Model-J__MAE__model_idx_0415 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
0a5570c verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.86,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0415/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": 2.0487678050994873,
"learning_rate": 9e-05,
"loss": 1.5306,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.7733333333333333,
"eval_loss": 0.8255800008773804,
"eval_runtime": 12.8781,
"eval_samples_per_second": 291.192,
"eval_steps_per_second": 4.581,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": Infinity,
"learning_rate": 9e-05,
"loss": 0.5073,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8402666666666667,
"eval_loss": 0.5525099039077759,
"eval_runtime": 12.2933,
"eval_samples_per_second": 305.045,
"eval_steps_per_second": 4.799,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 6.9540863037109375,
"learning_rate": 9e-05,
"loss": 0.3073,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.8306666666666667,
"eval_loss": 0.5574058890342712,
"eval_runtime": 12.0535,
"eval_samples_per_second": 311.112,
"eval_steps_per_second": 4.895,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 1.8724627494812012,
"learning_rate": 9e-05,
"loss": 0.2109,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.86,
"eval_loss": 0.4916815757751465,
"eval_runtime": 12.1908,
"eval_samples_per_second": 307.61,
"eval_steps_per_second": 4.84,
"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
}