Model-J__SupViT__model_idx_0013 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
e6beb7b verified
Invalid JSON: Unexpected token 'N', ..."ad_norm": NaN, "... is not valid JSON
{
"best_metric": 0.936,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0013/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": NaN,
"learning_rate": 0.0001,
"loss": 0.5852,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.9253333333333333,
"eval_loss": 0.25535422563552856,
"eval_runtime": 12.4917,
"eval_samples_per_second": 300.198,
"eval_steps_per_second": 4.723,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 26.048255920410156,
"learning_rate": 0.0001,
"loss": 0.1263,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.9354666666666667,
"eval_loss": 0.2389230728149414,
"eval_runtime": 12.1399,
"eval_samples_per_second": 308.898,
"eval_steps_per_second": 4.86,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 89.7683334350586,
"learning_rate": 0.0001,
"loss": 0.0701,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.9253333333333333,
"eval_loss": 0.27615222334861755,
"eval_runtime": 12.3311,
"eval_samples_per_second": 304.11,
"eval_steps_per_second": 4.785,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.5264580249786377,
"learning_rate": 0.0001,
"loss": 0.0456,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.936,
"eval_loss": 0.2622740566730499,
"eval_runtime": 12.4688,
"eval_samples_per_second": 300.75,
"eval_steps_per_second": 4.732,
"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
}