| { | |
| "best_global_step": 939, | |
| "best_metric": 0.005780160892754793, | |
| "best_model_checkpoint": "siglip2-finetune-full/checkpoint-939", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 939, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.996999699969997, | |
| "eval_loss": 0.016467241570353508, | |
| "eval_model_preparation_time": 0.0024, | |
| "eval_runtime": 238.0574, | |
| "eval_samples_per_second": 42.002, | |
| "eval_steps_per_second": 5.251, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 1.5974440894568689, | |
| "grad_norm": 0.022631017491221428, | |
| "learning_rate": 4.938132733408324e-07, | |
| "loss": 0.1409, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9974997499749975, | |
| "eval_loss": 0.008936488069593906, | |
| "eval_model_preparation_time": 0.0024, | |
| "eval_runtime": 239.1233, | |
| "eval_samples_per_second": 41.815, | |
| "eval_steps_per_second": 5.227, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.9988998899889989, | |
| "eval_loss": 0.005780160892754793, | |
| "eval_model_preparation_time": 0.0024, | |
| "eval_runtime": 241.1947, | |
| "eval_samples_per_second": 41.456, | |
| "eval_steps_per_second": 5.183, | |
| "step": 939 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 939, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.5124344421540475e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |