{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 24135,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 4.8968303293971416e-05,
      "loss": 3.4484,
      "step": 500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.7932463227677646e-05,
      "loss": 3.355,
      "step": 1000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.6896623161383883e-05,
      "loss": 3.3179,
      "step": 1500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.586078309509012e-05,
      "loss": 3.2949,
      "step": 2000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.482494302879636e-05,
      "loss": 3.2563,
      "step": 2500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.3789102962502596e-05,
      "loss": 3.2521,
      "step": 3000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.2753262896208826e-05,
      "loss": 3.2356,
      "step": 3500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1717422829915064e-05,
      "loss": 3.2286,
      "step": 4000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.06815827636213e-05,
      "loss": 3.2046,
      "step": 4500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.964574269732754e-05,
      "loss": 3.1894,
      "step": 5000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.860990263103377e-05,
      "loss": 3.1449,
      "step": 5500
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.7574062564740006e-05,
      "loss": 3.1414,
      "step": 6000
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.653822249844624e-05,
      "loss": 3.1392,
      "step": 6500
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.5502382432152474e-05,
      "loss": 3.1346,
      "step": 7000
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.446654236585871e-05,
      "loss": 3.1401,
      "step": 7500
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.343070229956495e-05,
      "loss": 3.1233,
      "step": 8000
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.239693391340377e-05,
      "loss": 3.1274,
      "step": 8500
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.136109384711001e-05,
      "loss": 3.1225,
      "step": 9000
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.0325253780816242e-05,
      "loss": 3.1163,
      "step": 9500
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.928941371452248e-05,
      "loss": 3.0867,
      "step": 10000
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.82556453283613e-05,
      "loss": 3.073,
      "step": 10500
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.721980526206754e-05,
      "loss": 3.0646,
      "step": 11000
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.6183965195773773e-05,
      "loss": 3.0672,
      "step": 11500
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.514812512948001e-05,
      "loss": 3.0713,
      "step": 12000
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.411642842345142e-05,
      "loss": 3.0715,
      "step": 12500
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.3080588357157657e-05,
      "loss": 3.064,
      "step": 13000
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.204474829086389e-05,
      "loss": 3.0619,
      "step": 13500
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.100890822457013e-05,
      "loss": 3.0659,
      "step": 14000
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.9973068158276362e-05,
      "loss": 3.0686,
      "step": 14500
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.8937228091982596e-05,
      "loss": 3.024,
      "step": 15000
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.7901388025688834e-05,
      "loss": 3.0245,
      "step": 15500
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.6867619639527656e-05,
      "loss": 3.0451,
      "step": 16000
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.5831779573233893e-05,
      "loss": 3.0224,
      "step": 16500
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.4795939506940129e-05,
      "loss": 3.0343,
      "step": 17000
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.3760099440646364e-05,
      "loss": 3.0188,
      "step": 17500
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.2726331054485188e-05,
      "loss": 3.0242,
      "step": 18000
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.1690490988191424e-05,
      "loss": 3.0264,
      "step": 18500
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.065465092189766e-05,
      "loss": 3.0394,
      "step": 19000
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.618810855603895e-06,
      "loss": 3.025,
      "step": 19500
    },
    {
      "epoch": 4.14,
      "learning_rate": 8.585042469442718e-06,
      "loss": 2.9995,
      "step": 20000
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.549202403148954e-06,
      "loss": 3.0081,
      "step": 20500
    },
    {
      "epoch": 4.35,
      "learning_rate": 6.51336233685519e-06,
      "loss": 2.9937,
      "step": 21000
    },
    {
      "epoch": 4.45,
      "learning_rate": 5.4775222705614255e-06,
      "loss": 3.0028,
      "step": 21500
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.441682204267662e-06,
      "loss": 3.0022,
      "step": 22000
    },
    {
      "epoch": 4.66,
      "learning_rate": 3.4079138181064847e-06,
      "loss": 3.0221,
      "step": 22500
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.37207375181272e-06,
      "loss": 2.9937,
      "step": 23000
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.336233685518956e-06,
      "loss": 3.0063,
      "step": 23500
    },
    {
      "epoch": 4.97,
      "learning_rate": 3.0039361922519165e-07,
      "loss": 3.0098,
      "step": 24000
    },
    {
      "epoch": 5.0,
      "step": 24135,
      "total_flos": 5.044634615808e+16,
      "train_loss": 3.103326614534778,
      "train_runtime": 18434.3928,
      "train_samples_per_second": 10.473,
      "train_steps_per_second": 1.309
    }
  ],
  "max_steps": 24135,
  "num_train_epochs": 5,
  "total_flos": 5.044634615808e+16,
  "trial_name": null,
  "trial_params": null
}