{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.99777942264989,
  "global_step": 16850,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_bleu": 1.7639,
      "eval_gen_len": 17.0476,
      "eval_loss": 2.9241816997528076,
      "eval_runtime": 161.7821,
      "eval_samples_per_second": 34.93,
      "eval_steps_per_second": 1.749,
      "step": 337
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.851632047477745e-05,
      "loss": 3.1098,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_bleu": 1.7951,
      "eval_gen_len": 15.622,
      "eval_loss": 2.894585609436035,
      "eval_runtime": 158.8854,
      "eval_samples_per_second": 35.567,
      "eval_steps_per_second": 1.781,
      "step": 674
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.70326409495549e-05,
      "loss": 2.8138,
      "step": 1000
    },
    {
      "epoch": 3.0,
      "eval_bleu": 1.9954,
      "eval_gen_len": 16.75,
      "eval_loss": 2.8898093700408936,
      "eval_runtime": 161.5184,
      "eval_samples_per_second": 34.987,
      "eval_steps_per_second": 1.752,
      "step": 1011
    },
    {
      "epoch": 4.0,
      "eval_bleu": 1.9243,
      "eval_gen_len": 16.4229,
      "eval_loss": 2.903644323348999,
      "eval_runtime": 160.1198,
      "eval_samples_per_second": 35.292,
      "eval_steps_per_second": 1.767,
      "step": 1348
    },
    {
      "epoch": 4.45,
      "learning_rate": 4.554896142433235e-05,
      "loss": 2.594,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "eval_bleu": 1.9376,
      "eval_gen_len": 16.8443,
      "eval_loss": 2.9253323078155518,
      "eval_runtime": 159.7166,
      "eval_samples_per_second": 35.381,
      "eval_steps_per_second": 1.772,
      "step": 1685
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.4065281899109794e-05,
      "loss": 2.4374,
      "step": 2000
    },
    {
      "epoch": 6.0,
      "eval_bleu": 2.0055,
      "eval_gen_len": 16.2769,
      "eval_loss": 2.9406914710998535,
      "eval_runtime": 159.0512,
      "eval_samples_per_second": 35.529,
      "eval_steps_per_second": 1.779,
      "step": 2022
    },
    {
      "epoch": 7.0,
      "eval_bleu": 1.9174,
      "eval_gen_len": 15.7742,
      "eval_loss": 2.978670358657837,
      "eval_runtime": 158.3136,
      "eval_samples_per_second": 35.695,
      "eval_steps_per_second": 1.788,
      "step": 2359
    },
    {
      "epoch": 7.42,
      "learning_rate": 4.258160237388724e-05,
      "loss": 2.2623,
      "step": 2500
    },
    {
      "epoch": 8.0,
      "eval_bleu": 1.9378,
      "eval_gen_len": 15.8482,
      "eval_loss": 3.0429296493530273,
      "eval_runtime": 157.9591,
      "eval_samples_per_second": 35.775,
      "eval_steps_per_second": 1.792,
      "step": 2696
    },
    {
      "epoch": 8.9,
      "learning_rate": 4.109792284866469e-05,
      "loss": 2.1323,
      "step": 3000
    },
    {
      "epoch": 9.0,
      "eval_bleu": 1.9663,
      "eval_gen_len": 16.4675,
      "eval_loss": 3.058178663253784,
      "eval_runtime": 158.2119,
      "eval_samples_per_second": 35.718,
      "eval_steps_per_second": 1.789,
      "step": 3033
    },
    {
      "epoch": 10.0,
      "eval_bleu": 1.9412,
      "eval_gen_len": 16.0646,
      "eval_loss": 3.107869863510132,
      "eval_runtime": 157.8461,
      "eval_samples_per_second": 35.801,
      "eval_steps_per_second": 1.793,
      "step": 3370
    },
    {
      "epoch": 10.38,
      "learning_rate": 3.961424332344214e-05,
      "loss": 1.9817,
      "step": 3500
    },
    {
      "epoch": 11.0,
      "eval_bleu": 2.0972,
      "eval_gen_len": 16.4849,
      "eval_loss": 3.1683783531188965,
      "eval_runtime": 158.1876,
      "eval_samples_per_second": 35.723,
      "eval_steps_per_second": 1.789,
      "step": 3707
    },
    {
      "epoch": 11.87,
      "learning_rate": 3.8130563798219586e-05,
      "loss": 1.864,
      "step": 4000
    },
    {
      "epoch": 12.0,
      "eval_bleu": 1.8631,
      "eval_gen_len": 16.3084,
      "eval_loss": 3.224560022354126,
      "eval_runtime": 158.1833,
      "eval_samples_per_second": 35.724,
      "eval_steps_per_second": 1.789,
      "step": 4044
    },
    {
      "epoch": 13.0,
      "eval_bleu": 1.9812,
      "eval_gen_len": 16.2702,
      "eval_loss": 3.256418228149414,
      "eval_runtime": 157.9844,
      "eval_samples_per_second": 35.769,
      "eval_steps_per_second": 1.791,
      "step": 4381
    },
    {
      "epoch": 13.35,
      "learning_rate": 3.664688427299703e-05,
      "loss": 1.7319,
      "step": 4500
    },
    {
      "epoch": 14.0,
      "eval_bleu": 1.8518,
      "eval_gen_len": 16.6208,
      "eval_loss": 3.321943759918213,
      "eval_runtime": 158.399,
      "eval_samples_per_second": 35.676,
      "eval_steps_per_second": 1.787,
      "step": 4718
    },
    {
      "epoch": 14.83,
      "learning_rate": 3.516320474777448e-05,
      "loss": 1.6291,
      "step": 5000
    },
    {
      "epoch": 15.0,
      "eval_bleu": 1.8638,
      "eval_gen_len": 15.8876,
      "eval_loss": 3.3599555492401123,
      "eval_runtime": 157.6851,
      "eval_samples_per_second": 35.837,
      "eval_steps_per_second": 1.795,
      "step": 5055
    },
    {
      "epoch": 16.0,
      "eval_bleu": 1.9703,
      "eval_gen_len": 16.2079,
      "eval_loss": 3.4334099292755127,
      "eval_runtime": 157.9443,
      "eval_samples_per_second": 35.778,
      "eval_steps_per_second": 1.792,
      "step": 5392
    },
    {
      "epoch": 16.32,
      "learning_rate": 3.3679525222551934e-05,
      "loss": 1.5185,
      "step": 5500
    },
    {
      "epoch": 17.0,
      "eval_bleu": 1.67,
      "eval_gen_len": 15.4355,
      "eval_loss": 3.4718282222747803,
      "eval_runtime": 157.5031,
      "eval_samples_per_second": 35.879,
      "eval_steps_per_second": 1.797,
      "step": 5729
    },
    {
      "epoch": 17.8,
      "learning_rate": 3.219584569732938e-05,
      "loss": 1.423,
      "step": 6000
    },
    {
      "epoch": 18.0,
      "eval_bleu": 1.834,
      "eval_gen_len": 16.5098,
      "eval_loss": 3.5062952041625977,
      "eval_runtime": 158.3097,
      "eval_samples_per_second": 35.696,
      "eval_steps_per_second": 1.788,
      "step": 6066
    },
    {
      "epoch": 19.0,
      "eval_bleu": 1.6657,
      "eval_gen_len": 16.0078,
      "eval_loss": 3.5681209564208984,
      "eval_runtime": 157.9659,
      "eval_samples_per_second": 35.774,
      "eval_steps_per_second": 1.792,
      "step": 6403
    },
    {
      "epoch": 19.29,
      "learning_rate": 3.071216617210683e-05,
      "loss": 1.3274,
      "step": 6500
    },
    {
      "epoch": 20.0,
      "eval_bleu": 1.9141,
      "eval_gen_len": 16.8841,
      "eval_loss": 3.612222194671631,
      "eval_runtime": 158.8209,
      "eval_samples_per_second": 35.581,
      "eval_steps_per_second": 1.782,
      "step": 6740
    },
    {
      "epoch": 20.77,
      "learning_rate": 2.9228486646884274e-05,
      "loss": 1.2461,
      "step": 7000
    },
    {
      "epoch": 21.0,
      "eval_bleu": 1.7505,
      "eval_gen_len": 16.1793,
      "eval_loss": 3.6515696048736572,
      "eval_runtime": 158.234,
      "eval_samples_per_second": 35.713,
      "eval_steps_per_second": 1.788,
      "step": 7077
    },
    {
      "epoch": 22.0,
      "eval_bleu": 1.6842,
      "eval_gen_len": 16.2152,
      "eval_loss": 3.6844277381896973,
      "eval_runtime": 158.2358,
      "eval_samples_per_second": 35.713,
      "eval_steps_per_second": 1.788,
      "step": 7414
    },
    {
      "epoch": 22.25,
      "learning_rate": 2.774480712166172e-05,
      "loss": 1.1661,
      "step": 7500
    },
    {
      "epoch": 23.0,
      "eval_bleu": 1.6847,
      "eval_gen_len": 16.0894,
      "eval_loss": 3.7410662174224854,
      "eval_runtime": 158.234,
      "eval_samples_per_second": 35.713,
      "eval_steps_per_second": 1.788,
      "step": 7751
    },
    {
      "epoch": 23.74,
      "learning_rate": 2.6261127596439174e-05,
      "loss": 1.092,
      "step": 8000
    },
    {
      "epoch": 24.0,
      "eval_bleu": 1.6662,
      "eval_gen_len": 16.6121,
      "eval_loss": 3.7708585262298584,
      "eval_runtime": 158.583,
      "eval_samples_per_second": 35.634,
      "eval_steps_per_second": 1.785,
      "step": 8088
    },
    {
      "epoch": 25.0,
      "eval_bleu": 1.7985,
      "eval_gen_len": 16.3143,
      "eval_loss": 3.8098456859588623,
      "eval_runtime": 158.302,
      "eval_samples_per_second": 35.698,
      "eval_steps_per_second": 1.788,
      "step": 8425
    },
    {
      "epoch": 25.22,
      "learning_rate": 2.4777448071216618e-05,
      "loss": 1.0302,
      "step": 8500
    },
    {
      "epoch": 26.0,
      "eval_bleu": 1.7239,
      "eval_gen_len": 16.3467,
      "eval_loss": 3.866185426712036,
      "eval_runtime": 158.3049,
      "eval_samples_per_second": 35.697,
      "eval_steps_per_second": 1.788,
      "step": 8762
    },
    {
      "epoch": 26.7,
      "learning_rate": 2.3293768545994066e-05,
      "loss": 0.9626,
      "step": 9000
    },
    {
      "epoch": 27.0,
      "eval_bleu": 1.7687,
      "eval_gen_len": 16.8252,
      "eval_loss": 3.9026572704315186,
      "eval_runtime": 158.776,
      "eval_samples_per_second": 35.591,
      "eval_steps_per_second": 1.782,
      "step": 9099
    },
    {
      "epoch": 28.0,
      "eval_bleu": 1.6071,
      "eval_gen_len": 16.8395,
      "eval_loss": 3.9381048679351807,
      "eval_runtime": 158.8794,
      "eval_samples_per_second": 35.568,
      "eval_steps_per_second": 1.781,
      "step": 9436
    },
    {
      "epoch": 28.19,
      "learning_rate": 2.1810089020771514e-05,
      "loss": 0.9076,
      "step": 9500
    },
    {
      "epoch": 29.0,
      "eval_bleu": 1.7271,
      "eval_gen_len": 16.5845,
      "eval_loss": 3.9773497581481934,
      "eval_runtime": 158.5329,
      "eval_samples_per_second": 35.646,
      "eval_steps_per_second": 1.785,
      "step": 9773
    },
    {
      "epoch": 29.67,
      "learning_rate": 2.0326409495548962e-05,
      "loss": 0.8523,
      "step": 10000
    },
    {
      "epoch": 30.0,
      "eval_bleu": 1.6689,
      "eval_gen_len": 16.5388,
      "eval_loss": 3.987967014312744,
      "eval_runtime": 158.4608,
      "eval_samples_per_second": 35.662,
      "eval_steps_per_second": 1.786,
      "step": 10110
    },
    {
      "epoch": 31.0,
      "eval_bleu": 1.6433,
      "eval_gen_len": 16.6489,
      "eval_loss": 4.056297302246094,
      "eval_runtime": 158.7288,
      "eval_samples_per_second": 35.602,
      "eval_steps_per_second": 1.783,
      "step": 10447
    },
    {
      "epoch": 31.16,
      "learning_rate": 1.884272997032641e-05,
      "loss": 0.8051,
      "step": 10500
    },
    {
      "epoch": 32.0,
      "eval_bleu": 1.6298,
      "eval_gen_len": 16.6305,
      "eval_loss": 4.071838855743408,
      "eval_runtime": 158.6637,
      "eval_samples_per_second": 35.616,
      "eval_steps_per_second": 1.784,
      "step": 10784
    },
    {
      "epoch": 32.64,
      "learning_rate": 1.7359050445103858e-05,
      "loss": 0.7621,
      "step": 11000
    },
    {
      "epoch": 33.0,
      "eval_bleu": 1.5555,
      "eval_gen_len": 16.7367,
      "eval_loss": 4.104060649871826,
      "eval_runtime": 158.7329,
      "eval_samples_per_second": 35.601,
      "eval_steps_per_second": 1.783,
      "step": 11121
    },
    {
      "epoch": 34.0,
      "eval_bleu": 1.5922,
      "eval_gen_len": 17.4091,
      "eval_loss": 4.128782272338867,
      "eval_runtime": 159.2897,
      "eval_samples_per_second": 35.476,
      "eval_steps_per_second": 1.777,
      "step": 11458
    },
    {
      "epoch": 34.12,
      "learning_rate": 1.5875370919881306e-05,
      "loss": 0.7257,
      "step": 11500
    },
    {
      "epoch": 35.0,
      "eval_bleu": 1.6243,
      "eval_gen_len": 16.8432,
      "eval_loss": 4.1702704429626465,
      "eval_runtime": 158.8223,
      "eval_samples_per_second": 35.581,
      "eval_steps_per_second": 1.782,
      "step": 11795
    },
    {
      "epoch": 35.61,
      "learning_rate": 1.4391691394658754e-05,
      "loss": 0.6835,
      "step": 12000
    },
    {
      "epoch": 36.0,
      "eval_bleu": 1.589,
      "eval_gen_len": 17.0446,
      "eval_loss": 4.1908135414123535,
      "eval_runtime": 159.0745,
      "eval_samples_per_second": 35.524,
      "eval_steps_per_second": 1.779,
      "step": 12132
    },
    {
      "epoch": 37.0,
      "eval_bleu": 1.5549,
      "eval_gen_len": 17.0649,
      "eval_loss": 4.204710006713867,
      "eval_runtime": 159.0216,
      "eval_samples_per_second": 35.536,
      "eval_steps_per_second": 1.78,
      "step": 12469
    },
    {
      "epoch": 37.09,
      "learning_rate": 1.29080118694362e-05,
      "loss": 0.6586,
      "step": 12500
    },
    {
      "epoch": 38.0,
      "eval_bleu": 1.5035,
      "eval_gen_len": 16.8954,
      "eval_loss": 4.236845970153809,
      "eval_runtime": 158.8765,
      "eval_samples_per_second": 35.569,
      "eval_steps_per_second": 1.781,
      "step": 12806
    },
    {
      "epoch": 38.57,
      "learning_rate": 1.142433234421365e-05,
      "loss": 0.6247,
      "step": 13000
    },
    {
      "epoch": 39.0,
      "eval_bleu": 1.5845,
      "eval_gen_len": 16.7616,
      "eval_loss": 4.257309436798096,
      "eval_runtime": 158.8065,
      "eval_samples_per_second": 35.584,
      "eval_steps_per_second": 1.782,
      "step": 13143
    },
    {
      "epoch": 40.0,
      "eval_bleu": 1.5144,
      "eval_gen_len": 16.803,
      "eval_loss": 4.269469738006592,
      "eval_runtime": 158.8495,
      "eval_samples_per_second": 35.575,
      "eval_steps_per_second": 1.782,
      "step": 13480
    },
    {
      "epoch": 40.06,
      "learning_rate": 9.940652818991098e-06,
      "loss": 0.604,
      "step": 13500
    },
    {
      "epoch": 41.0,
      "eval_bleu": 1.5398,
      "eval_gen_len": 17.1295,
      "eval_loss": 4.295245170593262,
      "eval_runtime": 159.0688,
      "eval_samples_per_second": 35.526,
      "eval_steps_per_second": 1.779,
      "step": 13817
    },
    {
      "epoch": 41.54,
      "learning_rate": 8.456973293768548e-06,
      "loss": 0.5781,
      "step": 14000
    },
    {
      "epoch": 42.0,
      "eval_bleu": 1.6057,
      "eval_gen_len": 17.0359,
      "eval_loss": 4.318183422088623,
      "eval_runtime": 158.942,
      "eval_samples_per_second": 35.554,
      "eval_steps_per_second": 1.781,
      "step": 14154
    },
    {
      "epoch": 43.0,
      "eval_bleu": 1.5114,
      "eval_gen_len": 16.6418,
      "eval_loss": 4.331396102905273,
      "eval_runtime": 158.745,
      "eval_samples_per_second": 35.598,
      "eval_steps_per_second": 1.783,
      "step": 14491
    },
    {
      "epoch": 43.03,
      "learning_rate": 6.973293768545995e-06,
      "loss": 0.5636,
      "step": 14500
    },
    {
      "epoch": 44.0,
      "eval_bleu": 1.5443,
      "eval_gen_len": 16.745,
      "eval_loss": 4.341202259063721,
      "eval_runtime": 158.7765,
      "eval_samples_per_second": 35.591,
      "eval_steps_per_second": 1.782,
      "step": 14828
    },
    {
      "epoch": 44.51,
      "learning_rate": 5.489614243323442e-06,
      "loss": 0.5429,
      "step": 15000
    },
    {
      "epoch": 45.0,
      "eval_bleu": 1.5719,
      "eval_gen_len": 16.8237,
      "eval_loss": 4.341938495635986,
      "eval_runtime": 158.7306,
      "eval_samples_per_second": 35.601,
      "eval_steps_per_second": 1.783,
      "step": 15165
    },
    {
      "epoch": 45.99,
      "learning_rate": 4.005934718100891e-06,
      "loss": 0.5309,
      "step": 15500
    },
    {
      "epoch": 46.0,
      "eval_bleu": 1.5445,
      "eval_gen_len": 16.7997,
      "eval_loss": 4.355504512786865,
      "eval_runtime": 158.7779,
      "eval_samples_per_second": 35.591,
      "eval_steps_per_second": 1.782,
      "step": 15502
    },
    {
      "epoch": 47.0,
      "eval_bleu": 1.5689,
      "eval_gen_len": 16.8358,
      "eval_loss": 4.362900733947754,
      "eval_runtime": 158.7996,
      "eval_samples_per_second": 35.586,
      "eval_steps_per_second": 1.782,
      "step": 15839
    },
    {
      "epoch": 47.48,
      "learning_rate": 2.5222551928783385e-06,
      "loss": 0.5206,
      "step": 16000
    },
    {
      "epoch": 48.0,
      "eval_bleu": 1.5208,
      "eval_gen_len": 17.0039,
      "eval_loss": 4.370855808258057,
      "eval_runtime": 158.9053,
      "eval_samples_per_second": 35.562,
      "eval_steps_per_second": 1.781,
      "step": 16176
    },
    {
      "epoch": 48.96,
      "learning_rate": 1.0385756676557863e-06,
      "loss": 0.51,
      "step": 16500
    },
    {
      "epoch": 49.0,
      "eval_bleu": 1.5581,
      "eval_gen_len": 16.8577,
      "eval_loss": 4.374205589294434,
      "eval_runtime": 158.7365,
      "eval_samples_per_second": 35.6,
      "eval_steps_per_second": 1.783,
      "step": 16513
    },
    {
      "epoch": 50.0,
      "eval_bleu": 1.5873,
      "eval_gen_len": 16.9289,
      "eval_loss": 4.376501083374023,
      "eval_runtime": 158.7713,
      "eval_samples_per_second": 35.592,
      "eval_steps_per_second": 1.782,
      "step": 16850
    },
    {
      "epoch": 50.0,
      "step": 16850,
      "total_flos": 1.8872801391015936e+17,
      "train_loss": 1.262480049472888,
      "train_runtime": 34907.6333,
      "train_samples_per_second": 38.699,
      "train_steps_per_second": 0.483
    }
  ],
  "max_steps": 16850,
  "num_train_epochs": 50,
  "total_flos": 1.8872801391015936e+17,
  "trial_name": null,
  "trial_params": null
}