Model-J__SupViT__model_idx_0851 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
97f8487 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.9216,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0851/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": 2.5176913738250732,
"learning_rate": 9e-05,
"loss": 0.6424,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.8984,
"eval_loss": 0.32811856269836426,
"eval_runtime": 12.5047,
"eval_samples_per_second": 299.887,
"eval_steps_per_second": 4.718,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": Infinity,
"learning_rate": 9e-05,
"loss": 0.1656,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.9085333333333333,
"eval_loss": 0.2993265688419342,
"eval_runtime": 11.3174,
"eval_samples_per_second": 331.348,
"eval_steps_per_second": 5.213,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 0.025355486199259758,
"learning_rate": 9e-05,
"loss": 0.0865,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.9141333333333334,
"eval_loss": 0.3132348656654358,
"eval_runtime": 12.1582,
"eval_samples_per_second": 308.433,
"eval_steps_per_second": 4.853,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 20.2607479095459,
"learning_rate": 9e-05,
"loss": 0.0537,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.9216,
"eval_loss": 0.30869144201278687,
"eval_runtime": 12.2266,
"eval_samples_per_second": 306.708,
"eval_steps_per_second": 4.826,
"step": 1332
}
],
"logging_steps": 500,
"max_steps": 2664,
"num_input_tokens_seen": 0,
"num_train_epochs": 8,
"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
}