| { | |
| "best_metric": 0.7581333333333333, | |
| "best_model_checkpoint": "./resnet_finetuned_models_dataset/CIFAR100/50_from_100/microsoft_resnet-101/model_idx_0386/checkpoints/checkpoint-999", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 999, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 80.84972381591797, | |
| "learning_rate": 4.267766952966369e-05, | |
| "loss": 3.6592, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.49173333333333336, | |
| "eval_loss": 3.1507740020751953, | |
| "eval_runtime": 5.595, | |
| "eval_samples_per_second": 670.245, | |
| "eval_steps_per_second": 10.545, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 111.78571319580078, | |
| "learning_rate": 2.5e-05, | |
| "loss": 2.0897, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6941333333333334, | |
| "eval_loss": 1.3331729173660278, | |
| "eval_runtime": 7.3606, | |
| "eval_samples_per_second": 509.468, | |
| "eval_steps_per_second": 8.016, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 194.47325134277344, | |
| "learning_rate": 7.3223304703363135e-06, | |
| "loss": 1.2055, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7581333333333333, | |
| "eval_loss": 1.039676547050476, | |
| "eval_runtime": 5.5269, | |
| "eval_samples_per_second": 678.502, | |
| "eval_steps_per_second": 10.675, | |
| "step": 999 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 1332, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "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": 2.4529287718656e+18, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |