trainer_state.json

{
 "best_metric": null,
 "best_model_checkpoint": null,
 "epoch": 2.37034227742486,
 "eval_steps": 50000,
 "global_step": 100000,
 "is_hyper_param_search": false,
 "is_local_process_zero": true,
 "is_world_process_zero": true,
 "log_history": [
 {
 "epoch": 1.19,
 "learning_rate": 4.940766331658292e-05,
 "loss": 1.1651,
 "step": 50000
 },
 {
 "epoch": 1.19,
 "eval_loss": 1.026120662689209,
 "eval_runtime": 1432.5553,
 "eval_samples_per_second": 104.708,
 "eval_steps_per_second": 1.636,
 "step": 50000
 },
 {
 "epoch": 2.37,
 "learning_rate": 4.940763961316014e-05,
 "loss": 1.0237,
 "step": 100000
 },
 {
 "epoch": 2.37,
 "eval_loss": 0.9803563356399536,
 "eval_runtime": 1376.4724,
 "eval_samples_per_second": 108.974,
 "eval_steps_per_second": 1.703,
 "step": 100000
 }
 ],
 "logging_steps": 50000,
 "max_steps": 4218800,
 "num_input_tokens_seen": 0,
 "num_train_epochs": 100,
 "save_steps": 50000,
 "total_flos": 2.09031535263744e+17,
 "train_batch_size": 32,
 "trial_name": null,
 "trial_params": null
}