Model-J__SupViT__model_idx_0438 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
20959d7 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.9328,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0438/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": 5.543400764465332,
"learning_rate": 2.4e-05,
"loss": 1.3075,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.9064,
"eval_loss": 0.3838406205177307,
"eval_runtime": 10.9979,
"eval_samples_per_second": 340.975,
"eval_steps_per_second": 5.365,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": Infinity,
"learning_rate": 1.801801801801802e-05,
"loss": 0.2327,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.9232,
"eval_loss": 0.2752654552459717,
"eval_runtime": 11.0089,
"eval_samples_per_second": 340.633,
"eval_steps_per_second": 5.359,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 0.21334803104400635,
"learning_rate": 1.2018018018018019e-05,
"loss": 0.1112,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.932,
"eval_loss": 0.2436397522687912,
"eval_runtime": 10.7188,
"eval_samples_per_second": 349.852,
"eval_steps_per_second": 5.504,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.34483709931373596,
"learning_rate": 6.018018018018018e-06,
"loss": 0.0622,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.9328,
"eval_loss": 0.23488226532936096,
"eval_runtime": 10.7194,
"eval_samples_per_second": 349.832,
"eval_steps_per_second": 5.504,
"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
}