trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 11.0,
  "eval_steps": 500,
  "global_step": 2409,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.28,
      "learning_rate": 3.9622249896222504e-05,
      "loss": 1.9626,
      "step": 500
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.9244499792445003e-05,
      "loss": 1.7347,
      "step": 1000
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.8866749688667497e-05,
      "loss": 1.4932,
      "step": 1500
    },
    {
      "epoch": 9.13,
      "learning_rate": 8.488999584889996e-06,
      "loss": 1.2625,
      "step": 2000
    },
    {
      "epoch": 11.0,
      "step": 2409,
      "total_flos": 1105239600655344.0,
      "train_loss": 1.5172127248242269,
      "train_runtime": 9372.6529,
      "train_samples_per_second": 2.056,
      "train_steps_per_second": 0.257
    }
  ],
  "logging_steps": 500,
  "max_steps": 2409,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 11,
  "save_steps": 500,
  "total_flos": 1105239600655344.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}