| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5266468686454935, | |
| "global_step": 4500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00015, | |
| "loss": 2.5393, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "eval_loss": 0.4428945779800415, | |
| "eval_phone_accuracy": 0.9003876271495096, | |
| "eval_runtime": 16.8778, | |
| "eval_samples_per_second": 11.85, | |
| "eval_steps_per_second": 2.962, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0003, | |
| "loss": 0.4474, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "eval_loss": 0.2071739137172699, | |
| "eval_phone_accuracy": 0.9347190787134594, | |
| "eval_runtime": 15.9974, | |
| "eval_samples_per_second": 12.502, | |
| "eval_steps_per_second": 3.126, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00028011664899257687, | |
| "loss": 0.3105, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "eval_loss": 0.174857959151268, | |
| "eval_phone_accuracy": 0.94312200282699, | |
| "eval_runtime": 16.5779, | |
| "eval_samples_per_second": 12.064, | |
| "eval_steps_per_second": 3.016, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0002602332979851537, | |
| "loss": 0.2682, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "eval_loss": 0.16749557852745056, | |
| "eval_phone_accuracy": 0.9447956502510471, | |
| "eval_runtime": 15.1084, | |
| "eval_samples_per_second": 13.238, | |
| "eval_steps_per_second": 3.309, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0002403499469777306, | |
| "loss": 0.2467, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "eval_loss": 0.15832921862602234, | |
| "eval_phone_accuracy": 0.9478567774048926, | |
| "eval_runtime": 14.9641, | |
| "eval_samples_per_second": 13.365, | |
| "eval_steps_per_second": 3.341, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00022046659597030753, | |
| "loss": 0.2324, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "eval_loss": 0.14310674369335175, | |
| "eval_phone_accuracy": 0.949885966509708, | |
| "eval_runtime": 15.0339, | |
| "eval_samples_per_second": 13.303, | |
| "eval_steps_per_second": 3.326, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0002005832449628844, | |
| "loss": 0.22, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "eval_loss": 0.13784204423427582, | |
| "eval_phone_accuracy": 0.9521579645672364, | |
| "eval_runtime": 16.1699, | |
| "eval_samples_per_second": 12.369, | |
| "eval_steps_per_second": 3.092, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0001806998939554613, | |
| "loss": 0.2109, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "eval_loss": 0.1341935396194458, | |
| "eval_phone_accuracy": 0.9543519168899642, | |
| "eval_runtime": 15.6229, | |
| "eval_samples_per_second": 12.802, | |
| "eval_steps_per_second": 3.2, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00016081654294803816, | |
| "loss": 0.2039, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_loss": 0.13459673523902893, | |
| "eval_phone_accuracy": 0.9526435824726623, | |
| "eval_runtime": 15.5354, | |
| "eval_samples_per_second": 12.874, | |
| "eval_steps_per_second": 3.218, | |
| "step": 4500 | |
| } | |
| ], | |
| "max_steps": 8544, | |
| "num_train_epochs": 1, | |
| "total_flos": 7.300030651149006e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |