{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 14286,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 4.8250034999300015e-05,
      "loss": 1.4008,
      "step": 500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.6500069998600034e-05,
      "loss": 0.9035,
      "step": 1000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.475010499790005e-05,
      "loss": 0.7891,
      "step": 1500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.300013999720006e-05,
      "loss": 0.7111,
      "step": 2000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.125017499650007e-05,
      "loss": 0.6457,
      "step": 2500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.9500209995800084e-05,
      "loss": 0.6018,
      "step": 3000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.77502449951001e-05,
      "loss": 0.5633,
      "step": 3500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.6000279994400116e-05,
      "loss": 0.5295,
      "step": 4000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.425031499370013e-05,
      "loss": 0.4894,
      "step": 4500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.250034999300014e-05,
      "loss": 0.4172,
      "step": 5000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.075038499230016e-05,
      "loss": 0.3773,
      "step": 5500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.900041999160017e-05,
      "loss": 0.3746,
      "step": 6000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7250454990900182e-05,
      "loss": 0.3461,
      "step": 6500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5500489990200198e-05,
      "loss": 0.3377,
      "step": 7000
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.375052498950021e-05,
      "loss": 0.3257,
      "step": 7500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2000559988800227e-05,
      "loss": 0.3138,
      "step": 8000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.025059498810024e-05,
      "loss": 0.3026,
      "step": 8500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8500629987400252e-05,
      "loss": 0.289,
      "step": 9000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6750664986700268e-05,
      "loss": 0.2805,
      "step": 9500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.500069998600028e-05,
      "loss": 0.2168,
      "step": 10000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3250734985300295e-05,
      "loss": 0.2078,
      "step": 10500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1500769984600309e-05,
      "loss": 0.207,
      "step": 11000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.750804983900321e-06,
      "loss": 0.1935,
      "step": 11500
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.000839983200337e-06,
      "loss": 0.1942,
      "step": 12000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.25087498250035e-06,
      "loss": 0.1903,
      "step": 12500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.500909981800364e-06,
      "loss": 0.1738,
      "step": 13000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7509449811003783e-06,
      "loss": 0.1779,
      "step": 13500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.000979980400392e-06,
      "loss": 0.1685,
      "step": 14000
    },
    {
      "epoch": 3.0,
      "step": 14286,
      "total_flos": 1.2205375337376e+17,
      "train_loss": 0.4136880006540494,
      "train_runtime": 34968.5642,
      "train_samples_per_second": 26.142,
      "train_steps_per_second": 0.409
    }
  ],
  "logging_steps": 500,
  "max_steps": 14286,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.2205375337376e+17,
  "trial_name": null,
  "trial_params": null
}