diff --git "a/checkpoint-1250/trainer_state.json" "b/checkpoint-1250/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1250/trainer_state.json"
@@ -0,0 +1,4408 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 1250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0032,
+      "grad_norm": 5.2346367835998535,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 5.8721,
+      "step": 2
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 6.022397994995117,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 5.775,
+      "step": 4
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 4.282025337219238,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 5.6253,
+      "step": 6
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 3.7371106147766113,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 5.1153,
+      "step": 8
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 8.722667694091797,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 5.3364,
+      "step": 10
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 4.185220718383789,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 5.3102,
+      "step": 12
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 3.5511889457702637,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 4.9739,
+      "step": 14
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 3.152355194091797,
+      "learning_rate": 2.5396825396825397e-05,
+      "loss": 4.5878,
+      "step": 16
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 2.925100326538086,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 4.7562,
+      "step": 18
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 3.192331552505493,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 4.6627,
+      "step": 20
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 2.958089590072632,
+      "learning_rate": 3.492063492063492e-05,
+      "loss": 4.6227,
+      "step": 22
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 4.2280592918396,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 4.4283,
+      "step": 24
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 2.748825788497925,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 4.2503,
+      "step": 26
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 2.705292224884033,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 4.4007,
+      "step": 28
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 2.5890371799468994,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 4.3219,
+      "step": 30
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 2.485609769821167,
+      "learning_rate": 5.0793650793650794e-05,
+      "loss": 4.2298,
+      "step": 32
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 2.5438082218170166,
+      "learning_rate": 5.396825396825397e-05,
+      "loss": 4.2862,
+      "step": 34
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 2.5402183532714844,
+      "learning_rate": 5.714285714285714e-05,
+      "loss": 4.0492,
+      "step": 36
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 2.5497469902038574,
+      "learning_rate": 6.0317460317460316e-05,
+      "loss": 4.0702,
+      "step": 38
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 2.3628146648406982,
+      "learning_rate": 6.349206349206349e-05,
+      "loss": 3.9895,
+      "step": 40
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 2.201223611831665,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 3.9743,
+      "step": 42
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 2.448514938354492,
+      "learning_rate": 6.984126984126984e-05,
+      "loss": 4.076,
+      "step": 44
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 2.386176824569702,
+      "learning_rate": 7.301587301587302e-05,
+      "loss": 3.912,
+      "step": 46
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 2.2783148288726807,
+      "learning_rate": 7.619047619047618e-05,
+      "loss": 3.8678,
+      "step": 48
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.472463369369507,
+      "learning_rate": 7.936507936507937e-05,
+      "loss": 3.9058,
+      "step": 50
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 2.548892021179199,
+      "learning_rate": 8.253968253968255e-05,
+      "loss": 3.7761,
+      "step": 52
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 2.4397833347320557,
+      "learning_rate": 8.571428571428571e-05,
+      "loss": 3.9437,
+      "step": 54
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 2.532597303390503,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 3.8437,
+      "step": 56
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 2.45221209526062,
+      "learning_rate": 9.206349206349206e-05,
+      "loss": 3.8429,
+      "step": 58
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 2.645132064819336,
+      "learning_rate": 9.523809523809524e-05,
+      "loss": 3.7965,
+      "step": 60
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 2.385370969772339,
+      "learning_rate": 9.841269841269841e-05,
+      "loss": 3.7968,
+      "step": 62
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 2.2052712440490723,
+      "learning_rate": 9.99998248790669e-05,
+      "loss": 3.7794,
+      "step": 64
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 2.3219361305236816,
+      "learning_rate": 9.999842391896222e-05,
+      "loss": 3.6841,
+      "step": 66
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 2.5465614795684814,
+      "learning_rate": 9.999562203800676e-05,
+      "loss": 3.5312,
+      "step": 68
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 2.497755527496338,
+      "learning_rate": 9.999141931470729e-05,
+      "loss": 3.6803,
+      "step": 70
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 2.157752752304077,
+      "learning_rate": 9.998581586682116e-05,
+      "loss": 3.5856,
+      "step": 72
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 2.1215410232543945,
+      "learning_rate": 9.997881185135307e-05,
+      "loss": 3.5429,
+      "step": 74
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 3.784062385559082,
+      "learning_rate": 9.997040746455062e-05,
+      "loss": 3.6365,
+      "step": 76
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 2.2366509437561035,
+      "learning_rate": 9.996060294189887e-05,
+      "loss": 3.5129,
+      "step": 78
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 2.240877628326416,
+      "learning_rate": 9.994939855811362e-05,
+      "loss": 3.6185,
+      "step": 80
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 2.176579475402832,
+      "learning_rate": 9.993679462713395e-05,
+      "loss": 3.4999,
+      "step": 82
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 2.3514060974121094,
+      "learning_rate": 9.992279150211314e-05,
+      "loss": 3.466,
+      "step": 84
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 2.2598917484283447,
+      "learning_rate": 9.990738957540896e-05,
+      "loss": 3.6413,
+      "step": 86
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 2.0476276874542236,
+      "learning_rate": 9.989058927857263e-05,
+      "loss": 3.5978,
+      "step": 88
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 2.1762521266937256,
+      "learning_rate": 9.987239108233668e-05,
+      "loss": 3.6667,
+      "step": 90
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 2.274958848953247,
+      "learning_rate": 9.985279549660185e-05,
+      "loss": 3.6054,
+      "step": 92
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 2.249992847442627,
+      "learning_rate": 9.983180307042274e-05,
+      "loss": 3.6287,
+      "step": 94
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 2.23592209815979,
+      "learning_rate": 9.980941439199246e-05,
+      "loss": 3.5967,
+      "step": 96
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 2.1270549297332764,
+      "learning_rate": 9.97856300886261e-05,
+      "loss": 3.4583,
+      "step": 98
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.140577554702759,
+      "learning_rate": 9.976045082674319e-05,
+      "loss": 3.4091,
+      "step": 100
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 2.1698827743530273,
+      "learning_rate": 9.973387731184902e-05,
+      "loss": 3.5535,
+      "step": 102
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 2.188966751098633,
+      "learning_rate": 9.97059102885149e-05,
+      "loss": 3.4673,
+      "step": 104
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 2.015054702758789,
+      "learning_rate": 9.967655054035727e-05,
+      "loss": 3.5025,
+      "step": 106
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 2.42785906791687,
+      "learning_rate": 9.964579889001569e-05,
+      "loss": 3.3789,
+      "step": 108
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 3.828245162963867,
+      "learning_rate": 9.961365619912989e-05,
+      "loss": 3.3673,
+      "step": 110
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 2.0954813957214355,
+      "learning_rate": 9.95801233683156e-05,
+      "loss": 3.501,
+      "step": 112
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 2.17081618309021,
+      "learning_rate": 9.954520133713924e-05,
+      "loss": 3.3926,
+      "step": 114
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 2.04852557182312,
+      "learning_rate": 9.950889108409172e-05,
+      "loss": 3.391,
+      "step": 116
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 2.426689386367798,
+      "learning_rate": 9.947119362656092e-05,
+      "loss": 3.4257,
+      "step": 118
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 3.680421829223633,
+      "learning_rate": 9.94321100208032e-05,
+      "loss": 3.2982,
+      "step": 120
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 2.1409482955932617,
+      "learning_rate": 9.939164136191384e-05,
+      "loss": 3.4619,
+      "step": 122
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 1.9399126768112183,
+      "learning_rate": 9.934978878379636e-05,
+      "loss": 3.3362,
+      "step": 124
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 1.954500675201416,
+      "learning_rate": 9.930655345913071e-05,
+      "loss": 3.1957,
+      "step": 126
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 2.1550300121307373,
+      "learning_rate": 9.926193659934043e-05,
+      "loss": 3.4578,
+      "step": 128
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 2.44838547706604,
+      "learning_rate": 9.921593945455869e-05,
+      "loss": 3.3975,
+      "step": 130
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 2.087881565093994,
+      "learning_rate": 9.916856331359335e-05,
+      "loss": 3.3682,
+      "step": 132
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 2.253127336502075,
+      "learning_rate": 9.911980950389067e-05,
+      "loss": 3.2451,
+      "step": 134
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 2.3103411197662354,
+      "learning_rate": 9.906967939149831e-05,
+      "loss": 3.3999,
+      "step": 136
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 2.2471373081207275,
+      "learning_rate": 9.901817438102695e-05,
+      "loss": 3.2925,
+      "step": 138
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 2.1333861351013184,
+      "learning_rate": 9.896529591561093e-05,
+      "loss": 3.3604,
+      "step": 140
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 1.914016842842102,
+      "learning_rate": 9.891104547686782e-05,
+      "loss": 3.1728,
+      "step": 142
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 2.061126232147217,
+      "learning_rate": 9.8855424584857e-05,
+      "loss": 3.2384,
+      "step": 144
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 2.153684139251709,
+      "learning_rate": 9.879843479803691e-05,
+      "loss": 3.2965,
+      "step": 146
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 2.109224557876587,
+      "learning_rate": 9.874007771322151e-05,
+      "loss": 3.2568,
+      "step": 148
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 3.588501214981079,
+      "learning_rate": 9.868035496553546e-05,
+      "loss": 3.223,
+      "step": 150
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 2.2085981369018555,
+      "learning_rate": 9.86192682283684e-05,
+      "loss": 3.3506,
+      "step": 152
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 1.9722400903701782,
+      "learning_rate": 9.855681921332793e-05,
+      "loss": 3.2184,
+      "step": 154
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 2.0837275981903076,
+      "learning_rate": 9.849300967019175e-05,
+      "loss": 3.2083,
+      "step": 156
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 1.799812912940979,
+      "learning_rate": 9.84278413868586e-05,
+      "loss": 3.2038,
+      "step": 158
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 2.0242528915405273,
+      "learning_rate": 9.836131618929819e-05,
+      "loss": 3.3193,
+      "step": 160
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 1.948160171508789,
+      "learning_rate": 9.82934359415e-05,
+      "loss": 3.2132,
+      "step": 162
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 2.1192405223846436,
+      "learning_rate": 9.822420254542108e-05,
+      "loss": 3.2444,
+      "step": 164
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 1.9564788341522217,
+      "learning_rate": 9.815361794093272e-05,
+      "loss": 3.105,
+      "step": 166
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 2.1899731159210205,
+      "learning_rate": 9.808168410576617e-05,
+      "loss": 3.1558,
+      "step": 168
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 2.0988922119140625,
+      "learning_rate": 9.800840305545715e-05,
+      "loss": 3.2485,
+      "step": 170
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 2.2193591594696045,
+      "learning_rate": 9.793377684328939e-05,
+      "loss": 3.296,
+      "step": 172
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 2.007413864135742,
+      "learning_rate": 9.785780756023714e-05,
+      "loss": 3.1287,
+      "step": 174
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 1.9749376773834229,
+      "learning_rate": 9.778049733490655e-05,
+      "loss": 3.0076,
+      "step": 176
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 2.059288263320923,
+      "learning_rate": 9.770184833347606e-05,
+      "loss": 3.1663,
+      "step": 178
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 1.96829354763031,
+      "learning_rate": 9.762186275963563e-05,
+      "loss": 3.2163,
+      "step": 180
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 2.0208017826080322,
+      "learning_rate": 9.754054285452506e-05,
+      "loss": 3.2242,
+      "step": 182
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 1.88913094997406,
+      "learning_rate": 9.745789089667121e-05,
+      "loss": 3.0072,
+      "step": 184
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 1.996383547782898,
+      "learning_rate": 9.737390920192408e-05,
+      "loss": 3.2108,
+      "step": 186
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 2.0667550563812256,
+      "learning_rate": 9.7288600123392e-05,
+      "loss": 3.108,
+      "step": 188
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 1.9526984691619873,
+      "learning_rate": 9.720196605137565e-05,
+      "loss": 3.067,
+      "step": 190
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 2.0643012523651123,
+      "learning_rate": 9.71140094133011e-05,
+      "loss": 3.2143,
+      "step": 192
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 2.187326431274414,
+      "learning_rate": 9.702473267365182e-05,
+      "loss": 3.1007,
+      "step": 194
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 2.1299145221710205,
+      "learning_rate": 9.693413833389956e-05,
+      "loss": 3.1868,
+      "step": 196
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 1.7644037008285522,
+      "learning_rate": 9.684222893243431e-05,
+      "loss": 2.9406,
+      "step": 198
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 2.112617015838623,
+      "learning_rate": 9.674900704449324e-05,
+      "loss": 3.1198,
+      "step": 200
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 1.8327059745788574,
+      "learning_rate": 9.665447528208836e-05,
+      "loss": 3.1278,
+      "step": 202
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 1.8569375276565552,
+      "learning_rate": 9.655863629393351e-05,
+      "loss": 3.2069,
+      "step": 204
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 1.7960104942321777,
+      "learning_rate": 9.64614927653701e-05,
+      "loss": 3.0708,
+      "step": 206
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 1.888593316078186,
+      "learning_rate": 9.636304741829181e-05,
+      "loss": 3.1365,
+      "step": 208
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 1.8564034700393677,
+      "learning_rate": 9.626330301106837e-05,
+      "loss": 3.0059,
+      "step": 210
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 1.9591517448425293,
+      "learning_rate": 9.616226233846828e-05,
+      "loss": 2.9778,
+      "step": 212
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 2.025777816772461,
+      "learning_rate": 9.605992823158046e-05,
+      "loss": 3.0969,
+      "step": 214
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 1.8839352130889893,
+      "learning_rate": 9.595630355773501e-05,
+      "loss": 3.1342,
+      "step": 216
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 5.388115882873535,
+      "learning_rate": 9.585139122042274e-05,
+      "loss": 3.1961,
+      "step": 218
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 2.056678533554077,
+      "learning_rate": 9.574519415921396e-05,
+      "loss": 3.1183,
+      "step": 220
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 3.0575530529022217,
+      "learning_rate": 9.5637715349676e-05,
+      "loss": 3.1446,
+      "step": 222
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 1.8165247440338135,
+      "learning_rate": 9.552895780328987e-05,
+      "loss": 3.0338,
+      "step": 224
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 1.845023512840271,
+      "learning_rate": 9.541892456736595e-05,
+      "loss": 3.194,
+      "step": 226
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 1.9389755725860596,
+      "learning_rate": 9.530761872495849e-05,
+      "loss": 3.0054,
+      "step": 228
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 1.9471769332885742,
+      "learning_rate": 9.519504339477932e-05,
+      "loss": 3.1499,
+      "step": 230
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 1.9367070198059082,
+      "learning_rate": 9.508120173111039e-05,
+      "loss": 3.0068,
+      "step": 232
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 2.018630027770996,
+      "learning_rate": 9.496609692371548e-05,
+      "loss": 3.1722,
+      "step": 234
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 2.0086734294891357,
+      "learning_rate": 9.484973219775074e-05,
+      "loss": 3.2773,
+      "step": 236
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 1.9771322011947632,
+      "learning_rate": 9.473211081367436e-05,
+      "loss": 3.0502,
+      "step": 238
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 1.9231762886047363,
+      "learning_rate": 9.46132360671552e-05,
+      "loss": 3.0415,
+      "step": 240
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 1.924302101135254,
+      "learning_rate": 9.449311128898049e-05,
+      "loss": 3.0794,
+      "step": 242
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 1.967323899269104,
+      "learning_rate": 9.437173984496246e-05,
+      "loss": 3.1527,
+      "step": 244
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 1.9681285619735718,
+      "learning_rate": 9.424912513584401e-05,
+      "loss": 3.1767,
+      "step": 246
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 1.908687710762024,
+      "learning_rate": 9.412527059720352e-05,
+      "loss": 2.9755,
+      "step": 248
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.9891773462295532,
+      "learning_rate": 9.400017969935848e-05,
+      "loss": 2.9644,
+      "step": 250
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 1.8200337886810303,
+      "learning_rate": 9.387385594726829e-05,
+      "loss": 3.0334,
+      "step": 252
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 1.9553104639053345,
+      "learning_rate": 9.374630288043614e-05,
+      "loss": 3.1011,
+      "step": 254
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 3.5282905101776123,
+      "learning_rate": 9.361752407280965e-05,
+      "loss": 2.9859,
+      "step": 256
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 1.9712797403335571,
+      "learning_rate": 9.348752313268093e-05,
+      "loss": 2.9472,
+      "step": 258
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 1.927635908126831,
+      "learning_rate": 9.335630370258533e-05,
+      "loss": 3.1396,
+      "step": 260
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 1.9233123064041138,
+      "learning_rate": 9.322386945919946e-05,
+      "loss": 3.1889,
+      "step": 262
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 1.870160460472107,
+      "learning_rate": 9.309022411323816e-05,
+      "loss": 3.0916,
+      "step": 264
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 1.7860538959503174,
+      "learning_rate": 9.295537140935049e-05,
+      "loss": 3.1584,
+      "step": 266
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 1.723097801208496,
+      "learning_rate": 9.281931512601485e-05,
+      "loss": 2.8587,
+      "step": 268
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 1.976706862449646,
+      "learning_rate": 9.26820590754331e-05,
+      "loss": 2.9942,
+      "step": 270
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 1.8147152662277222,
+      "learning_rate": 9.254360710342371e-05,
+      "loss": 3.087,
+      "step": 272
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 1.8946576118469238,
+      "learning_rate": 9.240396308931407e-05,
+      "loss": 3.0101,
+      "step": 274
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 1.8432953357696533,
+      "learning_rate": 9.226313094583173e-05,
+      "loss": 3.0351,
+      "step": 276
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 1.8600575923919678,
+      "learning_rate": 9.212111461899479e-05,
+      "loss": 3.0027,
+      "step": 278
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 1.7912688255310059,
+      "learning_rate": 9.197791808800135e-05,
+      "loss": 3.0568,
+      "step": 280
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 2.005932569503784,
+      "learning_rate": 9.183354536511803e-05,
+      "loss": 2.9778,
+      "step": 282
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 1.8989531993865967,
+      "learning_rate": 9.168800049556747e-05,
+      "loss": 2.9711,
+      "step": 284
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 1.7888331413269043,
+      "learning_rate": 9.154128755741509e-05,
+      "loss": 2.9901,
+      "step": 286
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 1.9094816446304321,
+      "learning_rate": 9.139341066145472e-05,
+      "loss": 3.0248,
+      "step": 288
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 1.7940737009048462,
+      "learning_rate": 9.124437395109353e-05,
+      "loss": 3.0141,
+      "step": 290
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 1.7626845836639404,
+      "learning_rate": 9.109418160223585e-05,
+      "loss": 2.9531,
+      "step": 292
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 1.9440515041351318,
+      "learning_rate": 9.094283782316619e-05,
+      "loss": 2.9732,
+      "step": 294
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 1.7515082359313965,
+      "learning_rate": 9.079034685443133e-05,
+      "loss": 2.8,
+      "step": 296
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 1.8595532178878784,
+      "learning_rate": 9.063671296872149e-05,
+      "loss": 2.9873,
+      "step": 298
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.9954842329025269,
+      "learning_rate": 9.048194047075069e-05,
+      "loss": 2.9793,
+      "step": 300
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 1.8819364309310913,
+      "learning_rate": 9.032603369713596e-05,
+      "loss": 2.8904,
+      "step": 302
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 1.75027596950531,
+      "learning_rate": 9.016899701627604e-05,
+      "loss": 2.9811,
+      "step": 304
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 1.9617975950241089,
+      "learning_rate": 9.00108348282288e-05,
+      "loss": 3.0418,
+      "step": 306
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 1.8097938299179077,
+      "learning_rate": 8.985155156458811e-05,
+      "loss": 3.0068,
+      "step": 308
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 2.008989095687866,
+      "learning_rate": 8.969115168835954e-05,
+      "loss": 2.8913,
+      "step": 310
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 1.8119149208068848,
+      "learning_rate": 8.952963969383538e-05,
+      "loss": 3.004,
+      "step": 312
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 1.9150359630584717,
+      "learning_rate": 8.93670201064687e-05,
+      "loss": 2.9404,
+      "step": 314
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 1.8803378343582153,
+      "learning_rate": 8.920329748274649e-05,
+      "loss": 2.9279,
+      "step": 316
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 1.7308014631271362,
+      "learning_rate": 8.903847641006218e-05,
+      "loss": 2.9482,
+      "step": 318
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 2.0764575004577637,
+      "learning_rate": 8.887256150658684e-05,
+      "loss": 2.9425,
+      "step": 320
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 1.8219692707061768,
+      "learning_rate": 8.870555742113998e-05,
+      "loss": 3.0176,
+      "step": 322
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 1.8642444610595703,
+      "learning_rate": 8.85374688330592e-05,
+      "loss": 3.0474,
+      "step": 324
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 1.9277112483978271,
+      "learning_rate": 8.836830045206911e-05,
+      "loss": 2.9487,
+      "step": 326
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 1.793232798576355,
+      "learning_rate": 8.81980570181494e-05,
+      "loss": 2.8907,
+      "step": 328
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 1.8513329029083252,
+      "learning_rate": 8.802674330140192e-05,
+      "loss": 2.9645,
+      "step": 330
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 1.7978984117507935,
+      "learning_rate": 8.785436410191714e-05,
+      "loss": 2.9939,
+      "step": 332
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 1.7157683372497559,
+      "learning_rate": 8.76809242496396e-05,
+      "loss": 2.8079,
+      "step": 334
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 1.7450584173202515,
+      "learning_rate": 8.750642860423262e-05,
+      "loss": 2.9477,
+      "step": 336
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 1.812904715538025,
+      "learning_rate": 8.733088205494205e-05,
+      "loss": 2.9842,
+      "step": 338
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 1.878509759902954,
+      "learning_rate": 8.715428952045936e-05,
+      "loss": 2.8992,
+      "step": 340
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 1.8090356588363647,
+      "learning_rate": 8.697665594878382e-05,
+      "loss": 2.9507,
+      "step": 342
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 1.8601405620574951,
+      "learning_rate": 8.679798631708375e-05,
+      "loss": 2.8263,
+      "step": 344
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 1.8406038284301758,
+      "learning_rate": 8.661828563155727e-05,
+      "loss": 2.8991,
+      "step": 346
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 1.7687346935272217,
+      "learning_rate": 8.643755892729179e-05,
+      "loss": 2.8437,
+      "step": 348
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.9318656921386719,
+      "learning_rate": 8.625581126812312e-05,
+      "loss": 3.0261,
+      "step": 350
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 1.8095970153808594,
+      "learning_rate": 8.607304774649349e-05,
+      "loss": 2.9269,
+      "step": 352
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 1.8424136638641357,
+      "learning_rate": 8.588927348330887e-05,
+      "loss": 2.7917,
+      "step": 354
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 1.9472522735595703,
+      "learning_rate": 8.57044936277955e-05,
+      "loss": 2.7557,
+      "step": 356
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 1.8774663209915161,
+      "learning_rate": 8.551871335735565e-05,
+      "loss": 2.8426,
+      "step": 358
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 2.0363616943359375,
+      "learning_rate": 8.533193787742251e-05,
+      "loss": 2.8605,
+      "step": 360
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 1.802148699760437,
+      "learning_rate": 8.51441724213143e-05,
+      "loss": 2.8999,
+      "step": 362
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 1.9002522230148315,
+      "learning_rate": 8.495542225008771e-05,
+      "loss": 2.854,
+      "step": 364
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 1.715409755706787,
+      "learning_rate": 8.476569265239046e-05,
+      "loss": 2.8574,
+      "step": 366
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 1.9336328506469727,
+      "learning_rate": 8.457498894431311e-05,
+      "loss": 2.7513,
+      "step": 368
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 1.8150614500045776,
+      "learning_rate": 8.438331646924013e-05,
+      "loss": 2.8648,
+      "step": 370
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 1.745450735092163,
+      "learning_rate": 8.419068059770011e-05,
+      "loss": 2.8102,
+      "step": 372
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 1.7649202346801758,
+      "learning_rate": 8.399708672721539e-05,
+      "loss": 2.943,
+      "step": 374
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 1.9029461145401,
+      "learning_rate": 8.380254028215076e-05,
+      "loss": 2.9549,
+      "step": 376
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 1.6569948196411133,
+      "learning_rate": 8.360704671356145e-05,
+      "loss": 2.762,
+      "step": 378
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 1.8082654476165771,
+      "learning_rate": 8.341061149904045e-05,
+      "loss": 2.8673,
+      "step": 380
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 2.0130746364593506,
+      "learning_rate": 8.321324014256504e-05,
+      "loss": 2.8208,
+      "step": 382
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 1.9243208169937134,
+      "learning_rate": 8.30149381743425e-05,
+      "loss": 2.8175,
+      "step": 384
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 1.7602218389511108,
+      "learning_rate": 8.28157111506552e-05,
+      "loss": 2.8133,
+      "step": 386
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 1.9011287689208984,
+      "learning_rate": 8.261556465370493e-05,
+      "loss": 2.915,
+      "step": 388
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 1.8260935544967651,
+      "learning_rate": 8.24145042914565e-05,
+      "loss": 2.7879,
+      "step": 390
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 2.0104498863220215,
+      "learning_rate": 8.221253569748055e-05,
+      "loss": 2.9628,
+      "step": 392
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 1.821583867073059,
+      "learning_rate": 8.200966453079575e-05,
+      "loss": 2.8391,
+      "step": 394
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 1.8291980028152466,
+      "learning_rate": 8.180589647571023e-05,
+      "loss": 2.8921,
+      "step": 396
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 1.8733659982681274,
+      "learning_rate": 8.16012372416623e-05,
+      "loss": 2.8988,
+      "step": 398
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.9064126014709473,
+      "learning_rate": 8.13956925630605e-05,
+      "loss": 2.8047,
+      "step": 400
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 1.7694967985153198,
+      "learning_rate": 8.118926819912287e-05,
+      "loss": 2.8175,
+      "step": 402
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 1.9278019666671753,
+      "learning_rate": 8.098196993371565e-05,
+      "loss": 2.7233,
+      "step": 404
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 1.7496165037155151,
+      "learning_rate": 8.077380357519115e-05,
+      "loss": 2.8196,
+      "step": 406
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 1.8212895393371582,
+      "learning_rate": 8.056477495622511e-05,
+      "loss": 2.8749,
+      "step": 408
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 1.8443467617034912,
+      "learning_rate": 8.035488993365312e-05,
+      "loss": 2.8373,
+      "step": 410
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 1.804992437362671,
+      "learning_rate": 8.014415438830667e-05,
+      "loss": 2.8248,
+      "step": 412
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 1.7744520902633667,
+      "learning_rate": 7.993257422484826e-05,
+      "loss": 2.8403,
+      "step": 414
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 1.9854934215545654,
+      "learning_rate": 7.972015537160602e-05,
+      "loss": 2.9588,
+      "step": 416
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 1.7830610275268555,
+      "learning_rate": 7.950690378040758e-05,
+      "loss": 2.7737,
+      "step": 418
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 1.8114231824874878,
+      "learning_rate": 7.929282542641325e-05,
+      "loss": 2.7003,
+      "step": 420
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 1.931700348854065,
+      "learning_rate": 7.907792630794876e-05,
+      "loss": 2.8088,
+      "step": 422
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 1.8684518337249756,
+      "learning_rate": 7.886221244633703e-05,
+      "loss": 2.878,
+      "step": 424
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 1.8997987508773804,
+      "learning_rate": 7.864568988572947e-05,
+      "loss": 2.9374,
+      "step": 426
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 1.7682809829711914,
+      "learning_rate": 7.842836469293673e-05,
+      "loss": 2.7694,
+      "step": 428
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 1.8019146919250488,
+      "learning_rate": 7.821024295725865e-05,
+      "loss": 2.8153,
+      "step": 430
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 1.8119292259216309,
+      "learning_rate": 7.79913307903136e-05,
+      "loss": 2.8072,
+      "step": 432
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 1.8016608953475952,
+      "learning_rate": 7.777163432586734e-05,
+      "loss": 2.7276,
+      "step": 434
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 1.8160144090652466,
+      "learning_rate": 7.755115971966104e-05,
+      "loss": 2.8539,
+      "step": 436
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 1.825020670890808,
+      "learning_rate": 7.732991314923891e-05,
+      "loss": 2.7796,
+      "step": 438
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 1.7632222175598145,
+      "learning_rate": 7.710790081377502e-05,
+      "loss": 2.7914,
+      "step": 440
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 1.8498951196670532,
+      "learning_rate": 7.688512893389964e-05,
+      "loss": 2.6861,
+      "step": 442
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 2.0129451751708984,
+      "learning_rate": 7.666160375152496e-05,
+      "loss": 2.8217,
+      "step": 444
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 1.780062198638916,
+      "learning_rate": 7.643733152967019e-05,
+      "loss": 2.8554,
+      "step": 446
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 1.726577877998352,
+      "learning_rate": 7.621231855228604e-05,
+      "loss": 2.831,
+      "step": 448
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.7651227712631226,
+      "learning_rate": 7.598657112407865e-05,
+      "loss": 2.8193,
+      "step": 450
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 1.7961740493774414,
+      "learning_rate": 7.576009557033304e-05,
+      "loss": 2.8149,
+      "step": 452
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 1.813366174697876,
+      "learning_rate": 7.553289823673568e-05,
+      "loss": 2.855,
+      "step": 454
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 1.8143000602722168,
+      "learning_rate": 7.530498548919693e-05,
+      "loss": 2.8651,
+      "step": 456
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 1.7585805654525757,
+      "learning_rate": 7.507636371367246e-05,
+      "loss": 3.0031,
+      "step": 458
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 2.2777135372161865,
+      "learning_rate": 7.484703931598445e-05,
+      "loss": 2.8548,
+      "step": 460
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 1.8288154602050781,
+      "learning_rate": 7.461701872164204e-05,
+      "loss": 2.7425,
+      "step": 462
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 1.8734841346740723,
+      "learning_rate": 7.438630837566133e-05,
+      "loss": 2.8703,
+      "step": 464
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 1.742242455482483,
+      "learning_rate": 7.415491474238475e-05,
+      "loss": 2.772,
+      "step": 466
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 1.7874287366867065,
+      "learning_rate": 7.39228443053e-05,
+      "loss": 2.7379,
+      "step": 468
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 1.8021794557571411,
+      "learning_rate": 7.369010356685833e-05,
+      "loss": 2.9262,
+      "step": 470
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 1.7524378299713135,
+      "learning_rate": 7.345669904829237e-05,
+      "loss": 2.6676,
+      "step": 472
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 1.8277724981307983,
+      "learning_rate": 7.32226372894334e-05,
+      "loss": 2.771,
+      "step": 474
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 1.7385722398757935,
+      "learning_rate": 7.298792484852808e-05,
+      "loss": 2.7508,
+      "step": 476
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 1.927331805229187,
+      "learning_rate": 7.27525683020548e-05,
+      "loss": 2.8279,
+      "step": 478
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 1.7165371179580688,
+      "learning_rate": 7.251657424453928e-05,
+      "loss": 2.6799,
+      "step": 480
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 1.6585590839385986,
+      "learning_rate": 7.227994928836988e-05,
+      "loss": 2.6849,
+      "step": 482
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 1.7066259384155273,
+      "learning_rate": 7.204270006361228e-05,
+      "loss": 2.7189,
+      "step": 484
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 1.8811277151107788,
+      "learning_rate": 7.180483321782374e-05,
+      "loss": 2.7771,
+      "step": 486
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 1.790667176246643,
+      "learning_rate": 7.156635541586682e-05,
+      "loss": 2.6777,
+      "step": 488
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 1.7074140310287476,
+      "learning_rate": 7.132727333972265e-05,
+      "loss": 2.7974,
+      "step": 490
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 1.6692975759506226,
+      "learning_rate": 7.108759368830371e-05,
+      "loss": 2.7194,
+      "step": 492
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 1.759386420249939,
+      "learning_rate": 7.084732317726611e-05,
+      "loss": 2.7475,
+      "step": 494
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 1.7140787839889526,
+      "learning_rate": 7.060646853882145e-05,
+      "loss": 2.7576,
+      "step": 496
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 1.6590884923934937,
+      "learning_rate": 7.036503652154812e-05,
+      "loss": 2.8157,
+      "step": 498
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.7052589654922485,
+      "learning_rate": 7.012303389020234e-05,
+      "loss": 2.7951,
+      "step": 500
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 1.730635166168213,
+      "learning_rate": 6.988046742552845e-05,
+      "loss": 2.8279,
+      "step": 502
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 1.7786180973052979,
+      "learning_rate": 6.963734392406907e-05,
+      "loss": 2.6559,
+      "step": 504
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 1.901053547859192,
+      "learning_rate": 6.93936701979746e-05,
+      "loss": 2.8896,
+      "step": 506
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 1.7321664094924927,
+      "learning_rate": 6.914945307481228e-05,
+      "loss": 2.795,
+      "step": 508
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 1.7901755571365356,
+      "learning_rate": 6.890469939737506e-05,
+      "loss": 2.7142,
+      "step": 510
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 1.7946327924728394,
+      "learning_rate": 6.865941602348966e-05,
+      "loss": 2.781,
+      "step": 512
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 1.8260494470596313,
+      "learning_rate": 6.841360982582463e-05,
+      "loss": 2.6868,
+      "step": 514
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 1.8089832067489624,
+      "learning_rate": 6.816728769169757e-05,
+      "loss": 2.7845,
+      "step": 516
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 1.711962342262268,
+      "learning_rate": 6.792045652288234e-05,
+      "loss": 2.7037,
+      "step": 518
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 1.7422336339950562,
+      "learning_rate": 6.767312323541555e-05,
+      "loss": 2.7938,
+      "step": 520
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 1.7964292764663696,
+      "learning_rate": 6.742529475940284e-05,
+      "loss": 2.6584,
+      "step": 522
+    },
+    {
+      "epoch": 0.8384,
+      "grad_norm": 1.7422146797180176,
+      "learning_rate": 6.717697803882467e-05,
+      "loss": 2.735,
+      "step": 524
+    },
+    {
+      "epoch": 0.8416,
+      "grad_norm": 1.7856199741363525,
+      "learning_rate": 6.692818003134184e-05,
+      "loss": 2.7753,
+      "step": 526
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 1.7519943714141846,
+      "learning_rate": 6.667890770810035e-05,
+      "loss": 2.7173,
+      "step": 528
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 1.705423355102539,
+      "learning_rate": 6.64291680535363e-05,
+      "loss": 2.7212,
+      "step": 530
+    },
+    {
+      "epoch": 0.8512,
+      "grad_norm": 1.787747859954834,
+      "learning_rate": 6.617896806518005e-05,
+      "loss": 2.7499,
+      "step": 532
+    },
+    {
+      "epoch": 0.8544,
+      "grad_norm": 1.652585744857788,
+      "learning_rate": 6.592831475346018e-05,
+      "loss": 2.5542,
+      "step": 534
+    },
+    {
+      "epoch": 0.8576,
+      "grad_norm": 1.8116321563720703,
+      "learning_rate": 6.56772151415071e-05,
+      "loss": 2.8155,
+      "step": 536
+    },
+    {
+      "epoch": 0.8608,
+      "grad_norm": 1.7901153564453125,
+      "learning_rate": 6.542567626495619e-05,
+      "loss": 2.7472,
+      "step": 538
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 1.7034342288970947,
+      "learning_rate": 6.517370517175081e-05,
+      "loss": 2.7116,
+      "step": 540
+    },
+    {
+      "epoch": 0.8672,
+      "grad_norm": 1.832322597503662,
+      "learning_rate": 6.492130892194461e-05,
+      "loss": 2.7618,
+      "step": 542
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 1.7125661373138428,
+      "learning_rate": 6.466849458750394e-05,
+      "loss": 2.6383,
+      "step": 544
+    },
+    {
+      "epoch": 0.8736,
+      "grad_norm": 1.801355004310608,
+      "learning_rate": 6.441526925210949e-05,
+      "loss": 2.5274,
+      "step": 546
+    },
+    {
+      "epoch": 0.8768,
+      "grad_norm": 1.7398046255111694,
+      "learning_rate": 6.416164001095799e-05,
+      "loss": 2.7207,
+      "step": 548
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.6901566982269287,
+      "learning_rate": 6.390761397056328e-05,
+      "loss": 2.7043,
+      "step": 550
+    },
+    {
+      "epoch": 0.8832,
+      "grad_norm": 1.6404509544372559,
+      "learning_rate": 6.365319824855727e-05,
+      "loss": 2.6003,
+      "step": 552
+    },
+    {
+      "epoch": 0.8864,
+      "grad_norm": 1.6571403741836548,
+      "learning_rate": 6.339839997349045e-05,
+      "loss": 2.7515,
+      "step": 554
+    },
+    {
+      "epoch": 0.8896,
+      "grad_norm": 1.64620041847229,
+      "learning_rate": 6.314322628463219e-05,
+      "loss": 2.619,
+      "step": 556
+    },
+    {
+      "epoch": 0.8928,
+      "grad_norm": 1.7471263408660889,
+      "learning_rate": 6.288768433177068e-05,
+      "loss": 2.6689,
+      "step": 558
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 1.7717193365097046,
+      "learning_rate": 6.26317812750126e-05,
+      "loss": 2.7607,
+      "step": 560
+    },
+    {
+      "epoch": 0.8992,
+      "grad_norm": 1.6764250993728638,
+      "learning_rate": 6.237552428458256e-05,
+      "loss": 2.6201,
+      "step": 562
+    },
+    {
+      "epoch": 0.9024,
+      "grad_norm": 1.6406267881393433,
+      "learning_rate": 6.21189205406221e-05,
+      "loss": 2.7044,
+      "step": 564
+    },
+    {
+      "epoch": 0.9056,
+      "grad_norm": 1.7228118181228638,
+      "learning_rate": 6.186197723298855e-05,
+      "loss": 2.8106,
+      "step": 566
+    },
+    {
+      "epoch": 0.9088,
+      "grad_norm": 1.6297177076339722,
+      "learning_rate": 6.160470156105362e-05,
+      "loss": 2.7442,
+      "step": 568
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 1.691156268119812,
+      "learning_rate": 6.134710073350156e-05,
+      "loss": 2.6915,
+      "step": 570
+    },
+    {
+      "epoch": 0.9152,
+      "grad_norm": 1.7338085174560547,
+      "learning_rate": 6.108918196812734e-05,
+      "loss": 2.6418,
+      "step": 572
+    },
+    {
+      "epoch": 0.9184,
+      "grad_norm": 1.7589360475540161,
+      "learning_rate": 6.083095249163424e-05,
+      "loss": 2.7577,
+      "step": 574
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 1.6277837753295898,
+      "learning_rate": 6.057241953943154e-05,
+      "loss": 2.6624,
+      "step": 576
+    },
+    {
+      "epoch": 0.9248,
+      "grad_norm": 1.7026268243789673,
+      "learning_rate": 6.031359035543158e-05,
+      "loss": 2.6401,
+      "step": 578
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 1.781171202659607,
+      "learning_rate": 6.005447219184702e-05,
+      "loss": 2.7218,
+      "step": 580
+    },
+    {
+      "epoch": 0.9312,
+      "grad_norm": 1.7018693685531616,
+      "learning_rate": 5.9795072308987485e-05,
+      "loss": 2.5756,
+      "step": 582
+    },
+    {
+      "epoch": 0.9344,
+      "grad_norm": 1.710750699043274,
+      "learning_rate": 5.9535397975056154e-05,
+      "loss": 2.738,
+      "step": 584
+    },
+    {
+      "epoch": 0.9376,
+      "grad_norm": 1.7480794191360474,
+      "learning_rate": 5.927545646594617e-05,
+      "loss": 2.5714,
+      "step": 586
+    },
+    {
+      "epoch": 0.9408,
+      "grad_norm": 1.7700692415237427,
+      "learning_rate": 5.901525506503668e-05,
+      "loss": 2.8079,
+      "step": 588
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 1.7823493480682373,
+      "learning_rate": 5.87548010629889e-05,
+      "loss": 2.6776,
+      "step": 590
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 1.6947530508041382,
+      "learning_rate": 5.8494101757541676e-05,
+      "loss": 2.6956,
+      "step": 592
+    },
+    {
+      "epoch": 0.9504,
+      "grad_norm": 3.6984357833862305,
+      "learning_rate": 5.8233164453307156e-05,
+      "loss": 2.7138,
+      "step": 594
+    },
+    {
+      "epoch": 0.9536,
+      "grad_norm": 1.7368056774139404,
+      "learning_rate": 5.797199646156596e-05,
+      "loss": 2.6646,
+      "step": 596
+    },
+    {
+      "epoch": 0.9568,
+      "grad_norm": 1.7944828271865845,
+      "learning_rate": 5.7710605100062485e-05,
+      "loss": 2.7675,
+      "step": 598
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.6918281316757202,
+      "learning_rate": 5.7448997692799764e-05,
+      "loss": 2.5737,
+      "step": 600
+    },
+    {
+      "epoch": 0.9632,
+      "grad_norm": 1.6261357069015503,
+      "learning_rate": 5.718718156983428e-05,
+      "loss": 2.6307,
+      "step": 602
+    },
+    {
+      "epoch": 0.9664,
+      "grad_norm": 1.7582976818084717,
+      "learning_rate": 5.69251640670706e-05,
+      "loss": 2.6687,
+      "step": 604
+    },
+    {
+      "epoch": 0.9696,
+      "grad_norm": 1.6695537567138672,
+      "learning_rate": 5.6662952526055793e-05,
+      "loss": 2.5779,
+      "step": 606
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 1.7661280632019043,
+      "learning_rate": 5.6400554293773744e-05,
+      "loss": 2.7142,
+      "step": 608
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 1.8265255689620972,
+      "learning_rate": 5.61379767224393e-05,
+      "loss": 2.749,
+      "step": 610
+    },
+    {
+      "epoch": 0.9792,
+      "grad_norm": 1.6744146347045898,
+      "learning_rate": 5.587522716929228e-05,
+      "loss": 2.5499,
+      "step": 612
+    },
+    {
+      "epoch": 0.9824,
+      "grad_norm": 1.6788204908370972,
+      "learning_rate": 5.561231299639127e-05,
+      "loss": 2.7511,
+      "step": 614
+    },
+    {
+      "epoch": 0.9856,
+      "grad_norm": 1.8337587118148804,
+      "learning_rate": 5.534924157040745e-05,
+      "loss": 2.7018,
+      "step": 616
+    },
+    {
+      "epoch": 0.9888,
+      "grad_norm": 1.706388235092163,
+      "learning_rate": 5.508602026241807e-05,
+      "loss": 2.6037,
+      "step": 618
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 1.699271559715271,
+      "learning_rate": 5.482265644769998e-05,
+      "loss": 2.7222,
+      "step": 620
+    },
+    {
+      "epoch": 0.9952,
+      "grad_norm": 1.67972993850708,
+      "learning_rate": 5.4559157505522985e-05,
+      "loss": 2.4653,
+      "step": 622
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 1.6633968353271484,
+      "learning_rate": 5.429553081894304e-05,
+      "loss": 2.6239,
+      "step": 624
+    },
+    {
+      "epoch": 1.0016,
+      "grad_norm": 1.6871411800384521,
+      "learning_rate": 5.4031783774595455e-05,
+      "loss": 2.4794,
+      "step": 626
+    },
+    {
+      "epoch": 1.0048,
+      "grad_norm": 1.5942273139953613,
+      "learning_rate": 5.3767923762487824e-05,
+      "loss": 2.1013,
+      "step": 628
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 1.6011604070663452,
+      "learning_rate": 5.3503958175793055e-05,
+      "loss": 2.139,
+      "step": 630
+    },
+    {
+      "epoch": 1.0112,
+      "grad_norm": 1.642553687095642,
+      "learning_rate": 5.323989441064216e-05,
+      "loss": 2.0493,
+      "step": 632
+    },
+    {
+      "epoch": 1.0144,
+      "grad_norm": 1.7297475337982178,
+      "learning_rate": 5.2975739865917074e-05,
+      "loss": 1.9964,
+      "step": 634
+    },
+    {
+      "epoch": 1.0176,
+      "grad_norm": 1.7036877870559692,
+      "learning_rate": 5.271150194304326e-05,
+      "loss": 2.0491,
+      "step": 636
+    },
+    {
+      "epoch": 1.0208,
+      "grad_norm": 1.6151326894760132,
+      "learning_rate": 5.244718804578246e-05,
+      "loss": 2.1586,
+      "step": 638
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 1.6182959079742432,
+      "learning_rate": 5.218280558002506e-05,
+      "loss": 2.0155,
+      "step": 640
+    },
+    {
+      "epoch": 1.0272,
+      "grad_norm": 1.5469590425491333,
+      "learning_rate": 5.191836195358278e-05,
+      "loss": 2.0697,
+      "step": 642
+    },
+    {
+      "epoch": 1.0304,
+      "grad_norm": 1.574875831604004,
+      "learning_rate": 5.165386457598099e-05,
+      "loss": 2.0498,
+      "step": 644
+    },
+    {
+      "epoch": 1.0336,
+      "grad_norm": 1.6611751317977905,
+      "learning_rate": 5.13893208582511e-05,
+      "loss": 2.0674,
+      "step": 646
+    },
+    {
+      "epoch": 1.0368,
+      "grad_norm": 1.622523546218872,
+      "learning_rate": 5.1124738212722966e-05,
+      "loss": 2.1196,
+      "step": 648
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.5410690307617188,
+      "learning_rate": 5.086012405281717e-05,
+      "loss": 2.1537,
+      "step": 650
+    },
+    {
+      "epoch": 1.0432,
+      "grad_norm": 1.564372181892395,
+      "learning_rate": 5.0595485792837305e-05,
+      "loss": 1.9727,
+      "step": 652
+    },
+    {
+      "epoch": 1.0464,
+      "grad_norm": 1.5646798610687256,
+      "learning_rate": 5.033083084776222e-05,
+      "loss": 2.1061,
+      "step": 654
+    },
+    {
+      "epoch": 1.0496,
+      "grad_norm": 1.661833643913269,
+      "learning_rate": 5.0066166633038305e-05,
+      "loss": 2.075,
+      "step": 656
+    },
+    {
+      "epoch": 1.0528,
+      "grad_norm": 1.6063848733901978,
+      "learning_rate": 4.980150056437163e-05,
+      "loss": 2.0384,
+      "step": 658
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 1.7570315599441528,
+      "learning_rate": 4.9536840057520224e-05,
+      "loss": 2.1477,
+      "step": 660
+    },
+    {
+      "epoch": 1.0592,
+      "grad_norm": 1.6103792190551758,
+      "learning_rate": 4.927219252808631e-05,
+      "loss": 2.025,
+      "step": 662
+    },
+    {
+      "epoch": 1.0624,
+      "grad_norm": 1.6376656293869019,
+      "learning_rate": 4.900756539130846e-05,
+      "loss": 2.0121,
+      "step": 664
+    },
+    {
+      "epoch": 1.0656,
+      "grad_norm": 1.6290217638015747,
+      "learning_rate": 4.874296606185387e-05,
+      "loss": 1.9497,
+      "step": 666
+    },
+    {
+      "epoch": 1.0688,
+      "grad_norm": 1.622162938117981,
+      "learning_rate": 4.847840195361058e-05,
+      "loss": 2.1003,
+      "step": 668
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 1.6676520109176636,
+      "learning_rate": 4.821388047947979e-05,
+      "loss": 2.0269,
+      "step": 670
+    },
+    {
+      "epoch": 1.0752,
+      "grad_norm": 1.6200871467590332,
+      "learning_rate": 4.7949409051168085e-05,
+      "loss": 2.186,
+      "step": 672
+    },
+    {
+      "epoch": 1.0784,
+      "grad_norm": 1.6458507776260376,
+      "learning_rate": 4.768499507897981e-05,
+      "loss": 2.1137,
+      "step": 674
+    },
+    {
+      "epoch": 1.0816,
+      "grad_norm": 1.5950709581375122,
+      "learning_rate": 4.742064597160948e-05,
+      "loss": 1.9582,
+      "step": 676
+    },
+    {
+      "epoch": 1.0848,
+      "grad_norm": 1.7151204347610474,
+      "learning_rate": 4.715636913593404e-05,
+      "loss": 2.0608,
+      "step": 678
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 1.6758900880813599,
+      "learning_rate": 4.689217197680554e-05,
+      "loss": 1.9813,
+      "step": 680
+    },
+    {
+      "epoch": 1.0912,
+      "grad_norm": 1.7488733530044556,
+      "learning_rate": 4.6628061896843474e-05,
+      "loss": 1.983,
+      "step": 682
+    },
+    {
+      "epoch": 1.0944,
+      "grad_norm": 1.6792566776275635,
+      "learning_rate": 4.6364046296227484e-05,
+      "loss": 2.0934,
+      "step": 684
+    },
+    {
+      "epoch": 1.0976,
+      "grad_norm": 1.595950722694397,
+      "learning_rate": 4.6100132572489915e-05,
+      "loss": 2.0593,
+      "step": 686
+    },
+    {
+      "epoch": 1.1008,
+      "grad_norm": 1.6133977174758911,
+      "learning_rate": 4.5836328120308674e-05,
+      "loss": 2.1856,
+      "step": 688
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 1.6371673345565796,
+      "learning_rate": 4.5572640331299875e-05,
+      "loss": 1.8256,
+      "step": 690
+    },
+    {
+      "epoch": 1.1072,
+      "grad_norm": 1.6217482089996338,
+      "learning_rate": 4.530907659381086e-05,
+      "loss": 2.0228,
+      "step": 692
+    },
+    {
+      "epoch": 1.1104,
+      "grad_norm": 1.8364999294281006,
+      "learning_rate": 4.504564429271311e-05,
+      "loss": 2.2268,
+      "step": 694
+    },
+    {
+      "epoch": 1.1136,
+      "grad_norm": 1.6585414409637451,
+      "learning_rate": 4.478235080919536e-05,
+      "loss": 2.1736,
+      "step": 696
+    },
+    {
+      "epoch": 1.1168,
+      "grad_norm": 1.7598919868469238,
+      "learning_rate": 4.451920352055678e-05,
+      "loss": 2.0017,
+      "step": 698
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.659216284751892,
+      "learning_rate": 4.425620980000026e-05,
+      "loss": 2.1184,
+      "step": 700
+    },
+    {
+      "epoch": 1.1232,
+      "grad_norm": 1.6339902877807617,
+      "learning_rate": 4.39933770164258e-05,
+      "loss": 2.0396,
+      "step": 702
+    },
+    {
+      "epoch": 1.1264,
+      "grad_norm": 1.713364839553833,
+      "learning_rate": 4.373071253422408e-05,
+      "loss": 2.0963,
+      "step": 704
+    },
+    {
+      "epoch": 1.1296,
+      "grad_norm": 1.728036642074585,
+      "learning_rate": 4.346822371307009e-05,
+      "loss": 2.026,
+      "step": 706
+    },
+    {
+      "epoch": 1.1328,
+      "grad_norm": 1.6496070623397827,
+      "learning_rate": 4.320591790771691e-05,
+      "loss": 2.0528,
+      "step": 708
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.7058520317077637,
+      "learning_rate": 4.294380246778966e-05,
+      "loss": 1.9498,
+      "step": 710
+    },
+    {
+      "epoch": 1.1392,
+      "grad_norm": 1.5882078409194946,
+      "learning_rate": 4.2681884737579524e-05,
+      "loss": 1.9813,
+      "step": 712
+    },
+    {
+      "epoch": 1.1424,
+      "grad_norm": 1.7214953899383545,
+      "learning_rate": 4.242017205583805e-05,
+      "loss": 2.0316,
+      "step": 714
+    },
+    {
+      "epoch": 1.1456,
+      "grad_norm": 1.6648966073989868,
+      "learning_rate": 4.215867175557142e-05,
+      "loss": 2.0019,
+      "step": 716
+    },
+    {
+      "epoch": 1.1488,
+      "grad_norm": 1.6206002235412598,
+      "learning_rate": 4.189739116383506e-05,
+      "loss": 1.9073,
+      "step": 718
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 1.6015231609344482,
+      "learning_rate": 4.163633760152834e-05,
+      "loss": 2.013,
+      "step": 720
+    },
+    {
+      "epoch": 1.1552,
+      "grad_norm": 1.6537818908691406,
+      "learning_rate": 4.137551838318936e-05,
+      "loss": 2.1472,
+      "step": 722
+    },
+    {
+      "epoch": 1.1584,
+      "grad_norm": 1.9408031702041626,
+      "learning_rate": 4.1114940816790135e-05,
+      "loss": 2.1239,
+      "step": 724
+    },
+    {
+      "epoch": 1.1616,
+      "grad_norm": 1.7310969829559326,
+      "learning_rate": 4.08546122035317e-05,
+      "loss": 2.1907,
+      "step": 726
+    },
+    {
+      "epoch": 1.1648,
+      "grad_norm": 1.6550313234329224,
+      "learning_rate": 4.059453983763967e-05,
+      "loss": 2.0719,
+      "step": 728
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 1.8182940483093262,
+      "learning_rate": 4.03347310061597e-05,
+      "loss": 2.0437,
+      "step": 730
+    },
+    {
+      "epoch": 1.1712,
+      "grad_norm": 1.6281683444976807,
+      "learning_rate": 4.007519298875347e-05,
+      "loss": 2.0488,
+      "step": 732
+    },
+    {
+      "epoch": 1.1743999999999999,
+      "grad_norm": 3.110891342163086,
+      "learning_rate": 3.98159330574946e-05,
+      "loss": 2.0592,
+      "step": 734
+    },
+    {
+      "epoch": 1.1776,
+      "grad_norm": 1.7148778438568115,
+      "learning_rate": 3.955695847666494e-05,
+      "loss": 2.0754,
+      "step": 736
+    },
+    {
+      "epoch": 1.1808,
+      "grad_norm": 1.6069791316986084,
+      "learning_rate": 3.929827650255104e-05,
+      "loss": 1.9366,
+      "step": 738
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 1.7344155311584473,
+      "learning_rate": 3.903989438324077e-05,
+      "loss": 2.1932,
+      "step": 740
+    },
+    {
+      "epoch": 1.1872,
+      "grad_norm": 1.7402360439300537,
+      "learning_rate": 3.878181935842033e-05,
+      "loss": 2.1352,
+      "step": 742
+    },
+    {
+      "epoch": 1.1904,
+      "grad_norm": 1.7004384994506836,
+      "learning_rate": 3.85240586591713e-05,
+      "loss": 1.9779,
+      "step": 744
+    },
+    {
+      "epoch": 1.1936,
+      "grad_norm": 1.8418961763381958,
+      "learning_rate": 3.8266619507768126e-05,
+      "loss": 2.2648,
+      "step": 746
+    },
+    {
+      "epoch": 1.1968,
+      "grad_norm": 1.7444120645523071,
+      "learning_rate": 3.800950911747565e-05,
+      "loss": 2.0055,
+      "step": 748
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.709296703338623,
+      "learning_rate": 3.775273469234712e-05,
+      "loss": 2.1322,
+      "step": 750
+    },
+    {
+      "epoch": 1.2032,
+      "grad_norm": 1.6528595685958862,
+      "learning_rate": 3.749630342702221e-05,
+      "loss": 2.0173,
+      "step": 752
+    },
+    {
+      "epoch": 1.2064,
+      "grad_norm": 1.86497163772583,
+      "learning_rate": 3.724022250652557e-05,
+      "loss": 2.2187,
+      "step": 754
+    },
+    {
+      "epoch": 1.2096,
+      "grad_norm": 1.6813569068908691,
+      "learning_rate": 3.698449910606536e-05,
+      "loss": 1.9149,
+      "step": 756
+    },
+    {
+      "epoch": 1.2128,
+      "grad_norm": 1.4694783687591553,
+      "learning_rate": 3.672914039083233e-05,
+      "loss": 1.9928,
+      "step": 758
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 1.6934654712677002,
+      "learning_rate": 3.6474153515799e-05,
+      "loss": 2.0018,
+      "step": 760
+    },
+    {
+      "epoch": 1.2192,
+      "grad_norm": 1.7023409605026245,
+      "learning_rate": 3.6219545625519145e-05,
+      "loss": 1.9795,
+      "step": 762
+    },
+    {
+      "epoch": 1.2224,
+      "grad_norm": 1.5944123268127441,
+      "learning_rate": 3.596532385392772e-05,
+      "loss": 1.9711,
+      "step": 764
+    },
+    {
+      "epoch": 1.2256,
+      "grad_norm": 1.665781855583191,
+      "learning_rate": 3.5711495324140845e-05,
+      "loss": 2.0697,
+      "step": 766
+    },
+    {
+      "epoch": 1.2288000000000001,
+      "grad_norm": 1.6670796871185303,
+      "learning_rate": 3.54580671482563e-05,
+      "loss": 2.0016,
+      "step": 768
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 1.81361985206604,
+      "learning_rate": 3.520504642715424e-05,
+      "loss": 2.1093,
+      "step": 770
+    },
+    {
+      "epoch": 1.2352,
+      "grad_norm": 1.6931331157684326,
+      "learning_rate": 3.495244025029822e-05,
+      "loss": 2.0959,
+      "step": 772
+    },
+    {
+      "epoch": 1.2384,
+      "grad_norm": 1.6072664260864258,
+      "learning_rate": 3.470025569553653e-05,
+      "loss": 1.9525,
+      "step": 774
+    },
+    {
+      "epoch": 1.2416,
+      "grad_norm": 1.5714666843414307,
+      "learning_rate": 3.444849982890393e-05,
+      "loss": 1.9701,
+      "step": 776
+    },
+    {
+      "epoch": 1.2448,
+      "grad_norm": 1.7283622026443481,
+      "learning_rate": 3.4197179704423656e-05,
+      "loss": 2.1773,
+      "step": 778
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 1.5926114320755005,
+      "learning_rate": 3.39463023639097e-05,
+      "loss": 1.9916,
+      "step": 780
+    },
+    {
+      "epoch": 1.2511999999999999,
+      "grad_norm": 1.6143758296966553,
+      "learning_rate": 3.36958748367696e-05,
+      "loss": 2.0623,
+      "step": 782
+    },
+    {
+      "epoch": 1.2544,
+      "grad_norm": 1.5428004264831543,
+      "learning_rate": 3.3445904139807405e-05,
+      "loss": 1.966,
+      "step": 784
+    },
+    {
+      "epoch": 1.2576,
+      "grad_norm": 1.7031697034835815,
+      "learning_rate": 3.319639727702716e-05,
+      "loss": 2.1779,
+      "step": 786
+    },
+    {
+      "epoch": 1.2608,
+      "grad_norm": 1.5835998058319092,
+      "learning_rate": 3.2947361239436525e-05,
+      "loss": 1.9549,
+      "step": 788
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 1.7775769233703613,
+      "learning_rate": 3.2698803004851026e-05,
+      "loss": 1.9312,
+      "step": 790
+    },
+    {
+      "epoch": 1.2671999999999999,
+      "grad_norm": 1.583977460861206,
+      "learning_rate": 3.245072953769844e-05,
+      "loss": 1.8452,
+      "step": 792
+    },
+    {
+      "epoch": 1.2704,
+      "grad_norm": 1.5791876316070557,
+      "learning_rate": 3.2203147788823764e-05,
+      "loss": 2.0222,
+      "step": 794
+    },
+    {
+      "epoch": 1.2736,
+      "grad_norm": 1.6620211601257324,
+      "learning_rate": 3.1956064695294305e-05,
+      "loss": 2.0612,
+      "step": 796
+    },
+    {
+      "epoch": 1.2768,
+      "grad_norm": 1.6324514150619507,
+      "learning_rate": 3.170948718020546e-05,
+      "loss": 2.0428,
+      "step": 798
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6789875030517578,
+      "learning_rate": 3.1463422152486674e-05,
+      "loss": 1.967,
+      "step": 800
+    },
+    {
+      "epoch": 1.2832,
+      "grad_norm": 2.000178337097168,
+      "learning_rate": 3.121787650670781e-05,
+      "loss": 1.9851,
+      "step": 802
+    },
+    {
+      "epoch": 1.2864,
+      "grad_norm": 1.7736194133758545,
+      "learning_rate": 3.097285712288605e-05,
+      "loss": 1.8732,
+      "step": 804
+    },
+    {
+      "epoch": 1.2896,
+      "grad_norm": 1.7309801578521729,
+      "learning_rate": 3.072837086629305e-05,
+      "loss": 2.0212,
+      "step": 806
+    },
+    {
+      "epoch": 1.2928,
+      "grad_norm": 1.7109277248382568,
+      "learning_rate": 3.0484424587262655e-05,
+      "loss": 2.0203,
+      "step": 808
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 1.6125346422195435,
+      "learning_rate": 3.024102512099889e-05,
+      "loss": 2.075,
+      "step": 810
+    },
+    {
+      "epoch": 1.2992,
+      "grad_norm": 1.661042332649231,
+      "learning_rate": 2.9998179287384485e-05,
+      "loss": 2.1594,
+      "step": 812
+    },
+    {
+      "epoch": 1.3024,
+      "grad_norm": 1.6753617525100708,
+      "learning_rate": 2.975589389078975e-05,
+      "loss": 2.0397,
+      "step": 814
+    },
+    {
+      "epoch": 1.3056,
+      "grad_norm": 1.6913373470306396,
+      "learning_rate": 2.9514175719881966e-05,
+      "loss": 2.0017,
+      "step": 816
+    },
+    {
+      "epoch": 1.3088,
+      "grad_norm": 1.6717087030410767,
+      "learning_rate": 2.9273031547435114e-05,
+      "loss": 2.1042,
+      "step": 818
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 1.7565637826919556,
+      "learning_rate": 2.9032468130140168e-05,
+      "loss": 2.1539,
+      "step": 820
+    },
+    {
+      "epoch": 1.3152,
+      "grad_norm": 1.6280590295791626,
+      "learning_rate": 2.8792492208415768e-05,
+      "loss": 1.9744,
+      "step": 822
+    },
+    {
+      "epoch": 1.3184,
+      "grad_norm": 1.5978885889053345,
+      "learning_rate": 2.8553110506219283e-05,
+      "loss": 2.0044,
+      "step": 824
+    },
+    {
+      "epoch": 1.3216,
+      "grad_norm": 1.6840837001800537,
+      "learning_rate": 2.831432973085848e-05,
+      "loss": 2.062,
+      "step": 826
+    },
+    {
+      "epoch": 1.3248,
+      "grad_norm": 1.5866726636886597,
+      "learning_rate": 2.8076156572803635e-05,
+      "loss": 1.9896,
+      "step": 828
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 1.5741474628448486,
+      "learning_rate": 2.783859770549996e-05,
+      "loss": 1.9465,
+      "step": 830
+    },
+    {
+      "epoch": 1.3312,
+      "grad_norm": 1.683282732963562,
+      "learning_rate": 2.760165978518067e-05,
+      "loss": 2.0189,
+      "step": 832
+    },
+    {
+      "epoch": 1.3344,
+      "grad_norm": 1.6024737358093262,
+      "learning_rate": 2.7365349450680466e-05,
+      "loss": 1.941,
+      "step": 834
+    },
+    {
+      "epoch": 1.3376000000000001,
+      "grad_norm": 1.6115479469299316,
+      "learning_rate": 2.7129673323249604e-05,
+      "loss": 1.9948,
+      "step": 836
+    },
+    {
+      "epoch": 1.3408,
+      "grad_norm": 1.6458511352539062,
+      "learning_rate": 2.689463800636824e-05,
+      "loss": 2.0941,
+      "step": 838
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.5999996662139893,
+      "learning_rate": 2.6660250085561457e-05,
+      "loss": 2.081,
+      "step": 840
+    },
+    {
+      "epoch": 1.3472,
+      "grad_norm": 1.6597297191619873,
+      "learning_rate": 2.6426516128214807e-05,
+      "loss": 1.8927,
+      "step": 842
+    },
+    {
+      "epoch": 1.3504,
+      "grad_norm": 1.7434853315353394,
+      "learning_rate": 2.619344268339021e-05,
+      "loss": 2.0059,
+      "step": 844
+    },
+    {
+      "epoch": 1.3536000000000001,
+      "grad_norm": 2.4289755821228027,
+      "learning_rate": 2.5961036281642493e-05,
+      "loss": 2.1133,
+      "step": 846
+    },
+    {
+      "epoch": 1.3568,
+      "grad_norm": 1.623465657234192,
+      "learning_rate": 2.572930343483637e-05,
+      "loss": 1.9284,
+      "step": 848
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.9334876537322998,
+      "learning_rate": 2.54982506359641e-05,
+      "loss": 2.0351,
+      "step": 850
+    },
+    {
+      "epoch": 1.3632,
+      "grad_norm": 1.7356194257736206,
+      "learning_rate": 2.526788435896339e-05,
+      "loss": 2.0995,
+      "step": 852
+    },
+    {
+      "epoch": 1.3664,
+      "grad_norm": 1.581726312637329,
+      "learning_rate": 2.5038211058536133e-05,
+      "loss": 1.9878,
+      "step": 854
+    },
+    {
+      "epoch": 1.3696,
+      "grad_norm": 1.7448982000350952,
+      "learning_rate": 2.4809237169967458e-05,
+      "loss": 2.0996,
+      "step": 856
+    },
+    {
+      "epoch": 1.3728,
+      "grad_norm": 1.6610596179962158,
+      "learning_rate": 2.4580969108945533e-05,
+      "loss": 1.9928,
+      "step": 858
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 1.6751465797424316,
+      "learning_rate": 2.435341327138168e-05,
+      "loss": 2.0282,
+      "step": 860
+    },
+    {
+      "epoch": 1.3792,
+      "grad_norm": 1.7888554334640503,
+      "learning_rate": 2.4126576033231208e-05,
+      "loss": 1.9525,
+      "step": 862
+    },
+    {
+      "epoch": 1.3824,
+      "grad_norm": 1.66560959815979,
+      "learning_rate": 2.3900463750314834e-05,
+      "loss": 1.984,
+      "step": 864
+    },
+    {
+      "epoch": 1.3856,
+      "grad_norm": 1.6778205633163452,
+      "learning_rate": 2.3675082758140475e-05,
+      "loss": 1.8466,
+      "step": 866
+    },
+    {
+      "epoch": 1.3888,
+      "grad_norm": 1.6759799718856812,
+      "learning_rate": 2.3450439371725825e-05,
+      "loss": 2.0435,
+      "step": 868
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 1.6375926733016968,
+      "learning_rate": 2.3226539885421343e-05,
+      "loss": 1.9285,
+      "step": 870
+    },
+    {
+      "epoch": 1.3952,
+      "grad_norm": 1.7163138389587402,
+      "learning_rate": 2.3003390572734006e-05,
+      "loss": 2.0509,
+      "step": 872
+    },
+    {
+      "epoch": 1.3984,
+      "grad_norm": 1.7290489673614502,
+      "learning_rate": 2.2780997686151378e-05,
+      "loss": 2.0219,
+      "step": 874
+    },
+    {
+      "epoch": 1.4016,
+      "grad_norm": 1.5673465728759766,
+      "learning_rate": 2.255936745696652e-05,
+      "loss": 2.0706,
+      "step": 876
+    },
+    {
+      "epoch": 1.4048,
+      "grad_norm": 1.7429379224777222,
+      "learning_rate": 2.2338506095103334e-05,
+      "loss": 1.9849,
+      "step": 878
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 1.6174511909484863,
+      "learning_rate": 2.2118419788942672e-05,
+      "loss": 1.958,
+      "step": 880
+    },
+    {
+      "epoch": 1.4112,
+      "grad_norm": 1.653412938117981,
+      "learning_rate": 2.189911470514881e-05,
+      "loss": 1.9639,
+      "step": 882
+    },
+    {
+      "epoch": 1.4144,
+      "grad_norm": 1.7575905323028564,
+      "learning_rate": 2.1680596988496705e-05,
+      "loss": 2.0275,
+      "step": 884
+    },
+    {
+      "epoch": 1.4176,
+      "grad_norm": 1.4820189476013184,
+      "learning_rate": 2.1462872761699905e-05,
+      "loss": 1.8751,
+      "step": 886
+    },
+    {
+      "epoch": 1.4208,
+      "grad_norm": 1.558645248413086,
+      "learning_rate": 2.1245948125238867e-05,
+      "loss": 1.8586,
+      "step": 888
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 1.6014423370361328,
+      "learning_rate": 2.1029829157190117e-05,
+      "loss": 1.9532,
+      "step": 890
+    },
+    {
+      "epoch": 1.4272,
+      "grad_norm": 1.5966588258743286,
+      "learning_rate": 2.081452191305587e-05,
+      "loss": 1.9198,
+      "step": 892
+    },
+    {
+      "epoch": 1.4304000000000001,
+      "grad_norm": 1.5865297317504883,
+      "learning_rate": 2.06000324255945e-05,
+      "loss": 1.907,
+      "step": 894
+    },
+    {
+      "epoch": 1.4336,
+      "grad_norm": 1.6410948038101196,
+      "learning_rate": 2.0386366704651315e-05,
+      "loss": 2.1222,
+      "step": 896
+    },
+    {
+      "epoch": 1.4368,
+      "grad_norm": 1.6076593399047852,
+      "learning_rate": 2.0173530736990304e-05,
+      "loss": 1.8874,
+      "step": 898
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.610167384147644,
+      "learning_rate": 1.9961530486126327e-05,
+      "loss": 2.074,
+      "step": 900
+    },
+    {
+      "epoch": 1.4432,
+      "grad_norm": 1.7789791822433472,
+      "learning_rate": 1.9750371892158103e-05,
+      "loss": 2.0396,
+      "step": 902
+    },
+    {
+      "epoch": 1.4464000000000001,
+      "grad_norm": 1.6089529991149902,
+      "learning_rate": 1.9540060871601646e-05,
+      "loss": 2.0706,
+      "step": 904
+    },
+    {
+      "epoch": 1.4496,
+      "grad_norm": 1.6056288480758667,
+      "learning_rate": 1.933060331722457e-05,
+      "loss": 1.8598,
+      "step": 906
+    },
+    {
+      "epoch": 1.4527999999999999,
+      "grad_norm": 1.5929008722305298,
+      "learning_rate": 1.9122005097881014e-05,
+      "loss": 1.9494,
+      "step": 908
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 1.6866933107376099,
+      "learning_rate": 1.8914272058347088e-05,
+      "loss": 1.8477,
+      "step": 910
+    },
+    {
+      "epoch": 1.4592,
+      "grad_norm": 1.614665150642395,
+      "learning_rate": 1.8707410019157196e-05,
+      "loss": 1.907,
+      "step": 912
+    },
+    {
+      "epoch": 1.4624,
+      "grad_norm": 1.5032063722610474,
+      "learning_rate": 1.8501424776440907e-05,
+      "loss": 1.9152,
+      "step": 914
+    },
+    {
+      "epoch": 1.4656,
+      "grad_norm": 1.6080886125564575,
+      "learning_rate": 1.829632210176061e-05,
+      "loss": 1.9224,
+      "step": 916
+    },
+    {
+      "epoch": 1.4687999999999999,
+      "grad_norm": 1.6803348064422607,
+      "learning_rate": 1.809210774194971e-05,
+      "loss": 1.9741,
+      "step": 918
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 1.7053711414337158,
+      "learning_rate": 1.7888787418951645e-05,
+      "loss": 2.0472,
+      "step": 920
+    },
+    {
+      "epoch": 1.4752,
+      "grad_norm": 1.6643372774124146,
+      "learning_rate": 1.7686366829659628e-05,
+      "loss": 2.0729,
+      "step": 922
+    },
+    {
+      "epoch": 1.4784,
+      "grad_norm": 1.6798537969589233,
+      "learning_rate": 1.74848516457569e-05,
+      "loss": 1.9989,
+      "step": 924
+    },
+    {
+      "epoch": 1.4816,
+      "grad_norm": 1.6798433065414429,
+      "learning_rate": 1.72842475135579e-05,
+      "loss": 2.0111,
+      "step": 926
+    },
+    {
+      "epoch": 1.4848,
+      "grad_norm": 1.5733171701431274,
+      "learning_rate": 1.7084560053850024e-05,
+      "loss": 2.0498,
+      "step": 928
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 1.6396392583847046,
+      "learning_rate": 1.6885794861736183e-05,
+      "loss": 1.9701,
+      "step": 930
+    },
+    {
+      "epoch": 1.4912,
+      "grad_norm": 1.628389596939087,
+      "learning_rate": 1.668795750647796e-05,
+      "loss": 2.0453,
+      "step": 932
+    },
+    {
+      "epoch": 1.4944,
+      "grad_norm": 1.580734372138977,
+      "learning_rate": 1.6491053531339607e-05,
+      "loss": 1.908,
+      "step": 934
+    },
+    {
+      "epoch": 1.4976,
+      "grad_norm": 1.6267027854919434,
+      "learning_rate": 1.62950884534327e-05,
+      "loss": 1.949,
+      "step": 936
+    },
+    {
+      "epoch": 1.5008,
+      "grad_norm": 1.5968527793884277,
+      "learning_rate": 1.6100067763561626e-05,
+      "loss": 1.9075,
+      "step": 938
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 1.6220203638076782,
+      "learning_rate": 1.5905996926069628e-05,
+      "loss": 1.8555,
+      "step": 940
+    },
+    {
+      "epoch": 1.5072,
+      "grad_norm": 1.6823129653930664,
+      "learning_rate": 1.5712881378685755e-05,
+      "loss": 2.083,
+      "step": 942
+    },
+    {
+      "epoch": 1.5104,
+      "grad_norm": 1.634656548500061,
+      "learning_rate": 1.5520726532372537e-05,
+      "loss": 1.9641,
+      "step": 944
+    },
+    {
+      "epoch": 1.5135999999999998,
+      "grad_norm": 1.5676469802856445,
+      "learning_rate": 1.532953777117429e-05,
+      "loss": 1.9694,
+      "step": 946
+    },
+    {
+      "epoch": 1.5168,
+      "grad_norm": 1.652653455734253,
+      "learning_rate": 1.5139320452066313e-05,
+      "loss": 1.9754,
+      "step": 948
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.5929044485092163,
+      "learning_rate": 1.4950079904804759e-05,
+      "loss": 1.9062,
+      "step": 950
+    },
+    {
+      "epoch": 1.5232,
+      "grad_norm": 1.6427743434906006,
+      "learning_rate": 1.4761821431777373e-05,
+      "loss": 2.0017,
+      "step": 952
+    },
+    {
+      "epoch": 1.5264,
+      "grad_norm": 1.5246950387954712,
+      "learning_rate": 1.4574550307854817e-05,
+      "loss": 1.9427,
+      "step": 954
+    },
+    {
+      "epoch": 1.5295999999999998,
+      "grad_norm": 1.6303575038909912,
+      "learning_rate": 1.4388271780242929e-05,
+      "loss": 1.8824,
+      "step": 956
+    },
+    {
+      "epoch": 1.5328,
+      "grad_norm": 1.690637469291687,
+      "learning_rate": 1.4202991068335697e-05,
+      "loss": 2.0095,
+      "step": 958
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 1.7552894353866577,
+      "learning_rate": 1.4018713363569035e-05,
+      "loss": 2.0734,
+      "step": 960
+    },
+    {
+      "epoch": 1.5392000000000001,
+      "grad_norm": 1.633931279182434,
+      "learning_rate": 1.3835443829275268e-05,
+      "loss": 1.8231,
+      "step": 962
+    },
+    {
+      "epoch": 1.5424,
+      "grad_norm": 1.586737036705017,
+      "learning_rate": 1.365318760053848e-05,
+      "loss": 1.9087,
+      "step": 964
+    },
+    {
+      "epoch": 1.5455999999999999,
+      "grad_norm": 1.5858091115951538,
+      "learning_rate": 1.3471949784050702e-05,
+      "loss": 1.8308,
+      "step": 966
+    },
+    {
+      "epoch": 1.5488,
+      "grad_norm": 1.6560001373291016,
+      "learning_rate": 1.3291735457968701e-05,
+      "loss": 2.0853,
+      "step": 968
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 1.6715130805969238,
+      "learning_rate": 1.3112549671771796e-05,
+      "loss": 2.0848,
+      "step": 970
+    },
+    {
+      "epoch": 1.5552000000000001,
+      "grad_norm": 1.6843191385269165,
+      "learning_rate": 1.2934397446120306e-05,
+      "loss": 2.0182,
+      "step": 972
+    },
+    {
+      "epoch": 1.5584,
+      "grad_norm": 1.616097092628479,
+      "learning_rate": 1.2757283772714957e-05,
+      "loss": 1.9008,
+      "step": 974
+    },
+    {
+      "epoch": 1.5615999999999999,
+      "grad_norm": 1.5122194290161133,
+      "learning_rate": 1.2581213614156928e-05,
+      "loss": 1.7905,
+      "step": 976
+    },
+    {
+      "epoch": 1.5648,
+      "grad_norm": 1.6870348453521729,
+      "learning_rate": 1.2406191903808844e-05,
+      "loss": 2.1301,
+      "step": 978
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 1.5509247779846191,
+      "learning_rate": 1.2232223545656552e-05,
+      "loss": 1.8702,
+      "step": 980
+    },
+    {
+      "epoch": 1.5712000000000002,
+      "grad_norm": 1.5840715169906616,
+      "learning_rate": 1.205931341417173e-05,
+      "loss": 2.0163,
+      "step": 982
+    },
+    {
+      "epoch": 1.5744,
+      "grad_norm": 1.659145712852478,
+      "learning_rate": 1.1887466354175253e-05,
+      "loss": 1.8171,
+      "step": 984
+    },
+    {
+      "epoch": 1.5776,
+      "grad_norm": 1.7107597589492798,
+      "learning_rate": 1.1716687180701474e-05,
+      "loss": 2.0593,
+      "step": 986
+    },
+    {
+      "epoch": 1.5808,
+      "grad_norm": 1.5526190996170044,
+      "learning_rate": 1.1546980678863361e-05,
+      "loss": 1.8897,
+      "step": 988
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 1.6252529621124268,
+      "learning_rate": 1.1378351603718312e-05,
+      "loss": 1.9001,
+      "step": 990
+    },
+    {
+      "epoch": 1.5872000000000002,
+      "grad_norm": 1.64818274974823,
+      "learning_rate": 1.1210804680135022e-05,
+      "loss": 2.0044,
+      "step": 992
+    },
+    {
+      "epoch": 1.5904,
+      "grad_norm": 1.6926898956298828,
+      "learning_rate": 1.1044344602661034e-05,
+      "loss": 2.1088,
+      "step": 994
+    },
+    {
+      "epoch": 1.5936,
+      "grad_norm": 1.7737877368927002,
+      "learning_rate": 1.0878976035391252e-05,
+      "loss": 1.9949,
+      "step": 996
+    },
+    {
+      "epoch": 1.5968,
+      "grad_norm": 1.6282092332839966,
+      "learning_rate": 1.0714703611837201e-05,
+      "loss": 1.9759,
+      "step": 998
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.5869003534317017,
+      "learning_rate": 1.0551531934797243e-05,
+      "loss": 2.0077,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6032,
+      "grad_norm": 1.5880354642868042,
+      "learning_rate": 1.0389465576227558e-05,
+      "loss": 2.0529,
+      "step": 1002
+    },
+    {
+      "epoch": 1.6064,
+      "grad_norm": 1.796076774597168,
+      "learning_rate": 1.0228509077114146e-05,
+      "loss": 1.9833,
+      "step": 1004
+    },
+    {
+      "epoch": 1.6096,
+      "grad_norm": 1.6073834896087646,
+      "learning_rate": 1.0068666947345456e-05,
+      "loss": 1.8307,
+      "step": 1006
+    },
+    {
+      "epoch": 1.6128,
+      "grad_norm": 1.637624740600586,
+      "learning_rate": 9.909943665586102e-06,
+      "loss": 1.9477,
+      "step": 1008
+    },
+    {
+      "epoch": 1.616,
+      "grad_norm": 1.6284286975860596,
+      "learning_rate": 9.752343679151399e-06,
+      "loss": 1.9163,
+      "step": 1010
+    },
+    {
+      "epoch": 1.6192,
+      "grad_norm": 1.5366007089614868,
+      "learning_rate": 9.595871403882661e-06,
+      "loss": 1.8232,
+      "step": 1012
+    },
+    {
+      "epoch": 1.6223999999999998,
+      "grad_norm": 1.5945929288864136,
+      "learning_rate": 9.440531224023552e-06,
+      "loss": 1.9313,
+      "step": 1014
+    },
+    {
+      "epoch": 1.6256,
+      "grad_norm": 1.589713454246521,
+      "learning_rate": 9.286327492097196e-06,
+      "loss": 1.8511,
+      "step": 1016
+    },
+    {
+      "epoch": 1.6288,
+      "grad_norm": 1.6770306825637817,
+      "learning_rate": 9.133264528784274e-06,
+      "loss": 2.1287,
+      "step": 1018
+    },
+    {
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.5880622863769531,
+      "learning_rate": 8.981346622801905e-06,
+      "loss": 1.9876,
+      "step": 1020
+    },
+    {
+      "epoch": 1.6352,
+      "grad_norm": 1.6425656080245972,
+      "learning_rate": 8.830578030783493e-06,
+      "loss": 1.9793,
+      "step": 1022
+    },
+    {
+      "epoch": 1.6383999999999999,
+      "grad_norm": 1.7399017810821533,
+      "learning_rate": 8.680962977159502e-06,
+      "loss": 1.9023,
+      "step": 1024
+    },
+    {
+      "epoch": 1.6416,
+      "grad_norm": 1.6427820920944214,
+      "learning_rate": 8.53250565403903e-06,
+      "loss": 1.897,
+      "step": 1026
+    },
+    {
+      "epoch": 1.6448,
+      "grad_norm": 1.5444337129592896,
+      "learning_rate": 8.385210221092382e-06,
+      "loss": 1.8902,
+      "step": 1028
+    },
+    {
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.6862332820892334,
+      "learning_rate": 8.239080805434513e-06,
+      "loss": 1.9339,
+      "step": 1030
+    },
+    {
+      "epoch": 1.6512,
+      "grad_norm": 1.5926483869552612,
+      "learning_rate": 8.094121501509399e-06,
+      "loss": 1.8791,
+      "step": 1032
+    },
+    {
+      "epoch": 1.6543999999999999,
+      "grad_norm": 1.634626865386963,
+      "learning_rate": 7.950336370975304e-06,
+      "loss": 1.9781,
+      "step": 1034
+    },
+    {
+      "epoch": 1.6576,
+      "grad_norm": 1.6866514682769775,
+      "learning_rate": 7.80772944259096e-06,
+      "loss": 1.9539,
+      "step": 1036
+    },
+    {
+      "epoch": 1.6608,
+      "grad_norm": 1.58961820602417,
+      "learning_rate": 7.666304712102695e-06,
+      "loss": 1.669,
+      "step": 1038
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.800206184387207,
+      "learning_rate": 7.526066142132521e-06,
+      "loss": 1.9265,
+      "step": 1040
+    },
+    {
+      "epoch": 1.6672,
+      "grad_norm": 1.6069746017456055,
+      "learning_rate": 7.3870176620670194e-06,
+      "loss": 1.8683,
+      "step": 1042
+    },
+    {
+      "epoch": 1.6703999999999999,
+      "grad_norm": 1.639652132987976,
+      "learning_rate": 7.249163167947287e-06,
+      "loss": 1.8093,
+      "step": 1044
+    },
+    {
+      "epoch": 1.6736,
+      "grad_norm": 1.6031360626220703,
+      "learning_rate": 7.1125065223598076e-06,
+      "loss": 2.0082,
+      "step": 1046
+    },
+    {
+      "epoch": 1.6768,
+      "grad_norm": 1.6937788724899292,
+      "learning_rate": 6.9770515543281455e-06,
+      "loss": 2.0,
+      "step": 1048
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.6093850135803223,
+      "learning_rate": 6.842802059205727e-06,
+      "loss": 2.0147,
+      "step": 1050
+    },
+    {
+      "epoch": 1.6832,
+      "grad_norm": 1.541329026222229,
+      "learning_rate": 6.709761798569442e-06,
+      "loss": 1.8566,
+      "step": 1052
+    },
+    {
+      "epoch": 1.6864,
+      "grad_norm": 1.5949456691741943,
+      "learning_rate": 6.577934500114335e-06,
+      "loss": 1.873,
+      "step": 1054
+    },
+    {
+      "epoch": 1.6896,
+      "grad_norm": 1.6678704023361206,
+      "learning_rate": 6.44732385754902e-06,
+      "loss": 1.8975,
+      "step": 1056
+    },
+    {
+      "epoch": 1.6928,
+      "grad_norm": 1.7647327184677124,
+      "learning_rate": 6.3179335304923095e-06,
+      "loss": 1.932,
+      "step": 1058
+    },
+    {
+      "epoch": 1.696,
+      "grad_norm": 1.6539088487625122,
+      "learning_rate": 6.189767144370645e-06,
+      "loss": 1.9846,
+      "step": 1060
+    },
+    {
+      "epoch": 1.6992,
+      "grad_norm": 2.910256862640381,
+      "learning_rate": 6.062828290316469e-06,
+      "loss": 2.0498,
+      "step": 1062
+    },
+    {
+      "epoch": 1.7024,
+      "grad_norm": 1.6432437896728516,
+      "learning_rate": 5.937120525067641e-06,
+      "loss": 1.9539,
+      "step": 1064
+    },
+    {
+      "epoch": 1.7056,
+      "grad_norm": 1.6407873630523682,
+      "learning_rate": 5.812647370867763e-06,
+      "loss": 1.9476,
+      "step": 1066
+    },
+    {
+      "epoch": 1.7088,
+      "grad_norm": 1.617720603942871,
+      "learning_rate": 5.689412315367543e-06,
+      "loss": 1.9902,
+      "step": 1068
+    },
+    {
+      "epoch": 1.712,
+      "grad_norm": 1.5690410137176514,
+      "learning_rate": 5.567418811526981e-06,
+      "loss": 1.8755,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7151999999999998,
+      "grad_norm": 1.602350115776062,
+      "learning_rate": 5.4466702775186785e-06,
+      "loss": 1.9992,
+      "step": 1072
+    },
+    {
+      "epoch": 1.7184,
+      "grad_norm": 1.511695384979248,
+      "learning_rate": 5.327170096632089e-06,
+      "loss": 1.749,
+      "step": 1074
+    },
+    {
+      "epoch": 1.7216,
+      "grad_norm": 1.547643780708313,
+      "learning_rate": 5.208921617178641e-06,
+      "loss": 1.8993,
+      "step": 1076
+    },
+    {
+      "epoch": 1.7248,
+      "grad_norm": 1.7196307182312012,
+      "learning_rate": 5.091928152397984e-06,
+      "loss": 1.9801,
+      "step": 1078
+    },
+    {
+      "epoch": 1.728,
+      "grad_norm": 1.7015469074249268,
+      "learning_rate": 4.976192980365124e-06,
+      "loss": 1.7825,
+      "step": 1080
+    },
+    {
+      "epoch": 1.7311999999999999,
+      "grad_norm": 1.688491702079773,
+      "learning_rate": 4.861719343898613e-06,
+      "loss": 2.0279,
+      "step": 1082
+    },
+    {
+      "epoch": 1.7344,
+      "grad_norm": 1.6017239093780518,
+      "learning_rate": 4.748510450469623e-06,
+      "loss": 1.8703,
+      "step": 1084
+    },
+    {
+      "epoch": 1.7376,
+      "grad_norm": 1.6276230812072754,
+      "learning_rate": 4.63656947211214e-06,
+      "loss": 1.9359,
+      "step": 1086
+    },
+    {
+      "epoch": 1.7408000000000001,
+      "grad_norm": 1.5692756175994873,
+      "learning_rate": 4.525899545334023e-06,
+      "loss": 1.9388,
+      "step": 1088
+    },
+    {
+      "epoch": 1.744,
+      "grad_norm": 1.6018142700195312,
+      "learning_rate": 4.416503771029201e-06,
+      "loss": 1.8337,
+      "step": 1090
+    },
+    {
+      "epoch": 1.7471999999999999,
+      "grad_norm": 1.6205910444259644,
+      "learning_rate": 4.308385214390709e-06,
+      "loss": 1.9551,
+      "step": 1092
+    },
+    {
+      "epoch": 1.7504,
+      "grad_norm": 1.5405246019363403,
+      "learning_rate": 4.2015469048248375e-06,
+      "loss": 1.824,
+      "step": 1094
+    },
+    {
+      "epoch": 1.7536,
+      "grad_norm": 1.6863925457000732,
+      "learning_rate": 4.095991835866275e-06,
+      "loss": 2.0388,
+      "step": 1096
+    },
+    {
+      "epoch": 1.7568000000000001,
+      "grad_norm": 1.59629225730896,
+      "learning_rate": 3.99172296509418e-06,
+      "loss": 1.9307,
+      "step": 1098
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.6163777112960815,
+      "learning_rate": 3.888743214049346e-06,
+      "loss": 1.9041,
+      "step": 1100
+    },
+    {
+      "epoch": 1.7631999999999999,
+      "grad_norm": 1.6829643249511719,
+      "learning_rate": 3.7870554681523287e-06,
+      "loss": 1.9971,
+      "step": 1102
+    },
+    {
+      "epoch": 1.7664,
+      "grad_norm": 1.614225149154663,
+      "learning_rate": 3.6866625766226293e-06,
+      "loss": 1.9732,
+      "step": 1104
+    },
+    {
+      "epoch": 1.7696,
+      "grad_norm": 1.631516933441162,
+      "learning_rate": 3.587567352398796e-06,
+      "loss": 1.8115,
+      "step": 1106
+    },
+    {
+      "epoch": 1.7728000000000002,
+      "grad_norm": 1.5938745737075806,
+      "learning_rate": 3.489772572059674e-06,
+      "loss": 1.9675,
+      "step": 1108
+    },
+    {
+      "epoch": 1.776,
+      "grad_norm": 1.580931544303894,
+      "learning_rate": 3.393280975746588e-06,
+      "loss": 1.9522,
+      "step": 1110
+    },
+    {
+      "epoch": 1.7792,
+      "grad_norm": 1.5829232931137085,
+      "learning_rate": 3.2980952670865317e-06,
+      "loss": 1.7989,
+      "step": 1112
+    },
+    {
+      "epoch": 1.7824,
+      "grad_norm": 1.695078730583191,
+      "learning_rate": 3.2042181131164528e-06,
+      "loss": 1.9354,
+      "step": 1114
+    },
+    {
+      "epoch": 1.7856,
+      "grad_norm": 1.579819679260254,
+      "learning_rate": 3.11165214420851e-06,
+      "loss": 1.8468,
+      "step": 1116
+    },
+    {
+      "epoch": 1.7888,
+      "grad_norm": 1.5718368291854858,
+      "learning_rate": 3.020399953996389e-06,
+      "loss": 1.8217,
+      "step": 1118
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 1.5997322797775269,
+      "learning_rate": 2.9304640993025988e-06,
+      "loss": 1.788,
+      "step": 1120
+    },
+    {
+      "epoch": 1.7952,
+      "grad_norm": 1.5769128799438477,
+      "learning_rate": 2.8418471000668523e-06,
+      "loss": 1.8006,
+      "step": 1122
+    },
+    {
+      "epoch": 1.7984,
+      "grad_norm": 1.602059006690979,
+      "learning_rate": 2.7545514392754437e-06,
+      "loss": 1.9391,
+      "step": 1124
+    },
+    {
+      "epoch": 1.8016,
+      "grad_norm": 1.7443532943725586,
+      "learning_rate": 2.6685795628917266e-06,
+      "loss": 2.0419,
+      "step": 1126
+    },
+    {
+      "epoch": 1.8048,
+      "grad_norm": 2.4460225105285645,
+      "learning_rate": 2.5839338797875036e-06,
+      "loss": 2.0401,
+      "step": 1128
+    },
+    {
+      "epoch": 1.808,
+      "grad_norm": 1.5434504747390747,
+      "learning_rate": 2.500616761675578e-06,
+      "loss": 1.8181,
+      "step": 1130
+    },
+    {
+      "epoch": 1.8112,
+      "grad_norm": 1.6132147312164307,
+      "learning_rate": 2.41863054304331e-06,
+      "loss": 2.0401,
+      "step": 1132
+    },
+    {
+      "epoch": 1.8144,
+      "grad_norm": 1.6048423051834106,
+      "learning_rate": 2.3379775210871648e-06,
+      "loss": 1.8892,
+      "step": 1134
+    },
+    {
+      "epoch": 1.8176,
+      "grad_norm": 1.5809746980667114,
+      "learning_rate": 2.2586599556483734e-06,
+      "loss": 1.8272,
+      "step": 1136
+    },
+    {
+      "epoch": 1.8208,
+      "grad_norm": 1.7967019081115723,
+      "learning_rate": 2.180680069149621e-06,
+      "loss": 1.9311,
+      "step": 1138
+    },
+    {
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.6240350008010864,
+      "learning_rate": 2.104040046532768e-06,
+      "loss": 1.9291,
+      "step": 1140
+    },
+    {
+      "epoch": 1.8272,
+      "grad_norm": 1.6555086374282837,
+      "learning_rate": 2.0287420351976063e-06,
+      "loss": 1.9264,
+      "step": 1142
+    },
+    {
+      "epoch": 1.8304,
+      "grad_norm": 1.6264870166778564,
+      "learning_rate": 1.954788144941727e-06,
+      "loss": 1.8148,
+      "step": 1144
+    },
+    {
+      "epoch": 1.8336000000000001,
+      "grad_norm": 1.6269564628601074,
+      "learning_rate": 1.8821804479013772e-06,
+      "loss": 1.7891,
+      "step": 1146
+    },
+    {
+      "epoch": 1.8368,
+      "grad_norm": 1.6172057390213013,
+      "learning_rate": 1.81092097849343e-06,
+      "loss": 2.0688,
+      "step": 1148
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.6283061504364014,
+      "learning_rate": 1.7410117333583498e-06,
+      "loss": 1.9176,
+      "step": 1150
+    },
+    {
+      "epoch": 1.8432,
+      "grad_norm": 1.6128772497177124,
+      "learning_rate": 1.6724546713042577e-06,
+      "loss": 1.7971,
+      "step": 1152
+    },
+    {
+      "epoch": 1.8464,
+      "grad_norm": 1.603771686553955,
+      "learning_rate": 1.6052517132520651e-06,
+      "loss": 1.9016,
+      "step": 1154
+    },
+    {
+      "epoch": 1.8496000000000001,
+      "grad_norm": 1.5155688524246216,
+      "learning_rate": 1.5394047421816327e-06,
+      "loss": 1.8073,
+      "step": 1156
+    },
+    {
+      "epoch": 1.8528,
+      "grad_norm": 1.6564611196517944,
+      "learning_rate": 1.4749156030790024e-06,
+      "loss": 1.9445,
+      "step": 1158
+    },
+    {
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.7882254123687744,
+      "learning_rate": 1.4117861028847267e-06,
+      "loss": 2.0257,
+      "step": 1160
+    },
+    {
+      "epoch": 1.8592,
+      "grad_norm": 1.6352617740631104,
+      "learning_rate": 1.3500180104432325e-06,
+      "loss": 2.1369,
+      "step": 1162
+    },
+    {
+      "epoch": 1.8624,
+      "grad_norm": 1.5756018161773682,
+      "learning_rate": 1.2896130564532427e-06,
+      "loss": 1.7884,
+      "step": 1164
+    },
+    {
+      "epoch": 1.8656000000000001,
+      "grad_norm": 1.6122503280639648,
+      "learning_rate": 1.2305729334192994e-06,
+      "loss": 1.9276,
+      "step": 1166
+    },
+    {
+      "epoch": 1.8688,
+      "grad_norm": 1.888556718826294,
+      "learning_rate": 1.1728992956043238e-06,
+      "loss": 2.0542,
+      "step": 1168
+    },
+    {
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.6685359477996826,
+      "learning_rate": 1.1165937589833087e-06,
+      "loss": 1.9068,
+      "step": 1170
+    },
+    {
+      "epoch": 1.8752,
+      "grad_norm": 1.647849202156067,
+      "learning_rate": 1.061657901197971e-06,
+      "loss": 1.905,
+      "step": 1172
+    },
+    {
+      "epoch": 1.8784,
+      "grad_norm": 1.5802327394485474,
+      "learning_rate": 1.008093261512616e-06,
+      "loss": 1.8536,
+      "step": 1174
+    },
+    {
+      "epoch": 1.8816000000000002,
+      "grad_norm": 1.6293805837631226,
+      "learning_rate": 9.559013407709595e-07,
+      "loss": 1.9312,
+      "step": 1176
+    },
+    {
+      "epoch": 1.8848,
+      "grad_norm": 1.5595163106918335,
+      "learning_rate": 9.050836013541009e-07,
+      "loss": 1.9464,
+      "step": 1178
+    },
+    {
+      "epoch": 1.888,
+      "grad_norm": 1.518751621246338,
+      "learning_rate": 8.55641467139534e-07,
+      "loss": 1.8809,
+      "step": 1180
+    },
+    {
+      "epoch": 1.8912,
+      "grad_norm": 1.6101138591766357,
+      "learning_rate": 8.075763234612622e-07,
+      "loss": 1.8115,
+      "step": 1182
+    },
+    {
+      "epoch": 1.8944,
+      "grad_norm": 1.6096612215042114,
+      "learning_rate": 7.60889517070984e-07,
+      "loss": 1.9574,
+      "step": 1184
+    },
+    {
+      "epoch": 1.8976,
+      "grad_norm": 1.5596234798431396,
+      "learning_rate": 7.155823561003361e-07,
+      "loss": 1.9247,
+      "step": 1186
+    },
+    {
+      "epoch": 1.9008,
+      "grad_norm": 1.63005530834198,
+      "learning_rate": 6.716561100242658e-07,
+      "loss": 1.9093,
+      "step": 1188
+    },
+    {
+      "epoch": 1.904,
+      "grad_norm": 1.932976245880127,
+      "learning_rate": 6.291120096254433e-07,
+      "loss": 1.9061,
+      "step": 1190
+    },
+    {
+      "epoch": 1.9072,
+      "grad_norm": 1.571678638458252,
+      "learning_rate": 5.879512469598058e-07,
+      "loss": 1.9505,
+      "step": 1192
+    },
+    {
+      "epoch": 1.9104,
+      "grad_norm": 1.6436798572540283,
+      "learning_rate": 5.481749753231124e-07,
+      "loss": 1.977,
+      "step": 1194
+    },
+    {
+      "epoch": 1.9136,
+      "grad_norm": 1.6677478551864624,
+      "learning_rate": 5.097843092186583e-07,
+      "loss": 1.9259,
+      "step": 1196
+    },
+    {
+      "epoch": 1.9167999999999998,
+      "grad_norm": 1.6268632411956787,
+      "learning_rate": 4.7278032432604425e-07,
+      "loss": 1.9393,
+      "step": 1198
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.673553228378296,
+      "learning_rate": 4.371640574710345e-07,
+      "loss": 1.9702,
+      "step": 1200
+    },
+    {
+      "epoch": 1.9232,
+      "grad_norm": 1.6786904335021973,
+      "learning_rate": 4.0293650659650184e-07,
+      "loss": 1.9319,
+      "step": 1202
+    },
+    {
+      "epoch": 1.9264000000000001,
+      "grad_norm": 1.6758121252059937,
+      "learning_rate": 3.7009863073446673e-07,
+      "loss": 1.9173,
+      "step": 1204
+    },
+    {
+      "epoch": 1.9296,
+      "grad_norm": 1.7078176736831665,
+      "learning_rate": 3.386513499792354e-07,
+      "loss": 2.1281,
+      "step": 1206
+    },
+    {
+      "epoch": 1.9327999999999999,
+      "grad_norm": 1.6237930059432983,
+      "learning_rate": 3.0859554546160965e-07,
+      "loss": 1.8814,
+      "step": 1208
+    },
+    {
+      "epoch": 1.936,
+      "grad_norm": 1.6799558401107788,
+      "learning_rate": 2.7993205932420053e-07,
+      "loss": 1.891,
+      "step": 1210
+    },
+    {
+      "epoch": 1.9392,
+      "grad_norm": 1.6336385011672974,
+      "learning_rate": 2.5266169469783105e-07,
+      "loss": 1.9337,
+      "step": 1212
+    },
+    {
+      "epoch": 1.9424000000000001,
+      "grad_norm": 1.6081047058105469,
+      "learning_rate": 2.2678521567903176e-07,
+      "loss": 1.8411,
+      "step": 1214
+    },
+    {
+      "epoch": 1.9456,
+      "grad_norm": 1.64824640750885,
+      "learning_rate": 2.023033473086411e-07,
+      "loss": 1.9833,
+      "step": 1216
+    },
+    {
+      "epoch": 1.9487999999999999,
+      "grad_norm": 1.6368495225906372,
+      "learning_rate": 1.7921677555147177e-07,
+      "loss": 1.8514,
+      "step": 1218
+    },
+    {
+      "epoch": 1.952,
+      "grad_norm": 1.6810824871063232,
+      "learning_rate": 1.5752614727712057e-07,
+      "loss": 2.0633,
+      "step": 1220
+    },
+    {
+      "epoch": 1.9552,
+      "grad_norm": 1.5862808227539062,
+      "learning_rate": 1.3723207024180507e-07,
+      "loss": 1.9975,
+      "step": 1222
+    },
+    {
+      "epoch": 1.9584000000000001,
+      "grad_norm": 3.4000141620635986,
+      "learning_rate": 1.1833511307136613e-07,
+      "loss": 2.0093,
+      "step": 1224
+    },
+    {
+      "epoch": 1.9616,
+      "grad_norm": 1.5674700736999512,
+      "learning_rate": 1.0083580524531955e-07,
+      "loss": 1.6954,
+      "step": 1226
+    },
+    {
+      "epoch": 1.9647999999999999,
+      "grad_norm": 1.700727105140686,
+      "learning_rate": 8.473463708202345e-08,
+      "loss": 2.0088,
+      "step": 1228
+    },
+    {
+      "epoch": 1.968,
+      "grad_norm": 1.6896897554397583,
+      "learning_rate": 7.003205972494486e-08,
+      "loss": 2.1039,
+      "step": 1230
+    },
+    {
+      "epoch": 1.9712,
+      "grad_norm": 1.693572998046875,
+      "learning_rate": 5.672848513000873e-08,
+      "loss": 1.942,
+      "step": 1232
+    },
+    {
+      "epoch": 1.9744000000000002,
+      "grad_norm": 1.662209153175354,
+      "learning_rate": 4.482428605407374e-08,
+      "loss": 1.8457,
+      "step": 1234
+    },
+    {
+      "epoch": 1.9776,
+      "grad_norm": 1.6514451503753662,
+      "learning_rate": 3.431979604445745e-08,
+      "loss": 1.9565,
+      "step": 1236
+    },
+    {
+      "epoch": 1.9808,
+      "grad_norm": 1.5132412910461426,
+      "learning_rate": 2.521530942962702e-08,
+      "loss": 1.66,
+      "step": 1238
+    },
+    {
+      "epoch": 1.984,
+      "grad_norm": 1.609984040260315,
+      "learning_rate": 1.7511081310922495e-08,
+      "loss": 1.8459,
+      "step": 1240
+    },
+    {
+      "epoch": 1.9872,
+      "grad_norm": 1.6186810731887817,
+      "learning_rate": 1.1207327555429192e-08,
+      "loss": 1.9105,
+      "step": 1242
+    },
+    {
+      "epoch": 1.9904,
+      "grad_norm": 1.6339943408966064,
+      "learning_rate": 6.304224789910329e-09,
+      "loss": 1.867,
+      "step": 1244
+    },
+    {
+      "epoch": 1.9936,
+      "grad_norm": 1.6734284162521362,
+      "learning_rate": 2.801910395877627e-09,
+      "loss": 1.936,
+      "step": 1246
+    },
+    {
+      "epoch": 1.9968,
+      "grad_norm": 1.6463286876678467,
+      "learning_rate": 7.004825057277398e-10,
+      "loss": 2.0955,
+      "step": 1248
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.7489490509033203,
+      "learning_rate": 0.0,
+      "loss": 1.8566,
+      "step": 1250
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1250,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 625,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1154802941978214e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}