trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "global_step": 2700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.26,
      "learning_rate": 4.074074074074074e-05,
      "loss": 1.7192,
      "step": 500
    },
    {
      "epoch": 18.52,
      "learning_rate": 3.148148148148148e-05,
      "loss": 0.9638,
      "step": 1000
    },
    {
      "epoch": 27.78,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.7646,
      "step": 1500
    },
    {
      "epoch": 37.04,
      "learning_rate": 1.2962962962962962e-05,
      "loss": 0.6811,
      "step": 2000
    },
    {
      "epoch": 46.3,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.6386,
      "step": 2500
    },
    {
      "epoch": 50.0,
      "step": 2700,
      "total_flos": 7.4663889076224e+17,
      "train_loss": 0.9289147666648582,
      "train_runtime": 1292.9621,
      "train_samples_per_second": 32.87,
      "train_steps_per_second": 2.088
    }
  ],
  "max_steps": 2700,
  "num_train_epochs": 50,
  "total_flos": 7.4663889076224e+17,
  "trial_name": null,
  "trial_params": null
}