{
  "best_metric": 0.6713286713286714,
  "best_model_checkpoint": "wav2vec2-5Class-train-test-finetune-V1/checkpoint-858",
  "epoch": 323.0769230769231,
  "eval_steps": 500,
  "global_step": 1050,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.92,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6025770902633667,
      "eval_runtime": 4.6997,
      "eval_samples_per_second": 60.854,
      "eval_steps_per_second": 0.638,
      "step": 3
    },
    {
      "epoch": 1.85,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6024432182312012,
      "eval_runtime": 4.9888,
      "eval_samples_per_second": 57.329,
      "eval_steps_per_second": 0.601,
      "step": 6
    },
    {
      "epoch": 2.77,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6022151708602905,
      "eval_runtime": 4.302,
      "eval_samples_per_second": 66.481,
      "eval_steps_per_second": 0.697,
      "step": 9
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6017613410949707,
      "eval_runtime": 4.5772,
      "eval_samples_per_second": 62.484,
      "eval_steps_per_second": 0.655,
      "step": 13
    },
    {
      "epoch": 4.92,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6013050079345703,
      "eval_runtime": 4.3565,
      "eval_samples_per_second": 65.649,
      "eval_steps_per_second": 0.689,
      "step": 16
    },
    {
      "epoch": 5.85,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6007716655731201,
      "eval_runtime": 4.1903,
      "eval_samples_per_second": 68.253,
      "eval_steps_per_second": 0.716,
      "step": 19
    },
    {
      "epoch": 6.77,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6001368761062622,
      "eval_runtime": 4.7263,
      "eval_samples_per_second": 60.513,
      "eval_steps_per_second": 0.635,
      "step": 22
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.599120020866394,
      "eval_runtime": 4.511,
      "eval_samples_per_second": 63.4,
      "eval_steps_per_second": 0.665,
      "step": 26
    },
    {
      "epoch": 8.92,
      "eval_accuracy": 0.16433566433566432,
      "eval_loss": 1.5982468128204346,
      "eval_runtime": 4.2484,
      "eval_samples_per_second": 67.319,
      "eval_steps_per_second": 0.706,
      "step": 29
    },
    {
      "epoch": 9.85,
      "eval_accuracy": 0.1853146853146853,
      "eval_loss": 1.597304105758667,
      "eval_runtime": 3.8273,
      "eval_samples_per_second": 74.726,
      "eval_steps_per_second": 0.784,
      "step": 32
    },
    {
      "epoch": 10.77,
      "eval_accuracy": 0.1888111888111888,
      "eval_loss": 1.5962693691253662,
      "eval_runtime": 4.8122,
      "eval_samples_per_second": 59.432,
      "eval_steps_per_second": 0.623,
      "step": 35
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.21678321678321677,
      "eval_loss": 1.5947297811508179,
      "eval_runtime": 4.7135,
      "eval_samples_per_second": 60.677,
      "eval_steps_per_second": 0.636,
      "step": 39
    },
    {
      "epoch": 12.92,
      "eval_accuracy": 0.23076923076923078,
      "eval_loss": 1.59347403049469,
      "eval_runtime": 4.362,
      "eval_samples_per_second": 65.567,
      "eval_steps_per_second": 0.688,
      "step": 42
    },
    {
      "epoch": 13.85,
      "eval_accuracy": 0.23076923076923078,
      "eval_loss": 1.592130184173584,
      "eval_runtime": 3.6426,
      "eval_samples_per_second": 78.514,
      "eval_steps_per_second": 0.824,
      "step": 45
    },
    {
      "epoch": 14.77,
      "eval_accuracy": 0.24825174825174826,
      "eval_loss": 1.5907161235809326,
      "eval_runtime": 3.8908,
      "eval_samples_per_second": 73.506,
      "eval_steps_per_second": 0.771,
      "step": 48
    },
    {
      "epoch": 15.38,
      "grad_norm": 0.9358195066452026,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 1.5896,
      "step": 50
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.27972027972027974,
      "eval_loss": 1.588749647140503,
      "eval_runtime": 4.5869,
      "eval_samples_per_second": 62.352,
      "eval_steps_per_second": 0.654,
      "step": 52
    },
    {
      "epoch": 16.92,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.5871775150299072,
      "eval_runtime": 3.7115,
      "eval_samples_per_second": 77.057,
      "eval_steps_per_second": 0.808,
      "step": 55
    },
    {
      "epoch": 17.85,
      "eval_accuracy": 0.3041958041958042,
      "eval_loss": 1.5855580568313599,
      "eval_runtime": 4.1287,
      "eval_samples_per_second": 69.271,
      "eval_steps_per_second": 0.727,
      "step": 58
    },
    {
      "epoch": 18.77,
      "eval_accuracy": 0.3356643356643357,
      "eval_loss": 1.5838947296142578,
      "eval_runtime": 4.2428,
      "eval_samples_per_second": 67.408,
      "eval_steps_per_second": 0.707,
      "step": 61
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.3706293706293706,
      "eval_loss": 1.5815041065216064,
      "eval_runtime": 4.1047,
      "eval_samples_per_second": 69.677,
      "eval_steps_per_second": 0.731,
      "step": 65
    },
    {
      "epoch": 20.92,
      "eval_accuracy": 0.3811188811188811,
      "eval_loss": 1.5795336961746216,
      "eval_runtime": 4.3419,
      "eval_samples_per_second": 65.869,
      "eval_steps_per_second": 0.691,
      "step": 68
    },
    {
      "epoch": 21.85,
      "eval_accuracy": 0.3776223776223776,
      "eval_loss": 1.5774352550506592,
      "eval_runtime": 5.0868,
      "eval_samples_per_second": 56.224,
      "eval_steps_per_second": 0.59,
      "step": 71
    },
    {
      "epoch": 22.77,
      "eval_accuracy": 0.36013986013986016,
      "eval_loss": 1.5753329992294312,
      "eval_runtime": 4.2391,
      "eval_samples_per_second": 67.468,
      "eval_steps_per_second": 0.708,
      "step": 74
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.3531468531468531,
      "eval_loss": 1.5723278522491455,
      "eval_runtime": 4.1913,
      "eval_samples_per_second": 68.237,
      "eval_steps_per_second": 0.716,
      "step": 78
    },
    {
      "epoch": 24.92,
      "eval_accuracy": 0.33916083916083917,
      "eval_loss": 1.5699166059494019,
      "eval_runtime": 4.1956,
      "eval_samples_per_second": 68.166,
      "eval_steps_per_second": 0.715,
      "step": 81
    },
    {
      "epoch": 25.85,
      "eval_accuracy": 0.32867132867132864,
      "eval_loss": 1.5674887895584106,
      "eval_runtime": 4.2222,
      "eval_samples_per_second": 67.737,
      "eval_steps_per_second": 0.711,
      "step": 84
    },
    {
      "epoch": 26.77,
      "eval_accuracy": 0.32167832167832167,
      "eval_loss": 1.564941167831421,
      "eval_runtime": 4.3597,
      "eval_samples_per_second": 65.6,
      "eval_steps_per_second": 0.688,
      "step": 87
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.3146853146853147,
      "eval_loss": 1.5612314939498901,
      "eval_runtime": 4.4489,
      "eval_samples_per_second": 64.285,
      "eval_steps_per_second": 0.674,
      "step": 91
    },
    {
      "epoch": 28.92,
      "eval_accuracy": 0.3111888111888112,
      "eval_loss": 1.5583022832870483,
      "eval_runtime": 3.9002,
      "eval_samples_per_second": 73.33,
      "eval_steps_per_second": 0.769,
      "step": 94
    },
    {
      "epoch": 29.85,
      "eval_accuracy": 0.3076923076923077,
      "eval_loss": 1.5552746057510376,
      "eval_runtime": 3.8211,
      "eval_samples_per_second": 74.848,
      "eval_steps_per_second": 0.785,
      "step": 97
    },
    {
      "epoch": 30.77,
      "grad_norm": 1.0354599952697754,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.5478,
      "step": 100
    },
    {
      "epoch": 30.77,
      "eval_accuracy": 0.3111888111888112,
      "eval_loss": 1.5521942377090454,
      "eval_runtime": 3.9174,
      "eval_samples_per_second": 73.008,
      "eval_steps_per_second": 0.766,
      "step": 100
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.3006993006993007,
      "eval_loss": 1.5478450059890747,
      "eval_runtime": 4.2795,
      "eval_samples_per_second": 66.83,
      "eval_steps_per_second": 0.701,
      "step": 104
    },
    {
      "epoch": 32.92,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.5445122718811035,
      "eval_runtime": 4.2852,
      "eval_samples_per_second": 66.742,
      "eval_steps_per_second": 0.7,
      "step": 107
    },
    {
      "epoch": 33.85,
      "eval_accuracy": 0.2867132867132867,
      "eval_loss": 1.5413419008255005,
      "eval_runtime": 4.7279,
      "eval_samples_per_second": 60.492,
      "eval_steps_per_second": 0.635,
      "step": 110
    },
    {
      "epoch": 34.77,
      "eval_accuracy": 0.2762237762237762,
      "eval_loss": 1.5382936000823975,
      "eval_runtime": 4.5512,
      "eval_samples_per_second": 62.841,
      "eval_steps_per_second": 0.659,
      "step": 113
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.2762237762237762,
      "eval_loss": 1.5340471267700195,
      "eval_runtime": 4.0909,
      "eval_samples_per_second": 69.911,
      "eval_steps_per_second": 0.733,
      "step": 117
    },
    {
      "epoch": 36.92,
      "eval_accuracy": 0.26573426573426573,
      "eval_loss": 1.5310546159744263,
      "eval_runtime": 3.6024,
      "eval_samples_per_second": 79.392,
      "eval_steps_per_second": 0.833,
      "step": 120
    },
    {
      "epoch": 37.85,
      "eval_accuracy": 0.2517482517482518,
      "eval_loss": 1.5282360315322876,
      "eval_runtime": 3.8247,
      "eval_samples_per_second": 74.777,
      "eval_steps_per_second": 0.784,
      "step": 123
    },
    {
      "epoch": 38.77,
      "eval_accuracy": 0.24475524475524477,
      "eval_loss": 1.5255342721939087,
      "eval_runtime": 4.5969,
      "eval_samples_per_second": 62.216,
      "eval_steps_per_second": 0.653,
      "step": 126
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.24125874125874125,
      "eval_loss": 1.5223983526229858,
      "eval_runtime": 4.5515,
      "eval_samples_per_second": 62.836,
      "eval_steps_per_second": 0.659,
      "step": 130
    },
    {
      "epoch": 40.92,
      "eval_accuracy": 0.23426573426573427,
      "eval_loss": 1.5204370021820068,
      "eval_runtime": 3.5155,
      "eval_samples_per_second": 81.354,
      "eval_steps_per_second": 0.853,
      "step": 133
    },
    {
      "epoch": 41.85,
      "eval_accuracy": 0.24475524475524477,
      "eval_loss": 1.5190935134887695,
      "eval_runtime": 4.1808,
      "eval_samples_per_second": 68.409,
      "eval_steps_per_second": 0.718,
      "step": 136
    },
    {
      "epoch": 42.77,
      "eval_accuracy": 0.23776223776223776,
      "eval_loss": 1.5183987617492676,
      "eval_runtime": 5.6726,
      "eval_samples_per_second": 50.418,
      "eval_steps_per_second": 0.529,
      "step": 139
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.23076923076923078,
      "eval_loss": 1.518118143081665,
      "eval_runtime": 4.9675,
      "eval_samples_per_second": 57.575,
      "eval_steps_per_second": 0.604,
      "step": 143
    },
    {
      "epoch": 44.92,
      "eval_accuracy": 0.23076923076923078,
      "eval_loss": 1.5188548564910889,
      "eval_runtime": 3.8693,
      "eval_samples_per_second": 73.915,
      "eval_steps_per_second": 0.775,
      "step": 146
    },
    {
      "epoch": 45.85,
      "eval_accuracy": 0.23776223776223776,
      "eval_loss": 1.5198808908462524,
      "eval_runtime": 4.5179,
      "eval_samples_per_second": 63.304,
      "eval_steps_per_second": 0.664,
      "step": 149
    },
    {
      "epoch": 46.15,
      "grad_norm": 0.8092480897903442,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.4365,
      "step": 150
    },
    {
      "epoch": 46.77,
      "eval_accuracy": 0.24825174825174826,
      "eval_loss": 1.5214567184448242,
      "eval_runtime": 4.1214,
      "eval_samples_per_second": 69.394,
      "eval_steps_per_second": 0.728,
      "step": 152
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.25874125874125875,
      "eval_loss": 1.5235881805419922,
      "eval_runtime": 4.2822,
      "eval_samples_per_second": 66.789,
      "eval_steps_per_second": 0.701,
      "step": 156
    },
    {
      "epoch": 48.92,
      "eval_accuracy": 0.26573426573426573,
      "eval_loss": 1.5250674486160278,
      "eval_runtime": 6.0487,
      "eval_samples_per_second": 47.283,
      "eval_steps_per_second": 0.496,
      "step": 159
    },
    {
      "epoch": 49.85,
      "eval_accuracy": 0.28321678321678323,
      "eval_loss": 1.5259337425231934,
      "eval_runtime": 4.1709,
      "eval_samples_per_second": 68.571,
      "eval_steps_per_second": 0.719,
      "step": 162
    },
    {
      "epoch": 50.77,
      "eval_accuracy": 0.27972027972027974,
      "eval_loss": 1.5261683464050293,
      "eval_runtime": 4.3522,
      "eval_samples_per_second": 65.713,
      "eval_steps_per_second": 0.689,
      "step": 165
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.5248568058013916,
      "eval_runtime": 4.8279,
      "eval_samples_per_second": 59.24,
      "eval_steps_per_second": 0.621,
      "step": 169
    },
    {
      "epoch": 52.92,
      "eval_accuracy": 0.3006993006993007,
      "eval_loss": 1.522719383239746,
      "eval_runtime": 4.205,
      "eval_samples_per_second": 68.014,
      "eval_steps_per_second": 0.713,
      "step": 172
    },
    {
      "epoch": 53.85,
      "eval_accuracy": 0.3076923076923077,
      "eval_loss": 1.5189520120620728,
      "eval_runtime": 5.4852,
      "eval_samples_per_second": 52.14,
      "eval_steps_per_second": 0.547,
      "step": 175
    },
    {
      "epoch": 54.77,
      "eval_accuracy": 0.32167832167832167,
      "eval_loss": 1.5137863159179688,
      "eval_runtime": 4.3909,
      "eval_samples_per_second": 65.134,
      "eval_steps_per_second": 0.683,
      "step": 178
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.34965034965034963,
      "eval_loss": 1.5053085088729858,
      "eval_runtime": 4.0994,
      "eval_samples_per_second": 69.766,
      "eval_steps_per_second": 0.732,
      "step": 182
    },
    {
      "epoch": 56.92,
      "eval_accuracy": 0.36013986013986016,
      "eval_loss": 1.497667908668518,
      "eval_runtime": 4.4275,
      "eval_samples_per_second": 64.596,
      "eval_steps_per_second": 0.678,
      "step": 185
    },
    {
      "epoch": 57.85,
      "eval_accuracy": 0.36013986013986016,
      "eval_loss": 1.4910008907318115,
      "eval_runtime": 4.4713,
      "eval_samples_per_second": 63.963,
      "eval_steps_per_second": 0.671,
      "step": 188
    },
    {
      "epoch": 58.77,
      "eval_accuracy": 0.36713286713286714,
      "eval_loss": 1.4839717149734497,
      "eval_runtime": 4.6741,
      "eval_samples_per_second": 61.189,
      "eval_steps_per_second": 0.642,
      "step": 191
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.3706293706293706,
      "eval_loss": 1.4754726886749268,
      "eval_runtime": 4.1736,
      "eval_samples_per_second": 68.526,
      "eval_steps_per_second": 0.719,
      "step": 195
    },
    {
      "epoch": 60.92,
      "eval_accuracy": 0.3811188811188811,
      "eval_loss": 1.4683842658996582,
      "eval_runtime": 3.9445,
      "eval_samples_per_second": 72.507,
      "eval_steps_per_second": 0.761,
      "step": 198
    },
    {
      "epoch": 61.54,
      "grad_norm": 0.41712912917137146,
      "learning_rate": 2.6984126984126984e-05,
      "loss": 1.2845,
      "step": 200
    },
    {
      "epoch": 61.85,
      "eval_accuracy": 0.38461538461538464,
      "eval_loss": 1.4626859426498413,
      "eval_runtime": 4.7292,
      "eval_samples_per_second": 60.475,
      "eval_steps_per_second": 0.634,
      "step": 201
    },
    {
      "epoch": 62.77,
      "eval_accuracy": 0.3881118881118881,
      "eval_loss": 1.4546846151351929,
      "eval_runtime": 4.5742,
      "eval_samples_per_second": 62.524,
      "eval_steps_per_second": 0.656,
      "step": 204
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.4020979020979021,
      "eval_loss": 1.4456309080123901,
      "eval_runtime": 4.1295,
      "eval_samples_per_second": 69.259,
      "eval_steps_per_second": 0.726,
      "step": 208
    },
    {
      "epoch": 64.92,
      "eval_accuracy": 0.40559440559440557,
      "eval_loss": 1.4385287761688232,
      "eval_runtime": 4.9759,
      "eval_samples_per_second": 57.477,
      "eval_steps_per_second": 0.603,
      "step": 211
    },
    {
      "epoch": 65.85,
      "eval_accuracy": 0.4090909090909091,
      "eval_loss": 1.4315868616104126,
      "eval_runtime": 4.7,
      "eval_samples_per_second": 60.851,
      "eval_steps_per_second": 0.638,
      "step": 214
    },
    {
      "epoch": 66.77,
      "eval_accuracy": 0.4160839160839161,
      "eval_loss": 1.423151969909668,
      "eval_runtime": 4.1887,
      "eval_samples_per_second": 68.28,
      "eval_steps_per_second": 0.716,
      "step": 217
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.42657342657342656,
      "eval_loss": 1.4132899045944214,
      "eval_runtime": 6.2189,
      "eval_samples_per_second": 45.989,
      "eval_steps_per_second": 0.482,
      "step": 221
    },
    {
      "epoch": 68.92,
      "eval_accuracy": 0.43006993006993005,
      "eval_loss": 1.4061568975448608,
      "eval_runtime": 5.3452,
      "eval_samples_per_second": 53.506,
      "eval_steps_per_second": 0.561,
      "step": 224
    },
    {
      "epoch": 69.85,
      "eval_accuracy": 0.43356643356643354,
      "eval_loss": 1.4003138542175293,
      "eval_runtime": 5.111,
      "eval_samples_per_second": 55.958,
      "eval_steps_per_second": 0.587,
      "step": 227
    },
    {
      "epoch": 70.77,
      "eval_accuracy": 0.43356643356643354,
      "eval_loss": 1.396316647529602,
      "eval_runtime": 4.6366,
      "eval_samples_per_second": 61.683,
      "eval_steps_per_second": 0.647,
      "step": 230
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.43356643356643354,
      "eval_loss": 1.3880172967910767,
      "eval_runtime": 4.6851,
      "eval_samples_per_second": 61.045,
      "eval_steps_per_second": 0.64,
      "step": 234
    },
    {
      "epoch": 72.92,
      "eval_accuracy": 0.4370629370629371,
      "eval_loss": 1.3800761699676514,
      "eval_runtime": 4.5739,
      "eval_samples_per_second": 62.528,
      "eval_steps_per_second": 0.656,
      "step": 237
    },
    {
      "epoch": 73.85,
      "eval_accuracy": 0.4405594405594406,
      "eval_loss": 1.3724888563156128,
      "eval_runtime": 4.9175,
      "eval_samples_per_second": 58.16,
      "eval_steps_per_second": 0.61,
      "step": 240
    },
    {
      "epoch": 74.77,
      "eval_accuracy": 0.44755244755244755,
      "eval_loss": 1.3655133247375488,
      "eval_runtime": 5.0566,
      "eval_samples_per_second": 56.559,
      "eval_steps_per_second": 0.593,
      "step": 243
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.45104895104895104,
      "eval_loss": 1.3560560941696167,
      "eval_runtime": 3.592,
      "eval_samples_per_second": 79.622,
      "eval_steps_per_second": 0.835,
      "step": 247
    },
    {
      "epoch": 76.92,
      "grad_norm": 0.3923046588897705,
      "learning_rate": 2.5396825396825397e-05,
      "loss": 1.1752,
      "step": 250
    },
    {
      "epoch": 76.92,
      "eval_accuracy": 0.45454545454545453,
      "eval_loss": 1.3477288484573364,
      "eval_runtime": 5.2706,
      "eval_samples_per_second": 54.264,
      "eval_steps_per_second": 0.569,
      "step": 250
    },
    {
      "epoch": 77.85,
      "eval_accuracy": 0.45454545454545453,
      "eval_loss": 1.3385549783706665,
      "eval_runtime": 4.3668,
      "eval_samples_per_second": 65.494,
      "eval_steps_per_second": 0.687,
      "step": 253
    },
    {
      "epoch": 78.77,
      "eval_accuracy": 0.45454545454545453,
      "eval_loss": 1.329953908920288,
      "eval_runtime": 4.3715,
      "eval_samples_per_second": 65.423,
      "eval_steps_per_second": 0.686,
      "step": 256
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.46153846153846156,
      "eval_loss": 1.3186790943145752,
      "eval_runtime": 4.9164,
      "eval_samples_per_second": 58.173,
      "eval_steps_per_second": 0.61,
      "step": 260
    },
    {
      "epoch": 80.92,
      "eval_accuracy": 0.47202797202797203,
      "eval_loss": 1.3101806640625,
      "eval_runtime": 4.5555,
      "eval_samples_per_second": 62.782,
      "eval_steps_per_second": 0.659,
      "step": 263
    },
    {
      "epoch": 81.85,
      "eval_accuracy": 0.4755244755244755,
      "eval_loss": 1.3015599250793457,
      "eval_runtime": 3.7938,
      "eval_samples_per_second": 75.387,
      "eval_steps_per_second": 0.791,
      "step": 266
    },
    {
      "epoch": 82.77,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.2916128635406494,
      "eval_runtime": 3.4076,
      "eval_samples_per_second": 83.931,
      "eval_steps_per_second": 0.88,
      "step": 269
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.2802271842956543,
      "eval_runtime": 3.947,
      "eval_samples_per_second": 72.46,
      "eval_steps_per_second": 0.76,
      "step": 273
    },
    {
      "epoch": 84.92,
      "eval_accuracy": 0.486013986013986,
      "eval_loss": 1.2717865705490112,
      "eval_runtime": 4.6584,
      "eval_samples_per_second": 61.394,
      "eval_steps_per_second": 0.644,
      "step": 276
    },
    {
      "epoch": 85.85,
      "eval_accuracy": 0.48951048951048953,
      "eval_loss": 1.2625547647476196,
      "eval_runtime": 3.7402,
      "eval_samples_per_second": 76.466,
      "eval_steps_per_second": 0.802,
      "step": 279
    },
    {
      "epoch": 86.77,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2543796300888062,
      "eval_runtime": 4.726,
      "eval_samples_per_second": 60.516,
      "eval_steps_per_second": 0.635,
      "step": 282
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2428691387176514,
      "eval_runtime": 3.9492,
      "eval_samples_per_second": 72.42,
      "eval_steps_per_second": 0.76,
      "step": 286
    },
    {
      "epoch": 88.92,
      "eval_accuracy": 0.4965034965034965,
      "eval_loss": 1.2338002920150757,
      "eval_runtime": 4.6146,
      "eval_samples_per_second": 61.978,
      "eval_steps_per_second": 0.65,
      "step": 289
    },
    {
      "epoch": 89.85,
      "eval_accuracy": 0.5034965034965035,
      "eval_loss": 1.223657488822937,
      "eval_runtime": 4.7664,
      "eval_samples_per_second": 60.003,
      "eval_steps_per_second": 0.629,
      "step": 292
    },
    {
      "epoch": 90.77,
      "eval_accuracy": 0.513986013986014,
      "eval_loss": 1.2134180068969727,
      "eval_runtime": 3.6509,
      "eval_samples_per_second": 78.337,
      "eval_steps_per_second": 0.822,
      "step": 295
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.5314685314685315,
      "eval_loss": 1.1996746063232422,
      "eval_runtime": 4.1379,
      "eval_samples_per_second": 69.118,
      "eval_steps_per_second": 0.725,
      "step": 299
    },
    {
      "epoch": 92.31,
      "grad_norm": 0.4333157241344452,
      "learning_rate": 2.380952380952381e-05,
      "loss": 1.0336,
      "step": 300
    },
    {
      "epoch": 92.92,
      "eval_accuracy": 0.534965034965035,
      "eval_loss": 1.1893947124481201,
      "eval_runtime": 4.3495,
      "eval_samples_per_second": 65.755,
      "eval_steps_per_second": 0.69,
      "step": 302
    },
    {
      "epoch": 93.85,
      "eval_accuracy": 0.5524475524475524,
      "eval_loss": 1.1794614791870117,
      "eval_runtime": 5.0315,
      "eval_samples_per_second": 56.842,
      "eval_steps_per_second": 0.596,
      "step": 305
    },
    {
      "epoch": 94.77,
      "eval_accuracy": 0.5629370629370629,
      "eval_loss": 1.1703894138336182,
      "eval_runtime": 4.5156,
      "eval_samples_per_second": 63.336,
      "eval_steps_per_second": 0.664,
      "step": 308
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.5629370629370629,
      "eval_loss": 1.1574207544326782,
      "eval_runtime": 4.2769,
      "eval_samples_per_second": 66.871,
      "eval_steps_per_second": 0.701,
      "step": 312
    },
    {
      "epoch": 96.92,
      "eval_accuracy": 0.5804195804195804,
      "eval_loss": 1.1478246450424194,
      "eval_runtime": 4.4503,
      "eval_samples_per_second": 64.266,
      "eval_steps_per_second": 0.674,
      "step": 315
    },
    {
      "epoch": 97.85,
      "eval_accuracy": 0.583916083916084,
      "eval_loss": 1.1388368606567383,
      "eval_runtime": 5.2516,
      "eval_samples_per_second": 54.46,
      "eval_steps_per_second": 0.571,
      "step": 318
    },
    {
      "epoch": 98.77,
      "eval_accuracy": 0.5874125874125874,
      "eval_loss": 1.129955530166626,
      "eval_runtime": 4.5307,
      "eval_samples_per_second": 63.125,
      "eval_steps_per_second": 0.662,
      "step": 321
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.5944055944055944,
      "eval_loss": 1.1171796321868896,
      "eval_runtime": 4.5358,
      "eval_samples_per_second": 63.054,
      "eval_steps_per_second": 0.661,
      "step": 325
    },
    {
      "epoch": 100.92,
      "eval_accuracy": 0.5979020979020979,
      "eval_loss": 1.1089597940444946,
      "eval_runtime": 4.7931,
      "eval_samples_per_second": 59.669,
      "eval_steps_per_second": 0.626,
      "step": 328
    },
    {
      "epoch": 101.85,
      "eval_accuracy": 0.5944055944055944,
      "eval_loss": 1.1001044511795044,
      "eval_runtime": 4.4134,
      "eval_samples_per_second": 64.802,
      "eval_steps_per_second": 0.68,
      "step": 331
    },
    {
      "epoch": 102.77,
      "eval_accuracy": 0.6048951048951049,
      "eval_loss": 1.0910252332687378,
      "eval_runtime": 4.8251,
      "eval_samples_per_second": 59.273,
      "eval_steps_per_second": 0.622,
      "step": 334
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 0.6013986013986014,
      "eval_loss": 1.0768800973892212,
      "eval_runtime": 5.528,
      "eval_samples_per_second": 51.736,
      "eval_steps_per_second": 0.543,
      "step": 338
    },
    {
      "epoch": 104.92,
      "eval_accuracy": 0.6048951048951049,
      "eval_loss": 1.0675218105316162,
      "eval_runtime": 3.5589,
      "eval_samples_per_second": 80.362,
      "eval_steps_per_second": 0.843,
      "step": 341
    },
    {
      "epoch": 105.85,
      "eval_accuracy": 0.6118881118881119,
      "eval_loss": 1.0602259635925293,
      "eval_runtime": 4.2603,
      "eval_samples_per_second": 67.132,
      "eval_steps_per_second": 0.704,
      "step": 344
    },
    {
      "epoch": 106.77,
      "eval_accuracy": 0.6153846153846154,
      "eval_loss": 1.0537227392196655,
      "eval_runtime": 4.4326,
      "eval_samples_per_second": 64.521,
      "eval_steps_per_second": 0.677,
      "step": 347
    },
    {
      "epoch": 107.69,
      "grad_norm": 0.46974265575408936,
      "learning_rate": 2.222222222222222e-05,
      "loss": 0.8927,
      "step": 350
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 0.6223776223776224,
      "eval_loss": 1.045584797859192,
      "eval_runtime": 3.8481,
      "eval_samples_per_second": 74.323,
      "eval_steps_per_second": 0.78,
      "step": 351
    },
    {
      "epoch": 108.92,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0394147634506226,
      "eval_runtime": 4.6438,
      "eval_samples_per_second": 61.588,
      "eval_steps_per_second": 0.646,
      "step": 354
    },
    {
      "epoch": 109.85,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0330853462219238,
      "eval_runtime": 4.1627,
      "eval_samples_per_second": 68.706,
      "eval_steps_per_second": 0.721,
      "step": 357
    },
    {
      "epoch": 110.77,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0267359018325806,
      "eval_runtime": 4.5108,
      "eval_samples_per_second": 63.403,
      "eval_steps_per_second": 0.665,
      "step": 360
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 0.6328671328671329,
      "eval_loss": 1.0192536115646362,
      "eval_runtime": 4.9039,
      "eval_samples_per_second": 58.321,
      "eval_steps_per_second": 0.612,
      "step": 364
    },
    {
      "epoch": 112.92,
      "eval_accuracy": 0.6363636363636364,
      "eval_loss": 1.0148723125457764,
      "eval_runtime": 3.8444,
      "eval_samples_per_second": 74.393,
      "eval_steps_per_second": 0.78,
      "step": 367
    },
    {
      "epoch": 113.85,
      "eval_accuracy": 0.6363636363636364,
      "eval_loss": 1.0099798440933228,
      "eval_runtime": 5.0977,
      "eval_samples_per_second": 56.104,
      "eval_steps_per_second": 0.589,
      "step": 370
    },
    {
      "epoch": 114.77,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 1.0047037601470947,
      "eval_runtime": 4.837,
      "eval_samples_per_second": 59.128,
      "eval_steps_per_second": 0.62,
      "step": 373
    },
    {
      "epoch": 116.0,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9991178512573242,
      "eval_runtime": 4.9051,
      "eval_samples_per_second": 58.307,
      "eval_steps_per_second": 0.612,
      "step": 377
    },
    {
      "epoch": 116.92,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9972890615463257,
      "eval_runtime": 4.3405,
      "eval_samples_per_second": 65.891,
      "eval_steps_per_second": 0.691,
      "step": 380
    },
    {
      "epoch": 117.85,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9936119914054871,
      "eval_runtime": 3.9596,
      "eval_samples_per_second": 72.229,
      "eval_steps_per_second": 0.758,
      "step": 383
    },
    {
      "epoch": 118.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9908724427223206,
      "eval_runtime": 4.961,
      "eval_samples_per_second": 57.65,
      "eval_steps_per_second": 0.605,
      "step": 386
    },
    {
      "epoch": 120.0,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9878212213516235,
      "eval_runtime": 4.8469,
      "eval_samples_per_second": 59.006,
      "eval_steps_per_second": 0.619,
      "step": 390
    },
    {
      "epoch": 120.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9840787053108215,
      "eval_runtime": 4.5014,
      "eval_samples_per_second": 63.536,
      "eval_steps_per_second": 0.666,
      "step": 393
    },
    {
      "epoch": 121.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9810080528259277,
      "eval_runtime": 3.7722,
      "eval_samples_per_second": 75.818,
      "eval_steps_per_second": 0.795,
      "step": 396
    },
    {
      "epoch": 122.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9769043922424316,
      "eval_runtime": 4.713,
      "eval_samples_per_second": 60.684,
      "eval_steps_per_second": 0.637,
      "step": 399
    },
    {
      "epoch": 123.08,
      "grad_norm": 0.48339083790779114,
      "learning_rate": 2.0634920634920633e-05,
      "loss": 0.7859,
      "step": 400
    },
    {
      "epoch": 124.0,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9750482439994812,
      "eval_runtime": 4.9657,
      "eval_samples_per_second": 57.595,
      "eval_steps_per_second": 0.604,
      "step": 403
    },
    {
      "epoch": 124.92,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9752389192581177,
      "eval_runtime": 4.6155,
      "eval_samples_per_second": 61.965,
      "eval_steps_per_second": 0.65,
      "step": 406
    },
    {
      "epoch": 125.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.975228488445282,
      "eval_runtime": 4.2181,
      "eval_samples_per_second": 67.803,
      "eval_steps_per_second": 0.711,
      "step": 409
    },
    {
      "epoch": 126.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.973861575126648,
      "eval_runtime": 5.1633,
      "eval_samples_per_second": 55.391,
      "eval_steps_per_second": 0.581,
      "step": 412
    },
    {
      "epoch": 128.0,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9697291254997253,
      "eval_runtime": 3.8938,
      "eval_samples_per_second": 73.451,
      "eval_steps_per_second": 0.77,
      "step": 416
    },
    {
      "epoch": 128.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9673418998718262,
      "eval_runtime": 4.9929,
      "eval_samples_per_second": 57.282,
      "eval_steps_per_second": 0.601,
      "step": 419
    },
    {
      "epoch": 129.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9654524922370911,
      "eval_runtime": 4.8097,
      "eval_samples_per_second": 59.463,
      "eval_steps_per_second": 0.624,
      "step": 422
    },
    {
      "epoch": 130.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9648922085762024,
      "eval_runtime": 4.6728,
      "eval_samples_per_second": 61.205,
      "eval_steps_per_second": 0.642,
      "step": 425
    },
    {
      "epoch": 132.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9624325633049011,
      "eval_runtime": 3.9139,
      "eval_samples_per_second": 73.073,
      "eval_steps_per_second": 0.766,
      "step": 429
    },
    {
      "epoch": 132.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.960906445980072,
      "eval_runtime": 5.0181,
      "eval_samples_per_second": 56.994,
      "eval_steps_per_second": 0.598,
      "step": 432
    },
    {
      "epoch": 133.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.958863377571106,
      "eval_runtime": 4.172,
      "eval_samples_per_second": 68.552,
      "eval_steps_per_second": 0.719,
      "step": 435
    },
    {
      "epoch": 134.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9576795697212219,
      "eval_runtime": 5.1205,
      "eval_samples_per_second": 55.854,
      "eval_steps_per_second": 0.586,
      "step": 438
    },
    {
      "epoch": 136.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9575643539428711,
      "eval_runtime": 3.9439,
      "eval_samples_per_second": 72.517,
      "eval_steps_per_second": 0.761,
      "step": 442
    },
    {
      "epoch": 136.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9587862491607666,
      "eval_runtime": 4.289,
      "eval_samples_per_second": 66.682,
      "eval_steps_per_second": 0.699,
      "step": 445
    },
    {
      "epoch": 137.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9579612016677856,
      "eval_runtime": 4.3725,
      "eval_samples_per_second": 65.409,
      "eval_steps_per_second": 0.686,
      "step": 448
    },
    {
      "epoch": 138.46,
      "grad_norm": 0.5380613803863525,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 0.7428,
      "step": 450
    },
    {
      "epoch": 138.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9572532176971436,
      "eval_runtime": 4.4425,
      "eval_samples_per_second": 64.378,
      "eval_steps_per_second": 0.675,
      "step": 451
    },
    {
      "epoch": 140.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9572116732597351,
      "eval_runtime": 5.4248,
      "eval_samples_per_second": 52.721,
      "eval_steps_per_second": 0.553,
      "step": 455
    },
    {
      "epoch": 140.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9591361284255981,
      "eval_runtime": 4.4268,
      "eval_samples_per_second": 64.606,
      "eval_steps_per_second": 0.678,
      "step": 458
    },
    {
      "epoch": 141.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9620410203933716,
      "eval_runtime": 5.1983,
      "eval_samples_per_second": 55.018,
      "eval_steps_per_second": 0.577,
      "step": 461
    },
    {
      "epoch": 142.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9638076424598694,
      "eval_runtime": 4.2721,
      "eval_samples_per_second": 66.945,
      "eval_steps_per_second": 0.702,
      "step": 464
    },
    {
      "epoch": 144.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9598986506462097,
      "eval_runtime": 4.7252,
      "eval_samples_per_second": 60.527,
      "eval_steps_per_second": 0.635,
      "step": 468
    },
    {
      "epoch": 144.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9564277529716492,
      "eval_runtime": 4.3597,
      "eval_samples_per_second": 65.601,
      "eval_steps_per_second": 0.688,
      "step": 471
    },
    {
      "epoch": 145.85,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9549375176429749,
      "eval_runtime": 4.7228,
      "eval_samples_per_second": 60.558,
      "eval_steps_per_second": 0.635,
      "step": 474
    },
    {
      "epoch": 146.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9566131830215454,
      "eval_runtime": 4.5206,
      "eval_samples_per_second": 63.267,
      "eval_steps_per_second": 0.664,
      "step": 477
    },
    {
      "epoch": 148.0,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9600043892860413,
      "eval_runtime": 4.5711,
      "eval_samples_per_second": 62.567,
      "eval_steps_per_second": 0.656,
      "step": 481
    },
    {
      "epoch": 148.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9608636498451233,
      "eval_runtime": 4.4191,
      "eval_samples_per_second": 64.719,
      "eval_steps_per_second": 0.679,
      "step": 484
    },
    {
      "epoch": 149.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.958177387714386,
      "eval_runtime": 5.4181,
      "eval_samples_per_second": 52.786,
      "eval_steps_per_second": 0.554,
      "step": 487
    },
    {
      "epoch": 150.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9540855288505554,
      "eval_runtime": 4.3306,
      "eval_samples_per_second": 66.041,
      "eval_steps_per_second": 0.693,
      "step": 490
    },
    {
      "epoch": 152.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.955094575881958,
      "eval_runtime": 4.3124,
      "eval_samples_per_second": 66.32,
      "eval_steps_per_second": 0.696,
      "step": 494
    },
    {
      "epoch": 152.92,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9549675583839417,
      "eval_runtime": 5.2148,
      "eval_samples_per_second": 54.844,
      "eval_steps_per_second": 0.575,
      "step": 497
    },
    {
      "epoch": 153.85,
      "grad_norm": 0.44109636545181274,
      "learning_rate": 1.746031746031746e-05,
      "loss": 0.7119,
      "step": 500
    },
    {
      "epoch": 153.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9533360600471497,
      "eval_runtime": 4.9161,
      "eval_samples_per_second": 58.177,
      "eval_steps_per_second": 0.61,
      "step": 500
    },
    {
      "epoch": 154.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9527105689048767,
      "eval_runtime": 4.9344,
      "eval_samples_per_second": 57.96,
      "eval_steps_per_second": 0.608,
      "step": 503
    },
    {
      "epoch": 156.0,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9555291533470154,
      "eval_runtime": 4.0853,
      "eval_samples_per_second": 70.008,
      "eval_steps_per_second": 0.734,
      "step": 507
    },
    {
      "epoch": 156.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9557806849479675,
      "eval_runtime": 4.5929,
      "eval_samples_per_second": 62.27,
      "eval_steps_per_second": 0.653,
      "step": 510
    },
    {
      "epoch": 157.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9578056931495667,
      "eval_runtime": 4.7785,
      "eval_samples_per_second": 59.851,
      "eval_steps_per_second": 0.628,
      "step": 513
    },
    {
      "epoch": 158.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9589922428131104,
      "eval_runtime": 3.9453,
      "eval_samples_per_second": 72.491,
      "eval_steps_per_second": 0.76,
      "step": 516
    },
    {
      "epoch": 160.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9553244113922119,
      "eval_runtime": 4.0583,
      "eval_samples_per_second": 70.473,
      "eval_steps_per_second": 0.739,
      "step": 520
    },
    {
      "epoch": 160.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9509897232055664,
      "eval_runtime": 4.3977,
      "eval_samples_per_second": 65.034,
      "eval_steps_per_second": 0.682,
      "step": 523
    },
    {
      "epoch": 161.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9446966052055359,
      "eval_runtime": 4.6204,
      "eval_samples_per_second": 61.899,
      "eval_steps_per_second": 0.649,
      "step": 526
    },
    {
      "epoch": 162.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9404750466346741,
      "eval_runtime": 4.132,
      "eval_samples_per_second": 69.216,
      "eval_steps_per_second": 0.726,
      "step": 529
    },
    {
      "epoch": 164.0,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9428921341896057,
      "eval_runtime": 4.3506,
      "eval_samples_per_second": 65.738,
      "eval_steps_per_second": 0.69,
      "step": 533
    },
    {
      "epoch": 164.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9472917914390564,
      "eval_runtime": 4.0197,
      "eval_samples_per_second": 71.15,
      "eval_steps_per_second": 0.746,
      "step": 536
    },
    {
      "epoch": 165.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.952194333076477,
      "eval_runtime": 4.7031,
      "eval_samples_per_second": 60.811,
      "eval_steps_per_second": 0.638,
      "step": 539
    },
    {
      "epoch": 166.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9533052444458008,
      "eval_runtime": 4.3489,
      "eval_samples_per_second": 65.764,
      "eval_steps_per_second": 0.69,
      "step": 542
    },
    {
      "epoch": 168.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9497887492179871,
      "eval_runtime": 4.2573,
      "eval_samples_per_second": 67.178,
      "eval_steps_per_second": 0.705,
      "step": 546
    },
    {
      "epoch": 168.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9472330808639526,
      "eval_runtime": 4.4709,
      "eval_samples_per_second": 63.969,
      "eval_steps_per_second": 0.671,
      "step": 549
    },
    {
      "epoch": 169.23,
      "grad_norm": 0.8247159123420715,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.6802,
      "step": 550
    },
    {
      "epoch": 169.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9484440684318542,
      "eval_runtime": 4.8437,
      "eval_samples_per_second": 59.046,
      "eval_steps_per_second": 0.619,
      "step": 552
    },
    {
      "epoch": 170.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.948759913444519,
      "eval_runtime": 4.0675,
      "eval_samples_per_second": 70.313,
      "eval_steps_per_second": 0.738,
      "step": 555
    },
    {
      "epoch": 172.0,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9508001208305359,
      "eval_runtime": 4.1371,
      "eval_samples_per_second": 69.13,
      "eval_steps_per_second": 0.725,
      "step": 559
    },
    {
      "epoch": 172.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9550370573997498,
      "eval_runtime": 5.1562,
      "eval_samples_per_second": 55.467,
      "eval_steps_per_second": 0.582,
      "step": 562
    },
    {
      "epoch": 173.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9578031301498413,
      "eval_runtime": 5.1826,
      "eval_samples_per_second": 55.185,
      "eval_steps_per_second": 0.579,
      "step": 565
    },
    {
      "epoch": 174.77,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9607414603233337,
      "eval_runtime": 3.9547,
      "eval_samples_per_second": 72.318,
      "eval_steps_per_second": 0.759,
      "step": 568
    },
    {
      "epoch": 176.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9589574933052063,
      "eval_runtime": 4.6913,
      "eval_samples_per_second": 60.963,
      "eval_steps_per_second": 0.639,
      "step": 572
    },
    {
      "epoch": 176.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9530745148658752,
      "eval_runtime": 4.4058,
      "eval_samples_per_second": 64.915,
      "eval_steps_per_second": 0.681,
      "step": 575
    },
    {
      "epoch": 177.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9497523903846741,
      "eval_runtime": 4.7135,
      "eval_samples_per_second": 60.677,
      "eval_steps_per_second": 0.636,
      "step": 578
    },
    {
      "epoch": 178.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9497215151786804,
      "eval_runtime": 4.021,
      "eval_samples_per_second": 71.127,
      "eval_steps_per_second": 0.746,
      "step": 581
    },
    {
      "epoch": 180.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9547012448310852,
      "eval_runtime": 4.666,
      "eval_samples_per_second": 61.295,
      "eval_steps_per_second": 0.643,
      "step": 585
    },
    {
      "epoch": 180.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9555142521858215,
      "eval_runtime": 4.1238,
      "eval_samples_per_second": 69.353,
      "eval_steps_per_second": 0.727,
      "step": 588
    },
    {
      "epoch": 181.85,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9561424255371094,
      "eval_runtime": 4.7157,
      "eval_samples_per_second": 60.648,
      "eval_steps_per_second": 0.636,
      "step": 591
    },
    {
      "epoch": 182.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9556129574775696,
      "eval_runtime": 4.9472,
      "eval_samples_per_second": 57.81,
      "eval_steps_per_second": 0.606,
      "step": 594
    },
    {
      "epoch": 184.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9521645307540894,
      "eval_runtime": 4.7357,
      "eval_samples_per_second": 60.393,
      "eval_steps_per_second": 0.633,
      "step": 598
    },
    {
      "epoch": 184.62,
      "grad_norm": 0.5078141689300537,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 0.6609,
      "step": 600
    },
    {
      "epoch": 184.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9505414366722107,
      "eval_runtime": 4.6852,
      "eval_samples_per_second": 61.043,
      "eval_steps_per_second": 0.64,
      "step": 601
    },
    {
      "epoch": 185.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9508634805679321,
      "eval_runtime": 4.5463,
      "eval_samples_per_second": 62.908,
      "eval_steps_per_second": 0.66,
      "step": 604
    },
    {
      "epoch": 186.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9513095021247864,
      "eval_runtime": 4.6608,
      "eval_samples_per_second": 61.363,
      "eval_steps_per_second": 0.644,
      "step": 607
    },
    {
      "epoch": 188.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9521496891975403,
      "eval_runtime": 4.2761,
      "eval_samples_per_second": 66.883,
      "eval_steps_per_second": 0.702,
      "step": 611
    },
    {
      "epoch": 188.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9504989981651306,
      "eval_runtime": 4.2369,
      "eval_samples_per_second": 67.502,
      "eval_steps_per_second": 0.708,
      "step": 614
    },
    {
      "epoch": 189.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9491674304008484,
      "eval_runtime": 4.201,
      "eval_samples_per_second": 68.079,
      "eval_steps_per_second": 0.714,
      "step": 617
    },
    {
      "epoch": 190.77,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9477587938308716,
      "eval_runtime": 5.2893,
      "eval_samples_per_second": 54.071,
      "eval_steps_per_second": 0.567,
      "step": 620
    },
    {
      "epoch": 192.0,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9457892775535583,
      "eval_runtime": 4.1359,
      "eval_samples_per_second": 69.151,
      "eval_steps_per_second": 0.725,
      "step": 624
    },
    {
      "epoch": 192.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9427019953727722,
      "eval_runtime": 4.8485,
      "eval_samples_per_second": 58.987,
      "eval_steps_per_second": 0.619,
      "step": 627
    },
    {
      "epoch": 193.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9433549046516418,
      "eval_runtime": 4.4182,
      "eval_samples_per_second": 64.732,
      "eval_steps_per_second": 0.679,
      "step": 630
    },
    {
      "epoch": 194.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9444069862365723,
      "eval_runtime": 4.503,
      "eval_samples_per_second": 63.513,
      "eval_steps_per_second": 0.666,
      "step": 633
    },
    {
      "epoch": 196.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9477455615997314,
      "eval_runtime": 4.7047,
      "eval_samples_per_second": 60.79,
      "eval_steps_per_second": 0.638,
      "step": 637
    },
    {
      "epoch": 196.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.948028564453125,
      "eval_runtime": 4.6368,
      "eval_samples_per_second": 61.681,
      "eval_steps_per_second": 0.647,
      "step": 640
    },
    {
      "epoch": 197.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9453917741775513,
      "eval_runtime": 4.5374,
      "eval_samples_per_second": 63.032,
      "eval_steps_per_second": 0.661,
      "step": 643
    },
    {
      "epoch": 198.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.944375216960907,
      "eval_runtime": 4.3393,
      "eval_samples_per_second": 65.909,
      "eval_steps_per_second": 0.691,
      "step": 646
    },
    {
      "epoch": 200.0,
      "grad_norm": 0.6592119336128235,
      "learning_rate": 1.2698412698412699e-05,
      "loss": 0.6402,
      "step": 650
    },
    {
      "epoch": 200.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9393660426139832,
      "eval_runtime": 4.2917,
      "eval_samples_per_second": 66.64,
      "eval_steps_per_second": 0.699,
      "step": 650
    },
    {
      "epoch": 200.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9392969012260437,
      "eval_runtime": 4.6663,
      "eval_samples_per_second": 61.291,
      "eval_steps_per_second": 0.643,
      "step": 653
    },
    {
      "epoch": 201.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9408543705940247,
      "eval_runtime": 3.7675,
      "eval_samples_per_second": 75.912,
      "eval_steps_per_second": 0.796,
      "step": 656
    },
    {
      "epoch": 202.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9433931112289429,
      "eval_runtime": 4.0263,
      "eval_samples_per_second": 71.034,
      "eval_steps_per_second": 0.745,
      "step": 659
    },
    {
      "epoch": 204.0,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9422336220741272,
      "eval_runtime": 5.0827,
      "eval_samples_per_second": 56.27,
      "eval_steps_per_second": 0.59,
      "step": 663
    },
    {
      "epoch": 204.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9422111511230469,
      "eval_runtime": 4.3449,
      "eval_samples_per_second": 65.824,
      "eval_steps_per_second": 0.69,
      "step": 666
    },
    {
      "epoch": 205.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9415052533149719,
      "eval_runtime": 4.3877,
      "eval_samples_per_second": 65.183,
      "eval_steps_per_second": 0.684,
      "step": 669
    },
    {
      "epoch": 206.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9403449892997742,
      "eval_runtime": 4.5384,
      "eval_samples_per_second": 63.018,
      "eval_steps_per_second": 0.661,
      "step": 672
    },
    {
      "epoch": 208.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9443973898887634,
      "eval_runtime": 4.8712,
      "eval_samples_per_second": 58.713,
      "eval_steps_per_second": 0.616,
      "step": 676
    },
    {
      "epoch": 208.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9434468746185303,
      "eval_runtime": 4.5262,
      "eval_samples_per_second": 63.188,
      "eval_steps_per_second": 0.663,
      "step": 679
    },
    {
      "epoch": 209.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9392986297607422,
      "eval_runtime": 3.8813,
      "eval_samples_per_second": 73.687,
      "eval_steps_per_second": 0.773,
      "step": 682
    },
    {
      "epoch": 210.77,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9383959174156189,
      "eval_runtime": 4.0949,
      "eval_samples_per_second": 69.843,
      "eval_steps_per_second": 0.733,
      "step": 685
    },
    {
      "epoch": 212.0,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9405562281608582,
      "eval_runtime": 4.0071,
      "eval_samples_per_second": 71.374,
      "eval_steps_per_second": 0.749,
      "step": 689
    },
    {
      "epoch": 212.92,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9427996873855591,
      "eval_runtime": 4.4042,
      "eval_samples_per_second": 64.938,
      "eval_steps_per_second": 0.681,
      "step": 692
    },
    {
      "epoch": 213.85,
      "eval_accuracy": 0.6573426573426573,
      "eval_loss": 0.9420264363288879,
      "eval_runtime": 5.0628,
      "eval_samples_per_second": 56.491,
      "eval_steps_per_second": 0.593,
      "step": 695
    },
    {
      "epoch": 214.77,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9402673840522766,
      "eval_runtime": 4.5344,
      "eval_samples_per_second": 63.073,
      "eval_steps_per_second": 0.662,
      "step": 698
    },
    {
      "epoch": 215.38,
      "grad_norm": 0.41818705201148987,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.632,
      "step": 700
    },
    {
      "epoch": 216.0,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.939588189125061,
      "eval_runtime": 4.3905,
      "eval_samples_per_second": 65.141,
      "eval_steps_per_second": 0.683,
      "step": 702
    },
    {
      "epoch": 216.92,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9378303289413452,
      "eval_runtime": 3.949,
      "eval_samples_per_second": 72.423,
      "eval_steps_per_second": 0.76,
      "step": 705
    },
    {
      "epoch": 217.85,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9359939694404602,
      "eval_runtime": 6.0391,
      "eval_samples_per_second": 47.358,
      "eval_steps_per_second": 0.497,
      "step": 708
    },
    {
      "epoch": 218.77,
      "eval_accuracy": 0.6608391608391608,
      "eval_loss": 0.9351831078529358,
      "eval_runtime": 3.9064,
      "eval_samples_per_second": 73.213,
      "eval_steps_per_second": 0.768,
      "step": 711
    },
    {
      "epoch": 220.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9344322085380554,
      "eval_runtime": 4.2696,
      "eval_samples_per_second": 66.986,
      "eval_steps_per_second": 0.703,
      "step": 715
    },
    {
      "epoch": 220.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9372223615646362,
      "eval_runtime": 4.3125,
      "eval_samples_per_second": 66.319,
      "eval_steps_per_second": 0.696,
      "step": 718
    },
    {
      "epoch": 221.85,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9403982162475586,
      "eval_runtime": 5.843,
      "eval_samples_per_second": 48.948,
      "eval_steps_per_second": 0.513,
      "step": 721
    },
    {
      "epoch": 222.77,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9428540468215942,
      "eval_runtime": 4.6219,
      "eval_samples_per_second": 61.879,
      "eval_steps_per_second": 0.649,
      "step": 724
    },
    {
      "epoch": 224.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9426971673965454,
      "eval_runtime": 4.3056,
      "eval_samples_per_second": 66.425,
      "eval_steps_per_second": 0.697,
      "step": 728
    },
    {
      "epoch": 224.92,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9425584673881531,
      "eval_runtime": 4.1148,
      "eval_samples_per_second": 69.506,
      "eval_steps_per_second": 0.729,
      "step": 731
    },
    {
      "epoch": 225.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9411899447441101,
      "eval_runtime": 5.1345,
      "eval_samples_per_second": 55.702,
      "eval_steps_per_second": 0.584,
      "step": 734
    },
    {
      "epoch": 226.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9402187466621399,
      "eval_runtime": 4.509,
      "eval_samples_per_second": 63.428,
      "eval_steps_per_second": 0.665,
      "step": 737
    },
    {
      "epoch": 228.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9380680918693542,
      "eval_runtime": 4.4438,
      "eval_samples_per_second": 64.359,
      "eval_steps_per_second": 0.675,
      "step": 741
    },
    {
      "epoch": 228.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9378944635391235,
      "eval_runtime": 4.7137,
      "eval_samples_per_second": 60.674,
      "eval_steps_per_second": 0.636,
      "step": 744
    },
    {
      "epoch": 229.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9393661618232727,
      "eval_runtime": 4.3516,
      "eval_samples_per_second": 65.723,
      "eval_steps_per_second": 0.689,
      "step": 747
    },
    {
      "epoch": 230.77,
      "grad_norm": 0.43979376554489136,
      "learning_rate": 9.523809523809523e-06,
      "loss": 0.6285,
      "step": 750
    },
    {
      "epoch": 230.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9395984411239624,
      "eval_runtime": 4.7533,
      "eval_samples_per_second": 60.169,
      "eval_steps_per_second": 0.631,
      "step": 750
    },
    {
      "epoch": 232.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.943848729133606,
      "eval_runtime": 4.0413,
      "eval_samples_per_second": 70.769,
      "eval_steps_per_second": 0.742,
      "step": 754
    },
    {
      "epoch": 232.92,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9464450478553772,
      "eval_runtime": 4.5457,
      "eval_samples_per_second": 62.916,
      "eval_steps_per_second": 0.66,
      "step": 757
    },
    {
      "epoch": 233.85,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9500763416290283,
      "eval_runtime": 4.0761,
      "eval_samples_per_second": 70.166,
      "eval_steps_per_second": 0.736,
      "step": 760
    },
    {
      "epoch": 234.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9517615437507629,
      "eval_runtime": 4.7434,
      "eval_samples_per_second": 60.295,
      "eval_steps_per_second": 0.632,
      "step": 763
    },
    {
      "epoch": 236.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9502771496772766,
      "eval_runtime": 4.4044,
      "eval_samples_per_second": 64.935,
      "eval_steps_per_second": 0.681,
      "step": 767
    },
    {
      "epoch": 236.92,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9494647979736328,
      "eval_runtime": 4.5179,
      "eval_samples_per_second": 63.304,
      "eval_steps_per_second": 0.664,
      "step": 770
    },
    {
      "epoch": 237.85,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9487398862838745,
      "eval_runtime": 5.4124,
      "eval_samples_per_second": 52.842,
      "eval_steps_per_second": 0.554,
      "step": 773
    },
    {
      "epoch": 238.77,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9492418169975281,
      "eval_runtime": 4.4487,
      "eval_samples_per_second": 64.289,
      "eval_steps_per_second": 0.674,
      "step": 776
    },
    {
      "epoch": 240.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9463514685630798,
      "eval_runtime": 4.2462,
      "eval_samples_per_second": 67.354,
      "eval_steps_per_second": 0.707,
      "step": 780
    },
    {
      "epoch": 240.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9432539343833923,
      "eval_runtime": 4.4574,
      "eval_samples_per_second": 64.163,
      "eval_steps_per_second": 0.673,
      "step": 783
    },
    {
      "epoch": 241.85,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9403123259544373,
      "eval_runtime": 4.1413,
      "eval_samples_per_second": 69.06,
      "eval_steps_per_second": 0.724,
      "step": 786
    },
    {
      "epoch": 242.77,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9371182322502136,
      "eval_runtime": 4.0874,
      "eval_samples_per_second": 69.971,
      "eval_steps_per_second": 0.734,
      "step": 789
    },
    {
      "epoch": 244.0,
      "eval_accuracy": 0.6643356643356644,
      "eval_loss": 0.9386875033378601,
      "eval_runtime": 4.427,
      "eval_samples_per_second": 64.603,
      "eval_steps_per_second": 0.678,
      "step": 793
    },
    {
      "epoch": 244.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9423165321350098,
      "eval_runtime": 4.5541,
      "eval_samples_per_second": 62.8,
      "eval_steps_per_second": 0.659,
      "step": 796
    },
    {
      "epoch": 245.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9452247619628906,
      "eval_runtime": 4.3478,
      "eval_samples_per_second": 65.78,
      "eval_steps_per_second": 0.69,
      "step": 799
    },
    {
      "epoch": 246.15,
      "grad_norm": 0.8526151776313782,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.6049,
      "step": 800
    },
    {
      "epoch": 246.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9474557638168335,
      "eval_runtime": 4.2324,
      "eval_samples_per_second": 67.574,
      "eval_steps_per_second": 0.709,
      "step": 802
    },
    {
      "epoch": 248.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9469492435455322,
      "eval_runtime": 4.2172,
      "eval_samples_per_second": 67.817,
      "eval_steps_per_second": 0.711,
      "step": 806
    },
    {
      "epoch": 248.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9462881088256836,
      "eval_runtime": 3.7956,
      "eval_samples_per_second": 75.351,
      "eval_steps_per_second": 0.79,
      "step": 809
    },
    {
      "epoch": 249.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9461723566055298,
      "eval_runtime": 4.331,
      "eval_samples_per_second": 66.036,
      "eval_steps_per_second": 0.693,
      "step": 812
    },
    {
      "epoch": 250.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.945976972579956,
      "eval_runtime": 5.0364,
      "eval_samples_per_second": 56.787,
      "eval_steps_per_second": 0.596,
      "step": 815
    },
    {
      "epoch": 252.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9463274478912354,
      "eval_runtime": 4.1886,
      "eval_samples_per_second": 68.28,
      "eval_steps_per_second": 0.716,
      "step": 819
    },
    {
      "epoch": 252.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9467651844024658,
      "eval_runtime": 4.3006,
      "eval_samples_per_second": 66.502,
      "eval_steps_per_second": 0.698,
      "step": 822
    },
    {
      "epoch": 253.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9466995596885681,
      "eval_runtime": 4.7073,
      "eval_samples_per_second": 60.757,
      "eval_steps_per_second": 0.637,
      "step": 825
    },
    {
      "epoch": 254.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9466331005096436,
      "eval_runtime": 4.2578,
      "eval_samples_per_second": 67.171,
      "eval_steps_per_second": 0.705,
      "step": 828
    },
    {
      "epoch": 256.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9450873732566833,
      "eval_runtime": 4.0297,
      "eval_samples_per_second": 70.973,
      "eval_steps_per_second": 0.744,
      "step": 832
    },
    {
      "epoch": 256.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9441204071044922,
      "eval_runtime": 4.3682,
      "eval_samples_per_second": 65.474,
      "eval_steps_per_second": 0.687,
      "step": 835
    },
    {
      "epoch": 257.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9426379203796387,
      "eval_runtime": 4.5099,
      "eval_samples_per_second": 63.417,
      "eval_steps_per_second": 0.665,
      "step": 838
    },
    {
      "epoch": 258.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9439011216163635,
      "eval_runtime": 4.1513,
      "eval_samples_per_second": 68.894,
      "eval_steps_per_second": 0.723,
      "step": 841
    },
    {
      "epoch": 260.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9443875551223755,
      "eval_runtime": 4.5454,
      "eval_samples_per_second": 62.921,
      "eval_steps_per_second": 0.66,
      "step": 845
    },
    {
      "epoch": 260.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9435145258903503,
      "eval_runtime": 4.6513,
      "eval_samples_per_second": 61.489,
      "eval_steps_per_second": 0.645,
      "step": 848
    },
    {
      "epoch": 261.54,
      "grad_norm": 0.46575862169265747,
      "learning_rate": 6.349206349206349e-06,
      "loss": 0.6024,
      "step": 850
    },
    {
      "epoch": 261.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9441516399383545,
      "eval_runtime": 4.3022,
      "eval_samples_per_second": 66.478,
      "eval_steps_per_second": 0.697,
      "step": 851
    },
    {
      "epoch": 262.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9440596699714661,
      "eval_runtime": 4.2068,
      "eval_samples_per_second": 67.986,
      "eval_steps_per_second": 0.713,
      "step": 854
    },
    {
      "epoch": 264.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9449206590652466,
      "eval_runtime": 3.6546,
      "eval_samples_per_second": 78.258,
      "eval_steps_per_second": 0.821,
      "step": 858
    },
    {
      "epoch": 264.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9438286423683167,
      "eval_runtime": 4.4082,
      "eval_samples_per_second": 64.879,
      "eval_steps_per_second": 0.681,
      "step": 861
    },
    {
      "epoch": 265.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9423307180404663,
      "eval_runtime": 5.0619,
      "eval_samples_per_second": 56.5,
      "eval_steps_per_second": 0.593,
      "step": 864
    },
    {
      "epoch": 266.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9406384229660034,
      "eval_runtime": 3.9231,
      "eval_samples_per_second": 72.902,
      "eval_steps_per_second": 0.765,
      "step": 867
    },
    {
      "epoch": 268.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9399996399879456,
      "eval_runtime": 4.3437,
      "eval_samples_per_second": 65.843,
      "eval_steps_per_second": 0.691,
      "step": 871
    },
    {
      "epoch": 268.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9407250285148621,
      "eval_runtime": 4.1958,
      "eval_samples_per_second": 68.163,
      "eval_steps_per_second": 0.715,
      "step": 874
    },
    {
      "epoch": 269.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9427564740180969,
      "eval_runtime": 4.1649,
      "eval_samples_per_second": 68.668,
      "eval_steps_per_second": 0.72,
      "step": 877
    },
    {
      "epoch": 270.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9453551769256592,
      "eval_runtime": 4.1929,
      "eval_samples_per_second": 68.21,
      "eval_steps_per_second": 0.715,
      "step": 880
    },
    {
      "epoch": 272.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9466239213943481,
      "eval_runtime": 4.787,
      "eval_samples_per_second": 59.745,
      "eval_steps_per_second": 0.627,
      "step": 884
    },
    {
      "epoch": 272.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9471932649612427,
      "eval_runtime": 5.1674,
      "eval_samples_per_second": 55.347,
      "eval_steps_per_second": 0.581,
      "step": 887
    },
    {
      "epoch": 273.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9461983442306519,
      "eval_runtime": 4.2961,
      "eval_samples_per_second": 66.572,
      "eval_steps_per_second": 0.698,
      "step": 890
    },
    {
      "epoch": 274.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9463549852371216,
      "eval_runtime": 4.214,
      "eval_samples_per_second": 67.869,
      "eval_steps_per_second": 0.712,
      "step": 893
    },
    {
      "epoch": 276.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9452849626541138,
      "eval_runtime": 4.7011,
      "eval_samples_per_second": 60.836,
      "eval_steps_per_second": 0.638,
      "step": 897
    },
    {
      "epoch": 276.92,
      "grad_norm": 0.5298266410827637,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.5966,
      "step": 900
    },
    {
      "epoch": 276.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.943517804145813,
      "eval_runtime": 4.7023,
      "eval_samples_per_second": 60.821,
      "eval_steps_per_second": 0.638,
      "step": 900
    },
    {
      "epoch": 277.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9417858123779297,
      "eval_runtime": 4.8013,
      "eval_samples_per_second": 59.568,
      "eval_steps_per_second": 0.625,
      "step": 903
    },
    {
      "epoch": 278.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9401264190673828,
      "eval_runtime": 4.2146,
      "eval_samples_per_second": 67.859,
      "eval_steps_per_second": 0.712,
      "step": 906
    },
    {
      "epoch": 280.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9374663829803467,
      "eval_runtime": 4.5877,
      "eval_samples_per_second": 62.34,
      "eval_steps_per_second": 0.654,
      "step": 910
    },
    {
      "epoch": 280.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9365966320037842,
      "eval_runtime": 4.7172,
      "eval_samples_per_second": 60.629,
      "eval_steps_per_second": 0.636,
      "step": 913
    },
    {
      "epoch": 281.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9357553720474243,
      "eval_runtime": 4.1128,
      "eval_samples_per_second": 69.539,
      "eval_steps_per_second": 0.729,
      "step": 916
    },
    {
      "epoch": 282.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9363517761230469,
      "eval_runtime": 4.5892,
      "eval_samples_per_second": 62.321,
      "eval_steps_per_second": 0.654,
      "step": 919
    },
    {
      "epoch": 284.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9368854761123657,
      "eval_runtime": 4.6233,
      "eval_samples_per_second": 61.86,
      "eval_steps_per_second": 0.649,
      "step": 923
    },
    {
      "epoch": 284.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9384199976921082,
      "eval_runtime": 4.5974,
      "eval_samples_per_second": 62.21,
      "eval_steps_per_second": 0.653,
      "step": 926
    },
    {
      "epoch": 285.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9411068558692932,
      "eval_runtime": 4.3443,
      "eval_samples_per_second": 65.834,
      "eval_steps_per_second": 0.691,
      "step": 929
    },
    {
      "epoch": 286.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9424038529396057,
      "eval_runtime": 4.1488,
      "eval_samples_per_second": 68.936,
      "eval_steps_per_second": 0.723,
      "step": 932
    },
    {
      "epoch": 288.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9443323016166687,
      "eval_runtime": 3.949,
      "eval_samples_per_second": 72.423,
      "eval_steps_per_second": 0.76,
      "step": 936
    },
    {
      "epoch": 288.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.945092499256134,
      "eval_runtime": 5.4756,
      "eval_samples_per_second": 52.232,
      "eval_steps_per_second": 0.548,
      "step": 939
    },
    {
      "epoch": 289.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9460843801498413,
      "eval_runtime": 4.3265,
      "eval_samples_per_second": 66.104,
      "eval_steps_per_second": 0.693,
      "step": 942
    },
    {
      "epoch": 290.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9465355277061462,
      "eval_runtime": 4.0418,
      "eval_samples_per_second": 70.761,
      "eval_steps_per_second": 0.742,
      "step": 945
    },
    {
      "epoch": 292.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9477852582931519,
      "eval_runtime": 4.5983,
      "eval_samples_per_second": 62.196,
      "eval_steps_per_second": 0.652,
      "step": 949
    },
    {
      "epoch": 292.31,
      "grad_norm": 0.48691806197166443,
      "learning_rate": 3.1746031746031746e-06,
      "loss": 0.5841,
      "step": 950
    },
    {
      "epoch": 292.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9479745626449585,
      "eval_runtime": 4.8468,
      "eval_samples_per_second": 59.008,
      "eval_steps_per_second": 0.619,
      "step": 952
    },
    {
      "epoch": 293.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9476760029792786,
      "eval_runtime": 4.2654,
      "eval_samples_per_second": 67.051,
      "eval_steps_per_second": 0.703,
      "step": 955
    },
    {
      "epoch": 294.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9466102719306946,
      "eval_runtime": 5.1835,
      "eval_samples_per_second": 55.175,
      "eval_steps_per_second": 0.579,
      "step": 958
    },
    {
      "epoch": 296.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9453703165054321,
      "eval_runtime": 4.2033,
      "eval_samples_per_second": 68.041,
      "eval_steps_per_second": 0.714,
      "step": 962
    },
    {
      "epoch": 296.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9448667764663696,
      "eval_runtime": 5.1456,
      "eval_samples_per_second": 55.582,
      "eval_steps_per_second": 0.583,
      "step": 965
    },
    {
      "epoch": 297.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.944092333316803,
      "eval_runtime": 4.8501,
      "eval_samples_per_second": 58.968,
      "eval_steps_per_second": 0.619,
      "step": 968
    },
    {
      "epoch": 298.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9439365267753601,
      "eval_runtime": 4.6304,
      "eval_samples_per_second": 61.765,
      "eval_steps_per_second": 0.648,
      "step": 971
    },
    {
      "epoch": 300.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9432730674743652,
      "eval_runtime": 4.2353,
      "eval_samples_per_second": 67.527,
      "eval_steps_per_second": 0.708,
      "step": 975
    },
    {
      "epoch": 300.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9432761669158936,
      "eval_runtime": 5.7923,
      "eval_samples_per_second": 49.376,
      "eval_steps_per_second": 0.518,
      "step": 978
    },
    {
      "epoch": 301.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9426748752593994,
      "eval_runtime": 5.2861,
      "eval_samples_per_second": 54.104,
      "eval_steps_per_second": 0.568,
      "step": 981
    },
    {
      "epoch": 302.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9422904253005981,
      "eval_runtime": 4.6604,
      "eval_samples_per_second": 61.368,
      "eval_steps_per_second": 0.644,
      "step": 984
    },
    {
      "epoch": 304.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.941644012928009,
      "eval_runtime": 4.5261,
      "eval_samples_per_second": 63.19,
      "eval_steps_per_second": 0.663,
      "step": 988
    },
    {
      "epoch": 304.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9412463903427124,
      "eval_runtime": 3.7658,
      "eval_samples_per_second": 75.946,
      "eval_steps_per_second": 0.797,
      "step": 991
    },
    {
      "epoch": 305.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9411738514900208,
      "eval_runtime": 4.5896,
      "eval_samples_per_second": 62.315,
      "eval_steps_per_second": 0.654,
      "step": 994
    },
    {
      "epoch": 306.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9409748315811157,
      "eval_runtime": 5.0951,
      "eval_samples_per_second": 56.133,
      "eval_steps_per_second": 0.589,
      "step": 997
    },
    {
      "epoch": 307.69,
      "grad_norm": 0.4341520071029663,
      "learning_rate": 1.5873015873015873e-06,
      "loss": 0.5913,
      "step": 1000
    },
    {
      "epoch": 308.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9409425258636475,
      "eval_runtime": 4.2421,
      "eval_samples_per_second": 67.419,
      "eval_steps_per_second": 0.707,
      "step": 1001
    },
    {
      "epoch": 308.92,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9412462115287781,
      "eval_runtime": 5.4761,
      "eval_samples_per_second": 52.227,
      "eval_steps_per_second": 0.548,
      "step": 1004
    },
    {
      "epoch": 309.85,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9415078163146973,
      "eval_runtime": 4.3021,
      "eval_samples_per_second": 66.479,
      "eval_steps_per_second": 0.697,
      "step": 1007
    },
    {
      "epoch": 310.77,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9418912529945374,
      "eval_runtime": 4.6753,
      "eval_samples_per_second": 61.173,
      "eval_steps_per_second": 0.642,
      "step": 1010
    },
    {
      "epoch": 312.0,
      "eval_accuracy": 0.6713286713286714,
      "eval_loss": 0.9425641298294067,
      "eval_runtime": 4.2216,
      "eval_samples_per_second": 67.746,
      "eval_steps_per_second": 0.711,
      "step": 1014
    },
    {
      "epoch": 312.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9430021047592163,
      "eval_runtime": 4.5627,
      "eval_samples_per_second": 62.682,
      "eval_steps_per_second": 0.658,
      "step": 1017
    },
    {
      "epoch": 313.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9434046149253845,
      "eval_runtime": 4.4071,
      "eval_samples_per_second": 64.896,
      "eval_steps_per_second": 0.681,
      "step": 1020
    },
    {
      "epoch": 314.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9436425566673279,
      "eval_runtime": 4.4183,
      "eval_samples_per_second": 64.731,
      "eval_steps_per_second": 0.679,
      "step": 1023
    },
    {
      "epoch": 316.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9438886046409607,
      "eval_runtime": 5.0213,
      "eval_samples_per_second": 56.958,
      "eval_steps_per_second": 0.597,
      "step": 1027
    },
    {
      "epoch": 316.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9438981413841248,
      "eval_runtime": 4.4087,
      "eval_samples_per_second": 64.872,
      "eval_steps_per_second": 0.68,
      "step": 1030
    },
    {
      "epoch": 317.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9438797831535339,
      "eval_runtime": 4.2787,
      "eval_samples_per_second": 66.843,
      "eval_steps_per_second": 0.701,
      "step": 1033
    },
    {
      "epoch": 318.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9440018534660339,
      "eval_runtime": 4.5412,
      "eval_samples_per_second": 62.979,
      "eval_steps_per_second": 0.661,
      "step": 1036
    },
    {
      "epoch": 320.0,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9439986348152161,
      "eval_runtime": 5.0195,
      "eval_samples_per_second": 56.978,
      "eval_steps_per_second": 0.598,
      "step": 1040
    },
    {
      "epoch": 320.92,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.944036602973938,
      "eval_runtime": 4.4758,
      "eval_samples_per_second": 63.899,
      "eval_steps_per_second": 0.67,
      "step": 1043
    },
    {
      "epoch": 321.85,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9441185593605042,
      "eval_runtime": 6.0328,
      "eval_samples_per_second": 47.408,
      "eval_steps_per_second": 0.497,
      "step": 1046
    },
    {
      "epoch": 322.77,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9441850781440735,
      "eval_runtime": 4.7198,
      "eval_samples_per_second": 60.596,
      "eval_steps_per_second": 0.636,
      "step": 1049
    },
    {
      "epoch": 323.08,
      "grad_norm": 0.7628626823425293,
      "learning_rate": 0.0,
      "loss": 0.5798,
      "step": 1050
    },
    {
      "epoch": 323.08,
      "eval_accuracy": 0.6678321678321678,
      "eval_loss": 0.9441931843757629,
      "eval_runtime": 4.692,
      "eval_samples_per_second": 60.955,
      "eval_steps_per_second": 0.639,
      "step": 1050
    },
    {
      "epoch": 323.08,
      "step": 1050,
      "total_flos": 4.380490432252032e+18,
      "train_loss": 0.8572167641775948,
      "train_runtime": 4856.1657,
      "train_samples_per_second": 111.57,
      "train_steps_per_second": 0.216
    }
  ],
  "logging_steps": 50,
  "max_steps": 1050,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 350,
  "save_steps": 500,
  "total_flos": 4.380490432252032e+18,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}