checkpoint-10/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.24096385542168675,
  "eval_steps": 11,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 6.5,
      "learning_rate": 1e-05,
      "loss": 0.5887,
      "step": 1
    },
    {
      "epoch": 0.02,
      "eval_loss": 0.5842224955558777,
      "eval_runtime": 11.8233,
      "eval_samples_per_second": 1.184,
      "eval_steps_per_second": 0.085,
      "step": 1
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.28125,
      "learning_rate": 2e-05,
      "loss": 0.6588,
      "step": 2
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.859375,
      "learning_rate": 3e-05,
      "loss": 0.6446,
      "step": 3
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1484375,
      "learning_rate": 4e-05,
      "loss": 0.5767,
      "step": 4
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9609375,
      "learning_rate": 5e-05,
      "loss": 0.537,
      "step": 5
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0078125,
      "learning_rate": 6e-05,
      "loss": 0.5291,
      "step": 6
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7265625,
      "learning_rate": 7e-05,
      "loss": 0.4168,
      "step": 7
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.66796875,
      "learning_rate": 8e-05,
      "loss": 0.3394,
      "step": 8
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2578125,
      "learning_rate": 9e-05,
      "loss": 0.3067,
      "step": 9
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001,
      "loss": 0.2549,
      "step": 10
    }
  ],
  "logging_steps": 1,
  "max_steps": 164,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 5,
  "total_flos": 1.3690622720344064e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}