|
{ |
|
"best_metric": 0.030113982036709785, |
|
"best_model_checkpoint": "swin-base-patch4-window7-224-in22k-Kontur-competition-1.3K/checkpoint-18", |
|
"epoch": 4.8, |
|
"eval_steps": 500, |
|
"global_step": 30, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.96, |
|
"eval_loss": 0.3283247947692871, |
|
"eval_runtime": 3.7851, |
|
"eval_samples_per_second": 34.874, |
|
"eval_steps_per_second": 0.793, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"grad_norm": 3.094001293182373, |
|
"learning_rate": 3.7037037037037037e-05, |
|
"loss": 0.4531, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"eval_loss": 0.08094710856676102, |
|
"eval_runtime": 2.5203, |
|
"eval_samples_per_second": 52.376, |
|
"eval_steps_per_second": 1.19, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 2.88, |
|
"eval_loss": 0.030113982036709785, |
|
"eval_runtime": 2.6775, |
|
"eval_samples_per_second": 49.299, |
|
"eval_steps_per_second": 1.12, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 3.2, |
|
"grad_norm": 5.238066673278809, |
|
"learning_rate": 1.8518518518518518e-05, |
|
"loss": 0.069, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"eval_loss": 0.12024853378534317, |
|
"eval_runtime": 2.6542, |
|
"eval_samples_per_second": 49.733, |
|
"eval_steps_per_second": 1.13, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 4.8, |
|
"grad_norm": 3.4171395301818848, |
|
"learning_rate": 0.0, |
|
"loss": 0.0364, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 4.8, |
|
"eval_loss": 0.0865083634853363, |
|
"eval_runtime": 2.5761, |
|
"eval_samples_per_second": 51.24, |
|
"eval_steps_per_second": 1.165, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 4.8, |
|
"step": 30, |
|
"total_flos": 4.4500275596722176e+17, |
|
"train_loss": 0.1861822952826818, |
|
"train_runtime": 258.3418, |
|
"train_samples_per_second": 22.838, |
|
"train_steps_per_second": 0.116 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 30, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 5, |
|
"save_steps": 500, |
|
"total_flos": 4.4500275596722176e+17, |
|
"train_batch_size": 48, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|