Model-J__DINO__model_idx_0979 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
2b5cdf4 verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.3208,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0979/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": Infinity,
"learning_rate": 0.0005,
"loss": 3.3224,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.22533333333333333,
"eval_loss": 2.9476027488708496,
"eval_runtime": 11.3964,
"eval_samples_per_second": 329.051,
"eval_steps_per_second": 5.177,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 21.716508865356445,
"learning_rate": 0.0005,
"loss": 2.8166,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.2613333333333333,
"eval_loss": 2.736382007598877,
"eval_runtime": 11.8726,
"eval_samples_per_second": 315.853,
"eval_steps_per_second": 4.969,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 20.99009895324707,
"learning_rate": 0.0005,
"loss": 2.6495,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.27386666666666665,
"eval_loss": 2.6893398761749268,
"eval_runtime": 11.9055,
"eval_samples_per_second": 314.979,
"eval_steps_per_second": 4.956,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 14.981003761291504,
"learning_rate": 0.0005,
"loss": 2.532,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.3208,
"eval_loss": 2.475480794906616,
"eval_runtime": 11.3575,
"eval_samples_per_second": 330.177,
"eval_steps_per_second": 5.195,
"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
}