| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.98906439854192, |
| "global_step": 615, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.05, |
| "learning_rate": 4.996738892723075e-05, |
| "loss": 3.1898, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.1, |
| "learning_rate": 4.986964078748837e-05, |
| "loss": 3.0387, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.15, |
| "learning_rate": 4.970701059450872e-05, |
| "loss": 3.1413, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.19, |
| "learning_rate": 4.9479922631896405e-05, |
| "loss": 3.0543, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 4.918896934621734e-05, |
| "loss": 3.0544, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.29, |
| "learning_rate": 4.8834909801373264e-05, |
| "loss": 2.9954, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 4.8418667698290696e-05, |
| "loss": 3.0829, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.39, |
| "learning_rate": 4.79413289650907e-05, |
| "loss": 3.0368, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.44, |
| "learning_rate": 4.740413892402639e-05, |
| "loss": 2.941, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.49, |
| "learning_rate": 4.680849904257938e-05, |
| "loss": 2.9253, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.53, |
| "learning_rate": 4.615596327719111e-05, |
| "loss": 2.9969, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.58, |
| "learning_rate": 4.5448234019167945e-05, |
| "loss": 2.9547, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.468715765333664e-05, |
| "loss": 3.021, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.387471974103713e-05, |
| "loss": 2.96, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 4.301303984001967e-05, |
| "loss": 3.0919, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.78, |
| "learning_rate": 4.210436597476076e-05, |
| "loss": 2.9032, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.83, |
| "learning_rate": 4.1151068771623866e-05, |
| "loss": 2.9535, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 4.015563527416595e-05, |
| "loss": 2.9935, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.92, |
| "learning_rate": 3.9120662454724836e-05, |
| "loss": 3.0283, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.97, |
| "learning_rate": 3.8048850439214844e-05, |
| "loss": 2.9129, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.02, |
| "learning_rate": 3.694299546280657e-05, |
| "loss": 2.9757, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.07, |
| "learning_rate": 3.580598257486867e-05, |
| "loss": 2.9132, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.12, |
| "learning_rate": 3.46407781122034e-05, |
| "loss": 2.8876, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.17, |
| "learning_rate": 3.3450421960212566e-05, |
| "loss": 2.9416, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.22, |
| "learning_rate": 3.223801962218372e-05, |
| "loss": 2.994, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.26, |
| "learning_rate": 3.100673411738652e-05, |
| "loss": 2.9024, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.31, |
| "learning_rate": 2.975977772911671e-05, |
| "loss": 2.9789, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.36, |
| "learning_rate": 2.8500403624215734e-05, |
| "loss": 2.9478, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.41, |
| "learning_rate": 2.723189736592986e-05, |
| "loss": 2.8663, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 2.595756834225089e-05, |
| "loss": 3.0051, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.51, |
| "learning_rate": 2.468074113210066e-05, |
| "loss": 2.8488, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.56, |
| "learning_rate": 2.340474683188429e-05, |
| "loss": 2.9825, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.6, |
| "learning_rate": 2.2132914365039993e-05, |
| "loss": 3.0427, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.65, |
| "learning_rate": 2.0868561797257878e-05, |
| "loss": 3.0026, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.7, |
| "learning_rate": 1.961498768002547e-05, |
| "loss": 2.984, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.75, |
| "learning_rate": 1.8375462445083464e-05, |
| "loss": 2.9845, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.8, |
| "learning_rate": 1.7153219872242727e-05, |
| "loss": 2.9139, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.85, |
| "learning_rate": 1.5951448652822047e-05, |
| "loss": 3.0234, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.9, |
| "learning_rate": 1.4773284070716503e-05, |
| "loss": 2.9504, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.94, |
| "learning_rate": 1.3621799822799788e-05, |
| "loss": 2.9035, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.99, |
| "learning_rate": 1.2500000000000006e-05, |
| "loss": 2.9339, |
| "step": 410 |
| }, |
| { |
| "epoch": 2.04, |
| "learning_rate": 1.1410811249969475e-05, |
| "loss": 2.9613, |
| "step": 420 |
| }, |
| { |
| "epoch": 2.09, |
| "learning_rate": 1.035707514179513e-05, |
| "loss": 2.9762, |
| "step": 430 |
| }, |
| { |
| "epoch": 2.14, |
| "learning_rate": 9.341540752669235e-06, |
| "loss": 2.8895, |
| "step": 440 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 8.36685749586087e-06, |
| "loss": 2.9219, |
| "step": 450 |
| }, |
| { |
| "epoch": 2.24, |
| "learning_rate": 7.435568208699203e-06, |
| "loss": 2.8863, |
| "step": 460 |
| }, |
| { |
| "epoch": 2.28, |
| "learning_rate": 6.55010251860127e-06, |
| "loss": 2.8696, |
| "step": 470 |
| }, |
| { |
| "epoch": 2.33, |
| "learning_rate": 5.712770504451426e-06, |
| "loss": 2.8808, |
| "step": 480 |
| }, |
| { |
| "epoch": 2.38, |
| "learning_rate": 4.925756669869314e-06, |
| "loss": 2.9685, |
| "step": 490 |
| }, |
| { |
| "epoch": 2.43, |
| "learning_rate": 4.19111424408932e-06, |
| "loss": 2.8873, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.48, |
| "learning_rate": 3.5107598253199758e-06, |
| "loss": 2.9854, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.53, |
| "learning_rate": 2.8864683805580133e-06, |
| "loss": 2.9411, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.58, |
| "learning_rate": 2.3198686149022013e-06, |
| "loss": 2.9353, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.62, |
| "learning_rate": 1.8124387224476347e-06, |
| "loss": 2.9348, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.67, |
| "learning_rate": 1.365502529846166e-06, |
| "loss": 2.8754, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.72, |
| "learning_rate": 9.802260425938099e-07, |
| "loss": 2.925, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.77, |
| "learning_rate": 6.576144030555259e-07, |
| "loss": 2.976, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.82, |
| "learning_rate": 3.9850926816357157e-07, |
| "loss": 2.913, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.87, |
| "learning_rate": 2.0358661363065746e-07, |
| "loss": 2.9485, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.92, |
| "learning_rate": 7.335497040648898e-08, |
| "loss": 2.946, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.96, |
| "learning_rate": 8.154097978591014e-09, |
| "loss": 2.982, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.99, |
| "step": 615, |
| "total_flos": 7.322328732745728e+16, |
| "train_loss": 2.964561536835461, |
| "train_runtime": 1710.3688, |
| "train_samples_per_second": 5.771, |
| "train_steps_per_second": 0.36 |
| } |
| ], |
| "max_steps": 615, |
| "num_train_epochs": 3, |
| "total_flos": 7.322328732745728e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|