{
  "best_metric": 0.6538461538461539,
  "best_model_checkpoint": "wav2vec2-5Class-train-test-finetune-V7/checkpoint-552",
  "epoch": 230.76923076923077,
  "eval_steps": 500,
  "global_step": 750,
  "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": 1.6025639772415161,
      "eval_runtime": 25.1775,
      "eval_samples_per_second": 11.359,
      "eval_steps_per_second": 0.119,
      "step": 3
    },
    {
      "epoch": 1.85,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6023768186569214,
      "eval_runtime": 13.2244,
      "eval_samples_per_second": 21.627,
      "eval_steps_per_second": 0.227,
      "step": 6
    },
    {
      "epoch": 2.77,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6020615100860596,
      "eval_runtime": 19.1459,
      "eval_samples_per_second": 14.938,
      "eval_steps_per_second": 0.157,
      "step": 9
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6014316082000732,
      "eval_runtime": 24.6871,
      "eval_samples_per_second": 11.585,
      "eval_steps_per_second": 0.122,
      "step": 13
    },
    {
      "epoch": 4.92,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.6007977724075317,
      "eval_runtime": 14.5871,
      "eval_samples_per_second": 19.606,
      "eval_steps_per_second": 0.206,
      "step": 16
    },
    {
      "epoch": 5.85,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.600056529045105,
      "eval_runtime": 15.3427,
      "eval_samples_per_second": 18.641,
      "eval_steps_per_second": 0.196,
      "step": 19
    },
    {
      "epoch": 6.77,
      "eval_accuracy": 0.16083916083916083,
      "eval_loss": 1.599173903465271,
      "eval_runtime": 23.1969,
      "eval_samples_per_second": 12.329,
      "eval_steps_per_second": 0.129,
      "step": 22
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.17482517482517482,
      "eval_loss": 1.5977518558502197,
      "eval_runtime": 24.249,
      "eval_samples_per_second": 11.794,
      "eval_steps_per_second": 0.124,
      "step": 26
    },
    {
      "epoch": 8.92,
      "eval_accuracy": 0.1888111888111888,
      "eval_loss": 1.5965358018875122,
      "eval_runtime": 14.7479,
      "eval_samples_per_second": 19.393,
      "eval_steps_per_second": 0.203,
      "step": 29
    },
    {
      "epoch": 9.85,
      "eval_accuracy": 0.2097902097902098,
      "eval_loss": 1.5952231884002686,
      "eval_runtime": 18.4251,
      "eval_samples_per_second": 15.522,
      "eval_steps_per_second": 0.163,
      "step": 32
    },
    {
      "epoch": 10.77,
      "eval_accuracy": 0.22727272727272727,
      "eval_loss": 1.593780517578125,
      "eval_runtime": 20.5455,
      "eval_samples_per_second": 13.92,
      "eval_steps_per_second": 0.146,
      "step": 35
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.23426573426573427,
      "eval_loss": 1.5916367769241333,
      "eval_runtime": 16.0385,
      "eval_samples_per_second": 17.832,
      "eval_steps_per_second": 0.187,
      "step": 39
    },
    {
      "epoch": 12.92,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.5898847579956055,
      "eval_runtime": 21.1768,
      "eval_samples_per_second": 13.505,
      "eval_steps_per_second": 0.142,
      "step": 42
    },
    {
      "epoch": 13.85,
      "eval_accuracy": 0.2727272727272727,
      "eval_loss": 1.5879966020584106,
      "eval_runtime": 13.5248,
      "eval_samples_per_second": 21.146,
      "eval_steps_per_second": 0.222,
      "step": 45
    },
    {
      "epoch": 14.77,
      "eval_accuracy": 0.3076923076923077,
      "eval_loss": 1.5860111713409424,
      "eval_runtime": 12.9444,
      "eval_samples_per_second": 22.094,
      "eval_steps_per_second": 0.232,
      "step": 48
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.35664335664335667,
      "eval_loss": 1.583256483078003,
      "eval_runtime": 25.0161,
      "eval_samples_per_second": 11.433,
      "eval_steps_per_second": 0.12,
      "step": 52
    },
    {
      "epoch": 16.92,
      "eval_accuracy": 0.3881118881118881,
      "eval_loss": 1.5810612440109253,
      "eval_runtime": 14.5666,
      "eval_samples_per_second": 19.634,
      "eval_steps_per_second": 0.206,
      "step": 55
    },
    {
      "epoch": 17.85,
      "eval_accuracy": 0.3811188811188811,
      "eval_loss": 1.578792929649353,
      "eval_runtime": 13.0625,
      "eval_samples_per_second": 21.895,
      "eval_steps_per_second": 0.23,
      "step": 58
    },
    {
      "epoch": 18.77,
      "eval_accuracy": 0.36713286713286714,
      "eval_loss": 1.576446294784546,
      "eval_runtime": 18.6648,
      "eval_samples_per_second": 15.323,
      "eval_steps_per_second": 0.161,
      "step": 61
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.34965034965034963,
      "eval_loss": 1.5730595588684082,
      "eval_runtime": 11.5386,
      "eval_samples_per_second": 24.786,
      "eval_steps_per_second": 0.26,
      "step": 65
    },
    {
      "epoch": 20.92,
      "eval_accuracy": 0.32867132867132864,
      "eval_loss": 1.5702308416366577,
      "eval_runtime": 15.7041,
      "eval_samples_per_second": 18.212,
      "eval_steps_per_second": 0.191,
      "step": 68
    },
    {
      "epoch": 21.85,
      "eval_accuracy": 0.32517482517482516,
      "eval_loss": 1.5671954154968262,
      "eval_runtime": 11.3961,
      "eval_samples_per_second": 25.096,
      "eval_steps_per_second": 0.263,
      "step": 71
    },
    {
      "epoch": 22.77,
      "eval_accuracy": 0.3146853146853147,
      "eval_loss": 1.5641109943389893,
      "eval_runtime": 15.1673,
      "eval_samples_per_second": 18.856,
      "eval_steps_per_second": 0.198,
      "step": 74
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.3111888111888112,
      "eval_loss": 1.5597317218780518,
      "eval_runtime": 24.7249,
      "eval_samples_per_second": 11.567,
      "eval_steps_per_second": 0.121,
      "step": 78
    },
    {
      "epoch": 24.92,
      "eval_accuracy": 0.3076923076923077,
      "eval_loss": 1.5563873052597046,
      "eval_runtime": 13.1713,
      "eval_samples_per_second": 21.714,
      "eval_steps_per_second": 0.228,
      "step": 81
    },
    {
      "epoch": 25.85,
      "eval_accuracy": 0.3041958041958042,
      "eval_loss": 1.553160548210144,
      "eval_runtime": 12.194,
      "eval_samples_per_second": 23.454,
      "eval_steps_per_second": 0.246,
      "step": 84
    },
    {
      "epoch": 26.77,
      "eval_accuracy": 0.2937062937062937,
      "eval_loss": 1.549930214881897,
      "eval_runtime": 25.3259,
      "eval_samples_per_second": 11.293,
      "eval_steps_per_second": 0.118,
      "step": 87
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.2902097902097902,
      "eval_loss": 1.5454081296920776,
      "eval_runtime": 10.4422,
      "eval_samples_per_second": 27.389,
      "eval_steps_per_second": 0.287,
      "step": 91
    },
    {
      "epoch": 28.92,
      "eval_accuracy": 0.2867132867132867,
      "eval_loss": 1.5419245958328247,
      "eval_runtime": 16.2077,
      "eval_samples_per_second": 17.646,
      "eval_steps_per_second": 0.185,
      "step": 94
    },
    {
      "epoch": 29.85,
      "eval_accuracy": 0.28321678321678323,
      "eval_loss": 1.5383468866348267,
      "eval_runtime": 27.5909,
      "eval_samples_per_second": 10.366,
      "eval_steps_per_second": 0.109,
      "step": 97
    },
    {
      "epoch": 30.77,
      "grad_norm": 65491.41796875,
      "learning_rate": 2.8888888888888888e-05,
      "loss": 1.5563,
      "step": 100
    },
    {
      "epoch": 30.77,
      "eval_accuracy": 0.2762237762237762,
      "eval_loss": 1.5348902940750122,
      "eval_runtime": 14.2201,
      "eval_samples_per_second": 20.112,
      "eval_steps_per_second": 0.211,
      "step": 100
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.27972027972027974,
      "eval_loss": 1.5304443836212158,
      "eval_runtime": 11.5662,
      "eval_samples_per_second": 24.727,
      "eval_steps_per_second": 0.259,
      "step": 104
    },
    {
      "epoch": 32.92,
      "eval_accuracy": 0.2762237762237762,
      "eval_loss": 1.5273348093032837,
      "eval_runtime": 10.0608,
      "eval_samples_per_second": 28.427,
      "eval_steps_per_second": 0.298,
      "step": 107
    },
    {
      "epoch": 33.85,
      "eval_accuracy": 0.26573426573426573,
      "eval_loss": 1.524675965309143,
      "eval_runtime": 16.3502,
      "eval_samples_per_second": 17.492,
      "eval_steps_per_second": 0.183,
      "step": 110
    },
    {
      "epoch": 34.77,
      "eval_accuracy": 0.2517482517482518,
      "eval_loss": 1.5223478078842163,
      "eval_runtime": 11.4239,
      "eval_samples_per_second": 25.035,
      "eval_steps_per_second": 0.263,
      "step": 113
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.24825174825174826,
      "eval_loss": 1.5194127559661865,
      "eval_runtime": 12.5757,
      "eval_samples_per_second": 22.742,
      "eval_steps_per_second": 0.239,
      "step": 117
    },
    {
      "epoch": 36.92,
      "eval_accuracy": 0.24125874125874125,
      "eval_loss": 1.5177949666976929,
      "eval_runtime": 29.9608,
      "eval_samples_per_second": 9.546,
      "eval_steps_per_second": 0.1,
      "step": 120
    },
    {
      "epoch": 37.85,
      "eval_accuracy": 0.23776223776223776,
      "eval_loss": 1.5167657136917114,
      "eval_runtime": 11.3556,
      "eval_samples_per_second": 25.186,
      "eval_steps_per_second": 0.264,
      "step": 123
    },
    {
      "epoch": 38.77,
      "eval_accuracy": 0.24475524475524477,
      "eval_loss": 1.5161939859390259,
      "eval_runtime": 8.3481,
      "eval_samples_per_second": 34.259,
      "eval_steps_per_second": 0.359,
      "step": 126
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.24475524475524477,
      "eval_loss": 1.5161994695663452,
      "eval_runtime": 11.5276,
      "eval_samples_per_second": 24.81,
      "eval_steps_per_second": 0.26,
      "step": 130
    },
    {
      "epoch": 40.92,
      "eval_accuracy": 0.24825174825174826,
      "eval_loss": 1.5166517496109009,
      "eval_runtime": 16.6347,
      "eval_samples_per_second": 17.193,
      "eval_steps_per_second": 0.18,
      "step": 133
    },
    {
      "epoch": 41.85,
      "eval_accuracy": 0.24825174825174826,
      "eval_loss": 1.518052101135254,
      "eval_runtime": 9.0049,
      "eval_samples_per_second": 31.76,
      "eval_steps_per_second": 0.333,
      "step": 136
    },
    {
      "epoch": 42.77,
      "eval_accuracy": 0.25874125874125875,
      "eval_loss": 1.520302414894104,
      "eval_runtime": 9.9294,
      "eval_samples_per_second": 28.803,
      "eval_steps_per_second": 0.302,
      "step": 139
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.2692307692307692,
      "eval_loss": 1.5227454900741577,
      "eval_runtime": 11.2365,
      "eval_samples_per_second": 25.453,
      "eval_steps_per_second": 0.267,
      "step": 143
    },
    {
      "epoch": 44.92,
      "eval_accuracy": 0.28321678321678323,
      "eval_loss": 1.5243264436721802,
      "eval_runtime": 19.5213,
      "eval_samples_per_second": 14.651,
      "eval_steps_per_second": 0.154,
      "step": 146
    },
    {
      "epoch": 45.85,
      "eval_accuracy": 0.27972027972027974,
      "eval_loss": 1.5238893032073975,
      "eval_runtime": 10.1208,
      "eval_samples_per_second": 28.259,
      "eval_steps_per_second": 0.296,
      "step": 149
    },
    {
      "epoch": 46.77,
      "eval_accuracy": 0.3006993006993007,
      "eval_loss": 1.5224095582962036,
      "eval_runtime": 11.1479,
      "eval_samples_per_second": 25.655,
      "eval_steps_per_second": 0.269,
      "step": 152
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.3076923076923077,
      "eval_loss": 1.5170344114303589,
      "eval_runtime": 23.385,
      "eval_samples_per_second": 12.23,
      "eval_steps_per_second": 0.128,
      "step": 156
    },
    {
      "epoch": 48.92,
      "eval_accuracy": 0.32867132867132864,
      "eval_loss": 1.5102859735488892,
      "eval_runtime": 13.363,
      "eval_samples_per_second": 21.402,
      "eval_steps_per_second": 0.225,
      "step": 159
    },
    {
      "epoch": 49.85,
      "eval_accuracy": 0.34965034965034963,
      "eval_loss": 1.5032000541687012,
      "eval_runtime": 10.5466,
      "eval_samples_per_second": 27.118,
      "eval_steps_per_second": 0.284,
      "step": 162
    },
    {
      "epoch": 50.77,
      "eval_accuracy": 0.36013986013986016,
      "eval_loss": 1.4958622455596924,
      "eval_runtime": 13.1319,
      "eval_samples_per_second": 21.779,
      "eval_steps_per_second": 0.228,
      "step": 165
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.36363636363636365,
      "eval_loss": 1.4857254028320312,
      "eval_runtime": 11.3199,
      "eval_samples_per_second": 25.265,
      "eval_steps_per_second": 0.265,
      "step": 169
    },
    {
      "epoch": 52.92,
      "eval_accuracy": 0.36713286713286714,
      "eval_loss": 1.4788074493408203,
      "eval_runtime": 10.1044,
      "eval_samples_per_second": 28.304,
      "eval_steps_per_second": 0.297,
      "step": 172
    },
    {
      "epoch": 53.85,
      "eval_accuracy": 0.3741258741258741,
      "eval_loss": 1.4713162183761597,
      "eval_runtime": 21.314,
      "eval_samples_per_second": 13.418,
      "eval_steps_per_second": 0.141,
      "step": 175
    },
    {
      "epoch": 54.77,
      "eval_accuracy": 0.3811188811188811,
      "eval_loss": 1.4641575813293457,
      "eval_runtime": 11.6546,
      "eval_samples_per_second": 24.54,
      "eval_steps_per_second": 0.257,
      "step": 178
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.3881118881118881,
      "eval_loss": 1.455277442932129,
      "eval_runtime": 11.9334,
      "eval_samples_per_second": 23.966,
      "eval_steps_per_second": 0.251,
      "step": 182
    },
    {
      "epoch": 56.92,
      "eval_accuracy": 0.3986013986013986,
      "eval_loss": 1.4481216669082642,
      "eval_runtime": 10.9879,
      "eval_samples_per_second": 26.029,
      "eval_steps_per_second": 0.273,
      "step": 185
    },
    {
      "epoch": 57.85,
      "eval_accuracy": 0.4020979020979021,
      "eval_loss": 1.4420974254608154,
      "eval_runtime": 8.8556,
      "eval_samples_per_second": 32.296,
      "eval_steps_per_second": 0.339,
      "step": 188
    },
    {
      "epoch": 58.77,
      "eval_accuracy": 0.4125874125874126,
      "eval_loss": 1.4357017278671265,
      "eval_runtime": 9.9507,
      "eval_samples_per_second": 28.742,
      "eval_steps_per_second": 0.301,
      "step": 191
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.4195804195804196,
      "eval_loss": 1.4284459352493286,
      "eval_runtime": 10.9332,
      "eval_samples_per_second": 26.159,
      "eval_steps_per_second": 0.274,
      "step": 195
    },
    {
      "epoch": 60.92,
      "eval_accuracy": 0.4195804195804196,
      "eval_loss": 1.4218381643295288,
      "eval_runtime": 15.3761,
      "eval_samples_per_second": 18.6,
      "eval_steps_per_second": 0.195,
      "step": 198
    },
    {
      "epoch": 61.54,
      "grad_norm": 26744.056640625,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 1.3138,
      "step": 200
    },
    {
      "epoch": 61.85,
      "eval_accuracy": 0.43006993006993005,
      "eval_loss": 1.4166817665100098,
      "eval_runtime": 13.1906,
      "eval_samples_per_second": 21.682,
      "eval_steps_per_second": 0.227,
      "step": 201
    },
    {
      "epoch": 62.77,
      "eval_accuracy": 0.43006993006993005,
      "eval_loss": 1.409144639968872,
      "eval_runtime": 21.121,
      "eval_samples_per_second": 13.541,
      "eval_steps_per_second": 0.142,
      "step": 204
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.4370629370629371,
      "eval_loss": 1.3994969129562378,
      "eval_runtime": 11.3707,
      "eval_samples_per_second": 25.152,
      "eval_steps_per_second": 0.264,
      "step": 208
    },
    {
      "epoch": 64.92,
      "eval_accuracy": 0.4405594405594406,
      "eval_loss": 1.3911248445510864,
      "eval_runtime": 15.8386,
      "eval_samples_per_second": 18.057,
      "eval_steps_per_second": 0.189,
      "step": 211
    },
    {
      "epoch": 65.85,
      "eval_accuracy": 0.43356643356643354,
      "eval_loss": 1.382477045059204,
      "eval_runtime": 15.9417,
      "eval_samples_per_second": 17.94,
      "eval_steps_per_second": 0.188,
      "step": 214
    },
    {
      "epoch": 66.77,
      "eval_accuracy": 0.44405594405594406,
      "eval_loss": 1.373460054397583,
      "eval_runtime": 22.8178,
      "eval_samples_per_second": 12.534,
      "eval_steps_per_second": 0.131,
      "step": 217
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.44755244755244755,
      "eval_loss": 1.3632004261016846,
      "eval_runtime": 8.8406,
      "eval_samples_per_second": 32.351,
      "eval_steps_per_second": 0.339,
      "step": 221
    },
    {
      "epoch": 68.92,
      "eval_accuracy": 0.45104895104895104,
      "eval_loss": 1.355635166168213,
      "eval_runtime": 22.4517,
      "eval_samples_per_second": 12.738,
      "eval_steps_per_second": 0.134,
      "step": 224
    },
    {
      "epoch": 69.85,
      "eval_accuracy": 0.45104895104895104,
      "eval_loss": 1.349207878112793,
      "eval_runtime": 19.129,
      "eval_samples_per_second": 14.951,
      "eval_steps_per_second": 0.157,
      "step": 227
    },
    {
      "epoch": 70.77,
      "eval_accuracy": 0.45104895104895104,
      "eval_loss": 1.3441168069839478,
      "eval_runtime": 10.9281,
      "eval_samples_per_second": 26.171,
      "eval_steps_per_second": 0.275,
      "step": 230
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.458041958041958,
      "eval_loss": 1.3352017402648926,
      "eval_runtime": 9.8412,
      "eval_samples_per_second": 29.062,
      "eval_steps_per_second": 0.305,
      "step": 234
    },
    {
      "epoch": 72.92,
      "eval_accuracy": 0.46153846153846156,
      "eval_loss": 1.326931357383728,
      "eval_runtime": 9.1291,
      "eval_samples_per_second": 31.329,
      "eval_steps_per_second": 0.329,
      "step": 237
    },
    {
      "epoch": 73.85,
      "eval_accuracy": 0.4755244755244755,
      "eval_loss": 1.3185527324676514,
      "eval_runtime": 8.2766,
      "eval_samples_per_second": 34.555,
      "eval_steps_per_second": 0.362,
      "step": 240
    },
    {
      "epoch": 74.77,
      "eval_accuracy": 0.4755244755244755,
      "eval_loss": 1.3105268478393555,
      "eval_runtime": 11.5807,
      "eval_samples_per_second": 24.696,
      "eval_steps_per_second": 0.259,
      "step": 243
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.479020979020979,
      "eval_loss": 1.299223780632019,
      "eval_runtime": 26.8999,
      "eval_samples_per_second": 10.632,
      "eval_steps_per_second": 0.112,
      "step": 247
    },
    {
      "epoch": 76.92,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.2896299362182617,
      "eval_runtime": 22.9313,
      "eval_samples_per_second": 12.472,
      "eval_steps_per_second": 0.131,
      "step": 250
    },
    {
      "epoch": 77.85,
      "eval_accuracy": 0.4825174825174825,
      "eval_loss": 1.2797133922576904,
      "eval_runtime": 12.0479,
      "eval_samples_per_second": 23.739,
      "eval_steps_per_second": 0.249,
      "step": 253
    },
    {
      "epoch": 78.77,
      "eval_accuracy": 0.486013986013986,
      "eval_loss": 1.270691156387329,
      "eval_runtime": 21.8623,
      "eval_samples_per_second": 13.082,
      "eval_steps_per_second": 0.137,
      "step": 256
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2586532831192017,
      "eval_runtime": 21.4963,
      "eval_samples_per_second": 13.305,
      "eval_steps_per_second": 0.14,
      "step": 260
    },
    {
      "epoch": 80.92,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2493596076965332,
      "eval_runtime": 16.0914,
      "eval_samples_per_second": 17.774,
      "eval_steps_per_second": 0.186,
      "step": 263
    },
    {
      "epoch": 81.85,
      "eval_accuracy": 0.493006993006993,
      "eval_loss": 1.2407375574111938,
      "eval_runtime": 16.0788,
      "eval_samples_per_second": 17.787,
      "eval_steps_per_second": 0.187,
      "step": 266
    },
    {
      "epoch": 82.77,
      "eval_accuracy": 0.5104895104895105,
      "eval_loss": 1.2313759326934814,
      "eval_runtime": 13.8028,
      "eval_samples_per_second": 20.72,
      "eval_steps_per_second": 0.217,
      "step": 269
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.513986013986014,
      "eval_loss": 1.2205084562301636,
      "eval_runtime": 15.155,
      "eval_samples_per_second": 18.872,
      "eval_steps_per_second": 0.198,
      "step": 273
    },
    {
      "epoch": 84.92,
      "eval_accuracy": 0.5209790209790209,
      "eval_loss": 1.2124475240707397,
      "eval_runtime": 24.3109,
      "eval_samples_per_second": 11.764,
      "eval_steps_per_second": 0.123,
      "step": 276
    },
    {
      "epoch": 85.85,
      "eval_accuracy": 0.5314685314685315,
      "eval_loss": 1.2043241262435913,
      "eval_runtime": 12.0961,
      "eval_samples_per_second": 23.644,
      "eval_steps_per_second": 0.248,
      "step": 279
    },
    {
      "epoch": 86.77,
      "eval_accuracy": 0.534965034965035,
      "eval_loss": 1.1973276138305664,
      "eval_runtime": 18.6151,
      "eval_samples_per_second": 15.364,
      "eval_steps_per_second": 0.161,
      "step": 282
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.5524475524475524,
      "eval_loss": 1.1870229244232178,
      "eval_runtime": 19.514,
      "eval_samples_per_second": 14.656,
      "eval_steps_per_second": 0.154,
      "step": 286
    },
    {
      "epoch": 88.92,
      "eval_accuracy": 0.5629370629370629,
      "eval_loss": 1.178816318511963,
      "eval_runtime": 17.8959,
      "eval_samples_per_second": 15.981,
      "eval_steps_per_second": 0.168,
      "step": 289
    },
    {
      "epoch": 89.85,
      "eval_accuracy": 0.5629370629370629,
      "eval_loss": 1.1699650287628174,
      "eval_runtime": 14.6752,
      "eval_samples_per_second": 19.489,
      "eval_steps_per_second": 0.204,
      "step": 292
    },
    {
      "epoch": 90.77,
      "eval_accuracy": 0.5699300699300699,
      "eval_loss": 1.1613417863845825,
      "eval_runtime": 23.2377,
      "eval_samples_per_second": 12.308,
      "eval_steps_per_second": 0.129,
      "step": 295
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.583916083916084,
      "eval_loss": 1.1498184204101562,
      "eval_runtime": 16.8037,
      "eval_samples_per_second": 17.02,
      "eval_steps_per_second": 0.179,
      "step": 299
    },
    {
      "epoch": 92.31,
      "grad_norm": 27893.916015625,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 1.047,
      "step": 300
    },
    {
      "epoch": 92.92,
      "eval_accuracy": 0.5874125874125874,
      "eval_loss": 1.1410824060440063,
      "eval_runtime": 17.7012,
      "eval_samples_per_second": 16.157,
      "eval_steps_per_second": 0.169,
      "step": 302
    },
    {
      "epoch": 93.85,
      "eval_accuracy": 0.5944055944055944,
      "eval_loss": 1.133009433746338,
      "eval_runtime": 21.3624,
      "eval_samples_per_second": 13.388,
      "eval_steps_per_second": 0.14,
      "step": 305
    },
    {
      "epoch": 94.77,
      "eval_accuracy": 0.5944055944055944,
      "eval_loss": 1.1261355876922607,
      "eval_runtime": 12.5743,
      "eval_samples_per_second": 22.745,
      "eval_steps_per_second": 0.239,
      "step": 308
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.6013986013986014,
      "eval_loss": 1.1161199808120728,
      "eval_runtime": 10.714,
      "eval_samples_per_second": 26.694,
      "eval_steps_per_second": 0.28,
      "step": 312
    },
    {
      "epoch": 96.92,
      "eval_accuracy": 0.6013986013986014,
      "eval_loss": 1.1083568334579468,
      "eval_runtime": 23.7041,
      "eval_samples_per_second": 12.065,
      "eval_steps_per_second": 0.127,
      "step": 315
    },
    {
      "epoch": 97.85,
      "eval_accuracy": 0.6048951048951049,
      "eval_loss": 1.1002540588378906,
      "eval_runtime": 18.2723,
      "eval_samples_per_second": 15.652,
      "eval_steps_per_second": 0.164,
      "step": 318
    },
    {
      "epoch": 98.77,
      "eval_accuracy": 0.6048951048951049,
      "eval_loss": 1.0926333665847778,
      "eval_runtime": 15.4521,
      "eval_samples_per_second": 18.509,
      "eval_steps_per_second": 0.194,
      "step": 321
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.6083916083916084,
      "eval_loss": 1.0821202993392944,
      "eval_runtime": 14.9648,
      "eval_samples_per_second": 19.112,
      "eval_steps_per_second": 0.2,
      "step": 325
    },
    {
      "epoch": 100.92,
      "eval_accuracy": 0.6083916083916084,
      "eval_loss": 1.075362205505371,
      "eval_runtime": 18.8741,
      "eval_samples_per_second": 15.153,
      "eval_steps_per_second": 0.159,
      "step": 328
    },
    {
      "epoch": 101.85,
      "eval_accuracy": 0.6083916083916084,
      "eval_loss": 1.06904935836792,
      "eval_runtime": 12.1488,
      "eval_samples_per_second": 23.541,
      "eval_steps_per_second": 0.247,
      "step": 331
    },
    {
      "epoch": 102.77,
      "eval_accuracy": 0.6153846153846154,
      "eval_loss": 1.0637034177780151,
      "eval_runtime": 21.2582,
      "eval_samples_per_second": 13.454,
      "eval_steps_per_second": 0.141,
      "step": 334
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 0.6188811188811189,
      "eval_loss": 1.0549243688583374,
      "eval_runtime": 19.5513,
      "eval_samples_per_second": 14.628,
      "eval_steps_per_second": 0.153,
      "step": 338
    },
    {
      "epoch": 104.92,
      "eval_accuracy": 0.6223776223776224,
      "eval_loss": 1.047833800315857,
      "eval_runtime": 17.2191,
      "eval_samples_per_second": 16.609,
      "eval_steps_per_second": 0.174,
      "step": 341
    },
    {
      "epoch": 105.85,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0420035123825073,
      "eval_runtime": 25.1625,
      "eval_samples_per_second": 11.366,
      "eval_steps_per_second": 0.119,
      "step": 344
    },
    {
      "epoch": 106.77,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0369622707366943,
      "eval_runtime": 14.0552,
      "eval_samples_per_second": 20.348,
      "eval_steps_per_second": 0.213,
      "step": 347
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0308306217193604,
      "eval_runtime": 15.7213,
      "eval_samples_per_second": 18.192,
      "eval_steps_per_second": 0.191,
      "step": 351
    },
    {
      "epoch": 108.92,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0263006687164307,
      "eval_runtime": 19.1306,
      "eval_samples_per_second": 14.95,
      "eval_steps_per_second": 0.157,
      "step": 354
    },
    {
      "epoch": 109.85,
      "eval_accuracy": 0.6258741258741258,
      "eval_loss": 1.0230927467346191,
      "eval_runtime": 15.401,
      "eval_samples_per_second": 18.57,
      "eval_steps_per_second": 0.195,
      "step": 357
    },
    {
      "epoch": 110.77,
      "eval_accuracy": 0.6328671328671329,
      "eval_loss": 1.0204286575317383,
      "eval_runtime": 11.4397,
      "eval_samples_per_second": 25.001,
      "eval_steps_per_second": 0.262,
      "step": 360
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0166879892349243,
      "eval_runtime": 14.7547,
      "eval_samples_per_second": 19.384,
      "eval_steps_per_second": 0.203,
      "step": 364
    },
    {
      "epoch": 112.92,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0145463943481445,
      "eval_runtime": 12.2752,
      "eval_samples_per_second": 23.299,
      "eval_steps_per_second": 0.244,
      "step": 367
    },
    {
      "epoch": 113.85,
      "eval_accuracy": 0.6328671328671329,
      "eval_loss": 1.0118980407714844,
      "eval_runtime": 29.1643,
      "eval_samples_per_second": 9.807,
      "eval_steps_per_second": 0.103,
      "step": 370
    },
    {
      "epoch": 114.77,
      "eval_accuracy": 0.6293706293706294,
      "eval_loss": 1.0077061653137207,
      "eval_runtime": 15.736,
      "eval_samples_per_second": 18.175,
      "eval_steps_per_second": 0.191,
      "step": 373
    },
    {
      "epoch": 116.0,
      "eval_accuracy": 0.6363636363636364,
      "eval_loss": 1.001206398010254,
      "eval_runtime": 13.5478,
      "eval_samples_per_second": 21.11,
      "eval_steps_per_second": 0.221,
      "step": 377
    },
    {
      "epoch": 116.92,
      "eval_accuracy": 0.6363636363636364,
      "eval_loss": 0.9975122809410095,
      "eval_runtime": 12.3425,
      "eval_samples_per_second": 23.172,
      "eval_steps_per_second": 0.243,
      "step": 380
    },
    {
      "epoch": 117.85,
      "eval_accuracy": 0.6363636363636364,
      "eval_loss": 0.993752121925354,
      "eval_runtime": 16.3699,
      "eval_samples_per_second": 17.471,
      "eval_steps_per_second": 0.183,
      "step": 383
    },
    {
      "epoch": 118.77,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9913222193717957,
      "eval_runtime": 15.9587,
      "eval_samples_per_second": 17.921,
      "eval_steps_per_second": 0.188,
      "step": 386
    },
    {
      "epoch": 120.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9886361360549927,
      "eval_runtime": 14.2728,
      "eval_samples_per_second": 20.038,
      "eval_steps_per_second": 0.21,
      "step": 390
    },
    {
      "epoch": 120.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.987022340297699,
      "eval_runtime": 28.4357,
      "eval_samples_per_second": 10.058,
      "eval_steps_per_second": 0.106,
      "step": 393
    },
    {
      "epoch": 121.85,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9861236810684204,
      "eval_runtime": 12.6739,
      "eval_samples_per_second": 22.566,
      "eval_steps_per_second": 0.237,
      "step": 396
    },
    {
      "epoch": 122.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9857375025749207,
      "eval_runtime": 14.4905,
      "eval_samples_per_second": 19.737,
      "eval_steps_per_second": 0.207,
      "step": 399
    },
    {
      "epoch": 123.08,
      "grad_norm": 28786.830078125,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 0.8183,
      "step": 400
    },
    {
      "epoch": 124.0,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9855088591575623,
      "eval_runtime": 23.5291,
      "eval_samples_per_second": 12.155,
      "eval_steps_per_second": 0.128,
      "step": 403
    },
    {
      "epoch": 124.92,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.986443817615509,
      "eval_runtime": 11.267,
      "eval_samples_per_second": 25.384,
      "eval_steps_per_second": 0.266,
      "step": 406
    },
    {
      "epoch": 125.85,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9856669902801514,
      "eval_runtime": 15.102,
      "eval_samples_per_second": 18.938,
      "eval_steps_per_second": 0.199,
      "step": 409
    },
    {
      "epoch": 126.77,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9817618131637573,
      "eval_runtime": 13.7486,
      "eval_samples_per_second": 20.802,
      "eval_steps_per_second": 0.218,
      "step": 412
    },
    {
      "epoch": 128.0,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9765364527702332,
      "eval_runtime": 15.0977,
      "eval_samples_per_second": 18.943,
      "eval_steps_per_second": 0.199,
      "step": 416
    },
    {
      "epoch": 128.92,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9739550352096558,
      "eval_runtime": 9.9261,
      "eval_samples_per_second": 28.813,
      "eval_steps_per_second": 0.302,
      "step": 419
    },
    {
      "epoch": 129.85,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9736567139625549,
      "eval_runtime": 23.6705,
      "eval_samples_per_second": 12.083,
      "eval_steps_per_second": 0.127,
      "step": 422
    },
    {
      "epoch": 130.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9753543734550476,
      "eval_runtime": 13.1489,
      "eval_samples_per_second": 21.751,
      "eval_steps_per_second": 0.228,
      "step": 425
    },
    {
      "epoch": 132.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.975328803062439,
      "eval_runtime": 14.1291,
      "eval_samples_per_second": 20.242,
      "eval_steps_per_second": 0.212,
      "step": 429
    },
    {
      "epoch": 132.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9739886522293091,
      "eval_runtime": 23.3295,
      "eval_samples_per_second": 12.259,
      "eval_steps_per_second": 0.129,
      "step": 432
    },
    {
      "epoch": 133.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9709957242012024,
      "eval_runtime": 12.6523,
      "eval_samples_per_second": 22.605,
      "eval_steps_per_second": 0.237,
      "step": 435
    },
    {
      "epoch": 134.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9685706496238708,
      "eval_runtime": 12.6672,
      "eval_samples_per_second": 22.578,
      "eval_steps_per_second": 0.237,
      "step": 438
    },
    {
      "epoch": 136.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9670514464378357,
      "eval_runtime": 14.1037,
      "eval_samples_per_second": 20.278,
      "eval_steps_per_second": 0.213,
      "step": 442
    },
    {
      "epoch": 136.92,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.966910719871521,
      "eval_runtime": 17.8781,
      "eval_samples_per_second": 15.997,
      "eval_steps_per_second": 0.168,
      "step": 445
    },
    {
      "epoch": 137.85,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9659475088119507,
      "eval_runtime": 14.4986,
      "eval_samples_per_second": 19.726,
      "eval_steps_per_second": 0.207,
      "step": 448
    },
    {
      "epoch": 138.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9662479758262634,
      "eval_runtime": 9.0453,
      "eval_samples_per_second": 31.619,
      "eval_steps_per_second": 0.332,
      "step": 451
    },
    {
      "epoch": 140.0,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.967424750328064,
      "eval_runtime": 15.45,
      "eval_samples_per_second": 18.511,
      "eval_steps_per_second": 0.194,
      "step": 455
    },
    {
      "epoch": 140.92,
      "eval_accuracy": 0.6398601398601399,
      "eval_loss": 0.9694240689277649,
      "eval_runtime": 25.3846,
      "eval_samples_per_second": 11.267,
      "eval_steps_per_second": 0.118,
      "step": 458
    },
    {
      "epoch": 141.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9716474413871765,
      "eval_runtime": 13.4283,
      "eval_samples_per_second": 21.298,
      "eval_steps_per_second": 0.223,
      "step": 461
    },
    {
      "epoch": 142.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9738869667053223,
      "eval_runtime": 14.4683,
      "eval_samples_per_second": 19.767,
      "eval_steps_per_second": 0.207,
      "step": 464
    },
    {
      "epoch": 144.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9711639881134033,
      "eval_runtime": 16.7189,
      "eval_samples_per_second": 17.106,
      "eval_steps_per_second": 0.179,
      "step": 468
    },
    {
      "epoch": 144.92,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9670231938362122,
      "eval_runtime": 17.2531,
      "eval_samples_per_second": 16.577,
      "eval_steps_per_second": 0.174,
      "step": 471
    },
    {
      "epoch": 145.85,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.9637404084205627,
      "eval_runtime": 7.2246,
      "eval_samples_per_second": 39.587,
      "eval_steps_per_second": 0.415,
      "step": 474
    },
    {
      "epoch": 146.77,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9625027775764465,
      "eval_runtime": 22.2959,
      "eval_samples_per_second": 12.827,
      "eval_steps_per_second": 0.135,
      "step": 477
    },
    {
      "epoch": 148.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9634200930595398,
      "eval_runtime": 13.0089,
      "eval_samples_per_second": 21.985,
      "eval_steps_per_second": 0.231,
      "step": 481
    },
    {
      "epoch": 148.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9658861756324768,
      "eval_runtime": 13.9249,
      "eval_samples_per_second": 20.539,
      "eval_steps_per_second": 0.215,
      "step": 484
    },
    {
      "epoch": 149.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9662710428237915,
      "eval_runtime": 26.5045,
      "eval_samples_per_second": 10.791,
      "eval_steps_per_second": 0.113,
      "step": 487
    },
    {
      "epoch": 150.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9648845195770264,
      "eval_runtime": 13.8454,
      "eval_samples_per_second": 20.657,
      "eval_steps_per_second": 0.217,
      "step": 490
    },
    {
      "epoch": 152.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9654809832572937,
      "eval_runtime": 11.8579,
      "eval_samples_per_second": 24.119,
      "eval_steps_per_second": 0.253,
      "step": 494
    },
    {
      "epoch": 152.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9648337364196777,
      "eval_runtime": 13.0859,
      "eval_samples_per_second": 21.856,
      "eval_steps_per_second": 0.229,
      "step": 497
    },
    {
      "epoch": 153.85,
      "grad_norm": 27909.80078125,
      "learning_rate": 1.111111111111111e-05,
      "loss": 0.7321,
      "step": 500
    },
    {
      "epoch": 153.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.963828980922699,
      "eval_runtime": 8.9676,
      "eval_samples_per_second": 31.892,
      "eval_steps_per_second": 0.335,
      "step": 500
    },
    {
      "epoch": 154.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9631070494651794,
      "eval_runtime": 32.863,
      "eval_samples_per_second": 8.703,
      "eval_steps_per_second": 0.091,
      "step": 503
    },
    {
      "epoch": 156.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.964668333530426,
      "eval_runtime": 11.8735,
      "eval_samples_per_second": 24.087,
      "eval_steps_per_second": 0.253,
      "step": 507
    },
    {
      "epoch": 156.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9652780294418335,
      "eval_runtime": 12.5571,
      "eval_samples_per_second": 22.776,
      "eval_steps_per_second": 0.239,
      "step": 510
    },
    {
      "epoch": 157.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9662439227104187,
      "eval_runtime": 22.8916,
      "eval_samples_per_second": 12.494,
      "eval_steps_per_second": 0.131,
      "step": 513
    },
    {
      "epoch": 158.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9679471254348755,
      "eval_runtime": 13.713,
      "eval_samples_per_second": 20.856,
      "eval_steps_per_second": 0.219,
      "step": 516
    },
    {
      "epoch": 160.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9674729704856873,
      "eval_runtime": 13.3059,
      "eval_samples_per_second": 21.494,
      "eval_steps_per_second": 0.225,
      "step": 520
    },
    {
      "epoch": 160.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9663581252098083,
      "eval_runtime": 14.0564,
      "eval_samples_per_second": 20.347,
      "eval_steps_per_second": 0.213,
      "step": 523
    },
    {
      "epoch": 161.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9654582738876343,
      "eval_runtime": 17.3488,
      "eval_samples_per_second": 16.485,
      "eval_steps_per_second": 0.173,
      "step": 526
    },
    {
      "epoch": 162.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9642163515090942,
      "eval_runtime": 28.2809,
      "eval_samples_per_second": 10.113,
      "eval_steps_per_second": 0.106,
      "step": 529
    },
    {
      "epoch": 164.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9635214805603027,
      "eval_runtime": 13.2297,
      "eval_samples_per_second": 21.618,
      "eval_steps_per_second": 0.227,
      "step": 533
    },
    {
      "epoch": 164.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9632530808448792,
      "eval_runtime": 14.959,
      "eval_samples_per_second": 19.119,
      "eval_steps_per_second": 0.201,
      "step": 536
    },
    {
      "epoch": 165.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9644965529441833,
      "eval_runtime": 27.3169,
      "eval_samples_per_second": 10.47,
      "eval_steps_per_second": 0.11,
      "step": 539
    },
    {
      "epoch": 166.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9649432897567749,
      "eval_runtime": 16.2467,
      "eval_samples_per_second": 17.604,
      "eval_steps_per_second": 0.185,
      "step": 542
    },
    {
      "epoch": 168.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9651181101799011,
      "eval_runtime": 14.8641,
      "eval_samples_per_second": 19.241,
      "eval_steps_per_second": 0.202,
      "step": 546
    },
    {
      "epoch": 168.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9657481908798218,
      "eval_runtime": 14.6637,
      "eval_samples_per_second": 19.504,
      "eval_steps_per_second": 0.205,
      "step": 549
    },
    {
      "epoch": 169.85,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9662555456161499,
      "eval_runtime": 16.1862,
      "eval_samples_per_second": 17.669,
      "eval_steps_per_second": 0.185,
      "step": 552
    },
    {
      "epoch": 170.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9652681946754456,
      "eval_runtime": 13.8601,
      "eval_samples_per_second": 20.635,
      "eval_steps_per_second": 0.216,
      "step": 555
    },
    {
      "epoch": 172.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9637671113014221,
      "eval_runtime": 14.0763,
      "eval_samples_per_second": 20.318,
      "eval_steps_per_second": 0.213,
      "step": 559
    },
    {
      "epoch": 172.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9616447687149048,
      "eval_runtime": 14.3281,
      "eval_samples_per_second": 19.961,
      "eval_steps_per_second": 0.209,
      "step": 562
    },
    {
      "epoch": 173.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9601487517356873,
      "eval_runtime": 14.8622,
      "eval_samples_per_second": 19.243,
      "eval_steps_per_second": 0.202,
      "step": 565
    },
    {
      "epoch": 174.77,
      "eval_accuracy": 0.6538461538461539,
      "eval_loss": 0.9610276818275452,
      "eval_runtime": 11.3604,
      "eval_samples_per_second": 25.175,
      "eval_steps_per_second": 0.264,
      "step": 568
    },
    {
      "epoch": 176.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.962989866733551,
      "eval_runtime": 33.9341,
      "eval_samples_per_second": 8.428,
      "eval_steps_per_second": 0.088,
      "step": 572
    },
    {
      "epoch": 176.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9633333683013916,
      "eval_runtime": 11.5929,
      "eval_samples_per_second": 24.67,
      "eval_steps_per_second": 0.259,
      "step": 575
    },
    {
      "epoch": 177.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9645780324935913,
      "eval_runtime": 13.4891,
      "eval_samples_per_second": 21.202,
      "eval_steps_per_second": 0.222,
      "step": 578
    },
    {
      "epoch": 178.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9654965996742249,
      "eval_runtime": 28.6917,
      "eval_samples_per_second": 9.968,
      "eval_steps_per_second": 0.105,
      "step": 581
    },
    {
      "epoch": 180.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9672998785972595,
      "eval_runtime": 14.4786,
      "eval_samples_per_second": 19.753,
      "eval_steps_per_second": 0.207,
      "step": 585
    },
    {
      "epoch": 180.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9679505825042725,
      "eval_runtime": 13.6779,
      "eval_samples_per_second": 20.91,
      "eval_steps_per_second": 0.219,
      "step": 588
    },
    {
      "epoch": 181.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9686614274978638,
      "eval_runtime": 21.421,
      "eval_samples_per_second": 13.351,
      "eval_steps_per_second": 0.14,
      "step": 591
    },
    {
      "epoch": 182.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9691935777664185,
      "eval_runtime": 12.6562,
      "eval_samples_per_second": 22.598,
      "eval_steps_per_second": 0.237,
      "step": 594
    },
    {
      "epoch": 184.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9683831334114075,
      "eval_runtime": 18.1378,
      "eval_samples_per_second": 15.768,
      "eval_steps_per_second": 0.165,
      "step": 598
    },
    {
      "epoch": 184.62,
      "grad_norm": 28079.01171875,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.6941,
      "step": 600
    },
    {
      "epoch": 184.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9676913022994995,
      "eval_runtime": 22.197,
      "eval_samples_per_second": 12.885,
      "eval_steps_per_second": 0.135,
      "step": 601
    },
    {
      "epoch": 185.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9673991203308105,
      "eval_runtime": 13.8487,
      "eval_samples_per_second": 20.652,
      "eval_steps_per_second": 0.217,
      "step": 604
    },
    {
      "epoch": 186.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9671489000320435,
      "eval_runtime": 16.6344,
      "eval_samples_per_second": 17.193,
      "eval_steps_per_second": 0.18,
      "step": 607
    },
    {
      "epoch": 188.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.96697998046875,
      "eval_runtime": 14.8813,
      "eval_samples_per_second": 19.219,
      "eval_steps_per_second": 0.202,
      "step": 611
    },
    {
      "epoch": 188.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9661853313446045,
      "eval_runtime": 15.8068,
      "eval_samples_per_second": 18.094,
      "eval_steps_per_second": 0.19,
      "step": 614
    },
    {
      "epoch": 189.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9652778506278992,
      "eval_runtime": 27.4384,
      "eval_samples_per_second": 10.423,
      "eval_steps_per_second": 0.109,
      "step": 617
    },
    {
      "epoch": 190.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9644954800605774,
      "eval_runtime": 12.0079,
      "eval_samples_per_second": 23.818,
      "eval_steps_per_second": 0.25,
      "step": 620
    },
    {
      "epoch": 192.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9648195505142212,
      "eval_runtime": 15.7494,
      "eval_samples_per_second": 18.159,
      "eval_steps_per_second": 0.19,
      "step": 624
    },
    {
      "epoch": 192.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9652000665664673,
      "eval_runtime": 14.0294,
      "eval_samples_per_second": 20.386,
      "eval_steps_per_second": 0.214,
      "step": 627
    },
    {
      "epoch": 193.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9662671089172363,
      "eval_runtime": 16.6283,
      "eval_samples_per_second": 17.2,
      "eval_steps_per_second": 0.18,
      "step": 630
    },
    {
      "epoch": 194.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9662363529205322,
      "eval_runtime": 12.434,
      "eval_samples_per_second": 23.001,
      "eval_steps_per_second": 0.241,
      "step": 633
    },
    {
      "epoch": 196.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.966452419757843,
      "eval_runtime": 27.5274,
      "eval_samples_per_second": 10.39,
      "eval_steps_per_second": 0.109,
      "step": 637
    },
    {
      "epoch": 196.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9667707085609436,
      "eval_runtime": 14.0792,
      "eval_samples_per_second": 20.314,
      "eval_steps_per_second": 0.213,
      "step": 640
    },
    {
      "epoch": 197.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9669297337532043,
      "eval_runtime": 11.4357,
      "eval_samples_per_second": 25.009,
      "eval_steps_per_second": 0.262,
      "step": 643
    },
    {
      "epoch": 198.77,
      "eval_accuracy": 0.6433566433566433,
      "eval_loss": 0.967424213886261,
      "eval_runtime": 19.968,
      "eval_samples_per_second": 14.323,
      "eval_steps_per_second": 0.15,
      "step": 646
    },
    {
      "epoch": 200.0,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9668986797332764,
      "eval_runtime": 11.3124,
      "eval_samples_per_second": 25.282,
      "eval_steps_per_second": 0.265,
      "step": 650
    },
    {
      "epoch": 200.92,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9671657681465149,
      "eval_runtime": 15.4547,
      "eval_samples_per_second": 18.506,
      "eval_steps_per_second": 0.194,
      "step": 653
    },
    {
      "epoch": 201.85,
      "eval_accuracy": 0.6468531468531469,
      "eval_loss": 0.9671434760093689,
      "eval_runtime": 11.8747,
      "eval_samples_per_second": 24.085,
      "eval_steps_per_second": 0.253,
      "step": 656
    },
    {
      "epoch": 202.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.967323362827301,
      "eval_runtime": 13.4706,
      "eval_samples_per_second": 21.231,
      "eval_steps_per_second": 0.223,
      "step": 659
    },
    {
      "epoch": 204.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9666022658348083,
      "eval_runtime": 11.809,
      "eval_samples_per_second": 24.219,
      "eval_steps_per_second": 0.254,
      "step": 663
    },
    {
      "epoch": 204.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.966048002243042,
      "eval_runtime": 30.3868,
      "eval_samples_per_second": 9.412,
      "eval_steps_per_second": 0.099,
      "step": 666
    },
    {
      "epoch": 205.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9656386971473694,
      "eval_runtime": 10.0795,
      "eval_samples_per_second": 28.375,
      "eval_steps_per_second": 0.298,
      "step": 669
    },
    {
      "epoch": 206.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9650998711585999,
      "eval_runtime": 11.1232,
      "eval_samples_per_second": 25.712,
      "eval_steps_per_second": 0.27,
      "step": 672
    },
    {
      "epoch": 208.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9660706520080566,
      "eval_runtime": 14.3325,
      "eval_samples_per_second": 19.955,
      "eval_steps_per_second": 0.209,
      "step": 676
    },
    {
      "epoch": 208.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9666641354560852,
      "eval_runtime": 18.7486,
      "eval_samples_per_second": 15.254,
      "eval_steps_per_second": 0.16,
      "step": 679
    },
    {
      "epoch": 209.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9667859077453613,
      "eval_runtime": 11.9596,
      "eval_samples_per_second": 23.914,
      "eval_steps_per_second": 0.251,
      "step": 682
    },
    {
      "epoch": 210.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9668753147125244,
      "eval_runtime": 21.049,
      "eval_samples_per_second": 13.587,
      "eval_steps_per_second": 0.143,
      "step": 685
    },
    {
      "epoch": 212.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665384292602539,
      "eval_runtime": 12.916,
      "eval_samples_per_second": 22.143,
      "eval_steps_per_second": 0.232,
      "step": 689
    },
    {
      "epoch": 212.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665247797966003,
      "eval_runtime": 13.5797,
      "eval_samples_per_second": 21.061,
      "eval_steps_per_second": 0.221,
      "step": 692
    },
    {
      "epoch": 213.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9663905501365662,
      "eval_runtime": 11.9452,
      "eval_samples_per_second": 23.943,
      "eval_steps_per_second": 0.251,
      "step": 695
    },
    {
      "epoch": 214.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9662503600120544,
      "eval_runtime": 11.3523,
      "eval_samples_per_second": 25.193,
      "eval_steps_per_second": 0.264,
      "step": 698
    },
    {
      "epoch": 215.38,
      "grad_norm": 25925.685546875,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.6696,
      "step": 700
    },
    {
      "epoch": 216.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9666140079498291,
      "eval_runtime": 11.4254,
      "eval_samples_per_second": 25.032,
      "eval_steps_per_second": 0.263,
      "step": 702
    },
    {
      "epoch": 216.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9667044878005981,
      "eval_runtime": 9.4078,
      "eval_samples_per_second": 30.4,
      "eval_steps_per_second": 0.319,
      "step": 705
    },
    {
      "epoch": 217.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665290713310242,
      "eval_runtime": 8.9463,
      "eval_samples_per_second": 31.969,
      "eval_steps_per_second": 0.335,
      "step": 708
    },
    {
      "epoch": 218.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9663181900978088,
      "eval_runtime": 12.4146,
      "eval_samples_per_second": 23.037,
      "eval_steps_per_second": 0.242,
      "step": 711
    },
    {
      "epoch": 220.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9661046862602234,
      "eval_runtime": 12.2235,
      "eval_samples_per_second": 23.398,
      "eval_steps_per_second": 0.245,
      "step": 715
    },
    {
      "epoch": 220.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.966101348400116,
      "eval_runtime": 12.7423,
      "eval_samples_per_second": 22.445,
      "eval_steps_per_second": 0.235,
      "step": 718
    },
    {
      "epoch": 221.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9661813974380493,
      "eval_runtime": 11.077,
      "eval_samples_per_second": 25.819,
      "eval_steps_per_second": 0.271,
      "step": 721
    },
    {
      "epoch": 222.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9663574695587158,
      "eval_runtime": 13.5789,
      "eval_samples_per_second": 21.062,
      "eval_steps_per_second": 0.221,
      "step": 724
    },
    {
      "epoch": 224.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9663906693458557,
      "eval_runtime": 11.7803,
      "eval_samples_per_second": 24.278,
      "eval_steps_per_second": 0.255,
      "step": 728
    },
    {
      "epoch": 224.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9664490222930908,
      "eval_runtime": 13.0369,
      "eval_samples_per_second": 21.938,
      "eval_steps_per_second": 0.23,
      "step": 731
    },
    {
      "epoch": 225.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665610790252686,
      "eval_runtime": 10.8415,
      "eval_samples_per_second": 26.38,
      "eval_steps_per_second": 0.277,
      "step": 734
    },
    {
      "epoch": 226.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665590524673462,
      "eval_runtime": 14.6312,
      "eval_samples_per_second": 19.547,
      "eval_steps_per_second": 0.205,
      "step": 737
    },
    {
      "epoch": 228.0,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665370583534241,
      "eval_runtime": 11.8502,
      "eval_samples_per_second": 24.135,
      "eval_steps_per_second": 0.253,
      "step": 741
    },
    {
      "epoch": 228.92,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665577411651611,
      "eval_runtime": 12.7745,
      "eval_samples_per_second": 22.388,
      "eval_steps_per_second": 0.235,
      "step": 744
    },
    {
      "epoch": 229.85,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665866494178772,
      "eval_runtime": 13.6538,
      "eval_samples_per_second": 20.947,
      "eval_steps_per_second": 0.22,
      "step": 747
    },
    {
      "epoch": 230.77,
      "eval_accuracy": 0.6503496503496503,
      "eval_loss": 0.9665878415107727,
      "eval_runtime": 11.2039,
      "eval_samples_per_second": 25.527,
      "eval_steps_per_second": 0.268,
      "step": 750
    },
    {
      "epoch": 230.77,
      "step": 750,
      "total_flos": 3.12964745557632e+18,
      "train_loss": 0.9558865051269532,
      "train_runtime": 10052.9365,
      "train_samples_per_second": 38.496,
      "train_steps_per_second": 0.075
    }
  ],
  "logging_steps": 100,
  "max_steps": 750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 250,
  "save_steps": 500,
  "total_flos": 3.12964745557632e+18,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}