diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,4459 +1,2416 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 1.0, - "global_step": 7393, + "epoch": 0.11911940976332462, + "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, - "learning_rate": 1.9972947382659272e-05, - "loss": 1.3959, + "learning_rate": 2.7443463702742214e-06, + "loss": 0.9983, "step": 10 }, { "epoch": 0.0, - "learning_rate": 1.9945894765318546e-05, - "loss": 1.3384, + "learning_rate": 3.6776215359677665e-06, + "loss": 0.9472, "step": 20 }, { "epoch": 0.0, - "learning_rate": 1.991884214797782e-05, - "loss": 1.4375, + "learning_rate": 4.205772220091474e-06, + "loss": 0.8982, "step": 30 }, { - "epoch": 0.01, - "learning_rate": 1.989178953063709e-05, - "loss": 1.3479, + "epoch": 0.0, + "learning_rate": 4.5758099602275054e-06, + "loss": 0.9054, "step": 40 }, { - "epoch": 0.01, - "learning_rate": 1.9864736913296363e-05, - "loss": 1.5122, + "epoch": 0.0, + "learning_rate": 4.860906353882052e-06, + "loss": 0.8821, "step": 50 }, { - "epoch": 0.01, - "learning_rate": 1.9837684295955634e-05, - "loss": 1.4171, + "epoch": 0.0, + "learning_rate": 5.071517109647391e-06, + "loss": 0.8922, "step": 60 }, { - "epoch": 0.01, - "learning_rate": 1.9810631678614907e-05, - "loss": 1.3833, + "epoch": 0.0, + "learning_rate": 5.270189845174036e-06, + "loss": 0.9029, "step": 70 }, { - "epoch": 0.01, - "learning_rate": 1.978357906127418e-05, - "loss": 1.4005, + "epoch": 0.0, + "learning_rate": 5.441554849783425e-06, + "loss": 0.9273, "step": 80 }, { - "epoch": 0.01, - "learning_rate": 1.975652644393345e-05, - "loss": 1.4811, + "epoch": 0.0, + "learning_rate": 5.592219913271158e-06, + "loss": 0.8706, "step": 90 }, { - "epoch": 0.01, - "learning_rate": 1.9729473826592725e-05, - "loss": 1.3316, + "epoch": 0.0, + "learning_rate": 5.726651243437972e-06, + "loss": 0.8855, "step": 100 }, { - "epoch": 0.01, - "learning_rate": 1.9702421209251995e-05, - "loss": 1.2898, + "epoch": 0.0, + "learning_rate": 5.848009334523169e-06, + "loss": 0.8569, "step": 110 }, { - "epoch": 0.02, - "learning_rate": 1.967536859191127e-05, - "loss": 1.5072, + "epoch": 0.0, + "learning_rate": 5.958613046622581e-06, + "loss": 0.87, "step": 120 }, { - "epoch": 0.02, - "learning_rate": 1.9648315974570543e-05, - "loss": 1.5131, + "epoch": 0.0, + "learning_rate": 6.0602142268914325e-06, + "loss": 0.8761, "step": 130 }, { - "epoch": 0.02, - "learning_rate": 1.9621263357229813e-05, - "loss": 1.4454, + "epoch": 0.0, + "learning_rate": 6.1541687207389654e-06, + "loss": 0.8512, "step": 140 }, { - "epoch": 0.02, - "learning_rate": 1.9594210739889087e-05, - "loss": 1.5047, + "epoch": 0.0, + "learning_rate": 6.241547407784983e-06, + "loss": 0.8554, "step": 150 }, { - "epoch": 0.02, - "learning_rate": 1.9567158122548357e-05, - "loss": 1.4437, + "epoch": 0.0, + "learning_rate": 6.323210857171233e-06, + "loss": 0.8594, "step": 160 }, { - "epoch": 0.02, - "learning_rate": 1.954010550520763e-05, - "loss": 1.4017, + "epoch": 0.01, + "learning_rate": 6.399861030745893e-06, + "loss": 0.8442, "step": 170 }, { - "epoch": 0.02, - "learning_rate": 1.9513052887866905e-05, - "loss": 1.4372, + "epoch": 0.01, + "learning_rate": 6.465041342203772e-06, + "loss": 0.8634, "step": 180 }, { - "epoch": 0.03, - "learning_rate": 1.9486000270526175e-05, - "loss": 1.4318, + "epoch": 0.01, + "learning_rate": 6.533685310665599e-06, + "loss": 0.8689, "step": 190 }, { - "epoch": 0.03, - "learning_rate": 1.945894765318545e-05, - "loss": 1.3671, + "epoch": 0.01, + "learning_rate": 6.598752410624751e-06, + "loss": 0.908, "step": 200 }, { - "epoch": 0.03, - "learning_rate": 1.943189503584472e-05, - "loss": 1.4456, + "epoch": 0.01, + "learning_rate": 6.660597016320156e-06, + "loss": 0.8696, "step": 210 }, { - "epoch": 0.03, - "learning_rate": 1.940484241850399e-05, - "loss": 1.5135, + "epoch": 0.01, + "learning_rate": 6.719523314515253e-06, + "loss": 0.862, "step": 220 }, { - "epoch": 0.03, - "learning_rate": 1.9377789801163266e-05, - "loss": 1.3754, + "epoch": 0.01, + "learning_rate": 6.775794355733718e-06, + "loss": 0.8188, "step": 230 }, { - "epoch": 0.03, - "learning_rate": 1.9350737183822537e-05, - "loss": 1.484, + "epoch": 0.01, + "learning_rate": 6.829639152752426e-06, + "loss": 0.8459, "step": 240 }, { - "epoch": 0.03, - "learning_rate": 1.9323684566481807e-05, - "loss": 1.4229, + "epoch": 0.01, + "learning_rate": 6.881258311058162e-06, + "loss": 0.8276, "step": 250 }, { - "epoch": 0.04, - "learning_rate": 1.929663194914108e-05, - "loss": 1.468, + "epoch": 0.01, + "learning_rate": 6.930828541168008e-06, + "loss": 0.8474, "step": 260 }, { - "epoch": 0.04, - "learning_rate": 1.9269579331800354e-05, - "loss": 1.4239, + "epoch": 0.01, + "learning_rate": 6.978506309122687e-06, + "loss": 0.8767, "step": 270 }, { - "epoch": 0.04, - "learning_rate": 1.9242526714459625e-05, - "loss": 1.4005, + "epoch": 0.01, + "learning_rate": 7.024430815431458e-06, + "loss": 0.846, "step": 280 }, { - "epoch": 0.04, - "learning_rate": 1.92154740971189e-05, - "loss": 1.4134, + "epoch": 0.01, + "learning_rate": 7.064366909249716e-06, + "loss": 0.8219, "step": 290 }, { - "epoch": 0.04, - "learning_rate": 1.918842147977817e-05, - "loss": 1.4436, + "epoch": 0.01, + "learning_rate": 7.107292297340902e-06, + "loss": 0.85, "step": 300 }, { - "epoch": 0.04, - "learning_rate": 1.9161368862437442e-05, - "loss": 1.425, + "epoch": 0.01, + "learning_rate": 7.148791325892338e-06, + "loss": 0.8484, "step": 310 }, { - "epoch": 0.04, - "learning_rate": 1.9134316245096716e-05, - "loss": 1.4789, + "epoch": 0.01, + "learning_rate": 7.188955746727152e-06, + "loss": 0.8553, "step": 320 }, { - "epoch": 0.04, - "learning_rate": 1.9107263627755987e-05, - "loss": 1.3974, + "epoch": 0.01, + "learning_rate": 7.227868733817176e-06, + "loss": 0.845, "step": 330 }, { - "epoch": 0.05, - "learning_rate": 1.908021101041526e-05, - "loss": 1.4658, + "epoch": 0.01, + "learning_rate": 7.265605920301813e-06, + "loss": 0.8831, "step": 340 }, { - "epoch": 0.05, - "learning_rate": 1.905315839307453e-05, - "loss": 1.5112, + "epoch": 0.01, + "learning_rate": 7.302236283380594e-06, + "loss": 0.8332, "step": 350 }, { - "epoch": 0.05, - "learning_rate": 1.9026105775733804e-05, - "loss": 1.4009, + "epoch": 0.01, + "learning_rate": 7.337822903097415e-06, + "loss": 0.8109, "step": 360 }, { - "epoch": 0.05, - "learning_rate": 1.8999053158393078e-05, - "loss": 1.6013, + "epoch": 0.01, + "learning_rate": 7.3724236159865746e-06, + "loss": 0.8536, "step": 370 }, { - "epoch": 0.05, - "learning_rate": 1.8972000541052348e-05, - "loss": 1.4251, + "epoch": 0.01, + "learning_rate": 7.406091580591952e-06, + "loss": 0.8525, "step": 380 }, { - "epoch": 0.05, - "learning_rate": 1.8944947923711622e-05, - "loss": 1.4459, + "epoch": 0.01, + "learning_rate": 7.438875768743633e-06, + "loss": 0.8468, "step": 390 }, { - "epoch": 0.05, - "learning_rate": 1.8917895306370892e-05, - "loss": 1.4584, + "epoch": 0.01, + "learning_rate": 7.47082139398946e-06, + "loss": 0.8792, "step": 400 }, { - "epoch": 0.06, - "learning_rate": 1.8890842689030166e-05, - "loss": 1.4266, + "epoch": 0.01, + "learning_rate": 7.501970286588418e-06, + "loss": 0.8434, "step": 410 }, { - "epoch": 0.06, - "learning_rate": 1.886379007168944e-05, - "loss": 1.504, + "epoch": 0.01, + "learning_rate": 7.53236122286999e-06, + "loss": 0.8485, "step": 420 }, { - "epoch": 0.06, - "learning_rate": 1.883673745434871e-05, - "loss": 1.4038, + "epoch": 0.01, + "learning_rate": 7.562030215465441e-06, + "loss": 0.8467, "step": 430 }, { - "epoch": 0.06, - "learning_rate": 1.8809684837007984e-05, - "loss": 1.4052, + "epoch": 0.01, + "learning_rate": 7.59101076985994e-06, + "loss": 0.8311, "step": 440 }, { - "epoch": 0.06, - "learning_rate": 1.8782632219667254e-05, - "loss": 1.4572, + "epoch": 0.01, + "learning_rate": 7.619334111849203e-06, + "loss": 0.8438, "step": 450 }, { - "epoch": 0.06, - "learning_rate": 1.8755579602326524e-05, - "loss": 1.5259, + "epoch": 0.01, + "learning_rate": 7.647029389772635e-06, + "loss": 0.875, "step": 460 }, { - "epoch": 0.06, - "learning_rate": 1.87285269849858e-05, - "loss": 1.4457, + "epoch": 0.01, + "learning_rate": 7.674123854806827e-06, + "loss": 0.8308, "step": 470 }, { - "epoch": 0.06, - "learning_rate": 1.8701474367645072e-05, - "loss": 1.4768, + "epoch": 0.01, + "learning_rate": 7.700643022115063e-06, + "loss": 0.8445, "step": 480 }, { - "epoch": 0.07, - "learning_rate": 1.8674421750304342e-05, - "loss": 1.3379, + "epoch": 0.01, + "learning_rate": 7.72661081524147e-06, + "loss": 0.8857, "step": 490 }, { - "epoch": 0.07, - "learning_rate": 1.8647369132963616e-05, - "loss": 1.5158, + "epoch": 0.01, + "learning_rate": 7.752049695797903e-06, + "loss": 0.8448, "step": 500 }, { - "epoch": 0.07, - "learning_rate": 1.8620316515622886e-05, - "loss": 1.4572, + "epoch": 0.02, + "learning_rate": 7.776980780205255e-06, + "loss": 0.827, "step": 510 }, { - "epoch": 0.07, - "learning_rate": 1.859326389828216e-05, - "loss": 1.354, + "epoch": 0.02, + "learning_rate": 7.801423945009691e-06, + "loss": 0.7976, "step": 520 }, { - "epoch": 0.07, - "learning_rate": 1.8566211280941434e-05, - "loss": 1.3889, + "epoch": 0.02, + "learning_rate": 7.825397922089763e-06, + "loss": 0.8469, "step": 530 }, { - "epoch": 0.07, - "learning_rate": 1.8539158663600704e-05, - "loss": 1.3632, + "epoch": 0.02, + "learning_rate": 7.848920384896975e-06, + "loss": 0.8477, "step": 540 }, { - "epoch": 0.07, - "learning_rate": 1.8512106046259978e-05, - "loss": 1.3859, + "epoch": 0.02, + "learning_rate": 7.87200802672445e-06, + "loss": 0.819, "step": 550 }, { - "epoch": 0.08, - "learning_rate": 1.8485053428919248e-05, - "loss": 1.3953, + "epoch": 0.02, + "learning_rate": 7.894676631871945e-06, + "loss": 0.822, "step": 560 }, { - "epoch": 0.08, - "learning_rate": 1.845800081157852e-05, - "loss": 1.4717, + "epoch": 0.02, + "learning_rate": 7.916941140467113e-06, + "loss": 0.8273, "step": 570 }, { - "epoch": 0.08, - "learning_rate": 1.8430948194237795e-05, - "loss": 1.4671, + "epoch": 0.02, + "learning_rate": 7.938815707609735e-06, + "loss": 0.8637, "step": 580 }, { - "epoch": 0.08, - "learning_rate": 1.8403895576897066e-05, - "loss": 1.4807, + "epoch": 0.02, + "learning_rate": 7.960313757425318e-06, + "loss": 0.8359, "step": 590 }, { - "epoch": 0.08, - "learning_rate": 1.837684295955634e-05, - "loss": 1.3847, + "epoch": 0.02, + "learning_rate": 7.98144803254492e-06, + "loss": 0.8353, "step": 600 }, { - "epoch": 0.08, - "learning_rate": 1.834979034221561e-05, - "loss": 1.4243, + "epoch": 0.02, + "learning_rate": 8.002230639467933e-06, + "loss": 0.8416, "step": 610 }, { - "epoch": 0.08, - "learning_rate": 1.8322737724874883e-05, - "loss": 1.4222, + "epoch": 0.02, + "learning_rate": 8.022673090212183e-06, + "loss": 0.8389, "step": 620 }, { - "epoch": 0.09, - "learning_rate": 1.8295685107534157e-05, - "loss": 1.4151, + "epoch": 0.02, + "learning_rate": 8.042786340610085e-06, + "loss": 0.835, "step": 630 }, { - "epoch": 0.09, - "learning_rate": 1.8268632490193427e-05, - "loss": 1.3298, + "epoch": 0.02, + "learning_rate": 8.062580825569854e-06, + "loss": 0.8712, "step": 640 }, { - "epoch": 0.09, - "learning_rate": 1.8241579872852698e-05, - "loss": 1.3483, + "epoch": 0.02, + "learning_rate": 8.082066491585883e-06, + "loss": 0.8419, "step": 650 }, { - "epoch": 0.09, - "learning_rate": 1.821452725551197e-05, - "loss": 1.4535, + "epoch": 0.02, + "learning_rate": 8.101252826751851e-06, + "loss": 0.8674, "step": 660 }, { - "epoch": 0.09, - "learning_rate": 1.8187474638171245e-05, - "loss": 1.3634, + "epoch": 0.02, + "learning_rate": 8.120148888503286e-06, + "loss": 0.8224, "step": 670 }, { - "epoch": 0.09, - "learning_rate": 1.8160422020830515e-05, - "loss": 1.2987, + "epoch": 0.02, + "learning_rate": 8.13876332929263e-06, + "loss": 0.828, "step": 680 }, { - "epoch": 0.09, - "learning_rate": 1.813336940348979e-05, - "loss": 1.5129, + "epoch": 0.02, + "learning_rate": 8.157104420378997e-06, + "loss": 0.8568, "step": 690 }, { - "epoch": 0.09, - "learning_rate": 1.810631678614906e-05, - "loss": 1.3101, + "epoch": 0.02, + "learning_rate": 8.17518007389634e-06, + "loss": 0.8216, "step": 700 }, { - "epoch": 0.1, - "learning_rate": 1.8079264168808333e-05, - "loss": 1.5473, + "epoch": 0.02, + "learning_rate": 8.192997863347404e-06, + "loss": 0.7905, "step": 710 }, { - "epoch": 0.1, - "learning_rate": 1.8052211551467607e-05, - "loss": 1.501, + "epoch": 0.02, + "learning_rate": 8.210565042656325e-06, + "loss": 0.8285, "step": 720 }, { - "epoch": 0.1, - "learning_rate": 1.8025158934126877e-05, - "loss": 1.5162, + "epoch": 0.02, + "learning_rate": 8.22788856389983e-06, + "loss": 0.7934, "step": 730 }, { - "epoch": 0.1, - "learning_rate": 1.799810631678615e-05, - "loss": 1.4877, + "epoch": 0.02, + "learning_rate": 8.244975093825531e-06, + "loss": 0.8617, "step": 740 }, { - "epoch": 0.1, - "learning_rate": 1.797105369944542e-05, - "loss": 1.4245, + "epoch": 0.02, + "learning_rate": 8.26183102925549e-06, + "loss": 0.8155, "step": 750 }, { - "epoch": 0.1, - "learning_rate": 1.7944001082104695e-05, - "loss": 1.503, + "epoch": 0.02, + "learning_rate": 8.278462511464196e-06, + "loss": 0.7782, "step": 760 }, { - "epoch": 0.1, - "learning_rate": 1.791694846476397e-05, - "loss": 1.4438, + "epoch": 0.02, + "learning_rate": 8.294875439611728e-06, + "loss": 0.8354, "step": 770 }, { - "epoch": 0.11, - "learning_rate": 1.788989584742324e-05, - "loss": 1.3682, + "epoch": 0.02, + "learning_rate": 8.311075483305743e-06, + "loss": 0.8133, "step": 780 }, { - "epoch": 0.11, - "learning_rate": 1.7862843230082513e-05, - "loss": 1.4456, + "epoch": 0.02, + "learning_rate": 8.32706809435913e-06, + "loss": 0.835, "step": 790 }, { - "epoch": 0.11, - "learning_rate": 1.7835790612741783e-05, - "loss": 1.4186, + "epoch": 0.02, + "learning_rate": 8.342858517804411e-06, + "loss": 0.8339, "step": 800 }, { - "epoch": 0.11, - "learning_rate": 1.7808737995401057e-05, - "loss": 1.4397, + "epoch": 0.02, + "learning_rate": 8.358451802220541e-06, + "loss": 0.8192, "step": 810 }, { - "epoch": 0.11, - "learning_rate": 1.778168537806033e-05, - "loss": 1.4031, + "epoch": 0.02, + "learning_rate": 8.373852809422983e-06, + "loss": 0.8187, "step": 820 }, { - "epoch": 0.11, - "learning_rate": 1.77546327607196e-05, - "loss": 1.401, + "epoch": 0.02, + "learning_rate": 8.389066223563605e-06, + "loss": 0.8202, "step": 830 }, { - "epoch": 0.11, - "learning_rate": 1.7727580143378874e-05, - "loss": 1.4405, + "epoch": 0.03, + "learning_rate": 8.404096559683005e-06, + "loss": 0.8373, "step": 840 }, { - "epoch": 0.11, - "learning_rate": 1.7700527526038145e-05, - "loss": 1.4139, + "epoch": 0.03, + "learning_rate": 8.418948171754334e-06, + "loss": 0.8085, "step": 850 }, { - "epoch": 0.12, - "learning_rate": 1.767347490869742e-05, - "loss": 1.4095, + "epoch": 0.03, + "learning_rate": 8.433625260254492e-06, + "loss": 0.8292, "step": 860 }, { - "epoch": 0.12, - "learning_rate": 1.7646422291356692e-05, - "loss": 1.464, + "epoch": 0.03, + "learning_rate": 8.448131879295593e-06, + "loss": 0.8149, "step": 870 }, { - "epoch": 0.12, - "learning_rate": 1.7619369674015962e-05, - "loss": 1.4072, + "epoch": 0.03, + "learning_rate": 8.462471943347044e-06, + "loss": 0.8174, "step": 880 }, { - "epoch": 0.12, - "learning_rate": 1.7592317056675233e-05, - "loss": 1.3421, + "epoch": 0.03, + "learning_rate": 8.476649233576078e-06, + "loss": 0.822, "step": 890 }, { - "epoch": 0.12, - "learning_rate": 1.7565264439334506e-05, - "loss": 1.3978, + "epoch": 0.03, + "learning_rate": 8.49066740383246e-06, + "loss": 0.8188, "step": 900 }, { - "epoch": 0.12, - "learning_rate": 1.753821182199378e-05, - "loss": 1.3789, + "epoch": 0.03, + "learning_rate": 8.504529986301062e-06, + "loss": 0.8184, "step": 910 }, { - "epoch": 0.12, - "learning_rate": 1.751115920465305e-05, - "loss": 1.4517, + "epoch": 0.03, + "learning_rate": 8.518240396844167e-06, + "loss": 0.8405, "step": 920 }, { - "epoch": 0.13, - "learning_rate": 1.7484106587312324e-05, - "loss": 1.3778, + "epoch": 0.03, + "learning_rate": 8.531801940053751e-06, + "loss": 0.8127, "step": 930 }, { - "epoch": 0.13, - "learning_rate": 1.7457053969971595e-05, - "loss": 1.3721, + "epoch": 0.03, + "learning_rate": 8.545217814032373e-06, + "loss": 0.8087, "step": 940 }, { - "epoch": 0.13, - "learning_rate": 1.7430001352630868e-05, - "loss": 1.3519, + "epoch": 0.03, + "learning_rate": 8.558491114920063e-06, + "loss": 0.7969, "step": 950 }, { - "epoch": 0.13, - "learning_rate": 1.7402948735290142e-05, - "loss": 1.4252, + "epoch": 0.03, + "learning_rate": 8.571624841183157e-06, + "loss": 0.7868, "step": 960 }, { - "epoch": 0.13, - "learning_rate": 1.7375896117949412e-05, - "loss": 1.4174, + "epoch": 0.03, + "learning_rate": 8.58462189767999e-06, + "loss": 0.803, "step": 970 }, { - "epoch": 0.13, - "learning_rate": 1.7348843500608686e-05, - "loss": 1.4173, + "epoch": 0.03, + "learning_rate": 8.597485099517221e-06, + "loss": 0.8116, "step": 980 }, { - "epoch": 0.13, - "learning_rate": 1.7321790883267956e-05, - "loss": 1.382, + "epoch": 0.03, + "learning_rate": 8.61021717570959e-06, + "loss": 0.8084, "step": 990 }, { - "epoch": 0.14, - "learning_rate": 1.729473826592723e-05, - "loss": 1.4501, + "epoch": 0.03, + "learning_rate": 8.622820772655021e-06, + "loss": 0.8238, "step": 1000 }, { - "epoch": 0.14, - "learning_rate": 1.7267685648586504e-05, - "loss": 1.4216, + "epoch": 0.03, + "learning_rate": 8.635298457436133e-06, + "loss": 0.8182, "step": 1010 }, { - "epoch": 0.14, - "learning_rate": 1.7240633031245774e-05, - "loss": 1.4291, + "epoch": 0.03, + "learning_rate": 8.647652720958468e-06, + "loss": 0.8431, "step": 1020 }, { - "epoch": 0.14, - "learning_rate": 1.7213580413905048e-05, - "loss": 1.4538, + "epoch": 0.03, + "learning_rate": 8.659885980935016e-06, + "loss": 0.8601, "step": 1030 }, { - "epoch": 0.14, - "learning_rate": 1.7186527796564318e-05, - "loss": 1.4791, + "epoch": 0.03, + "learning_rate": 8.67200058472601e-06, + "loss": 0.8029, "step": 1040 }, { - "epoch": 0.14, - "learning_rate": 1.7159475179223592e-05, - "loss": 1.4549, + "epoch": 0.03, + "learning_rate": 8.683998812042295e-06, + "loss": 0.8336, "step": 1050 }, { - "epoch": 0.14, - "learning_rate": 1.7132422561882865e-05, - "loss": 1.3336, + "epoch": 0.03, + "learning_rate": 8.695882877520106e-06, + "loss": 0.8158, "step": 1060 }, { - "epoch": 0.14, - "learning_rate": 1.7105369944542136e-05, - "loss": 1.4338, + "epoch": 0.03, + "learning_rate": 8.70765493317449e-06, + "loss": 0.7954, "step": 1070 }, { - "epoch": 0.15, - "learning_rate": 1.707831732720141e-05, - "loss": 1.3757, + "epoch": 0.03, + "learning_rate": 8.719317070738179e-06, + "loss": 0.8156, "step": 1080 }, { - "epoch": 0.15, - "learning_rate": 1.705126470986068e-05, - "loss": 1.2891, + "epoch": 0.03, + "learning_rate": 8.73087132389229e-06, + "loss": 0.826, "step": 1090 }, { - "epoch": 0.15, - "learning_rate": 1.702421209251995e-05, - "loss": 1.4253, + "epoch": 0.03, + "learning_rate": 8.742319670394788e-06, + "loss": 0.8407, "step": 1100 }, { - "epoch": 0.15, - "learning_rate": 1.6997159475179227e-05, - "loss": 1.434, + "epoch": 0.03, + "learning_rate": 8.753664034112258e-06, + "loss": 0.8149, "step": 1110 }, { - "epoch": 0.15, - "learning_rate": 1.6970106857838498e-05, - "loss": 1.3248, + "epoch": 0.03, + "learning_rate": 8.764906286960285e-06, + "loss": 0.7951, "step": 1120 }, { - "epoch": 0.15, - "learning_rate": 1.6943054240497768e-05, - "loss": 1.4, + "epoch": 0.03, + "learning_rate": 8.776048250757231e-06, + "loss": 0.8036, "step": 1130 }, { - "epoch": 0.15, - "learning_rate": 1.691600162315704e-05, - "loss": 1.3843, + "epoch": 0.03, + "learning_rate": 8.787091698996089e-06, + "loss": 0.8013, "step": 1140 }, { - "epoch": 0.16, - "learning_rate": 1.6888949005816312e-05, - "loss": 1.4056, + "epoch": 0.03, + "learning_rate": 8.798038358538676e-06, + "loss": 0.8109, "step": 1150 }, { - "epoch": 0.16, - "learning_rate": 1.6861896388475586e-05, - "loss": 1.3752, + "epoch": 0.03, + "learning_rate": 8.808889911236233e-06, + "loss": 0.8091, "step": 1160 }, { - "epoch": 0.16, - "learning_rate": 1.683484377113486e-05, - "loss": 1.4068, + "epoch": 0.03, + "learning_rate": 8.819647995480226e-06, + "loss": 0.799, "step": 1170 }, { - "epoch": 0.16, - "learning_rate": 1.680779115379413e-05, - "loss": 1.4207, + "epoch": 0.04, + "learning_rate": 8.83031420768692e-06, + "loss": 0.8179, "step": 1180 }, { - "epoch": 0.16, - "learning_rate": 1.6780738536453403e-05, - "loss": 1.3498, + "epoch": 0.04, + "learning_rate": 8.840890103719111e-06, + "loss": 0.7954, "step": 1190 }, { - "epoch": 0.16, - "learning_rate": 1.6753685919112677e-05, - "loss": 1.4908, + "epoch": 0.04, + "learning_rate": 8.851377200248169e-06, + "loss": 0.8444, "step": 1200 }, { - "epoch": 0.16, - "learning_rate": 1.6726633301771947e-05, - "loss": 1.3216, + "epoch": 0.04, + "learning_rate": 8.861776976059358e-06, + "loss": 0.8211, "step": 1210 }, { - "epoch": 0.17, - "learning_rate": 1.669958068443122e-05, - "loss": 1.391, + "epoch": 0.04, + "learning_rate": 8.872090873303281e-06, + "loss": 0.8336, "step": 1220 }, { - "epoch": 0.17, - "learning_rate": 1.667252806709049e-05, - "loss": 1.4478, + "epoch": 0.04, + "learning_rate": 8.88232029869606e-06, + "loss": 0.8114, "step": 1230 }, { - "epoch": 0.17, - "learning_rate": 1.6645475449749765e-05, - "loss": 1.4299, + "epoch": 0.04, + "learning_rate": 8.892466624670776e-06, + "loss": 0.7884, "step": 1240 }, { - "epoch": 0.17, - "learning_rate": 1.661842283240904e-05, - "loss": 1.4335, + "epoch": 0.04, + "learning_rate": 8.90253119048252e-06, + "loss": 0.7746, "step": 1250 }, { - "epoch": 0.17, - "learning_rate": 1.659137021506831e-05, - "loss": 1.4385, + "epoch": 0.04, + "learning_rate": 8.912515303269253e-06, + "loss": 0.8045, "step": 1260 }, { - "epoch": 0.17, - "learning_rate": 1.6564317597727583e-05, - "loss": 1.4116, + "epoch": 0.04, + "learning_rate": 8.922420239070625e-06, + "loss": 0.8118, "step": 1270 }, { - "epoch": 0.17, - "learning_rate": 1.6537264980386853e-05, - "loss": 1.3132, + "epoch": 0.04, + "learning_rate": 8.932247243806712e-06, + "loss": 0.8386, "step": 1280 }, { - "epoch": 0.17, - "learning_rate": 1.6510212363046127e-05, - "loss": 1.3827, + "epoch": 0.04, + "learning_rate": 8.941997534218523e-06, + "loss": 0.7903, "step": 1290 }, { - "epoch": 0.18, - "learning_rate": 1.64831597457054e-05, - "loss": 1.4384, + "epoch": 0.04, + "learning_rate": 8.951672298772117e-06, + "loss": 0.803, "step": 1300 }, { - "epoch": 0.18, - "learning_rate": 1.645610712836467e-05, - "loss": 1.3981, + "epoch": 0.04, + "learning_rate": 8.961272698527948e-06, + "loss": 0.7786, "step": 1310 }, { - "epoch": 0.18, - "learning_rate": 1.6429054511023945e-05, - "loss": 1.3039, + "epoch": 0.04, + "learning_rate": 8.97079986797706e-06, + "loss": 0.8029, "step": 1320 }, { - "epoch": 0.18, - "learning_rate": 1.6402001893683215e-05, - "loss": 1.4895, + "epoch": 0.04, + "learning_rate": 8.980254915845623e-06, + "loss": 0.8261, "step": 1330 }, { - "epoch": 0.18, - "learning_rate": 1.6374949276342485e-05, - "loss": 1.4101, + "epoch": 0.04, + "learning_rate": 8.98963892586927e-06, + "loss": 0.7883, "step": 1340 }, { - "epoch": 0.18, - "learning_rate": 1.6347896659001762e-05, - "loss": 1.3817, + "epoch": 0.04, + "learning_rate": 8.99895295753853e-06, + "loss": 0.82, "step": 1350 }, { - "epoch": 0.18, - "learning_rate": 1.6320844041661033e-05, - "loss": 1.3831, + "epoch": 0.04, + "learning_rate": 9.008198046816687e-06, + "loss": 0.8107, "step": 1360 }, { - "epoch": 0.19, - "learning_rate": 1.6293791424320303e-05, - "loss": 1.4333, + "epoch": 0.04, + "learning_rate": 9.017375206831278e-06, + "loss": 0.8036, "step": 1370 }, { - "epoch": 0.19, - "learning_rate": 1.6266738806979577e-05, - "loss": 1.4683, + "epoch": 0.04, + "learning_rate": 9.026485428540338e-06, + "loss": 0.7946, "step": 1380 }, { - "epoch": 0.19, - "learning_rate": 1.6239686189638847e-05, - "loss": 1.2967, + "epoch": 0.04, + "learning_rate": 9.035529681374568e-06, + "loss": 0.8063, "step": 1390 }, { - "epoch": 0.19, - "learning_rate": 1.621263357229812e-05, - "loss": 1.4496, + "epoch": 0.04, + "learning_rate": 9.044508913856369e-06, + "loss": 0.8009, "step": 1400 }, { - "epoch": 0.19, - "learning_rate": 1.6185580954957394e-05, - "loss": 1.4164, + "epoch": 0.04, + "learning_rate": 9.053424054196832e-06, + "loss": 0.8255, "step": 1410 }, { - "epoch": 0.19, - "learning_rate": 1.6158528337616665e-05, - "loss": 1.3667, + "epoch": 0.04, + "learning_rate": 9.06227601087153e-06, + "loss": 0.831, "step": 1420 }, { - "epoch": 0.19, - "learning_rate": 1.613147572027594e-05, - "loss": 1.4397, + "epoch": 0.04, + "learning_rate": 9.07106567317608e-06, + "loss": 0.8133, "step": 1430 }, { - "epoch": 0.19, - "learning_rate": 1.610442310293521e-05, - "loss": 1.4361, + "epoch": 0.04, + "learning_rate": 9.079793911762282e-06, + "loss": 0.8305, "step": 1440 }, { - "epoch": 0.2, - "learning_rate": 1.6077370485594482e-05, - "loss": 1.4322, + "epoch": 0.04, + "learning_rate": 9.088461579155646e-06, + "loss": 0.8283, "step": 1450 }, { - "epoch": 0.2, - "learning_rate": 1.6050317868253756e-05, - "loss": 1.3377, + "epoch": 0.04, + "learning_rate": 9.097069510255097e-06, + "loss": 0.815, "step": 1460 }, { - "epoch": 0.2, - "learning_rate": 1.6023265250913026e-05, - "loss": 1.4274, + "epoch": 0.04, + "learning_rate": 9.105618522815576e-06, + "loss": 0.8059, "step": 1470 }, { - "epoch": 0.2, - "learning_rate": 1.59962126335723e-05, - "loss": 1.316, + "epoch": 0.04, + "learning_rate": 9.114109417914234e-06, + "loss": 0.7878, "step": 1480 }, { - "epoch": 0.2, - "learning_rate": 1.596916001623157e-05, - "loss": 1.4283, + "epoch": 0.04, + "learning_rate": 9.122542980400885e-06, + "loss": 0.8045, "step": 1490 }, { - "epoch": 0.2, - "learning_rate": 1.5942107398890844e-05, - "loss": 1.4837, + "epoch": 0.04, + "learning_rate": 9.130919979333354e-06, + "loss": 0.8232, "step": 1500 }, { - "epoch": 0.2, - "learning_rate": 1.5915054781550118e-05, - "loss": 1.3838, + "epoch": 0.04, + "learning_rate": 9.139241168398338e-06, + "loss": 0.812, "step": 1510 }, { - "epoch": 0.21, - "learning_rate": 1.5888002164209388e-05, - "loss": 1.339, + "epoch": 0.05, + "learning_rate": 9.14668313197706e-06, + "loss": 0.8241, "step": 1520 }, { - "epoch": 0.21, - "learning_rate": 1.5860949546868662e-05, - "loss": 1.3901, + "epoch": 0.05, + "learning_rate": 9.154900305423636e-06, + "loss": 0.8143, "step": 1530 }, { - "epoch": 0.21, - "learning_rate": 1.5833896929527932e-05, - "loss": 1.4442, + "epoch": 0.05, + "learning_rate": 9.163063771470579e-06, + "loss": 0.8081, "step": 1540 }, { - "epoch": 0.21, - "learning_rate": 1.5806844312187206e-05, - "loss": 1.4565, + "epoch": 0.05, + "learning_rate": 9.171174227623984e-06, + "loss": 0.8189, "step": 1550 }, { - "epoch": 0.21, - "learning_rate": 1.577979169484648e-05, - "loss": 1.2944, + "epoch": 0.05, + "learning_rate": 9.179232357889644e-06, + "loss": 0.8298, "step": 1560 }, { - "epoch": 0.21, - "learning_rate": 1.575273907750575e-05, - "loss": 1.3916, + "epoch": 0.05, + "learning_rate": 9.187238833119211e-06, + "loss": 0.7892, "step": 1570 }, { - "epoch": 0.21, - "learning_rate": 1.572568646016502e-05, - "loss": 1.4557, + "epoch": 0.05, + "learning_rate": 9.195194311345337e-06, + "loss": 0.7882, "step": 1580 }, { - "epoch": 0.22, - "learning_rate": 1.5698633842824294e-05, - "loss": 1.3379, + "epoch": 0.05, + "learning_rate": 9.203099438106217e-06, + "loss": 0.8157, "step": 1590 }, { - "epoch": 0.22, - "learning_rate": 1.5671581225483568e-05, - "loss": 1.3807, + "epoch": 0.05, + "learning_rate": 9.21095484675992e-06, + "loss": 0.7796, "step": 1600 }, { - "epoch": 0.22, - "learning_rate": 1.5644528608142838e-05, - "loss": 1.4264, + "epoch": 0.05, + "learning_rate": 9.21876115878889e-06, + "loss": 0.8145, "step": 1610 }, { - "epoch": 0.22, - "learning_rate": 1.5617475990802112e-05, - "loss": 1.4578, + "epoch": 0.05, + "learning_rate": 9.226518984095009e-06, + "loss": 0.8507, "step": 1620 }, { - "epoch": 0.22, - "learning_rate": 1.5590423373461382e-05, - "loss": 1.397, + "epoch": 0.05, + "learning_rate": 9.234228921285531e-06, + "loss": 0.8343, "step": 1630 }, { - "epoch": 0.22, - "learning_rate": 1.5563370756120656e-05, - "loss": 1.4386, + "epoch": 0.05, + "learning_rate": 9.241891557950258e-06, + "loss": 0.8324, "step": 1640 }, { - "epoch": 0.22, - "learning_rate": 1.553631813877993e-05, - "loss": 1.3933, + "epoch": 0.05, + "learning_rate": 9.249507470930265e-06, + "loss": 0.8095, "step": 1650 }, { - "epoch": 0.22, - "learning_rate": 1.55092655214392e-05, - "loss": 1.4173, + "epoch": 0.05, + "learning_rate": 9.257077226578481e-06, + "loss": 0.8039, "step": 1660 }, { - "epoch": 0.23, - "learning_rate": 1.5482212904098473e-05, - "loss": 1.2836, + "epoch": 0.05, + "learning_rate": 9.264601381012412e-06, + "loss": 0.8153, "step": 1670 }, { - "epoch": 0.23, - "learning_rate": 1.5455160286757744e-05, - "loss": 1.3643, + "epoch": 0.05, + "learning_rate": 9.272080480359307e-06, + "loss": 0.8062, "step": 1680 }, { - "epoch": 0.23, - "learning_rate": 1.5428107669417018e-05, - "loss": 1.4037, + "epoch": 0.05, + "learning_rate": 9.279515060994034e-06, + "loss": 0.8195, "step": 1690 }, { - "epoch": 0.23, - "learning_rate": 1.540105505207629e-05, - "loss": 1.2814, + "epoch": 0.05, + "learning_rate": 9.286905649769893e-06, + "loss": 0.8274, "step": 1700 }, { - "epoch": 0.23, - "learning_rate": 1.537400243473556e-05, - "loss": 1.4156, + "epoch": 0.05, + "learning_rate": 9.294252764242674e-06, + "loss": 0.8118, "step": 1710 }, { - "epoch": 0.23, - "learning_rate": 1.5346949817394835e-05, - "loss": 1.3672, + "epoch": 0.05, + "learning_rate": 9.301556912888123e-06, + "loss": 0.8065, "step": 1720 }, { - "epoch": 0.23, - "learning_rate": 1.5319897200054106e-05, - "loss": 1.5362, + "epoch": 0.05, + "learning_rate": 9.308818595313135e-06, + "loss": 0.8255, "step": 1730 }, { - "epoch": 0.24, - "learning_rate": 1.529284458271338e-05, - "loss": 1.4041, + "epoch": 0.05, + "learning_rate": 9.3160383024608e-06, + "loss": 0.7981, "step": 1740 }, { - "epoch": 0.24, - "learning_rate": 1.5265791965372653e-05, - "loss": 1.4587, + "epoch": 0.05, + "learning_rate": 9.323216516809576e-06, + "loss": 0.7966, "step": 1750 }, { - "epoch": 0.24, - "learning_rate": 1.5238739348031923e-05, - "loss": 1.5285, + "epoch": 0.05, + "learning_rate": 9.330353712566771e-06, + "loss": 0.7948, "step": 1760 }, { - "epoch": 0.24, - "learning_rate": 1.5211686730691195e-05, - "loss": 1.4492, + "epoch": 0.05, + "learning_rate": 9.33745035585653e-06, + "loss": 0.8022, "step": 1770 }, { - "epoch": 0.24, - "learning_rate": 1.5184634113350467e-05, - "loss": 1.4173, + "epoch": 0.05, + "learning_rate": 9.344506904902502e-06, + "loss": 0.7971, "step": 1780 }, { - "epoch": 0.24, - "learning_rate": 1.5157581496009741e-05, - "loss": 1.4211, + "epoch": 0.05, + "learning_rate": 9.35152381020538e-06, + "loss": 0.83, "step": 1790 }, { - "epoch": 0.24, - "learning_rate": 1.5130528878669013e-05, - "loss": 1.3379, + "epoch": 0.05, + "learning_rate": 9.358501514715491e-06, + "loss": 0.8204, "step": 1800 }, { - "epoch": 0.24, - "learning_rate": 1.5103476261328285e-05, - "loss": 1.3973, + "epoch": 0.05, + "learning_rate": 9.365440454000577e-06, + "loss": 0.7879, "step": 1810 }, { - "epoch": 0.25, - "learning_rate": 1.5076423643987557e-05, - "loss": 1.3135, + "epoch": 0.05, + "learning_rate": 9.372341056408958e-06, + "loss": 0.8028, "step": 1820 }, { - "epoch": 0.25, - "learning_rate": 1.5049371026646829e-05, - "loss": 1.5027, + "epoch": 0.05, + "learning_rate": 9.379203743228185e-06, + "loss": 0.812, "step": 1830 }, { - "epoch": 0.25, - "learning_rate": 1.5022318409306103e-05, - "loss": 1.3825, + "epoch": 0.05, + "learning_rate": 9.386028928839394e-06, + "loss": 0.7885, "step": 1840 }, { - "epoch": 0.25, - "learning_rate": 1.4995265791965375e-05, - "loss": 1.4282, + "epoch": 0.06, + "learning_rate": 9.39281702086744e-06, + "loss": 0.8224, "step": 1850 }, { - "epoch": 0.25, - "learning_rate": 1.4968213174624647e-05, - "loss": 1.4295, + "epoch": 0.06, + "learning_rate": 9.399568420326976e-06, + "loss": 0.8072, "step": 1860 }, { - "epoch": 0.25, - "learning_rate": 1.4941160557283919e-05, - "loss": 1.4623, + "epoch": 0.06, + "learning_rate": 9.406283521764617e-06, + "loss": 0.8267, "step": 1870 }, { - "epoch": 0.25, - "learning_rate": 1.4914107939943189e-05, - "loss": 1.3697, + "epoch": 0.06, + "learning_rate": 9.41296271339729e-06, + "loss": 0.8174, "step": 1880 }, { - "epoch": 0.26, - "learning_rate": 1.4887055322602465e-05, - "loss": 1.3479, + "epoch": 0.06, + "learning_rate": 9.419606377246914e-06, + "loss": 0.7966, "step": 1890 }, { - "epoch": 0.26, - "learning_rate": 1.4860002705261737e-05, - "loss": 1.3305, + "epoch": 0.06, + "learning_rate": 9.42621488927151e-06, + "loss": 0.8085, "step": 1900 }, { - "epoch": 0.26, - "learning_rate": 1.4832950087921007e-05, - "loss": 1.3113, + "epoch": 0.06, + "learning_rate": 9.432788619492866e-06, + "loss": 0.7715, "step": 1910 }, { - "epoch": 0.26, - "learning_rate": 1.4805897470580279e-05, - "loss": 1.3887, + "epoch": 0.06, + "learning_rate": 9.439327932120854e-06, + "loss": 0.7899, "step": 1920 }, { - "epoch": 0.26, - "learning_rate": 1.4778844853239551e-05, - "loss": 1.4085, + "epoch": 0.06, + "learning_rate": 9.445833185674515e-06, + "loss": 0.8234, "step": 1930 }, { - "epoch": 0.26, - "learning_rate": 1.4751792235898825e-05, - "loss": 1.4331, + "epoch": 0.06, + "learning_rate": 9.45230473310001e-06, + "loss": 0.7594, "step": 1940 }, { - "epoch": 0.26, - "learning_rate": 1.4724739618558097e-05, - "loss": 1.4234, + "epoch": 0.06, + "learning_rate": 9.45874292188551e-06, + "loss": 0.7989, "step": 1950 }, { - "epoch": 0.27, - "learning_rate": 1.4697687001217369e-05, - "loss": 1.3118, + "epoch": 0.06, + "learning_rate": 9.465148094173177e-06, + "loss": 0.8325, "step": 1960 }, { - "epoch": 0.27, - "learning_rate": 1.467063438387664e-05, - "loss": 1.4936, + "epoch": 0.06, + "learning_rate": 9.471520586868255e-06, + "loss": 0.784, "step": 1970 }, { - "epoch": 0.27, - "learning_rate": 1.4643581766535913e-05, - "loss": 1.3971, + "epoch": 0.06, + "learning_rate": 9.477860731745404e-06, + "loss": 0.7925, "step": 1980 }, { - "epoch": 0.27, - "learning_rate": 1.4616529149195186e-05, - "loss": 1.4079, + "epoch": 0.06, + "learning_rate": 9.484168855552354e-06, + "loss": 0.7905, "step": 1990 }, { - "epoch": 0.27, - "learning_rate": 1.4589476531854458e-05, - "loss": 1.2972, + "epoch": 0.06, + "learning_rate": 9.490445280110947e-06, + "loss": 0.839, "step": 2000 }, { - "epoch": 0.27, - "learning_rate": 1.456242391451373e-05, - "loss": 1.4365, + "epoch": 0.06, + "learning_rate": 9.496690322415662e-06, + "loss": 0.7983, "step": 2010 }, { - "epoch": 0.27, - "learning_rate": 1.4535371297173002e-05, - "loss": 1.3128, + "epoch": 0.06, + "learning_rate": 9.502904294729667e-06, + "loss": 0.7627, "step": 2020 }, { - "epoch": 0.27, - "learning_rate": 1.4508318679832274e-05, - "loss": 1.3887, + "epoch": 0.06, + "learning_rate": 9.509087504678532e-06, + "loss": 0.7854, "step": 2030 }, { - "epoch": 0.28, - "learning_rate": 1.4481266062491548e-05, - "loss": 1.3066, + "epoch": 0.06, + "learning_rate": 9.515240255341591e-06, + "loss": 0.8019, "step": 2040 }, { - "epoch": 0.28, - "learning_rate": 1.445421344515082e-05, - "loss": 1.4259, + "epoch": 0.06, + "learning_rate": 9.521362845341097e-06, + "loss": 0.8063, "step": 2050 }, { - "epoch": 0.28, - "learning_rate": 1.4427160827810092e-05, - "loss": 1.374, + "epoch": 0.06, + "learning_rate": 9.527455568929186e-06, + "loss": 0.7965, "step": 2060 }, { - "epoch": 0.28, - "learning_rate": 1.4400108210469364e-05, - "loss": 1.2865, + "epoch": 0.06, + "learning_rate": 9.533518716072731e-06, + "loss": 0.7828, "step": 2070 }, { - "epoch": 0.28, - "learning_rate": 1.4373055593128634e-05, - "loss": 1.2938, + "epoch": 0.06, + "learning_rate": 9.539552572536162e-06, + "loss": 0.7979, "step": 2080 }, { - "epoch": 0.28, - "learning_rate": 1.434600297578791e-05, - "loss": 1.429, + "epoch": 0.06, + "learning_rate": 9.545557419962264e-06, + "loss": 0.7961, "step": 2090 }, { - "epoch": 0.28, - "learning_rate": 1.4318950358447182e-05, - "loss": 1.4477, + "epoch": 0.06, + "learning_rate": 9.551533535951078e-06, + "loss": 0.8017, "step": 2100 }, { - "epoch": 0.29, - "learning_rate": 1.4291897741106452e-05, - "loss": 1.3928, + "epoch": 0.06, + "learning_rate": 9.5574811941369e-06, + "loss": 0.7808, "step": 2110 }, { - "epoch": 0.29, - "learning_rate": 1.4264845123765724e-05, - "loss": 1.3417, + "epoch": 0.06, + "learning_rate": 9.56340066426346e-06, + "loss": 0.8078, "step": 2120 }, { - "epoch": 0.29, - "learning_rate": 1.4237792506425e-05, - "loss": 1.3849, + "epoch": 0.06, + "learning_rate": 9.569292212257332e-06, + "loss": 0.7956, "step": 2130 }, { - "epoch": 0.29, - "learning_rate": 1.421073988908427e-05, - "loss": 1.319, + "epoch": 0.06, + "learning_rate": 9.575156100299618e-06, + "loss": 0.792, "step": 2140 }, { - "epoch": 0.29, - "learning_rate": 1.4183687271743542e-05, - "loss": 1.4355, + "epoch": 0.06, + "learning_rate": 9.580992586895953e-06, + "loss": 0.8122, "step": 2150 }, { - "epoch": 0.29, - "learning_rate": 1.4156634654402814e-05, - "loss": 1.4974, + "epoch": 0.06, + "learning_rate": 9.586801926944894e-06, + "loss": 0.7944, "step": 2160 }, { - "epoch": 0.29, - "learning_rate": 1.4129582037062086e-05, - "loss": 1.4579, + "epoch": 0.06, + "learning_rate": 9.592584371804719e-06, + "loss": 0.8075, "step": 2170 }, { - "epoch": 0.29, - "learning_rate": 1.410252941972136e-05, - "loss": 1.4073, + "epoch": 0.06, + "learning_rate": 9.598340169358675e-06, + "loss": 0.8205, "step": 2180 }, { - "epoch": 0.3, - "learning_rate": 1.4075476802380632e-05, - "loss": 1.402, + "epoch": 0.07, + "learning_rate": 9.604069564078745e-06, + "loss": 0.8142, "step": 2190 }, { - "epoch": 0.3, - "learning_rate": 1.4048424185039904e-05, - "loss": 1.3516, + "epoch": 0.07, + "learning_rate": 9.609772797087972e-06, + "loss": 0.8124, "step": 2200 }, { - "epoch": 0.3, - "learning_rate": 1.4021371567699176e-05, - "loss": 1.3967, + "epoch": 0.07, + "learning_rate": 9.615450106221318e-06, + "loss": 0.8008, "step": 2210 }, { - "epoch": 0.3, - "learning_rate": 1.3994318950358448e-05, - "loss": 1.3478, + "epoch": 0.07, + "learning_rate": 9.621101726085202e-06, + "loss": 0.8091, "step": 2220 }, { - "epoch": 0.3, - "learning_rate": 1.3967266333017721e-05, - "loss": 1.4304, + "epoch": 0.07, + "learning_rate": 9.626727888115666e-06, + "loss": 0.7764, "step": 2230 }, { - "epoch": 0.3, - "learning_rate": 1.3940213715676993e-05, - "loss": 1.3697, + "epoch": 0.07, + "learning_rate": 9.632328820635236e-06, + "loss": 0.8022, "step": 2240 }, { - "epoch": 0.3, - "learning_rate": 1.3913161098336265e-05, - "loss": 1.4548, + "epoch": 0.07, + "learning_rate": 9.637904748908531e-06, + "loss": 0.8228, "step": 2250 }, { - "epoch": 0.31, - "learning_rate": 1.3886108480995537e-05, - "loss": 1.4578, + "epoch": 0.07, + "learning_rate": 9.643455895196622e-06, + "loss": 0.7955, "step": 2260 }, { - "epoch": 0.31, - "learning_rate": 1.385905586365481e-05, - "loss": 1.3775, + "epoch": 0.07, + "learning_rate": 9.648982478810187e-06, + "loss": 0.7859, "step": 2270 }, { - "epoch": 0.31, - "learning_rate": 1.3832003246314083e-05, - "loss": 1.2729, + "epoch": 0.07, + "learning_rate": 9.654484716161512e-06, + "loss": 0.7956, "step": 2280 }, { - "epoch": 0.31, - "learning_rate": 1.3804950628973355e-05, - "loss": 1.3702, + "epoch": 0.07, + "learning_rate": 9.659962820815331e-06, + "loss": 0.8379, "step": 2290 }, { - "epoch": 0.31, - "learning_rate": 1.3777898011632627e-05, - "loss": 1.3805, + "epoch": 0.07, + "learning_rate": 9.665417003538574e-06, + "loss": 0.7958, "step": 2300 }, { - "epoch": 0.31, - "learning_rate": 1.37508453942919e-05, - "loss": 1.3707, + "epoch": 0.07, + "learning_rate": 9.670847472349013e-06, + "loss": 0.8078, "step": 2310 }, { - "epoch": 0.31, - "learning_rate": 1.372379277695117e-05, - "loss": 1.3759, + "epoch": 0.07, + "learning_rate": 9.676254432562877e-06, + "loss": 0.7764, "step": 2320 }, { - "epoch": 0.32, - "learning_rate": 1.3696740159610445e-05, - "loss": 1.4317, + "epoch": 0.07, + "learning_rate": 9.681638086841426e-06, + "loss": 0.7853, "step": 2330 }, { - "epoch": 0.32, - "learning_rate": 1.3669687542269717e-05, - "loss": 1.384, + "epoch": 0.07, + "learning_rate": 9.686998635236535e-06, + "loss": 0.7811, "step": 2340 }, { - "epoch": 0.32, - "learning_rate": 1.3642634924928987e-05, - "loss": 1.357, + "epoch": 0.07, + "learning_rate": 9.69233627523529e-06, + "loss": 0.7867, "step": 2350 }, { - "epoch": 0.32, - "learning_rate": 1.361558230758826e-05, - "loss": 1.2858, + "epoch": 0.07, + "learning_rate": 9.697651201803651e-06, + "loss": 0.8049, "step": 2360 }, { - "epoch": 0.32, - "learning_rate": 1.3588529690247531e-05, - "loss": 1.5065, + "epoch": 0.07, + "learning_rate": 9.70294360742919e-06, + "loss": 0.7858, "step": 2370 }, { - "epoch": 0.32, - "learning_rate": 1.3561477072906805e-05, - "loss": 1.37, + "epoch": 0.07, + "learning_rate": 9.708213682162902e-06, + "loss": 0.8043, "step": 2380 }, { - "epoch": 0.32, - "learning_rate": 1.3534424455566077e-05, - "loss": 1.4009, + "epoch": 0.07, + "learning_rate": 9.713461613660172e-06, + "loss": 0.7544, "step": 2390 }, { - "epoch": 0.32, - "learning_rate": 1.3507371838225349e-05, - "loss": 1.3726, + "epoch": 0.07, + "learning_rate": 9.718687587220866e-06, + "loss": 0.7901, "step": 2400 }, { - "epoch": 0.33, - "learning_rate": 1.3480319220884621e-05, - "loss": 1.361, + "epoch": 0.07, + "learning_rate": 9.723891785828587e-06, + "loss": 0.8242, "step": 2410 }, { - "epoch": 0.33, - "learning_rate": 1.3453266603543893e-05, - "loss": 1.4044, + "epoch": 0.07, + "learning_rate": 9.729074390189116e-06, + "loss": 0.8037, "step": 2420 }, { - "epoch": 0.33, - "learning_rate": 1.3426213986203167e-05, - "loss": 1.3882, + "epoch": 0.07, + "learning_rate": 9.734235578768077e-06, + "loss": 0.7992, "step": 2430 }, { - "epoch": 0.33, - "learning_rate": 1.3399161368862439e-05, - "loss": 1.2583, + "epoch": 0.07, + "learning_rate": 9.739375527827789e-06, + "loss": 0.8321, "step": 2440 }, { - "epoch": 0.33, - "learning_rate": 1.337210875152171e-05, - "loss": 1.3283, + "epoch": 0.07, + "learning_rate": 9.744494411463412e-06, + "loss": 0.7946, "step": 2450 }, { - "epoch": 0.33, - "learning_rate": 1.3345056134180983e-05, - "loss": 1.2969, + "epoch": 0.07, + "learning_rate": 9.749592401638308e-06, + "loss": 0.8284, "step": 2460 }, { - "epoch": 0.33, - "learning_rate": 1.3318003516840255e-05, - "loss": 1.3566, + "epoch": 0.07, + "learning_rate": 9.754669668218715e-06, + "loss": 0.8193, "step": 2470 }, { - "epoch": 0.34, - "learning_rate": 1.3290950899499529e-05, - "loss": 1.4263, + "epoch": 0.07, + "learning_rate": 9.759726379007711e-06, + "loss": 0.7918, "step": 2480 }, { - "epoch": 0.34, - "learning_rate": 1.32638982821588e-05, - "loss": 1.3724, + "epoch": 0.07, + "learning_rate": 9.764762699778483e-06, + "loss": 0.7733, "step": 2490 }, { - "epoch": 0.34, - "learning_rate": 1.3236845664818073e-05, - "loss": 1.3994, + "epoch": 0.07, + "learning_rate": 9.76977879430694e-06, + "loss": 0.7545, "step": 2500 }, { - "epoch": 0.34, - "learning_rate": 1.3209793047477345e-05, - "loss": 1.3567, + "epoch": 0.07, + "learning_rate": 9.774774824403668e-06, + "loss": 0.7995, "step": 2510 }, { - "epoch": 0.34, - "learning_rate": 1.3182740430136617e-05, - "loss": 1.4803, + "epoch": 0.08, + "learning_rate": 9.779750949945259e-06, + "loss": 0.822, "step": 2520 }, { - "epoch": 0.34, - "learning_rate": 1.315568781279589e-05, - "loss": 1.2974, + "epoch": 0.08, + "learning_rate": 9.784707328904993e-06, + "loss": 0.7891, "step": 2530 }, { - "epoch": 0.34, - "learning_rate": 1.3128635195455162e-05, - "loss": 1.3667, + "epoch": 0.08, + "learning_rate": 9.789644117382946e-06, + "loss": 0.8081, "step": 2540 }, { - "epoch": 0.34, - "learning_rate": 1.3101582578114433e-05, - "loss": 1.4112, + "epoch": 0.08, + "learning_rate": 9.794561469635479e-06, + "loss": 0.7719, "step": 2550 }, { - "epoch": 0.35, - "learning_rate": 1.3074529960773705e-05, - "loss": 1.3731, + "epoch": 0.08, + "learning_rate": 9.799459538104166e-06, + "loss": 0.7964, "step": 2560 }, { - "epoch": 0.35, - "learning_rate": 1.3047477343432977e-05, - "loss": 1.3438, + "epoch": 0.08, + "learning_rate": 9.804338473444142e-06, + "loss": 0.7849, "step": 2570 }, { - "epoch": 0.35, - "learning_rate": 1.302042472609225e-05, - "loss": 1.4988, + "epoch": 0.08, + "learning_rate": 9.80919842455192e-06, + "loss": 0.7943, "step": 2580 }, { - "epoch": 0.35, - "learning_rate": 1.2993372108751522e-05, - "loss": 1.3883, + "epoch": 0.08, + "learning_rate": 9.814039538592641e-06, + "loss": 0.8179, "step": 2590 }, { - "epoch": 0.35, - "learning_rate": 1.2966319491410794e-05, - "loss": 1.4478, + "epoch": 0.08, + "learning_rate": 9.81886196102683e-06, + "loss": 0.7809, "step": 2600 }, { - "epoch": 0.35, - "learning_rate": 1.2939266874070066e-05, - "loss": 1.309, + "epoch": 0.08, + "learning_rate": 9.823665835636607e-06, + "loss": 0.8191, "step": 2610 }, { - "epoch": 0.35, - "learning_rate": 1.291221425672934e-05, - "loss": 1.3898, + "epoch": 0.08, + "learning_rate": 9.82845130455142e-06, + "loss": 0.774, "step": 2620 }, { - "epoch": 0.36, - "learning_rate": 1.2885161639388612e-05, - "loss": 1.3766, + "epoch": 0.08, + "learning_rate": 9.83321850827327e-06, + "loss": 0.8062, "step": 2630 }, { - "epoch": 0.36, - "learning_rate": 1.2858109022047884e-05, - "loss": 1.3857, + "epoch": 0.08, + "learning_rate": 9.83796758570147e-06, + "loss": 0.7829, "step": 2640 }, { - "epoch": 0.36, - "learning_rate": 1.2831056404707156e-05, - "loss": 1.3507, + "epoch": 0.08, + "learning_rate": 9.842698674156915e-06, + "loss": 0.7831, "step": 2650 }, { - "epoch": 0.36, - "learning_rate": 1.2804003787366428e-05, - "loss": 1.45, + "epoch": 0.08, + "learning_rate": 9.847411909405928e-06, + "loss": 0.7677, "step": 2660 }, { - "epoch": 0.36, - "learning_rate": 1.2776951170025702e-05, - "loss": 1.3714, + "epoch": 0.08, + "learning_rate": 9.852107425683615e-06, + "loss": 0.8183, "step": 2670 }, { - "epoch": 0.36, - "learning_rate": 1.2749898552684974e-05, - "loss": 1.2946, + "epoch": 0.08, + "learning_rate": 9.856785355716825e-06, + "loss": 0.7847, "step": 2680 }, { - "epoch": 0.36, - "learning_rate": 1.2722845935344246e-05, - "loss": 1.4648, + "epoch": 0.08, + "learning_rate": 9.861445830746643e-06, + "loss": 0.7636, "step": 2690 }, { - "epoch": 0.37, - "learning_rate": 1.2695793318003518e-05, - "loss": 1.3191, + "epoch": 0.08, + "learning_rate": 9.866088980550497e-06, + "loss": 0.8071, "step": 2700 }, { - "epoch": 0.37, - "learning_rate": 1.266874070066279e-05, - "loss": 1.3008, + "epoch": 0.08, + "learning_rate": 9.870714933463837e-06, + "loss": 0.8048, "step": 2710 }, { - "epoch": 0.37, - "learning_rate": 1.2641688083322064e-05, - "loss": 1.3335, + "epoch": 0.08, + "learning_rate": 9.875323816401418e-06, + "loss": 0.7643, "step": 2720 }, { - "epoch": 0.37, - "learning_rate": 1.2614635465981336e-05, - "loss": 1.3381, + "epoch": 0.08, + "learning_rate": 9.879915754878191e-06, + "loss": 0.8177, "step": 2730 }, { - "epoch": 0.37, - "learning_rate": 1.2587582848640608e-05, - "loss": 1.326, + "epoch": 0.08, + "learning_rate": 9.884490873029821e-06, + "loss": 0.8113, "step": 2740 }, { - "epoch": 0.37, - "learning_rate": 1.256053023129988e-05, - "loss": 1.3833, + "epoch": 0.08, + "learning_rate": 9.889049293632811e-06, + "loss": 0.8018, "step": 2750 }, { - "epoch": 0.37, - "learning_rate": 1.253347761395915e-05, - "loss": 1.4497, + "epoch": 0.08, + "learning_rate": 9.893591138124284e-06, + "loss": 0.7857, "step": 2760 }, { - "epoch": 0.37, - "learning_rate": 1.2506424996618425e-05, - "loss": 1.3892, + "epoch": 0.08, + "learning_rate": 9.898116526621384e-06, + "loss": 0.7321, "step": 2770 }, { - "epoch": 0.38, - "learning_rate": 1.2479372379277697e-05, - "loss": 1.4059, + "epoch": 0.08, + "learning_rate": 9.902625577940345e-06, + "loss": 0.7987, "step": 2780 }, { - "epoch": 0.38, - "learning_rate": 1.2452319761936968e-05, - "loss": 1.3615, + "epoch": 0.08, + "learning_rate": 9.907118409615204e-06, + "loss": 0.795, "step": 2790 }, { - "epoch": 0.38, - "learning_rate": 1.242526714459624e-05, - "loss": 1.3792, + "epoch": 0.08, + "learning_rate": 9.911595137916184e-06, + "loss": 0.7824, "step": 2800 }, { - "epoch": 0.38, - "learning_rate": 1.2398214527255512e-05, - "loss": 1.3617, + "epoch": 0.08, + "learning_rate": 9.916055877867754e-06, + "loss": 0.7657, "step": 2810 }, { - "epoch": 0.38, - "learning_rate": 1.2371161909914785e-05, - "loss": 1.3857, + "epoch": 0.08, + "learning_rate": 9.920500743266357e-06, + "loss": 0.7808, "step": 2820 }, { - "epoch": 0.38, - "learning_rate": 1.2344109292574057e-05, - "loss": 1.3659, + "epoch": 0.08, + "learning_rate": 9.924929846697829e-06, + "loss": 0.7936, "step": 2830 }, { - "epoch": 0.38, - "learning_rate": 1.231705667523333e-05, - "loss": 1.3111, + "epoch": 0.08, + "learning_rate": 9.929343299554512e-06, + "loss": 0.7835, "step": 2840 }, { - "epoch": 0.39, - "learning_rate": 1.2290004057892601e-05, - "loss": 1.3687, + "epoch": 0.08, + "learning_rate": 9.933741212052067e-06, + "loss": 0.7982, "step": 2850 }, { - "epoch": 0.39, - "learning_rate": 1.2262951440551873e-05, - "loss": 1.4139, + "epoch": 0.09, + "learning_rate": 9.938123693245983e-06, + "loss": 0.7795, "step": 2860 }, { - "epoch": 0.39, - "learning_rate": 1.2235898823211147e-05, - "loss": 1.2684, + "epoch": 0.09, + "learning_rate": 9.9424908510478e-06, + "loss": 0.7723, "step": 2870 }, { - "epoch": 0.39, - "learning_rate": 1.220884620587042e-05, - "loss": 1.4344, + "epoch": 0.09, + "learning_rate": 9.94684279224107e-06, + "loss": 0.7899, "step": 2880 }, { - "epoch": 0.39, - "learning_rate": 1.2181793588529691e-05, - "loss": 1.3254, + "epoch": 0.09, + "learning_rate": 9.951179622497001e-06, + "loss": 0.7733, "step": 2890 }, { - "epoch": 0.39, - "learning_rate": 1.2154740971188963e-05, - "loss": 1.2692, + "epoch": 0.09, + "learning_rate": 9.955501446389869e-06, + "loss": 0.7995, "step": 2900 }, { - "epoch": 0.39, - "learning_rate": 1.2127688353848235e-05, - "loss": 1.4458, + "epoch": 0.09, + "learning_rate": 9.959808367412143e-06, + "loss": 0.8119, "step": 2910 }, { - "epoch": 0.39, - "learning_rate": 1.2100635736507509e-05, - "loss": 1.3394, + "epoch": 0.09, + "learning_rate": 9.964100487989352e-06, + "loss": 0.8068, "step": 2920 }, { - "epoch": 0.4, - "learning_rate": 1.2073583119166781e-05, - "loss": 1.3478, + "epoch": 0.09, + "learning_rate": 9.968377909494707e-06, + "loss": 0.7862, "step": 2930 }, { - "epoch": 0.4, - "learning_rate": 1.2046530501826053e-05, - "loss": 1.3344, + "epoch": 0.09, + "learning_rate": 9.97264073226346e-06, + "loss": 0.7889, "step": 2940 }, { - "epoch": 0.4, - "learning_rate": 1.2019477884485325e-05, - "loss": 1.3622, + "epoch": 0.09, + "learning_rate": 9.976889055607038e-06, + "loss": 0.7854, "step": 2950 }, { - "epoch": 0.4, - "learning_rate": 1.1992425267144597e-05, - "loss": 1.4438, + "epoch": 0.09, + "learning_rate": 9.981122977826906e-06, + "loss": 0.7908, "step": 2960 }, { - "epoch": 0.4, - "learning_rate": 1.196537264980387e-05, - "loss": 1.4282, + "epoch": 0.09, + "learning_rate": 9.985342596228237e-06, + "loss": 0.761, "step": 2970 }, { - "epoch": 0.4, - "learning_rate": 1.1938320032463143e-05, - "loss": 1.4355, + "epoch": 0.09, + "learning_rate": 9.989548007133313e-06, + "loss": 0.806, "step": 2980 }, { - "epoch": 0.4, - "learning_rate": 1.1911267415122415e-05, - "loss": 1.442, + "epoch": 0.09, + "learning_rate": 9.993739305894728e-06, + "loss": 0.7941, "step": 2990 }, { - "epoch": 0.41, - "learning_rate": 1.1884214797781685e-05, - "loss": 1.4028, + "epoch": 0.09, + "learning_rate": 9.997916586908357e-06, + "loss": 0.8242, "step": 3000 }, { - "epoch": 0.41, - "learning_rate": 1.1857162180440957e-05, - "loss": 1.3572, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8142, "step": 3010 }, { - "epoch": 0.41, - "learning_rate": 1.183010956310023e-05, - "loss": 1.3916, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8038, "step": 3020 }, { - "epoch": 0.41, - "learning_rate": 1.1803056945759503e-05, - "loss": 1.3676, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7903, "step": 3030 }, { - "epoch": 0.41, - "learning_rate": 1.1776004328418775e-05, - "loss": 1.3295, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7369, "step": 3040 }, { - "epoch": 0.41, - "learning_rate": 1.1748951711078047e-05, - "loss": 1.3886, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7631, "step": 3050 }, { - "epoch": 0.41, - "learning_rate": 1.1721899093737319e-05, - "loss": 1.3099, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8063, "step": 3060 }, { - "epoch": 0.42, - "learning_rate": 1.1694846476396593e-05, - "loss": 1.4045, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7876, "step": 3070 }, { - "epoch": 0.42, - "learning_rate": 1.1667793859055865e-05, - "loss": 1.36, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7775, "step": 3080 }, { - "epoch": 0.42, - "learning_rate": 1.1640741241715137e-05, - "loss": 1.4901, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8087, "step": 3090 }, { - "epoch": 0.42, - "learning_rate": 1.1613688624374409e-05, - "loss": 1.3389, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7963, "step": 3100 }, { - "epoch": 0.42, - "learning_rate": 1.1586636007033682e-05, - "loss": 1.4128, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7648, "step": 3110 }, { - "epoch": 0.42, - "learning_rate": 1.1559583389692954e-05, - "loss": 1.3237, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7854, "step": 3120 }, { - "epoch": 0.42, - "learning_rate": 1.1532530772352226e-05, - "loss": 1.4016, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8144, "step": 3130 }, { - "epoch": 0.42, - "learning_rate": 1.1505478155011498e-05, - "loss": 1.339, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7879, "step": 3140 }, { - "epoch": 0.43, - "learning_rate": 1.147842553767077e-05, - "loss": 1.2514, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7973, "step": 3150 }, { - "epoch": 0.43, - "learning_rate": 1.1451372920330044e-05, - "loss": 1.3511, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7661, "step": 3160 }, { - "epoch": 0.43, - "learning_rate": 1.1424320302989316e-05, - "loss": 1.2333, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8059, "step": 3170 }, { - "epoch": 0.43, - "learning_rate": 1.1397267685648588e-05, - "loss": 1.2941, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.7926, "step": 3180 }, { - "epoch": 0.43, - "learning_rate": 1.137021506830786e-05, - "loss": 1.2298, + "epoch": 0.09, + "learning_rate": 1e-05, + "loss": 0.8048, "step": 3190 }, { - "epoch": 0.43, - "learning_rate": 1.134316245096713e-05, - "loss": 1.3611, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8069, "step": 3200 }, { - "epoch": 0.43, - "learning_rate": 1.1316109833626406e-05, - "loss": 1.3905, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7716, "step": 3210 }, { - "epoch": 0.44, - "learning_rate": 1.1289057216285678e-05, - "loss": 1.2703, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7904, "step": 3220 }, { - "epoch": 0.44, - "learning_rate": 1.1262004598944948e-05, - "loss": 1.3445, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7995, "step": 3230 }, { - "epoch": 0.44, - "learning_rate": 1.123495198160422e-05, - "loss": 1.3907, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8062, "step": 3240 }, { - "epoch": 0.44, - "learning_rate": 1.1207899364263492e-05, - "loss": 1.4032, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7894, "step": 3250 }, { - "epoch": 0.44, - "learning_rate": 1.1180846746922766e-05, - "loss": 1.3875, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7912, "step": 3260 }, { - "epoch": 0.44, - "learning_rate": 1.1153794129582038e-05, - "loss": 1.3915, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7735, "step": 3270 }, { - "epoch": 0.44, - "learning_rate": 1.112674151224131e-05, - "loss": 1.3339, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7838, "step": 3280 }, { - "epoch": 0.45, - "learning_rate": 1.1099688894900582e-05, - "loss": 1.2987, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7779, "step": 3290 }, { - "epoch": 0.45, - "learning_rate": 1.1072636277559854e-05, - "loss": 1.4916, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7471, "step": 3300 }, { - "epoch": 0.45, - "learning_rate": 1.1045583660219128e-05, - "loss": 1.4523, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7632, "step": 3310 }, { - "epoch": 0.45, - "learning_rate": 1.10185310428784e-05, - "loss": 1.3857, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8012, "step": 3320 }, { - "epoch": 0.45, - "learning_rate": 1.0991478425537672e-05, - "loss": 1.2006, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7941, "step": 3330 }, { - "epoch": 0.45, - "learning_rate": 1.0964425808196944e-05, - "loss": 1.3905, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7791, "step": 3340 }, { - "epoch": 0.45, - "learning_rate": 1.0937373190856216e-05, - "loss": 1.4326, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8017, "step": 3350 }, { - "epoch": 0.45, - "learning_rate": 1.091032057351549e-05, - "loss": 1.3051, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8195, "step": 3360 }, { - "epoch": 0.46, - "learning_rate": 1.0883267956174761e-05, - "loss": 1.4052, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7782, "step": 3370 }, { - "epoch": 0.46, - "learning_rate": 1.0856215338834033e-05, - "loss": 1.2806, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7576, "step": 3380 }, { - "epoch": 0.46, - "learning_rate": 1.0829162721493305e-05, - "loss": 1.4119, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7975, "step": 3390 }, { - "epoch": 0.46, - "learning_rate": 1.0802110104152577e-05, - "loss": 1.4438, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7682, "step": 3400 }, { - "epoch": 0.46, - "learning_rate": 1.0775057486811851e-05, - "loss": 1.3909, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7719, "step": 3410 }, { - "epoch": 0.46, - "learning_rate": 1.0748004869471123e-05, - "loss": 1.3446, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7625, "step": 3420 }, { - "epoch": 0.46, - "learning_rate": 1.0720952252130395e-05, - "loss": 1.42, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7799, "step": 3430 }, { - "epoch": 0.47, - "learning_rate": 1.0693899634789665e-05, - "loss": 1.2771, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7617, "step": 3440 }, { - "epoch": 0.47, - "learning_rate": 1.0666847017448937e-05, - "loss": 1.3805, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.762, "step": 3450 }, { - "epoch": 0.47, - "learning_rate": 1.0639794400108213e-05, - "loss": 1.3742, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8049, "step": 3460 }, { - "epoch": 0.47, - "learning_rate": 1.0612741782767483e-05, - "loss": 1.3671, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7971, "step": 3470 }, { - "epoch": 0.47, - "learning_rate": 1.0585689165426755e-05, - "loss": 1.3164, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.8157, "step": 3480 }, { - "epoch": 0.47, - "learning_rate": 1.0558636548086027e-05, - "loss": 1.3741, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7716, "step": 3490 }, { - "epoch": 0.47, - "learning_rate": 1.05315839307453e-05, - "loss": 1.3697, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7746, "step": 3500 }, { - "epoch": 0.47, - "learning_rate": 1.0504531313404573e-05, - "loss": 1.4629, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7614, "step": 3510 }, { - "epoch": 0.48, - "learning_rate": 1.0477478696063845e-05, - "loss": 1.3748, + "epoch": 0.1, + "learning_rate": 1e-05, + "loss": 0.7895, "step": 3520 }, { - "epoch": 0.48, - "learning_rate": 1.0450426078723117e-05, - "loss": 1.3946, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.8076, "step": 3530 }, { - "epoch": 0.48, - "learning_rate": 1.0423373461382389e-05, - "loss": 1.4994, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7599, "step": 3540 }, { - "epoch": 0.48, - "learning_rate": 1.0396320844041663e-05, - "loss": 1.321, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7956, "step": 3550 }, { - "epoch": 0.48, - "learning_rate": 1.0369268226700935e-05, - "loss": 1.2496, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7751, "step": 3560 }, { - "epoch": 0.48, - "learning_rate": 1.0342215609360207e-05, - "loss": 1.3156, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7697, "step": 3570 }, { - "epoch": 0.48, - "learning_rate": 1.0315162992019479e-05, - "loss": 1.4947, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7982, "step": 3580 }, { - "epoch": 0.49, - "learning_rate": 1.028811037467875e-05, - "loss": 1.333, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7622, "step": 3590 }, { - "epoch": 0.49, - "learning_rate": 1.0261057757338024e-05, - "loss": 1.3308, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7716, "step": 3600 }, { - "epoch": 0.49, - "learning_rate": 1.0234005139997296e-05, - "loss": 1.339, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7761, "step": 3610 }, { - "epoch": 0.49, - "learning_rate": 1.0206952522656568e-05, - "loss": 1.3775, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.8177, "step": 3620 }, { - "epoch": 0.49, - "learning_rate": 1.017989990531584e-05, - "loss": 1.4347, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7722, "step": 3630 }, { - "epoch": 0.49, - "learning_rate": 1.015284728797511e-05, - "loss": 1.3368, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7809, "step": 3640 }, { - "epoch": 0.49, - "learning_rate": 1.0125794670634386e-05, - "loss": 1.2931, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.8001, "step": 3650 }, { - "epoch": 0.5, - "learning_rate": 1.0098742053293658e-05, - "loss": 1.2733, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7641, "step": 3660 }, { - "epoch": 0.5, - "learning_rate": 1.0071689435952929e-05, - "loss": 1.3117, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7582, "step": 3670 }, { - "epoch": 0.5, - "learning_rate": 1.00446368186122e-05, - "loss": 1.3899, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7644, "step": 3680 }, { - "epoch": 0.5, - "learning_rate": 1.0017584201271473e-05, - "loss": 1.3196, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7709, "step": 3690 }, { - "epoch": 0.5, - "learning_rate": 9.990531583930746e-06, - "loss": 1.3301, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7986, "step": 3700 }, { - "epoch": 0.5, - "learning_rate": 9.963478966590018e-06, - "loss": 1.4633, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7853, "step": 3710 }, { - "epoch": 0.5, - "learning_rate": 9.93642634924929e-06, - "loss": 1.2581, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7782, "step": 3720 }, { - "epoch": 0.5, - "learning_rate": 9.909373731908562e-06, - "loss": 1.212, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.747, "step": 3730 }, { - "epoch": 0.51, - "learning_rate": 9.882321114567836e-06, - "loss": 1.4123, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7427, "step": 3740 }, { - "epoch": 0.51, - "learning_rate": 9.855268497227108e-06, - "loss": 1.3047, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7813, "step": 3750 }, { - "epoch": 0.51, - "learning_rate": 9.82821587988638e-06, - "loss": 1.305, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7691, "step": 3760 }, { - "epoch": 0.51, - "learning_rate": 9.801163262545652e-06, - "loss": 1.3485, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7912, "step": 3770 }, { - "epoch": 0.51, - "learning_rate": 9.774110645204924e-06, - "loss": 1.3869, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7849, "step": 3780 }, { - "epoch": 0.51, - "learning_rate": 9.747058027864196e-06, - "loss": 1.3148, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7848, "step": 3790 }, { - "epoch": 0.51, - "learning_rate": 9.720005410523468e-06, - "loss": 1.1635, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7946, "step": 3800 }, { - "epoch": 0.52, - "learning_rate": 9.692952793182742e-06, - "loss": 1.3119, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7852, "step": 3810 }, { - "epoch": 0.52, - "learning_rate": 9.665900175842014e-06, - "loss": 1.3421, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7828, "step": 3820 }, { - "epoch": 0.52, - "learning_rate": 9.638847558501286e-06, - "loss": 1.3168, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7714, "step": 3830 }, { - "epoch": 0.52, - "learning_rate": 9.611794941160558e-06, - "loss": 1.2631, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.8218, "step": 3840 }, { - "epoch": 0.52, - "learning_rate": 9.58474232381983e-06, - "loss": 1.2664, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.8017, "step": 3850 }, { - "epoch": 0.52, - "learning_rate": 9.557689706479104e-06, - "loss": 1.3792, + "epoch": 0.11, + "learning_rate": 1e-05, + "loss": 0.7707, "step": 3860 }, { - "epoch": 0.52, - "learning_rate": 9.530637089138376e-06, - "loss": 1.4046, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7678, "step": 3870 }, { - "epoch": 0.52, - "learning_rate": 9.503584471797646e-06, - "loss": 1.3742, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7662, "step": 3880 }, { - "epoch": 0.53, - "learning_rate": 9.47653185445692e-06, - "loss": 1.3315, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7681, "step": 3890 }, { - "epoch": 0.53, - "learning_rate": 9.449479237116192e-06, - "loss": 1.3023, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7641, "step": 3900 }, { - "epoch": 0.53, - "learning_rate": 9.422426619775464e-06, - "loss": 1.3177, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7553, "step": 3910 }, { - "epoch": 0.53, - "learning_rate": 9.395374002434736e-06, - "loss": 1.3046, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7591, "step": 3920 }, { - "epoch": 0.53, - "learning_rate": 9.36832138509401e-06, - "loss": 1.2978, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7863, "step": 3930 }, { - "epoch": 0.53, - "learning_rate": 9.341268767753281e-06, - "loss": 1.4031, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7862, "step": 3940 }, { - "epoch": 0.53, - "learning_rate": 9.314216150412553e-06, - "loss": 1.3274, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7807, "step": 3950 }, { - "epoch": 0.54, - "learning_rate": 9.287163533071825e-06, - "loss": 1.3498, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.777, "step": 3960 }, { - "epoch": 0.54, - "learning_rate": 9.260110915731097e-06, - "loss": 1.3236, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7829, "step": 3970 }, { - "epoch": 0.54, - "learning_rate": 9.233058298390371e-06, - "loss": 1.2199, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7952, "step": 3980 }, { - "epoch": 0.54, - "learning_rate": 9.206005681049643e-06, - "loss": 1.2987, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7829, "step": 3990 }, { - "epoch": 0.54, - "learning_rate": 9.178953063708913e-06, - "loss": 1.3872, + "epoch": 0.12, + "learning_rate": 1e-05, + "loss": 0.7675, "step": 4000 - }, - { - "epoch": 0.54, - "learning_rate": 9.151900446368187e-06, - "loss": 1.3344, - "step": 4010 - }, - { - "epoch": 0.54, - "learning_rate": 9.124847829027459e-06, - "loss": 1.3608, - "step": 4020 - }, - { - "epoch": 0.55, - "learning_rate": 9.097795211686731e-06, - "loss": 1.352, - "step": 4030 - }, - { - "epoch": 0.55, - "learning_rate": 9.070742594346003e-06, - "loss": 1.3153, - "step": 4040 - }, - { - "epoch": 0.55, - "learning_rate": 9.043689977005275e-06, - "loss": 1.3213, - "step": 4050 - }, - { - "epoch": 0.55, - "learning_rate": 9.016637359664549e-06, - "loss": 1.3446, - "step": 4060 - }, - { - "epoch": 0.55, - "learning_rate": 8.989584742323821e-06, - "loss": 1.2776, - "step": 4070 - }, - { - "epoch": 0.55, - "learning_rate": 8.962532124983093e-06, - "loss": 1.2492, - "step": 4080 - }, - { - "epoch": 0.55, - "learning_rate": 8.935479507642365e-06, - "loss": 1.3352, - "step": 4090 - }, - { - "epoch": 0.55, - "learning_rate": 8.908426890301637e-06, - "loss": 1.4349, - "step": 4100 - }, - { - "epoch": 0.56, - "learning_rate": 8.88137427296091e-06, - "loss": 1.2922, - "step": 4110 - }, - { - "epoch": 0.56, - "learning_rate": 8.854321655620181e-06, - "loss": 1.2631, - "step": 4120 - }, - { - "epoch": 0.56, - "learning_rate": 8.827269038279455e-06, - "loss": 1.3137, - "step": 4130 - }, - { - "epoch": 0.56, - "learning_rate": 8.800216420938727e-06, - "loss": 1.4448, - "step": 4140 - }, - { - "epoch": 0.56, - "learning_rate": 8.773163803597999e-06, - "loss": 1.2777, - "step": 4150 - }, - { - "epoch": 0.56, - "learning_rate": 8.74611118625727e-06, - "loss": 1.3405, - "step": 4160 - }, - { - "epoch": 0.56, - "learning_rate": 8.719058568916543e-06, - "loss": 1.3843, - "step": 4170 - }, - { - "epoch": 0.57, - "learning_rate": 8.692005951575816e-06, - "loss": 1.2743, - "step": 4180 - }, - { - "epoch": 0.57, - "learning_rate": 8.664953334235088e-06, - "loss": 1.282, - "step": 4190 - }, - { - "epoch": 0.57, - "learning_rate": 8.63790071689436e-06, - "loss": 1.2945, - "step": 4200 - }, - { - "epoch": 0.57, - "learning_rate": 8.610848099553632e-06, - "loss": 1.3465, - "step": 4210 - }, - { - "epoch": 0.57, - "learning_rate": 8.583795482212904e-06, - "loss": 1.4186, - "step": 4220 - }, - { - "epoch": 0.57, - "learning_rate": 8.556742864872177e-06, - "loss": 1.3309, - "step": 4230 - }, - { - "epoch": 0.57, - "learning_rate": 8.529690247531449e-06, - "loss": 1.3802, - "step": 4240 - }, - { - "epoch": 0.57, - "learning_rate": 8.502637630190722e-06, - "loss": 1.4322, - "step": 4250 - }, - { - "epoch": 0.58, - "learning_rate": 8.475585012849994e-06, - "loss": 1.3568, - "step": 4260 - }, - { - "epoch": 0.58, - "learning_rate": 8.448532395509266e-06, - "loss": 1.1623, - "step": 4270 - }, - { - "epoch": 0.58, - "learning_rate": 8.421479778168538e-06, - "loss": 1.3236, - "step": 4280 - }, - { - "epoch": 0.58, - "learning_rate": 8.39442716082781e-06, - "loss": 1.3883, - "step": 4290 - }, - { - "epoch": 0.58, - "learning_rate": 8.367374543487084e-06, - "loss": 1.2427, - "step": 4300 - }, - { - "epoch": 0.58, - "learning_rate": 8.340321926146356e-06, - "loss": 1.3012, - "step": 4310 - }, - { - "epoch": 0.58, - "learning_rate": 8.313269308805626e-06, - "loss": 1.4305, - "step": 4320 - }, - { - "epoch": 0.59, - "learning_rate": 8.2862166914649e-06, - "loss": 1.4563, - "step": 4330 - }, - { - "epoch": 0.59, - "learning_rate": 8.259164074124172e-06, - "loss": 1.3257, - "step": 4340 - }, - { - "epoch": 0.59, - "learning_rate": 8.232111456783444e-06, - "loss": 1.2835, - "step": 4350 - }, - { - "epoch": 0.59, - "learning_rate": 8.205058839442716e-06, - "loss": 1.3389, - "step": 4360 - }, - { - "epoch": 0.59, - "learning_rate": 8.17800622210199e-06, - "loss": 1.4013, - "step": 4370 - }, - { - "epoch": 0.59, - "learning_rate": 8.150953604761262e-06, - "loss": 1.3558, - "step": 4380 - }, - { - "epoch": 0.59, - "learning_rate": 8.123900987420534e-06, - "loss": 1.3387, - "step": 4390 - }, - { - "epoch": 0.6, - "learning_rate": 8.096848370079806e-06, - "loss": 1.3122, - "step": 4400 - }, - { - "epoch": 0.6, - "learning_rate": 8.069795752739078e-06, - "loss": 1.3243, - "step": 4410 - }, - { - "epoch": 0.6, - "learning_rate": 8.042743135398352e-06, - "loss": 1.3679, - "step": 4420 - }, - { - "epoch": 0.6, - "learning_rate": 8.015690518057624e-06, - "loss": 1.2976, - "step": 4430 - }, - { - "epoch": 0.6, - "learning_rate": 7.988637900716894e-06, - "loss": 1.3655, - "step": 4440 - }, - { - "epoch": 0.6, - "learning_rate": 7.961585283376168e-06, - "loss": 1.2666, - "step": 4450 - }, - { - "epoch": 0.6, - "learning_rate": 7.93453266603544e-06, - "loss": 1.2528, - "step": 4460 - }, - { - "epoch": 0.6, - "learning_rate": 7.907480048694712e-06, - "loss": 1.3767, - "step": 4470 - }, - { - "epoch": 0.61, - "learning_rate": 7.880427431353984e-06, - "loss": 1.3033, - "step": 4480 - }, - { - "epoch": 0.61, - "learning_rate": 7.853374814013256e-06, - "loss": 1.2887, - "step": 4490 - }, - { - "epoch": 0.61, - "learning_rate": 7.82632219667253e-06, - "loss": 1.3784, - "step": 4500 - }, - { - "epoch": 0.61, - "learning_rate": 7.799269579331801e-06, - "loss": 1.3615, - "step": 4510 - }, - { - "epoch": 0.61, - "learning_rate": 7.772216961991073e-06, - "loss": 1.2267, - "step": 4520 - }, - { - "epoch": 0.61, - "learning_rate": 7.745164344650345e-06, - "loss": 1.2735, - "step": 4530 - }, - { - "epoch": 0.61, - "learning_rate": 7.718111727309617e-06, - "loss": 1.2746, - "step": 4540 - }, - { - "epoch": 0.62, - "learning_rate": 7.691059109968891e-06, - "loss": 1.2991, - "step": 4550 - }, - { - "epoch": 0.62, - "learning_rate": 7.664006492628161e-06, - "loss": 1.3383, - "step": 4560 - }, - { - "epoch": 0.62, - "learning_rate": 7.636953875287435e-06, - "loss": 1.3418, - "step": 4570 - }, - { - "epoch": 0.62, - "learning_rate": 7.609901257946707e-06, - "loss": 1.3049, - "step": 4580 - }, - { - "epoch": 0.62, - "learning_rate": 7.582848640605978e-06, - "loss": 1.2943, - "step": 4590 - }, - { - "epoch": 0.62, - "learning_rate": 7.555796023265252e-06, - "loss": 1.2644, - "step": 4600 - }, - { - "epoch": 0.62, - "learning_rate": 7.528743405924523e-06, - "loss": 1.3723, - "step": 4610 - }, - { - "epoch": 0.62, - "learning_rate": 7.501690788583796e-06, - "loss": 1.2647, - "step": 4620 - }, - { - "epoch": 0.63, - "learning_rate": 7.474638171243068e-06, - "loss": 1.3336, - "step": 4630 - }, - { - "epoch": 0.63, - "learning_rate": 7.447585553902341e-06, - "loss": 1.355, - "step": 4640 - }, - { - "epoch": 0.63, - "learning_rate": 7.420532936561613e-06, - "loss": 1.2595, - "step": 4650 - }, - { - "epoch": 0.63, - "learning_rate": 7.393480319220885e-06, - "loss": 1.3246, - "step": 4660 - }, - { - "epoch": 0.63, - "learning_rate": 7.366427701880158e-06, - "loss": 1.2293, - "step": 4670 - }, - { - "epoch": 0.63, - "learning_rate": 7.33937508453943e-06, - "loss": 1.3004, - "step": 4680 - }, - { - "epoch": 0.63, - "learning_rate": 7.312322467198703e-06, - "loss": 1.3278, - "step": 4690 - }, - { - "epoch": 0.64, - "learning_rate": 7.285269849857975e-06, - "loss": 1.3041, - "step": 4700 - }, - { - "epoch": 0.64, - "learning_rate": 7.258217232517246e-06, - "loss": 1.3277, - "step": 4710 - }, - { - "epoch": 0.64, - "learning_rate": 7.2311646151765195e-06, - "loss": 1.3714, - "step": 4720 - }, - { - "epoch": 0.64, - "learning_rate": 7.204111997835791e-06, - "loss": 1.3716, - "step": 4730 - }, - { - "epoch": 0.64, - "learning_rate": 7.1770593804950635e-06, - "loss": 1.2244, - "step": 4740 - }, - { - "epoch": 0.64, - "learning_rate": 7.1500067631543356e-06, - "loss": 1.2683, - "step": 4750 - }, - { - "epoch": 0.64, - "learning_rate": 7.1229541458136076e-06, - "loss": 1.3482, - "step": 4760 - }, - { - "epoch": 0.65, - "learning_rate": 7.0959015284728804e-06, - "loss": 1.3238, - "step": 4770 - }, - { - "epoch": 0.65, - "learning_rate": 7.0688489111321524e-06, - "loss": 1.3791, - "step": 4780 - }, - { - "epoch": 0.65, - "learning_rate": 7.041796293791425e-06, - "loss": 1.2705, - "step": 4790 - }, - { - "epoch": 0.65, - "learning_rate": 7.014743676450697e-06, - "loss": 1.3447, - "step": 4800 - }, - { - "epoch": 0.65, - "learning_rate": 6.987691059109969e-06, - "loss": 1.3656, - "step": 4810 - }, - { - "epoch": 0.65, - "learning_rate": 6.960638441769242e-06, - "loss": 1.3313, - "step": 4820 - }, - { - "epoch": 0.65, - "learning_rate": 6.933585824428513e-06, - "loss": 1.2603, - "step": 4830 - }, - { - "epoch": 0.65, - "learning_rate": 6.906533207087786e-06, - "loss": 1.2819, - "step": 4840 - }, - { - "epoch": 0.66, - "learning_rate": 6.879480589747058e-06, - "loss": 1.3329, - "step": 4850 - }, - { - "epoch": 0.66, - "learning_rate": 6.852427972406331e-06, - "loss": 1.239, - "step": 4860 - }, - { - "epoch": 0.66, - "learning_rate": 6.825375355065603e-06, - "loss": 1.248, - "step": 4870 - }, - { - "epoch": 0.66, - "learning_rate": 6.798322737724875e-06, - "loss": 1.3815, - "step": 4880 - }, - { - "epoch": 0.66, - "learning_rate": 6.771270120384148e-06, - "loss": 1.3893, - "step": 4890 - }, - { - "epoch": 0.66, - "learning_rate": 6.74421750304342e-06, - "loss": 1.372, - "step": 4900 - }, - { - "epoch": 0.66, - "learning_rate": 6.717164885702693e-06, - "loss": 1.2889, - "step": 4910 - }, - { - "epoch": 0.67, - "learning_rate": 6.690112268361965e-06, - "loss": 1.304, - "step": 4920 - }, - { - "epoch": 0.67, - "learning_rate": 6.663059651021236e-06, - "loss": 1.2241, - "step": 4930 - }, - { - "epoch": 0.67, - "learning_rate": 6.63600703368051e-06, - "loss": 1.3402, - "step": 4940 - }, - { - "epoch": 0.67, - "learning_rate": 6.608954416339781e-06, - "loss": 1.2631, - "step": 4950 - }, - { - "epoch": 0.67, - "learning_rate": 6.581901798999054e-06, - "loss": 1.2667, - "step": 4960 - }, - { - "epoch": 0.67, - "learning_rate": 6.554849181658326e-06, - "loss": 1.3114, - "step": 4970 - }, - { - "epoch": 0.67, - "learning_rate": 6.527796564317598e-06, - "loss": 1.3055, - "step": 4980 - }, - { - "epoch": 0.67, - "learning_rate": 6.500743946976871e-06, - "loss": 1.3026, - "step": 4990 - }, - { - "epoch": 0.68, - "learning_rate": 6.473691329636143e-06, - "loss": 1.3487, - "step": 5000 - }, - { - "epoch": 0.68, - "learning_rate": 6.4466387122954155e-06, - "loss": 1.2515, - "step": 5010 - }, - { - "epoch": 0.68, - "learning_rate": 6.4195860949546875e-06, - "loss": 1.2152, - "step": 5020 - }, - { - "epoch": 0.68, - "learning_rate": 6.3925334776139595e-06, - "loss": 1.4006, - "step": 5030 - }, - { - "epoch": 0.68, - "learning_rate": 6.365480860273232e-06, - "loss": 1.2256, - "step": 5040 - }, - { - "epoch": 0.68, - "learning_rate": 6.3384282429325036e-06, - "loss": 1.3478, - "step": 5050 - }, - { - "epoch": 0.68, - "learning_rate": 6.311375625591777e-06, - "loss": 1.3279, - "step": 5060 - }, - { - "epoch": 0.69, - "learning_rate": 6.2843230082510484e-06, - "loss": 1.3142, - "step": 5070 - }, - { - "epoch": 0.69, - "learning_rate": 6.257270390910321e-06, - "loss": 1.2511, - "step": 5080 - }, - { - "epoch": 0.69, - "learning_rate": 6.230217773569593e-06, - "loss": 1.2952, - "step": 5090 - }, - { - "epoch": 0.69, - "learning_rate": 6.203165156228865e-06, - "loss": 1.3287, - "step": 5100 - }, - { - "epoch": 0.69, - "learning_rate": 6.176112538888138e-06, - "loss": 1.2532, - "step": 5110 - }, - { - "epoch": 0.69, - "learning_rate": 6.14905992154741e-06, - "loss": 1.1895, - "step": 5120 - }, - { - "epoch": 0.69, - "learning_rate": 6.122007304206683e-06, - "loss": 1.309, - "step": 5130 - }, - { - "epoch": 0.7, - "learning_rate": 6.094954686865955e-06, - "loss": 1.348, - "step": 5140 - }, - { - "epoch": 0.7, - "learning_rate": 6.067902069525226e-06, - "loss": 1.2705, - "step": 5150 - }, - { - "epoch": 0.7, - "learning_rate": 6.0408494521845e-06, - "loss": 1.3303, - "step": 5160 - }, - { - "epoch": 0.7, - "learning_rate": 6.013796834843771e-06, - "loss": 1.2834, - "step": 5170 - }, - { - "epoch": 0.7, - "learning_rate": 5.986744217503044e-06, - "loss": 1.2206, - "step": 5180 - }, - { - "epoch": 0.7, - "learning_rate": 5.959691600162316e-06, - "loss": 1.3875, - "step": 5190 - }, - { - "epoch": 0.7, - "learning_rate": 5.932638982821588e-06, - "loss": 1.2656, - "step": 5200 - }, - { - "epoch": 0.7, - "learning_rate": 5.905586365480861e-06, - "loss": 1.3478, - "step": 5210 - }, - { - "epoch": 0.71, - "learning_rate": 5.878533748140133e-06, - "loss": 1.3352, - "step": 5220 - }, - { - "epoch": 0.71, - "learning_rate": 5.851481130799406e-06, - "loss": 1.2967, - "step": 5230 - }, - { - "epoch": 0.71, - "learning_rate": 5.824428513458678e-06, - "loss": 1.2273, - "step": 5240 - }, - { - "epoch": 0.71, - "learning_rate": 5.79737589611795e-06, - "loss": 1.2247, - "step": 5250 - }, - { - "epoch": 0.71, - "learning_rate": 5.770323278777223e-06, - "loss": 1.3172, - "step": 5260 - }, - { - "epoch": 0.71, - "learning_rate": 5.743270661436494e-06, - "loss": 1.2982, - "step": 5270 - }, - { - "epoch": 0.71, - "learning_rate": 5.7162180440957675e-06, - "loss": 1.2272, - "step": 5280 - }, - { - "epoch": 0.72, - "learning_rate": 5.689165426755039e-06, - "loss": 1.2338, - "step": 5290 - }, - { - "epoch": 0.72, - "learning_rate": 5.6621128094143115e-06, - "loss": 1.2788, - "step": 5300 - }, - { - "epoch": 0.72, - "learning_rate": 5.6350601920735835e-06, - "loss": 1.2713, - "step": 5310 - }, - { - "epoch": 0.72, - "learning_rate": 5.6080075747328555e-06, - "loss": 1.3878, - "step": 5320 - }, - { - "epoch": 0.72, - "learning_rate": 5.580954957392128e-06, - "loss": 1.3369, - "step": 5330 - }, - { - "epoch": 0.72, - "learning_rate": 5.5539023400514e-06, - "loss": 1.3504, - "step": 5340 - }, - { - "epoch": 0.72, - "learning_rate": 5.526849722710673e-06, - "loss": 1.3086, - "step": 5350 - }, - { - "epoch": 0.73, - "learning_rate": 5.499797105369945e-06, - "loss": 1.2101, - "step": 5360 - }, - { - "epoch": 0.73, - "learning_rate": 5.472744488029217e-06, - "loss": 1.2773, - "step": 5370 - }, - { - "epoch": 0.73, - "learning_rate": 5.44569187068849e-06, - "loss": 1.1808, - "step": 5380 - }, - { - "epoch": 0.73, - "learning_rate": 5.418639253347761e-06, - "loss": 1.4258, - "step": 5390 - }, - { - "epoch": 0.73, - "learning_rate": 5.391586636007034e-06, - "loss": 1.2997, - "step": 5400 - }, - { - "epoch": 0.73, - "learning_rate": 5.364534018666306e-06, - "loss": 1.3596, - "step": 5410 - }, - { - "epoch": 0.73, - "learning_rate": 5.337481401325578e-06, - "loss": 1.324, - "step": 5420 - }, - { - "epoch": 0.73, - "learning_rate": 5.310428783984851e-06, - "loss": 1.3513, - "step": 5430 - }, - { - "epoch": 0.74, - "learning_rate": 5.283376166644123e-06, - "loss": 1.2955, - "step": 5440 - }, - { - "epoch": 0.74, - "learning_rate": 5.256323549303396e-06, - "loss": 1.3214, - "step": 5450 - }, - { - "epoch": 0.74, - "learning_rate": 5.229270931962668e-06, - "loss": 1.2305, - "step": 5460 - }, - { - "epoch": 0.74, - "learning_rate": 5.20221831462194e-06, - "loss": 1.2032, - "step": 5470 - }, - { - "epoch": 0.74, - "learning_rate": 5.175165697281213e-06, - "loss": 1.2551, - "step": 5480 - }, - { - "epoch": 0.74, - "learning_rate": 5.148113079940484e-06, - "loss": 1.2703, - "step": 5490 - }, - { - "epoch": 0.74, - "learning_rate": 5.121060462599758e-06, - "loss": 1.3463, - "step": 5500 - }, - { - "epoch": 0.75, - "learning_rate": 5.094007845259029e-06, - "loss": 1.2445, - "step": 5510 - }, - { - "epoch": 0.75, - "learning_rate": 5.066955227918301e-06, - "loss": 1.3853, - "step": 5520 - }, - { - "epoch": 0.75, - "learning_rate": 5.039902610577574e-06, - "loss": 1.2805, - "step": 5530 - }, - { - "epoch": 0.75, - "learning_rate": 5.012849993236846e-06, - "loss": 1.2073, - "step": 5540 - }, - { - "epoch": 0.75, - "learning_rate": 4.985797375896118e-06, - "loss": 1.2761, - "step": 5550 - }, - { - "epoch": 0.75, - "learning_rate": 4.958744758555391e-06, - "loss": 1.3558, - "step": 5560 - }, - { - "epoch": 0.75, - "learning_rate": 4.931692141214663e-06, - "loss": 1.3822, - "step": 5570 - }, - { - "epoch": 0.75, - "learning_rate": 4.9046395238739355e-06, - "loss": 1.3899, - "step": 5580 - }, - { - "epoch": 0.76, - "learning_rate": 4.8775869065332075e-06, - "loss": 1.2224, - "step": 5590 - }, - { - "epoch": 0.76, - "learning_rate": 4.85053428919248e-06, - "loss": 1.3231, - "step": 5600 - }, - { - "epoch": 0.76, - "learning_rate": 4.8234816718517515e-06, - "loss": 1.3951, - "step": 5610 - }, - { - "epoch": 0.76, - "learning_rate": 4.796429054511024e-06, - "loss": 1.2355, - "step": 5620 - }, - { - "epoch": 0.76, - "learning_rate": 4.769376437170296e-06, - "loss": 1.2818, - "step": 5630 - }, - { - "epoch": 0.76, - "learning_rate": 4.742323819829569e-06, - "loss": 1.4235, - "step": 5640 - }, - { - "epoch": 0.76, - "learning_rate": 4.715271202488841e-06, - "loss": 1.241, - "step": 5650 - }, - { - "epoch": 0.77, - "learning_rate": 4.688218585148113e-06, - "loss": 1.3052, - "step": 5660 - }, - { - "epoch": 0.77, - "learning_rate": 4.661165967807385e-06, - "loss": 1.2379, - "step": 5670 - }, - { - "epoch": 0.77, - "learning_rate": 4.634113350466658e-06, - "loss": 1.1818, - "step": 5680 - }, - { - "epoch": 0.77, - "learning_rate": 4.60706073312593e-06, - "loss": 1.3041, - "step": 5690 - }, - { - "epoch": 0.77, - "learning_rate": 4.580008115785203e-06, - "loss": 1.2873, - "step": 5700 - }, - { - "epoch": 0.77, - "learning_rate": 4.552955498444475e-06, - "loss": 1.2344, - "step": 5710 - }, - { - "epoch": 0.77, - "learning_rate": 4.525902881103747e-06, - "loss": 1.2907, - "step": 5720 - }, - { - "epoch": 0.78, - "learning_rate": 4.498850263763019e-06, - "loss": 1.2444, - "step": 5730 - }, - { - "epoch": 0.78, - "learning_rate": 4.471797646422292e-06, - "loss": 1.2851, - "step": 5740 - }, - { - "epoch": 0.78, - "learning_rate": 4.444745029081564e-06, - "loss": 1.261, - "step": 5750 - }, - { - "epoch": 0.78, - "learning_rate": 4.417692411740837e-06, - "loss": 1.2401, - "step": 5760 - }, - { - "epoch": 0.78, - "learning_rate": 4.390639794400108e-06, - "loss": 1.2664, - "step": 5770 - }, - { - "epoch": 0.78, - "learning_rate": 4.363587177059381e-06, - "loss": 1.2852, - "step": 5780 - }, - { - "epoch": 0.78, - "learning_rate": 4.336534559718653e-06, - "loss": 1.2273, - "step": 5790 - }, - { - "epoch": 0.78, - "learning_rate": 4.309481942377926e-06, - "loss": 1.235, - "step": 5800 - }, - { - "epoch": 0.79, - "learning_rate": 4.282429325037198e-06, - "loss": 1.2836, - "step": 5810 - }, - { - "epoch": 0.79, - "learning_rate": 4.25537670769647e-06, - "loss": 1.2458, - "step": 5820 - }, - { - "epoch": 0.79, - "learning_rate": 4.228324090355742e-06, - "loss": 1.2887, - "step": 5830 - }, - { - "epoch": 0.79, - "learning_rate": 4.201271473015015e-06, - "loss": 1.3712, - "step": 5840 - }, - { - "epoch": 0.79, - "learning_rate": 4.174218855674287e-06, - "loss": 1.2536, - "step": 5850 - }, - { - "epoch": 0.79, - "learning_rate": 4.1471662383335595e-06, - "loss": 1.2423, - "step": 5860 - }, - { - "epoch": 0.79, - "learning_rate": 4.1201136209928315e-06, - "loss": 1.3013, - "step": 5870 - }, - { - "epoch": 0.8, - "learning_rate": 4.0930610036521035e-06, - "loss": 1.2787, - "step": 5880 - }, - { - "epoch": 0.8, - "learning_rate": 4.0660083863113755e-06, - "loss": 1.2041, - "step": 5890 - }, - { - "epoch": 0.8, - "learning_rate": 4.038955768970648e-06, - "loss": 1.2638, - "step": 5900 - }, - { - "epoch": 0.8, - "learning_rate": 4.01190315162992e-06, - "loss": 1.355, - "step": 5910 - }, - { - "epoch": 0.8, - "learning_rate": 3.984850534289193e-06, - "loss": 1.4058, - "step": 5920 - }, - { - "epoch": 0.8, - "learning_rate": 3.957797916948465e-06, - "loss": 1.2593, - "step": 5930 - }, - { - "epoch": 0.8, - "learning_rate": 3.930745299607737e-06, - "loss": 1.2869, - "step": 5940 - }, - { - "epoch": 0.8, - "learning_rate": 3.903692682267009e-06, - "loss": 1.278, - "step": 5950 - }, - { - "epoch": 0.81, - "learning_rate": 3.876640064926282e-06, - "loss": 1.2861, - "step": 5960 - }, - { - "epoch": 0.81, - "learning_rate": 3.849587447585554e-06, - "loss": 1.3489, - "step": 5970 - }, - { - "epoch": 0.81, - "learning_rate": 3.822534830244827e-06, - "loss": 1.2948, - "step": 5980 - }, - { - "epoch": 0.81, - "learning_rate": 3.7954822129040986e-06, - "loss": 1.2842, - "step": 5990 - }, - { - "epoch": 0.81, - "learning_rate": 3.768429595563371e-06, - "loss": 1.2563, - "step": 6000 - }, - { - "epoch": 0.81, - "learning_rate": 3.7413769782226435e-06, - "loss": 1.278, - "step": 6010 - }, - { - "epoch": 0.81, - "learning_rate": 3.714324360881916e-06, - "loss": 1.2912, - "step": 6020 - }, - { - "epoch": 0.82, - "learning_rate": 3.687271743541188e-06, - "loss": 1.2033, - "step": 6030 - }, - { - "epoch": 0.82, - "learning_rate": 3.66021912620046e-06, - "loss": 1.4748, - "step": 6040 - }, - { - "epoch": 0.82, - "learning_rate": 3.6331665088597324e-06, - "loss": 1.2688, - "step": 6050 - }, - { - "epoch": 0.82, - "learning_rate": 3.606113891519005e-06, - "loss": 1.2803, - "step": 6060 - }, - { - "epoch": 0.82, - "learning_rate": 3.5790612741782772e-06, - "loss": 1.3041, - "step": 6070 - }, - { - "epoch": 0.82, - "learning_rate": 3.5520086568375497e-06, - "loss": 1.1707, - "step": 6080 - }, - { - "epoch": 0.82, - "learning_rate": 3.5249560394968217e-06, - "loss": 1.2963, - "step": 6090 - }, - { - "epoch": 0.83, - "learning_rate": 3.4979034221560937e-06, - "loss": 1.2933, - "step": 6100 - }, - { - "epoch": 0.83, - "learning_rate": 3.470850804815366e-06, - "loss": 1.2344, - "step": 6110 - }, - { - "epoch": 0.83, - "learning_rate": 3.4437981874746386e-06, - "loss": 1.1929, - "step": 6120 - }, - { - "epoch": 0.83, - "learning_rate": 3.416745570133911e-06, - "loss": 1.3239, - "step": 6130 - }, - { - "epoch": 0.83, - "learning_rate": 3.389692952793183e-06, - "loss": 1.2799, - "step": 6140 - }, - { - "epoch": 0.83, - "learning_rate": 3.362640335452455e-06, - "loss": 1.3147, - "step": 6150 - }, - { - "epoch": 0.83, - "learning_rate": 3.3355877181117275e-06, - "loss": 1.2602, - "step": 6160 - }, - { - "epoch": 0.83, - "learning_rate": 3.308535100771e-06, - "loss": 1.3501, - "step": 6170 - }, - { - "epoch": 0.84, - "learning_rate": 3.2814824834302723e-06, - "loss": 1.316, - "step": 6180 - }, - { - "epoch": 0.84, - "learning_rate": 3.2544298660895448e-06, - "loss": 1.3737, - "step": 6190 - }, - { - "epoch": 0.84, - "learning_rate": 3.2273772487488168e-06, - "loss": 1.2829, - "step": 6200 - }, - { - "epoch": 0.84, - "learning_rate": 3.200324631408089e-06, - "loss": 1.3458, - "step": 6210 - }, - { - "epoch": 0.84, - "learning_rate": 3.1732720140673612e-06, - "loss": 1.2351, - "step": 6220 - }, - { - "epoch": 0.84, - "learning_rate": 3.1462193967266337e-06, - "loss": 1.2876, - "step": 6230 - }, - { - "epoch": 0.84, - "learning_rate": 3.119166779385906e-06, - "loss": 1.2534, - "step": 6240 - }, - { - "epoch": 0.85, - "learning_rate": 3.0921141620451785e-06, - "loss": 1.2838, - "step": 6250 - }, - { - "epoch": 0.85, - "learning_rate": 3.06506154470445e-06, - "loss": 1.326, - "step": 6260 - }, - { - "epoch": 0.85, - "learning_rate": 3.0380089273637226e-06, - "loss": 1.3398, - "step": 6270 - }, - { - "epoch": 0.85, - "learning_rate": 3.010956310022995e-06, - "loss": 1.1882, - "step": 6280 - }, - { - "epoch": 0.85, - "learning_rate": 2.9839036926822674e-06, - "loss": 1.3127, - "step": 6290 - }, - { - "epoch": 0.85, - "learning_rate": 2.95685107534154e-06, - "loss": 1.268, - "step": 6300 - }, - { - "epoch": 0.85, - "learning_rate": 2.929798458000812e-06, - "loss": 1.2426, - "step": 6310 - }, - { - "epoch": 0.85, - "learning_rate": 2.902745840660084e-06, - "loss": 1.228, - "step": 6320 - }, - { - "epoch": 0.86, - "learning_rate": 2.8756932233193563e-06, - "loss": 1.2987, - "step": 6330 - }, - { - "epoch": 0.86, - "learning_rate": 2.8486406059786288e-06, - "loss": 1.2554, - "step": 6340 - }, - { - "epoch": 0.86, - "learning_rate": 2.821587988637901e-06, - "loss": 1.1762, - "step": 6350 - }, - { - "epoch": 0.86, - "learning_rate": 2.7945353712971736e-06, - "loss": 1.3805, - "step": 6360 - }, - { - "epoch": 0.86, - "learning_rate": 2.7674827539564452e-06, - "loss": 1.3917, - "step": 6370 - }, - { - "epoch": 0.86, - "learning_rate": 2.7404301366157177e-06, - "loss": 1.2458, - "step": 6380 - }, - { - "epoch": 0.86, - "learning_rate": 2.71337751927499e-06, - "loss": 1.3998, - "step": 6390 - }, - { - "epoch": 0.87, - "learning_rate": 2.6863249019342625e-06, - "loss": 1.2591, - "step": 6400 - }, - { - "epoch": 0.87, - "learning_rate": 2.659272284593535e-06, - "loss": 1.3082, - "step": 6410 - }, - { - "epoch": 0.87, - "learning_rate": 2.632219667252807e-06, - "loss": 1.3033, - "step": 6420 - }, - { - "epoch": 0.87, - "learning_rate": 2.605167049912079e-06, - "loss": 1.3704, - "step": 6430 - }, - { - "epoch": 0.87, - "learning_rate": 2.5781144325713514e-06, - "loss": 1.2005, - "step": 6440 - }, - { - "epoch": 0.87, - "learning_rate": 2.551061815230624e-06, - "loss": 1.3414, - "step": 6450 - }, - { - "epoch": 0.87, - "learning_rate": 2.5240091978898963e-06, - "loss": 1.2884, - "step": 6460 - }, - { - "epoch": 0.88, - "learning_rate": 2.4969565805491683e-06, - "loss": 1.271, - "step": 6470 - }, - { - "epoch": 0.88, - "learning_rate": 2.4699039632084408e-06, - "loss": 1.2048, - "step": 6480 - }, - { - "epoch": 0.88, - "learning_rate": 2.4428513458677128e-06, - "loss": 1.2517, - "step": 6490 - }, - { - "epoch": 0.88, - "learning_rate": 2.4157987285269852e-06, - "loss": 1.2662, - "step": 6500 - }, - { - "epoch": 0.88, - "learning_rate": 2.3887461111862576e-06, - "loss": 1.3133, - "step": 6510 - }, - { - "epoch": 0.88, - "learning_rate": 2.3616934938455297e-06, - "loss": 1.2662, - "step": 6520 - }, - { - "epoch": 0.88, - "learning_rate": 2.334640876504802e-06, - "loss": 1.3612, - "step": 6530 - }, - { - "epoch": 0.88, - "learning_rate": 2.3075882591640745e-06, - "loss": 1.1793, - "step": 6540 - }, - { - "epoch": 0.89, - "learning_rate": 2.2805356418233465e-06, - "loss": 1.2251, - "step": 6550 - }, - { - "epoch": 0.89, - "learning_rate": 2.253483024482619e-06, - "loss": 1.2811, - "step": 6560 - }, - { - "epoch": 0.89, - "learning_rate": 2.2264304071418914e-06, - "loss": 1.3258, - "step": 6570 - }, - { - "epoch": 0.89, - "learning_rate": 2.1993777898011634e-06, - "loss": 1.3123, - "step": 6580 - }, - { - "epoch": 0.89, - "learning_rate": 2.172325172460436e-06, - "loss": 1.3606, - "step": 6590 - }, - { - "epoch": 0.89, - "learning_rate": 2.145272555119708e-06, - "loss": 1.213, - "step": 6600 - }, - { - "epoch": 0.89, - "learning_rate": 2.1182199377789803e-06, - "loss": 1.3343, - "step": 6610 - }, - { - "epoch": 0.9, - "learning_rate": 2.0911673204382528e-06, - "loss": 1.3229, - "step": 6620 - }, - { - "epoch": 0.9, - "learning_rate": 2.0641147030975248e-06, - "loss": 1.3807, - "step": 6630 - }, - { - "epoch": 0.9, - "learning_rate": 2.037062085756797e-06, - "loss": 1.3573, - "step": 6640 - }, - { - "epoch": 0.9, - "learning_rate": 2.0100094684160696e-06, - "loss": 1.2112, - "step": 6650 - }, - { - "epoch": 0.9, - "learning_rate": 1.9829568510753417e-06, - "loss": 1.302, - "step": 6660 - }, - { - "epoch": 0.9, - "learning_rate": 1.955904233734614e-06, - "loss": 1.1786, - "step": 6670 - }, - { - "epoch": 0.9, - "learning_rate": 1.928851616393886e-06, - "loss": 1.3572, - "step": 6680 - }, - { - "epoch": 0.9, - "learning_rate": 1.9017989990531585e-06, - "loss": 1.3131, - "step": 6690 - }, - { - "epoch": 0.91, - "learning_rate": 1.874746381712431e-06, - "loss": 1.249, - "step": 6700 - }, - { - "epoch": 0.91, - "learning_rate": 1.847693764371703e-06, - "loss": 1.2392, - "step": 6710 - }, - { - "epoch": 0.91, - "learning_rate": 1.8206411470309754e-06, - "loss": 1.2833, - "step": 6720 - }, - { - "epoch": 0.91, - "learning_rate": 1.7935885296902479e-06, - "loss": 1.2991, - "step": 6730 - }, - { - "epoch": 0.91, - "learning_rate": 1.7665359123495199e-06, - "loss": 1.3772, - "step": 6740 - }, - { - "epoch": 0.91, - "learning_rate": 1.7394832950087923e-06, - "loss": 1.2383, - "step": 6750 - }, - { - "epoch": 0.91, - "learning_rate": 1.7124306776680647e-06, - "loss": 1.3294, - "step": 6760 - }, - { - "epoch": 0.92, - "learning_rate": 1.6853780603273368e-06, - "loss": 1.2221, - "step": 6770 - }, - { - "epoch": 0.92, - "learning_rate": 1.6583254429866092e-06, - "loss": 1.2485, - "step": 6780 - }, - { - "epoch": 0.92, - "learning_rate": 1.6312728256458812e-06, - "loss": 1.2928, - "step": 6790 - }, - { - "epoch": 0.92, - "learning_rate": 1.6042202083051536e-06, - "loss": 1.3282, - "step": 6800 - }, - { - "epoch": 0.92, - "learning_rate": 1.577167590964426e-06, - "loss": 1.3335, - "step": 6810 - }, - { - "epoch": 0.92, - "learning_rate": 1.550114973623698e-06, - "loss": 1.3003, - "step": 6820 - }, - { - "epoch": 0.92, - "learning_rate": 1.5230623562829705e-06, - "loss": 1.1809, - "step": 6830 - }, - { - "epoch": 0.93, - "learning_rate": 1.496009738942243e-06, - "loss": 1.3068, - "step": 6840 - }, - { - "epoch": 0.93, - "learning_rate": 1.468957121601515e-06, - "loss": 1.2019, - "step": 6850 - }, - { - "epoch": 0.93, - "learning_rate": 1.4419045042607874e-06, - "loss": 1.2414, - "step": 6860 - }, - { - "epoch": 0.93, - "learning_rate": 1.4148518869200598e-06, - "loss": 1.3198, - "step": 6870 - }, - { - "epoch": 0.93, - "learning_rate": 1.3877992695793319e-06, - "loss": 1.1741, - "step": 6880 - }, - { - "epoch": 0.93, - "learning_rate": 1.3607466522386043e-06, - "loss": 1.2822, - "step": 6890 - }, - { - "epoch": 0.93, - "learning_rate": 1.3336940348978763e-06, - "loss": 1.2552, - "step": 6900 - }, - { - "epoch": 0.93, - "learning_rate": 1.3066414175571487e-06, - "loss": 1.325, - "step": 6910 - }, - { - "epoch": 0.94, - "learning_rate": 1.2795888002164212e-06, - "loss": 1.2532, - "step": 6920 - }, - { - "epoch": 0.94, - "learning_rate": 1.2525361828756932e-06, - "loss": 1.1989, - "step": 6930 - }, - { - "epoch": 0.94, - "learning_rate": 1.2254835655349656e-06, - "loss": 1.1572, - "step": 6940 - }, - { - "epoch": 0.94, - "learning_rate": 1.1984309481942379e-06, - "loss": 1.3367, - "step": 6950 - }, - { - "epoch": 0.94, - "learning_rate": 1.17137833085351e-06, - "loss": 1.2946, - "step": 6960 - }, - { - "epoch": 0.94, - "learning_rate": 1.1443257135127825e-06, - "loss": 1.2744, - "step": 6970 - }, - { - "epoch": 0.94, - "learning_rate": 1.1172730961720547e-06, - "loss": 1.2555, - "step": 6980 - }, - { - "epoch": 0.95, - "learning_rate": 1.090220478831327e-06, - "loss": 1.2799, - "step": 6990 - }, - { - "epoch": 0.95, - "learning_rate": 1.0631678614905994e-06, - "loss": 1.2302, - "step": 7000 - }, - { - "epoch": 0.95, - "learning_rate": 1.0361152441498716e-06, - "loss": 1.2972, - "step": 7010 - }, - { - "epoch": 0.95, - "learning_rate": 1.0090626268091438e-06, - "loss": 1.2348, - "step": 7020 - }, - { - "epoch": 0.95, - "learning_rate": 9.82010009468416e-07, - "loss": 1.2838, - "step": 7030 - }, - { - "epoch": 0.95, - "learning_rate": 9.549573921276885e-07, - "loss": 1.2531, - "step": 7040 - }, - { - "epoch": 0.95, - "learning_rate": 9.279047747869607e-07, - "loss": 1.3645, - "step": 7050 - }, - { - "epoch": 0.95, - "learning_rate": 9.00852157446233e-07, - "loss": 1.3312, - "step": 7060 - }, - { - "epoch": 0.96, - "learning_rate": 8.737995401055052e-07, - "loss": 1.2169, - "step": 7070 - }, - { - "epoch": 0.96, - "learning_rate": 8.467469227647776e-07, - "loss": 1.2424, - "step": 7080 - }, - { - "epoch": 0.96, - "learning_rate": 8.196943054240498e-07, - "loss": 1.3582, - "step": 7090 - }, - { - "epoch": 0.96, - "learning_rate": 7.926416880833221e-07, - "loss": 1.3237, - "step": 7100 - }, - { - "epoch": 0.96, - "learning_rate": 7.655890707425945e-07, - "loss": 1.3479, - "step": 7110 - }, - { - "epoch": 0.96, - "learning_rate": 7.385364534018667e-07, - "loss": 1.3575, - "step": 7120 - }, - { - "epoch": 0.96, - "learning_rate": 7.11483836061139e-07, - "loss": 1.2859, - "step": 7130 - }, - { - "epoch": 0.97, - "learning_rate": 6.844312187204112e-07, - "loss": 1.372, - "step": 7140 - }, - { - "epoch": 0.97, - "learning_rate": 6.573786013796836e-07, - "loss": 1.2061, - "step": 7150 - }, - { - "epoch": 0.97, - "learning_rate": 6.303259840389558e-07, - "loss": 1.237, - "step": 7160 - }, - { - "epoch": 0.97, - "learning_rate": 6.032733666982281e-07, - "loss": 1.1609, - "step": 7170 - }, - { - "epoch": 0.97, - "learning_rate": 5.762207493575004e-07, - "loss": 1.3026, - "step": 7180 - }, - { - "epoch": 0.97, - "learning_rate": 5.491681320167726e-07, - "loss": 1.3402, - "step": 7190 - }, - { - "epoch": 0.97, - "learning_rate": 5.221155146760449e-07, - "loss": 1.2512, - "step": 7200 - }, - { - "epoch": 0.98, - "learning_rate": 4.950628973353172e-07, - "loss": 1.2137, - "step": 7210 - }, - { - "epoch": 0.98, - "learning_rate": 4.680102799945895e-07, - "loss": 1.3718, - "step": 7220 - }, - { - "epoch": 0.98, - "learning_rate": 4.409576626538618e-07, - "loss": 1.1741, - "step": 7230 - }, - { - "epoch": 0.98, - "learning_rate": 4.1390504531313406e-07, - "loss": 1.2558, - "step": 7240 - }, - { - "epoch": 0.98, - "learning_rate": 3.868524279724064e-07, - "loss": 1.3625, - "step": 7250 - }, - { - "epoch": 0.98, - "learning_rate": 3.5979981063167866e-07, - "loss": 1.2939, - "step": 7260 - }, - { - "epoch": 0.98, - "learning_rate": 3.3274719329095094e-07, - "loss": 1.2041, - "step": 7270 - }, - { - "epoch": 0.98, - "learning_rate": 3.056945759502232e-07, - "loss": 1.2374, - "step": 7280 - }, - { - "epoch": 0.99, - "learning_rate": 2.786419586094955e-07, - "loss": 1.3662, - "step": 7290 - }, - { - "epoch": 0.99, - "learning_rate": 2.5158934126876777e-07, - "loss": 1.3344, - "step": 7300 - }, - { - "epoch": 0.99, - "learning_rate": 2.2453672392804008e-07, - "loss": 1.2437, - "step": 7310 - }, - { - "epoch": 0.99, - "learning_rate": 1.9748410658731235e-07, - "loss": 1.2795, - "step": 7320 - }, - { - "epoch": 0.99, - "learning_rate": 1.7043148924658463e-07, - "loss": 1.2735, - "step": 7330 - }, - { - "epoch": 0.99, - "learning_rate": 1.433788719058569e-07, - "loss": 1.223, - "step": 7340 - }, - { - "epoch": 0.99, - "learning_rate": 1.163262545651292e-07, - "loss": 1.3147, - "step": 7350 - }, - { - "epoch": 1.0, - "learning_rate": 8.927363722440148e-08, - "loss": 1.3607, - "step": 7360 - }, - { - "epoch": 1.0, - "learning_rate": 6.222101988367375e-08, - "loss": 1.347, - "step": 7370 - }, - { - "epoch": 1.0, - "learning_rate": 3.516840254294603e-08, - "loss": 1.3362, - "step": 7380 - }, - { - "epoch": 1.0, - "learning_rate": 8.115785202218315e-09, - "loss": 1.214, - "step": 7390 - }, - { - "epoch": 1.0, - "step": 7393, - "total_flos": 3.018363584615809e+18, - "train_loss": 1.3462718177773738, - "train_runtime": 28665.7785, - "train_samples_per_second": 4.126, - "train_steps_per_second": 0.258 } ], - "max_steps": 7393, + "max_steps": 33579, "num_train_epochs": 1, - "total_flos": 3.018363584615809e+18, + "total_flos": 8.143508134258803e+18, "trial_name": null, "trial_params": null }