Model-J__MAE__model_idx_0112 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
34328e2 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.8669333333333333,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0112/checkpoints/checkpoint-1332",
"epoch": 4.0,
"eval_steps": 500,
"global_step": 1332,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"grad_norm": Infinity,
"learning_rate": 3e-05,
"loss": 2.2125,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.7397333333333334,
"eval_loss": 1.1247119903564453,
"eval_runtime": 6.011,
"eval_samples_per_second": 623.859,
"eval_steps_per_second": 9.815,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 5.245887279510498,
"learning_rate": 3e-05,
"loss": 0.7472,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8205333333333333,
"eval_loss": 0.6599355340003967,
"eval_runtime": 5.8597,
"eval_samples_per_second": 639.963,
"eval_steps_per_second": 10.069,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 2.676271677017212,
"learning_rate": 3e-05,
"loss": 0.4173,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.8648,
"eval_loss": 0.4947391152381897,
"eval_runtime": 5.9637,
"eval_samples_per_second": 628.806,
"eval_steps_per_second": 9.893,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 1.1752407550811768,
"learning_rate": 3e-05,
"loss": 0.2706,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.8669333333333333,
"eval_loss": 0.4515393376350403,
"eval_runtime": 5.3294,
"eval_samples_per_second": 703.638,
"eval_steps_per_second": 11.071,
"step": 1332
}
],
"logging_steps": 500,
"max_steps": 2331,
"num_input_tokens_seen": 0,
"num_train_epochs": 7,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 6.58965282490368e+18,
"train_batch_size": 64,
"trial_name": null,
"trial_params": null
}