| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 10.0, |
| "eval_steps": 94.0, |
| "global_step": 940, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.010666666666666666, |
| "grad_norm": 7.547696113586426, |
| "learning_rate": 2.1276595744680853e-06, |
| "loss": 2.038382053375244, |
| "step": 1, |
| "token_acc": 0.7836630504148053 |
| }, |
| { |
| "epoch": 0.23466666666666666, |
| "grad_norm": 1.0874807834625244, |
| "learning_rate": 4.680851063829788e-05, |
| "loss": 1.306983402797154, |
| "step": 22, |
| "token_acc": 0.8200718489643049 |
| }, |
| { |
| "epoch": 0.4693333333333333, |
| "grad_norm": 0.5040778517723083, |
| "learning_rate": 9.361702127659576e-05, |
| "loss": 0.2126124772158536, |
| "step": 44, |
| "token_acc": 0.9077583539603961 |
| }, |
| { |
| "epoch": 0.704, |
| "grad_norm": 0.6848906874656677, |
| "learning_rate": 9.988834393115767e-05, |
| "loss": 0.18232445283369583, |
| "step": 66, |
| "token_acc": 0.9224162264607444 |
| }, |
| { |
| "epoch": 0.9386666666666666, |
| "grad_norm": 0.4289134442806244, |
| "learning_rate": 9.948077993573981e-05, |
| "loss": 0.1672248515215787, |
| "step": 88, |
| "token_acc": 0.9286029973080477 |
| }, |
| { |
| "epoch": 1.1706666666666667, |
| "grad_norm": 0.5577853322029114, |
| "learning_rate": 9.877696350987952e-05, |
| "loss": 0.1583817113529552, |
| "step": 110, |
| "token_acc": 0.9318281542941481 |
| }, |
| { |
| "epoch": 1.4053333333333333, |
| "grad_norm": 0.35486462712287903, |
| "learning_rate": 9.778110855899658e-05, |
| "loss": 0.1497676264155995, |
| "step": 132, |
| "token_acc": 0.937381329113924 |
| }, |
| { |
| "epoch": 1.6400000000000001, |
| "grad_norm": 0.9932478666305542, |
| "learning_rate": 9.649917748808843e-05, |
| "loss": 0.13864619081670587, |
| "step": 154, |
| "token_acc": 0.9411919237408992 |
| }, |
| { |
| "epoch": 1.8746666666666667, |
| "grad_norm": 0.25615832209587097, |
| "learning_rate": 9.493884550348589e-05, |
| "loss": 0.14688751914284445, |
| "step": 176, |
| "token_acc": 0.9386684033098376 |
| }, |
| { |
| "epoch": 2.1066666666666665, |
| "grad_norm": 0.5196049213409424, |
| "learning_rate": 9.31094546596864e-05, |
| "loss": 0.12925255298614502, |
| "step": 198, |
| "token_acc": 0.9438259706541835 |
| }, |
| { |
| "epoch": 2.3413333333333335, |
| "grad_norm": 0.5867960453033447, |
| "learning_rate": 9.102195792639676e-05, |
| "loss": 0.12472861463373358, |
| "step": 220, |
| "token_acc": 0.947999520364523 |
| }, |
| { |
| "epoch": 2.576, |
| "grad_norm": 0.5322598218917847, |
| "learning_rate": 8.868885361066834e-05, |
| "loss": 0.1263554963198575, |
| "step": 242, |
| "token_acc": 0.9466004101593312 |
| }, |
| { |
| "epoch": 2.8106666666666666, |
| "grad_norm": 0.453080415725708, |
| "learning_rate": 8.6124110526754e-05, |
| "loss": 0.12584879181601785, |
| "step": 264, |
| "token_acc": 0.9469967319033845 |
| }, |
| { |
| "epoch": 3.042666666666667, |
| "grad_norm": 0.4214714467525482, |
| "learning_rate": 8.334308436171159e-05, |
| "loss": 0.12248275496742943, |
| "step": 286, |
| "token_acc": 0.9486226581357036 |
| }, |
| { |
| "epoch": 3.2773333333333334, |
| "grad_norm": 1.330133080482483, |
| "learning_rate": 8.036242573749142e-05, |
| "loss": 0.10653089393268932, |
| "step": 308, |
| "token_acc": 0.9579908314890926 |
| }, |
| { |
| "epoch": 3.512, |
| "grad_norm": 0.6831464171409607, |
| "learning_rate": 7.719998051996087e-05, |
| "loss": 0.11062429168007591, |
| "step": 330, |
| "token_acc": 0.954451425068955 |
| }, |
| { |
| "epoch": 3.7466666666666666, |
| "grad_norm": 0.4873807430267334, |
| "learning_rate": 7.387468297173764e-05, |
| "loss": 0.10803152214397084, |
| "step": 352, |
| "token_acc": 0.9542289335732729 |
| }, |
| { |
| "epoch": 3.981333333333333, |
| "grad_norm": 0.5915949940681458, |
| "learning_rate": 7.040644238855e-05, |
| "loss": 0.1053880669853904, |
| "step": 374, |
| "token_acc": 0.9552102027568826 |
| }, |
| { |
| "epoch": 4.213333333333333, |
| "grad_norm": 0.6838200688362122, |
| "learning_rate": 6.681602389785683e-05, |
| "loss": 0.09514215859499844, |
| "step": 396, |
| "token_acc": 0.9615836133263166 |
| }, |
| { |
| "epoch": 4.448, |
| "grad_norm": 1.012934684753418, |
| "learning_rate": 6.312492413341274e-05, |
| "loss": 0.09638077562505548, |
| "step": 418, |
| "token_acc": 0.961881908658196 |
| }, |
| { |
| "epoch": 4.682666666666667, |
| "grad_norm": 0.4542749524116516, |
| "learning_rate": 5.9355242530142626e-05, |
| "loss": 0.0886035832491788, |
| "step": 440, |
| "token_acc": 0.9629837562691819 |
| }, |
| { |
| "epoch": 4.917333333333334, |
| "grad_norm": 0.9579441547393799, |
| "learning_rate": 5.552954900991139e-05, |
| "loss": 0.08472504399039528, |
| "step": 462, |
| "token_acc": 0.9649002327267178 |
| }, |
| { |
| "epoch": 5.149333333333334, |
| "grad_norm": 0.5529518723487854, |
| "learning_rate": 5.167074885038373e-05, |
| "loss": 0.07990209622816606, |
| "step": 484, |
| "token_acc": 0.9683850418780229 |
| }, |
| { |
| "epoch": 5.384, |
| "grad_norm": 1.123564600944519, |
| "learning_rate": 4.780194554603444e-05, |
| "loss": 0.06963089921257713, |
| "step": 506, |
| "token_acc": 0.9725745762054385 |
| }, |
| { |
| "epoch": 5.618666666666667, |
| "grad_norm": 1.0872821807861328, |
| "learning_rate": 4.3946302482390287e-05, |
| "loss": 0.06721324270421808, |
| "step": 528, |
| "token_acc": 0.9737011878107092 |
| }, |
| { |
| "epoch": 5.8533333333333335, |
| "grad_norm": 0.6262445449829102, |
| "learning_rate": 4.0126904251691044e-05, |
| "loss": 0.07399303262883966, |
| "step": 550, |
| "token_acc": 0.9713927914530583 |
| }, |
| { |
| "epoch": 6.085333333333334, |
| "grad_norm": 0.46083056926727295, |
| "learning_rate": 3.6366618440303434e-05, |
| "loss": 0.06862657720392401, |
| "step": 572, |
| "token_acc": 0.972604921557697 |
| }, |
| { |
| "epoch": 6.32, |
| "grad_norm": 1.431113362312317, |
| "learning_rate": 3.2687958715397374e-05, |
| "loss": 0.05023954131386497, |
| "step": 594, |
| "token_acc": 0.9803737263601158 |
| }, |
| { |
| "epoch": 6.554666666666667, |
| "grad_norm": 1.3620234727859497, |
| "learning_rate": 2.911295003061526e-05, |
| "loss": 0.04778737913478504, |
| "step": 616, |
| "token_acc": 0.9810086358368537 |
| }, |
| { |
| "epoch": 6.789333333333333, |
| "grad_norm": 1.1580841541290283, |
| "learning_rate": 2.566299675777771e-05, |
| "loss": 0.052371426062150436, |
| "step": 638, |
| "token_acc": 0.9782565488730771 |
| }, |
| { |
| "epoch": 7.021333333333334, |
| "grad_norm": 0.5715532898902893, |
| "learning_rate": 2.235875453415075e-05, |
| "loss": 0.052268792282451286, |
| "step": 660, |
| "token_acc": 0.9788740994016364 |
| }, |
| { |
| "epoch": 7.256, |
| "grad_norm": 2.0458245277404785, |
| "learning_rate": 1.9220006592553074e-05, |
| "loss": 0.03598861802708019, |
| "step": 682, |
| "token_acc": 0.9860324506352365 |
| }, |
| { |
| "epoch": 7.490666666666667, |
| "grad_norm": 0.8864838480949402, |
| "learning_rate": 1.6265545314742835e-05, |
| "loss": 0.034599818966605446, |
| "step": 704, |
| "token_acc": 0.9864760730523233 |
| }, |
| { |
| "epoch": 7.725333333333333, |
| "grad_norm": 0.8155806064605713, |
| "learning_rate": 1.3513059717250038e-05, |
| "loss": 0.033493843945589935, |
| "step": 726, |
| "token_acc": 0.9862047705290741 |
| }, |
| { |
| "epoch": 7.96, |
| "grad_norm": 1.2418785095214844, |
| "learning_rate": 1.0979029543301717e-05, |
| "loss": 0.037197887897491455, |
| "step": 748, |
| "token_acc": 0.9842913524285882 |
| }, |
| { |
| "epoch": 8.192, |
| "grad_norm": 0.944339394569397, |
| "learning_rate": 8.678626594935384e-06, |
| "loss": 0.030396569858897816, |
| "step": 770, |
| "token_acc": 0.9886741351717346 |
| }, |
| { |
| "epoch": 8.426666666666666, |
| "grad_norm": 0.6999959349632263, |
| "learning_rate": 6.625623896047101e-06, |
| "loss": 0.02375302802432667, |
| "step": 792, |
| "token_acc": 0.9904859763291091 |
| }, |
| { |
| "epoch": 8.661333333333333, |
| "grad_norm": 2.007450580596924, |
| "learning_rate": 4.832313230235253e-06, |
| "loss": 0.023021272637627342, |
| "step": 814, |
| "token_acc": 0.9902346707040122 |
| }, |
| { |
| "epoch": 8.896, |
| "grad_norm": 0.8154206275939941, |
| "learning_rate": 3.3094315471594194e-06, |
| "loss": 0.022192109714854847, |
| "step": 836, |
| "token_acc": 0.9908001879110554 |
| }, |
| { |
| "epoch": 9.128, |
| "grad_norm": 0.8900894522666931, |
| "learning_rate": 2.066096678035523e-06, |
| "loss": 0.02179607613520189, |
| "step": 858, |
| "token_acc": 0.9918525549754785 |
| }, |
| { |
| "epoch": 9.362666666666666, |
| "grad_norm": 0.5602992177009583, |
| "learning_rate": 1.109752745153292e-06, |
| "loss": 0.01696674796667966, |
| "step": 880, |
| "token_acc": 0.9937646436399366 |
| }, |
| { |
| "epoch": 9.597333333333333, |
| "grad_norm": 0.6423795223236084, |
| "learning_rate": 4.461255922609986e-07, |
| "loss": 0.018755904652855614, |
| "step": 902, |
| "token_acc": 0.9920016077170418 |
| }, |
| { |
| "epoch": 9.832, |
| "grad_norm": 1.2169914245605469, |
| "learning_rate": 7.918850266596111e-08, |
| "loss": 0.015993623570962387, |
| "step": 924, |
| "token_acc": 0.9939921670025477 |
| } |
| ], |
| "logging_steps": 22, |
| "max_steps": 940, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 10, |
| "save_steps": 94, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 2.3182069090304563e+18, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
