{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.999138673557278,
  "eval_steps": 500,
  "global_step": 14510,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "learning_rate": 4.9800000000000004e-05,
      "loss": 0.0562,
      "step": 500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.984428128696318e-05,
      "loss": 0.0703,
      "step": 1000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.9376579221363204e-05,
      "loss": 0.0744,
      "step": 1500
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.8602764529345966e-05,
      "loss": 0.0807,
      "step": 2000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.7534982401683416e-05,
      "loss": 0.082,
      "step": 2500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.618236648123212e-05,
      "loss": 0.072,
      "step": 3000
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.4563749749985335e-05,
      "loss": 0.0638,
      "step": 3500
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.269945823540843e-05,
      "loss": 0.0627,
      "step": 4000
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.061290306367919e-05,
      "loss": 0.0638,
      "step": 4500
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.8330286470885907e-05,
      "loss": 0.0652,
      "step": 5000
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.58802727644352e-05,
      "loss": 0.0622,
      "step": 5500
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.3298916668553426e-05,
      "loss": 0.0555,
      "step": 6000
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.0608298703277996e-05,
      "loss": 0.0467,
      "step": 6500
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.7847253668223704e-05,
      "loss": 0.0428,
      "step": 7000
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.5050453809223793e-05,
      "loss": 0.0473,
      "step": 7500
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.2253020368788445e-05,
      "loss": 0.0452,
      "step": 8000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.9495550817504742e-05,
      "loss": 0.0412,
      "step": 8500
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.6801632122617095e-05,
      "loss": 0.0385,
      "step": 9000
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.4210665687785734e-05,
      "loss": 0.0311,
      "step": 9500
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.1755187970777065e-05,
      "loss": 0.0317,
      "step": 10000
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.470426834280832e-06,
      "loss": 0.0309,
      "step": 10500
    },
    {
      "epoch": 3.79,
      "learning_rate": 7.3759257710121985e-06,
      "loss": 0.0296,
      "step": 11000
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.5027417186915175e-06,
      "loss": 0.0261,
      "step": 11500
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.877395681105919e-06,
      "loss": 0.0268,
      "step": 12000
    },
    {
      "epoch": 4.31,
      "learning_rate": 2.513790616397402e-06,
      "loss": 0.0261,
      "step": 12500
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.4325595693915228e-06,
      "loss": 0.022,
      "step": 13000
    },
    {
      "epoch": 4.65,
      "learning_rate": 6.472802645326154e-07,
      "loss": 0.0205,
      "step": 13500
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.6846304762653532e-07,
      "loss": 0.025,
      "step": 14000
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.0364694921504168e-10,
      "loss": 0.0274,
      "step": 14500
    }
  ],
  "logging_steps": 500,
  "max_steps": 14510,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 1.2300127834804224e+17,
  "trial_name": null,
  "trial_params": null
}