trainer_state.json

{
  "best_metric": 1.1399264335632324,
  "best_model_checkpoint": "Action_model/checkpoint-100",
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 134,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15,
      "grad_norm": 1.509538173675537,
      "learning_rate": 8.507462686567164e-05,
      "loss": 2.1872,
      "step": 20
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6843363046646118,
      "learning_rate": 7.014925373134329e-05,
      "loss": 1.872,
      "step": 40
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.891447901725769,
      "learning_rate": 5.5223880597014934e-05,
      "loss": 1.5872,
      "step": 60
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.9932177066802979,
      "learning_rate": 4.029850746268657e-05,
      "loss": 1.3864,
      "step": 80
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7128252983093262,
      "learning_rate": 2.537313432835821e-05,
      "loss": 1.2948,
      "step": 100
    },
    {
      "epoch": 0.75,
      "eval_accuracy": 0.789103690685413,
      "eval_loss": 1.1399264335632324,
      "eval_runtime": 11.9209,
      "eval_samples_per_second": 47.731,
      "eval_steps_per_second": 6.04,
      "step": 100
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.182009696960449,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 1.2271,
      "step": 120
    },
    {
      "epoch": 1.0,
      "step": 134,
      "total_flos": 3.3230947683690086e+17,
      "train_loss": 1.549544946471257,
      "train_runtime": 145.4848,
      "train_samples_per_second": 29.474,
      "train_steps_per_second": 0.921
    }
  ],
  "logging_steps": 20,
  "max_steps": 134,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 3.3230947683690086e+17,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}