checkpoint-10/trainer_state.json

{
  "best_metric": 2.228980779647827,
  "best_model_checkpoint": "/scratch/kwamea/llama-output/checkpoint-10",
  "epoch": 0.7142857142857143,
  "eval_steps": 5,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.36,
      "learning_rate": 6.428571428571429e-05,
      "loss": 2.3035,
      "step": 5
    },
    {
      "epoch": 0.36,
      "eval_loss": 2.2832000255584717,
      "eval_runtime": 2.2848,
      "eval_samples_per_second": 2.626,
      "eval_steps_per_second": 0.438,
      "step": 5
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.857142857142857e-05,
      "loss": 2.201,
      "step": 10
    },
    {
      "epoch": 0.71,
      "eval_loss": 2.228980779647827,
      "eval_runtime": 2.2845,
      "eval_samples_per_second": 2.626,
      "eval_steps_per_second": 0.438,
      "step": 10
    }
  ],
  "logging_steps": 5,
  "max_steps": 14,
  "num_train_epochs": 1,
  "save_steps": 10,
  "total_flos": 3249703118438400.0,
  "trial_name": null,
  "trial_params": null
}