| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 4371, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.77121940059483e-05, | |
| "loss": 4.4507, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5424388011896593e-05, | |
| "loss": 3.4418, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.5406472086906433, | |
| "eval_loss": 2.3865766525268555, | |
| "eval_runtime": 2.3607, | |
| "eval_samples_per_second": 536.704, | |
| "eval_steps_per_second": 67.353, | |
| "step": 1457 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.3136582017844888e-05, | |
| "loss": 2.7157, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.0848776023793184e-05, | |
| "loss": 2.0734, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.56097002974148e-06, | |
| "loss": 1.7742, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6069455146789551, | |
| "eval_loss": 1.9364700317382812, | |
| "eval_runtime": 2.4048, | |
| "eval_samples_per_second": 526.856, | |
| "eval_steps_per_second": 66.117, | |
| "step": 2914 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.273164035689774e-06, | |
| "loss": 1.518, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.98535804163807e-06, | |
| "loss": 1.2078, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.697552047586365e-06, | |
| "loss": 1.1313, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6322020292282104, | |
| "eval_loss": 1.8354707956314087, | |
| "eval_runtime": 2.3743, | |
| "eval_samples_per_second": 533.633, | |
| "eval_steps_per_second": 66.967, | |
| "step": 4371 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 4371, | |
| "total_flos": 1151412634864512.0, | |
| "train_loss": 2.186672953385506, | |
| "train_runtime": 418.7818, | |
| "train_samples_per_second": 83.464, | |
| "train_steps_per_second": 10.437 | |
| } | |
| ], | |
| "max_steps": 4371, | |
| "num_train_epochs": 3, | |
| "total_flos": 1151412634864512.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |