
-      "loss": 1.9286,
-      "step": 9300
+      "epoch": 1.93248,
+      "grad_norm": 45.77558135986328,
+      "learning_rate": 7.502222222222223e-07,
+      "loss": 1.592,
+      "step": 60390
     },
     {
-      "epoch": 0.59712,
-      "grad_norm": 3.3262417316436768,
-      "learning_rate": 4.476603612572892e-06,
-      "loss": 1.9105,
-      "step": 9330
+      "epoch": 1.9327999999999999,
+      "grad_norm": 44.50881576538086,
+      "learning_rate": 7.466666666666668e-07,
+      "loss": 1.5838,
+      "step": 60400
     },
     {
-      "epoch": 0.59904,
-      "grad_norm": 3.342047691345215,
-      "learning_rate": 4.45526952069407e-06,
-      "loss": 1.904,
-      "step": 9360
+      "epoch": 1.93312,
+      "grad_norm": 45.751163482666016,
+      "learning_rate": 7.431111111111111e-07,
+      "loss": 1.5825,
+      "step": 60410
     },
     {
-      "epoch": 0.60096,
-      "grad_norm": 3.308074951171875,
-      "learning_rate": 4.4339354288152465e-06,
-      "loss": 1.9244,
-      "step": 9390
+      "epoch": 1.93344,
+      "grad_norm": 47.217376708984375,
+      "learning_rate": 7.395555555555556e-07,
+      "loss": 1.5713,
+      "step": 60420
     },
     {
-      "epoch": 0.60288,
-      "grad_norm": 3.3136653900146484,
-      "learning_rate": 4.412601336936424e-06,
-      "loss": 1.9109,
-      "step": 9420
+      "epoch": 1.93376,
+      "grad_norm": 46.75165939331055,
+      "learning_rate": 7.36e-07,
+      "loss": 1.5891,
+      "step": 60430
     },
     {
-      "epoch": 0.6048,
-      "grad_norm": 3.191739320755005,
-      "learning_rate": 4.391267245057602e-06,
-      "loss": 1.9027,
-      "step": 9450
+      "epoch": 1.93408,
+      "grad_norm": 45.729713439941406,
+      "learning_rate": 7.324444444444445e-07,
+      "loss": 1.5803,
+      "step": 60440
     },
     {
-      "epoch": 0.60672,
-      "grad_norm": 3.285452127456665,
-      "learning_rate": 4.3699331531787804e-06,
-      "loss": 1.9083,
-      "step": 9480
+      "epoch": 1.9344000000000001,
+      "grad_norm": 44.836177825927734,
+      "learning_rate": 7.28888888888889e-07,
+      "loss": 1.5715,
+      "step": 60450
     },
     {
-      "epoch": 0.60864,
-      "grad_norm": 3.2615160942077637,
-      "learning_rate": 4.348599061299958e-06,
-      "loss": 1.9052,
-      "step": 9510
+      "epoch": 1.93472,
+      "grad_norm": 46.619300842285156,
+      "learning_rate": 7.253333333333335e-07,
+      "loss": 1.5787,
+      "step": 60460
     },
     {
-      "epoch": 0.61056,
-      "grad_norm": 3.286186695098877,
-      "learning_rate": 4.327264969421136e-06,
-      "loss": 1.9092,
-      "step": 9540
+      "epoch": 1.9350399999999999,
+      "grad_norm": 45.582069396972656,
+      "learning_rate": 7.217777777777778e-07,
+      "loss": 1.5786,
+      "step": 60470
     },
     {
-      "epoch": 0.61248,
-      "grad_norm": 3.311403512954712,
-      "learning_rate": 4.3059308775423135e-06,
-      "loss": 1.9158,
-      "step": 9570
+      "epoch": 1.93536,
+      "grad_norm": 43.99139404296875,
+      "learning_rate": 7.182222222222223e-07,
+      "loss": 1.6028,
+      "step": 60480
     },
     {
-      "epoch": 0.6144,
-      "grad_norm": 3.357612133026123,
-      "learning_rate": 4.28459678566349e-06,
-      "loss": 1.9001,
-      "step": 9600
+      "epoch": 1.93568,
+      "grad_norm": 47.75700759887695,
+      "learning_rate": 7.146666666666667e-07,
+      "loss": 1.5846,
+      "step": 60490
     },
     {
-      "epoch": 0.61632,
-      "grad_norm": 3.3893637657165527,
-      "learning_rate": 4.263262693784668e-06,
-      "loss": 1.9127,
-      "step": 9630
+      "epoch": 1.936,
+      "grad_norm": 47.97227096557617,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 1.5992,
+      "step": 60500
     },
     {
-      "epoch": 0.61824,
-      "grad_norm": 3.1436495780944824,
-      "learning_rate": 4.241928601905846e-06,
-      "loss": 1.9247,
-      "step": 9660
+      "epoch": 1.93632,
+      "grad_norm": 45.68336486816406,
+      "learning_rate": 7.075555555555556e-07,
+      "loss": 1.5879,
+      "step": 60510
     },
     {
-      "epoch": 0.62016,
-      "grad_norm": 3.3548660278320312,
-      "learning_rate": 4.220594510027023e-06,
-      "loss": 1.9114,
-      "step": 9690
+      "epoch": 1.9366400000000001,
+      "grad_norm": 44.39118194580078,
+      "learning_rate": 7.040000000000001e-07,
+      "loss": 1.613,
+      "step": 60520
     },
     {
-      "epoch": 0.62208,
-      "grad_norm": 3.2861545085906982,
-      "learning_rate": 4.199260418148201e-06,
-      "loss": 1.9032,
-      "step": 9720
+      "epoch": 1.93696,
+      "grad_norm": 47.586116790771484,
+      "learning_rate": 7.004444444444446e-07,
+      "loss": 1.569,
+      "step": 60530
     },
     {
-      "epoch": 0.624,
-      "grad_norm": 3.383060932159424,
-      "learning_rate": 4.177926326269379e-06,
-      "loss": 1.9043,
-      "step": 9750
+      "epoch": 1.93728,
+      "grad_norm": 46.94290542602539,
+      "learning_rate": 6.968888888888888e-07,
+      "loss": 1.5928,
+      "step": 60540
     },
     {
-      "epoch": 0.62592,
-      "grad_norm": 3.490474224090576,
-      "learning_rate": 4.1565922343905564e-06,
-      "loss": 1.9303,
-      "step": 9780
+      "epoch": 1.9376,
+      "grad_norm": 44.617095947265625,
+      "learning_rate": 6.933333333333334e-07,
+      "loss": 1.5557,
+      "step": 60550
     },
     {
-      "epoch": 0.62784,
-      "grad_norm": 3.2612555027008057,
-      "learning_rate": 4.135258142511734e-06,
-      "loss": 1.9156,
-      "step": 9810
+      "epoch": 1.93792,
+      "grad_norm": 45.4676513671875,
+      "learning_rate": 6.897777777777778e-07,
+      "loss": 1.5778,
+      "step": 60560
     },
     {
-      "epoch": 0.62976,
-      "grad_norm": 3.328390121459961,
-      "learning_rate": 4.113924050632912e-06,
-      "loss": 1.9181,
-      "step": 9840
+      "epoch": 1.93824,
+      "grad_norm": 46.230648040771484,
+      "learning_rate": 6.862222222222223e-07,
+      "loss": 1.5916,
+      "step": 60570
     },
     {
-      "epoch": 0.63168,
-      "grad_norm": 3.4136643409729004,
-      "learning_rate": 4.0925899587540895e-06,
-      "loss": 1.909,
-      "step": 9870
+      "epoch": 1.9385599999999998,
+      "grad_norm": 45.70417404174805,
+      "learning_rate": 6.826666666666667e-07,
+      "loss": 1.5775,
+      "step": 60580
     },
     {
-      "epoch": 0.6336,
-      "grad_norm": 3.4976086616516113,
-      "learning_rate": 4.071255866875267e-06,
-      "loss": 1.9377,
-      "step": 9900
+      "epoch": 1.9388800000000002,
+      "grad_norm": 46.56899642944336,
+      "learning_rate": 6.791111111111112e-07,
+      "loss": 1.5836,
+      "step": 60590
     },
     {
-      "epoch": 0.63552,
-      "grad_norm": 3.3502397537231445,
-      "learning_rate": 4.049921774996445e-06,
-      "loss": 1.9098,
-      "step": 9930
+      "epoch": 1.9392,
+      "grad_norm": 44.46154022216797,
+      "learning_rate": 6.755555555555555e-07,
+      "loss": 1.5723,
+      "step": 60600
     },
     {
-      "epoch": 0.63744,
-      "grad_norm": 3.301764726638794,
-      "learning_rate": 4.0285876831176226e-06,
-      "loss": 1.9106,
-      "step": 9960
+      "epoch": 1.93952,
+      "grad_norm": 45.880916595458984,
+      "learning_rate": 6.72e-07,
+      "loss": 1.573,
+      "step": 60610
     },
     {
-      "epoch": 0.63936,
-      "grad_norm": 3.3253800868988037,
-      "learning_rate": 4.0072535912388e-06,
-      "loss": 1.9049,
-      "step": 9990
+      "epoch": 1.93984,
+      "grad_norm": 46.16361618041992,
+      "learning_rate": 6.684444444444444e-07,
+      "loss": 1.5931,
+      "step": 60620
     },
     {
-      "epoch": 0.64128,
-      "grad_norm": 3.4913229942321777,
-      "learning_rate": 3.985919499359978e-06,
-      "loss": 1.9004,
-      "step": 10020
+      "epoch": 1.94016,
+      "grad_norm": 43.51244354248047,
+      "learning_rate": 6.64888888888889e-07,
+      "loss": 1.5756,
+      "step": 60630
     },
     {
-      "epoch": 0.6432,
-      "grad_norm": 3.363783121109009,
-      "learning_rate": 3.964585407481155e-06,
-      "loss": 1.9058,
-      "step": 10050
+      "epoch": 1.94048,
+      "grad_norm": 44.64166259765625,
+      "learning_rate": 6.613333333333334e-07,
+      "loss": 1.5582,
+      "step": 60640
     },
     {
-      "epoch": 0.64512,
-      "grad_norm": 3.4047791957855225,
-      "learning_rate": 3.9432513156023324e-06,
-      "loss": 1.9062,
-      "step": 10080
+      "epoch": 1.9407999999999999,
+      "grad_norm": 46.329654693603516,
+      "learning_rate": 6.577777777777779e-07,
+      "loss": 1.5762,
+      "step": 60650
     },
     {
-      "epoch": 0.64704,
-      "grad_norm": 3.248828411102295,
-      "learning_rate": 3.92191722372351e-06,
-      "loss": 1.9145,
-      "step": 10110
+      "epoch": 1.94112,
+      "grad_norm": 47.477203369140625,
+      "learning_rate": 6.542222222222223e-07,
+      "loss": 1.582,
+      "step": 60660
     },
     {
-      "epoch": 0.64896,
-      "grad_norm": 3.595554828643799,
-      "learning_rate": 3.900583131844688e-06,
-      "loss": 1.9162,
-      "step": 10140
+      "epoch": 1.94144,
+      "grad_norm": 44.95314407348633,
+      "learning_rate": 6.506666666666667e-07,
+      "loss": 1.5694,
+      "step": 60670
     },
     {
-      "epoch": 0.65088,
-      "grad_norm": 3.4337172508239746,
-      "learning_rate": 3.8792490399658655e-06,
-      "loss": 1.9162,
-      "step": 10170
+      "epoch": 1.94176,
+      "grad_norm": 46.40793228149414,
+      "learning_rate": 6.471111111111111e-07,
+      "loss": 1.5762,
+      "step": 60680
     },
     {
-      "epoch": 0.6528,
-      "grad_norm": 3.1357638835906982,
-      "learning_rate": 3.857914948087043e-06,
-      "loss": 1.9186,
-      "step": 10200
+      "epoch": 1.94208,
+      "grad_norm": 47.274681091308594,
+      "learning_rate": 6.435555555555556e-07,
+      "loss": 1.5578,
+      "step": 60690
     },
     {
-      "epoch": 0.65472,
-      "grad_norm": 3.47369647026062,
-      "learning_rate": 3.836580856208222e-06,
-      "loss": 1.9051,
-      "step": 10230
+      "epoch": 1.9424000000000001,
+      "grad_norm": 45.94260787963867,
+      "learning_rate": 6.4e-07,
+      "loss": 1.5773,
+      "step": 60700
     },
     {
-      "epoch": 0.65664,
-      "grad_norm": 3.5025222301483154,
-      "learning_rate": 3.8152467643293985e-06,
-      "loss": 1.9101,
-      "step": 10260
+      "epoch": 1.94272,
+      "grad_norm": 45.475677490234375,
+      "learning_rate": 6.364444444444446e-07,
+      "loss": 1.5859,
+      "step": 60710
     },
     {
-      "epoch": 0.65856,
-      "grad_norm": 3.3291850090026855,
-      "learning_rate": 3.7939126724505767e-06,
-      "loss": 1.9129,
-      "step": 10290
+      "epoch": 1.9430399999999999,
+      "grad_norm": 49.357112884521484,
+      "learning_rate": 6.32888888888889e-07,
+      "loss": 1.5858,
+      "step": 60720
     },
     {
-      "epoch": 0.66048,
-      "grad_norm": 3.334468364715576,
-      "learning_rate": 3.772578580571754e-06,
-      "loss": 1.9116,
-      "step": 10320
+      "epoch": 1.94336,
+      "grad_norm": 45.360870361328125,
+      "learning_rate": 6.293333333333334e-07,
+      "loss": 1.5734,
+      "step": 60730
     },
     {
-      "epoch": 0.6624,
-      "grad_norm": 3.216153860092163,
-      "learning_rate": 3.7512444886929316e-06,
-      "loss": 1.8991,
-      "step": 10350
+      "epoch": 1.94368,
+      "grad_norm": 46.292423248291016,
+      "learning_rate": 6.257777777777778e-07,
+      "loss": 1.5622,
+      "step": 60740
     },
     {
-      "epoch": 0.66432,
-      "grad_norm": 3.2956454753875732,
-      "learning_rate": 3.7299103968141093e-06,
-      "loss": 1.9092,
-      "step": 10380
+      "epoch": 1.944,
+      "grad_norm": 46.921993255615234,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 1.5833,
+      "step": 60750
     },
     {
-      "epoch": 0.66624,
-      "grad_norm": 3.658242702484131,
-      "learning_rate": 3.708576304935287e-06,
-      "loss": 1.9186,
-      "step": 10410
+      "epoch": 1.94432,
+      "grad_norm": 45.2756462097168,
+      "learning_rate": 6.186666666666667e-07,
+      "loss": 1.5684,
+      "step": 60760
     },
     {
-      "epoch": 0.66816,
-      "grad_norm": 3.3961496353149414,
-      "learning_rate": 3.6872422130564647e-06,
-      "loss": 1.908,
-      "step": 10440
+      "epoch": 1.9446400000000001,
+      "grad_norm": 44.58122634887695,
+      "learning_rate": 6.151111111111111e-07,
+      "loss": 1.5887,
+      "step": 60770
     },
     {
-      "epoch": 0.67008,
-      "grad_norm": 3.2990338802337646,
-      "learning_rate": 3.665908121177642e-06,
-      "loss": 1.9061,
-      "step": 10470
+      "epoch": 1.94496,
+      "grad_norm": 46.2196044921875,
+      "learning_rate": 6.115555555555556e-07,
+      "loss": 1.5575,
+      "step": 60780
     },
     {
-      "epoch": 0.672,
-      "grad_norm": 3.2165703773498535,
-      "learning_rate": 3.6445740292988196e-06,
-      "loss": 1.8989,
-      "step": 10500
+      "epoch": 1.94528,
+      "grad_norm": 45.56631851196289,
+      "learning_rate": 6.08e-07,
+      "loss": 1.5961,
+      "step": 60790
     },
     {
-      "epoch": 0.67392,
-      "grad_norm": 3.3282055854797363,
-      "learning_rate": 3.6232399374199973e-06,
-      "loss": 1.9058,
-      "step": 10530
+      "epoch": 1.9456,
+      "grad_norm": 46.696372985839844,
+      "learning_rate": 6.044444444444445e-07,
+      "loss": 1.5915,
+      "step": 60800
     },
     {
-      "epoch": 0.67584,
-      "grad_norm": 3.469121217727661,
-      "learning_rate": 3.601905845541175e-06,
-      "loss": 1.9011,
-      "step": 10560
+      "epoch": 1.94592,
+      "grad_norm": 46.68487548828125,
+      "learning_rate": 6.008888888888889e-07,
+      "loss": 1.5701,
+      "step": 60810
     },
     {
-      "epoch": 0.67776,
-      "grad_norm": 3.3346922397613525,
-      "learning_rate": 3.580571753662353e-06,
-      "loss": 1.9031,
-      "step": 10590
+      "epoch": 1.94624,
+      "grad_norm": 46.34423828125,
+      "learning_rate": 5.973333333333334e-07,
+      "loss": 1.585,
+      "step": 60820
     },
     {
-      "epoch": 0.67968,
-      "grad_norm": 3.366013526916504,
-      "learning_rate": 3.5592376617835308e-06,
-      "loss": 1.8915,
-      "step": 10620
+      "epoch": 1.9465599999999998,
+      "grad_norm": 45.44008255004883,
+      "learning_rate": 5.937777777777778e-07,
+      "loss": 1.5653,
+      "step": 60830
     },
     {
-      "epoch": 0.6816,
-      "grad_norm": 3.3305037021636963,
-      "learning_rate": 3.537903569904708e-06,
-      "loss": 1.9082,
-      "step": 10650
+      "epoch": 1.9468800000000002,
+      "grad_norm": 46.109352111816406,
+      "learning_rate": 5.902222222222223e-07,
+      "loss": 1.603,
+      "step": 60840
     },
     {
-      "epoch": 0.68352,
-      "grad_norm": 3.393563747406006,
-      "learning_rate": 3.5165694780258857e-06,
-      "loss": 1.9209,
-      "step": 10680
+      "epoch": 1.9472,
+      "grad_norm": 47.0877799987793,
+      "learning_rate": 5.866666666666667e-07,
+      "loss": 1.5975,
+      "step": 60850
     },
     {
-      "epoch": 0.68544,
-      "grad_norm": 3.2924818992614746,
-      "learning_rate": 3.4952353861470634e-06,
-      "loss": 1.9187,
-      "step": 10710
+      "epoch": 1.94752,
+      "grad_norm": 45.722328186035156,
+      "learning_rate": 5.831111111111112e-07,
+      "loss": 1.5726,
+      "step": 60860
     },
     {
-      "epoch": 0.68736,
-      "grad_norm": 3.1849982738494873,
-      "learning_rate": 3.473901294268241e-06,
-      "loss": 1.8997,
-      "step": 10740
+      "epoch": 1.94784,
+      "grad_norm": 47.501426696777344,
+      "learning_rate": 5.795555555555556e-07,
+      "loss": 1.6096,
+      "step": 60870
     },
     {
-      "epoch": 0.68928,
-      "grad_norm": 3.3936803340911865,
-      "learning_rate": 3.4525672023894188e-06,
-      "loss": 1.9039,
-      "step": 10770
+      "epoch": 1.9481600000000001,
+      "grad_norm": 46.5894889831543,
+      "learning_rate": 5.760000000000001e-07,
+      "loss": 1.5495,
+      "step": 60880
     },
     {
-      "epoch": 0.6912,
-      "grad_norm": 3.3271491527557373,
-      "learning_rate": 3.431233110510596e-06,
-      "loss": 1.9166,
-      "step": 10800
+      "epoch": 1.94848,
+      "grad_norm": 44.740455627441406,
+      "learning_rate": 5.724444444444445e-07,
+      "loss": 1.5948,
+      "step": 60890
     },
     {
-      "epoch": 0.69312,
-      "grad_norm": 3.367232322692871,
-      "learning_rate": 3.4098990186317737e-06,
-      "loss": 1.9078,
-      "step": 10830
+      "epoch": 1.9487999999999999,
+      "grad_norm": 45.47201919555664,
+      "learning_rate": 5.68888888888889e-07,
+      "loss": 1.5641,
+      "step": 60900
     },
     {
-      "epoch": 0.69504,
-      "grad_norm": 3.2815470695495605,
-      "learning_rate": 3.3885649267529514e-06,
-      "loss": 1.9063,
-      "step": 10860
+      "epoch": 1.94912,
+      "grad_norm": 45.011924743652344,
+      "learning_rate": 5.653333333333333e-07,
+      "loss": 1.5768,
+      "step": 60910
     },
     {
-      "epoch": 0.69696,
-      "grad_norm": 3.5337584018707275,
-      "learning_rate": 3.367230834874129e-06,
-      "loss": 1.9121,
-      "step": 10890
+      "epoch": 1.94944,
+      "grad_norm": 45.84285354614258,
+      "learning_rate": 5.617777777777779e-07,
+      "loss": 1.5938,
+      "step": 60920
     },
     {
-      "epoch": 0.69888,
-      "grad_norm": 3.3683621883392334,
-      "learning_rate": 3.3458967429953063e-06,
-      "loss": 1.9117,
-      "step": 10920
+      "epoch": 1.94976,
+      "grad_norm": 46.208534240722656,
+      "learning_rate": 5.582222222222223e-07,
+      "loss": 1.5976,
+      "step": 60930
     },
     {
-      "epoch": 0.7008,
-      "grad_norm": 3.2509257793426514,
-      "learning_rate": 3.324562651116485e-06,
-      "loss": 1.9014,
-      "step": 10950
+      "epoch": 1.95008,
+      "grad_norm": 46.11305618286133,
+      "learning_rate": 5.546666666666667e-07,
+      "loss": 1.5793,
+      "step": 60940
     },
     {
-      "epoch": 0.70272,
-      "grad_norm": 3.33274245262146,
-      "learning_rate": 3.303228559237662e-06,
-      "loss": 1.9054,
-      "step": 10980
+      "epoch": 1.9504000000000001,
+      "grad_norm": 45.28938674926758,
+      "learning_rate": 5.511111111111111e-07,
+      "loss": 1.5667,
+      "step": 60950
     },
     {
-      "epoch": 0.70464,
-      "grad_norm": 3.2902543544769287,
-      "learning_rate": 3.28189446735884e-06,
-      "loss": 1.8939,
-      "step": 11010
+      "epoch": 1.95072,
+      "grad_norm": 46.1634521484375,
+      "learning_rate": 5.475555555555557e-07,
+      "loss": 1.597,
+      "step": 60960
     },
     {
-      "epoch": 0.70656,
-      "grad_norm": 3.267218828201294,
-      "learning_rate": 3.2605603754800175e-06,
-      "loss": 1.9137,
-      "step": 11040
+      "epoch": 1.9510399999999999,
+      "grad_norm": 44.47576904296875,
+      "learning_rate": 5.44e-07,
+      "loss": 1.5682,
+      "step": 60970
     },
     {
-      "epoch": 0.70848,
-      "grad_norm": 3.272397041320801,
-      "learning_rate": 3.239226283601195e-06,
-      "loss": 1.9059,
-      "step": 11070
+      "epoch": 1.95136,
+      "grad_norm": 44.76446533203125,
+      "learning_rate": 5.404444444444444e-07,
+      "loss": 1.5536,
+      "step": 60980
     },
     {
-      "epoch": 0.7104,
-      "grad_norm": 3.419252872467041,
-      "learning_rate": 3.217892191722373e-06,
-      "loss": 1.8978,
-      "step": 11100
+      "epoch": 1.95168,
+      "grad_norm": 46.57415771484375,
+      "learning_rate": 5.368888888888889e-07,
+      "loss": 1.5834,
+      "step": 60990
     },
     {
-      "epoch": 0.71232,
-      "grad_norm": 3.4831225872039795,
-      "learning_rate": 3.19655809984355e-06,
-      "loss": 1.9247,
-      "step": 11130
+      "epoch": 1.952,
+      "grad_norm": 46.719879150390625,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.5883,
+      "step": 61000
     },
     {
-      "epoch": 0.71424,
-      "grad_norm": 3.237103223800659,
-      "learning_rate": 3.175224007964728e-06,
-      "loss": 1.9014,
-      "step": 11160
+      "epoch": 1.95232,
+      "grad_norm": 45.457401275634766,
+      "learning_rate": 5.297777777777778e-07,
+      "loss": 1.6105,
+      "step": 61010
     },
     {
-      "epoch": 0.71616,
-      "grad_norm": 3.282529592514038,
-      "learning_rate": 3.1538899160859055e-06,
-      "loss": 1.9072,
-      "step": 11190
+      "epoch": 1.9526400000000002,
+      "grad_norm": 46.55323028564453,
+      "learning_rate": 5.262222222222222e-07,
+      "loss": 1.5661,
+      "step": 61020
     },
     {
-      "epoch": 0.71808,
-      "grad_norm": 3.2297005653381348,
-      "learning_rate": 3.132555824207083e-06,
-      "loss": 1.9131,
-      "step": 11220
+      "epoch": 1.95296,
+      "grad_norm": 46.279483795166016,
+      "learning_rate": 5.226666666666667e-07,
+      "loss": 1.6017,
+      "step": 61030
     },
     {
-      "epoch": 0.72,
-      "grad_norm": 3.4319121837615967,
-      "learning_rate": 3.1112217323282604e-06,
-      "loss": 1.9111,
-      "step": 11250
+      "epoch": 1.95328,
+      "grad_norm": 45.748374938964844,
+      "learning_rate": 5.191111111111111e-07,
+      "loss": 1.5772,
+      "step": 61040
     },
     {
-      "epoch": 0.72192,
-      "grad_norm": 3.3072924613952637,
-      "learning_rate": 3.089887640449438e-06,
-      "loss": 1.9184,
-      "step": 11280
+      "epoch": 1.9536,
+      "grad_norm": 46.23691177368164,
+      "learning_rate": 5.155555555555556e-07,
+      "loss": 1.5956,
+      "step": 61050
     },
     {
-      "epoch": 0.72384,
-      "grad_norm": 3.4026801586151123,
-      "learning_rate": 3.068553548570616e-06,
-      "loss": 1.8978,
-      "step": 11310
+      "epoch": 1.95392,
+      "grad_norm": 45.71986389160156,
+      "learning_rate": 5.12e-07,
+      "loss": 1.5746,
+      "step": 61060
     },
     {
-      "epoch": 0.72576,
-      "grad_norm": 3.3368184566497803,
-      "learning_rate": 3.047219456691794e-06,
-      "loss": 1.8931,
-      "step": 11340
+      "epoch": 1.95424,
+      "grad_norm": 46.47679138183594,
+      "learning_rate": 5.084444444444445e-07,
+      "loss": 1.5706,
+      "step": 61070
     },
     {
-      "epoch": 0.72768,
-      "grad_norm": 3.2491300106048584,
-      "learning_rate": 3.0258853648129716e-06,
-      "loss": 1.9008,
-      "step": 11370
+      "epoch": 1.9545599999999999,
+      "grad_norm": 44.26342010498047,
+      "learning_rate": 5.048888888888889e-07,
+      "loss": 1.5738,
+      "step": 61080
     },
     {
-      "epoch": 0.7296,
-      "grad_norm": 3.3876442909240723,
-      "learning_rate": 3.0045512729341493e-06,
-      "loss": 1.9075,
-      "step": 11400
+      "epoch": 1.95488,
+      "grad_norm": 46.3544921875,
+      "learning_rate": 5.013333333333334e-07,
+      "loss": 1.6082,
+      "step": 61090
     },
     {
-      "epoch": 0.73152,
-      "grad_norm": 3.233262777328491,
-      "learning_rate": 2.983217181055327e-06,
-      "loss": 1.8927,
-      "step": 11430
+      "epoch": 1.9552,
+      "grad_norm": 47.279273986816406,
+      "learning_rate": 4.977777777777778e-07,
+      "loss": 1.5935,
+      "step": 61100
     },
     {
-      "epoch": 0.73344,
-      "grad_norm": 3.400437116622925,
-      "learning_rate": 2.9618830891765042e-06,
-      "loss": 1.8926,
-      "step": 11460
+      "epoch": 1.95552,
+      "grad_norm": 45.07588577270508,
+      "learning_rate": 4.942222222222223e-07,
+      "loss": 1.5689,
+      "step": 61110
     },
     {
-      "epoch": 0.73536,
-      "grad_norm": 3.355999708175659,
-      "learning_rate": 2.940548997297682e-06,
-      "loss": 1.9085,
-      "step": 11490
+      "epoch": 1.95584,
+      "grad_norm": 46.14258575439453,
+      "learning_rate": 4.906666666666667e-07,
+      "loss": 1.5729,
+      "step": 61120
     },
     {
-      "epoch": 0.73728,
-      "grad_norm": 3.339808702468872,
-      "learning_rate": 2.9192149054188596e-06,
-      "loss": 1.903,
-      "step": 11520
+      "epoch": 1.9561600000000001,
+      "grad_norm": 46.78408432006836,
+      "learning_rate": 4.871111111111112e-07,
+      "loss": 1.5916,
+      "step": 61130
     },
     {
-      "epoch": 0.7392,
-      "grad_norm": 3.3403186798095703,
-      "learning_rate": 2.8978808135400373e-06,
-      "loss": 1.9111,
-      "step": 11550
+      "epoch": 1.95648,
+      "grad_norm": 46.98505783081055,
+      "learning_rate": 4.835555555555556e-07,
+      "loss": 1.5717,
+      "step": 61140
     },
     {
-      "epoch": 0.74112,
-      "grad_norm": 3.515465259552002,
-      "learning_rate": 2.8765467216612145e-06,
-      "loss": 1.8907,
-      "step": 11580
+      "epoch": 1.9567999999999999,
+      "grad_norm": 47.45516586303711,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.5884,
+      "step": 61150
     },
     {
-      "epoch": 0.74304,
-      "grad_norm": 3.2402706146240234,
-      "learning_rate": 2.8552126297823922e-06,
-      "loss": 1.89,
-      "step": 11610
+      "epoch": 1.95712,
+      "grad_norm": 45.89262390136719,
+      "learning_rate": 4.764444444444445e-07,
+      "loss": 1.5847,
+      "step": 61160
     },
     {
-      "epoch": 0.74496,
-      "grad_norm": 3.2288084030151367,
-      "learning_rate": 2.83387853790357e-06,
-      "loss": 1.9053,
-      "step": 11640
+      "epoch": 1.95744,
+      "grad_norm": 46.40864181518555,
+      "learning_rate": 4.728888888888889e-07,
+      "loss": 1.5957,
+      "step": 61170
     },
     {
-      "epoch": 0.74688,
-      "grad_norm": 3.44089412689209,
-      "learning_rate": 2.8125444460247476e-06,
-      "loss": 1.9179,
-      "step": 11670
+      "epoch": 1.95776,
+      "grad_norm": 46.90511703491211,
+      "learning_rate": 4.6933333333333335e-07,
+      "loss": 1.5829,
+      "step": 61180
     },
     {
-      "epoch": 0.7488,
-      "grad_norm": 3.160490036010742,
-      "learning_rate": 2.7912103541459257e-06,
-      "loss": 1.9038,
-      "step": 11700
+      "epoch": 1.95808,
+      "grad_norm": 47.575321197509766,
+      "learning_rate": 4.6577777777777785e-07,
+      "loss": 1.5846,
+      "step": 61190
     },
     {
-      "epoch": 0.75072,
-      "grad_norm": 3.2737534046173096,
-      "learning_rate": 2.7698762622671034e-06,
-      "loss": 1.9218,
-      "step": 11730
+      "epoch": 1.9584000000000001,
+      "grad_norm": 46.93019485473633,
+      "learning_rate": 4.6222222222222225e-07,
+      "loss": 1.5949,
+      "step": 61200
     },
     {
-      "epoch": 0.75264,
-      "grad_norm": 3.3526477813720703,
-      "learning_rate": 2.748542170388281e-06,
-      "loss": 1.9002,
-      "step": 11760
+      "epoch": 1.95872,
+      "grad_norm": 44.054134368896484,
+      "learning_rate": 4.586666666666667e-07,
+      "loss": 1.6058,
+      "step": 61210
     },
     {
-      "epoch": 0.75456,
-      "grad_norm": 3.3810863494873047,
-      "learning_rate": 2.7272080785094583e-06,
-      "loss": 1.8998,
-      "step": 11790
+      "epoch": 1.95904,
+      "grad_norm": 43.94884490966797,
+      "learning_rate": 4.5511111111111115e-07,
+      "loss": 1.5917,
+      "step": 61220
     },
     {
-      "epoch": 0.75648,
-      "grad_norm": 3.4697837829589844,
-      "learning_rate": 2.705873986630636e-06,
-      "loss": 1.9003,
-      "step": 11820
+      "epoch": 1.95936,
+      "grad_norm": 47.810829162597656,
+      "learning_rate": 4.5155555555555554e-07,
+      "loss": 1.5809,
+      "step": 61230
     },
     {
-      "epoch": 0.7584,
-      "grad_norm": 3.4439995288848877,
-      "learning_rate": 2.6845398947518137e-06,
-      "loss": 1.9138,
-      "step": 11850
+      "epoch": 1.95968,
+      "grad_norm": 46.7258415222168,
+      "learning_rate": 4.4800000000000004e-07,
+      "loss": 1.5671,
+      "step": 61240
     },
     {
-      "epoch": 0.76032,
-      "grad_norm": 3.241582155227661,
-      "learning_rate": 2.6632058028729914e-06,
-      "loss": 1.8976,
-      "step": 11880
+      "epoch": 1.96,
+      "grad_norm": 46.42376708984375,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 1.5741,
+      "step": 61250
     },
     {
-      "epoch": 0.76224,
-      "grad_norm": 3.2096734046936035,
-      "learning_rate": 2.6418717109941686e-06,
-      "loss": 1.8932,
-      "step": 11910
+      "epoch": 1.9603199999999998,
+      "grad_norm": 44.41875457763672,
+      "learning_rate": 4.408888888888889e-07,
+      "loss": 1.5793,
+      "step": 61260
     },
     {
-      "epoch": 0.76416,
-      "grad_norm": 3.399176597595215,
-      "learning_rate": 2.6205376191153463e-06,
-      "loss": 1.9297,
-      "step": 11940
+      "epoch": 1.9606400000000002,
+      "grad_norm": 46.66280746459961,
+      "learning_rate": 4.373333333333334e-07,
+      "loss": 1.587,
+      "step": 61270
     },
     {
-      "epoch": 0.76608,
-      "grad_norm": 3.4943864345550537,
-      "learning_rate": 2.599203527236524e-06,
-      "loss": 1.9056,
-      "step": 11970
+      "epoch": 1.96096,
+      "grad_norm": 46.54970169067383,
+      "learning_rate": 4.3377777777777783e-07,
+      "loss": 1.5802,
+      "step": 61280
     },
     {
-      "epoch": 0.768,
-      "grad_norm": 3.349480628967285,
-      "learning_rate": 2.5778694353577017e-06,
-      "loss": 1.938,
-      "step": 12000
+      "epoch": 1.96128,
+      "grad_norm": 46.72460174560547,
+      "learning_rate": 4.302222222222223e-07,
+      "loss": 1.5841,
+      "step": 61290
     },
     {
-      "epoch": 0.76992,
-      "grad_norm": 3.375905752182007,
-      "learning_rate": 2.5565353434788794e-06,
-      "loss": 1.9132,
-      "step": 12030
+      "epoch": 1.9616,
+      "grad_norm": 46.453678131103516,
+      "learning_rate": 4.266666666666667e-07,
+      "loss": 1.5926,
+      "step": 61300
     },
     {
-      "epoch": 0.77184,
-      "grad_norm": 3.4605822563171387,
-      "learning_rate": 2.5352012516000575e-06,
-      "loss": 1.9036,
-      "step": 12060
+      "epoch": 1.96192,
+      "grad_norm": 44.05784225463867,
+      "learning_rate": 4.231111111111112e-07,
+      "loss": 1.5745,
+      "step": 61310
     },
     {
-      "epoch": 0.77376,
-      "grad_norm": 3.35102915763855,
-      "learning_rate": 2.5138671597212348e-06,
-      "loss": 1.8944,
-      "step": 12090
+      "epoch": 1.96224,
+      "grad_norm": 46.23289108276367,
+      "learning_rate": 4.195555555555556e-07,
+      "loss": 1.561,
+      "step": 61320
     },
     {
-      "epoch": 0.77568,
-      "grad_norm": 3.368705987930298,
-      "learning_rate": 2.4925330678424124e-06,
-      "loss": 1.9094,
-      "step": 12120
+      "epoch": 1.9625599999999999,
+      "grad_norm": 47.4354362487793,
+      "learning_rate": 4.16e-07,
+      "loss": 1.6063,
+      "step": 61330
     },
     {
-      "epoch": 0.7776,
-      "grad_norm": 3.3293306827545166,
-      "learning_rate": 2.47119897596359e-06,
-      "loss": 1.8971,
-      "step": 12150
+      "epoch": 1.96288,
+      "grad_norm": 46.875816345214844,
+      "learning_rate": 4.1244444444444447e-07,
+      "loss": 1.5885,
+      "step": 61340
     },
     {
-      "epoch": 0.77952,
-      "grad_norm": 3.4994451999664307,
-      "learning_rate": 2.449864884084768e-06,
-      "loss": 1.9122,
-      "step": 12180
+      "epoch": 1.9632,
+      "grad_norm": 46.69877243041992,
+      "learning_rate": 4.0888888888888897e-07,
+      "loss": 1.5855,
+      "step": 61350
     },
     {
-      "epoch": 0.78144,
-      "grad_norm": 3.353395462036133,
-      "learning_rate": 2.4285307922059455e-06,
-      "loss": 1.9003,
-      "step": 12210
+      "epoch": 1.96352,
+      "grad_norm": 45.40488052368164,
+      "learning_rate": 4.0533333333333336e-07,
+      "loss": 1.5825,
+      "step": 61360
     },
     {
-      "epoch": 0.78336,
-      "grad_norm": 3.344747304916382,
-      "learning_rate": 2.4071967003271228e-06,
-      "loss": 1.9075,
-      "step": 12240
+      "epoch": 1.96384,
+      "grad_norm": 44.27571105957031,
+      "learning_rate": 4.017777777777778e-07,
+      "loss": 1.5694,
+      "step": 61370
     },
     {
-      "epoch": 0.78528,
-      "grad_norm": 3.3555798530578613,
-      "learning_rate": 2.3858626084483004e-06,
-      "loss": 1.9026,
-      "step": 12270
+      "epoch": 1.9641600000000001,
+      "grad_norm": 48.688377380371094,
+      "learning_rate": 3.9822222222222226e-07,
+      "loss": 1.5695,
+      "step": 61380
     },
     {
-      "epoch": 0.7872,
-      "grad_norm": 3.2820401191711426,
-      "learning_rate": 2.364528516569478e-06,
-      "loss": 1.8766,
-      "step": 12300
+      "epoch": 1.96448,
+      "grad_norm": 46.94468688964844,
+      "learning_rate": 3.9466666666666665e-07,
+      "loss": 1.6029,
+      "step": 61390
     },
     {
-      "epoch": 0.78912,
-      "grad_norm": 3.4789068698883057,
-      "learning_rate": 2.343194424690656e-06,
-      "loss": 1.9038,
-      "step": 12330
+      "epoch": 1.9647999999999999,
+      "grad_norm": 46.42035675048828,
+      "learning_rate": 3.9111111111111115e-07,
+      "loss": 1.6019,
+      "step": 61400
     },
     {
-      "epoch": 0.79104,
-      "grad_norm": 3.4297752380371094,
-      "learning_rate": 2.3218603328118335e-06,
-      "loss": 1.8996,
-      "step": 12360
+      "epoch": 1.96512,
+      "grad_norm": 47.09354782104492,
+      "learning_rate": 3.875555555555556e-07,
+      "loss": 1.6001,
+      "step": 61410
     },
     {
-      "epoch": 0.79296,
-      "grad_norm": 3.3933277130126953,
-      "learning_rate": 2.300526240933011e-06,
-      "loss": 1.9004,
-      "step": 12390
+      "epoch": 1.96544,
+      "grad_norm": 44.141197204589844,
+      "learning_rate": 3.84e-07,
+      "loss": 1.5785,
+      "step": 61420
     },
     {
-      "epoch": 0.79488,
-      "grad_norm": 3.2792038917541504,
-      "learning_rate": 2.279192149054189e-06,
-      "loss": 1.9102,
-      "step": 12420
+      "epoch": 1.96576,
+      "grad_norm": 48.216793060302734,
+      "learning_rate": 3.8044444444444445e-07,
+      "loss": 1.588,
+      "step": 61430
     },
     {
-      "epoch": 0.7968,
-      "grad_norm": 3.3389816284179688,
-      "learning_rate": 2.2578580571753665e-06,
-      "loss": 1.9084,
-      "step": 12450
+      "epoch": 1.96608,
+      "grad_norm": 46.14058303833008,
+      "learning_rate": 3.7688888888888895e-07,
+      "loss": 1.5482,
+      "step": 61440
     },
     {
-      "epoch": 0.79872,
-      "grad_norm": 3.2868168354034424,
-      "learning_rate": 2.236523965296544e-06,
-      "loss": 1.9123,
-      "step": 12480
+      "epoch": 1.9664000000000001,
+      "grad_norm": 46.50981140136719,
+      "learning_rate": 3.733333333333334e-07,
+      "loss": 1.5704,
+      "step": 61450
     },
     {
-      "epoch": 0.80064,
-      "grad_norm": 3.4595017433166504,
-      "learning_rate": 2.2151898734177215e-06,
-      "loss": 1.9092,
-      "step": 12510
+      "epoch": 1.96672,
+      "grad_norm": 46.209110260009766,
+      "learning_rate": 3.697777777777778e-07,
+      "loss": 1.5685,
+      "step": 61460
     },
     {
-      "epoch": 0.80256,
-      "grad_norm": 3.3249852657318115,
-      "learning_rate": 2.1938557815388996e-06,
-      "loss": 1.8966,
-      "step": 12540
+      "epoch": 1.96704,
+      "grad_norm": 45.74169158935547,
+      "learning_rate": 3.6622222222222224e-07,
+      "loss": 1.62,
+      "step": 61470
     },
     {
-      "epoch": 0.80448,
-      "grad_norm": 3.424227714538574,
-      "learning_rate": 2.172521689660077e-06,
-      "loss": 1.8931,
-      "step": 12570
+      "epoch": 1.96736,
+      "grad_norm": 46.35404968261719,
+      "learning_rate": 3.6266666666666674e-07,
+      "loss": 1.5676,
+      "step": 61480
     },
     {
-      "epoch": 0.8064,
-      "grad_norm": 3.3327128887176514,
-      "learning_rate": 2.1511875977812545e-06,
-      "loss": 1.8952,
-      "step": 12600
+      "epoch": 1.96768,
+      "grad_norm": 47.3962287902832,
+      "learning_rate": 3.5911111111111113e-07,
+      "loss": 1.591,
+      "step": 61490
     },
     {
-      "epoch": 0.80832,
-      "grad_norm": 3.277287483215332,
-      "learning_rate": 2.1298535059024322e-06,
-      "loss": 1.89,
-      "step": 12630
+      "epoch": 1.968,
+      "grad_norm": 46.52360534667969,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 1.5589,
+      "step": 61500
     },
     {
-      "epoch": 0.81024,
-      "grad_norm": 3.407421588897705,
-      "learning_rate": 2.10851941402361e-06,
-      "loss": 1.9148,
-      "step": 12660
+      "epoch": 1.9683199999999998,
+      "grad_norm": 44.55923843383789,
+      "learning_rate": 3.5200000000000003e-07,
+      "loss": 1.588,
+      "step": 61510
     },
     {
-      "epoch": 0.81216,
-      "grad_norm": 3.3559443950653076,
-      "learning_rate": 2.0871853221447876e-06,
-      "loss": 1.9029,
-      "step": 12690
+      "epoch": 1.96864,
+      "grad_norm": 45.096378326416016,
+      "learning_rate": 3.484444444444444e-07,
+      "loss": 1.6024,
+      "step": 61520
     },
     {
-      "epoch": 0.81408,
-      "grad_norm": 3.2470178604125977,
-      "learning_rate": 2.0658512302659653e-06,
-      "loss": 1.896,
-      "step": 12720
+      "epoch": 1.96896,
+      "grad_norm": 44.80331802368164,
+      "learning_rate": 3.448888888888889e-07,
+      "loss": 1.5675,
+      "step": 61530
     },
     {
-      "epoch": 0.816,
-      "grad_norm": 3.3119843006134033,
-      "learning_rate": 2.044517138387143e-06,
-      "loss": 1.9002,
-      "step": 12750
+      "epoch": 1.96928,
+      "grad_norm": 45.03395462036133,
+      "learning_rate": 3.4133333333333337e-07,
+      "loss": 1.6069,
+      "step": 61540
     },
     {
-      "epoch": 0.81792,
-      "grad_norm": 3.318296194076538,
-      "learning_rate": 2.0231830465083206e-06,
-      "loss": 1.9117,
-      "step": 12780
+      "epoch": 1.9696,
+      "grad_norm": 46.5980224609375,
+      "learning_rate": 3.3777777777777777e-07,
+      "loss": 1.612,
+      "step": 61550
     },
     {
-      "epoch": 0.81984,
-      "grad_norm": 3.3162431716918945,
-      "learning_rate": 2.001848954629498e-06,
-      "loss": 1.9073,
-      "step": 12810
+      "epoch": 1.9699200000000001,
+      "grad_norm": 46.33894348144531,
+      "learning_rate": 3.342222222222222e-07,
+      "loss": 1.5674,
+      "step": 61560
     },
     {
-      "epoch": 0.82176,
-      "grad_norm": 3.295182704925537,
-      "learning_rate": 1.9805148627506756e-06,
-      "loss": 1.896,
-      "step": 12840
+      "epoch": 1.97024,
+      "grad_norm": 45.43359375,
+      "learning_rate": 3.306666666666667e-07,
+      "loss": 1.5893,
+      "step": 61570
     },
     {
-      "epoch": 0.82368,
-      "grad_norm": 3.37908935546875,
-      "learning_rate": 1.9591807708718533e-06,
-      "loss": 1.9176,
-      "step": 12870
+      "epoch": 1.9705599999999999,
+      "grad_norm": 44.497108459472656,
+      "learning_rate": 3.2711111111111116e-07,
+      "loss": 1.5841,
+      "step": 61580
     },
     {
-      "epoch": 0.8256,
-      "grad_norm": 3.519794225692749,
-      "learning_rate": 1.937846678993031e-06,
-      "loss": 1.8864,
-      "step": 12900
+      "epoch": 1.97088,
+      "grad_norm": 45.044307708740234,
+      "learning_rate": 3.2355555555555556e-07,
+      "loss": 1.5929,
+      "step": 61590
     },
     {
-      "epoch": 0.82752,
-      "grad_norm": 3.2552225589752197,
-      "learning_rate": 1.9165125871142086e-06,
-      "loss": 1.907,
-      "step": 12930
+      "epoch": 1.9712,
+      "grad_norm": 45.46745681762695,
+      "learning_rate": 3.2e-07,
+      "loss": 1.5928,
+      "step": 61600
     },
     {
-      "epoch": 0.82944,
-      "grad_norm": 3.339785575866699,
-      "learning_rate": 1.8951784952353863e-06,
-      "loss": 1.9199,
-      "step": 12960
+      "epoch": 1.97152,
+      "grad_norm": 44.18543243408203,
+      "learning_rate": 3.164444444444445e-07,
+      "loss": 1.5643,
+      "step": 61610
     },
     {
-      "epoch": 0.83136,
-      "grad_norm": 3.304584264755249,
-      "learning_rate": 1.873844403356564e-06,
-      "loss": 1.8803,
-      "step": 12990
+      "epoch": 1.97184,
+      "grad_norm": 48.65835189819336,
+      "learning_rate": 3.128888888888889e-07,
+      "loss": 1.5772,
+      "step": 61620
     },
     {
-      "epoch": 0.83328,
-      "grad_norm": 3.345766067504883,
-      "learning_rate": 1.8525103114777415e-06,
-      "loss": 1.9116,
-      "step": 13020
+      "epoch": 1.9721600000000001,
+      "grad_norm": 43.52328872680664,
+      "learning_rate": 3.0933333333333335e-07,
+      "loss": 1.5517,
+      "step": 61630
     },
     {
-      "epoch": 0.8352,
-      "grad_norm": 3.291215658187866,
-      "learning_rate": 1.8311762195989192e-06,
-      "loss": 1.9135,
-      "step": 13050
+      "epoch": 1.97248,
+      "grad_norm": 45.04435729980469,
+      "learning_rate": 3.057777777777778e-07,
+      "loss": 1.5833,
+      "step": 61640
     },
     {
-      "epoch": 0.83712,
-      "grad_norm": 3.1401145458221436,
-      "learning_rate": 1.809842127720097e-06,
-      "loss": 1.8967,
-      "step": 13080
+      "epoch": 1.9727999999999999,
+      "grad_norm": 49.300537109375,
+      "learning_rate": 3.0222222222222225e-07,
+      "loss": 1.581,
+      "step": 61650
     },
     {
-      "epoch": 0.83904,
-      "grad_norm": 3.31388521194458,
-      "learning_rate": 1.7885080358412745e-06,
-      "loss": 1.9126,
-      "step": 13110
+      "epoch": 1.97312,
+      "grad_norm": 45.710182189941406,
+      "learning_rate": 2.986666666666667e-07,
+      "loss": 1.6062,
+      "step": 61660
     },
     {
-      "epoch": 0.84096,
-      "grad_norm": 3.2397680282592773,
-      "learning_rate": 1.7671739439624522e-06,
-      "loss": 1.9009,
-      "step": 13140
+      "epoch": 1.97344,
+      "grad_norm": 45.44401550292969,
+      "learning_rate": 2.9511111111111114e-07,
+      "loss": 1.5691,
+      "step": 61670
     },
     {
-      "epoch": 0.84288,
-      "grad_norm": 3.5296459197998047,
-      "learning_rate": 1.7458398520836297e-06,
-      "loss": 1.8951,
-      "step": 13170
+      "epoch": 1.97376,
+      "grad_norm": 46.06452178955078,
+      "learning_rate": 2.915555555555556e-07,
+      "loss": 1.5492,
+      "step": 61680
     },
     {
-      "epoch": 0.8448,
-      "grad_norm": 3.4343180656433105,
-      "learning_rate": 1.7245057602048074e-06,
-      "loss": 1.8987,
-      "step": 13200
+      "epoch": 1.9740799999999998,
+      "grad_norm": 45.38791275024414,
+      "learning_rate": 2.8800000000000004e-07,
+      "loss": 1.5955,
+      "step": 61690
     },
     {
-      "epoch": 0.84672,
-      "grad_norm": 3.2717978954315186,
-      "learning_rate": 1.703171668325985e-06,
-      "loss": 1.8934,
-      "step": 13230
+      "epoch": 1.9744000000000002,
+      "grad_norm": 45.86880874633789,
+      "learning_rate": 2.844444444444445e-07,
+      "loss": 1.6253,
+      "step": 61700
     },
     {
-      "epoch": 0.84864,
-      "grad_norm": 3.2265992164611816,
-      "learning_rate": 1.6818375764471627e-06,
-      "loss": 1.9097,
-      "step": 13260
+      "epoch": 1.97472,
+      "grad_norm": 46.864437103271484,
+      "learning_rate": 2.8088888888888893e-07,
+      "loss": 1.5992,
+      "step": 61710
     },
     {
-      "epoch": 0.85056,
-      "grad_norm": 3.249821662902832,
-      "learning_rate": 1.6605034845683404e-06,
-      "loss": 1.89,
-      "step": 13290
+      "epoch": 1.97504,
+      "grad_norm": 46.647247314453125,
+      "learning_rate": 2.7733333333333333e-07,
+      "loss": 1.5904,
+      "step": 61720
     },
     {
-      "epoch": 0.85248,
-      "grad_norm": 3.239107131958008,
-      "learning_rate": 1.6391693926895181e-06,
-      "loss": 1.9062,
-      "step": 13320
+      "epoch": 1.97536,
+      "grad_norm": 46.546382904052734,
+      "learning_rate": 2.7377777777777783e-07,
+      "loss": 1.6046,
+      "step": 61730
     },
     {
-      "epoch": 0.8544,
-      "grad_norm": 3.2462799549102783,
-      "learning_rate": 1.6178353008106956e-06,
-      "loss": 1.8826,
-      "step": 13350
+      "epoch": 1.97568,
+      "grad_norm": 45.93708038330078,
+      "learning_rate": 2.702222222222222e-07,
+      "loss": 1.5922,
+      "step": 61740
     },
     {
-      "epoch": 0.85632,
-      "grad_norm": 3.1941652297973633,
-      "learning_rate": 1.5965012089318733e-06,
-      "loss": 1.8957,
-      "step": 13380
+      "epoch": 1.976,
+      "grad_norm": 47.503761291503906,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1.5745,
+      "step": 61750
     },
     {
-      "epoch": 0.85824,
-      "grad_norm": 3.190826416015625,
-      "learning_rate": 1.5751671170530507e-06,
-      "loss": 1.9277,
-      "step": 13410
+      "epoch": 1.9763199999999999,
+      "grad_norm": 47.48344039916992,
+      "learning_rate": 2.631111111111111e-07,
+      "loss": 1.5903,
+      "step": 61760
     },
     {
-      "epoch": 0.86016,
-      "grad_norm": 3.2493083477020264,
-      "learning_rate": 1.5538330251742286e-06,
-      "loss": 1.9073,
-      "step": 13440
+      "epoch": 1.97664,
+      "grad_norm": 46.735836029052734,
+      "learning_rate": 2.5955555555555557e-07,
+      "loss": 1.5687,
+      "step": 61770
     },
     {
-      "epoch": 0.86208,
-      "grad_norm": 3.2959301471710205,
-      "learning_rate": 1.5324989332954063e-06,
-      "loss": 1.9124,
-      "step": 13470
+      "epoch": 1.97696,
+      "grad_norm": 45.55669403076172,
+      "learning_rate": 2.56e-07,
+      "loss": 1.5991,
+      "step": 61780
     },
     {
-      "epoch": 0.864,
-      "grad_norm": 3.4026668071746826,
-      "learning_rate": 1.5111648414165838e-06,
-      "loss": 1.9026,
-      "step": 13500
+      "epoch": 1.97728,
+      "grad_norm": 45.0802001953125,
+      "learning_rate": 2.5244444444444446e-07,
+      "loss": 1.5932,
+      "step": 61790
     },
     {
-      "epoch": 0.86592,
-      "grad_norm": 3.32248854637146,
-      "learning_rate": 1.4898307495377615e-06,
-      "loss": 1.9182,
-      "step": 13530
+      "epoch": 1.9776,
+      "grad_norm": 44.832820892333984,
+      "learning_rate": 2.488888888888889e-07,
+      "loss": 1.5915,
+      "step": 61800
     },
     {
-      "epoch": 0.86784,
-      "grad_norm": 3.368107795715332,
-      "learning_rate": 1.4684966576589392e-06,
-      "loss": 1.8966,
-      "step": 13560
+      "epoch": 1.9779200000000001,
+      "grad_norm": 46.13886642456055,
+      "learning_rate": 2.4533333333333336e-07,
+      "loss": 1.6123,
+      "step": 61810
     },
     {
-      "epoch": 0.86976,
-      "grad_norm": 3.2843072414398193,
-      "learning_rate": 1.4471625657801166e-06,
-      "loss": 1.8861,
-      "step": 13590
+      "epoch": 1.97824,
+      "grad_norm": 44.888118743896484,
+      "learning_rate": 2.417777777777778e-07,
+      "loss": 1.5735,
+      "step": 61820
     },
     {
-      "epoch": 0.87168,
-      "grad_norm": 3.1779439449310303,
-      "learning_rate": 1.4258284739012943e-06,
-      "loss": 1.9177,
-      "step": 13620
+      "epoch": 1.9785599999999999,
+      "grad_norm": 46.51178741455078,
+      "learning_rate": 2.3822222222222226e-07,
+      "loss": 1.5806,
+      "step": 61830
     },
     {
-      "epoch": 0.8736,
-      "grad_norm": 3.3681070804595947,
-      "learning_rate": 1.4044943820224722e-06,
-      "loss": 1.9085,
-      "step": 13650
+      "epoch": 1.97888,
+      "grad_norm": 46.9849967956543,
+      "learning_rate": 2.3466666666666668e-07,
+      "loss": 1.6006,
+      "step": 61840
     },
     {
-      "epoch": 0.87552,
-      "grad_norm": 3.276012897491455,
-      "learning_rate": 1.3831602901436497e-06,
-      "loss": 1.885,
-      "step": 13680
+      "epoch": 1.9792,
+      "grad_norm": 45.56910705566406,
+      "learning_rate": 2.3111111111111112e-07,
+      "loss": 1.5665,
+      "step": 61850
     },
     {
-      "epoch": 0.87744,
-      "grad_norm": 3.436389446258545,
-      "learning_rate": 1.3618261982648274e-06,
-      "loss": 1.8941,
-      "step": 13710
+      "epoch": 1.97952,
+      "grad_norm": 46.16838073730469,
+      "learning_rate": 2.2755555555555557e-07,
+      "loss": 1.5849,
+      "step": 61860
     },
     {
-      "epoch": 0.87936,
-      "grad_norm": 3.2149569988250732,
-      "learning_rate": 1.3404921063860048e-06,
-      "loss": 1.9043,
-      "step": 13740
+      "epoch": 1.97984,
+      "grad_norm": 46.8780403137207,
+      "learning_rate": 2.2400000000000002e-07,
+      "loss": 1.5849,
+      "step": 61870
     },
     {
-      "epoch": 0.88128,
-      "grad_norm": 3.4241669178009033,
-      "learning_rate": 1.3191580145071825e-06,
-      "loss": 1.8985,
-      "step": 13770
+      "epoch": 1.9801600000000001,
+      "grad_norm": 46.16302490234375,
+      "learning_rate": 2.2044444444444444e-07,
+      "loss": 1.5572,
+      "step": 61880
     },
     {
-      "epoch": 0.8832,
-      "grad_norm": 3.268451452255249,
-      "learning_rate": 1.29782392262836e-06,
-      "loss": 1.8847,
-      "step": 13800
+      "epoch": 1.98048,
+      "grad_norm": 46.81636428833008,
+      "learning_rate": 2.1688888888888892e-07,
+      "loss": 1.5691,
+      "step": 61890
     },
     {
-      "epoch": 0.88512,
-      "grad_norm": 3.230989456176758,
-      "learning_rate": 1.276489830749538e-06,
-      "loss": 1.9186,
-      "step": 13830
+      "epoch": 1.9808,
+      "grad_norm": 45.55078887939453,
+      "learning_rate": 2.1333333333333334e-07,
+      "loss": 1.5925,
+      "step": 61900
     },
     {
-      "epoch": 0.88704,
-      "grad_norm": 3.310396432876587,
-      "learning_rate": 1.2551557388707156e-06,
-      "loss": 1.9179,
-      "step": 13860
+      "epoch": 1.98112,
+      "grad_norm": 46.17424011230469,
+      "learning_rate": 2.097777777777778e-07,
+      "loss": 1.606,
+      "step": 61910
     },
     {
-      "epoch": 0.88896,
-      "grad_norm": 3.3383021354675293,
-      "learning_rate": 1.233821646991893e-06,
-      "loss": 1.9001,
-      "step": 13890
+      "epoch": 1.98144,
+      "grad_norm": 46.1997184753418,
+      "learning_rate": 2.0622222222222223e-07,
+      "loss": 1.5906,
+      "step": 61920
     },
     {
-      "epoch": 0.89088,
-      "grad_norm": 3.285322904586792,
-      "learning_rate": 1.2124875551130707e-06,
-      "loss": 1.8897,
-      "step": 13920
+      "epoch": 1.98176,
+      "grad_norm": 46.45069885253906,
+      "learning_rate": 2.0266666666666668e-07,
+      "loss": 1.5905,
+      "step": 61930
     },
     {
-      "epoch": 0.8928,
-      "grad_norm": 3.306582450866699,
-      "learning_rate": 1.1911534632342484e-06,
-      "loss": 1.8756,
-      "step": 13950
+      "epoch": 1.9820799999999998,
+      "grad_norm": 47.12847137451172,
+      "learning_rate": 1.9911111111111113e-07,
+      "loss": 1.5989,
+      "step": 61940
     },
     {
-      "epoch": 0.89472,
-      "grad_norm": 3.2524266242980957,
-      "learning_rate": 1.1698193713554261e-06,
-      "loss": 1.8965,
-      "step": 13980
+      "epoch": 1.9824000000000002,
+      "grad_norm": 47.08571243286133,
+      "learning_rate": 1.9555555555555558e-07,
+      "loss": 1.5649,
+      "step": 61950
     },
     {
-      "epoch": 0.89664,
-      "grad_norm": 3.372833728790283,
-      "learning_rate": 1.1484852794766036e-06,
-      "loss": 1.9046,
-      "step": 14010
+      "epoch": 1.98272,
+      "grad_norm": 43.01755905151367,
+      "learning_rate": 1.92e-07,
+      "loss": 1.5765,
+      "step": 61960
     },
     {
-      "epoch": 0.89856,
-      "grad_norm": 3.3202505111694336,
-      "learning_rate": 1.1271511875977815e-06,
-      "loss": 1.8967,
-      "step": 14040
+      "epoch": 1.98304,
+      "grad_norm": 45.58803176879883,
+      "learning_rate": 1.8844444444444447e-07,
+      "loss": 1.5654,
+      "step": 61970
     },
     {
-      "epoch": 0.90048,
-      "grad_norm": 3.2793309688568115,
-      "learning_rate": 1.105817095718959e-06,
-      "loss": 1.8962,
-      "step": 14070
+      "epoch": 1.98336,
+      "grad_norm": 46.55001449584961,
+      "learning_rate": 1.848888888888889e-07,
+      "loss": 1.5973,
+      "step": 61980
     },
     {
-      "epoch": 0.9024,
-      "grad_norm": 3.314178943634033,
-      "learning_rate": 1.0844830038401366e-06,
-      "loss": 1.8928,
-      "step": 14100
+      "epoch": 1.98368,
+      "grad_norm": 48.73170471191406,
+      "learning_rate": 1.8133333333333337e-07,
+      "loss": 1.5939,
+      "step": 61990
     },
     {
-      "epoch": 0.90432,
-      "grad_norm": 3.432605743408203,
-      "learning_rate": 1.0631489119613141e-06,
-      "loss": 1.8927,
-      "step": 14130
+      "epoch": 1.984,
+      "grad_norm": 46.5644416809082,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 1.6406,
+      "step": 62000
     },
     {
-      "epoch": 0.90624,
-      "grad_norm": 3.2980775833129883,
-      "learning_rate": 1.041814820082492e-06,
-      "loss": 1.9031,
-      "step": 14160
+      "epoch": 1.9843199999999999,
+      "grad_norm": 47.24319076538086,
+      "learning_rate": 1.742222222222222e-07,
+      "loss": 1.5819,
+      "step": 62010
     },
     {
-      "epoch": 0.90816,
-      "grad_norm": 3.344231605529785,
-      "learning_rate": 1.0204807282036695e-06,
-      "loss": 1.9069,
-      "step": 14190
+      "epoch": 1.98464,
+      "grad_norm": 46.026180267333984,
+      "learning_rate": 1.7066666666666669e-07,
+      "loss": 1.6059,
+      "step": 62020
     },
     {
-      "epoch": 0.91008,
-      "grad_norm": 3.254805326461792,
-      "learning_rate": 9.991466363248472e-07,
-      "loss": 1.8905,
-      "step": 14220
+      "epoch": 1.98496,
+      "grad_norm": 47.73154830932617,
+      "learning_rate": 1.671111111111111e-07,
+      "loss": 1.5876,
+      "step": 62030
     },
     {
-      "epoch": 0.912,
-      "grad_norm": 3.436722993850708,
-      "learning_rate": 9.778125444460248e-07,
-      "loss": 1.8984,
-      "step": 14250
+      "epoch": 1.98528,
+      "grad_norm": 47.258453369140625,
+      "learning_rate": 1.6355555555555558e-07,
+      "loss": 1.5868,
+      "step": 62040
     },
     {
-      "epoch": 0.91392,
-      "grad_norm": 3.4136459827423096,
-      "learning_rate": 9.564784525672025e-07,
-      "loss": 1.8816,
-      "step": 14280
+      "epoch": 1.9856,
+      "grad_norm": 44.5970344543457,
+      "learning_rate": 1.6e-07,
+      "loss": 1.5916,
+      "step": 62050
     },
     {
-      "epoch": 0.91584,
-      "grad_norm": 3.34030818939209,
-      "learning_rate": 9.3514436068838e-07,
-      "loss": 1.9127,
-      "step": 14310
+      "epoch": 1.9859200000000001,
+      "grad_norm": 46.769012451171875,
+      "learning_rate": 1.5644444444444445e-07,
+      "loss": 1.5834,
+      "step": 62060
     },
     {
-      "epoch": 0.91776,
-      "grad_norm": 3.219815731048584,
-      "learning_rate": 9.138102688095578e-07,
-      "loss": 1.8991,
-      "step": 14340
+      "epoch": 1.98624,
+      "grad_norm": 46.680118560791016,
+      "learning_rate": 1.528888888888889e-07,
+      "loss": 1.5683,
+      "step": 62070
     },
     {
-      "epoch": 0.91968,
-      "grad_norm": 3.4473705291748047,
-      "learning_rate": 8.924761769307354e-07,
-      "loss": 1.8872,
-      "step": 14370
+      "epoch": 1.9865599999999999,
+      "grad_norm": 44.251956939697266,
+      "learning_rate": 1.4933333333333335e-07,
+      "loss": 1.5555,
+      "step": 62080
     },
     {
-      "epoch": 0.9216,
-      "grad_norm": 3.3151919841766357,
-      "learning_rate": 8.71142085051913e-07,
-      "loss": 1.8829,
-      "step": 14400
+      "epoch": 1.98688,
+      "grad_norm": 44.65584182739258,
+      "learning_rate": 1.457777777777778e-07,
+      "loss": 1.5926,
+      "step": 62090
     },
     {
-      "epoch": 0.92352,
-      "grad_norm": 3.348175048828125,
-      "learning_rate": 8.498079931730907e-07,
-      "loss": 1.9113,
-      "step": 14430
+      "epoch": 1.9872,
+      "grad_norm": 46.639320373535156,
+      "learning_rate": 1.4222222222222224e-07,
+      "loss": 1.5981,
+      "step": 62100
     },
     {
-      "epoch": 0.92544,
-      "grad_norm": 3.3792388439178467,
-      "learning_rate": 8.284739012942683e-07,
-      "loss": 1.8937,
-      "step": 14460
+      "epoch": 1.98752,
+      "grad_norm": 47.51844024658203,
+      "learning_rate": 1.3866666666666666e-07,
+      "loss": 1.613,
+      "step": 62110
     },
     {
-      "epoch": 0.92736,
-      "grad_norm": 3.2966620922088623,
-      "learning_rate": 8.071398094154459e-07,
-      "loss": 1.9059,
-      "step": 14490
+      "epoch": 1.98784,
+      "grad_norm": 46.52462387084961,
+      "learning_rate": 1.351111111111111e-07,
+      "loss": 1.5845,
+      "step": 62120
     },
     {
-      "epoch": 0.92928,
-      "grad_norm": 3.2808728218078613,
-      "learning_rate": 7.858057175366236e-07,
-      "loss": 1.9048,
-      "step": 14520
+      "epoch": 1.9881600000000001,
+      "grad_norm": 44.17267990112305,
+      "learning_rate": 1.3155555555555556e-07,
+      "loss": 1.5731,
+      "step": 62130
     },
     {
-      "epoch": 0.9312,
-      "grad_norm": 3.333638906478882,
-      "learning_rate": 7.644716256578013e-07,
-      "loss": 1.8945,
-      "step": 14550
+      "epoch": 1.98848,
+      "grad_norm": 45.80207824707031,
+      "learning_rate": 1.28e-07,
+      "loss": 1.5949,
+      "step": 62140
     },
     {
-      "epoch": 0.93312,
-      "grad_norm": 3.270951509475708,
-      "learning_rate": 7.431375337789788e-07,
-      "loss": 1.9099,
-      "step": 14580
+      "epoch": 1.9888,
+      "grad_norm": 47.79655456542969,
+      "learning_rate": 1.2444444444444446e-07,
+      "loss": 1.6052,
+      "step": 62150
     },
     {
-      "epoch": 0.93504,
-      "grad_norm": 3.416135311126709,
-      "learning_rate": 7.218034419001565e-07,
-      "loss": 1.9126,
-      "step": 14610
+      "epoch": 1.98912,
+      "grad_norm": 44.944766998291016,
+      "learning_rate": 1.208888888888889e-07,
+      "loss": 1.5771,
+      "step": 62160
     },
     {
-      "epoch": 0.93696,
-      "grad_norm": 3.491899013519287,
-      "learning_rate": 7.004693500213341e-07,
-      "loss": 1.8937,
-      "step": 14640
+      "epoch": 1.98944,
+      "grad_norm": 46.573734283447266,
+      "learning_rate": 1.1733333333333334e-07,
+      "loss": 1.5782,
+      "step": 62170
     },
     {
-      "epoch": 0.93888,
-      "grad_norm": 3.3975751399993896,
-      "learning_rate": 6.791352581425118e-07,
-      "loss": 1.8949,
-      "step": 14670
+      "epoch": 1.98976,
+      "grad_norm": 47.87807083129883,
+      "learning_rate": 1.1377777777777779e-07,
+      "loss": 1.5974,
+      "step": 62180
     },
     {
-      "epoch": 0.9408,
-      "grad_norm": 3.2626681327819824,
-      "learning_rate": 6.578011662636895e-07,
-      "loss": 1.9054,
-      "step": 14700
+      "epoch": 1.9900799999999998,
+      "grad_norm": 45.32151412963867,
+      "learning_rate": 1.1022222222222222e-07,
+      "loss": 1.5836,
+      "step": 62190
     },
     {
-      "epoch": 0.94272,
-      "grad_norm": 3.307912826538086,
-      "learning_rate": 6.364670743848671e-07,
-      "loss": 1.9059,
-      "step": 14730
+      "epoch": 1.9904,
+      "grad_norm": 46.42742919921875,
+      "learning_rate": 1.0666666666666667e-07,
+      "loss": 1.5909,
+      "step": 62200
     },
     {
-      "epoch": 0.94464,
-      "grad_norm": 3.2694568634033203,
-      "learning_rate": 6.151329825060447e-07,
-      "loss": 1.8621,
-      "step": 14760
+      "epoch": 1.99072,
+      "grad_norm": 48.42001724243164,
+      "learning_rate": 1.0311111111111112e-07,
+      "loss": 1.5701,
+      "step": 62210
     },
     {
-      "epoch": 0.94656,
-      "grad_norm": 3.3107099533081055,
-      "learning_rate": 5.937988906272223e-07,
-      "loss": 1.9115,
-      "step": 14790
+      "epoch": 1.99104,
+      "grad_norm": 43.873111724853516,
+      "learning_rate": 9.955555555555556e-08,
+      "loss": 1.5792,
+      "step": 62220
     },
     {
-      "epoch": 0.94848,
-      "grad_norm": 3.2878222465515137,
-      "learning_rate": 5.724647987484e-07,
-      "loss": 1.9161,
-      "step": 14820
+      "epoch": 1.99136,
+      "grad_norm": 46.204490661621094,
+      "learning_rate": 9.6e-08,
+      "loss": 1.6342,
+      "step": 62230
     },
     {
-      "epoch": 0.9504,
-      "grad_norm": 3.4034011363983154,
-      "learning_rate": 5.511307068695777e-07,
-      "loss": 1.8935,
-      "step": 14850
+      "epoch": 1.9916800000000001,
+      "grad_norm": 47.35697555541992,
+      "learning_rate": 9.244444444444445e-08,
+      "loss": 1.6296,
+      "step": 62240
     },
     {
-      "epoch": 0.95232,
-      "grad_norm": 3.319850206375122,
-      "learning_rate": 5.297966149907553e-07,
-      "loss": 1.8944,
-      "step": 14880
+      "epoch": 1.992,
+      "grad_norm": 44.651161193847656,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 1.6051,
+      "step": 62250
     },
     {
-      "epoch": 0.95424,
-      "grad_norm": 3.2024145126342773,
-      "learning_rate": 5.08462523111933e-07,
-      "loss": 1.8907,
-      "step": 14910
+      "epoch": 1.9923199999999999,
+      "grad_norm": 47.13521194458008,
+      "learning_rate": 8.533333333333334e-08,
+      "loss": 1.6059,
+      "step": 62260
     },
     {
-      "epoch": 0.95616,
-      "grad_norm": 3.376710891723633,
-      "learning_rate": 4.871284312331105e-07,
-      "loss": 1.8945,
-      "step": 14940
+      "epoch": 1.99264,
+      "grad_norm": 44.870391845703125,
+      "learning_rate": 8.177777777777779e-08,
+      "loss": 1.5809,
+      "step": 62270
     },
     {
-      "epoch": 0.95808,
-      "grad_norm": 3.3908932209014893,
-      "learning_rate": 4.6579433935428816e-07,
-      "loss": 1.8881,
-      "step": 14970
+      "epoch": 1.99296,
+      "grad_norm": 45.75646209716797,
+      "learning_rate": 7.822222222222223e-08,
+      "loss": 1.5928,
+      "step": 62280
     },
     {
-      "epoch": 0.96,
-      "grad_norm": 3.4702765941619873,
-      "learning_rate": 4.4446024747546584e-07,
-      "loss": 1.9067,
-      "step": 15000
+      "epoch": 1.99328,
+      "grad_norm": 46.337249755859375,
+      "learning_rate": 7.466666666666667e-08,
+      "loss": 1.592,
+      "step": 62290
     },
     {
-      "epoch": 0.96192,
-      "grad_norm": 3.435137987136841,
-      "learning_rate": 4.231261555966434e-07,
-      "loss": 1.8958,
-      "step": 15030
+      "epoch": 1.9936,
+      "grad_norm": 43.975128173828125,
+      "learning_rate": 7.111111111111112e-08,
+      "loss": 1.5866,
+      "step": 62300
     },
     {
-      "epoch": 0.96384,
-      "grad_norm": 3.2179274559020996,
-      "learning_rate": 4.017920637178211e-07,
-      "loss": 1.8896,
-      "step": 15060
+      "epoch": 1.9939200000000001,
+      "grad_norm": 47.980926513671875,
+      "learning_rate": 6.755555555555556e-08,
+      "loss": 1.5795,
+      "step": 62310
     },
     {
-      "epoch": 0.96576,
-      "grad_norm": 3.309631586074829,
-      "learning_rate": 3.804579718389988e-07,
-      "loss": 1.9174,
-      "step": 15090
+      "epoch": 1.99424,
+      "grad_norm": 48.362449645996094,
+      "learning_rate": 6.4e-08,
+      "loss": 1.6139,
+      "step": 62320
     },
     {
-      "epoch": 0.96768,
-      "grad_norm": 3.351106882095337,
-      "learning_rate": 3.5912387996017637e-07,
-      "loss": 1.8939,
-      "step": 15120
+      "epoch": 1.9945599999999999,
+      "grad_norm": 47.03620529174805,
+      "learning_rate": 6.044444444444445e-08,
+      "loss": 1.6054,
+      "step": 62330
     },
     {
-      "epoch": 0.9696,
-      "grad_norm": 3.3022665977478027,
-      "learning_rate": 3.37789788081354e-07,
-      "loss": 1.9005,
-      "step": 15150
+      "epoch": 1.99488,
+      "grad_norm": 47.1905632019043,
+      "learning_rate": 5.688888888888889e-08,
+      "loss": 1.591,
+      "step": 62340
     },
     {
-      "epoch": 0.97152,
-      "grad_norm": 3.4056894779205322,
-      "learning_rate": 3.164556962025317e-07,
-      "loss": 1.9181,
-      "step": 15180
+      "epoch": 1.9952,
+      "grad_norm": 45.94517517089844,
+      "learning_rate": 5.3333333333333334e-08,
+      "loss": 1.5694,
+      "step": 62350
     },
     {
-      "epoch": 0.97344,
-      "grad_norm": 3.35656476020813,
-      "learning_rate": 2.951216043237093e-07,
-      "loss": 1.9076,
-      "step": 15210
+      "epoch": 1.99552,
+      "grad_norm": 47.92856979370117,
+      "learning_rate": 4.977777777777778e-08,
+      "loss": 1.565,
+      "step": 62360
     },
     {
-      "epoch": 0.97536,
-      "grad_norm": 3.170989990234375,
-      "learning_rate": 2.7378751244488695e-07,
-      "loss": 1.8873,
-      "step": 15240
+      "epoch": 1.9958399999999998,
+      "grad_norm": 45.670406341552734,
+      "learning_rate": 4.6222222222222224e-08,
+      "loss": 1.571,
+      "step": 62370
     },
     {
-      "epoch": 0.97728,
-      "grad_norm": 3.3478634357452393,
-      "learning_rate": 2.524534205660646e-07,
-      "loss": 1.9018,
-      "step": 15270
+      "epoch": 1.9961600000000002,
+      "grad_norm": 46.5335693359375,
+      "learning_rate": 4.266666666666667e-08,
+      "loss": 1.5581,
+      "step": 62380
     },
     {
-      "epoch": 0.9792,
-      "grad_norm": 3.204563856124878,
-      "learning_rate": 2.3111932868724224e-07,
-      "loss": 1.9164,
-      "step": 15300
+      "epoch": 1.99648,
+      "grad_norm": 48.223350524902344,
+      "learning_rate": 3.911111111111111e-08,
+      "loss": 1.6012,
+      "step": 62390
     },
     {
-      "epoch": 0.98112,
-      "grad_norm": 3.293850898742676,
-      "learning_rate": 2.0978523680841987e-07,
-      "loss": 1.906,
-      "step": 15330
+      "epoch": 1.9968,
+      "grad_norm": 45.21166229248047,
+      "learning_rate": 3.555555555555556e-08,
+      "loss": 1.5798,
+      "step": 62400
     },
     {
-      "epoch": 0.98304,
-      "grad_norm": 3.27933406829834,
-      "learning_rate": 1.884511449295975e-07,
-      "loss": 1.9,
-      "step": 15360
+      "epoch": 1.99712,
+      "grad_norm": 45.345008850097656,
+      "learning_rate": 3.2e-08,
+      "loss": 1.5842,
+      "step": 62410
     },
     {
-      "epoch": 0.98496,
-      "grad_norm": 3.435851573944092,
-      "learning_rate": 1.6711705305077513e-07,
-      "loss": 1.8855,
-      "step": 15390
+      "epoch": 1.99744,
+      "grad_norm": 47.583457946777344,
+      "learning_rate": 2.8444444444444447e-08,
+      "loss": 1.5653,
+      "step": 62420
     },
     {
-      "epoch": 0.98688,
-      "grad_norm": 3.3113291263580322,
-      "learning_rate": 1.457829611719528e-07,
-      "loss": 1.8843,
-      "step": 15420
+      "epoch": 1.99776,
+      "grad_norm": 45.92597579956055,
+      "learning_rate": 2.488888888888889e-08,
+      "loss": 1.5627,
+      "step": 62430
     },
     {
-      "epoch": 0.9888,
-      "grad_norm": 3.3031797409057617,
-      "learning_rate": 1.2444886929313042e-07,
-      "loss": 1.9055,
-      "step": 15450
+      "epoch": 1.9980799999999999,
+      "grad_norm": 44.429718017578125,
+      "learning_rate": 2.1333333333333336e-08,
+      "loss": 1.5901,
+      "step": 62440
     },
     {
-      "epoch": 0.99072,
-      "grad_norm": 3.2720372676849365,
-      "learning_rate": 1.0311477741430808e-07,
-      "loss": 1.9021,
-      "step": 15480
+      "epoch": 1.9984,
+      "grad_norm": 46.34611511230469,
+      "learning_rate": 1.777777777777778e-08,
+      "loss": 1.5488,
+      "step": 62450
     },
     {
-      "epoch": 0.99264,
-      "grad_norm": 3.4496352672576904,
-      "learning_rate": 8.178068553548571e-08,
-      "loss": 1.8833,
-      "step": 15510
+      "epoch": 1.99872,
+      "grad_norm": 46.130760192871094,
+      "learning_rate": 1.4222222222222223e-08,
+      "loss": 1.5986,
+      "step": 62460
     },
     {
-      "epoch": 0.99456,
-      "grad_norm": 3.1687467098236084,
-      "learning_rate": 6.044659365666336e-08,
-      "loss": 1.8954,
-      "step": 15540
+      "epoch": 1.99904,
+      "grad_norm": 46.61756896972656,
+      "learning_rate": 1.0666666666666668e-08,
+      "loss": 1.5739,
+      "step": 62470
     },
     {
-      "epoch": 0.99648,
-      "grad_norm": 3.9161536693573,
-      "learning_rate": 3.9112501777840994e-08,
-      "loss": 1.8956,
-      "step": 15570
+      "epoch": 1.99936,
+      "grad_norm": 47.170257568359375,
+      "learning_rate": 7.111111111111112e-09,
+      "loss": 1.5921,
+      "step": 62480
     },
     {
-      "epoch": 0.9984,
-      "grad_norm": 3.415825605392456,
-      "learning_rate": 1.7778409899018632e-08,
-      "loss": 1.8921,
-      "step": 15600
+      "epoch": 1.9996800000000001,
+      "grad_norm": 48.82705307006836,
+      "learning_rate": 3.555555555555556e-09,
+      "loss": 1.5959,
+      "step": 62490
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 45.47258758544922,
+      "learning_rate": 0.0,
+      "loss": 1.5795,
+      "step": 62500
     }
   ],
-  "logging_steps": 30,
-  "max_steps": 15625,
+  "logging_steps": 10,
+  "max_steps": 62500,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
-  "save_steps": 300,
+  "num_train_epochs": 2,
+  "save_steps": 100,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
@@ -3666,7 +43776,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.6387905536e+17,
+  "total_flos": 1.05551622144e+18,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null