checkpoint-50/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5,
  "eval_steps": 20,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 1e-05,
      "loss": 0.588,
      "step": 10
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.5793,
      "step": 20
    },
    {
      "epoch": 0.2,
      "eval_loss": 0.5849277377128601,
      "eval_runtime": 13.5251,
      "eval_samples_per_second": 1.848,
      "eval_steps_per_second": 0.296,
      "step": 20
    },
    {
      "epoch": 0.3,
      "learning_rate": 3e-05,
      "loss": 0.5065,
      "step": 30
    },
    {
      "epoch": 0.4,
      "learning_rate": 4e-05,
      "loss": 0.5404,
      "step": 40
    },
    {
      "epoch": 0.4,
      "eval_loss": 0.5613898634910583,
      "eval_runtime": 13.3646,
      "eval_samples_per_second": 1.871,
      "eval_steps_per_second": 0.299,
      "step": 40
    },
    {
      "epoch": 0.5,
      "learning_rate": 5e-05,
      "loss": 0.4523,
      "step": 50
    }
  ],
  "logging_steps": 10,
  "max_steps": 1000,
  "num_train_epochs": 10,
  "save_steps": 50,
  "total_flos": 7147125638258688.0,
  "trial_name": null,
  "trial_params": null
}