Model-J__DINO__model_idx_0242 / trainer_state.json
Eliahu's picture
Upload folder using huggingface_hub
8d6006b verified
{
"best_metric": 0.9293333333333333,
"best_model_checkpoint": "./vit_finetuned_models_dataset/CIFAR100/50_from_100/facebook_dino-vitb16/model_idx_0242/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": 1.0355011224746704,
"learning_rate": 9.050623058764834e-05,
"loss": 0.8278,
"step": 333
},
{
"epoch": 1.0,
"eval_accuracy": 0.8637333333333334,
"eval_loss": 0.4547056555747986,
"eval_runtime": 12.07,
"eval_samples_per_second": 310.689,
"eval_steps_per_second": 4.888,
"step": 333
},
{
"epoch": 2.0,
"grad_norm": 0.4433557391166687,
"learning_rate": 6.563018865678952e-05,
"loss": 0.2613,
"step": 666
},
{
"epoch": 2.0,
"eval_accuracy": 0.8861333333333333,
"eval_loss": 0.3860296607017517,
"eval_runtime": 11.109,
"eval_samples_per_second": 337.564,
"eval_steps_per_second": 5.311,
"step": 666
},
{
"epoch": 3.0,
"grad_norm": 0.036628685891628265,
"learning_rate": 3.47287092499105e-05,
"loss": 0.0949,
"step": 999
},
{
"epoch": 3.0,
"eval_accuracy": 0.9037333333333334,
"eval_loss": 0.3757302165031433,
"eval_runtime": 11.2611,
"eval_samples_per_second": 333.005,
"eval_steps_per_second": 5.239,
"step": 999
},
{
"epoch": 4.0,
"grad_norm": 0.028482496738433838,
"learning_rate": 9.6603438574838e-06,
"loss": 0.0215,
"step": 1332
},
{
"epoch": 4.0,
"eval_accuracy": 0.9293333333333333,
"eval_loss": 0.2990959584712982,
"eval_runtime": 11.545,
"eval_samples_per_second": 324.816,
"eval_steps_per_second": 5.11,
"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
}