trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.608695652173913,
  "eval_steps": 500,
  "global_step": 15,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "grad_norm": 1.3876663980871808,
      "learning_rate": 1e-05,
      "loss": 1.1674,
      "step": 1
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8785244493337074,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 1.1532,
      "step": 5
    },
    {
      "epoch": 0.87,
      "eval_loss": 1.0827027559280396,
      "eval_runtime": 39.1576,
      "eval_samples_per_second": 11.313,
      "eval_steps_per_second": 0.715,
      "step": 5
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8699830361300466,
      "learning_rate": 6.453951129574644e-06,
      "loss": 0.9871,
      "step": 10
    },
    {
      "epoch": 1.91,
      "eval_loss": 0.9193618297576904,
      "eval_runtime": 39.0986,
      "eval_samples_per_second": 11.33,
      "eval_steps_per_second": 0.716,
      "step": 11
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.6963222548011768,
      "learning_rate": 0.0,
      "loss": 0.8631,
      "step": 15
    },
    {
      "epoch": 2.61,
      "eval_loss": 0.8974742889404297,
      "eval_runtime": 39.2796,
      "eval_samples_per_second": 11.278,
      "eval_steps_per_second": 0.713,
      "step": 15
    },
    {
      "epoch": 2.61,
      "step": 15,
      "total_flos": 3088349921280.0,
      "train_loss": 1.0021109501520793,
      "train_runtime": 367.7198,
      "train_samples_per_second": 2.986,
      "train_steps_per_second": 0.041
    }
  ],
  "logging_steps": 5,
  "max_steps": 15,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 3088349921280.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}