Model-J__MAE__model_idx_0942 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
b91b1d9 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.7994666666666667,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_vit-mae-base/model_idx_0942/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": 2.1098766326904297,
"learning_rate": 0.0003,
"loss": 1.5851,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.7101333333333333,
"eval_loss": 0.9029722809791565,
"eval_runtime": 7.1787,
"eval_samples_per_second": 522.38,
"eval_steps_per_second": 8.219,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": Infinity,
"learning_rate": 0.0003,
"loss": 0.6877,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.7810666666666667,
"eval_loss": 0.7443170547485352,
"eval_runtime": 6.7183,
"eval_samples_per_second": 558.18,
"eval_steps_per_second": 8.782,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 1.323656678199768,
"learning_rate": 0.0003,
"loss": 0.476,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.784,
"eval_loss": 0.7082349061965942,
"eval_runtime": 7.4625,
"eval_samples_per_second": 502.512,
"eval_steps_per_second": 7.906,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 15.230979919433594,
"learning_rate": 0.0003,
"loss": 0.3567,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.7994666666666667,
"eval_loss": 0.674713134765625,
"eval_runtime": 7.1511,
"eval_samples_per_second": 524.392,
"eval_steps_per_second": 8.25,
"step": 1332
}
],
"logging_steps": 500,
"max_steps": 1665,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"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
}