| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.29631824579598487, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.9999729068921297e-05, | |
| "loss": 1.8898, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9998916281557476e-05, | |
| "loss": 1.7273, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999756165552527e-05, | |
| "loss": 1.6799, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999566522018553e-05, | |
| "loss": 1.6431, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.999322701664249e-05, | |
| "loss": 1.6153, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9990247097742984e-05, | |
| "loss": 1.5933, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9986725528075205e-05, | |
| "loss": 1.5913, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.998266238396737e-05, | |
| "loss": 1.5434, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.997805775348605e-05, | |
| "loss": 1.5304, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.997291173643424e-05, | |
| "loss": 1.5531, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.996722444434921e-05, | |
| "loss": 1.5446, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.99609960005001e-05, | |
| "loss": 1.5352, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.995422653988524e-05, | |
| "loss": 1.5303, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.994691620922919e-05, | |
| "loss": 1.5449, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.993906516697964e-05, | |
| "loss": 1.5114, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9930673583303865e-05, | |
| "loss": 1.5043, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.992174164008515e-05, | |
| "loss": 1.5476, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.991226953091877e-05, | |
| "loss": 1.5107, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9902257461107824e-05, | |
| "loss": 1.5104, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.9891705647658795e-05, | |
| "loss": 1.5298, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.988061431927681e-05, | |
| "loss": 1.4907, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.986898371636071e-05, | |
| "loss": 1.5127, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.985681409099784e-05, | |
| "loss": 1.5037, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.984410570695858e-05, | |
| "loss": 1.5029, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.983085883969063e-05, | |
| "loss": 1.4725, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.981707377631303e-05, | |
| "loss": 1.5148, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9802750815609936e-05, | |
| "loss": 1.4993, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.978789026802419e-05, | |
| "loss": 1.5006, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9772492455650494e-05, | |
| "loss": 1.4885, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.975655771222855e-05, | |
| "loss": 1.4898, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9740086383135706e-05, | |
| "loss": 1.4906, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.97230788253796e-05, | |
| "loss": 1.4796, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.970553540759028e-05, | |
| "loss": 1.4861, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.968745651001231e-05, | |
| "loss": 1.4827, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.9668842524496526e-05, | |
| "loss": 1.4884, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.964969385449149e-05, | |
| "loss": 1.4873, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.96300109150348e-05, | |
| "loss": 1.4848, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.960979413274404e-05, | |
| "loss": 1.4881, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9589043945807594e-05, | |
| "loss": 1.4618, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9567760803975105e-05, | |
| "loss": 1.4858, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.954594516854773e-05, | |
| "loss": 1.4777, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.952359751236817e-05, | |
| "loss": 1.4828, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.950071831981038e-05, | |
| "loss": 1.4571, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9477308086769117e-05, | |
| "loss": 1.4724, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.945336732064915e-05, | |
| "loss": 1.4771, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.9428896540354294e-05, | |
| "loss": 1.4604, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.940389627627613e-05, | |
| "loss": 1.4815, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.937836707028255e-05, | |
| "loss": 1.4859, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.935230947570597e-05, | |
| "loss": 1.4715, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.932572405733137e-05, | |
| "loss": 1.4759, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.929861139138404e-05, | |
| "loss": 1.4678, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9270972065517083e-05, | |
| "loss": 1.4754, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.924280667879869e-05, | |
| "loss": 1.462, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.921411584169915e-05, | |
| "loss": 1.4704, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.918490017607761e-05, | |
| "loss": 1.4661, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.915516031516863e-05, | |
| "loss": 1.471, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.912489690356841e-05, | |
| "loss": 1.451, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.909411059722084e-05, | |
| "loss": 1.4411, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.9062802063403316e-05, | |
| "loss": 1.456, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.90309719807122e-05, | |
| "loss": 1.4678, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.8998621039048205e-05, | |
| "loss": 1.479, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.896574993960136e-05, | |
| "loss": 1.4471, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.893235939483587e-05, | |
| "loss": 1.453, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8898450128474626e-05, | |
| "loss": 1.4696, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.886402287548357e-05, | |
| "loss": 1.4526, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8829078382055725e-05, | |
| "loss": 1.4429, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8793617405595025e-05, | |
| "loss": 1.4491, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8757640714699924e-05, | |
| "loss": 1.4411, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.872114908914671e-05, | |
| "loss": 1.4543, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.8684143319872636e-05, | |
| "loss": 1.4556, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.864662420895873e-05, | |
| "loss": 1.4506, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.860859256961244e-05, | |
| "loss": 1.4671, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.857004922615002e-05, | |
| "loss": 1.4469, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.8530995013978645e-05, | |
| "loss": 1.4554, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.84914307795783e-05, | |
| "loss": 1.4671, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.845135738048343e-05, | |
| "loss": 1.445, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.841077568526439e-05, | |
| "loss": 1.4469, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.836968657350857e-05, | |
| "loss": 1.4677, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.832809093580135e-05, | |
| "loss": 1.4653, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8285989673706826e-05, | |
| "loss": 1.4342, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.824338369974822e-05, | |
| "loss": 1.458, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.8200273937388126e-05, | |
| "loss": 1.4541, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.81566613210085e-05, | |
| "loss": 1.4324, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.81125467958904e-05, | |
| "loss": 1.4405, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.80679313181935e-05, | |
| "loss": 1.4408, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.8022815854935356e-05, | |
| "loss": 1.4395, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.797720138397045e-05, | |
| "loss": 1.4359, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.793108889396902e-05, | |
| "loss": 1.442, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.7884479384395594e-05, | |
| "loss": 1.4566, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7837373865487345e-05, | |
| "loss": 1.4257, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.77897733582322e-05, | |
| "loss": 1.4755, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.774167889434671e-05, | |
| "loss": 1.4476, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.769309151625366e-05, | |
| "loss": 1.4531, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.7644012277059516e-05, | |
| "loss": 1.447, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.7594442240531574e-05, | |
| "loss": 1.4201, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.754438248107491e-05, | |
| "loss": 1.4323, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7493834083709104e-05, | |
| "loss": 1.4432, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.7442798144044695e-05, | |
| "loss": 1.4339, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.739127576825945e-05, | |
| "loss": 1.4477, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.733926807307441e-05, | |
| "loss": 1.4242, | |
| "step": 1000 | |
| } | |
| ], | |
| "max_steps": 6748, | |
| "num_train_epochs": 2, | |
| "total_flos": 1.731069029080105e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |