flytet2ycrlv6/local_flytekit/faa0c0e4bee4bd512a5fe448c325d4ac/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 66.66666666666667,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.67,
      "learning_rate": 2e-05,
      "loss": 1.9669,
      "step": 10
    },
    {
      "epoch": 13.33,
      "learning_rate": 2e-05,
      "loss": 0.9447,
      "step": 20
    },
    {
      "epoch": 20.0,
      "learning_rate": 2e-05,
      "loss": 0.1908,
      "step": 30
    },
    {
      "epoch": 26.67,
      "learning_rate": 2e-05,
      "loss": 0.0666,
      "step": 40
    },
    {
      "epoch": 33.33,
      "learning_rate": 2e-05,
      "loss": 0.0441,
      "step": 50
    },
    {
      "epoch": 40.0,
      "learning_rate": 2e-05,
      "loss": 0.0329,
      "step": 60
    },
    {
      "epoch": 46.67,
      "learning_rate": 2e-05,
      "loss": 0.0251,
      "step": 70
    },
    {
      "epoch": 53.33,
      "learning_rate": 2e-05,
      "loss": 0.019,
      "step": 80
    },
    {
      "epoch": 60.0,
      "learning_rate": 2e-05,
      "loss": 0.0166,
      "step": 90
    },
    {
      "epoch": 66.67,
      "learning_rate": 2e-05,
      "loss": 0.0133,
      "step": 100
    },
    {
      "epoch": 66.67,
      "step": 100,
      "total_flos": 42050959441920.0,
      "train_loss": 0.3319866207242012,
      "train_runtime": 4269.4137,
      "train_samples_per_second": 2.998,
      "train_steps_per_second": 0.023
    }
  ],
  "max_steps": 100,
  "num_train_epochs": 100,
  "total_flos": 42050959441920.0,
  "trial_name": null,
  "trial_params": null
}