{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 22176,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 2.9323593073593074e-05,
      "loss": 2.4787,
      "step": 500
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.8647186147186148e-05,
      "loss": 1.7517,
      "step": 1000
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.797077922077922e-05,
      "loss": 1.5948,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.7294372294372295e-05,
      "loss": 1.501,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.661796536796537e-05,
      "loss": 1.4643,
      "step": 2500
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.594155844155844e-05,
      "loss": 1.3776,
      "step": 3000
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.5265151515151516e-05,
      "loss": 1.3432,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.458874458874459e-05,
      "loss": 1.2899,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.3912337662337663e-05,
      "loss": 1.2779,
      "step": 4500
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.3235930735930737e-05,
      "loss": 1.2761,
      "step": 5000
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.255952380952381e-05,
      "loss": 1.2415,
      "step": 5500
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.1883116883116884e-05,
      "loss": 1.2253,
      "step": 6000
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1206709956709957e-05,
      "loss": 1.2359,
      "step": 6500
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.053030303030303e-05,
      "loss": 1.1721,
      "step": 7000
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9853896103896105e-05,
      "loss": 1.2008,
      "step": 7500
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9177489177489178e-05,
      "loss": 1.1782,
      "step": 8000
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.8501082251082252e-05,
      "loss": 1.1297,
      "step": 8500
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7824675324675322e-05,
      "loss": 1.1093,
      "step": 9000
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.71482683982684e-05,
      "loss": 1.1485,
      "step": 9500
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6471861471861473e-05,
      "loss": 1.0913,
      "step": 10000
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5795454545454546e-05,
      "loss": 1.0968,
      "step": 10500
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.511904761904762e-05,
      "loss": 1.0904,
      "step": 11000
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.4442640692640693e-05,
      "loss": 0.8584,
      "step": 11500
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.3766233766233767e-05,
      "loss": 0.7981,
      "step": 12000
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.308982683982684e-05,
      "loss": 0.8166,
      "step": 12500
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.2413419913419914e-05,
      "loss": 0.791,
      "step": 13000
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.1737012987012986e-05,
      "loss": 0.8021,
      "step": 13500
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.106060606060606e-05,
      "loss": 0.7986,
      "step": 14000
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.0384199134199135e-05,
      "loss": 0.7991,
      "step": 14500
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.707792207792209e-06,
      "loss": 0.7904,
      "step": 15000
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.031385281385282e-06,
      "loss": 0.7787,
      "step": 15500
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.354978354978356e-06,
      "loss": 0.749,
      "step": 16000
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.678571428571428e-06,
      "loss": 0.8107,
      "step": 16500
    },
    {
      "epoch": 1.53,
      "learning_rate": 7.002164502164502e-06,
      "loss": 0.7838,
      "step": 17000
    },
    {
      "epoch": 1.58,
      "learning_rate": 6.325757575757576e-06,
      "loss": 0.8014,
      "step": 17500
    },
    {
      "epoch": 1.62,
      "learning_rate": 5.64935064935065e-06,
      "loss": 0.7691,
      "step": 18000
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.972943722943723e-06,
      "loss": 0.7493,
      "step": 18500
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.2965367965367965e-06,
      "loss": 0.7534,
      "step": 19000
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.62012987012987e-06,
      "loss": 0.7561,
      "step": 19500
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.943722943722944e-06,
      "loss": 0.751,
      "step": 20000
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.2673160173160173e-06,
      "loss": 0.7453,
      "step": 20500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.590909090909091e-06,
      "loss": 0.7336,
      "step": 21000
    },
    {
      "epoch": 1.94,
      "learning_rate": 9.145021645021645e-07,
      "loss": 0.7211,
      "step": 21500
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.380952380952381e-07,
      "loss": 0.7327,
      "step": 22000
    },
    {
      "epoch": 2.0,
      "step": 22176,
      "total_flos": 5.214986800612762e+16,
      "train_loss": 1.0508460692341022,
      "train_runtime": 7174.5051,
      "train_samples_per_second": 37.091,
      "train_steps_per_second": 3.091
    }
  ],
  "max_steps": 22176,
  "num_train_epochs": 2,
  "total_flos": 5.214986800612762e+16,
  "trial_name": null,
  "trial_params": null
}