{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.4078254326561326,
  "eval_steps": 20,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 5e-07,
      "loss": 1.1046,
      "step": 10
    },
    {
      "epoch": 0.06,
      "learning_rate": 1e-06,
      "loss": 1.0664,
      "step": 20
    },
    {
      "epoch": 0.06,
      "eval_accuracy": 0.44479243019924036,
      "eval_loss": 1.065091609954834,
      "eval_runtime": 50.348,
      "eval_samples_per_second": 596.131,
      "eval_steps_per_second": 1.172,
      "step": 20
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5e-06,
      "loss": 1.0672,
      "step": 30
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-06,
      "loss": 1.0423,
      "step": 40
    },
    {
      "epoch": 0.12,
      "eval_accuracy": 0.5033650962883988,
      "eval_loss": 1.0188277959823608,
      "eval_runtime": 49.614,
      "eval_samples_per_second": 604.95,
      "eval_steps_per_second": 1.189,
      "step": 40
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5e-06,
      "loss": 1.027,
      "step": 50
    },
    {
      "epoch": 0.18,
      "learning_rate": 3e-06,
      "loss": 1.0137,
      "step": 60
    },
    {
      "epoch": 0.18,
      "eval_accuracy": 0.5279203038581995,
      "eval_loss": 0.9871189594268799,
      "eval_runtime": 49.5201,
      "eval_samples_per_second": 606.098,
      "eval_steps_per_second": 1.191,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 3.5e-06,
      "loss": 1.0005,
      "step": 70
    },
    {
      "epoch": 0.24,
      "learning_rate": 4e-06,
      "loss": 1.0027,
      "step": 80
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.5308189511561271,
      "eval_loss": 0.9888613224029541,
      "eval_runtime": 49.4469,
      "eval_samples_per_second": 606.994,
      "eval_steps_per_second": 1.193,
      "step": 80
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5e-06,
      "loss": 0.9937,
      "step": 90
    },
    {
      "epoch": 0.3,
      "learning_rate": 5e-06,
      "loss": 0.9914,
      "step": 100
    },
    {
      "epoch": 0.3,
      "eval_accuracy": 0.5307856333710935,
      "eval_loss": 0.9762536883354187,
      "eval_runtime": 49.4804,
      "eval_samples_per_second": 606.583,
      "eval_steps_per_second": 1.192,
      "step": 100
    },
    {
      "epoch": 0.33,
      "learning_rate": 5.5e-06,
      "loss": 0.9884,
      "step": 110
    },
    {
      "epoch": 0.36,
      "learning_rate": 6e-06,
      "loss": 0.9826,
      "step": 120
    },
    {
      "epoch": 0.36,
      "eval_accuracy": 0.5387819017791697,
      "eval_loss": 0.9713281989097595,
      "eval_runtime": 49.4983,
      "eval_samples_per_second": 606.364,
      "eval_steps_per_second": 1.192,
      "step": 120
    },
    {
      "epoch": 0.39,
      "learning_rate": 6.5e-06,
      "loss": 0.9882,
      "step": 130
    },
    {
      "epoch": 0.42,
      "learning_rate": 7e-06,
      "loss": 0.9788,
      "step": 140
    },
    {
      "epoch": 0.42,
      "eval_accuracy": 0.5312520823615646,
      "eval_loss": 0.9766249656677246,
      "eval_runtime": 49.5213,
      "eval_samples_per_second": 606.083,
      "eval_steps_per_second": 1.191,
      "step": 140
    },
    {
      "epoch": 0.45,
      "learning_rate": 7.5e-06,
      "loss": 0.9854,
      "step": 150
    },
    {
      "epoch": 0.48,
      "learning_rate": 8e-06,
      "loss": 0.984,
      "step": 160
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.5398147531152129,
      "eval_loss": 0.9589501619338989,
      "eval_runtime": 49.4977,
      "eval_samples_per_second": 606.372,
      "eval_steps_per_second": 1.192,
      "step": 160
    },
    {
      "epoch": 0.51,
      "learning_rate": 8.500000000000002e-06,
      "loss": 0.9795,
      "step": 170
    },
    {
      "epoch": 0.54,
      "learning_rate": 9e-06,
      "loss": 0.9694,
      "step": 180
    },
    {
      "epoch": 0.54,
      "eval_accuracy": 0.5423469047777704,
      "eval_loss": 0.953514814376831,
      "eval_runtime": 49.4511,
      "eval_samples_per_second": 606.943,
      "eval_steps_per_second": 1.193,
      "step": 180
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.5e-06,
      "loss": 0.9757,
      "step": 190
    },
    {
      "epoch": 0.6,
      "learning_rate": 1e-05,
      "loss": 0.9676,
      "step": 200
    },
    {
      "epoch": 0.6,
      "eval_accuracy": 0.567201972412874,
      "eval_loss": 0.9273685812950134,
      "eval_runtime": 49.4122,
      "eval_samples_per_second": 607.421,
      "eval_steps_per_second": 1.194,
      "step": 200
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 0.9708,
      "step": 210
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.1e-05,
      "loss": 0.9753,
      "step": 220
    },
    {
      "epoch": 0.66,
      "eval_accuracy": 0.573598987139335,
      "eval_loss": 0.912590503692627,
      "eval_runtime": 49.2827,
      "eval_samples_per_second": 609.017,
      "eval_steps_per_second": 1.197,
      "step": 220
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.15e-05,
      "loss": 0.9586,
      "step": 230
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.2e-05,
      "loss": 0.9557,
      "step": 240
    },
    {
      "epoch": 0.72,
      "eval_accuracy": 0.5759645498767242,
      "eval_loss": 0.9052607417106628,
      "eval_runtime": 49.4457,
      "eval_samples_per_second": 607.01,
      "eval_steps_per_second": 1.193,
      "step": 240
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.25e-05,
      "loss": 0.9518,
      "step": 250
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.3e-05,
      "loss": 0.9508,
      "step": 260
    },
    {
      "epoch": 0.78,
      "eval_accuracy": 0.5766975411474645,
      "eval_loss": 0.9178985953330994,
      "eval_runtime": 49.3132,
      "eval_samples_per_second": 608.641,
      "eval_steps_per_second": 1.196,
      "step": 260
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.35e-05,
      "loss": 0.9405,
      "step": 270
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4e-05,
      "loss": 0.9355,
      "step": 280
    },
    {
      "epoch": 0.84,
      "eval_accuracy": 0.58915839275005,
      "eval_loss": 0.8937407732009888,
      "eval_runtime": 49.4342,
      "eval_samples_per_second": 607.15,
      "eval_steps_per_second": 1.194,
      "step": 280
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.4500000000000002e-05,
      "loss": 0.9235,
      "step": 290
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.5e-05,
      "loss": 0.9,
      "step": 300
    },
    {
      "epoch": 0.9,
      "eval_accuracy": 0.613013926834144,
      "eval_loss": 0.8468813300132751,
      "eval_runtime": 49.5046,
      "eval_samples_per_second": 606.287,
      "eval_steps_per_second": 1.192,
      "step": 300
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.55e-05,
      "loss": 0.8857,
      "step": 310
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6e-05,
      "loss": 0.993,
      "step": 320
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.6046511627906976,
      "eval_loss": 0.8615403771400452,
      "eval_runtime": 49.5041,
      "eval_samples_per_second": 606.294,
      "eval_steps_per_second": 1.192,
      "step": 320
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.65e-05,
      "loss": 0.8418,
      "step": 330
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.8527,
      "step": 340
    },
    {
      "epoch": 1.02,
      "eval_accuracy": 0.6439328313453722,
      "eval_loss": 0.7896137237548828,
      "eval_runtime": 49.4448,
      "eval_samples_per_second": 607.02,
      "eval_steps_per_second": 1.193,
      "step": 340
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.8982,
      "step": 350
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.8e-05,
      "loss": 0.966,
      "step": 360
    },
    {
      "epoch": 1.08,
      "eval_accuracy": 0.5315852602119011,
      "eval_loss": 1.0123510360717773,
      "eval_runtime": 49.4838,
      "eval_samples_per_second": 606.542,
      "eval_steps_per_second": 1.192,
      "step": 360
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.85e-05,
      "loss": 0.9413,
      "step": 370
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.9e-05,
      "loss": 0.8441,
      "step": 380
    },
    {
      "epoch": 1.14,
      "eval_accuracy": 0.6488638635303525,
      "eval_loss": 0.791083574295044,
      "eval_runtime": 49.4701,
      "eval_samples_per_second": 606.71,
      "eval_steps_per_second": 1.193,
      "step": 380
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.95e-05,
      "loss": 0.8223,
      "step": 390
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.8226,
      "step": 400
    },
    {
      "epoch": 1.2,
      "eval_accuracy": 0.6699873392416872,
      "eval_loss": 0.7472424507141113,
      "eval_runtime": 49.4938,
      "eval_samples_per_second": 606.419,
      "eval_steps_per_second": 1.192,
      "step": 400
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.05e-05,
      "loss": 0.7924,
      "step": 410
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 0.7948,
      "step": 420
    },
    {
      "epoch": 1.26,
      "eval_accuracy": 0.6580595721996402,
      "eval_loss": 0.7663838863372803,
      "eval_runtime": 49.5064,
      "eval_samples_per_second": 606.265,
      "eval_steps_per_second": 1.192,
      "step": 420
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.1499999999999997e-05,
      "loss": 0.776,
      "step": 430
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.2e-05,
      "loss": 0.7428,
      "step": 440
    },
    {
      "epoch": 1.32,
      "eval_accuracy": 0.6991737189311654,
      "eval_loss": 0.6993714570999146,
      "eval_runtime": 49.4888,
      "eval_samples_per_second": 606.481,
      "eval_steps_per_second": 1.192,
      "step": 440
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.2499999999999998e-05,
      "loss": 0.7512,
      "step": 450
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.3e-05,
      "loss": 0.7109,
      "step": 460
    },
    {
      "epoch": 1.38,
      "eval_accuracy": 0.7283600986206437,
      "eval_loss": 0.6510820984840393,
      "eval_runtime": 49.5999,
      "eval_samples_per_second": 605.122,
      "eval_steps_per_second": 1.19,
      "step": 460
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.3500000000000002e-05,
      "loss": 0.701,
      "step": 470
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.4e-05,
      "loss": 0.6882,
      "step": 480
    },
    {
      "epoch": 1.44,
      "eval_accuracy": 0.7576797494502565,
      "eval_loss": 0.5987845063209534,
      "eval_runtime": 49.5123,
      "eval_samples_per_second": 606.192,
      "eval_steps_per_second": 1.192,
      "step": 480
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.4500000000000003e-05,
      "loss": 0.651,
      "step": 490
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5e-05,
      "loss": 0.7296,
      "step": 500
    },
    {
      "epoch": 1.5,
      "eval_accuracy": 0.7564136736189778,
      "eval_loss": 0.5992804765701294,
      "eval_runtime": 49.5641,
      "eval_samples_per_second": 605.56,
      "eval_steps_per_second": 1.19,
      "step": 500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.55e-05,
      "loss": 0.5943,
      "step": 510
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.6e-05,
      "loss": 0.5677,
      "step": 520
    },
    {
      "epoch": 1.57,
      "eval_accuracy": 0.8125874591857133,
      "eval_loss": 0.5067983865737915,
      "eval_runtime": 49.3712,
      "eval_samples_per_second": 607.925,
      "eval_steps_per_second": 1.195,
      "step": 520
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.65e-05,
      "loss": 0.5818,
      "step": 530
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.7e-05,
      "loss": 0.5096,
      "step": 540
    },
    {
      "epoch": 1.63,
      "eval_accuracy": 0.8520023988805224,
      "eval_loss": 0.42726200819015503,
      "eval_runtime": 49.3448,
      "eval_samples_per_second": 608.251,
      "eval_steps_per_second": 1.196,
      "step": 540
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.75e-05,
      "loss": 0.462,
      "step": 550
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.8e-05,
      "loss": 0.4452,
      "step": 560
    },
    {
      "epoch": 1.69,
      "eval_accuracy": 0.8721929766109149,
      "eval_loss": 0.37960025668144226,
      "eval_runtime": 49.2893,
      "eval_samples_per_second": 608.936,
      "eval_steps_per_second": 1.197,
      "step": 560
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.85e-05,
      "loss": 0.4267,
      "step": 570
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.9000000000000004e-05,
      "loss": 0.3836,
      "step": 580
    },
    {
      "epoch": 1.75,
      "eval_accuracy": 0.8756580262544146,
      "eval_loss": 0.3854842483997345,
      "eval_runtime": 49.5207,
      "eval_samples_per_second": 606.09,
      "eval_steps_per_second": 1.191,
      "step": 580
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.95e-05,
      "loss": 0.3828,
      "step": 590
    },
    {
      "epoch": 1.81,
      "learning_rate": 3e-05,
      "loss": 0.3783,
      "step": 600
    },
    {
      "epoch": 1.81,
      "eval_accuracy": 0.8893849536882789,
      "eval_loss": 0.35855188965797424,
      "eval_runtime": 49.3852,
      "eval_samples_per_second": 607.753,
      "eval_steps_per_second": 1.195,
      "step": 600
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.05e-05,
      "loss": 0.3466,
      "step": 610
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1e-05,
      "loss": 0.3496,
      "step": 620
    },
    {
      "epoch": 1.87,
      "eval_accuracy": 0.8971813153861531,
      "eval_loss": 0.321043998003006,
      "eval_runtime": 49.4982,
      "eval_samples_per_second": 606.366,
      "eval_steps_per_second": 1.192,
      "step": 620
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.15e-05,
      "loss": 0.3421,
      "step": 630
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.2e-05,
      "loss": 0.3585,
      "step": 640
    },
    {
      "epoch": 1.93,
      "eval_accuracy": 0.9035450123275804,
      "eval_loss": 0.3006099760532379,
      "eval_runtime": 49.5179,
      "eval_samples_per_second": 606.124,
      "eval_steps_per_second": 1.191,
      "step": 640
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.3442,
      "step": 650
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.3e-05,
      "loss": 0.345,
      "step": 660
    },
    {
      "epoch": 1.99,
      "eval_accuracy": 0.9014126740854268,
      "eval_loss": 0.30540063977241516,
      "eval_runtime": 49.5044,
      "eval_samples_per_second": 606.289,
      "eval_steps_per_second": 1.192,
      "step": 660
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.35e-05,
      "loss": 0.3246,
      "step": 670
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.3327,
      "step": 680
    },
    {
      "epoch": 2.05,
      "eval_accuracy": 0.8912507496501633,
      "eval_loss": 0.3174073100090027,
      "eval_runtime": 49.4656,
      "eval_samples_per_second": 606.765,
      "eval_steps_per_second": 1.193,
      "step": 680
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.4500000000000005e-05,
      "loss": 0.3172,
      "step": 690
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 0.2962,
      "step": 700
    },
    {
      "epoch": 2.11,
      "eval_accuracy": 0.9122076364363297,
      "eval_loss": 0.2769884169101715,
      "eval_runtime": 49.5225,
      "eval_samples_per_second": 606.068,
      "eval_steps_per_second": 1.191,
      "step": 700
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.5499999999999996e-05,
      "loss": 0.287,
      "step": 710
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.6e-05,
      "loss": 0.3032,
      "step": 720
    },
    {
      "epoch": 2.17,
      "eval_accuracy": 0.9062437529153062,
      "eval_loss": 0.2979203760623932,
      "eval_runtime": 49.2758,
      "eval_samples_per_second": 609.102,
      "eval_steps_per_second": 1.197,
      "step": 720
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.65e-05,
      "loss": 0.3017,
      "step": 730
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.7e-05,
      "loss": 0.27,
      "step": 740
    },
    {
      "epoch": 2.23,
      "eval_accuracy": 0.8997801026187779,
      "eval_loss": 0.29734131693840027,
      "eval_runtime": 49.4882,
      "eval_samples_per_second": 606.488,
      "eval_steps_per_second": 1.192,
      "step": 740
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.75e-05,
      "loss": 0.3042,
      "step": 750
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.8e-05,
      "loss": 0.2912,
      "step": 760
    },
    {
      "epoch": 2.29,
      "eval_accuracy": 0.9221696541613914,
      "eval_loss": 0.24671417474746704,
      "eval_runtime": 49.5036,
      "eval_samples_per_second": 606.299,
      "eval_steps_per_second": 1.192,
      "step": 760
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.85e-05,
      "loss": 0.2609,
      "step": 770
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.9e-05,
      "loss": 0.2412,
      "step": 780
    },
    {
      "epoch": 2.35,
      "eval_accuracy": 0.9113080562404211,
      "eval_loss": 0.2760636806488037,
      "eval_runtime": 49.4644,
      "eval_samples_per_second": 606.78,
      "eval_steps_per_second": 1.193,
      "step": 780
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.95e-05,
      "loss": 0.2755,
      "step": 790
    },
    {
      "epoch": 2.41,
      "learning_rate": 4e-05,
      "loss": 0.2746,
      "step": 800
    },
    {
      "epoch": 2.41,
      "eval_accuracy": 0.9260011994402612,
      "eval_loss": 0.24102580547332764,
      "eval_runtime": 49.464,
      "eval_samples_per_second": 606.785,
      "eval_steps_per_second": 1.193,
      "step": 800
    }
  ],
  "logging_steps": 10,
  "max_steps": 996,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 1.077670425216942e+17,
  "trial_name": null,
  "trial_params": null
}