{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.999725902238465,
  "eval_steps": 2000,
  "global_step": 16416,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09,
      "learning_rate": 4.8477095516569206e-05,
      "loss": 5.1276,
      "step": 500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.69541910331384e-05,
      "loss": 3.7227,
      "step": 1000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.54312865497076e-05,
      "loss": 3.343,
      "step": 1500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.39083820662768e-05,
      "loss": 3.0755,
      "step": 2000
    },
    {
      "epoch": 0.37,
      "eval_loss": 2.8688604831695557,
      "eval_runtime": 5.135,
      "eval_samples_per_second": 19.474,
      "eval_steps_per_second": 4.869,
      "step": 2000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.238547758284601e-05,
      "loss": 2.9129,
      "step": 2500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.086257309941521e-05,
      "loss": 2.7827,
      "step": 3000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.933966861598441e-05,
      "loss": 2.6733,
      "step": 3500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7816764132553604e-05,
      "loss": 2.5673,
      "step": 4000
    },
    {
      "epoch": 0.73,
      "eval_loss": 2.4161481857299805,
      "eval_runtime": 5.1947,
      "eval_samples_per_second": 19.251,
      "eval_steps_per_second": 4.813,
      "step": 4000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.629385964912281e-05,
      "loss": 2.5116,
      "step": 4500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.477095516569201e-05,
      "loss": 2.4446,
      "step": 5000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3248050682261215e-05,
      "loss": 2.3883,
      "step": 5500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.172514619883041e-05,
      "loss": 2.2716,
      "step": 6000
    },
    {
      "epoch": 1.1,
      "eval_loss": 2.184684991836548,
      "eval_runtime": 5.2889,
      "eval_samples_per_second": 18.908,
      "eval_steps_per_second": 4.727,
      "step": 6000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0202241715399608e-05,
      "loss": 2.2207,
      "step": 6500
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8679337231968812e-05,
      "loss": 2.1887,
      "step": 7000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7156432748538012e-05,
      "loss": 2.1625,
      "step": 7500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5633528265107216e-05,
      "loss": 2.1425,
      "step": 8000
    },
    {
      "epoch": 1.46,
      "eval_loss": 2.050933837890625,
      "eval_runtime": 5.1426,
      "eval_samples_per_second": 19.446,
      "eval_steps_per_second": 4.861,
      "step": 8000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4110623781676416e-05,
      "loss": 2.1204,
      "step": 8500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.2587719298245616e-05,
      "loss": 2.0851,
      "step": 9000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1064814814814816e-05,
      "loss": 2.0784,
      "step": 9500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9541910331384016e-05,
      "loss": 2.0479,
      "step": 10000
    },
    {
      "epoch": 1.83,
      "eval_loss": 1.95827054977417,
      "eval_runtime": 5.3452,
      "eval_samples_per_second": 18.708,
      "eval_steps_per_second": 4.677,
      "step": 10000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8019005847953216e-05,
      "loss": 2.0349,
      "step": 10500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6496101364522417e-05,
      "loss": 2.0088,
      "step": 11000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.497319688109162e-05,
      "loss": 1.9606,
      "step": 11500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3450292397660819e-05,
      "loss": 1.9389,
      "step": 12000
    },
    {
      "epoch": 2.19,
      "eval_loss": 1.9141876697540283,
      "eval_runtime": 5.2433,
      "eval_samples_per_second": 19.072,
      "eval_steps_per_second": 4.768,
      "step": 12000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.192738791423002e-05,
      "loss": 1.9075,
      "step": 12500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.040448343079922e-05,
      "loss": 1.9313,
      "step": 13000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.881578947368421e-06,
      "loss": 1.9012,
      "step": 13500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.358674463937622e-06,
      "loss": 1.9041,
      "step": 14000
    },
    {
      "epoch": 2.56,
      "eval_loss": 1.8739492893218994,
      "eval_runtime": 5.3315,
      "eval_samples_per_second": 18.756,
      "eval_steps_per_second": 4.689,
      "step": 14000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.835769980506823e-06,
      "loss": 1.9066,
      "step": 14500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.312865497076023e-06,
      "loss": 1.9035,
      "step": 15000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7899610136452243e-06,
      "loss": 1.8883,
      "step": 15500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.2670565302144249e-06,
      "loss": 1.8992,
      "step": 16000
    },
    {
      "epoch": 2.92,
      "eval_loss": 1.8671270608901978,
      "eval_runtime": 5.2706,
      "eval_samples_per_second": 18.973,
      "eval_steps_per_second": 4.743,
      "step": 16000
    },
    {
      "epoch": 3.0,
      "step": 16416,
      "total_flos": 6.325314943920538e+17,
      "train_loss": 2.364324833682174,
      "train_runtime": 42522.3186,
      "train_samples_per_second": 6.177,
      "train_steps_per_second": 0.386
    }
  ],
  "logging_steps": 500,
  "max_steps": 16416,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 6.325314943920538e+17,
  "trial_name": null,
  "trial_params": null
}