checkpoint-12/trainer_state.json

{
 "best_metric": null,
 "best_model_checkpoint": null,
 "epoch": 0.96,
 "eval_steps": 500,
 "global_step": 12,
 "is_hyper_param_search": false,
 "is_local_process_zero": true,
 "is_world_process_zero": true,
 "log_history": [
 {
 "epoch": 0.32,
 "grad_norm": 1.5884294509887695,
 "learning_rate": 2.4e-05,
 "loss": 1.9447,
 "step": 4
 },
 {
 "epoch": 0.64,
 "grad_norm": 1.335665225982666,
 "learning_rate": 1.2e-05,
 "loss": 1.7186,
 "step": 8
 },
 {
 "epoch": 0.96,
 "grad_norm": 1.2305809259414673,
 "learning_rate": 0.0,
 "loss": 1.7538,
 "step": 12
 }
 ],
 "logging_steps": 4,
 "max_steps": 12,
 "num_input_tokens_seen": 0,
 "num_train_epochs": 1,
 "save_steps": 500,
 "total_flos": 2099029034926080.0,
 "train_batch_size": 1,
 "trial_name": null,
 "trial_params": null
}