| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.383378016085791, | |
| "eval_steps": 500, | |
| "global_step": 3500, | |
| "is_hyper_param_search": true, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.758632242679596, | |
| "eval_loss": 0.4715859889984131, | |
| "eval_runtime": 9.5006, | |
| "eval_samples_per_second": 313.981, | |
| "eval_steps_per_second": 19.683, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 1.3404825737265416, | |
| "grad_norm": 9.62618637084961, | |
| "learning_rate": 1.4776902224208135e-05, | |
| "loss": 0.5101, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.7727120518684387, | |
| "eval_loss": 0.47308608889579773, | |
| "eval_runtime": 9.4715, | |
| "eval_samples_per_second": 314.944, | |
| "eval_steps_per_second": 19.743, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 2.680965147453083, | |
| "grad_norm": 17.83710289001465, | |
| "learning_rate": 1.2489456059470035e-05, | |
| "loss": 0.2485, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.7743881940841675, | |
| "eval_loss": 0.7006397843360901, | |
| "eval_runtime": 9.4464, | |
| "eval_samples_per_second": 315.781, | |
| "eval_steps_per_second": 19.796, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7750586867332458, | |
| "eval_loss": 0.8257986903190613, | |
| "eval_runtime": 9.4942, | |
| "eval_samples_per_second": 314.193, | |
| "eval_steps_per_second": 19.696, | |
| "step": 1492 | |
| }, | |
| { | |
| "epoch": 4.021447721179625, | |
| "grad_norm": 0.4933696985244751, | |
| "learning_rate": 1.0202009894731932e-05, | |
| "loss": 0.0969, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.782433807849884, | |
| "eval_loss": 0.9035575985908508, | |
| "eval_runtime": 9.4688, | |
| "eval_samples_per_second": 315.034, | |
| "eval_steps_per_second": 19.749, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 5.361930294906166, | |
| "grad_norm": 15.212503433227539, | |
| "learning_rate": 7.91456372999383e-06, | |
| "loss": 0.0396, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7844451665878296, | |
| "eval_loss": 1.023715853691101, | |
| "eval_runtime": 9.6126, | |
| "eval_samples_per_second": 310.323, | |
| "eval_steps_per_second": 19.454, | |
| "step": 2238 | |
| }, | |
| { | |
| "epoch": 6.702412868632708, | |
| "grad_norm": 1.1762652397155762, | |
| "learning_rate": 5.627117565255729e-06, | |
| "loss": 0.0231, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.7760643362998962, | |
| "eval_loss": 1.3074647188186646, | |
| "eval_runtime": 9.435, | |
| "eval_samples_per_second": 316.164, | |
| "eval_steps_per_second": 19.82, | |
| "step": 2611 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.783774733543396, | |
| "eval_loss": 1.3387647867202759, | |
| "eval_runtime": 9.4842, | |
| "eval_samples_per_second": 314.525, | |
| "eval_steps_per_second": 19.717, | |
| "step": 2984 | |
| }, | |
| { | |
| "epoch": 8.04289544235925, | |
| "grad_norm": 3.214566707611084, | |
| "learning_rate": 3.3396714005176278e-06, | |
| "loss": 0.0136, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.7814280986785889, | |
| "eval_loss": 1.466342806816101, | |
| "eval_runtime": 9.5941, | |
| "eval_samples_per_second": 310.922, | |
| "eval_steps_per_second": 19.491, | |
| "step": 3357 | |
| }, | |
| { | |
| "epoch": 9.383378016085791, | |
| "grad_norm": 0.25926658511161804, | |
| "learning_rate": 1.0522252357795266e-06, | |
| "loss": 0.0084, | |
| "step": 3500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 3730, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 2.00380947674166e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": { | |
| "learning_rate": 1.7064348388946237e-05, | |
| "per_device_train_batch_size": 32 | |
| } | |
| } | |