checkpoint-32/trainer_state.json

{
 "best_metric": null,
 "best_model_checkpoint": null,
 "epoch": 0.9922480620155039,
 "eval_steps": 500,
 "global_step": 32,
 "is_hyper_param_search": false,
 "is_local_process_zero": true,
 "is_world_process_zero": true,
 "log_history": [
 {
 "epoch": 0.78,
 "grad_norm": 1.0154805183410645,
 "learning_rate": 5e-05,
 "loss": 3.6644,
 "step": 25
 }
 ],
 "logging_steps": 25,
 "max_steps": 32,
 "num_input_tokens_seen": 0,
 "num_train_epochs": 1,
 "save_steps": 500,
 "total_flos": 5202849294188544.0,
 "train_batch_size": 1,
 "trial_name": null,
 "trial_params": null
}