| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "global_step": 675, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.07, |
| "learning_rate": 1.9703703703703704e-05, |
| "loss": 1.1019, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 1.9407407407407407e-05, |
| "loss": 1.0854, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.22, |
| "learning_rate": 1.9111111111111113e-05, |
| "loss": 1.0717, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.3, |
| "learning_rate": 1.8814814814814816e-05, |
| "loss": 0.9272, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.37, |
| "learning_rate": 1.851851851851852e-05, |
| "loss": 0.7433, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 1.8222222222222224e-05, |
| "loss": 0.6434, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.52, |
| "learning_rate": 1.7925925925925927e-05, |
| "loss": 0.5779, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.59, |
| "learning_rate": 1.7629629629629633e-05, |
| "loss": 0.5034, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.67, |
| "learning_rate": 1.7333333333333336e-05, |
| "loss": 0.526, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.74, |
| "learning_rate": 1.7037037037037038e-05, |
| "loss": 0.376, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.81, |
| "learning_rate": 1.674074074074074e-05, |
| "loss": 0.4096, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.89, |
| "learning_rate": 1.6444444444444444e-05, |
| "loss": 0.3543, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.96, |
| "learning_rate": 1.614814814814815e-05, |
| "loss": 0.4828, |
| "step": 130 |
| }, |
| { |
| "epoch": 1.04, |
| "learning_rate": 1.5851851851851852e-05, |
| "loss": 0.3712, |
| "step": 140 |
| }, |
| { |
| "epoch": 1.11, |
| "learning_rate": 1.555555555555556e-05, |
| "loss": 0.2876, |
| "step": 150 |
| }, |
| { |
| "epoch": 1.19, |
| "learning_rate": 1.525925925925926e-05, |
| "loss": 0.292, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 1.4962962962962964e-05, |
| "loss": 0.3807, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.33, |
| "learning_rate": 1.4666666666666666e-05, |
| "loss": 0.3449, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 1.4370370370370372e-05, |
| "loss": 0.2524, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.48, |
| "learning_rate": 1.4074074074074075e-05, |
| "loss": 0.214, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 1.377777777777778e-05, |
| "loss": 0.3225, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.63, |
| "learning_rate": 1.3481481481481482e-05, |
| "loss": 0.2482, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 1.3185185185185185e-05, |
| "loss": 0.2422, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.78, |
| "learning_rate": 1.288888888888889e-05, |
| "loss": 0.3402, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.2592592592592593e-05, |
| "loss": 0.3747, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.93, |
| "learning_rate": 1.2296296296296298e-05, |
| "loss": 0.3456, |
| "step": 260 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 1.2e-05, |
| "loss": 0.3144, |
| "step": 270 |
| }, |
| { |
| "epoch": 2.07, |
| "learning_rate": 1.1703703703703703e-05, |
| "loss": 0.2441, |
| "step": 280 |
| }, |
| { |
| "epoch": 2.15, |
| "learning_rate": 1.1407407407407409e-05, |
| "loss": 0.2079, |
| "step": 290 |
| }, |
| { |
| "epoch": 2.22, |
| "learning_rate": 1.1111111111111113e-05, |
| "loss": 0.2248, |
| "step": 300 |
| }, |
| { |
| "epoch": 2.3, |
| "learning_rate": 1.0814814814814816e-05, |
| "loss": 0.2352, |
| "step": 310 |
| }, |
| { |
| "epoch": 2.37, |
| "learning_rate": 1.0518518518518519e-05, |
| "loss": 0.1649, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.44, |
| "learning_rate": 1.0222222222222223e-05, |
| "loss": 0.2001, |
| "step": 330 |
| }, |
| { |
| "epoch": 2.52, |
| "learning_rate": 9.925925925925927e-06, |
| "loss": 0.2592, |
| "step": 340 |
| }, |
| { |
| "epoch": 2.59, |
| "learning_rate": 9.62962962962963e-06, |
| "loss": 0.1567, |
| "step": 350 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 9.333333333333334e-06, |
| "loss": 0.1798, |
| "step": 360 |
| }, |
| { |
| "epoch": 2.74, |
| "learning_rate": 9.037037037037037e-06, |
| "loss": 0.2292, |
| "step": 370 |
| }, |
| { |
| "epoch": 2.81, |
| "learning_rate": 8.740740740740741e-06, |
| "loss": 0.2495, |
| "step": 380 |
| }, |
| { |
| "epoch": 2.89, |
| "learning_rate": 8.444444444444446e-06, |
| "loss": 0.167, |
| "step": 390 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 8.148148148148148e-06, |
| "loss": 0.1521, |
| "step": 400 |
| }, |
| { |
| "epoch": 3.04, |
| "learning_rate": 7.851851851851853e-06, |
| "loss": 0.1815, |
| "step": 410 |
| }, |
| { |
| "epoch": 3.11, |
| "learning_rate": 7.555555555555556e-06, |
| "loss": 0.1172, |
| "step": 420 |
| }, |
| { |
| "epoch": 3.19, |
| "learning_rate": 7.2592592592592605e-06, |
| "loss": 0.1175, |
| "step": 430 |
| }, |
| { |
| "epoch": 3.26, |
| "learning_rate": 6.962962962962964e-06, |
| "loss": 0.1655, |
| "step": 440 |
| }, |
| { |
| "epoch": 3.33, |
| "learning_rate": 6.666666666666667e-06, |
| "loss": 0.1596, |
| "step": 450 |
| }, |
| { |
| "epoch": 3.41, |
| "learning_rate": 6.370370370370371e-06, |
| "loss": 0.144, |
| "step": 460 |
| }, |
| { |
| "epoch": 3.48, |
| "learning_rate": 6.0740740740740745e-06, |
| "loss": 0.1975, |
| "step": 470 |
| }, |
| { |
| "epoch": 3.56, |
| "learning_rate": 5.777777777777778e-06, |
| "loss": 0.1337, |
| "step": 480 |
| }, |
| { |
| "epoch": 3.63, |
| "learning_rate": 5.481481481481482e-06, |
| "loss": 0.1635, |
| "step": 490 |
| }, |
| { |
| "epoch": 3.7, |
| "learning_rate": 5.185185185185185e-06, |
| "loss": 0.1982, |
| "step": 500 |
| }, |
| { |
| "epoch": 3.78, |
| "learning_rate": 4.888888888888889e-06, |
| "loss": 0.1802, |
| "step": 510 |
| }, |
| { |
| "epoch": 3.85, |
| "learning_rate": 4.592592592592593e-06, |
| "loss": 0.1706, |
| "step": 520 |
| }, |
| { |
| "epoch": 3.93, |
| "learning_rate": 4.296296296296296e-06, |
| "loss": 0.1807, |
| "step": 530 |
| }, |
| { |
| "epoch": 4.0, |
| "learning_rate": 4.000000000000001e-06, |
| "loss": 0.1827, |
| "step": 540 |
| }, |
| { |
| "epoch": 4.07, |
| "learning_rate": 3.7037037037037037e-06, |
| "loss": 0.121, |
| "step": 550 |
| }, |
| { |
| "epoch": 4.15, |
| "learning_rate": 3.4074074074074077e-06, |
| "loss": 0.0976, |
| "step": 560 |
| }, |
| { |
| "epoch": 4.22, |
| "learning_rate": 3.1111111111111116e-06, |
| "loss": 0.1254, |
| "step": 570 |
| }, |
| { |
| "epoch": 4.3, |
| "learning_rate": 2.814814814814815e-06, |
| "loss": 0.0885, |
| "step": 580 |
| }, |
| { |
| "epoch": 4.37, |
| "learning_rate": 2.5185185185185186e-06, |
| "loss": 0.0957, |
| "step": 590 |
| }, |
| { |
| "epoch": 4.44, |
| "learning_rate": 2.222222222222222e-06, |
| "loss": 0.0798, |
| "step": 600 |
| }, |
| { |
| "epoch": 4.52, |
| "learning_rate": 1.925925925925926e-06, |
| "loss": 0.2056, |
| "step": 610 |
| }, |
| { |
| "epoch": 4.59, |
| "learning_rate": 1.62962962962963e-06, |
| "loss": 0.1089, |
| "step": 620 |
| }, |
| { |
| "epoch": 4.67, |
| "learning_rate": 1.3333333333333334e-06, |
| "loss": 0.0792, |
| "step": 630 |
| }, |
| { |
| "epoch": 4.74, |
| "learning_rate": 1.0370370370370371e-06, |
| "loss": 0.1138, |
| "step": 640 |
| }, |
| { |
| "epoch": 4.81, |
| "learning_rate": 7.407407407407407e-07, |
| "loss": 0.1337, |
| "step": 650 |
| }, |
| { |
| "epoch": 4.89, |
| "learning_rate": 4.444444444444445e-07, |
| "loss": 0.1142, |
| "step": 660 |
| }, |
| { |
| "epoch": 4.96, |
| "learning_rate": 1.4814814814814817e-07, |
| "loss": 0.1845, |
| "step": 670 |
| }, |
| { |
| "epoch": 5.0, |
| "step": 675, |
| "total_flos": 1420483563959040.0, |
| "train_loss": 0.29182871209250555, |
| "train_runtime": 187.2401, |
| "train_samples_per_second": 115.333, |
| "train_steps_per_second": 3.605 |
| } |
| ], |
| "max_steps": 675, |
| "num_train_epochs": 5, |
| "total_flos": 1420483563959040.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
