{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.994219653179191,
  "eval_steps": 500,
  "global_step": 777,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04,
      "learning_rate": 1.9999666208982518e-05,
      "loss": 28.454,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9987985862949325e-05,
      "loss": 8.849,
      "step": 20
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9959638242644855e-05,
      "loss": 8.2489,
      "step": 30
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.991467065265775e-05,
      "loss": 4.2986,
      "step": 40
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.985315813185629e-05,
      "loss": 4.8588,
      "step": 50
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9775203328168643e-05,
      "loss": 9.355,
      "step": 60
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9680936327290924e-05,
      "loss": 5.7211,
      "step": 70
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.957051443560902e-05,
      "loss": 10.0795,
      "step": 80
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9444121917696335e-05,
      "loss": 3.6404,
      "step": 90
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.930196968882556e-05,
      "loss": 30.7862,
      "step": 100
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9144294963007542e-05,
      "loss": 6.9002,
      "step": 110
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.8971360857144616e-05,
      "loss": 8.5785,
      "step": 120
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.8783455951958948e-05,
      "loss": 5.644,
      "step": 130
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8580893810428562e-05,
      "loss": 14.8062,
      "step": 140
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8364012454534687e-05,
      "loss": 14.919,
      "step": 150
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.813317380119356e-05,
      "loss": 6.5775,
      "step": 160
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.7888763058314016e-05,
      "loss": 6.5815,
      "step": 170
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.763118808198859e-05,
      "loss": 4.1731,
      "step": 180
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.736087869589092e-05,
      "loss": 9.1267,
      "step": 190
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7078285974015103e-05,
      "loss": 3.9908,
      "step": 200
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.678388148795397e-05,
      "loss": 5.6814,
      "step": 210
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6478156519972354e-05,
      "loss": 5.0232,
      "step": 220
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6161621243188528e-05,
      "loss": 3.5724,
      "step": 230
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.5834803870231846e-05,
      "loss": 11.7568,
      "step": 240
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.549824977179731e-05,
      "loss": 5.393,
      "step": 250
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5152520566567873e-05,
      "loss": 6.7241,
      "step": 260
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.4798193184023233e-05,
      "loss": 3.7898,
      "step": 270
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4435858901698995e-05,
      "loss": 15.7335,
      "step": 280
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4066122358502772e-05,
      "loss": 2.6745,
      "step": 290
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.3689600545733713e-05,
      "loss": 5.8789,
      "step": 300
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.330692177748925e-05,
      "loss": 2.7451,
      "step": 310
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.2918724642177054e-05,
      "loss": 3.0655,
      "step": 320
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.252565693688198e-05,
      "loss": 1.6089,
      "step": 330
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2128374586366159e-05,
      "loss": 9.6673,
      "step": 340
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.172754054850619e-05,
      "loss": 1.9807,
      "step": 350
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1323823707993937e-05,
      "loss": 0.7278,
      "step": 360
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.091789776014706e-05,
      "loss": 5.7635,
      "step": 370
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0510440086691911e-05,
      "loss": 3.151,
      "step": 380
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0102130625394776e-05,
      "loss": 16.2403,
      "step": 390
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.693650735427808e-06,
      "loss": 2.06,
      "step": 400
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.285682060362974e-06,
      "loss": 3.2416,
      "step": 410
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.878905390691437e-06,
      "loss": 1.7038,
      "step": 420
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.473999527766503e-06,
      "loss": 1.4826,
      "step": 430
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.071640151065902e-06,
      "loss": 5.8947,
      "step": 440
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.672498690663632e-06,
      "loss": 6.4248,
      "step": 450
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.277241206792944e-06,
      "loss": 4.6138,
      "step": 460
    },
    {
      "epoch": 1.81,
      "learning_rate": 6.886527278370131e-06,
      "loss": 2.0887,
      "step": 470
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.501008902333912e-06,
      "loss": 3.9643,
      "step": 480
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.121329405637111e-06,
      "loss": 4.3215,
      "step": 490
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.748122371706198e-06,
      "loss": 3.4914,
      "step": 500
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.382010583160201e-06,
      "loss": 13.7674,
      "step": 510
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.0236049825532355e-06,
      "loss": 3.2402,
      "step": 520
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.673503652874977e-06,
      "loss": 0.7369,
      "step": 530
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.33229081951025e-06,
      "loss": 0.707,
      "step": 540
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.000535875323307e-06,
      "loss": 1.7793,
      "step": 550
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.6787924304935696e-06,
      "loss": 0.5325,
      "step": 560
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.3675973886884506e-06,
      "loss": 0.4873,
      "step": 570
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.0674700511149057e-06,
      "loss": 0.3436,
      "step": 580
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.7789112499447312e-06,
      "loss": 0.3319,
      "step": 590
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.502402512559773e-06,
      "loss": 0.4382,
      "step": 600
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.2384052580116465e-06,
      "loss": 0.2655,
      "step": 610
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.9873600270368664e-06,
      "loss": 1.3714,
      "step": 620
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.749685746912323e-06,
      "loss": 0.1256,
      "step": 630
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.52577903237781e-06,
      "loss": 0.2344,
      "step": 640
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.3160135237922011e-06,
      "loss": 0.4926,
      "step": 650
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.1207392636277502e-06,
      "loss": 0.369,
      "step": 660
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.402821123429017e-07,
      "loss": 0.1464,
      "step": 670
    },
    {
      "epoch": 2.62,
      "learning_rate": 7.749432046084471e-07,
      "loss": 0.5273,
      "step": 680
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.249984467943737e-07,
      "loss": 0.4998,
      "step": 690
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.906980565560004e-07,
      "loss": 2.2237,
      "step": 700
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.722661452877163e-07,
      "loss": 0.2594,
      "step": 710
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.699003441410508e-07,
      "loss": 0.3388,
      "step": 720
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.8377147423120467e-07,
      "loss": 0.4497,
      "step": 730
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.1402326158234e-07,
      "loss": 0.3246,
      "step": 740
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.077209728732492e-08,
      "loss": 2.0268,
      "step": 750
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.4106843282165615e-08,
      "loss": 0.1469,
      "step": 760
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.088684059220249e-09,
      "loss": 0.5057,
      "step": 770
    },
    {
      "epoch": 2.99,
      "step": 777,
      "total_flos": 1.0509820178622054e+17,
      "train_loss": 4.876755037362972,
      "train_runtime": 2316.9087,
      "train_samples_per_second": 21.497,
      "train_steps_per_second": 0.335
    }
  ],
  "logging_steps": 10,
  "max_steps": 777,
  "num_train_epochs": 3,
  "save_steps": 300,
  "total_flos": 1.0509820178622054e+17,
  "trial_name": null,
  "trial_params": null
}