{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.3083005260197837,
  "global_step": 25000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "learning_rate": 4.9172842773954474e-05,
      "loss": 7.5606,
      "step": 500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8345685547908946e-05,
      "loss": 5.6106,
      "step": 1000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.751852832186342e-05,
      "loss": 4.6836,
      "step": 1500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.669137109581789e-05,
      "loss": 4.0373,
      "step": 2000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.586421386977237e-05,
      "loss": 3.6003,
      "step": 2500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.503705664372684e-05,
      "loss": 3.2726,
      "step": 3000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.4209899417681313e-05,
      "loss": 3.0363,
      "step": 3500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.338274219163579e-05,
      "loss": 2.8696,
      "step": 4000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.2555584965590264e-05,
      "loss": 2.7368,
      "step": 4500
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.1728427739544736e-05,
      "loss": 2.6329,
      "step": 5000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.090127051349921e-05,
      "loss": 2.5426,
      "step": 5500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.007411328745368e-05,
      "loss": 2.4677,
      "step": 6000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.924695606140815e-05,
      "loss": 2.4017,
      "step": 6500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.8419798835362624e-05,
      "loss": 2.3466,
      "step": 7000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.75926416093171e-05,
      "loss": 2.2957,
      "step": 7500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.6765484383271575e-05,
      "loss": 2.2443,
      "step": 8000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.593832715722605e-05,
      "loss": 2.2067,
      "step": 8500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.511116993118052e-05,
      "loss": 2.17,
      "step": 9000
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.428401270513499e-05,
      "loss": 2.1339,
      "step": 9500
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.345685547908946e-05,
      "loss": 2.1097,
      "step": 10000
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.2629698253043935e-05,
      "loss": 2.0808,
      "step": 10500
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.1802541026998414e-05,
      "loss": 2.0546,
      "step": 11000
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.0975383800952886e-05,
      "loss": 2.0371,
      "step": 11500
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.014822657490736e-05,
      "loss": 2.0158,
      "step": 12000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.9321069348861834e-05,
      "loss": 1.9924,
      "step": 12500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.849391212281631e-05,
      "loss": 1.9762,
      "step": 13000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.766675489677078e-05,
      "loss": 1.9568,
      "step": 13500
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.6839597670725253e-05,
      "loss": 1.9439,
      "step": 14000
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.6012440444679725e-05,
      "loss": 1.9298,
      "step": 14500
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.5185283218634197e-05,
      "loss": 1.9172,
      "step": 15000
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.4358125992588673e-05,
      "loss": 1.893,
      "step": 15500
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.3530968766543145e-05,
      "loss": 1.8762,
      "step": 16000
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.270381154049762e-05,
      "loss": 1.8717,
      "step": 16500
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.1876654314452092e-05,
      "loss": 1.8579,
      "step": 17000
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.1049497088406564e-05,
      "loss": 1.8454,
      "step": 17500
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.022233986236104e-05,
      "loss": 1.8387,
      "step": 18000
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.939518263631551e-05,
      "loss": 1.8287,
      "step": 18500
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.8568025410269984e-05,
      "loss": 1.8253,
      "step": 19000
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.7740868184224456e-05,
      "loss": 1.8145,
      "step": 19500
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.691371095817893e-05,
      "loss": 1.8095,
      "step": 20000
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.6086553732133407e-05,
      "loss": 1.8027,
      "step": 20500
    },
    {
      "epoch": 2.78,
      "learning_rate": 1.525939650608788e-05,
      "loss": 1.7982,
      "step": 21000
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.443223928004235e-05,
      "loss": 1.7908,
      "step": 21500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.3605082053996823e-05,
      "loss": 1.7863,
      "step": 22000
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.2777924827951298e-05,
      "loss": 1.7812,
      "step": 22500
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.1950767601905772e-05,
      "loss": 1.7693,
      "step": 23000
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.1123610375860244e-05,
      "loss": 1.7593,
      "step": 23500
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.0296453149814718e-05,
      "loss": 1.757,
      "step": 24000
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.469295923769191e-06,
      "loss": 1.7535,
      "step": 24500
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.642138697723663e-06,
      "loss": 1.7508,
      "step": 25000
    }
  ],
  "max_steps": 30224,
  "num_train_epochs": 4,
  "total_flos": 7.166643794309775e+17,
  "trial_name": null,
  "trial_params": null
}