|
{ |
|
"best_metric": 0.1375, |
|
"best_model_checkpoint": "model/image_classification/run-3/checkpoint-20", |
|
"epoch": 3.0, |
|
"eval_steps": 500, |
|
"global_step": 30, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 0.8163558840751648, |
|
"learning_rate": 1.96917647162269e-06, |
|
"loss": 2.0789, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_accuracy": 0.11875, |
|
"eval_loss": 2.078460693359375, |
|
"eval_runtime": 0.7985, |
|
"eval_samples_per_second": 200.371, |
|
"eval_steps_per_second": 12.523, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 0.6774879693984985, |
|
"learning_rate": 9.84588235811345e-07, |
|
"loss": 2.074, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_accuracy": 0.1375, |
|
"eval_loss": 2.078967332839966, |
|
"eval_runtime": 0.9313, |
|
"eval_samples_per_second": 171.797, |
|
"eval_steps_per_second": 10.737, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"grad_norm": 0.8743719458580017, |
|
"learning_rate": 0.0, |
|
"loss": 2.0732, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_accuracy": 0.09375, |
|
"eval_loss": 2.0778565406799316, |
|
"eval_runtime": 1.6395, |
|
"eval_samples_per_second": 97.588, |
|
"eval_steps_per_second": 6.099, |
|
"step": 30 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 30, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 3, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 1.4879262111694848e+17, |
|
"train_batch_size": 32, |
|
"trial_name": null, |
|
"trial_params": { |
|
"learning_rate": 2.6583882366906316e-06, |
|
"per_device_train_batch_size": 32 |
|
} |
|
} |
|
|