|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.05102040816326531, |
|
"eval_steps": 2, |
|
"global_step": 5, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01020408163265306, |
|
"grad_norm": 11.50589656829834, |
|
"learning_rate": 1e-05, |
|
"loss": 1.0181, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.01020408163265306, |
|
"eval_loss": 1.6529520750045776, |
|
"eval_runtime": 194.8465, |
|
"eval_samples_per_second": 0.847, |
|
"eval_steps_per_second": 0.426, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.02040816326530612, |
|
"grad_norm": 10.689672470092773, |
|
"learning_rate": 2e-05, |
|
"loss": 1.2269, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.02040816326530612, |
|
"eval_loss": 1.477925181388855, |
|
"eval_runtime": 4.704, |
|
"eval_samples_per_second": 35.077, |
|
"eval_steps_per_second": 17.645, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.030612244897959183, |
|
"grad_norm": 9.542746543884277, |
|
"learning_rate": 3e-05, |
|
"loss": 1.1125, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.04081632653061224, |
|
"grad_norm": 8.531493186950684, |
|
"learning_rate": 4e-05, |
|
"loss": 2.2516, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.04081632653061224, |
|
"eval_loss": 1.1037262678146362, |
|
"eval_runtime": 4.713, |
|
"eval_samples_per_second": 35.009, |
|
"eval_steps_per_second": 17.611, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.05102040816326531, |
|
"grad_norm": 7.03532600402832, |
|
"learning_rate": 5e-05, |
|
"loss": 0.5774, |
|
"step": 5 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 5, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 2, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 4156623830384640.0, |
|
"train_batch_size": 2, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|