pythia-1B-deduped-wikipedia-8bit
/
flyteva2a9_9t
/local_flytekit
/df86818aaa4263bdf5800c2d18124919
/trainer_state.json
{ | |
"best_metric": null, | |
"best_model_checkpoint": null, | |
"epoch": 3.3333333333333335, | |
"global_step": 10, | |
"is_hyper_param_search": false, | |
"is_local_process_zero": true, | |
"is_world_process_zero": true, | |
"log_history": [ | |
{ | |
"epoch": 3.33, | |
"learning_rate": 2e-05, | |
"loss": 2.1837, | |
"step": 10 | |
}, | |
{ | |
"epoch": 3.33, | |
"step": 10, | |
"total_flos": 835002040320.0, | |
"train_loss": 2.1836517333984373, | |
"train_runtime": 74.2827, | |
"train_samples_per_second": 8.616, | |
"train_steps_per_second": 0.135 | |
} | |
], | |
"max_steps": 10, | |
"num_train_epochs": 4, | |
"total_flos": 835002040320.0, | |
"trial_name": null, | |
"trial_params": null | |
} | |