| { | |
| "best_metric": 10.36404160475483, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 58.8235294117647, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.5136083400296205e-06, | |
| "loss": 0.0024, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 1.8687587131475301e-06, | |
| "loss": 0.0024, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 2.0711488350670174e-06, | |
| "loss": 0.0023, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 2.213317753617305e-06, | |
| "loss": 0.0023, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 2.3230029693718747e-06, | |
| "loss": 0.002, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 2.412322158351148e-06, | |
| "loss": 0.002, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 10.29, | |
| "learning_rate": 2.4876668872198717e-06, | |
| "loss": 0.0022, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 11.76, | |
| "learning_rate": 2.552824062407326e-06, | |
| "loss": 0.0021, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 13.24, | |
| "learning_rate": 2.610223373296667e-06, | |
| "loss": 0.0034, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 14.71, | |
| "learning_rate": 2.661517182828361e-06, | |
| "loss": 0.0019, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 16.18, | |
| "learning_rate": 2.7078803874740543e-06, | |
| "loss": 0.0018, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 17.65, | |
| "learning_rate": 2.750178319990197e-06, | |
| "loss": 0.0023, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 19.12, | |
| "learning_rate": 2.7890667754365044e-06, | |
| "loss": 0.0019, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 20.59, | |
| "learning_rate": 2.8250546392106077e-06, | |
| "loss": 0.0021, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 22.06, | |
| "learning_rate": 2.8585447348549113e-06, | |
| "loss": 0.0023, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 23.53, | |
| "learning_rate": 2.889861392935294e-06, | |
| "loss": 0.0021, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 2.9192696063561725e-06, | |
| "loss": 0.0016, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 26.47, | |
| "learning_rate": 2.946988676871634e-06, | |
| "loss": 0.0018, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 27.94, | |
| "learning_rate": 2.973202150939645e-06, | |
| "loss": 0.0022, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 29.41, | |
| "learning_rate": 2.998065193492142e-06, | |
| "loss": 0.0018, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 30.88, | |
| "learning_rate": 2.9559999999999997e-06, | |
| "loss": 0.0018, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 32.35, | |
| "learning_rate": 2.9060000000000002e-06, | |
| "loss": 0.0021, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 33.82, | |
| "learning_rate": 2.856e-06, | |
| "loss": 0.0017, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 35.29, | |
| "learning_rate": 2.8060000000000003e-06, | |
| "loss": 0.0018, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 36.76, | |
| "learning_rate": 2.756e-06, | |
| "loss": 0.0016, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 38.24, | |
| "learning_rate": 2.706e-06, | |
| "loss": 0.0017, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 39.71, | |
| "learning_rate": 2.656e-06, | |
| "loss": 0.0016, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 41.18, | |
| "learning_rate": 2.606e-06, | |
| "loss": 0.0013, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 42.65, | |
| "learning_rate": 2.556e-06, | |
| "loss": 0.0013, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 44.12, | |
| "learning_rate": 2.5060000000000002e-06, | |
| "loss": 0.0012, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 45.59, | |
| "learning_rate": 2.456e-06, | |
| "loss": 0.0013, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 47.06, | |
| "learning_rate": 2.4060000000000003e-06, | |
| "loss": 0.0013, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 48.53, | |
| "learning_rate": 2.356e-06, | |
| "loss": 0.0013, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 2.306e-06, | |
| "loss": 0.0011, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 51.47, | |
| "learning_rate": 2.256e-06, | |
| "loss": 0.0013, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 52.94, | |
| "learning_rate": 2.2059999999999997e-06, | |
| "loss": 0.0014, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 54.41, | |
| "learning_rate": 2.156e-06, | |
| "loss": 0.0011, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 55.88, | |
| "learning_rate": 2.106e-06, | |
| "loss": 0.0024, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 57.35, | |
| "learning_rate": 2.0560000000000003e-06, | |
| "loss": 0.0014, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 58.82, | |
| "learning_rate": 2.006e-06, | |
| "loss": 0.0014, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 58.82, | |
| "eval_loss": 0.494384765625, | |
| "eval_runtime": 154.3699, | |
| "eval_samples_per_second": 1.762, | |
| "eval_steps_per_second": 0.11, | |
| "eval_wer": 10.36404160475483, | |
| "step": 1000 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 118, | |
| "total_flos": 3.0824308756665336e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |