trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 345,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.43,
      "learning_rate": 1.710144927536232e-05,
      "loss": 0.615,
      "step": 50
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.420289855072464e-05,
      "loss": 0.5381,
      "step": 100
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 0.4313,
      "step": 150
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.405797101449275e-06,
      "loss": 0.4235,
      "step": 200
    },
    {
      "epoch": 2.17,
      "learning_rate": 5.507246376811595e-06,
      "loss": 0.3594,
      "step": 250
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.293,
      "step": 300
    },
    {
      "epoch": 3.0,
      "step": 345,
      "total_flos": 611246037148560.0,
      "train_loss": 0.42235686260720956,
      "train_runtime": 52.7528,
      "train_samples_per_second": 208.596,
      "train_steps_per_second": 6.54
    }
  ],
  "logging_steps": 50,
  "max_steps": 345,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 611246037148560.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}