Model-J__MAE__model_idx_0671 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
3becaf3 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.8752,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0671/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": 2.2447,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.7653333333333333,
"eval_loss": 1.070206880569458,
"eval_runtime": 3.9861,
"eval_samples_per_second": 940.767,
"eval_steps_per_second": 14.801,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 1.9856637716293335,
"learning_rate": 3e-05,
"loss": 0.7057,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8512,
"eval_loss": 0.574259877204895,
"eval_runtime": 4.1703,
"eval_samples_per_second": 899.22,
"eval_steps_per_second": 14.148,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 3.1409900188446045,
"learning_rate": 3e-05,
"loss": 0.3497,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.8666666666666667,
"eval_loss": 0.47451186180114746,
"eval_runtime": 3.9425,
"eval_samples_per_second": 951.166,
"eval_steps_per_second": 14.965,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": Infinity,
"learning_rate": 3e-05,
"loss": 0.1953,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.8752,
"eval_loss": 0.4481125473976135,
"eval_runtime": 4.165,
"eval_samples_per_second": 900.364,
"eval_steps_per_second": 14.166,
"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
}