Model-J__SupViT__model_idx_0334 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
4659db0 verified
{
"best_metric": 0.9394666666666667,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/google_vit-base-patch16-224/model_idx_0334/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": 0.15233968198299408,
"learning_rate": 8.554359905560886e-05,
"loss": 0.7157,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.9144,
"eval_loss": 0.286166787147522,
"eval_runtime": 6.9525,
"eval_samples_per_second": 539.371,
"eval_steps_per_second": 8.486,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 0.16568335890769958,
"learning_rate": 7.305704108364301e-05,
"loss": 0.1535,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.9165333333333333,
"eval_loss": 0.28322097659111023,
"eval_runtime": 6.585,
"eval_samples_per_second": 569.476,
"eval_steps_per_second": 8.96,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 0.023154545575380325,
"learning_rate": 5.501344202803415e-05,
"loss": 0.0705,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.9304,
"eval_loss": 0.2544439435005188,
"eval_runtime": 6.6384,
"eval_samples_per_second": 564.894,
"eval_steps_per_second": 8.888,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.046178340911865234,
"learning_rate": 3.498655797196586e-05,
"loss": 0.0298,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.9394666666666667,
"eval_loss": 0.24743057787418365,
"eval_runtime": 6.2886,
"eval_samples_per_second": 596.317,
"eval_steps_per_second": 9.382,
"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
}