Model-J__DINO__model_idx_0704 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
d6d14cd verified
Invalid JSON: Unexpected token 'I', ..."ad_norm": Infinity, "... is not valid JSON
{
"best_metric": 0.8141333333333334,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0704/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": 28.748106002807617,
"learning_rate": 0.0001,
"loss": 1.0197,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.7781333333333333,
"eval_loss": 0.7233564257621765,
"eval_runtime": 12.4793,
"eval_samples_per_second": 300.497,
"eval_steps_per_second": 4.728,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": Infinity,
"learning_rate": 0.0001,
"loss": 0.47,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8061333333333334,
"eval_loss": 0.6672335267066956,
"eval_runtime": 13.1161,
"eval_samples_per_second": 285.907,
"eval_steps_per_second": 4.498,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 9.647558212280273,
"learning_rate": 0.0001,
"loss": 0.3155,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.7909333333333334,
"eval_loss": 0.731598973274231,
"eval_runtime": 12.1324,
"eval_samples_per_second": 309.09,
"eval_steps_per_second": 4.863,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 41.799102783203125,
"learning_rate": 0.0001,
"loss": 0.2469,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.8141333333333334,
"eval_loss": 0.68025141954422,
"eval_runtime": 12.0191,
"eval_samples_per_second": 312.003,
"eval_steps_per_second": 4.909,
"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
}