| { | |
| "best_global_step": 3186, | |
| "best_metric": 0.1701197475194931, | |
| "best_model_checkpoint": "siglip2-finetune-full/checkpoint-3186", | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 3186, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.4708097928436911, | |
| "grad_norm": 10.774321556091309, | |
| "learning_rate": 8.565051020408164e-05, | |
| "loss": 1.047, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.9416195856873822, | |
| "grad_norm": 4.252574920654297, | |
| "learning_rate": 6.970663265306123e-05, | |
| "loss": 0.7732, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6644597457627118, | |
| "eval_loss": 0.7104782462120056, | |
| "eval_model_preparation_time": 0.0038, | |
| "eval_runtime": 419.7148, | |
| "eval_samples_per_second": 80.969, | |
| "eval_steps_per_second": 10.121, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 1.4124293785310735, | |
| "grad_norm": 4.366853713989258, | |
| "learning_rate": 5.3762755102040816e-05, | |
| "loss": 0.6655, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.8832391713747647, | |
| "grad_norm": 7.878406047821045, | |
| "learning_rate": 3.781887755102041e-05, | |
| "loss": 0.5723, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8096162900188324, | |
| "eval_loss": 0.43832501769065857, | |
| "eval_model_preparation_time": 0.0038, | |
| "eval_runtime": 419.9197, | |
| "eval_samples_per_second": 80.93, | |
| "eval_steps_per_second": 10.116, | |
| "step": 2124 | |
| }, | |
| { | |
| "epoch": 2.354048964218456, | |
| "grad_norm": 11.523202896118164, | |
| "learning_rate": 2.1875e-05, | |
| "loss": 0.4148, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.824858757062147, | |
| "grad_norm": 4.949053764343262, | |
| "learning_rate": 5.931122448979593e-06, | |
| "loss": 0.2886, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.931379472693032, | |
| "eval_loss": 0.1701197475194931, | |
| "eval_model_preparation_time": 0.0038, | |
| "eval_runtime": 415.1862, | |
| "eval_samples_per_second": 81.852, | |
| "eval_steps_per_second": 10.232, | |
| "step": 3186 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 3186, | |
| "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": 8.539181928682095e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |