{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 6158,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.9675219227021763e-05,
      "loss": 2.0573,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.935043845404352e-05,
      "loss": 0.2273,
      "step": 200
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.902565768106528e-05,
      "loss": 0.1299,
      "step": 300
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.870087690808704e-05,
      "loss": 0.0909,
      "step": 400
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.8376096135108803e-05,
      "loss": 0.0606,
      "step": 500
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8051315362130565e-05,
      "loss": 0.053,
      "step": 600
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7726534589152323e-05,
      "loss": 0.0426,
      "step": 700
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.740175381617408e-05,
      "loss": 0.0228,
      "step": 800
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.7076973043195843e-05,
      "loss": 0.0266,
      "step": 900
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6752192270217605e-05,
      "loss": 0.033,
      "step": 1000
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.6427411497239367e-05,
      "loss": 0.0226,
      "step": 1100
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.6102630724261125e-05,
      "loss": 0.03,
      "step": 1200
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.5777849951282887e-05,
      "loss": 0.0236,
      "step": 1300
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.5453069178304645e-05,
      "loss": 0.0328,
      "step": 1400
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.5128288405326405e-05,
      "loss": 0.0103,
      "step": 1500
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.4803507632348167e-05,
      "loss": 0.0135,
      "step": 1600
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.4478726859369926e-05,
      "loss": 0.0058,
      "step": 1700
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.4153946086391688e-05,
      "loss": 0.0212,
      "step": 1800
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.3829165313413446e-05,
      "loss": 0.0179,
      "step": 1900
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.3504384540435206e-05,
      "loss": 0.0216,
      "step": 2000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.3179603767456968e-05,
      "loss": 0.0249,
      "step": 2100
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.2854822994478728e-05,
      "loss": 0.0455,
      "step": 2200
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.253004222150049e-05,
      "loss": 0.0123,
      "step": 2300
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.2205261448522248e-05,
      "loss": 0.0148,
      "step": 2400
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.1880480675544008e-05,
      "loss": 0.0255,
      "step": 2500
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.155569990256577e-05,
      "loss": 0.0055,
      "step": 2600
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.123091912958753e-05,
      "loss": 0.0061,
      "step": 2700
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.090613835660929e-05,
      "loss": 0.0229,
      "step": 2800
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.058135758363105e-05,
      "loss": 0.007,
      "step": 2900
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.025657681065281e-05,
      "loss": 0.0144,
      "step": 3000
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.93179603767457e-06,
      "loss": 0.007,
      "step": 3100
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.607015264696331e-06,
      "loss": 0.0056,
      "step": 3200
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.282234491718091e-06,
      "loss": 0.0063,
      "step": 3300
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.957453718739851e-06,
      "loss": 0.0193,
      "step": 3400
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.632672945761611e-06,
      "loss": 0.0115,
      "step": 3500
    },
    {
      "epoch": 1.17,
      "learning_rate": 8.307892172783371e-06,
      "loss": 0.0035,
      "step": 3600
    },
    {
      "epoch": 1.2,
      "learning_rate": 7.983111399805133e-06,
      "loss": 0.009,
      "step": 3700
    },
    {
      "epoch": 1.23,
      "learning_rate": 7.658330626826891e-06,
      "loss": 0.0122,
      "step": 3800
    },
    {
      "epoch": 1.27,
      "learning_rate": 7.333549853848653e-06,
      "loss": 0.001,
      "step": 3900
    },
    {
      "epoch": 1.3,
      "learning_rate": 7.008769080870414e-06,
      "loss": 0.0003,
      "step": 4000
    },
    {
      "epoch": 1.33,
      "learning_rate": 6.683988307892173e-06,
      "loss": 0.0114,
      "step": 4100
    },
    {
      "epoch": 1.36,
      "learning_rate": 6.359207534913934e-06,
      "loss": 0.0036,
      "step": 4200
    },
    {
      "epoch": 1.4,
      "learning_rate": 6.034426761935695e-06,
      "loss": 0.0109,
      "step": 4300
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.709645988957454e-06,
      "loss": 0.0006,
      "step": 4400
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.384865215979215e-06,
      "loss": 0.005,
      "step": 4500
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.0600844430009746e-06,
      "loss": 0.0046,
      "step": 4600
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.735303670022735e-06,
      "loss": 0.0064,
      "step": 4700
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.410522897044495e-06,
      "loss": 0.0063,
      "step": 4800
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.085742124066255e-06,
      "loss": 0.0004,
      "step": 4900
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.760961351088016e-06,
      "loss": 0.0106,
      "step": 5000
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.436180578109776e-06,
      "loss": 0.004,
      "step": 5100
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.111399805131536e-06,
      "loss": 0.0053,
      "step": 5200
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.7866190321532966e-06,
      "loss": 0.0096,
      "step": 5300
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.461838259175057e-06,
      "loss": 0.0009,
      "step": 5400
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.1370574861968174e-06,
      "loss": 0.0068,
      "step": 5500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.8122767132185776e-06,
      "loss": 0.0054,
      "step": 5600
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.487495940240338e-06,
      "loss": 0.0005,
      "step": 5700
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.1627151672620982e-06,
      "loss": 0.0012,
      "step": 5800
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.379343942838584e-07,
      "loss": 0.0093,
      "step": 5900
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.131536213056187e-07,
      "loss": 0.0036,
      "step": 6000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.8837284832737905e-07,
      "loss": 0.0005,
      "step": 6100
    }
  ],
  "max_steps": 6158,
  "num_train_epochs": 2,
  "total_flos": 2.1662462190359347e+19,
  "trial_name": null,
  "trial_params": null
}