{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 24495,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 4.898550724637682e-05,
      "loss": 5.4391,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.79648907940396e-05,
      "loss": 3.2254,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.694427434170239e-05,
      "loss": 2.7119,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.592365788936518e-05,
      "loss": 2.4259,
      "step": 2000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.490304143702796e-05,
      "loss": 2.206,
      "step": 2500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.3884466217595436e-05,
      "loss": 2.1015,
      "step": 3000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.286384976525822e-05,
      "loss": 1.99,
      "step": 3500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.184527454582568e-05,
      "loss": 1.9223,
      "step": 4000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.082465809348847e-05,
      "loss": 1.8566,
      "step": 4500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9804041641151255e-05,
      "loss": 1.7873,
      "step": 5000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.8783425188814046e-05,
      "loss": 1.7276,
      "step": 5500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.776280873647684e-05,
      "loss": 1.6822,
      "step": 6000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.674219228413962e-05,
      "loss": 1.6448,
      "step": 6500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.572157583180241e-05,
      "loss": 1.5829,
      "step": 7000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.47009593794652e-05,
      "loss": 1.5763,
      "step": 7500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.368034292712799e-05,
      "loss": 1.5781,
      "step": 8000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.265972647479077e-05,
      "loss": 1.5427,
      "step": 8500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.1639110022453563e-05,
      "loss": 1.5025,
      "step": 9000
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.0618493570116355e-05,
      "loss": 1.4859,
      "step": 9500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.959787711777914e-05,
      "loss": 1.4483,
      "step": 10000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.857726066544193e-05,
      "loss": 1.4282,
      "step": 10500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7558685446009393e-05,
      "loss": 1.4401,
      "step": 11000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6538068993672177e-05,
      "loss": 1.3864,
      "step": 11500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5517452541334968e-05,
      "loss": 1.4,
      "step": 12000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4496836088997756e-05,
      "loss": 1.3831,
      "step": 12500
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3476219636660544e-05,
      "loss": 1.3708,
      "step": 13000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.245560318432333e-05,
      "loss": 1.3737,
      "step": 13500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.143498673198612e-05,
      "loss": 1.3142,
      "step": 14000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.041437027964891e-05,
      "loss": 1.333,
      "step": 14500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.939579506021637e-05,
      "loss": 1.3242,
      "step": 15000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.837517860787916e-05,
      "loss": 1.3056,
      "step": 15500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7356603388446624e-05,
      "loss": 1.2832,
      "step": 16000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.633598693610941e-05,
      "loss": 1.2872,
      "step": 16500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.53153704837722e-05,
      "loss": 1.2703,
      "step": 17000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4294754031434987e-05,
      "loss": 1.261,
      "step": 17500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3274137579097776e-05,
      "loss": 1.2512,
      "step": 18000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.2253521126760564e-05,
      "loss": 1.2212,
      "step": 18500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1232904674423353e-05,
      "loss": 1.2351,
      "step": 19000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0212288222086141e-05,
      "loss": 1.2337,
      "step": 19500
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.191671769748929e-06,
      "loss": 1.2258,
      "step": 20000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.171055317411718e-06,
      "loss": 1.2216,
      "step": 20500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.150438865074505e-06,
      "loss": 1.2057,
      "step": 21000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.129822412737294e-06,
      "loss": 1.2135,
      "step": 21500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.111247193304756e-06,
      "loss": 1.2054,
      "step": 22000
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.090630740967544e-06,
      "loss": 1.2021,
      "step": 22500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.0700142886303327e-06,
      "loss": 1.1868,
      "step": 23000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.049397836293121e-06,
      "loss": 1.1735,
      "step": 23500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0308226168605838e-06,
      "loss": 1.2107,
      "step": 24000
    },
    {
      "epoch": 3.0,
      "step": 24495,
      "total_flos": 1.290813800478588e+16,
      "train_loss": 1.6035245578856974,
      "train_runtime": 14452.6928,
      "train_samples_per_second": 1.695,
      "train_steps_per_second": 1.695
    }
  ],
  "max_steps": 24495,
  "num_train_epochs": 3,
  "total_flos": 1.290813800478588e+16,
  "trial_name": null,
  "trial_params": null
}