Model-J__DINO__model_idx_0304 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
1a1075c verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.8994666666666666,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0304/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": 3e-05,
"loss": 0.8335,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.872,
"eval_loss": 0.4301021695137024,
"eval_runtime": 7.0214,
"eval_samples_per_second": 534.081,
"eval_steps_per_second": 8.403,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 3.7990753650665283,
"learning_rate": 3e-05,
"loss": 0.2235,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8882666666666666,
"eval_loss": 0.39018714427948,
"eval_runtime": 7.1024,
"eval_samples_per_second": 527.99,
"eval_steps_per_second": 8.307,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 2.072925090789795,
"learning_rate": 3e-05,
"loss": 0.1117,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.8933333333333333,
"eval_loss": 0.3855521082878113,
"eval_runtime": 6.397,
"eval_samples_per_second": 586.215,
"eval_steps_per_second": 9.223,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.9043755531311035,
"learning_rate": 3e-05,
"loss": 0.0753,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.8994666666666666,
"eval_loss": 0.3963587284088135,
"eval_runtime": 6.338,
"eval_samples_per_second": 591.667,
"eval_steps_per_second": 9.309,
"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
}