diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -2,13174 +2,688 @@ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, - "global_step": 21900, + "global_step": 10950, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { - "epoch": 0.0, - "learning_rate": 2.9986301369863015e-05, - "loss": 0.1722, - "step": 10 + "epoch": 0.02, + "learning_rate": 2.9726027397260275e-05, + "loss": 0.2472, + "step": 100 }, { - "epoch": 0.0, - "learning_rate": 2.9972602739726027e-05, - "loss": 0.0498, - "step": 20 + "epoch": 0.04, + "learning_rate": 2.945205479452055e-05, + "loss": 0.179, + "step": 200 }, { - "epoch": 0.0, - "learning_rate": 2.995890410958904e-05, - "loss": 0.0421, - "step": 30 + "epoch": 0.05, + "learning_rate": 2.9178082191780824e-05, + "loss": 0.169, + "step": 300 }, { - "epoch": 0.0, - "learning_rate": 2.9945205479452056e-05, - "loss": 0.0348, - "step": 40 + "epoch": 0.07, + "learning_rate": 2.8904109589041095e-05, + "loss": 0.1476, + "step": 400 }, { - "epoch": 0.0, - "learning_rate": 2.993150684931507e-05, - "loss": 0.0394, - "step": 50 + "epoch": 0.09, + "learning_rate": 2.863013698630137e-05, + "loss": 0.1419, + "step": 500 }, { - "epoch": 0.01, - "learning_rate": 2.9917808219178085e-05, - "loss": 0.044, - "step": 60 + "epoch": 0.11, + "learning_rate": 2.8356164383561644e-05, + "loss": 0.1517, + "step": 600 }, { - "epoch": 0.01, - "learning_rate": 2.9904109589041096e-05, - "loss": 0.0366, - "step": 70 + "epoch": 0.13, + "learning_rate": 2.808219178082192e-05, + "loss": 0.1466, + "step": 700 }, { - "epoch": 0.01, - "learning_rate": 2.989041095890411e-05, - "loss": 0.0373, - "step": 80 + "epoch": 0.15, + "learning_rate": 2.7808219178082193e-05, + "loss": 0.1398, + "step": 800 }, { - "epoch": 0.01, - "learning_rate": 2.9876712328767122e-05, - "loss": 0.0467, - "step": 90 + "epoch": 0.16, + "learning_rate": 2.7534246575342468e-05, + "loss": 0.1409, + "step": 900 }, { - "epoch": 0.01, - "learning_rate": 2.9863013698630136e-05, - "loss": 0.0294, - "step": 100 + "epoch": 0.18, + "learning_rate": 2.726027397260274e-05, + "loss": 0.1381, + "step": 1000 }, { - "epoch": 0.01, - "learning_rate": 2.9849315068493154e-05, - "loss": 0.0377, - "step": 110 + "epoch": 0.2, + "learning_rate": 2.6986301369863014e-05, + "loss": 0.1349, + "step": 1100 }, { - "epoch": 0.01, - "learning_rate": 2.9835616438356165e-05, - "loss": 0.0346, - "step": 120 + "epoch": 0.22, + "learning_rate": 2.6712328767123288e-05, + "loss": 0.1375, + "step": 1200 }, { - "epoch": 0.01, - "learning_rate": 2.982191780821918e-05, - "loss": 0.0265, - "step": 130 + "epoch": 0.24, + "learning_rate": 2.6438356164383563e-05, + "loss": 0.1397, + "step": 1300 }, { - "epoch": 0.01, - "learning_rate": 2.980821917808219e-05, - "loss": 0.0927, - "step": 140 + "epoch": 0.26, + "learning_rate": 2.6164383561643837e-05, + "loss": 0.1354, + "step": 1400 }, { - "epoch": 0.01, - "learning_rate": 2.9794520547945206e-05, - "loss": 0.0429, - "step": 150 + "epoch": 0.27, + "learning_rate": 2.589041095890411e-05, + "loss": 0.1288, + "step": 1500 }, { - "epoch": 0.01, - "learning_rate": 2.9780821917808217e-05, - "loss": 0.026, - "step": 160 + "epoch": 0.29, + "learning_rate": 2.5616438356164386e-05, + "loss": 0.1216, + "step": 1600 }, { - "epoch": 0.02, - "learning_rate": 2.9767123287671235e-05, - "loss": 0.0325, - "step": 170 + "epoch": 0.31, + "learning_rate": 2.5342465753424657e-05, + "loss": 0.1282, + "step": 1700 }, { - "epoch": 0.02, - "learning_rate": 2.975342465753425e-05, - "loss": 0.0411, - "step": 180 + "epoch": 0.33, + "learning_rate": 2.5068493150684932e-05, + "loss": 0.1302, + "step": 1800 }, { - "epoch": 0.02, - "learning_rate": 2.973972602739726e-05, - "loss": 0.048, - "step": 190 + "epoch": 0.35, + "learning_rate": 2.4794520547945206e-05, + "loss": 0.1289, + "step": 1900 }, { - "epoch": 0.02, - "learning_rate": 2.9726027397260275e-05, - "loss": 0.0418, - "step": 200 + "epoch": 0.37, + "learning_rate": 2.452054794520548e-05, + "loss": 0.1209, + "step": 2000 }, { - "epoch": 0.02, - "learning_rate": 2.9712328767123286e-05, - "loss": 0.0283, - "step": 210 + "epoch": 0.38, + "learning_rate": 2.4246575342465755e-05, + "loss": 0.1253, + "step": 2100 }, { - "epoch": 0.02, - "learning_rate": 2.96986301369863e-05, - "loss": 0.027, - "step": 220 + "epoch": 0.4, + "learning_rate": 2.397260273972603e-05, + "loss": 0.1268, + "step": 2200 }, { - "epoch": 0.02, - "learning_rate": 2.9684931506849316e-05, - "loss": 0.0249, - "step": 230 + "epoch": 0.42, + "learning_rate": 2.36986301369863e-05, + "loss": 0.1234, + "step": 2300 }, { - "epoch": 0.02, - "learning_rate": 2.967123287671233e-05, - "loss": 0.0207, - "step": 240 + "epoch": 0.44, + "learning_rate": 2.3424657534246575e-05, + "loss": 0.1225, + "step": 2400 }, { - "epoch": 0.02, - "learning_rate": 2.9657534246575345e-05, - "loss": 0.03, - "step": 250 + "epoch": 0.46, + "learning_rate": 2.315068493150685e-05, + "loss": 0.1148, + "step": 2500 }, { - "epoch": 0.02, - "learning_rate": 2.9643835616438356e-05, - "loss": 0.0351, - "step": 260 + "epoch": 0.47, + "learning_rate": 2.2876712328767124e-05, + "loss": 0.1228, + "step": 2600 }, { - "epoch": 0.02, - "learning_rate": 2.963013698630137e-05, - "loss": 0.0212, - "step": 270 + "epoch": 0.49, + "learning_rate": 2.26027397260274e-05, + "loss": 0.1141, + "step": 2700 }, { - "epoch": 0.03, - "learning_rate": 2.9616438356164385e-05, - "loss": 0.0226, - "step": 280 + "epoch": 0.51, + "learning_rate": 2.2328767123287673e-05, + "loss": 0.1142, + "step": 2800 }, { - "epoch": 0.03, - "learning_rate": 2.9602739726027396e-05, - "loss": 0.0233, - "step": 290 + "epoch": 0.53, + "learning_rate": 2.2054794520547945e-05, + "loss": 0.11, + "step": 2900 }, { - "epoch": 0.03, - "learning_rate": 2.958904109589041e-05, - "loss": 0.0354, - "step": 300 + "epoch": 0.55, + "learning_rate": 2.178082191780822e-05, + "loss": 0.1142, + "step": 3000 }, { - "epoch": 0.03, - "learning_rate": 2.9575342465753425e-05, - "loss": 0.0274, - "step": 310 + "epoch": 0.57, + "learning_rate": 2.1506849315068494e-05, + "loss": 0.1118, + "step": 3100 }, { - "epoch": 0.03, - "learning_rate": 2.956164383561644e-05, - "loss": 0.0226, - "step": 320 + "epoch": 0.58, + "learning_rate": 2.1232876712328768e-05, + "loss": 0.1126, + "step": 3200 }, { - "epoch": 0.03, - "learning_rate": 2.9547945205479455e-05, - "loss": 0.0301, - "step": 330 + "epoch": 0.6, + "learning_rate": 2.0958904109589043e-05, + "loss": 0.1147, + "step": 3300 }, { - "epoch": 0.03, - "learning_rate": 2.9534246575342466e-05, - "loss": 0.0354, - "step": 340 + "epoch": 0.62, + "learning_rate": 2.0684931506849317e-05, + "loss": 0.1134, + "step": 3400 }, { - "epoch": 0.03, - "learning_rate": 2.952054794520548e-05, - "loss": 0.0296, - "step": 350 + "epoch": 0.64, + "learning_rate": 2.041095890410959e-05, + "loss": 0.1139, + "step": 3500 }, { - "epoch": 0.03, - "learning_rate": 2.950684931506849e-05, - "loss": 0.0301, - "step": 360 + "epoch": 0.66, + "learning_rate": 2.0136986301369863e-05, + "loss": 0.1136, + "step": 3600 }, { - "epoch": 0.03, - "learning_rate": 2.949315068493151e-05, - "loss": 0.0284, - "step": 370 + "epoch": 0.68, + "learning_rate": 1.9863013698630137e-05, + "loss": 0.1034, + "step": 3700 }, { - "epoch": 0.03, - "learning_rate": 2.947945205479452e-05, - "loss": 0.0227, - "step": 380 + "epoch": 0.69, + "learning_rate": 1.9589041095890412e-05, + "loss": 0.1063, + "step": 3800 }, { - "epoch": 0.04, - "learning_rate": 2.9465753424657535e-05, - "loss": 0.0341, - "step": 390 + "epoch": 0.71, + "learning_rate": 1.9315068493150686e-05, + "loss": 0.1075, + "step": 3900 }, { - "epoch": 0.04, - "learning_rate": 2.945205479452055e-05, - "loss": 0.0268, - "step": 400 + "epoch": 0.73, + "learning_rate": 1.904109589041096e-05, + "loss": 0.1076, + "step": 4000 }, { - "epoch": 0.04, - "learning_rate": 2.943835616438356e-05, - "loss": 0.0232, - "step": 410 + "epoch": 0.75, + "learning_rate": 1.8767123287671235e-05, + "loss": 0.1057, + "step": 4100 }, { - "epoch": 0.04, - "learning_rate": 2.9424657534246576e-05, - "loss": 0.0224, - "step": 420 + "epoch": 0.77, + "learning_rate": 1.8493150684931506e-05, + "loss": 0.1016, + "step": 4200 }, { - "epoch": 0.04, - "learning_rate": 2.9410958904109587e-05, - "loss": 0.0259, - "step": 430 + "epoch": 0.79, + "learning_rate": 1.821917808219178e-05, + "loss": 0.1058, + "step": 4300 }, { - "epoch": 0.04, - "learning_rate": 2.9397260273972605e-05, - "loss": 0.053, - "step": 440 + "epoch": 0.8, + "learning_rate": 1.7945205479452055e-05, + "loss": 0.1099, + "step": 4400 }, { - "epoch": 0.04, - "learning_rate": 2.938356164383562e-05, - "loss": 0.0266, - "step": 450 + "epoch": 0.82, + "learning_rate": 1.767123287671233e-05, + "loss": 0.1082, + "step": 4500 }, { - "epoch": 0.04, - "learning_rate": 2.936986301369863e-05, - "loss": 0.032, - "step": 460 + "epoch": 0.84, + "learning_rate": 1.7397260273972604e-05, + "loss": 0.1009, + "step": 4600 }, { - "epoch": 0.04, - "learning_rate": 2.9356164383561645e-05, - "loss": 0.026, - "step": 470 + "epoch": 0.86, + "learning_rate": 1.712328767123288e-05, + "loss": 0.1076, + "step": 4700 }, { - "epoch": 0.04, - "learning_rate": 2.9342465753424656e-05, - "loss": 0.0307, - "step": 480 + "epoch": 0.88, + "learning_rate": 1.6849315068493153e-05, + "loss": 0.106, + "step": 4800 }, { - "epoch": 0.04, - "learning_rate": 2.932876712328767e-05, - "loss": 0.0329, - "step": 490 + "epoch": 0.89, + "learning_rate": 1.6575342465753425e-05, + "loss": 0.103, + "step": 4900 }, { - "epoch": 0.05, - "learning_rate": 2.9315068493150685e-05, - "loss": 0.0221, - "step": 500 + "epoch": 0.91, + "learning_rate": 1.63013698630137e-05, + "loss": 0.1111, + "step": 5000 }, { - "epoch": 0.05, - "learning_rate": 2.93013698630137e-05, - "loss": 0.0433, - "step": 510 + "epoch": 0.93, + "learning_rate": 1.6027397260273974e-05, + "loss": 0.1039, + "step": 5100 }, { - "epoch": 0.05, - "learning_rate": 2.9287671232876714e-05, - "loss": 0.0965, - "step": 520 + "epoch": 0.95, + "learning_rate": 1.5753424657534248e-05, + "loss": 0.1004, + "step": 5200 }, { - "epoch": 0.05, - "learning_rate": 2.9273972602739726e-05, - "loss": 0.0208, - "step": 530 + "epoch": 0.97, + "learning_rate": 1.5479452054794523e-05, + "loss": 0.1039, + "step": 5300 }, { - "epoch": 0.05, - "learning_rate": 2.926027397260274e-05, - "loss": 0.0218, - "step": 540 + "epoch": 0.99, + "learning_rate": 1.5205479452054795e-05, + "loss": 0.1053, + "step": 5400 }, { - "epoch": 0.05, - "learning_rate": 2.9246575342465755e-05, - "loss": 0.0278, - "step": 550 + "epoch": 1.0, + "eval_exact_match": 73.0085146641438, + "eval_f1": 83.09249441563516, + "step": 5475 }, { - "epoch": 0.05, - "learning_rate": 2.9232876712328766e-05, - "loss": 0.0272, - "step": 560 + "epoch": 1.0, + "learning_rate": 1.4931506849315068e-05, + "loss": 0.0943, + "step": 5500 }, { - "epoch": 0.05, - "learning_rate": 2.9219178082191784e-05, - "loss": 0.0213, - "step": 570 + "epoch": 1.02, + "learning_rate": 1.4657534246575343e-05, + "loss": 0.0767, + "step": 5600 }, { - "epoch": 0.05, - "learning_rate": 2.9205479452054795e-05, - "loss": 0.0277, - "step": 580 + "epoch": 1.04, + "learning_rate": 1.4383561643835615e-05, + "loss": 0.0762, + "step": 5700 }, { - "epoch": 0.05, - "learning_rate": 2.919178082191781e-05, - "loss": 0.0219, - "step": 590 + "epoch": 1.06, + "learning_rate": 1.410958904109589e-05, + "loss": 0.0759, + "step": 5800 }, { - "epoch": 0.05, - "learning_rate": 2.9178082191780824e-05, - "loss": 0.0246, - "step": 600 + "epoch": 1.08, + "learning_rate": 1.3835616438356164e-05, + "loss": 0.0801, + "step": 5900 }, { - "epoch": 0.06, - "learning_rate": 2.9164383561643835e-05, - "loss": 0.0222, - "step": 610 + "epoch": 1.1, + "learning_rate": 1.3561643835616437e-05, + "loss": 0.0719, + "step": 6000 }, { - "epoch": 0.06, - "learning_rate": 2.915068493150685e-05, - "loss": 0.0275, - "step": 620 + "epoch": 1.11, + "learning_rate": 1.3287671232876712e-05, + "loss": 0.0754, + "step": 6100 }, { - "epoch": 0.06, - "learning_rate": 2.913698630136986e-05, - "loss": 0.0225, - "step": 630 + "epoch": 1.13, + "learning_rate": 1.3013698630136986e-05, + "loss": 0.0781, + "step": 6200 }, { - "epoch": 0.06, - "learning_rate": 2.912328767123288e-05, - "loss": 0.0266, - "step": 640 + "epoch": 1.15, + "learning_rate": 1.2739726027397259e-05, + "loss": 0.0823, + "step": 6300 }, { - "epoch": 0.06, - "learning_rate": 2.910958904109589e-05, - "loss": 0.0209, - "step": 650 + "epoch": 1.17, + "learning_rate": 1.2465753424657534e-05, + "loss": 0.0827, + "step": 6400 }, { - "epoch": 0.06, - "learning_rate": 2.9095890410958905e-05, - "loss": 0.0261, - "step": 660 + "epoch": 1.19, + "learning_rate": 1.2191780821917808e-05, + "loss": 0.0718, + "step": 6500 }, { - "epoch": 0.06, - "learning_rate": 2.908219178082192e-05, - "loss": 0.0207, - "step": 670 + "epoch": 1.21, + "learning_rate": 1.1917808219178083e-05, + "loss": 0.0755, + "step": 6600 }, { - "epoch": 0.06, - "learning_rate": 2.906849315068493e-05, - "loss": 0.0245, - "step": 680 + "epoch": 1.22, + "learning_rate": 1.1643835616438355e-05, + "loss": 0.0807, + "step": 6700 }, { - "epoch": 0.06, - "learning_rate": 2.9054794520547945e-05, - "loss": 0.0187, - "step": 690 + "epoch": 1.24, + "learning_rate": 1.136986301369863e-05, + "loss": 0.083, + "step": 6800 }, { - "epoch": 0.06, - "learning_rate": 2.9041095890410956e-05, - "loss": 0.0249, - "step": 700 + "epoch": 1.26, + "learning_rate": 1.1095890410958904e-05, + "loss": 0.0741, + "step": 6900 }, { - "epoch": 0.06, - "learning_rate": 2.9027397260273974e-05, - "loss": 0.0225, - "step": 710 + "epoch": 1.28, + "learning_rate": 1.0821917808219177e-05, + "loss": 0.0782, + "step": 7000 }, { - "epoch": 0.07, - "learning_rate": 2.901369863013699e-05, - "loss": 0.0207, - "step": 720 - }, - { - "epoch": 0.07, - "learning_rate": 2.9e-05, - "loss": 0.0183, - "step": 730 - }, - { - "epoch": 0.07, - "learning_rate": 2.8986301369863015e-05, - "loss": 0.022, - "step": 740 - }, - { - "epoch": 0.07, - "learning_rate": 2.8972602739726026e-05, - "loss": 0.0218, - "step": 750 - }, - { - "epoch": 0.07, - "learning_rate": 2.895890410958904e-05, - "loss": 0.0384, - "step": 760 - }, - { - "epoch": 0.07, - "learning_rate": 2.894520547945206e-05, - "loss": 0.0236, - "step": 770 - }, - { - "epoch": 0.07, - "learning_rate": 2.893150684931507e-05, - "loss": 0.0223, - "step": 780 - }, - { - "epoch": 0.07, - "learning_rate": 2.8917808219178084e-05, - "loss": 0.0273, - "step": 790 - }, - { - "epoch": 0.07, - "learning_rate": 2.8904109589041095e-05, - "loss": 0.0199, - "step": 800 - }, - { - "epoch": 0.07, - "learning_rate": 2.889041095890411e-05, - "loss": 0.0205, - "step": 810 - }, - { - "epoch": 0.07, - "learning_rate": 2.8876712328767125e-05, - "loss": 0.0206, - "step": 820 - }, - { - "epoch": 0.08, - "learning_rate": 2.8863013698630136e-05, - "loss": 0.022, - "step": 830 - }, - { - "epoch": 0.08, - "learning_rate": 2.8849315068493154e-05, - "loss": 0.027, - "step": 840 - }, - { - "epoch": 0.08, - "learning_rate": 2.8835616438356165e-05, - "loss": 0.019, - "step": 850 - }, - { - "epoch": 0.08, - "learning_rate": 2.882191780821918e-05, - "loss": 0.0225, - "step": 860 - }, - { - "epoch": 0.08, - "learning_rate": 2.8808219178082194e-05, - "loss": 0.0234, - "step": 870 - }, - { - "epoch": 0.08, - "learning_rate": 2.8794520547945205e-05, - "loss": 0.028, - "step": 880 - }, - { - "epoch": 0.08, - "learning_rate": 2.878082191780822e-05, - "loss": 0.0219, - "step": 890 - }, - { - "epoch": 0.08, - "learning_rate": 2.876712328767123e-05, - "loss": 0.022, - "step": 900 - }, - { - "epoch": 0.08, - "learning_rate": 2.875342465753425e-05, - "loss": 0.02, - "step": 910 - }, - { - "epoch": 0.08, - "learning_rate": 2.873972602739726e-05, - "loss": 0.0486, - "step": 920 - }, - { - "epoch": 0.08, - "learning_rate": 2.8726027397260275e-05, - "loss": 0.0193, - "step": 930 - }, - { - "epoch": 0.09, - "learning_rate": 2.871232876712329e-05, - "loss": 0.0174, - "step": 940 - }, - { - "epoch": 0.09, - "learning_rate": 2.86986301369863e-05, - "loss": 0.0209, - "step": 950 - }, - { - "epoch": 0.09, - "learning_rate": 2.8684931506849315e-05, - "loss": 0.0257, - "step": 960 - }, - { - "epoch": 0.09, - "learning_rate": 2.867123287671233e-05, - "loss": 0.0259, - "step": 970 - }, - { - "epoch": 0.09, - "learning_rate": 2.8657534246575344e-05, - "loss": 0.0186, - "step": 980 - }, - { - "epoch": 0.09, - "learning_rate": 2.864383561643836e-05, - "loss": 0.021, - "step": 990 - }, - { - "epoch": 0.09, - "learning_rate": 2.863013698630137e-05, - "loss": 0.0169, - "step": 1000 - }, - { - "epoch": 0.09, - "learning_rate": 2.8616438356164384e-05, - "loss": 0.0204, - "step": 1010 - }, - { - "epoch": 0.09, - "learning_rate": 2.8602739726027396e-05, - "loss": 0.0661, - "step": 1020 - }, - { - "epoch": 0.09, - "learning_rate": 2.858904109589041e-05, - "loss": 0.0425, - "step": 1030 - }, - { - "epoch": 0.09, - "learning_rate": 2.8575342465753428e-05, - "loss": 0.0173, - "step": 1040 - }, - { - "epoch": 0.1, - "learning_rate": 2.856164383561644e-05, - "loss": 0.0203, - "step": 1050 - }, - { - "epoch": 0.1, - "learning_rate": 2.8547945205479454e-05, - "loss": 0.0236, - "step": 1060 - }, - { - "epoch": 0.1, - "learning_rate": 2.8534246575342465e-05, - "loss": 0.0419, - "step": 1070 - }, - { - "epoch": 0.1, - "learning_rate": 2.852054794520548e-05, - "loss": 0.0166, - "step": 1080 - }, - { - "epoch": 0.1, - "learning_rate": 2.8506849315068494e-05, - "loss": 0.0269, - "step": 1090 - }, - { - "epoch": 0.1, - "learning_rate": 2.8493150684931505e-05, - "loss": 0.0275, - "step": 1100 - }, - { - "epoch": 0.1, - "learning_rate": 2.8479452054794523e-05, - "loss": 0.0218, - "step": 1110 - }, - { - "epoch": 0.1, - "learning_rate": 2.8465753424657535e-05, - "loss": 0.0183, - "step": 1120 - }, - { - "epoch": 0.1, - "learning_rate": 2.845205479452055e-05, - "loss": 0.0179, - "step": 1130 - }, - { - "epoch": 0.1, - "learning_rate": 2.8438356164383564e-05, - "loss": 0.0238, - "step": 1140 - }, - { - "epoch": 0.11, - "learning_rate": 2.8424657534246575e-05, - "loss": 0.0206, - "step": 1150 - }, - { - "epoch": 0.11, - "learning_rate": 2.841095890410959e-05, - "loss": 0.0212, - "step": 1160 - }, - { - "epoch": 0.11, - "learning_rate": 2.8397260273972604e-05, - "loss": 0.0233, - "step": 1170 - }, - { - "epoch": 0.11, - "learning_rate": 2.838356164383562e-05, - "loss": 0.026, - "step": 1180 - }, - { - "epoch": 0.11, - "learning_rate": 2.836986301369863e-05, - "loss": 0.0219, - "step": 1190 - }, - { - "epoch": 0.11, - "learning_rate": 2.8356164383561644e-05, - "loss": 0.0211, - "step": 1200 - }, - { - "epoch": 0.11, - "learning_rate": 2.834246575342466e-05, - "loss": 0.0212, - "step": 1210 - }, - { - "epoch": 0.11, - "learning_rate": 2.832876712328767e-05, - "loss": 0.0246, - "step": 1220 - }, - { - "epoch": 0.11, - "learning_rate": 2.8315068493150685e-05, - "loss": 0.0176, - "step": 1230 - }, - { - "epoch": 0.11, - "learning_rate": 2.83013698630137e-05, - "loss": 0.0176, - "step": 1240 - }, - { - "epoch": 0.11, - "learning_rate": 2.8287671232876714e-05, - "loss": 0.0201, - "step": 1250 - }, - { - "epoch": 0.12, - "learning_rate": 2.827397260273973e-05, - "loss": 0.0274, - "step": 1260 - }, - { - "epoch": 0.12, - "learning_rate": 2.826027397260274e-05, - "loss": 0.0259, - "step": 1270 - }, - { - "epoch": 0.12, - "learning_rate": 2.8246575342465754e-05, - "loss": 0.0212, - "step": 1280 - }, - { - "epoch": 0.12, - "learning_rate": 2.8232876712328765e-05, - "loss": 0.0187, - "step": 1290 - }, - { - "epoch": 0.12, - "learning_rate": 2.821917808219178e-05, - "loss": 0.0227, - "step": 1300 - }, - { - "epoch": 0.12, - "learning_rate": 2.8205479452054798e-05, - "loss": 0.0213, - "step": 1310 - }, - { - "epoch": 0.12, - "learning_rate": 2.819178082191781e-05, - "loss": 0.0201, - "step": 1320 - }, - { - "epoch": 0.12, - "learning_rate": 2.8178082191780824e-05, - "loss": 0.0256, - "step": 1330 - }, - { - "epoch": 0.12, - "learning_rate": 2.8164383561643835e-05, - "loss": 0.023, - "step": 1340 - }, - { - "epoch": 0.12, - "learning_rate": 2.815068493150685e-05, - "loss": 0.0433, - "step": 1350 - }, - { - "epoch": 0.12, - "learning_rate": 2.8136986301369864e-05, - "loss": 0.0233, - "step": 1360 - }, - { - "epoch": 0.13, - "learning_rate": 2.812328767123288e-05, - "loss": 0.0147, - "step": 1370 - }, - { - "epoch": 0.13, - "learning_rate": 2.8109589041095893e-05, - "loss": 0.0214, - "step": 1380 - }, - { - "epoch": 0.13, - "learning_rate": 2.8095890410958904e-05, - "loss": 0.0174, - "step": 1390 - }, - { - "epoch": 0.13, - "learning_rate": 2.808219178082192e-05, - "loss": 0.0174, - "step": 1400 - }, - { - "epoch": 0.13, - "learning_rate": 2.8068493150684933e-05, - "loss": 0.0195, - "step": 1410 - }, - { - "epoch": 0.13, - "learning_rate": 2.8054794520547945e-05, - "loss": 0.0181, - "step": 1420 - }, - { - "epoch": 0.13, - "learning_rate": 2.804109589041096e-05, - "loss": 0.019, - "step": 1430 - }, - { - "epoch": 0.13, - "learning_rate": 2.8027397260273974e-05, - "loss": 0.0217, - "step": 1440 - }, - { - "epoch": 0.13, - "learning_rate": 2.801369863013699e-05, - "loss": 0.0209, - "step": 1450 - }, - { - "epoch": 0.13, - "learning_rate": 2.8e-05, - "loss": 0.02, - "step": 1460 - }, - { - "epoch": 0.13, - "learning_rate": 2.7986301369863014e-05, - "loss": 0.021, - "step": 1470 - }, - { - "epoch": 0.14, - "learning_rate": 2.797260273972603e-05, - "loss": 0.02, - "step": 1480 - }, - { - "epoch": 0.14, - "learning_rate": 2.795890410958904e-05, - "loss": 0.0357, - "step": 1490 - }, - { - "epoch": 0.14, - "learning_rate": 2.7945205479452054e-05, - "loss": 0.0176, - "step": 1500 - }, - { - "epoch": 0.14, - "learning_rate": 2.793150684931507e-05, - "loss": 0.0191, - "step": 1510 - }, - { - "epoch": 0.14, - "learning_rate": 2.7917808219178084e-05, - "loss": 0.0211, - "step": 1520 - }, - { - "epoch": 0.14, - "learning_rate": 2.7904109589041098e-05, - "loss": 0.0212, - "step": 1530 - }, - { - "epoch": 0.14, - "learning_rate": 2.789041095890411e-05, - "loss": 0.0385, - "step": 1540 - }, - { - "epoch": 0.14, - "learning_rate": 2.7876712328767124e-05, - "loss": 0.016, - "step": 1550 - }, - { - "epoch": 0.14, - "learning_rate": 2.7863013698630135e-05, - "loss": 0.0201, - "step": 1560 - }, - { - "epoch": 0.14, - "learning_rate": 2.7849315068493153e-05, - "loss": 0.0191, - "step": 1570 - }, - { - "epoch": 0.14, - "learning_rate": 2.7835616438356168e-05, - "loss": 0.0213, - "step": 1580 - }, - { - "epoch": 0.15, - "learning_rate": 2.782191780821918e-05, - "loss": 0.0216, - "step": 1590 - }, - { - "epoch": 0.15, - "learning_rate": 2.7808219178082193e-05, - "loss": 0.0194, - "step": 1600 - }, - { - "epoch": 0.15, - "learning_rate": 2.7794520547945205e-05, - "loss": 0.0247, - "step": 1610 - }, - { - "epoch": 0.15, - "learning_rate": 2.778082191780822e-05, - "loss": 0.0194, - "step": 1620 - }, - { - "epoch": 0.15, - "learning_rate": 2.7767123287671234e-05, - "loss": 0.0242, - "step": 1630 - }, - { - "epoch": 0.15, - "learning_rate": 2.775342465753425e-05, - "loss": 0.0157, - "step": 1640 - }, - { - "epoch": 0.15, - "learning_rate": 2.7739726027397263e-05, - "loss": 0.0156, - "step": 1650 - }, - { - "epoch": 0.15, - "learning_rate": 2.7726027397260274e-05, - "loss": 0.0244, - "step": 1660 - }, - { - "epoch": 0.15, - "learning_rate": 2.771232876712329e-05, - "loss": 0.0191, - "step": 1670 - }, - { - "epoch": 0.15, - "learning_rate": 2.76986301369863e-05, - "loss": 0.0249, - "step": 1680 - }, - { - "epoch": 0.15, - "learning_rate": 2.7684931506849314e-05, - "loss": 0.0233, - "step": 1690 - }, - { - "epoch": 0.16, - "learning_rate": 2.767123287671233e-05, - "loss": 0.1308, - "step": 1700 - }, - { - "epoch": 0.16, - "learning_rate": 2.7657534246575344e-05, - "loss": 0.0171, - "step": 1710 - }, - { - "epoch": 0.16, - "learning_rate": 2.7643835616438358e-05, - "loss": 0.0171, - "step": 1720 - }, - { - "epoch": 0.16, - "learning_rate": 2.763013698630137e-05, - "loss": 0.0164, - "step": 1730 - }, - { - "epoch": 0.16, - "learning_rate": 2.7616438356164384e-05, - "loss": 0.0488, - "step": 1740 - }, - { - "epoch": 0.16, - "learning_rate": 2.76027397260274e-05, - "loss": 0.0152, - "step": 1750 - }, - { - "epoch": 0.16, - "learning_rate": 2.758904109589041e-05, - "loss": 0.0149, - "step": 1760 - }, - { - "epoch": 0.16, - "learning_rate": 2.7575342465753428e-05, - "loss": 0.0136, - "step": 1770 - }, - { - "epoch": 0.16, - "learning_rate": 2.756164383561644e-05, - "loss": 0.0194, - "step": 1780 - }, - { - "epoch": 0.16, - "learning_rate": 2.7547945205479453e-05, - "loss": 0.0194, - "step": 1790 - }, - { - "epoch": 0.16, - "learning_rate": 2.7534246575342468e-05, - "loss": 0.0204, - "step": 1800 - }, - { - "epoch": 0.17, - "learning_rate": 2.752054794520548e-05, - "loss": 0.0212, - "step": 1810 - }, - { - "epoch": 0.17, - "learning_rate": 2.7506849315068494e-05, - "loss": 0.0182, - "step": 1820 - }, - { - "epoch": 0.17, - "learning_rate": 2.7493150684931505e-05, - "loss": 0.0167, - "step": 1830 - }, - { - "epoch": 0.17, - "learning_rate": 2.7479452054794523e-05, - "loss": 0.0195, - "step": 1840 - }, - { - "epoch": 0.17, - "learning_rate": 2.7465753424657537e-05, - "loss": 0.0168, - "step": 1850 - }, - { - "epoch": 0.17, - "learning_rate": 2.745205479452055e-05, - "loss": 0.0213, - "step": 1860 - }, - { - "epoch": 0.17, - "learning_rate": 2.7438356164383563e-05, - "loss": 0.0192, - "step": 1870 - }, - { - "epoch": 0.17, - "learning_rate": 2.7424657534246574e-05, - "loss": 0.018, - "step": 1880 - }, - { - "epoch": 0.17, - "learning_rate": 2.741095890410959e-05, - "loss": 0.0185, - "step": 1890 - }, - { - "epoch": 0.17, - "learning_rate": 2.7397260273972603e-05, - "loss": 0.0169, - "step": 1900 - }, - { - "epoch": 0.17, - "learning_rate": 2.7383561643835618e-05, - "loss": 0.0221, - "step": 1910 - }, - { - "epoch": 0.18, - "learning_rate": 2.7369863013698633e-05, - "loss": 0.1066, - "step": 1920 - }, - { - "epoch": 0.18, - "learning_rate": 2.7356164383561644e-05, - "loss": 0.0227, - "step": 1930 - }, - { - "epoch": 0.18, - "learning_rate": 2.734246575342466e-05, - "loss": 0.0171, - "step": 1940 - }, - { - "epoch": 0.18, - "learning_rate": 2.732876712328767e-05, - "loss": 0.0227, - "step": 1950 - }, - { - "epoch": 0.18, - "learning_rate": 2.7315068493150684e-05, - "loss": 0.0223, - "step": 1960 - }, - { - "epoch": 0.18, - "learning_rate": 2.7301369863013702e-05, - "loss": 0.0201, - "step": 1970 - }, - { - "epoch": 0.18, - "learning_rate": 2.7287671232876713e-05, - "loss": 0.02, - "step": 1980 - }, - { - "epoch": 0.18, - "learning_rate": 2.7273972602739728e-05, - "loss": 0.0218, - "step": 1990 - }, - { - "epoch": 0.18, - "learning_rate": 2.726027397260274e-05, - "loss": 0.0196, - "step": 2000 - }, - { - "epoch": 0.18, - "learning_rate": 2.7246575342465754e-05, - "loss": 0.018, - "step": 2010 - }, - { - "epoch": 0.18, - "learning_rate": 2.7232876712328768e-05, - "loss": 0.0202, - "step": 2020 - }, - { - "epoch": 0.19, - "learning_rate": 2.721917808219178e-05, - "loss": 0.0189, - "step": 2030 - }, - { - "epoch": 0.19, - "learning_rate": 2.7205479452054797e-05, - "loss": 0.0405, - "step": 2040 - }, - { - "epoch": 0.19, - "learning_rate": 2.719178082191781e-05, - "loss": 0.0663, - "step": 2050 - }, - { - "epoch": 0.19, - "learning_rate": 2.7178082191780823e-05, - "loss": 0.0199, - "step": 2060 - }, - { - "epoch": 0.19, - "learning_rate": 2.7164383561643838e-05, - "loss": 0.0216, - "step": 2070 - }, - { - "epoch": 0.19, - "learning_rate": 2.715068493150685e-05, - "loss": 0.0183, - "step": 2080 - }, - { - "epoch": 0.19, - "learning_rate": 2.7136986301369863e-05, - "loss": 0.0173, - "step": 2090 - }, - { - "epoch": 0.19, - "learning_rate": 2.7123287671232875e-05, - "loss": 0.0204, - "step": 2100 - }, - { - "epoch": 0.19, - "learning_rate": 2.7109589041095893e-05, - "loss": 0.0201, - "step": 2110 - }, - { - "epoch": 0.19, - "learning_rate": 2.7095890410958907e-05, - "loss": 0.0243, - "step": 2120 - }, - { - "epoch": 0.19, - "learning_rate": 2.708219178082192e-05, - "loss": 0.0214, - "step": 2130 - }, - { - "epoch": 0.2, - "learning_rate": 2.7068493150684933e-05, - "loss": 0.0173, - "step": 2140 - }, - { - "epoch": 0.2, - "learning_rate": 2.7054794520547944e-05, - "loss": 0.0146, - "step": 2150 - }, - { - "epoch": 0.2, - "learning_rate": 2.704109589041096e-05, - "loss": 0.0246, - "step": 2160 - }, - { - "epoch": 0.2, - "learning_rate": 2.7027397260273977e-05, - "loss": 0.0178, - "step": 2170 - }, - { - "epoch": 0.2, - "learning_rate": 2.7013698630136988e-05, - "loss": 0.0158, - "step": 2180 - }, - { - "epoch": 0.2, - "learning_rate": 2.7000000000000002e-05, - "loss": 0.0178, - "step": 2190 - }, - { - "epoch": 0.2, - "learning_rate": 2.6986301369863014e-05, - "loss": 0.024, - "step": 2200 - }, - { - "epoch": 0.2, - "learning_rate": 2.6972602739726028e-05, - "loss": 0.0266, - "step": 2210 - }, - { - "epoch": 0.2, - "learning_rate": 2.695890410958904e-05, - "loss": 0.0183, - "step": 2220 - }, - { - "epoch": 0.2, - "learning_rate": 2.6945205479452054e-05, - "loss": 0.02, - "step": 2230 - }, - { - "epoch": 0.2, - "learning_rate": 2.6931506849315072e-05, - "loss": 0.0198, - "step": 2240 - }, - { - "epoch": 0.21, - "learning_rate": 2.6917808219178083e-05, - "loss": 0.019, - "step": 2250 - }, - { - "epoch": 0.21, - "learning_rate": 2.6904109589041098e-05, - "loss": 0.0187, - "step": 2260 - }, - { - "epoch": 0.21, - "learning_rate": 2.689041095890411e-05, - "loss": 0.023, - "step": 2270 - }, - { - "epoch": 0.21, - "learning_rate": 2.6876712328767123e-05, - "loss": 0.0168, - "step": 2280 - }, - { - "epoch": 0.21, - "learning_rate": 2.6863013698630138e-05, - "loss": 0.0153, - "step": 2290 - }, - { - "epoch": 0.21, - "learning_rate": 2.684931506849315e-05, - "loss": 0.0543, - "step": 2300 - }, - { - "epoch": 0.21, - "learning_rate": 2.6835616438356167e-05, - "loss": 0.0179, - "step": 2310 - }, - { - "epoch": 0.21, - "learning_rate": 2.6821917808219178e-05, - "loss": 0.0192, - "step": 2320 - }, - { - "epoch": 0.21, - "learning_rate": 2.6808219178082193e-05, - "loss": 0.023, - "step": 2330 - }, - { - "epoch": 0.21, - "learning_rate": 2.6794520547945207e-05, - "loss": 0.016, - "step": 2340 - }, - { - "epoch": 0.21, - "learning_rate": 2.678082191780822e-05, - "loss": 0.0151, - "step": 2350 - }, - { - "epoch": 0.22, - "learning_rate": 2.6767123287671233e-05, - "loss": 0.0178, - "step": 2360 - }, - { - "epoch": 0.22, - "learning_rate": 2.6753424657534244e-05, - "loss": 0.0178, - "step": 2370 - }, - { - "epoch": 0.22, - "learning_rate": 2.6739726027397262e-05, - "loss": 0.0182, - "step": 2380 - }, - { - "epoch": 0.22, - "learning_rate": 2.6726027397260277e-05, - "loss": 0.029, - "step": 2390 - }, - { - "epoch": 0.22, - "learning_rate": 2.6712328767123288e-05, - "loss": 0.0226, - "step": 2400 - }, - { - "epoch": 0.22, - "learning_rate": 2.6698630136986303e-05, - "loss": 0.0767, - "step": 2410 - }, - { - "epoch": 0.22, - "learning_rate": 2.6684931506849314e-05, - "loss": 0.0169, - "step": 2420 - }, - { - "epoch": 0.22, - "learning_rate": 2.667123287671233e-05, - "loss": 0.0231, - "step": 2430 - }, - { - "epoch": 0.22, - "learning_rate": 2.6657534246575343e-05, - "loss": 0.0197, - "step": 2440 - }, - { - "epoch": 0.22, - "learning_rate": 2.6643835616438358e-05, - "loss": 0.0172, - "step": 2450 - }, - { - "epoch": 0.22, - "learning_rate": 2.6630136986301372e-05, - "loss": 0.0193, - "step": 2460 - }, - { - "epoch": 0.23, - "learning_rate": 2.6616438356164383e-05, - "loss": 0.0168, - "step": 2470 - }, - { - "epoch": 0.23, - "learning_rate": 2.6602739726027398e-05, - "loss": 0.0191, - "step": 2480 - }, - { - "epoch": 0.23, - "learning_rate": 2.658904109589041e-05, - "loss": 0.0214, - "step": 2490 - }, - { - "epoch": 0.23, - "learning_rate": 2.6575342465753424e-05, - "loss": 0.0458, - "step": 2500 - }, - { - "epoch": 0.23, - "learning_rate": 2.656164383561644e-05, - "loss": 0.0171, - "step": 2510 - }, - { - "epoch": 0.23, - "learning_rate": 2.6547945205479453e-05, - "loss": 0.0171, - "step": 2520 - }, - { - "epoch": 0.23, - "learning_rate": 2.6534246575342467e-05, - "loss": 0.0143, - "step": 2530 - }, - { - "epoch": 0.23, - "learning_rate": 2.652054794520548e-05, - "loss": 0.0153, - "step": 2540 - }, - { - "epoch": 0.23, - "learning_rate": 2.6506849315068493e-05, - "loss": 0.02, - "step": 2550 - }, - { - "epoch": 0.23, - "learning_rate": 2.6493150684931508e-05, - "loss": 0.0221, - "step": 2560 - }, - { - "epoch": 0.23, - "learning_rate": 2.647945205479452e-05, - "loss": 0.0193, - "step": 2570 - }, - { - "epoch": 0.24, - "learning_rate": 2.6465753424657537e-05, - "loss": 0.0161, - "step": 2580 - }, - { - "epoch": 0.24, - "learning_rate": 2.6452054794520548e-05, - "loss": 0.0235, - "step": 2590 - }, - { - "epoch": 0.24, - "learning_rate": 2.6438356164383563e-05, - "loss": 0.0174, - "step": 2600 - }, - { - "epoch": 0.24, - "learning_rate": 2.6424657534246577e-05, - "loss": 0.0145, - "step": 2610 - }, - { - "epoch": 0.24, - "learning_rate": 2.641095890410959e-05, - "loss": 0.0245, - "step": 2620 - }, - { - "epoch": 0.24, - "learning_rate": 2.6397260273972603e-05, - "loss": 0.0181, - "step": 2630 - }, - { - "epoch": 0.24, - "learning_rate": 2.6383561643835617e-05, - "loss": 0.0197, - "step": 2640 - }, - { - "epoch": 0.24, - "learning_rate": 2.6369863013698632e-05, - "loss": 0.0311, - "step": 2650 - }, - { - "epoch": 0.24, - "learning_rate": 2.6356164383561647e-05, - "loss": 0.023, - "step": 2660 - }, - { - "epoch": 0.24, - "learning_rate": 2.6342465753424658e-05, - "loss": 0.0195, - "step": 2670 - }, - { - "epoch": 0.24, - "learning_rate": 2.6328767123287672e-05, - "loss": 0.017, - "step": 2680 - }, - { - "epoch": 0.25, - "learning_rate": 2.6315068493150684e-05, - "loss": 0.0252, - "step": 2690 - }, - { - "epoch": 0.25, - "learning_rate": 2.6301369863013698e-05, - "loss": 0.0163, - "step": 2700 - }, - { - "epoch": 0.25, - "learning_rate": 2.6287671232876713e-05, - "loss": 0.0199, - "step": 2710 - }, - { - "epoch": 0.25, - "learning_rate": 2.6273972602739727e-05, - "loss": 0.0207, - "step": 2720 - }, - { - "epoch": 0.25, - "learning_rate": 2.6260273972602742e-05, - "loss": 0.0182, - "step": 2730 - }, - { - "epoch": 0.25, - "learning_rate": 2.6246575342465753e-05, - "loss": 0.0233, - "step": 2740 - }, - { - "epoch": 0.25, - "learning_rate": 2.6232876712328768e-05, - "loss": 0.0461, - "step": 2750 - }, - { - "epoch": 0.25, - "learning_rate": 2.621917808219178e-05, - "loss": 0.0222, - "step": 2760 - }, - { - "epoch": 0.25, - "learning_rate": 2.6205479452054793e-05, - "loss": 0.0232, - "step": 2770 - }, - { - "epoch": 0.25, - "learning_rate": 2.619178082191781e-05, - "loss": 0.0174, - "step": 2780 - }, - { - "epoch": 0.25, - "learning_rate": 2.6178082191780823e-05, - "loss": 0.0224, - "step": 2790 - }, - { - "epoch": 0.26, - "learning_rate": 2.6164383561643837e-05, - "loss": 0.0201, - "step": 2800 - }, - { - "epoch": 0.26, - "learning_rate": 2.6150684931506848e-05, - "loss": 0.0532, - "step": 2810 - }, - { - "epoch": 0.26, - "learning_rate": 2.6136986301369863e-05, - "loss": 0.0273, - "step": 2820 - }, - { - "epoch": 0.26, - "learning_rate": 2.6123287671232877e-05, - "loss": 0.0214, - "step": 2830 - }, - { - "epoch": 0.26, - "learning_rate": 2.6109589041095892e-05, - "loss": 0.0147, - "step": 2840 - }, - { - "epoch": 0.26, - "learning_rate": 2.6095890410958907e-05, - "loss": 0.0186, - "step": 2850 - }, - { - "epoch": 0.26, - "learning_rate": 2.6082191780821918e-05, - "loss": 0.0186, - "step": 2860 - }, - { - "epoch": 0.26, - "learning_rate": 2.6068493150684932e-05, - "loss": 0.0169, - "step": 2870 - }, - { - "epoch": 0.26, - "learning_rate": 2.6054794520547947e-05, - "loss": 0.022, - "step": 2880 - }, - { - "epoch": 0.26, - "learning_rate": 2.6041095890410958e-05, - "loss": 0.0282, - "step": 2890 - }, - { - "epoch": 0.26, - "learning_rate": 2.6027397260273973e-05, - "loss": 0.0194, - "step": 2900 - }, - { - "epoch": 0.27, - "learning_rate": 2.6013698630136987e-05, - "loss": 0.0797, - "step": 2910 - }, - { - "epoch": 0.27, - "learning_rate": 2.6000000000000002e-05, - "loss": 0.0177, - "step": 2920 - }, - { - "epoch": 0.27, - "learning_rate": 2.5986301369863016e-05, - "loss": 0.0139, - "step": 2930 - }, - { - "epoch": 0.27, - "learning_rate": 2.5972602739726028e-05, - "loss": 0.0133, - "step": 2940 - }, - { - "epoch": 0.27, - "learning_rate": 2.5958904109589042e-05, - "loss": 0.0178, - "step": 2950 - }, - { - "epoch": 0.27, - "learning_rate": 2.5945205479452053e-05, - "loss": 0.0198, - "step": 2960 - }, - { - "epoch": 0.27, - "learning_rate": 2.5931506849315068e-05, - "loss": 0.0211, - "step": 2970 - }, - { - "epoch": 0.27, - "learning_rate": 2.5917808219178082e-05, - "loss": 0.0235, - "step": 2980 - }, - { - "epoch": 0.27, - "learning_rate": 2.5904109589041097e-05, - "loss": 0.0169, - "step": 2990 - }, - { - "epoch": 0.27, - "learning_rate": 2.589041095890411e-05, - "loss": 0.0142, - "step": 3000 - }, - { - "epoch": 0.27, - "learning_rate": 2.5876712328767123e-05, - "loss": 0.0132, - "step": 3010 - }, - { - "epoch": 0.28, - "learning_rate": 2.5863013698630137e-05, - "loss": 0.0132, - "step": 3020 - }, - { - "epoch": 0.28, - "learning_rate": 2.584931506849315e-05, - "loss": 0.0177, - "step": 3030 - }, - { - "epoch": 0.28, - "learning_rate": 2.5835616438356166e-05, - "loss": 0.0138, - "step": 3040 - }, - { - "epoch": 0.28, - "learning_rate": 2.582191780821918e-05, - "loss": 0.0167, - "step": 3050 - }, - { - "epoch": 0.28, - "learning_rate": 2.5808219178082192e-05, - "loss": 0.0466, - "step": 3060 - }, - { - "epoch": 0.28, - "learning_rate": 2.5794520547945207e-05, - "loss": 0.013, - "step": 3070 - }, - { - "epoch": 0.28, - "learning_rate": 2.5780821917808218e-05, - "loss": 0.0167, - "step": 3080 - }, - { - "epoch": 0.28, - "learning_rate": 2.5767123287671233e-05, - "loss": 0.0183, - "step": 3090 - }, - { - "epoch": 0.28, - "learning_rate": 2.5753424657534247e-05, - "loss": 0.017, - "step": 3100 - }, - { - "epoch": 0.28, - "learning_rate": 2.5739726027397262e-05, - "loss": 0.0176, - "step": 3110 - }, - { - "epoch": 0.28, - "learning_rate": 2.5726027397260276e-05, - "loss": 0.0202, - "step": 3120 - }, - { - "epoch": 0.29, - "learning_rate": 2.5712328767123287e-05, - "loss": 0.0196, - "step": 3130 - }, - { - "epoch": 0.29, - "learning_rate": 2.5698630136986302e-05, - "loss": 0.0167, - "step": 3140 - }, - { - "epoch": 0.29, - "learning_rate": 2.5684931506849317e-05, - "loss": 0.034, - "step": 3150 - }, - { - "epoch": 0.29, - "learning_rate": 2.5671232876712328e-05, - "loss": 0.0168, - "step": 3160 - }, - { - "epoch": 0.29, - "learning_rate": 2.5657534246575342e-05, - "loss": 0.0192, - "step": 3170 - }, - { - "epoch": 0.29, - "learning_rate": 2.5643835616438357e-05, - "loss": 0.0152, - "step": 3180 - }, - { - "epoch": 0.29, - "learning_rate": 2.563013698630137e-05, - "loss": 0.0143, - "step": 3190 - }, - { - "epoch": 0.29, - "learning_rate": 2.5616438356164386e-05, - "loss": 0.0178, - "step": 3200 - }, - { - "epoch": 0.29, - "learning_rate": 2.5602739726027397e-05, - "loss": 0.0182, - "step": 3210 - }, - { - "epoch": 0.29, - "learning_rate": 2.5589041095890412e-05, - "loss": 0.0151, - "step": 3220 - }, - { - "epoch": 0.29, - "learning_rate": 2.5575342465753423e-05, - "loss": 0.0158, - "step": 3230 - }, - { - "epoch": 0.3, - "learning_rate": 2.556164383561644e-05, - "loss": 0.0178, - "step": 3240 - }, - { - "epoch": 0.3, - "learning_rate": 2.5547945205479452e-05, - "loss": 0.0175, - "step": 3250 - }, - { - "epoch": 0.3, - "learning_rate": 2.5534246575342467e-05, - "loss": 0.016, - "step": 3260 - }, - { - "epoch": 0.3, - "learning_rate": 2.552054794520548e-05, - "loss": 0.0171, - "step": 3270 - }, - { - "epoch": 0.3, - "learning_rate": 2.5506849315068493e-05, - "loss": 0.0147, - "step": 3280 - }, - { - "epoch": 0.3, - "learning_rate": 2.5493150684931507e-05, - "loss": 0.0192, - "step": 3290 - }, - { - "epoch": 0.3, - "learning_rate": 2.5479452054794518e-05, - "loss": 0.0447, - "step": 3300 - }, - { - "epoch": 0.3, - "learning_rate": 2.5465753424657536e-05, - "loss": 0.0214, - "step": 3310 - }, - { - "epoch": 0.3, - "learning_rate": 2.545205479452055e-05, - "loss": 0.0165, - "step": 3320 - }, - { - "epoch": 0.3, - "learning_rate": 2.5438356164383562e-05, - "loss": 0.0149, - "step": 3330 - }, - { - "epoch": 0.31, - "learning_rate": 2.5424657534246577e-05, - "loss": 0.0184, - "step": 3340 - }, - { - "epoch": 0.31, - "learning_rate": 2.5410958904109588e-05, - "loss": 0.0182, - "step": 3350 - }, - { - "epoch": 0.31, - "learning_rate": 2.5397260273972602e-05, - "loss": 0.0193, - "step": 3360 - }, - { - "epoch": 0.31, - "learning_rate": 2.5383561643835617e-05, - "loss": 0.0212, - "step": 3370 - }, - { - "epoch": 0.31, - "learning_rate": 2.536986301369863e-05, - "loss": 0.0174, - "step": 3380 - }, - { - "epoch": 0.31, - "learning_rate": 2.5356164383561646e-05, - "loss": 0.0161, - "step": 3390 - }, - { - "epoch": 0.31, - "learning_rate": 2.5342465753424657e-05, - "loss": 0.019, - "step": 3400 - }, - { - "epoch": 0.31, - "learning_rate": 2.5328767123287672e-05, - "loss": 0.0277, - "step": 3410 - }, - { - "epoch": 0.31, - "learning_rate": 2.5315068493150686e-05, - "loss": 0.0165, - "step": 3420 - }, - { - "epoch": 0.31, - "learning_rate": 2.5301369863013698e-05, - "loss": 0.0138, - "step": 3430 - }, - { - "epoch": 0.31, - "learning_rate": 2.5287671232876716e-05, - "loss": 0.0185, - "step": 3440 - }, - { - "epoch": 0.32, - "learning_rate": 2.5273972602739727e-05, - "loss": 0.0192, - "step": 3450 - }, - { - "epoch": 0.32, - "learning_rate": 2.526027397260274e-05, - "loss": 0.0169, - "step": 3460 - }, - { - "epoch": 0.32, - "learning_rate": 2.5246575342465756e-05, - "loss": 0.0209, - "step": 3470 - }, - { - "epoch": 0.32, - "learning_rate": 2.5232876712328767e-05, - "loss": 0.0141, - "step": 3480 - }, - { - "epoch": 0.32, - "learning_rate": 2.521917808219178e-05, - "loss": 0.019, - "step": 3490 - }, - { - "epoch": 0.32, - "learning_rate": 2.5205479452054793e-05, - "loss": 0.0185, - "step": 3500 - }, - { - "epoch": 0.32, - "learning_rate": 2.519178082191781e-05, - "loss": 0.0216, - "step": 3510 - }, - { - "epoch": 0.32, - "learning_rate": 2.5178082191780822e-05, - "loss": 0.0183, - "step": 3520 - }, - { - "epoch": 0.32, - "learning_rate": 2.5164383561643836e-05, - "loss": 0.017, - "step": 3530 - }, - { - "epoch": 0.32, - "learning_rate": 2.515068493150685e-05, - "loss": 0.0201, - "step": 3540 - }, - { - "epoch": 0.32, - "learning_rate": 2.5136986301369862e-05, - "loss": 0.0143, - "step": 3550 - }, - { - "epoch": 0.33, - "learning_rate": 2.5123287671232877e-05, - "loss": 0.0164, - "step": 3560 - }, - { - "epoch": 0.33, - "learning_rate": 2.5109589041095888e-05, - "loss": 0.0144, - "step": 3570 - }, - { - "epoch": 0.33, - "learning_rate": 2.5095890410958906e-05, - "loss": 0.0174, - "step": 3580 - }, - { - "epoch": 0.33, - "learning_rate": 2.508219178082192e-05, - "loss": 0.0191, - "step": 3590 - }, - { - "epoch": 0.33, - "learning_rate": 2.5068493150684932e-05, - "loss": 0.0165, - "step": 3600 - }, - { - "epoch": 0.33, - "learning_rate": 2.5054794520547946e-05, - "loss": 0.0178, - "step": 3610 - }, - { - "epoch": 0.33, - "learning_rate": 2.5041095890410957e-05, - "loss": 0.0182, - "step": 3620 - }, - { - "epoch": 0.33, - "learning_rate": 2.5027397260273972e-05, - "loss": 0.0177, - "step": 3630 - }, - { - "epoch": 0.33, - "learning_rate": 2.501369863013699e-05, - "loss": 0.0149, - "step": 3640 - }, - { - "epoch": 0.33, - "learning_rate": 2.5e-05, - "loss": 0.015, - "step": 3650 - }, - { - "epoch": 0.33, - "learning_rate": 2.4986301369863016e-05, - "loss": 0.0183, - "step": 3660 - }, - { - "epoch": 0.34, - "learning_rate": 2.4972602739726027e-05, - "loss": 0.0235, - "step": 3670 - }, - { - "epoch": 0.34, - "learning_rate": 2.495890410958904e-05, - "loss": 0.0195, - "step": 3680 - }, - { - "epoch": 0.34, - "learning_rate": 2.4945205479452056e-05, - "loss": 0.0189, - "step": 3690 - }, - { - "epoch": 0.34, - "learning_rate": 2.4931506849315067e-05, - "loss": 0.0161, - "step": 3700 - }, - { - "epoch": 0.34, - "learning_rate": 2.4917808219178085e-05, - "loss": 0.0203, - "step": 3710 - }, - { - "epoch": 0.34, - "learning_rate": 2.4904109589041096e-05, - "loss": 0.0185, - "step": 3720 - }, - { - "epoch": 0.34, - "learning_rate": 2.489041095890411e-05, - "loss": 0.0241, - "step": 3730 - }, - { - "epoch": 0.34, - "learning_rate": 2.4876712328767122e-05, - "loss": 0.0156, - "step": 3740 - }, - { - "epoch": 0.34, - "learning_rate": 2.4863013698630137e-05, - "loss": 0.0173, - "step": 3750 - }, - { - "epoch": 0.34, - "learning_rate": 2.484931506849315e-05, - "loss": 0.0178, - "step": 3760 - }, - { - "epoch": 0.34, - "learning_rate": 2.4835616438356163e-05, - "loss": 0.0164, - "step": 3770 - }, - { - "epoch": 0.35, - "learning_rate": 2.482191780821918e-05, - "loss": 0.0533, - "step": 3780 - }, - { - "epoch": 0.35, - "learning_rate": 2.480821917808219e-05, - "loss": 0.0185, - "step": 3790 - }, - { - "epoch": 0.35, - "learning_rate": 2.4794520547945206e-05, - "loss": 0.0205, - "step": 3800 - }, - { - "epoch": 0.35, - "learning_rate": 2.478082191780822e-05, - "loss": 0.0157, - "step": 3810 - }, - { - "epoch": 0.35, - "learning_rate": 2.4767123287671232e-05, - "loss": 0.0153, - "step": 3820 - }, - { - "epoch": 0.35, - "learning_rate": 2.4753424657534247e-05, - "loss": 0.0125, - "step": 3830 - }, - { - "epoch": 0.35, - "learning_rate": 2.473972602739726e-05, - "loss": 0.0208, - "step": 3840 - }, - { - "epoch": 0.35, - "learning_rate": 2.4726027397260276e-05, - "loss": 0.0191, - "step": 3850 - }, - { - "epoch": 0.35, - "learning_rate": 2.471232876712329e-05, - "loss": 0.0179, - "step": 3860 - }, - { - "epoch": 0.35, - "learning_rate": 2.46986301369863e-05, - "loss": 0.0232, - "step": 3870 - }, - { - "epoch": 0.35, - "learning_rate": 2.4684931506849316e-05, - "loss": 0.0342, - "step": 3880 - }, - { - "epoch": 0.36, - "learning_rate": 2.4671232876712327e-05, - "loss": 0.0152, - "step": 3890 - }, - { - "epoch": 0.36, - "learning_rate": 2.4657534246575342e-05, - "loss": 0.0247, - "step": 3900 - }, - { - "epoch": 0.36, - "learning_rate": 2.464383561643836e-05, - "loss": 0.0164, - "step": 3910 - }, - { - "epoch": 0.36, - "learning_rate": 2.463013698630137e-05, - "loss": 0.016, - "step": 3920 - }, - { - "epoch": 0.36, - "learning_rate": 2.4616438356164386e-05, - "loss": 0.0179, - "step": 3930 - }, - { - "epoch": 0.36, - "learning_rate": 2.4602739726027397e-05, - "loss": 0.0213, - "step": 3940 - }, - { - "epoch": 0.36, - "learning_rate": 2.458904109589041e-05, - "loss": 0.0179, - "step": 3950 - }, - { - "epoch": 0.36, - "learning_rate": 2.4575342465753426e-05, - "loss": 0.0158, - "step": 3960 - }, - { - "epoch": 0.36, - "learning_rate": 2.4561643835616437e-05, - "loss": 0.0173, - "step": 3970 - }, - { - "epoch": 0.36, - "learning_rate": 2.4547945205479455e-05, - "loss": 0.0167, - "step": 3980 - }, - { - "epoch": 0.36, - "learning_rate": 2.4534246575342466e-05, - "loss": 0.0162, - "step": 3990 - }, - { - "epoch": 0.37, - "learning_rate": 2.452054794520548e-05, - "loss": 0.0173, - "step": 4000 - }, - { - "epoch": 0.37, - "learning_rate": 2.4506849315068492e-05, - "loss": 0.0152, - "step": 4010 - }, - { - "epoch": 0.37, - "learning_rate": 2.4493150684931506e-05, - "loss": 0.0197, - "step": 4020 - }, - { - "epoch": 0.37, - "learning_rate": 2.447945205479452e-05, - "loss": 0.0232, - "step": 4030 - }, - { - "epoch": 0.37, - "learning_rate": 2.4465753424657536e-05, - "loss": 0.0191, - "step": 4040 - }, - { - "epoch": 0.37, - "learning_rate": 2.445205479452055e-05, - "loss": 0.0198, - "step": 4050 - }, - { - "epoch": 0.37, - "learning_rate": 2.443835616438356e-05, - "loss": 0.0164, - "step": 4060 - }, - { - "epoch": 0.37, - "learning_rate": 2.4424657534246576e-05, - "loss": 0.0169, - "step": 4070 - }, - { - "epoch": 0.37, - "learning_rate": 2.441095890410959e-05, - "loss": 0.0167, - "step": 4080 - }, - { - "epoch": 0.37, - "learning_rate": 2.4397260273972602e-05, - "loss": 0.0378, - "step": 4090 - }, - { - "epoch": 0.37, - "learning_rate": 2.4383561643835616e-05, - "loss": 0.0162, - "step": 4100 - }, - { - "epoch": 0.38, - "learning_rate": 2.436986301369863e-05, - "loss": 0.0182, - "step": 4110 - }, - { - "epoch": 0.38, - "learning_rate": 2.4356164383561645e-05, - "loss": 0.013, - "step": 4120 - }, - { - "epoch": 0.38, - "learning_rate": 2.434246575342466e-05, - "loss": 0.019, - "step": 4130 - }, - { - "epoch": 0.38, - "learning_rate": 2.432876712328767e-05, - "loss": 0.0157, - "step": 4140 - }, - { - "epoch": 0.38, - "learning_rate": 2.4315068493150686e-05, - "loss": 0.0209, - "step": 4150 - }, - { - "epoch": 0.38, - "learning_rate": 2.4301369863013697e-05, - "loss": 0.0123, - "step": 4160 - }, - { - "epoch": 0.38, - "learning_rate": 2.428767123287671e-05, - "loss": 0.0784, - "step": 4170 - }, - { - "epoch": 0.38, - "learning_rate": 2.427397260273973e-05, - "loss": 0.0184, - "step": 4180 - }, - { - "epoch": 0.38, - "learning_rate": 2.426027397260274e-05, - "loss": 0.0139, - "step": 4190 - }, - { - "epoch": 0.38, - "learning_rate": 2.4246575342465755e-05, - "loss": 0.0149, - "step": 4200 - }, - { - "epoch": 0.38, - "learning_rate": 2.4232876712328766e-05, - "loss": 0.0173, - "step": 4210 - }, - { - "epoch": 0.39, - "learning_rate": 2.421917808219178e-05, - "loss": 0.0316, - "step": 4220 - }, - { - "epoch": 0.39, - "learning_rate": 2.4205479452054796e-05, - "loss": 0.0156, - "step": 4230 - }, - { - "epoch": 0.39, - "learning_rate": 2.419178082191781e-05, - "loss": 0.023, - "step": 4240 - }, - { - "epoch": 0.39, - "learning_rate": 2.4178082191780825e-05, - "loss": 0.0163, - "step": 4250 - }, - { - "epoch": 0.39, - "learning_rate": 2.4164383561643836e-05, - "loss": 0.0229, - "step": 4260 - }, - { - "epoch": 0.39, - "learning_rate": 2.415068493150685e-05, - "loss": 0.018, - "step": 4270 - }, - { - "epoch": 0.39, - "learning_rate": 2.413698630136986e-05, - "loss": 0.0149, - "step": 4280 - }, - { - "epoch": 0.39, - "learning_rate": 2.4123287671232876e-05, - "loss": 0.0217, - "step": 4290 - }, - { - "epoch": 0.39, - "learning_rate": 2.410958904109589e-05, - "loss": 0.0153, - "step": 4300 - }, - { - "epoch": 0.39, - "learning_rate": 2.4095890410958905e-05, - "loss": 0.017, - "step": 4310 - }, - { - "epoch": 0.39, - "learning_rate": 2.408219178082192e-05, - "loss": 0.0267, - "step": 4320 - }, - { - "epoch": 0.4, - "learning_rate": 2.406849315068493e-05, - "loss": 0.0168, - "step": 4330 - }, - { - "epoch": 0.4, - "learning_rate": 2.4054794520547946e-05, - "loss": 0.0155, - "step": 4340 - }, - { - "epoch": 0.4, - "learning_rate": 2.404109589041096e-05, - "loss": 0.0155, - "step": 4350 - }, - { - "epoch": 0.4, - "learning_rate": 2.402739726027397e-05, - "loss": 0.0137, - "step": 4360 - }, - { - "epoch": 0.4, - "learning_rate": 2.4013698630136986e-05, - "loss": 0.016, - "step": 4370 - }, - { - "epoch": 0.4, - "learning_rate": 2.4e-05, - "loss": 0.0176, - "step": 4380 - }, - { - "epoch": 0.4, - "learning_rate": 2.3986301369863015e-05, - "loss": 0.0195, - "step": 4390 - }, - { - "epoch": 0.4, - "learning_rate": 2.397260273972603e-05, - "loss": 0.0126, - "step": 4400 - }, - { - "epoch": 0.4, - "learning_rate": 2.395890410958904e-05, - "loss": 0.0163, - "step": 4410 - }, - { - "epoch": 0.4, - "learning_rate": 2.3945205479452056e-05, - "loss": 0.0137, - "step": 4420 - }, - { - "epoch": 0.4, - "learning_rate": 2.3931506849315067e-05, - "loss": 0.0167, - "step": 4430 - }, - { - "epoch": 0.41, - "learning_rate": 2.391780821917808e-05, - "loss": 0.0232, - "step": 4440 - }, - { - "epoch": 0.41, - "learning_rate": 2.39041095890411e-05, - "loss": 0.0232, - "step": 4450 - }, - { - "epoch": 0.41, - "learning_rate": 2.389041095890411e-05, - "loss": 0.018, - "step": 4460 - }, - { - "epoch": 0.41, - "learning_rate": 2.3876712328767125e-05, - "loss": 0.0142, - "step": 4470 - }, - { - "epoch": 0.41, - "learning_rate": 2.3863013698630136e-05, - "loss": 0.0153, - "step": 4480 - }, - { - "epoch": 0.41, - "learning_rate": 2.384931506849315e-05, - "loss": 0.015, - "step": 4490 - }, - { - "epoch": 0.41, - "learning_rate": 2.3835616438356165e-05, - "loss": 0.0212, - "step": 4500 - }, - { - "epoch": 0.41, - "learning_rate": 2.382191780821918e-05, - "loss": 0.017, - "step": 4510 - }, - { - "epoch": 0.41, - "learning_rate": 2.3808219178082194e-05, - "loss": 0.0176, - "step": 4520 - }, - { - "epoch": 0.41, - "learning_rate": 2.3794520547945206e-05, - "loss": 0.015, - "step": 4530 - }, - { - "epoch": 0.41, - "learning_rate": 2.378082191780822e-05, - "loss": 0.0197, - "step": 4540 - }, - { - "epoch": 0.42, - "learning_rate": 2.376712328767123e-05, - "loss": 0.0191, - "step": 4550 - }, - { - "epoch": 0.42, - "learning_rate": 2.3753424657534246e-05, - "loss": 0.0174, - "step": 4560 - }, - { - "epoch": 0.42, - "learning_rate": 2.373972602739726e-05, - "loss": 0.0268, - "step": 4570 - }, - { - "epoch": 0.42, - "learning_rate": 2.3726027397260275e-05, - "loss": 0.0177, - "step": 4580 - }, - { - "epoch": 0.42, - "learning_rate": 2.371232876712329e-05, - "loss": 0.019, - "step": 4590 - }, - { - "epoch": 0.42, - "learning_rate": 2.36986301369863e-05, - "loss": 0.0401, - "step": 4600 - }, - { - "epoch": 0.42, - "learning_rate": 2.3684931506849315e-05, - "loss": 0.0144, - "step": 4610 - }, - { - "epoch": 0.42, - "learning_rate": 2.367123287671233e-05, - "loss": 0.0175, - "step": 4620 - }, - { - "epoch": 0.42, - "learning_rate": 2.365753424657534e-05, - "loss": 0.0178, - "step": 4630 - }, - { - "epoch": 0.42, - "learning_rate": 2.3643835616438356e-05, - "loss": 0.0163, - "step": 4640 - }, - { - "epoch": 0.42, - "learning_rate": 2.363013698630137e-05, - "loss": 0.0163, - "step": 4650 - }, - { - "epoch": 0.43, - "learning_rate": 2.3616438356164385e-05, - "loss": 0.018, - "step": 4660 - }, - { - "epoch": 0.43, - "learning_rate": 2.36027397260274e-05, - "loss": 0.0217, - "step": 4670 - }, - { - "epoch": 0.43, - "learning_rate": 2.358904109589041e-05, - "loss": 0.0192, - "step": 4680 - }, - { - "epoch": 0.43, - "learning_rate": 2.3575342465753425e-05, - "loss": 0.0179, - "step": 4690 - }, - { - "epoch": 0.43, - "learning_rate": 2.3561643835616436e-05, - "loss": 0.0175, - "step": 4700 - }, - { - "epoch": 0.43, - "learning_rate": 2.3547945205479454e-05, - "loss": 0.0158, - "step": 4710 - }, - { - "epoch": 0.43, - "learning_rate": 2.353424657534247e-05, - "loss": 0.0151, - "step": 4720 - }, - { - "epoch": 0.43, - "learning_rate": 2.352054794520548e-05, - "loss": 0.0199, - "step": 4730 - }, - { - "epoch": 0.43, - "learning_rate": 2.3506849315068495e-05, - "loss": 0.0209, - "step": 4740 - }, - { - "epoch": 0.43, - "learning_rate": 2.3493150684931506e-05, - "loss": 0.0113, - "step": 4750 - }, - { - "epoch": 0.43, - "learning_rate": 2.347945205479452e-05, - "loss": 0.0154, - "step": 4760 - }, - { - "epoch": 0.44, - "learning_rate": 2.3465753424657535e-05, - "loss": 0.0219, - "step": 4770 - }, - { - "epoch": 0.44, - "learning_rate": 2.345205479452055e-05, - "loss": 0.0183, - "step": 4780 - }, - { - "epoch": 0.44, - "learning_rate": 2.3438356164383564e-05, - "loss": 0.0158, - "step": 4790 - }, - { - "epoch": 0.44, - "learning_rate": 2.3424657534246575e-05, - "loss": 0.0205, - "step": 4800 - }, - { - "epoch": 0.44, - "learning_rate": 2.341095890410959e-05, - "loss": 0.0162, - "step": 4810 - }, - { - "epoch": 0.44, - "learning_rate": 2.33972602739726e-05, - "loss": 0.0129, - "step": 4820 - }, - { - "epoch": 0.44, - "learning_rate": 2.3383561643835616e-05, - "loss": 0.014, - "step": 4830 - }, - { - "epoch": 0.44, - "learning_rate": 2.336986301369863e-05, - "loss": 0.0165, - "step": 4840 - }, - { - "epoch": 0.44, - "learning_rate": 2.3356164383561645e-05, - "loss": 0.0166, - "step": 4850 - }, - { - "epoch": 0.44, - "learning_rate": 2.334246575342466e-05, - "loss": 0.0151, - "step": 4860 - }, - { - "epoch": 0.44, - "learning_rate": 2.332876712328767e-05, - "loss": 0.0184, - "step": 4870 - }, - { - "epoch": 0.45, - "learning_rate": 2.3315068493150685e-05, - "loss": 0.0132, - "step": 4880 - }, - { - "epoch": 0.45, - "learning_rate": 2.33013698630137e-05, - "loss": 0.0139, - "step": 4890 - }, - { - "epoch": 0.45, - "learning_rate": 2.328767123287671e-05, - "loss": 0.0189, - "step": 4900 - }, - { - "epoch": 0.45, - "learning_rate": 2.327397260273973e-05, - "loss": 0.0458, - "step": 4910 - }, - { - "epoch": 0.45, - "learning_rate": 2.326027397260274e-05, - "loss": 0.0143, - "step": 4920 - }, - { - "epoch": 0.45, - "learning_rate": 2.3246575342465755e-05, - "loss": 0.0162, - "step": 4930 - }, - { - "epoch": 0.45, - "learning_rate": 2.323287671232877e-05, - "loss": 0.0194, - "step": 4940 - }, - { - "epoch": 0.45, - "learning_rate": 2.321917808219178e-05, - "loss": 0.0188, - "step": 4950 - }, - { - "epoch": 0.45, - "learning_rate": 2.3205479452054795e-05, - "loss": 0.0158, - "step": 4960 - }, - { - "epoch": 0.45, - "learning_rate": 2.3191780821917806e-05, - "loss": 0.0156, - "step": 4970 - }, - { - "epoch": 0.45, - "learning_rate": 2.3178082191780824e-05, - "loss": 0.0151, - "step": 4980 - }, - { - "epoch": 0.46, - "learning_rate": 2.316438356164384e-05, - "loss": 0.0181, - "step": 4990 - }, - { - "epoch": 0.46, - "learning_rate": 2.315068493150685e-05, - "loss": 0.0187, - "step": 5000 - }, - { - "epoch": 0.46, - "learning_rate": 2.3136986301369864e-05, - "loss": 0.017, - "step": 5010 - }, - { - "epoch": 0.46, - "learning_rate": 2.3123287671232876e-05, - "loss": 0.019, - "step": 5020 - }, - { - "epoch": 0.46, - "learning_rate": 2.310958904109589e-05, - "loss": 0.0143, - "step": 5030 - }, - { - "epoch": 0.46, - "learning_rate": 2.30958904109589e-05, - "loss": 0.0155, - "step": 5040 - }, - { - "epoch": 0.46, - "learning_rate": 2.308219178082192e-05, - "loss": 0.0173, - "step": 5050 - }, - { - "epoch": 0.46, - "learning_rate": 2.3068493150684934e-05, - "loss": 0.0188, - "step": 5060 - }, - { - "epoch": 0.46, - "learning_rate": 2.3054794520547945e-05, - "loss": 0.0189, - "step": 5070 - }, - { - "epoch": 0.46, - "learning_rate": 2.304109589041096e-05, - "loss": 0.0161, - "step": 5080 - }, - { - "epoch": 0.46, - "learning_rate": 2.302739726027397e-05, - "loss": 0.0143, - "step": 5090 - }, - { - "epoch": 0.47, - "learning_rate": 2.3013698630136985e-05, - "loss": 0.0123, - "step": 5100 - }, - { - "epoch": 0.47, - "learning_rate": 2.3000000000000003e-05, - "loss": 0.0234, - "step": 5110 - }, - { - "epoch": 0.47, - "learning_rate": 2.2986301369863015e-05, - "loss": 0.0149, - "step": 5120 - }, - { - "epoch": 0.47, - "learning_rate": 2.297260273972603e-05, - "loss": 0.0106, - "step": 5130 - }, - { - "epoch": 0.47, - "learning_rate": 2.295890410958904e-05, - "loss": 0.0269, - "step": 5140 - }, - { - "epoch": 0.47, - "learning_rate": 2.2945205479452055e-05, - "loss": 0.038, - "step": 5150 - }, - { - "epoch": 0.47, - "learning_rate": 2.293150684931507e-05, - "loss": 0.0138, - "step": 5160 - }, - { - "epoch": 0.47, - "learning_rate": 2.291780821917808e-05, - "loss": 0.0148, - "step": 5170 - }, - { - "epoch": 0.47, - "learning_rate": 2.29041095890411e-05, - "loss": 0.017, - "step": 5180 - }, - { - "epoch": 0.47, - "learning_rate": 2.289041095890411e-05, - "loss": 0.0227, - "step": 5190 - }, - { - "epoch": 0.47, - "learning_rate": 2.2876712328767124e-05, - "loss": 0.0164, - "step": 5200 - }, - { - "epoch": 0.48, - "learning_rate": 2.286301369863014e-05, - "loss": 0.0146, - "step": 5210 - }, - { - "epoch": 0.48, - "learning_rate": 2.284931506849315e-05, - "loss": 0.013, - "step": 5220 - }, - { - "epoch": 0.48, - "learning_rate": 2.2835616438356165e-05, - "loss": 0.0152, - "step": 5230 - }, - { - "epoch": 0.48, - "learning_rate": 2.2821917808219176e-05, - "loss": 0.0168, - "step": 5240 - }, - { - "epoch": 0.48, - "learning_rate": 2.2808219178082194e-05, - "loss": 0.0139, - "step": 5250 - }, - { - "epoch": 0.48, - "learning_rate": 2.279452054794521e-05, - "loss": 0.0171, - "step": 5260 - }, - { - "epoch": 0.48, - "learning_rate": 2.278082191780822e-05, - "loss": 0.0204, - "step": 5270 - }, - { - "epoch": 0.48, - "learning_rate": 2.2767123287671234e-05, - "loss": 0.0198, - "step": 5280 - }, - { - "epoch": 0.48, - "learning_rate": 2.2753424657534245e-05, - "loss": 0.0182, - "step": 5290 - }, - { - "epoch": 0.48, - "learning_rate": 2.273972602739726e-05, - "loss": 0.0167, - "step": 5300 - }, - { - "epoch": 0.48, - "learning_rate": 2.2726027397260275e-05, - "loss": 0.0145, - "step": 5310 - }, - { - "epoch": 0.49, - "learning_rate": 2.271232876712329e-05, - "loss": 0.0425, - "step": 5320 - }, - { - "epoch": 0.49, - "learning_rate": 2.2698630136986304e-05, - "loss": 0.0246, - "step": 5330 - }, - { - "epoch": 0.49, - "learning_rate": 2.2684931506849315e-05, - "loss": 0.0146, - "step": 5340 - }, - { - "epoch": 0.49, - "learning_rate": 2.267123287671233e-05, - "loss": 0.0151, - "step": 5350 - }, - { - "epoch": 0.49, - "learning_rate": 2.265753424657534e-05, - "loss": 0.017, - "step": 5360 - }, - { - "epoch": 0.49, - "learning_rate": 2.2643835616438355e-05, - "loss": 0.015, - "step": 5370 - }, - { - "epoch": 0.49, - "learning_rate": 2.2630136986301373e-05, - "loss": 0.0158, - "step": 5380 - }, - { - "epoch": 0.49, - "learning_rate": 2.2616438356164384e-05, - "loss": 0.0165, - "step": 5390 - }, - { - "epoch": 0.49, - "learning_rate": 2.26027397260274e-05, - "loss": 0.0174, - "step": 5400 - }, - { - "epoch": 0.49, - "learning_rate": 2.258904109589041e-05, - "loss": 0.0162, - "step": 5410 - }, - { - "epoch": 0.49, - "learning_rate": 2.2575342465753425e-05, - "loss": 0.017, - "step": 5420 - }, - { - "epoch": 0.5, - "learning_rate": 2.256164383561644e-05, - "loss": 0.0138, - "step": 5430 - }, - { - "epoch": 0.5, - "learning_rate": 2.254794520547945e-05, - "loss": 0.0207, - "step": 5440 - }, - { - "epoch": 0.5, - "learning_rate": 2.253424657534247e-05, - "loss": 0.0178, - "step": 5450 - }, - { - "epoch": 0.5, - "learning_rate": 2.252054794520548e-05, - "loss": 0.0177, - "step": 5460 - }, - { - "epoch": 0.5, - "learning_rate": 2.2506849315068494e-05, - "loss": 0.0156, - "step": 5470 - }, - { - "epoch": 0.5, - "learning_rate": 2.249315068493151e-05, - "loss": 0.0181, - "step": 5480 - }, - { - "epoch": 0.5, - "learning_rate": 2.247945205479452e-05, - "loss": 0.0155, - "step": 5490 - }, - { - "epoch": 0.5, - "learning_rate": 2.2465753424657534e-05, - "loss": 0.0364, - "step": 5500 - }, - { - "epoch": 0.5, - "learning_rate": 2.245205479452055e-05, - "loss": 0.0118, - "step": 5510 - }, - { - "epoch": 0.5, - "learning_rate": 2.2438356164383564e-05, - "loss": 0.0184, - "step": 5520 - }, - { - "epoch": 0.51, - "learning_rate": 2.2424657534246578e-05, - "loss": 0.0171, - "step": 5530 - }, - { - "epoch": 0.51, - "learning_rate": 2.241095890410959e-05, - "loss": 0.0157, - "step": 5540 - }, - { - "epoch": 0.51, - "learning_rate": 2.2397260273972604e-05, - "loss": 0.0141, - "step": 5550 - }, - { - "epoch": 0.51, - "learning_rate": 2.2383561643835615e-05, - "loss": 0.0114, - "step": 5560 - }, - { - "epoch": 0.51, - "learning_rate": 2.236986301369863e-05, - "loss": 0.0155, - "step": 5570 - }, - { - "epoch": 0.51, - "learning_rate": 2.2356164383561644e-05, - "loss": 0.0322, - "step": 5580 - }, - { - "epoch": 0.51, - "learning_rate": 2.234246575342466e-05, - "loss": 0.0173, - "step": 5590 - }, - { - "epoch": 0.51, - "learning_rate": 2.2328767123287673e-05, - "loss": 0.0181, - "step": 5600 - }, - { - "epoch": 0.51, - "learning_rate": 2.2315068493150685e-05, - "loss": 0.0135, - "step": 5610 - }, - { - "epoch": 0.51, - "learning_rate": 2.23013698630137e-05, - "loss": 0.0166, - "step": 5620 - }, - { - "epoch": 0.51, - "learning_rate": 2.228767123287671e-05, - "loss": 0.0212, - "step": 5630 - }, - { - "epoch": 0.52, - "learning_rate": 2.2273972602739725e-05, - "loss": 0.0154, - "step": 5640 - }, - { - "epoch": 0.52, - "learning_rate": 2.2260273972602743e-05, - "loss": 0.0171, - "step": 5650 - }, - { - "epoch": 0.52, - "learning_rate": 2.2246575342465754e-05, - "loss": 0.0149, - "step": 5660 - }, - { - "epoch": 0.52, - "learning_rate": 2.223287671232877e-05, - "loss": 0.0135, - "step": 5670 - }, - { - "epoch": 0.52, - "learning_rate": 2.221917808219178e-05, - "loss": 0.0223, - "step": 5680 - }, - { - "epoch": 0.52, - "learning_rate": 2.2205479452054794e-05, - "loss": 0.0142, - "step": 5690 - }, - { - "epoch": 0.52, - "learning_rate": 2.219178082191781e-05, - "loss": 0.0149, - "step": 5700 - }, - { - "epoch": 0.52, - "learning_rate": 2.2178082191780824e-05, - "loss": 0.0125, - "step": 5710 - }, - { - "epoch": 0.52, - "learning_rate": 2.2164383561643838e-05, - "loss": 0.036, - "step": 5720 - }, - { - "epoch": 0.52, - "learning_rate": 2.215068493150685e-05, - "loss": 0.0351, - "step": 5730 - }, - { - "epoch": 0.52, - "learning_rate": 2.2136986301369864e-05, - "loss": 0.0309, - "step": 5740 - }, - { - "epoch": 0.53, - "learning_rate": 2.212328767123288e-05, - "loss": 0.0146, - "step": 5750 - }, - { - "epoch": 0.53, - "learning_rate": 2.210958904109589e-05, - "loss": 0.0137, - "step": 5760 - }, - { - "epoch": 0.53, - "learning_rate": 2.2095890410958904e-05, - "loss": 0.0171, - "step": 5770 - }, - { - "epoch": 0.53, - "learning_rate": 2.208219178082192e-05, - "loss": 0.0147, - "step": 5780 - }, - { - "epoch": 0.53, - "learning_rate": 2.2068493150684933e-05, - "loss": 0.0411, - "step": 5790 - }, - { - "epoch": 0.53, - "learning_rate": 2.2054794520547945e-05, - "loss": 0.0232, - "step": 5800 - }, - { - "epoch": 0.53, - "learning_rate": 2.204109589041096e-05, - "loss": 0.0161, - "step": 5810 - }, - { - "epoch": 0.53, - "learning_rate": 2.2027397260273974e-05, - "loss": 0.0181, - "step": 5820 - }, - { - "epoch": 0.53, - "learning_rate": 2.2013698630136985e-05, - "loss": 0.0189, - "step": 5830 - }, - { - "epoch": 0.53, - "learning_rate": 2.2e-05, - "loss": 0.012, - "step": 5840 - }, - { - "epoch": 0.53, - "learning_rate": 2.1986301369863014e-05, - "loss": 0.0163, - "step": 5850 - }, - { - "epoch": 0.54, - "learning_rate": 2.197260273972603e-05, - "loss": 0.0156, - "step": 5860 - }, - { - "epoch": 0.54, - "learning_rate": 2.1958904109589043e-05, - "loss": 0.074, - "step": 5870 - }, - { - "epoch": 0.54, - "learning_rate": 2.1945205479452054e-05, - "loss": 0.0162, - "step": 5880 - }, - { - "epoch": 0.54, - "learning_rate": 2.193150684931507e-05, - "loss": 0.0145, - "step": 5890 - }, - { - "epoch": 0.54, - "learning_rate": 2.191780821917808e-05, - "loss": 0.0288, - "step": 5900 - }, - { - "epoch": 0.54, - "learning_rate": 2.1904109589041098e-05, - "loss": 0.0166, - "step": 5910 - }, - { - "epoch": 0.54, - "learning_rate": 2.1890410958904113e-05, - "loss": 0.0184, - "step": 5920 - }, - { - "epoch": 0.54, - "learning_rate": 2.1876712328767124e-05, - "loss": 0.0554, - "step": 5930 - }, - { - "epoch": 0.54, - "learning_rate": 2.186301369863014e-05, - "loss": 0.021, - "step": 5940 - }, - { - "epoch": 0.54, - "learning_rate": 2.184931506849315e-05, - "loss": 0.0158, - "step": 5950 - }, - { - "epoch": 0.54, - "learning_rate": 2.1835616438356164e-05, - "loss": 0.0201, - "step": 5960 - }, - { - "epoch": 0.55, - "learning_rate": 2.182191780821918e-05, - "loss": 0.0165, - "step": 5970 - }, - { - "epoch": 0.55, - "learning_rate": 2.1808219178082193e-05, - "loss": 0.016, - "step": 5980 - }, - { - "epoch": 0.55, - "learning_rate": 2.1794520547945208e-05, - "loss": 0.021, - "step": 5990 - }, - { - "epoch": 0.55, - "learning_rate": 2.178082191780822e-05, - "loss": 0.0165, - "step": 6000 - }, - { - "epoch": 0.55, - "learning_rate": 2.1767123287671234e-05, - "loss": 0.0317, - "step": 6010 - }, - { - "epoch": 0.55, - "learning_rate": 2.1753424657534248e-05, - "loss": 0.018, - "step": 6020 - }, - { - "epoch": 0.55, - "learning_rate": 2.173972602739726e-05, - "loss": 0.0156, - "step": 6030 - }, - { - "epoch": 0.55, - "learning_rate": 2.1726027397260274e-05, - "loss": 0.014, - "step": 6040 - }, - { - "epoch": 0.55, - "learning_rate": 2.171232876712329e-05, - "loss": 0.0131, - "step": 6050 - }, - { - "epoch": 0.55, - "learning_rate": 2.1698630136986303e-05, - "loss": 0.0125, - "step": 6060 - }, - { - "epoch": 0.55, - "learning_rate": 2.1684931506849314e-05, - "loss": 0.0119, - "step": 6070 - }, - { - "epoch": 0.56, - "learning_rate": 2.167123287671233e-05, - "loss": 0.0141, - "step": 6080 - }, - { - "epoch": 0.56, - "learning_rate": 2.1657534246575343e-05, - "loss": 0.0168, - "step": 6090 - }, - { - "epoch": 0.56, - "learning_rate": 2.1643835616438355e-05, - "loss": 0.0151, - "step": 6100 - }, - { - "epoch": 0.56, - "learning_rate": 2.1630136986301373e-05, - "loss": 0.0158, - "step": 6110 - }, - { - "epoch": 0.56, - "learning_rate": 2.1616438356164384e-05, - "loss": 0.0154, - "step": 6120 - }, - { - "epoch": 0.56, - "learning_rate": 2.16027397260274e-05, - "loss": 0.015, - "step": 6130 - }, - { - "epoch": 0.56, - "learning_rate": 2.1589041095890413e-05, - "loss": 0.0144, - "step": 6140 - }, - { - "epoch": 0.56, - "learning_rate": 2.1575342465753424e-05, - "loss": 0.015, - "step": 6150 - }, - { - "epoch": 0.56, - "learning_rate": 2.156164383561644e-05, - "loss": 0.0147, - "step": 6160 - }, - { - "epoch": 0.56, - "learning_rate": 2.154794520547945e-05, - "loss": 0.0139, - "step": 6170 - }, - { - "epoch": 0.56, - "learning_rate": 2.1534246575342468e-05, - "loss": 0.0135, - "step": 6180 - }, - { - "epoch": 0.57, - "learning_rate": 2.1520547945205482e-05, - "loss": 0.0112, - "step": 6190 - }, - { - "epoch": 0.57, - "learning_rate": 2.1506849315068494e-05, - "loss": 0.0174, - "step": 6200 - }, - { - "epoch": 0.57, - "learning_rate": 2.1493150684931508e-05, - "loss": 0.0375, - "step": 6210 - }, - { - "epoch": 0.57, - "learning_rate": 2.147945205479452e-05, - "loss": 0.0168, - "step": 6220 - }, - { - "epoch": 0.57, - "learning_rate": 2.1465753424657534e-05, - "loss": 0.0147, - "step": 6230 - }, - { - "epoch": 0.57, - "learning_rate": 2.145205479452055e-05, - "loss": 0.0136, - "step": 6240 - }, - { - "epoch": 0.57, - "learning_rate": 2.1438356164383563e-05, - "loss": 0.0172, - "step": 6250 - }, - { - "epoch": 0.57, - "learning_rate": 2.1424657534246578e-05, - "loss": 0.0182, - "step": 6260 - }, - { - "epoch": 0.57, - "learning_rate": 2.141095890410959e-05, - "loss": 0.0165, - "step": 6270 - }, - { - "epoch": 0.57, - "learning_rate": 2.1397260273972603e-05, - "loss": 0.0153, - "step": 6280 - }, - { - "epoch": 0.57, - "learning_rate": 2.1383561643835618e-05, - "loss": 0.0104, - "step": 6290 - }, - { - "epoch": 0.58, - "learning_rate": 2.136986301369863e-05, - "loss": 0.0108, - "step": 6300 - }, - { - "epoch": 0.58, - "learning_rate": 2.1356164383561647e-05, - "loss": 0.0145, - "step": 6310 - }, - { - "epoch": 0.58, - "learning_rate": 2.1342465753424658e-05, - "loss": 0.0139, - "step": 6320 - }, - { - "epoch": 0.58, - "learning_rate": 2.1328767123287673e-05, - "loss": 0.0147, - "step": 6330 - }, - { - "epoch": 0.58, - "learning_rate": 2.1315068493150684e-05, - "loss": 0.0185, - "step": 6340 - }, - { - "epoch": 0.58, - "learning_rate": 2.13013698630137e-05, - "loss": 0.0178, - "step": 6350 - }, - { - "epoch": 0.58, - "learning_rate": 2.1287671232876713e-05, - "loss": 0.0151, - "step": 6360 - }, - { - "epoch": 0.58, - "learning_rate": 2.1273972602739724e-05, - "loss": 0.0113, - "step": 6370 - }, - { - "epoch": 0.58, - "learning_rate": 2.1260273972602742e-05, - "loss": 0.0196, - "step": 6380 - }, - { - "epoch": 0.58, - "learning_rate": 2.1246575342465753e-05, - "loss": 0.0158, - "step": 6390 - }, - { - "epoch": 0.58, - "learning_rate": 2.1232876712328768e-05, - "loss": 0.015, - "step": 6400 - }, - { - "epoch": 0.59, - "learning_rate": 2.1219178082191783e-05, - "loss": 0.0178, - "step": 6410 - }, - { - "epoch": 0.59, - "learning_rate": 2.1205479452054794e-05, - "loss": 0.0118, - "step": 6420 - }, - { - "epoch": 0.59, - "learning_rate": 2.119178082191781e-05, - "loss": 0.0209, - "step": 6430 - }, - { - "epoch": 0.59, - "learning_rate": 2.117808219178082e-05, - "loss": 0.0269, - "step": 6440 - }, - { - "epoch": 0.59, - "learning_rate": 2.1164383561643838e-05, - "loss": 0.0186, - "step": 6450 - }, - { - "epoch": 0.59, - "learning_rate": 2.1150684931506852e-05, - "loss": 0.0127, - "step": 6460 - }, - { - "epoch": 0.59, - "learning_rate": 2.1136986301369863e-05, - "loss": 0.0148, - "step": 6470 - }, - { - "epoch": 0.59, - "learning_rate": 2.1123287671232878e-05, - "loss": 0.0265, - "step": 6480 - }, - { - "epoch": 0.59, - "learning_rate": 2.110958904109589e-05, - "loss": 0.0515, - "step": 6490 - }, - { - "epoch": 0.59, - "learning_rate": 2.1095890410958904e-05, - "loss": 0.015, - "step": 6500 - }, - { - "epoch": 0.59, - "learning_rate": 2.108219178082192e-05, - "loss": 0.0162, - "step": 6510 - }, - { - "epoch": 0.6, - "learning_rate": 2.1068493150684933e-05, - "loss": 0.0185, - "step": 6520 - }, - { - "epoch": 0.6, - "learning_rate": 2.1054794520547947e-05, - "loss": 0.0155, - "step": 6530 - }, - { - "epoch": 0.6, - "learning_rate": 2.104109589041096e-05, - "loss": 0.0288, - "step": 6540 - }, - { - "epoch": 0.6, - "learning_rate": 2.1027397260273973e-05, - "loss": 0.0127, - "step": 6550 - }, - { - "epoch": 0.6, - "learning_rate": 2.1013698630136988e-05, - "loss": 0.0133, - "step": 6560 - }, - { - "epoch": 0.6, - "learning_rate": 2.1e-05, - "loss": 0.0198, - "step": 6570 - }, - { - "epoch": 0.6, - "learning_rate": 2.0986301369863017e-05, - "loss": 0.0128, - "step": 6580 - }, - { - "epoch": 0.6, - "learning_rate": 2.0972602739726028e-05, - "loss": 0.015, - "step": 6590 - }, - { - "epoch": 0.6, - "learning_rate": 2.0958904109589043e-05, - "loss": 0.0176, - "step": 6600 - }, - { - "epoch": 0.6, - "learning_rate": 2.0945205479452054e-05, - "loss": 0.014, - "step": 6610 - }, - { - "epoch": 0.6, - "learning_rate": 2.093150684931507e-05, - "loss": 0.0147, - "step": 6620 - }, - { - "epoch": 0.61, - "learning_rate": 2.0917808219178083e-05, - "loss": 0.0266, - "step": 6630 - }, - { - "epoch": 0.61, - "learning_rate": 2.0904109589041094e-05, - "loss": 0.0135, - "step": 6640 - }, - { - "epoch": 0.61, - "learning_rate": 2.0890410958904112e-05, - "loss": 0.0148, - "step": 6650 - }, - { - "epoch": 0.61, - "learning_rate": 2.0876712328767123e-05, - "loss": 0.0162, - "step": 6660 - }, - { - "epoch": 0.61, - "learning_rate": 2.0863013698630138e-05, - "loss": 0.0162, - "step": 6670 - }, - { - "epoch": 0.61, - "learning_rate": 2.0849315068493152e-05, - "loss": 0.02, - "step": 6680 - }, - { - "epoch": 0.61, - "learning_rate": 2.0835616438356164e-05, - "loss": 0.0142, - "step": 6690 - }, - { - "epoch": 0.61, - "learning_rate": 2.0821917808219178e-05, - "loss": 0.0153, - "step": 6700 - }, - { - "epoch": 0.61, - "learning_rate": 2.080821917808219e-05, - "loss": 0.0128, - "step": 6710 - }, - { - "epoch": 0.61, - "learning_rate": 2.0794520547945207e-05, - "loss": 0.0111, - "step": 6720 - }, - { - "epoch": 0.61, - "learning_rate": 2.0780821917808222e-05, - "loss": 0.016, - "step": 6730 - }, - { - "epoch": 0.62, - "learning_rate": 2.0767123287671233e-05, - "loss": 0.0163, - "step": 6740 - }, - { - "epoch": 0.62, - "learning_rate": 2.0753424657534248e-05, - "loss": 0.0358, - "step": 6750 - }, - { - "epoch": 0.62, - "learning_rate": 2.073972602739726e-05, - "loss": 0.0151, - "step": 6760 - }, - { - "epoch": 0.62, - "learning_rate": 2.0726027397260273e-05, - "loss": 0.012, - "step": 6770 - }, - { - "epoch": 0.62, - "learning_rate": 2.071232876712329e-05, - "loss": 0.0199, - "step": 6780 - }, - { - "epoch": 0.62, - "learning_rate": 2.0698630136986302e-05, - "loss": 0.0134, - "step": 6790 - }, - { - "epoch": 0.62, - "learning_rate": 2.0684931506849317e-05, - "loss": 0.0147, - "step": 6800 - }, - { - "epoch": 0.62, - "learning_rate": 2.0671232876712328e-05, - "loss": 0.0132, - "step": 6810 - }, - { - "epoch": 0.62, - "learning_rate": 2.0657534246575343e-05, - "loss": 0.0156, - "step": 6820 - }, - { - "epoch": 0.62, - "learning_rate": 2.0643835616438357e-05, - "loss": 0.0187, - "step": 6830 - }, - { - "epoch": 0.62, - "learning_rate": 2.063013698630137e-05, - "loss": 0.0126, - "step": 6840 - }, - { - "epoch": 0.63, - "learning_rate": 2.0616438356164387e-05, - "loss": 0.0129, - "step": 6850 - }, - { - "epoch": 0.63, - "learning_rate": 2.0602739726027398e-05, - "loss": 0.0166, - "step": 6860 - }, - { - "epoch": 0.63, - "learning_rate": 2.0589041095890412e-05, - "loss": 0.0148, - "step": 6870 - }, - { - "epoch": 0.63, - "learning_rate": 2.0575342465753423e-05, - "loss": 0.0172, - "step": 6880 - }, - { - "epoch": 0.63, - "learning_rate": 2.0561643835616438e-05, - "loss": 0.039, - "step": 6890 - }, - { - "epoch": 0.63, - "learning_rate": 2.0547945205479453e-05, - "loss": 0.0143, - "step": 6900 - }, - { - "epoch": 0.63, - "learning_rate": 2.0534246575342464e-05, - "loss": 0.0116, - "step": 6910 - }, - { - "epoch": 0.63, - "learning_rate": 2.0520547945205482e-05, - "loss": 0.0159, - "step": 6920 - }, - { - "epoch": 0.63, - "learning_rate": 2.0506849315068493e-05, - "loss": 0.0196, - "step": 6930 - }, - { - "epoch": 0.63, - "learning_rate": 2.0493150684931508e-05, - "loss": 0.0158, - "step": 6940 - }, - { - "epoch": 0.63, - "learning_rate": 2.0479452054794522e-05, - "loss": 0.0194, - "step": 6950 - }, - { - "epoch": 0.64, - "learning_rate": 2.0465753424657533e-05, - "loss": 0.0154, - "step": 6960 - }, - { - "epoch": 0.64, - "learning_rate": 2.0452054794520548e-05, - "loss": 0.0168, - "step": 6970 - }, - { - "epoch": 0.64, - "learning_rate": 2.0438356164383562e-05, - "loss": 0.019, - "step": 6980 - }, - { - "epoch": 0.64, - "learning_rate": 2.0424657534246577e-05, - "loss": 0.0161, - "step": 6990 - }, - { - "epoch": 0.64, - "learning_rate": 2.041095890410959e-05, - "loss": 0.0148, - "step": 7000 - }, - { - "epoch": 0.64, - "learning_rate": 2.0397260273972603e-05, - "loss": 0.0136, - "step": 7010 - }, - { - "epoch": 0.64, - "learning_rate": 2.0383561643835617e-05, - "loss": 0.018, - "step": 7020 - }, - { - "epoch": 0.64, - "learning_rate": 2.036986301369863e-05, - "loss": 0.0133, - "step": 7030 - }, - { - "epoch": 0.64, - "learning_rate": 2.0356164383561643e-05, - "loss": 0.0168, - "step": 7040 - }, - { - "epoch": 0.64, - "learning_rate": 2.034246575342466e-05, - "loss": 0.0152, - "step": 7050 - }, - { - "epoch": 0.64, - "learning_rate": 2.0328767123287672e-05, - "loss": 0.0192, - "step": 7060 - }, - { - "epoch": 0.65, - "learning_rate": 2.0315068493150687e-05, - "loss": 0.0139, - "step": 7070 - }, - { - "epoch": 0.65, - "learning_rate": 2.0301369863013698e-05, - "loss": 0.0099, - "step": 7080 - }, - { - "epoch": 0.65, - "learning_rate": 2.0287671232876713e-05, - "loss": 0.0179, - "step": 7090 - }, - { - "epoch": 0.65, - "learning_rate": 2.0273972602739724e-05, - "loss": 0.0114, - "step": 7100 - }, - { - "epoch": 0.65, - "learning_rate": 2.026027397260274e-05, - "loss": 0.0134, - "step": 7110 - }, - { - "epoch": 0.65, - "learning_rate": 2.0246575342465756e-05, - "loss": 0.0141, - "step": 7120 - }, - { - "epoch": 0.65, - "learning_rate": 2.0232876712328767e-05, - "loss": 0.0159, - "step": 7130 - }, - { - "epoch": 0.65, - "learning_rate": 2.0219178082191782e-05, - "loss": 0.0134, - "step": 7140 - }, - { - "epoch": 0.65, - "learning_rate": 2.0205479452054793e-05, - "loss": 0.0168, - "step": 7150 - }, - { - "epoch": 0.65, - "learning_rate": 2.0191780821917808e-05, - "loss": 0.0194, - "step": 7160 - }, - { - "epoch": 0.65, - "learning_rate": 2.0178082191780822e-05, - "loss": 0.0187, - "step": 7170 - }, - { - "epoch": 0.66, - "learning_rate": 2.0164383561643837e-05, - "loss": 0.018, - "step": 7180 - }, - { - "epoch": 0.66, - "learning_rate": 2.015068493150685e-05, - "loss": 0.0187, - "step": 7190 - }, - { - "epoch": 0.66, - "learning_rate": 2.0136986301369863e-05, - "loss": 0.0175, - "step": 7200 - }, - { - "epoch": 0.66, - "learning_rate": 2.0123287671232877e-05, - "loss": 0.0171, - "step": 7210 - }, - { - "epoch": 0.66, - "learning_rate": 2.0109589041095892e-05, - "loss": 0.0141, - "step": 7220 - }, - { - "epoch": 0.66, - "learning_rate": 2.0095890410958903e-05, - "loss": 0.0141, - "step": 7230 - }, - { - "epoch": 0.66, - "learning_rate": 2.0082191780821918e-05, - "loss": 0.0256, - "step": 7240 - }, - { - "epoch": 0.66, - "learning_rate": 2.0068493150684932e-05, - "loss": 0.0142, - "step": 7250 - }, - { - "epoch": 0.66, - "learning_rate": 2.0054794520547947e-05, - "loss": 0.0164, - "step": 7260 - }, - { - "epoch": 0.66, - "learning_rate": 2.004109589041096e-05, - "loss": 0.0126, - "step": 7270 - }, - { - "epoch": 0.66, - "learning_rate": 2.0027397260273972e-05, - "loss": 0.0112, - "step": 7280 - }, - { - "epoch": 0.67, - "learning_rate": 2.0013698630136987e-05, - "loss": 0.0154, - "step": 7290 - }, - { - "epoch": 0.67, - "learning_rate": 1.9999999999999998e-05, - "loss": 0.0129, - "step": 7300 - }, - { - "epoch": 0.67, - "learning_rate": 1.9986301369863013e-05, - "loss": 0.0133, - "step": 7310 - }, - { - "epoch": 0.67, - "learning_rate": 1.997260273972603e-05, - "loss": 0.0137, - "step": 7320 - }, - { - "epoch": 0.67, - "learning_rate": 1.9958904109589042e-05, - "loss": 0.0119, - "step": 7330 - }, - { - "epoch": 0.67, - "learning_rate": 1.9945205479452057e-05, - "loss": 0.0134, - "step": 7340 - }, - { - "epoch": 0.67, - "learning_rate": 1.9931506849315068e-05, - "loss": 0.0156, - "step": 7350 - }, - { - "epoch": 0.67, - "learning_rate": 1.9917808219178082e-05, - "loss": 0.0159, - "step": 7360 - }, - { - "epoch": 0.67, - "learning_rate": 1.9904109589041093e-05, - "loss": 0.0168, - "step": 7370 - }, - { - "epoch": 0.67, - "learning_rate": 1.989041095890411e-05, - "loss": 0.0147, - "step": 7380 - }, - { - "epoch": 0.67, - "learning_rate": 1.9876712328767126e-05, - "loss": 0.0162, - "step": 7390 - }, - { - "epoch": 0.68, - "learning_rate": 1.9863013698630137e-05, - "loss": 0.0098, - "step": 7400 - }, - { - "epoch": 0.68, - "learning_rate": 1.9849315068493152e-05, - "loss": 0.0191, - "step": 7410 - }, - { - "epoch": 0.68, - "learning_rate": 1.9835616438356163e-05, - "loss": 0.0126, - "step": 7420 - }, - { - "epoch": 0.68, - "learning_rate": 1.9821917808219178e-05, - "loss": 0.0164, - "step": 7430 - }, - { - "epoch": 0.68, - "learning_rate": 1.9808219178082192e-05, - "loss": 0.0136, - "step": 7440 - }, - { - "epoch": 0.68, - "learning_rate": 1.9794520547945207e-05, - "loss": 0.0132, - "step": 7450 - }, - { - "epoch": 0.68, - "learning_rate": 1.978082191780822e-05, - "loss": 0.0131, - "step": 7460 - }, - { - "epoch": 0.68, - "learning_rate": 1.9767123287671232e-05, - "loss": 0.0149, - "step": 7470 - }, - { - "epoch": 0.68, - "learning_rate": 1.9753424657534247e-05, - "loss": 0.0095, - "step": 7480 - }, - { - "epoch": 0.68, - "learning_rate": 1.973972602739726e-05, - "loss": 0.0239, - "step": 7490 - }, - { - "epoch": 0.68, - "learning_rate": 1.9726027397260273e-05, - "loss": 0.0173, - "step": 7500 - }, - { - "epoch": 0.69, - "learning_rate": 1.9712328767123287e-05, - "loss": 0.018, - "step": 7510 - }, - { - "epoch": 0.69, - "learning_rate": 1.9698630136986302e-05, - "loss": 0.0161, - "step": 7520 - }, - { - "epoch": 0.69, - "learning_rate": 1.9684931506849316e-05, - "loss": 0.0185, - "step": 7530 - }, - { - "epoch": 0.69, - "learning_rate": 1.967123287671233e-05, - "loss": 0.0098, - "step": 7540 - }, - { - "epoch": 0.69, - "learning_rate": 1.9657534246575342e-05, - "loss": 0.016, - "step": 7550 - }, - { - "epoch": 0.69, - "learning_rate": 1.9643835616438357e-05, - "loss": 0.0141, - "step": 7560 - }, - { - "epoch": 0.69, - "learning_rate": 1.9630136986301368e-05, - "loss": 0.0162, - "step": 7570 - }, - { - "epoch": 0.69, - "learning_rate": 1.9616438356164386e-05, - "loss": 0.0154, - "step": 7580 - }, - { - "epoch": 0.69, - "learning_rate": 1.96027397260274e-05, - "loss": 0.0123, - "step": 7590 - }, - { - "epoch": 0.69, - "learning_rate": 1.9589041095890412e-05, - "loss": 0.0171, - "step": 7600 - }, - { - "epoch": 0.69, - "learning_rate": 1.9575342465753426e-05, - "loss": 0.017, - "step": 7610 - }, - { - "epoch": 0.7, - "learning_rate": 1.9561643835616437e-05, - "loss": 0.0222, - "step": 7620 - }, - { - "epoch": 0.7, - "learning_rate": 1.9547945205479452e-05, - "loss": 0.0109, - "step": 7630 - }, - { - "epoch": 0.7, - "learning_rate": 1.9534246575342463e-05, - "loss": 0.0765, - "step": 7640 - }, - { - "epoch": 0.7, - "learning_rate": 1.952054794520548e-05, - "loss": 0.0149, - "step": 7650 - }, - { - "epoch": 0.7, - "learning_rate": 1.9506849315068496e-05, - "loss": 0.0135, - "step": 7660 - }, - { - "epoch": 0.7, - "learning_rate": 1.9493150684931507e-05, - "loss": 0.0138, - "step": 7670 - }, - { - "epoch": 0.7, - "learning_rate": 1.947945205479452e-05, - "loss": 0.0172, - "step": 7680 - }, - { - "epoch": 0.7, - "learning_rate": 1.9465753424657533e-05, - "loss": 0.0143, - "step": 7690 - }, - { - "epoch": 0.7, - "learning_rate": 1.9452054794520547e-05, - "loss": 0.013, - "step": 7700 - }, - { - "epoch": 0.7, - "learning_rate": 1.9438356164383562e-05, - "loss": 0.0161, - "step": 7710 - }, - { - "epoch": 0.71, - "learning_rate": 1.9424657534246576e-05, - "loss": 0.0178, - "step": 7720 - }, - { - "epoch": 0.71, - "learning_rate": 1.941095890410959e-05, - "loss": 0.0186, - "step": 7730 - }, - { - "epoch": 0.71, - "learning_rate": 1.9397260273972602e-05, - "loss": 0.0112, - "step": 7740 - }, - { - "epoch": 0.71, - "learning_rate": 1.9383561643835617e-05, - "loss": 0.0138, - "step": 7750 - }, - { - "epoch": 0.71, - "learning_rate": 1.936986301369863e-05, - "loss": 0.0175, - "step": 7760 - }, - { - "epoch": 0.71, - "learning_rate": 1.9356164383561642e-05, - "loss": 0.0107, - "step": 7770 - }, - { - "epoch": 0.71, - "learning_rate": 1.934246575342466e-05, - "loss": 0.0338, - "step": 7780 - }, - { - "epoch": 0.71, - "learning_rate": 1.932876712328767e-05, - "loss": 0.016, - "step": 7790 - }, - { - "epoch": 0.71, - "learning_rate": 1.9315068493150686e-05, - "loss": 0.0483, - "step": 7800 - }, - { - "epoch": 0.71, - "learning_rate": 1.93013698630137e-05, - "loss": 0.0214, - "step": 7810 - }, - { - "epoch": 0.71, - "learning_rate": 1.9287671232876712e-05, - "loss": 0.016, - "step": 7820 - }, - { - "epoch": 0.72, - "learning_rate": 1.9273972602739727e-05, - "loss": 0.0146, - "step": 7830 - }, - { - "epoch": 0.72, - "learning_rate": 1.9260273972602738e-05, - "loss": 0.0148, - "step": 7840 - }, - { - "epoch": 0.72, - "learning_rate": 1.9246575342465756e-05, - "loss": 0.0161, - "step": 7850 - }, - { - "epoch": 0.72, - "learning_rate": 1.9232876712328767e-05, - "loss": 0.016, - "step": 7860 - }, - { - "epoch": 0.72, - "learning_rate": 1.921917808219178e-05, - "loss": 0.0123, - "step": 7870 - }, - { - "epoch": 0.72, - "learning_rate": 1.9205479452054796e-05, - "loss": 0.0138, - "step": 7880 - }, - { - "epoch": 0.72, - "learning_rate": 1.9191780821917807e-05, - "loss": 0.0148, - "step": 7890 - }, - { - "epoch": 0.72, - "learning_rate": 1.9178082191780822e-05, - "loss": 0.0152, - "step": 7900 - }, - { - "epoch": 0.72, - "learning_rate": 1.9164383561643833e-05, - "loss": 0.0151, - "step": 7910 - }, - { - "epoch": 0.72, - "learning_rate": 1.915068493150685e-05, - "loss": 0.0148, - "step": 7920 - }, - { - "epoch": 0.72, - "learning_rate": 1.9136986301369865e-05, - "loss": 0.0111, - "step": 7930 - }, - { - "epoch": 0.73, - "learning_rate": 1.9123287671232877e-05, - "loss": 0.0101, - "step": 7940 - }, - { - "epoch": 0.73, - "learning_rate": 1.910958904109589e-05, - "loss": 0.0185, - "step": 7950 - }, - { - "epoch": 0.73, - "learning_rate": 1.9095890410958902e-05, - "loss": 0.0132, - "step": 7960 - }, - { - "epoch": 0.73, - "learning_rate": 1.9082191780821917e-05, - "loss": 0.0133, - "step": 7970 - }, - { - "epoch": 0.73, - "learning_rate": 1.9068493150684935e-05, - "loss": 0.0116, - "step": 7980 - }, - { - "epoch": 0.73, - "learning_rate": 1.9054794520547946e-05, - "loss": 0.015, - "step": 7990 - }, - { - "epoch": 0.73, - "learning_rate": 1.904109589041096e-05, - "loss": 0.0362, - "step": 8000 - }, - { - "epoch": 0.73, - "learning_rate": 1.9027397260273972e-05, - "loss": 0.0143, - "step": 8010 - }, - { - "epoch": 0.73, - "learning_rate": 1.9013698630136986e-05, - "loss": 0.012, - "step": 8020 - }, - { - "epoch": 0.73, - "learning_rate": 1.9e-05, - "loss": 0.0138, - "step": 8030 - }, - { - "epoch": 0.73, - "learning_rate": 1.8986301369863012e-05, - "loss": 0.0187, - "step": 8040 - }, - { - "epoch": 0.74, - "learning_rate": 1.897260273972603e-05, - "loss": 0.0174, - "step": 8050 - }, - { - "epoch": 0.74, - "learning_rate": 1.895890410958904e-05, - "loss": 0.0128, - "step": 8060 - }, - { - "epoch": 0.74, - "learning_rate": 1.8945205479452056e-05, - "loss": 0.0128, - "step": 8070 - }, - { - "epoch": 0.74, - "learning_rate": 1.893150684931507e-05, - "loss": 0.0158, - "step": 8080 - }, - { - "epoch": 0.74, - "learning_rate": 1.8917808219178082e-05, - "loss": 0.0154, - "step": 8090 - }, - { - "epoch": 0.74, - "learning_rate": 1.8904109589041096e-05, - "loss": 0.0649, - "step": 8100 - }, - { - "epoch": 0.74, - "learning_rate": 1.8890410958904107e-05, - "loss": 0.0144, - "step": 8110 - }, - { - "epoch": 0.74, - "learning_rate": 1.8876712328767125e-05, - "loss": 0.0152, - "step": 8120 - }, - { - "epoch": 0.74, - "learning_rate": 1.8863013698630137e-05, - "loss": 0.0118, - "step": 8130 - }, - { - "epoch": 0.74, - "learning_rate": 1.884931506849315e-05, - "loss": 0.0131, - "step": 8140 - }, - { - "epoch": 0.74, - "learning_rate": 1.8835616438356166e-05, - "loss": 0.0146, - "step": 8150 - }, - { - "epoch": 0.75, - "learning_rate": 1.8821917808219177e-05, - "loss": 0.0131, - "step": 8160 - }, - { - "epoch": 0.75, - "learning_rate": 1.880821917808219e-05, - "loss": 0.0104, - "step": 8170 - }, - { - "epoch": 0.75, - "learning_rate": 1.8794520547945206e-05, - "loss": 0.0153, - "step": 8180 - }, - { - "epoch": 0.75, - "learning_rate": 1.878082191780822e-05, - "loss": 0.0171, - "step": 8190 - }, - { - "epoch": 0.75, - "learning_rate": 1.8767123287671235e-05, - "loss": 0.015, - "step": 8200 - }, - { - "epoch": 0.75, - "learning_rate": 1.8753424657534246e-05, - "loss": 0.0162, - "step": 8210 - }, - { - "epoch": 0.75, - "learning_rate": 1.873972602739726e-05, - "loss": 0.0148, - "step": 8220 - }, - { - "epoch": 0.75, - "learning_rate": 1.8726027397260272e-05, - "loss": 0.0148, - "step": 8230 - }, - { - "epoch": 0.75, - "learning_rate": 1.8712328767123287e-05, - "loss": 0.0123, - "step": 8240 - }, - { - "epoch": 0.75, - "learning_rate": 1.8698630136986305e-05, - "loss": 0.0135, - "step": 8250 - }, - { - "epoch": 0.75, - "learning_rate": 1.8684931506849316e-05, - "loss": 0.0183, - "step": 8260 - }, - { - "epoch": 0.76, - "learning_rate": 1.867123287671233e-05, - "loss": 0.0157, - "step": 8270 - }, - { - "epoch": 0.76, - "learning_rate": 1.865753424657534e-05, - "loss": 0.0376, - "step": 8280 - }, - { - "epoch": 0.76, - "learning_rate": 1.8643835616438356e-05, - "loss": 0.0178, - "step": 8290 - }, - { - "epoch": 0.76, - "learning_rate": 1.863013698630137e-05, - "loss": 0.0111, - "step": 8300 - }, - { - "epoch": 0.76, - "learning_rate": 1.8616438356164382e-05, - "loss": 0.0159, - "step": 8310 - }, - { - "epoch": 0.76, - "learning_rate": 1.86027397260274e-05, - "loss": 0.0211, - "step": 8320 - }, - { - "epoch": 0.76, - "learning_rate": 1.858904109589041e-05, - "loss": 0.0118, - "step": 8330 - }, - { - "epoch": 0.76, - "learning_rate": 1.8575342465753426e-05, - "loss": 0.0159, - "step": 8340 - }, - { - "epoch": 0.76, - "learning_rate": 1.856164383561644e-05, - "loss": 0.0144, - "step": 8350 - }, - { - "epoch": 0.76, - "learning_rate": 1.854794520547945e-05, - "loss": 0.0123, - "step": 8360 - }, - { - "epoch": 0.76, - "learning_rate": 1.8534246575342466e-05, - "loss": 0.0185, - "step": 8370 - }, - { - "epoch": 0.77, - "learning_rate": 1.852054794520548e-05, - "loss": 0.0108, - "step": 8380 - }, - { - "epoch": 0.77, - "learning_rate": 1.8506849315068495e-05, - "loss": 0.0172, - "step": 8390 - }, - { - "epoch": 0.77, - "learning_rate": 1.8493150684931506e-05, - "loss": 0.0161, - "step": 8400 - }, - { - "epoch": 0.77, - "learning_rate": 1.847945205479452e-05, - "loss": 0.064, - "step": 8410 - }, - { - "epoch": 0.77, - "learning_rate": 1.8465753424657535e-05, - "loss": 0.0132, - "step": 8420 - }, - { - "epoch": 0.77, - "learning_rate": 1.8452054794520547e-05, - "loss": 0.0122, - "step": 8430 - }, - { - "epoch": 0.77, - "learning_rate": 1.843835616438356e-05, - "loss": 0.0145, - "step": 8440 - }, - { - "epoch": 0.77, - "learning_rate": 1.8424657534246576e-05, - "loss": 0.0269, - "step": 8450 - }, - { - "epoch": 0.77, - "learning_rate": 1.841095890410959e-05, - "loss": 0.0178, - "step": 8460 - }, - { - "epoch": 0.77, - "learning_rate": 1.8397260273972605e-05, - "loss": 0.01, - "step": 8470 - }, - { - "epoch": 0.77, - "learning_rate": 1.8383561643835616e-05, - "loss": 0.0164, - "step": 8480 - }, - { - "epoch": 0.78, - "learning_rate": 1.836986301369863e-05, - "loss": 0.0099, - "step": 8490 - }, - { - "epoch": 0.78, - "learning_rate": 1.8356164383561642e-05, - "loss": 0.0349, - "step": 8500 - }, - { - "epoch": 0.78, - "learning_rate": 1.8342465753424656e-05, - "loss": 0.0319, - "step": 8510 - }, - { - "epoch": 0.78, - "learning_rate": 1.8328767123287674e-05, - "loss": 0.0155, - "step": 8520 - }, - { - "epoch": 0.78, - "learning_rate": 1.8315068493150686e-05, - "loss": 0.0177, - "step": 8530 - }, - { - "epoch": 0.78, - "learning_rate": 1.83013698630137e-05, - "loss": 0.0132, - "step": 8540 - }, - { - "epoch": 0.78, - "learning_rate": 1.828767123287671e-05, - "loss": 0.0149, - "step": 8550 - }, - { - "epoch": 0.78, - "learning_rate": 1.8273972602739726e-05, - "loss": 0.012, - "step": 8560 - }, - { - "epoch": 0.78, - "learning_rate": 1.826027397260274e-05, - "loss": 0.0156, - "step": 8570 - }, - { - "epoch": 0.78, - "learning_rate": 1.8246575342465755e-05, - "loss": 0.0185, - "step": 8580 - }, - { - "epoch": 0.78, - "learning_rate": 1.823287671232877e-05, - "loss": 0.0178, - "step": 8590 - }, - { - "epoch": 0.79, - "learning_rate": 1.821917808219178e-05, - "loss": 0.0121, - "step": 8600 - }, - { - "epoch": 0.79, - "learning_rate": 1.8205479452054795e-05, - "loss": 0.015, - "step": 8610 - }, - { - "epoch": 0.79, - "learning_rate": 1.819178082191781e-05, - "loss": 0.0107, - "step": 8620 - }, - { - "epoch": 0.79, - "learning_rate": 1.817808219178082e-05, - "loss": 0.016, - "step": 8630 - }, - { - "epoch": 0.79, - "learning_rate": 1.8164383561643836e-05, - "loss": 0.0143, - "step": 8640 - }, - { - "epoch": 0.79, - "learning_rate": 1.815068493150685e-05, - "loss": 0.0264, - "step": 8650 - }, - { - "epoch": 0.79, - "learning_rate": 1.8136986301369865e-05, - "loss": 0.0162, - "step": 8660 - }, - { - "epoch": 0.79, - "learning_rate": 1.8123287671232876e-05, - "loss": 0.0157, - "step": 8670 - }, - { - "epoch": 0.79, - "learning_rate": 1.810958904109589e-05, - "loss": 0.0118, - "step": 8680 - }, - { - "epoch": 0.79, - "learning_rate": 1.8095890410958905e-05, - "loss": 0.0138, - "step": 8690 - }, - { - "epoch": 0.79, - "learning_rate": 1.8082191780821916e-05, - "loss": 0.0146, - "step": 8700 - }, - { - "epoch": 0.8, - "learning_rate": 1.806849315068493e-05, - "loss": 0.0511, - "step": 8710 - }, - { - "epoch": 0.8, - "learning_rate": 1.8054794520547946e-05, - "loss": 0.0154, - "step": 8720 - }, - { - "epoch": 0.8, - "learning_rate": 1.804109589041096e-05, - "loss": 0.0143, - "step": 8730 - }, - { - "epoch": 0.8, - "learning_rate": 1.8027397260273975e-05, - "loss": 0.0154, - "step": 8740 - }, - { - "epoch": 0.8, - "learning_rate": 1.8013698630136986e-05, - "loss": 0.0149, - "step": 8750 - }, - { - "epoch": 0.8, - "learning_rate": 1.8e-05, - "loss": 0.0157, - "step": 8760 - }, - { - "epoch": 0.8, - "learning_rate": 1.798630136986301e-05, - "loss": 0.0185, - "step": 8770 - }, - { - "epoch": 0.8, - "learning_rate": 1.7972602739726026e-05, - "loss": 0.0144, - "step": 8780 - }, - { - "epoch": 0.8, - "learning_rate": 1.7958904109589044e-05, - "loss": 0.0146, - "step": 8790 - }, - { - "epoch": 0.8, - "learning_rate": 1.7945205479452055e-05, - "loss": 0.0409, - "step": 8800 - }, - { - "epoch": 0.8, - "learning_rate": 1.793150684931507e-05, - "loss": 0.0202, - "step": 8810 - }, - { - "epoch": 0.81, - "learning_rate": 1.791780821917808e-05, - "loss": 0.0121, - "step": 8820 - }, - { - "epoch": 0.81, - "learning_rate": 1.7904109589041096e-05, - "loss": 0.0122, - "step": 8830 - }, - { - "epoch": 0.81, - "learning_rate": 1.789041095890411e-05, - "loss": 0.0137, - "step": 8840 - }, - { - "epoch": 0.81, - "learning_rate": 1.7876712328767125e-05, - "loss": 0.0147, - "step": 8850 - }, - { - "epoch": 0.81, - "learning_rate": 1.786301369863014e-05, - "loss": 0.0159, - "step": 8860 - }, - { - "epoch": 0.81, - "learning_rate": 1.784931506849315e-05, - "loss": 0.0133, - "step": 8870 - }, - { - "epoch": 0.81, - "learning_rate": 1.7835616438356165e-05, - "loss": 0.0158, - "step": 8880 - }, - { - "epoch": 0.81, - "learning_rate": 1.782191780821918e-05, - "loss": 0.0139, - "step": 8890 - }, - { - "epoch": 0.81, - "learning_rate": 1.780821917808219e-05, - "loss": 0.0152, - "step": 8900 - }, - { - "epoch": 0.81, - "learning_rate": 1.7794520547945205e-05, - "loss": 0.0151, - "step": 8910 - }, - { - "epoch": 0.81, - "learning_rate": 1.778082191780822e-05, - "loss": 0.0121, - "step": 8920 - }, - { - "epoch": 0.82, - "learning_rate": 1.7767123287671235e-05, - "loss": 0.0145, - "step": 8930 - }, - { - "epoch": 0.82, - "learning_rate": 1.7753424657534246e-05, - "loss": 0.0144, - "step": 8940 - }, - { - "epoch": 0.82, - "learning_rate": 1.773972602739726e-05, - "loss": 0.014, - "step": 8950 - }, - { - "epoch": 0.82, - "learning_rate": 1.7726027397260275e-05, - "loss": 0.0134, - "step": 8960 - }, - { - "epoch": 0.82, - "learning_rate": 1.7712328767123286e-05, - "loss": 0.0152, - "step": 8970 - }, - { - "epoch": 0.82, - "learning_rate": 1.76986301369863e-05, - "loss": 0.0132, - "step": 8980 - }, - { - "epoch": 0.82, - "learning_rate": 1.7684931506849315e-05, - "loss": 0.0199, - "step": 8990 - }, - { - "epoch": 0.82, - "learning_rate": 1.767123287671233e-05, - "loss": 0.0206, - "step": 9000 - }, - { - "epoch": 0.82, - "learning_rate": 1.7657534246575344e-05, - "loss": 0.014, - "step": 9010 - }, - { - "epoch": 0.82, - "learning_rate": 1.7643835616438356e-05, - "loss": 0.0212, - "step": 9020 - }, - { - "epoch": 0.82, - "learning_rate": 1.763013698630137e-05, - "loss": 0.0165, - "step": 9030 - }, - { - "epoch": 0.83, - "learning_rate": 1.761643835616438e-05, - "loss": 0.0153, - "step": 9040 - }, - { - "epoch": 0.83, - "learning_rate": 1.76027397260274e-05, - "loss": 0.0148, - "step": 9050 - }, - { - "epoch": 0.83, - "learning_rate": 1.7589041095890414e-05, - "loss": 0.0091, - "step": 9060 - }, - { - "epoch": 0.83, - "learning_rate": 1.7575342465753425e-05, - "loss": 0.0165, - "step": 9070 - }, - { - "epoch": 0.83, - "learning_rate": 1.756164383561644e-05, - "loss": 0.0156, - "step": 9080 - }, - { - "epoch": 0.83, - "learning_rate": 1.754794520547945e-05, - "loss": 0.0116, - "step": 9090 - }, - { - "epoch": 0.83, - "learning_rate": 1.7534246575342465e-05, - "loss": 0.0173, - "step": 9100 - }, - { - "epoch": 0.83, - "learning_rate": 1.752054794520548e-05, - "loss": 0.0121, - "step": 9110 - }, - { - "epoch": 0.83, - "learning_rate": 1.7506849315068495e-05, - "loss": 0.0132, - "step": 9120 - }, - { - "epoch": 0.83, - "learning_rate": 1.749315068493151e-05, - "loss": 0.015, - "step": 9130 - }, - { - "epoch": 0.83, - "learning_rate": 1.747945205479452e-05, - "loss": 0.0285, - "step": 9140 - }, - { - "epoch": 0.84, - "learning_rate": 1.7465753424657535e-05, - "loss": 0.0167, - "step": 9150 - }, - { - "epoch": 0.84, - "learning_rate": 1.7452054794520546e-05, - "loss": 0.0361, - "step": 9160 - }, - { - "epoch": 0.84, - "learning_rate": 1.743835616438356e-05, - "loss": 0.0138, - "step": 9170 - }, - { - "epoch": 0.84, - "learning_rate": 1.7424657534246575e-05, - "loss": 0.0164, - "step": 9180 - }, - { - "epoch": 0.84, - "learning_rate": 1.741095890410959e-05, - "loss": 0.013, - "step": 9190 - }, - { - "epoch": 0.84, - "learning_rate": 1.7397260273972604e-05, - "loss": 0.0125, - "step": 9200 - }, - { - "epoch": 0.84, - "learning_rate": 1.7383561643835616e-05, - "loss": 0.0191, - "step": 9210 - }, - { - "epoch": 0.84, - "learning_rate": 1.736986301369863e-05, - "loss": 0.0138, - "step": 9220 - }, - { - "epoch": 0.84, - "learning_rate": 1.7356164383561645e-05, - "loss": 0.0124, - "step": 9230 - }, - { - "epoch": 0.84, - "learning_rate": 1.7342465753424656e-05, - "loss": 0.0128, - "step": 9240 - }, - { - "epoch": 0.84, - "learning_rate": 1.7328767123287674e-05, - "loss": 0.0135, - "step": 9250 - }, - { - "epoch": 0.85, - "learning_rate": 1.7315068493150685e-05, - "loss": 0.0154, - "step": 9260 - }, - { - "epoch": 0.85, - "learning_rate": 1.73013698630137e-05, - "loss": 0.0116, - "step": 9270 - }, - { - "epoch": 0.85, - "learning_rate": 1.7287671232876714e-05, - "loss": 0.0175, - "step": 9280 - }, - { - "epoch": 0.85, - "learning_rate": 1.7273972602739725e-05, - "loss": 0.0139, - "step": 9290 - }, - { - "epoch": 0.85, - "learning_rate": 1.726027397260274e-05, - "loss": 0.0109, - "step": 9300 - }, - { - "epoch": 0.85, - "learning_rate": 1.724657534246575e-05, - "loss": 0.014, - "step": 9310 - }, - { - "epoch": 0.85, - "learning_rate": 1.723287671232877e-05, - "loss": 0.026, - "step": 9320 - }, - { - "epoch": 0.85, - "learning_rate": 1.7219178082191784e-05, - "loss": 0.0153, - "step": 9330 - }, - { - "epoch": 0.85, - "learning_rate": 1.7205479452054795e-05, - "loss": 0.0157, - "step": 9340 - }, - { - "epoch": 0.85, - "learning_rate": 1.719178082191781e-05, - "loss": 0.0164, - "step": 9350 - }, - { - "epoch": 0.85, - "learning_rate": 1.717808219178082e-05, - "loss": 0.0136, - "step": 9360 - }, - { - "epoch": 0.86, - "learning_rate": 1.7164383561643835e-05, - "loss": 0.0165, - "step": 9370 - }, - { - "epoch": 0.86, - "learning_rate": 1.715068493150685e-05, - "loss": 0.012, - "step": 9380 - }, - { - "epoch": 0.86, - "learning_rate": 1.7136986301369864e-05, - "loss": 0.0119, - "step": 9390 - }, - { - "epoch": 0.86, - "learning_rate": 1.712328767123288e-05, - "loss": 0.0189, - "step": 9400 - }, - { - "epoch": 0.86, - "learning_rate": 1.710958904109589e-05, - "loss": 0.0116, - "step": 9410 - }, - { - "epoch": 0.86, - "learning_rate": 1.7095890410958905e-05, - "loss": 0.0149, - "step": 9420 - }, - { - "epoch": 0.86, - "learning_rate": 1.7082191780821916e-05, - "loss": 0.0187, - "step": 9430 - }, - { - "epoch": 0.86, - "learning_rate": 1.706849315068493e-05, - "loss": 0.0166, - "step": 9440 - }, - { - "epoch": 0.86, - "learning_rate": 1.705479452054795e-05, - "loss": 0.0138, - "step": 9450 - }, - { - "epoch": 0.86, - "learning_rate": 1.704109589041096e-05, - "loss": 0.0136, - "step": 9460 - }, - { - "epoch": 0.86, - "learning_rate": 1.7027397260273974e-05, - "loss": 0.0167, - "step": 9470 - }, - { - "epoch": 0.87, - "learning_rate": 1.7013698630136985e-05, - "loss": 0.0149, - "step": 9480 - }, - { - "epoch": 0.87, - "learning_rate": 1.7e-05, - "loss": 0.0115, - "step": 9490 - }, - { - "epoch": 0.87, - "learning_rate": 1.6986301369863014e-05, - "loss": 0.015, - "step": 9500 - }, - { - "epoch": 0.87, - "learning_rate": 1.6972602739726026e-05, - "loss": 0.0122, - "step": 9510 - }, - { - "epoch": 0.87, - "learning_rate": 1.6958904109589044e-05, - "loss": 0.0174, - "step": 9520 - }, - { - "epoch": 0.87, - "learning_rate": 1.6945205479452055e-05, - "loss": 0.0112, - "step": 9530 - }, - { - "epoch": 0.87, - "learning_rate": 1.693150684931507e-05, - "loss": 0.0142, - "step": 9540 - }, - { - "epoch": 0.87, - "learning_rate": 1.6917808219178084e-05, - "loss": 0.0133, - "step": 9550 - }, - { - "epoch": 0.87, - "learning_rate": 1.6904109589041095e-05, - "loss": 0.019, - "step": 9560 - }, - { - "epoch": 0.87, - "learning_rate": 1.689041095890411e-05, - "loss": 0.0153, - "step": 9570 - }, - { - "epoch": 0.87, - "learning_rate": 1.687671232876712e-05, - "loss": 0.0122, - "step": 9580 - }, - { - "epoch": 0.88, - "learning_rate": 1.686301369863014e-05, - "loss": 0.0105, - "step": 9590 - }, - { - "epoch": 0.88, - "learning_rate": 1.6849315068493153e-05, - "loss": 0.0146, - "step": 9600 - }, - { - "epoch": 0.88, - "learning_rate": 1.6835616438356165e-05, - "loss": 0.0143, - "step": 9610 - }, - { - "epoch": 0.88, - "learning_rate": 1.682191780821918e-05, - "loss": 0.0149, - "step": 9620 - }, - { - "epoch": 0.88, - "learning_rate": 1.680821917808219e-05, - "loss": 0.0599, - "step": 9630 - }, - { - "epoch": 0.88, - "learning_rate": 1.6794520547945205e-05, - "loss": 0.0126, - "step": 9640 - }, - { - "epoch": 0.88, - "learning_rate": 1.6780821917808223e-05, - "loss": 0.0156, - "step": 9650 - }, - { - "epoch": 0.88, - "learning_rate": 1.6767123287671234e-05, - "loss": 0.0135, - "step": 9660 - }, - { - "epoch": 0.88, - "learning_rate": 1.675342465753425e-05, - "loss": 0.0168, - "step": 9670 - }, - { - "epoch": 0.88, - "learning_rate": 1.673972602739726e-05, - "loss": 0.0299, - "step": 9680 - }, - { - "epoch": 0.88, - "learning_rate": 1.6726027397260274e-05, - "loss": 0.0105, - "step": 9690 - }, - { - "epoch": 0.89, - "learning_rate": 1.6712328767123286e-05, - "loss": 0.0126, - "step": 9700 - }, - { - "epoch": 0.89, - "learning_rate": 1.66986301369863e-05, - "loss": 0.0117, - "step": 9710 - }, - { - "epoch": 0.89, - "learning_rate": 1.6684931506849318e-05, - "loss": 0.0106, - "step": 9720 - }, - { - "epoch": 0.89, - "learning_rate": 1.667123287671233e-05, - "loss": 0.0124, - "step": 9730 - }, - { - "epoch": 0.89, - "learning_rate": 1.6657534246575344e-05, - "loss": 0.0096, - "step": 9740 - }, - { - "epoch": 0.89, - "learning_rate": 1.6643835616438355e-05, - "loss": 0.0162, - "step": 9750 - }, - { - "epoch": 0.89, - "learning_rate": 1.663013698630137e-05, - "loss": 0.0169, - "step": 9760 - }, - { - "epoch": 0.89, - "learning_rate": 1.6616438356164384e-05, - "loss": 0.0134, - "step": 9770 - }, - { - "epoch": 0.89, - "learning_rate": 1.6602739726027395e-05, - "loss": 0.0174, - "step": 9780 - }, - { - "epoch": 0.89, - "learning_rate": 1.6589041095890413e-05, - "loss": 0.0147, - "step": 9790 - }, - { - "epoch": 0.89, - "learning_rate": 1.6575342465753425e-05, - "loss": 0.0145, - "step": 9800 - }, - { - "epoch": 0.9, - "learning_rate": 1.656164383561644e-05, - "loss": 0.0148, - "step": 9810 - }, - { - "epoch": 0.9, - "learning_rate": 1.6547945205479454e-05, - "loss": 0.0145, - "step": 9820 - }, - { - "epoch": 0.9, - "learning_rate": 1.6534246575342465e-05, - "loss": 0.0147, - "step": 9830 - }, - { - "epoch": 0.9, - "learning_rate": 1.652054794520548e-05, - "loss": 0.01, - "step": 9840 - }, - { - "epoch": 0.9, - "learning_rate": 1.6506849315068494e-05, - "loss": 0.0163, - "step": 9850 - }, - { - "epoch": 0.9, - "learning_rate": 1.649315068493151e-05, - "loss": 0.0172, - "step": 9860 - }, - { - "epoch": 0.9, - "learning_rate": 1.6479452054794523e-05, - "loss": 0.0143, - "step": 9870 - }, - { - "epoch": 0.9, - "learning_rate": 1.6465753424657534e-05, - "loss": 0.017, - "step": 9880 - }, - { - "epoch": 0.9, - "learning_rate": 1.645205479452055e-05, - "loss": 0.0144, - "step": 9890 - }, - { - "epoch": 0.9, - "learning_rate": 1.643835616438356e-05, - "loss": 0.0976, - "step": 9900 - }, - { - "epoch": 0.91, - "learning_rate": 1.6424657534246575e-05, - "loss": 0.0203, - "step": 9910 - }, - { - "epoch": 0.91, - "learning_rate": 1.641095890410959e-05, - "loss": 0.0174, - "step": 9920 - }, - { - "epoch": 0.91, - "learning_rate": 1.6397260273972604e-05, - "loss": 0.0133, - "step": 9930 - }, - { - "epoch": 0.91, - "learning_rate": 1.638356164383562e-05, - "loss": 0.0096, - "step": 9940 - }, - { - "epoch": 0.91, - "learning_rate": 1.636986301369863e-05, - "loss": 0.0152, - "step": 9950 - }, - { - "epoch": 0.91, - "learning_rate": 1.6356164383561644e-05, - "loss": 0.0117, - "step": 9960 - }, - { - "epoch": 0.91, - "learning_rate": 1.6342465753424655e-05, - "loss": 0.0148, - "step": 9970 - }, - { - "epoch": 0.91, - "learning_rate": 1.632876712328767e-05, - "loss": 0.014, - "step": 9980 - }, - { - "epoch": 0.91, - "learning_rate": 1.6315068493150688e-05, - "loss": 0.0155, - "step": 9990 - }, - { - "epoch": 0.91, - "learning_rate": 1.63013698630137e-05, - "loss": 0.0143, - "step": 10000 - }, - { - "epoch": 0.91, - "learning_rate": 1.6287671232876714e-05, - "loss": 0.0115, - "step": 10010 - }, - { - "epoch": 0.92, - "learning_rate": 1.6273972602739725e-05, - "loss": 0.0109, - "step": 10020 - }, - { - "epoch": 0.92, - "learning_rate": 1.626027397260274e-05, - "loss": 0.0121, - "step": 10030 - }, - { - "epoch": 0.92, - "learning_rate": 1.6246575342465754e-05, - "loss": 0.0125, - "step": 10040 - }, - { - "epoch": 0.92, - "learning_rate": 1.623287671232877e-05, - "loss": 0.0137, - "step": 10050 - }, - { - "epoch": 0.92, - "learning_rate": 1.6219178082191783e-05, - "loss": 0.0131, - "step": 10060 - }, - { - "epoch": 0.92, - "learning_rate": 1.6205479452054794e-05, - "loss": 0.0104, - "step": 10070 - }, - { - "epoch": 0.92, - "learning_rate": 1.619178082191781e-05, - "loss": 0.0182, - "step": 10080 - }, - { - "epoch": 0.92, - "learning_rate": 1.6178082191780823e-05, - "loss": 0.0117, - "step": 10090 - }, - { - "epoch": 0.92, - "learning_rate": 1.6164383561643835e-05, - "loss": 0.0129, - "step": 10100 - }, - { - "epoch": 0.92, - "learning_rate": 1.615068493150685e-05, - "loss": 0.0179, - "step": 10110 - }, - { - "epoch": 0.92, - "learning_rate": 1.6136986301369864e-05, - "loss": 0.0128, - "step": 10120 - }, - { - "epoch": 0.93, - "learning_rate": 1.6123287671232878e-05, - "loss": 0.0147, - "step": 10130 - }, - { - "epoch": 0.93, - "learning_rate": 1.6109589041095893e-05, - "loss": 0.0138, - "step": 10140 - }, - { - "epoch": 0.93, - "learning_rate": 1.6095890410958904e-05, - "loss": 0.0151, - "step": 10150 - }, - { - "epoch": 0.93, - "learning_rate": 1.608219178082192e-05, - "loss": 0.0147, - "step": 10160 - }, - { - "epoch": 0.93, - "learning_rate": 1.606849315068493e-05, - "loss": 0.0126, - "step": 10170 - }, - { - "epoch": 0.93, - "learning_rate": 1.6054794520547944e-05, - "loss": 0.0132, - "step": 10180 - }, - { - "epoch": 0.93, - "learning_rate": 1.604109589041096e-05, - "loss": 0.0159, - "step": 10190 - }, - { - "epoch": 0.93, - "learning_rate": 1.6027397260273974e-05, - "loss": 0.0177, - "step": 10200 - }, - { - "epoch": 0.93, - "learning_rate": 1.6013698630136988e-05, - "loss": 0.0131, - "step": 10210 - }, - { - "epoch": 0.93, - "learning_rate": 1.6e-05, - "loss": 0.0104, - "step": 10220 - }, - { - "epoch": 0.93, - "learning_rate": 1.5986301369863014e-05, - "loss": 0.0146, - "step": 10230 - }, - { - "epoch": 0.94, - "learning_rate": 1.5972602739726025e-05, - "loss": 0.0145, - "step": 10240 - }, - { - "epoch": 0.94, - "learning_rate": 1.5958904109589043e-05, - "loss": 0.0146, - "step": 10250 - }, - { - "epoch": 0.94, - "learning_rate": 1.5945205479452058e-05, - "loss": 0.012, - "step": 10260 - }, - { - "epoch": 0.94, - "learning_rate": 1.593150684931507e-05, - "loss": 0.0145, - "step": 10270 - }, - { - "epoch": 0.94, - "learning_rate": 1.5917808219178083e-05, - "loss": 0.0159, - "step": 10280 - }, - { - "epoch": 0.94, - "learning_rate": 1.5904109589041095e-05, - "loss": 0.0097, - "step": 10290 - }, - { - "epoch": 0.94, - "learning_rate": 1.589041095890411e-05, - "loss": 0.0146, - "step": 10300 - }, - { - "epoch": 0.94, - "learning_rate": 1.5876712328767124e-05, - "loss": 0.0144, - "step": 10310 - }, - { - "epoch": 0.94, - "learning_rate": 1.5863013698630138e-05, - "loss": 0.0146, - "step": 10320 - }, - { - "epoch": 0.94, - "learning_rate": 1.5849315068493153e-05, - "loss": 0.0145, - "step": 10330 - }, - { - "epoch": 0.94, - "learning_rate": 1.5835616438356164e-05, - "loss": 0.0166, - "step": 10340 - }, - { - "epoch": 0.95, - "learning_rate": 1.582191780821918e-05, - "loss": 0.0171, - "step": 10350 - }, - { - "epoch": 0.95, - "learning_rate": 1.5808219178082193e-05, - "loss": 0.0135, - "step": 10360 - }, - { - "epoch": 0.95, - "learning_rate": 1.5794520547945204e-05, - "loss": 0.0134, - "step": 10370 - }, - { - "epoch": 0.95, - "learning_rate": 1.578082191780822e-05, - "loss": 0.0112, - "step": 10380 - }, - { - "epoch": 0.95, - "learning_rate": 1.5767123287671233e-05, - "loss": 0.0137, - "step": 10390 - }, - { - "epoch": 0.95, - "learning_rate": 1.5753424657534248e-05, - "loss": 0.0139, - "step": 10400 - }, - { - "epoch": 0.95, - "learning_rate": 1.5739726027397263e-05, - "loss": 0.0114, - "step": 10410 - }, - { - "epoch": 0.95, - "learning_rate": 1.5726027397260274e-05, - "loss": 0.0139, - "step": 10420 - }, - { - "epoch": 0.95, - "learning_rate": 1.571232876712329e-05, - "loss": 0.0142, - "step": 10430 - }, - { - "epoch": 0.95, - "learning_rate": 1.56986301369863e-05, - "loss": 0.0169, - "step": 10440 - }, - { - "epoch": 0.95, - "learning_rate": 1.5684931506849318e-05, - "loss": 0.0128, - "step": 10450 - }, - { - "epoch": 0.96, - "learning_rate": 1.567123287671233e-05, - "loss": 0.016, - "step": 10460 - }, - { - "epoch": 0.96, - "learning_rate": 1.5657534246575343e-05, - "loss": 0.015, - "step": 10470 - }, - { - "epoch": 0.96, - "learning_rate": 1.5643835616438358e-05, - "loss": 0.0475, - "step": 10480 - }, - { - "epoch": 0.96, - "learning_rate": 1.563013698630137e-05, - "loss": 0.013, - "step": 10490 - }, - { - "epoch": 0.96, - "learning_rate": 1.5616438356164384e-05, - "loss": 0.0122, - "step": 10500 - }, - { - "epoch": 0.96, - "learning_rate": 1.5602739726027395e-05, - "loss": 0.0118, - "step": 10510 - }, - { - "epoch": 0.96, - "learning_rate": 1.5589041095890413e-05, - "loss": 0.0122, - "step": 10520 - }, - { - "epoch": 0.96, - "learning_rate": 1.5575342465753427e-05, - "loss": 0.0123, - "step": 10530 - }, - { - "epoch": 0.96, - "learning_rate": 1.556164383561644e-05, - "loss": 0.0135, - "step": 10540 - }, - { - "epoch": 0.96, - "learning_rate": 1.5547945205479453e-05, - "loss": 0.0159, - "step": 10550 - }, - { - "epoch": 0.96, - "learning_rate": 1.5534246575342464e-05, - "loss": 0.0106, - "step": 10560 - }, - { - "epoch": 0.97, - "learning_rate": 1.552054794520548e-05, - "loss": 0.016, - "step": 10570 - }, - { - "epoch": 0.97, - "learning_rate": 1.5506849315068493e-05, - "loss": 0.0154, - "step": 10580 - }, - { - "epoch": 0.97, - "learning_rate": 1.5493150684931508e-05, - "loss": 0.0144, - "step": 10590 - }, - { - "epoch": 0.97, - "learning_rate": 1.5479452054794523e-05, - "loss": 0.0111, - "step": 10600 - }, - { - "epoch": 0.97, - "learning_rate": 1.5465753424657534e-05, - "loss": 0.0112, - "step": 10610 - }, - { - "epoch": 0.97, - "learning_rate": 1.5452054794520548e-05, - "loss": 0.0121, - "step": 10620 - }, - { - "epoch": 0.97, - "learning_rate": 1.5438356164383563e-05, - "loss": 0.057, - "step": 10630 - }, - { - "epoch": 0.97, - "learning_rate": 1.5424657534246574e-05, - "loss": 0.0157, - "step": 10640 - }, - { - "epoch": 0.97, - "learning_rate": 1.5410958904109592e-05, - "loss": 0.0112, - "step": 10650 - }, - { - "epoch": 0.97, - "learning_rate": 1.5397260273972603e-05, - "loss": 0.0143, - "step": 10660 - }, - { - "epoch": 0.97, - "learning_rate": 1.5383561643835618e-05, - "loss": 0.0099, - "step": 10670 - }, - { - "epoch": 0.98, - "learning_rate": 1.5369863013698632e-05, - "loss": 0.0149, - "step": 10680 - }, - { - "epoch": 0.98, - "learning_rate": 1.5356164383561644e-05, - "loss": 0.0127, - "step": 10690 - }, - { - "epoch": 0.98, - "learning_rate": 1.5342465753424658e-05, - "loss": 0.0127, - "step": 10700 - }, - { - "epoch": 0.98, - "learning_rate": 1.532876712328767e-05, - "loss": 0.0152, - "step": 10710 - }, - { - "epoch": 0.98, - "learning_rate": 1.5315068493150687e-05, - "loss": 0.0115, - "step": 10720 - }, - { - "epoch": 0.98, - "learning_rate": 1.53013698630137e-05, - "loss": 0.0139, - "step": 10730 - }, - { - "epoch": 0.98, - "learning_rate": 1.5287671232876713e-05, - "loss": 0.0147, - "step": 10740 - }, - { - "epoch": 0.98, - "learning_rate": 1.5273972602739728e-05, - "loss": 0.0161, - "step": 10750 - }, - { - "epoch": 0.98, - "learning_rate": 1.526027397260274e-05, - "loss": 0.0114, - "step": 10760 - }, - { - "epoch": 0.98, - "learning_rate": 1.5246575342465753e-05, - "loss": 0.013, - "step": 10770 - }, - { - "epoch": 0.98, - "learning_rate": 1.5232876712328766e-05, - "loss": 0.0305, - "step": 10780 - }, - { - "epoch": 0.99, - "learning_rate": 1.5219178082191782e-05, - "loss": 0.0127, - "step": 10790 - }, - { - "epoch": 0.99, - "learning_rate": 1.5205479452054795e-05, - "loss": 0.0169, - "step": 10800 - }, - { - "epoch": 0.99, - "learning_rate": 1.5191780821917808e-05, - "loss": 0.0142, - "step": 10810 - }, - { - "epoch": 0.99, - "learning_rate": 1.5178082191780823e-05, - "loss": 0.0176, - "step": 10820 - }, - { - "epoch": 0.99, - "learning_rate": 1.5164383561643836e-05, - "loss": 0.0139, - "step": 10830 - }, - { - "epoch": 0.99, - "learning_rate": 1.5150684931506849e-05, - "loss": 0.0143, - "step": 10840 - }, - { - "epoch": 0.99, - "learning_rate": 1.5136986301369865e-05, - "loss": 0.011, - "step": 10850 - }, - { - "epoch": 0.99, - "learning_rate": 1.5123287671232878e-05, - "loss": 0.0168, - "step": 10860 - }, - { - "epoch": 0.99, - "learning_rate": 1.5109589041095892e-05, - "loss": 0.0134, - "step": 10870 - }, - { - "epoch": 0.99, - "learning_rate": 1.5095890410958905e-05, - "loss": 0.0134, - "step": 10880 - }, - { - "epoch": 0.99, - "learning_rate": 1.5082191780821918e-05, - "loss": 0.0138, - "step": 10890 - }, - { - "epoch": 1.0, - "learning_rate": 1.5068493150684931e-05, - "loss": 0.0117, - "step": 10900 - }, - { - "epoch": 1.0, - "learning_rate": 1.5054794520547944e-05, - "loss": 0.0144, - "step": 10910 - }, - { - "epoch": 1.0, - "learning_rate": 1.504109589041096e-05, - "loss": 0.0169, - "step": 10920 - }, - { - "epoch": 1.0, - "learning_rate": 1.5027397260273975e-05, - "loss": 0.012, - "step": 10930 - }, - { - "epoch": 1.0, - "learning_rate": 1.5013698630136988e-05, - "loss": 0.0147, - "step": 10940 - }, - { - "epoch": 1.0, - "learning_rate": 1.5e-05, - "loss": 0.0142, - "step": 10950 - }, - { - "epoch": 1.0, - "eval_exact_match": 58.751182592242195, - "eval_f1": 75.30221684543821, - "step": 10950 - }, - { - "epoch": 1.0, - "learning_rate": 1.4986301369863013e-05, - "loss": 0.0078, - "step": 10960 - }, - { - "epoch": 1.0, - "learning_rate": 1.4972602739726028e-05, - "loss": 0.0083, - "step": 10970 - }, - { - "epoch": 1.0, - "learning_rate": 1.4958904109589042e-05, - "loss": 0.0133, - "step": 10980 - }, - { - "epoch": 1.0, - "learning_rate": 1.4945205479452055e-05, - "loss": 0.0088, - "step": 10990 - }, - { - "epoch": 1.0, - "learning_rate": 1.4931506849315068e-05, - "loss": 0.0092, - "step": 11000 - }, - { - "epoch": 1.01, - "learning_rate": 1.4917808219178083e-05, - "loss": 0.0101, - "step": 11010 - }, - { - "epoch": 1.01, - "learning_rate": 1.4904109589041096e-05, - "loss": 0.0102, - "step": 11020 - }, - { - "epoch": 1.01, - "learning_rate": 1.4890410958904108e-05, - "loss": 0.0117, - "step": 11030 - }, - { - "epoch": 1.01, - "learning_rate": 1.4876712328767125e-05, - "loss": 0.0135, - "step": 11040 - }, - { - "epoch": 1.01, - "learning_rate": 1.4863013698630138e-05, - "loss": 0.0113, - "step": 11050 - }, - { - "epoch": 1.01, - "learning_rate": 1.484931506849315e-05, - "loss": 0.0073, - "step": 11060 - }, - { - "epoch": 1.01, - "learning_rate": 1.4835616438356165e-05, - "loss": 0.0098, - "step": 11070 - }, - { - "epoch": 1.01, - "learning_rate": 1.4821917808219178e-05, - "loss": 0.0128, - "step": 11080 - }, - { - "epoch": 1.01, - "learning_rate": 1.4808219178082193e-05, - "loss": 0.0168, - "step": 11090 - }, - { - "epoch": 1.01, - "learning_rate": 1.4794520547945205e-05, - "loss": 0.0102, - "step": 11100 - }, - { - "epoch": 1.01, - "learning_rate": 1.478082191780822e-05, - "loss": 0.0204, - "step": 11110 - }, - { - "epoch": 1.02, - "learning_rate": 1.4767123287671233e-05, - "loss": 0.0101, - "step": 11120 - }, - { - "epoch": 1.02, - "learning_rate": 1.4753424657534246e-05, - "loss": 0.0071, - "step": 11130 - }, - { - "epoch": 1.02, - "learning_rate": 1.473972602739726e-05, - "loss": 0.0096, - "step": 11140 - }, - { - "epoch": 1.02, - "learning_rate": 1.4726027397260275e-05, - "loss": 0.0117, - "step": 11150 - }, - { - "epoch": 1.02, - "learning_rate": 1.4712328767123288e-05, - "loss": 0.0107, - "step": 11160 - }, - { - "epoch": 1.02, - "learning_rate": 1.4698630136986302e-05, - "loss": 0.0101, - "step": 11170 - }, - { - "epoch": 1.02, - "learning_rate": 1.4684931506849315e-05, - "loss": 0.0106, - "step": 11180 - }, - { - "epoch": 1.02, - "learning_rate": 1.4671232876712328e-05, - "loss": 0.0076, - "step": 11190 - }, - { - "epoch": 1.02, - "learning_rate": 1.4657534246575343e-05, - "loss": 0.0151, - "step": 11200 - }, - { - "epoch": 1.02, - "learning_rate": 1.4643835616438357e-05, - "loss": 0.0093, - "step": 11210 - }, - { - "epoch": 1.02, - "learning_rate": 1.463013698630137e-05, - "loss": 0.0075, - "step": 11220 - }, - { - "epoch": 1.03, - "learning_rate": 1.4616438356164383e-05, - "loss": 0.0212, - "step": 11230 - }, - { - "epoch": 1.03, - "learning_rate": 1.4602739726027398e-05, - "loss": 0.041, - "step": 11240 - }, - { - "epoch": 1.03, - "learning_rate": 1.4589041095890412e-05, - "loss": 0.0084, - "step": 11250 - }, - { - "epoch": 1.03, - "learning_rate": 1.4575342465753425e-05, - "loss": 0.007, - "step": 11260 - }, - { - "epoch": 1.03, - "learning_rate": 1.456164383561644e-05, - "loss": 0.0115, - "step": 11270 - }, - { - "epoch": 1.03, - "learning_rate": 1.4547945205479452e-05, - "loss": 0.0131, - "step": 11280 - }, - { - "epoch": 1.03, - "learning_rate": 1.4534246575342465e-05, - "loss": 0.011, - "step": 11290 - }, - { - "epoch": 1.03, - "learning_rate": 1.4520547945205478e-05, - "loss": 0.0094, - "step": 11300 - }, - { - "epoch": 1.03, - "learning_rate": 1.4506849315068494e-05, - "loss": 0.0105, - "step": 11310 - }, - { - "epoch": 1.03, - "learning_rate": 1.4493150684931507e-05, - "loss": 0.0105, - "step": 11320 - }, - { - "epoch": 1.03, - "learning_rate": 1.447945205479452e-05, - "loss": 0.0117, - "step": 11330 - }, - { - "epoch": 1.04, - "learning_rate": 1.4465753424657535e-05, - "loss": 0.0117, - "step": 11340 - }, - { - "epoch": 1.04, - "learning_rate": 1.4452054794520548e-05, - "loss": 0.0106, - "step": 11350 - }, - { - "epoch": 1.04, - "learning_rate": 1.4438356164383562e-05, - "loss": 0.0252, - "step": 11360 - }, - { - "epoch": 1.04, - "learning_rate": 1.4424657534246577e-05, - "loss": 0.0097, - "step": 11370 - }, - { - "epoch": 1.04, - "learning_rate": 1.441095890410959e-05, - "loss": 0.0094, - "step": 11380 - }, - { - "epoch": 1.04, - "learning_rate": 1.4397260273972603e-05, - "loss": 0.0109, - "step": 11390 - }, - { - "epoch": 1.04, - "learning_rate": 1.4383561643835615e-05, - "loss": 0.0123, - "step": 11400 - }, - { - "epoch": 1.04, - "learning_rate": 1.436986301369863e-05, - "loss": 0.01, - "step": 11410 - }, - { - "epoch": 1.04, - "learning_rate": 1.4356164383561645e-05, - "loss": 0.0081, - "step": 11420 - }, - { - "epoch": 1.04, - "learning_rate": 1.4342465753424658e-05, - "loss": 0.0097, - "step": 11430 - }, - { - "epoch": 1.04, - "learning_rate": 1.4328767123287672e-05, - "loss": 0.0103, - "step": 11440 - }, - { - "epoch": 1.05, - "learning_rate": 1.4315068493150685e-05, - "loss": 0.0105, - "step": 11450 - }, - { - "epoch": 1.05, - "learning_rate": 1.4301369863013698e-05, - "loss": 0.0104, - "step": 11460 - }, - { - "epoch": 1.05, - "learning_rate": 1.4287671232876714e-05, - "loss": 0.0102, - "step": 11470 - }, - { - "epoch": 1.05, - "learning_rate": 1.4273972602739727e-05, - "loss": 0.0095, - "step": 11480 - }, - { - "epoch": 1.05, - "learning_rate": 1.426027397260274e-05, - "loss": 0.0079, - "step": 11490 - }, - { - "epoch": 1.05, - "learning_rate": 1.4246575342465753e-05, - "loss": 0.0098, - "step": 11500 - }, - { - "epoch": 1.05, - "learning_rate": 1.4232876712328767e-05, - "loss": 0.0112, - "step": 11510 - }, - { - "epoch": 1.05, - "learning_rate": 1.4219178082191782e-05, - "loss": 0.0119, - "step": 11520 - }, - { - "epoch": 1.05, - "learning_rate": 1.4205479452054795e-05, - "loss": 0.0094, - "step": 11530 - }, - { - "epoch": 1.05, - "learning_rate": 1.419178082191781e-05, - "loss": 0.0107, - "step": 11540 - }, - { - "epoch": 1.05, - "learning_rate": 1.4178082191780822e-05, - "loss": 0.0111, - "step": 11550 - }, - { - "epoch": 1.06, - "learning_rate": 1.4164383561643835e-05, - "loss": 0.0076, - "step": 11560 - }, - { - "epoch": 1.06, - "learning_rate": 1.415068493150685e-05, - "loss": 0.0126, - "step": 11570 - }, - { - "epoch": 1.06, - "learning_rate": 1.4136986301369864e-05, - "loss": 0.0105, - "step": 11580 - }, - { - "epoch": 1.06, - "learning_rate": 1.4123287671232877e-05, - "loss": 0.0124, - "step": 11590 - }, - { - "epoch": 1.06, - "learning_rate": 1.410958904109589e-05, - "loss": 0.01, - "step": 11600 - }, - { - "epoch": 1.06, - "learning_rate": 1.4095890410958905e-05, - "loss": 0.0098, - "step": 11610 - }, - { - "epoch": 1.06, - "learning_rate": 1.4082191780821917e-05, - "loss": 0.0112, - "step": 11620 - }, - { - "epoch": 1.06, - "learning_rate": 1.4068493150684932e-05, - "loss": 0.0105, - "step": 11630 - }, - { - "epoch": 1.06, - "learning_rate": 1.4054794520547947e-05, - "loss": 0.0095, - "step": 11640 - }, - { - "epoch": 1.06, - "learning_rate": 1.404109589041096e-05, - "loss": 0.0112, - "step": 11650 - }, - { - "epoch": 1.06, - "learning_rate": 1.4027397260273972e-05, - "loss": 0.0094, - "step": 11660 - }, - { - "epoch": 1.07, - "learning_rate": 1.4013698630136987e-05, - "loss": 0.0117, - "step": 11670 - }, - { - "epoch": 1.07, - "learning_rate": 1.4e-05, - "loss": 0.0123, - "step": 11680 - }, - { - "epoch": 1.07, - "learning_rate": 1.3986301369863014e-05, - "loss": 0.0094, - "step": 11690 - }, - { - "epoch": 1.07, - "learning_rate": 1.3972602739726027e-05, - "loss": 0.0169, - "step": 11700 - }, - { - "epoch": 1.07, - "learning_rate": 1.3958904109589042e-05, - "loss": 0.0125, - "step": 11710 - }, - { - "epoch": 1.07, - "learning_rate": 1.3945205479452055e-05, - "loss": 0.0102, - "step": 11720 - }, - { - "epoch": 1.07, - "learning_rate": 1.3931506849315068e-05, - "loss": 0.0094, - "step": 11730 - }, - { - "epoch": 1.07, - "learning_rate": 1.3917808219178084e-05, - "loss": 0.0082, - "step": 11740 - }, - { - "epoch": 1.07, - "learning_rate": 1.3904109589041097e-05, - "loss": 0.0109, - "step": 11750 - }, - { - "epoch": 1.07, - "learning_rate": 1.389041095890411e-05, - "loss": 0.0193, - "step": 11760 - }, - { - "epoch": 1.07, - "learning_rate": 1.3876712328767124e-05, - "loss": 0.009, - "step": 11770 - }, - { - "epoch": 1.08, - "learning_rate": 1.3863013698630137e-05, - "loss": 0.0081, - "step": 11780 - }, - { - "epoch": 1.08, - "learning_rate": 1.384931506849315e-05, - "loss": 0.01, - "step": 11790 - }, - { - "epoch": 1.08, - "learning_rate": 1.3835616438356164e-05, - "loss": 0.0078, - "step": 11800 - }, - { - "epoch": 1.08, - "learning_rate": 1.3821917808219179e-05, - "loss": 0.0088, - "step": 11810 - }, - { - "epoch": 1.08, - "learning_rate": 1.3808219178082192e-05, - "loss": 0.014, - "step": 11820 - }, - { - "epoch": 1.08, - "learning_rate": 1.3794520547945205e-05, - "loss": 0.0097, - "step": 11830 - }, - { - "epoch": 1.08, - "learning_rate": 1.378082191780822e-05, - "loss": 0.012, - "step": 11840 - }, - { - "epoch": 1.08, - "learning_rate": 1.3767123287671234e-05, - "loss": 0.0085, - "step": 11850 - }, - { - "epoch": 1.08, - "learning_rate": 1.3753424657534247e-05, - "loss": 0.0211, - "step": 11860 - }, - { - "epoch": 1.08, - "learning_rate": 1.3739726027397261e-05, - "loss": 0.0107, - "step": 11870 - }, - { - "epoch": 1.08, - "learning_rate": 1.3726027397260274e-05, - "loss": 0.0096, - "step": 11880 - }, - { - "epoch": 1.09, - "learning_rate": 1.3712328767123287e-05, - "loss": 0.0095, - "step": 11890 - }, - { - "epoch": 1.09, - "learning_rate": 1.3698630136986302e-05, - "loss": 0.0117, - "step": 11900 - }, - { - "epoch": 1.09, - "learning_rate": 1.3684931506849316e-05, - "loss": 0.0112, - "step": 11910 - }, - { - "epoch": 1.09, - "learning_rate": 1.367123287671233e-05, - "loss": 0.0098, - "step": 11920 - }, - { - "epoch": 1.09, - "learning_rate": 1.3657534246575342e-05, - "loss": 0.0113, - "step": 11930 - }, - { - "epoch": 1.09, - "learning_rate": 1.3643835616438357e-05, - "loss": 0.0109, - "step": 11940 - }, - { - "epoch": 1.09, - "learning_rate": 1.363013698630137e-05, - "loss": 0.0094, - "step": 11950 - }, - { - "epoch": 1.09, - "learning_rate": 1.3616438356164384e-05, - "loss": 0.0133, - "step": 11960 - }, - { - "epoch": 1.09, - "learning_rate": 1.3602739726027399e-05, - "loss": 0.0114, - "step": 11970 - }, - { - "epoch": 1.09, - "learning_rate": 1.3589041095890412e-05, - "loss": 0.009, - "step": 11980 - }, - { - "epoch": 1.09, - "learning_rate": 1.3575342465753424e-05, - "loss": 0.011, - "step": 11990 - }, - { - "epoch": 1.1, - "learning_rate": 1.3561643835616437e-05, - "loss": 0.0239, - "step": 12000 - }, - { - "epoch": 1.1, - "learning_rate": 1.3547945205479454e-05, - "loss": 0.0098, - "step": 12010 - }, - { - "epoch": 1.1, - "learning_rate": 1.3534246575342466e-05, - "loss": 0.0119, - "step": 12020 - }, - { - "epoch": 1.1, - "learning_rate": 1.352054794520548e-05, - "loss": 0.0139, - "step": 12030 - }, - { - "epoch": 1.1, - "learning_rate": 1.3506849315068494e-05, - "loss": 0.0105, - "step": 12040 - }, - { - "epoch": 1.1, - "learning_rate": 1.3493150684931507e-05, - "loss": 0.0114, - "step": 12050 - }, - { - "epoch": 1.1, - "learning_rate": 1.347945205479452e-05, - "loss": 0.0095, - "step": 12060 - }, - { - "epoch": 1.1, - "learning_rate": 1.3465753424657536e-05, - "loss": 0.01, - "step": 12070 - }, - { - "epoch": 1.1, - "learning_rate": 1.3452054794520549e-05, - "loss": 0.0081, - "step": 12080 - }, - { - "epoch": 1.1, - "learning_rate": 1.3438356164383562e-05, - "loss": 0.0121, - "step": 12090 - }, - { - "epoch": 1.11, - "learning_rate": 1.3424657534246575e-05, - "loss": 0.0133, - "step": 12100 - }, - { - "epoch": 1.11, - "learning_rate": 1.3410958904109589e-05, - "loss": 0.0095, - "step": 12110 - }, - { - "epoch": 1.11, - "learning_rate": 1.3397260273972604e-05, - "loss": 0.0114, - "step": 12120 - }, - { - "epoch": 1.11, - "learning_rate": 1.3383561643835617e-05, - "loss": 0.0118, - "step": 12130 - }, - { - "epoch": 1.11, - "learning_rate": 1.3369863013698631e-05, - "loss": 0.0098, - "step": 12140 - }, - { - "epoch": 1.11, - "learning_rate": 1.3356164383561644e-05, - "loss": 0.0144, - "step": 12150 - }, - { - "epoch": 1.11, - "learning_rate": 1.3342465753424657e-05, - "loss": 0.0106, - "step": 12160 - }, - { - "epoch": 1.11, - "learning_rate": 1.3328767123287671e-05, - "loss": 0.0091, - "step": 12170 - }, - { - "epoch": 1.11, - "learning_rate": 1.3315068493150686e-05, - "loss": 0.0064, - "step": 12180 - }, - { - "epoch": 1.11, - "learning_rate": 1.3301369863013699e-05, - "loss": 0.0099, - "step": 12190 - }, - { - "epoch": 1.11, - "learning_rate": 1.3287671232876712e-05, - "loss": 0.0109, - "step": 12200 - }, - { - "epoch": 1.12, - "learning_rate": 1.3273972602739726e-05, - "loss": 0.0118, - "step": 12210 - }, - { - "epoch": 1.12, - "learning_rate": 1.326027397260274e-05, - "loss": 0.0106, - "step": 12220 - }, - { - "epoch": 1.12, - "learning_rate": 1.3246575342465754e-05, - "loss": 0.0097, - "step": 12230 - }, - { - "epoch": 1.12, - "learning_rate": 1.3232876712328768e-05, - "loss": 0.0101, - "step": 12240 - }, - { - "epoch": 1.12, - "learning_rate": 1.3219178082191781e-05, - "loss": 0.0104, - "step": 12250 - }, - { - "epoch": 1.12, - "learning_rate": 1.3205479452054794e-05, - "loss": 0.0104, - "step": 12260 - }, - { - "epoch": 1.12, - "learning_rate": 1.3191780821917809e-05, - "loss": 0.0105, - "step": 12270 - }, - { - "epoch": 1.12, - "learning_rate": 1.3178082191780823e-05, - "loss": 0.0091, - "step": 12280 - }, - { - "epoch": 1.12, - "learning_rate": 1.3164383561643836e-05, - "loss": 0.0099, - "step": 12290 - }, - { - "epoch": 1.12, - "learning_rate": 1.3150684931506849e-05, - "loss": 0.0083, - "step": 12300 - }, - { - "epoch": 1.12, - "learning_rate": 1.3136986301369864e-05, - "loss": 0.0111, - "step": 12310 - }, - { - "epoch": 1.13, - "learning_rate": 1.3123287671232877e-05, - "loss": 0.0359, - "step": 12320 - }, - { - "epoch": 1.13, - "learning_rate": 1.310958904109589e-05, - "loss": 0.0099, - "step": 12330 - }, - { - "epoch": 1.13, - "learning_rate": 1.3095890410958906e-05, - "loss": 0.0084, - "step": 12340 - }, - { - "epoch": 1.13, - "learning_rate": 1.3082191780821919e-05, - "loss": 0.0134, - "step": 12350 - }, - { - "epoch": 1.13, - "learning_rate": 1.3068493150684931e-05, - "loss": 0.0121, - "step": 12360 - }, - { - "epoch": 1.13, - "learning_rate": 1.3054794520547946e-05, - "loss": 0.0115, - "step": 12370 - }, - { - "epoch": 1.13, - "learning_rate": 1.3041095890410959e-05, - "loss": 0.0135, - "step": 12380 - }, - { - "epoch": 1.13, - "learning_rate": 1.3027397260273973e-05, - "loss": 0.0111, - "step": 12390 - }, - { - "epoch": 1.13, - "learning_rate": 1.3013698630136986e-05, - "loss": 0.0102, - "step": 12400 - }, - { - "epoch": 1.13, - "learning_rate": 1.3000000000000001e-05, - "loss": 0.0115, - "step": 12410 - }, - { - "epoch": 1.13, - "learning_rate": 1.2986301369863014e-05, - "loss": 0.0115, - "step": 12420 - }, - { - "epoch": 1.14, - "learning_rate": 1.2972602739726027e-05, - "loss": 0.0119, - "step": 12430 - }, - { - "epoch": 1.14, - "learning_rate": 1.2958904109589041e-05, - "loss": 0.0177, - "step": 12440 - }, - { - "epoch": 1.14, - "learning_rate": 1.2945205479452056e-05, - "loss": 0.0119, - "step": 12450 - }, - { - "epoch": 1.14, - "learning_rate": 1.2931506849315069e-05, - "loss": 0.0104, - "step": 12460 - }, - { - "epoch": 1.14, - "learning_rate": 1.2917808219178083e-05, - "loss": 0.0133, - "step": 12470 - }, - { - "epoch": 1.14, - "learning_rate": 1.2904109589041096e-05, - "loss": 0.0128, - "step": 12480 - }, - { - "epoch": 1.14, - "learning_rate": 1.2890410958904109e-05, - "loss": 0.0136, - "step": 12490 - }, - { - "epoch": 1.14, - "learning_rate": 1.2876712328767124e-05, - "loss": 0.0102, - "step": 12500 - }, - { - "epoch": 1.14, - "learning_rate": 1.2863013698630138e-05, - "loss": 0.0094, - "step": 12510 - }, - { - "epoch": 1.14, - "learning_rate": 1.2849315068493151e-05, - "loss": 0.0132, - "step": 12520 - }, - { - "epoch": 1.14, - "learning_rate": 1.2835616438356164e-05, - "loss": 0.0096, - "step": 12530 - }, - { - "epoch": 1.15, - "learning_rate": 1.2821917808219178e-05, - "loss": 0.012, - "step": 12540 - }, - { - "epoch": 1.15, - "learning_rate": 1.2808219178082193e-05, - "loss": 0.01, - "step": 12550 - }, - { - "epoch": 1.15, - "learning_rate": 1.2794520547945206e-05, - "loss": 0.0142, - "step": 12560 - }, - { - "epoch": 1.15, - "learning_rate": 1.278082191780822e-05, - "loss": 0.0095, - "step": 12570 - }, - { - "epoch": 1.15, - "learning_rate": 1.2767123287671233e-05, - "loss": 0.0085, - "step": 12580 - }, - { - "epoch": 1.15, - "learning_rate": 1.2753424657534246e-05, - "loss": 0.0148, - "step": 12590 - }, - { - "epoch": 1.15, - "learning_rate": 1.2739726027397259e-05, - "loss": 0.0095, - "step": 12600 - }, - { - "epoch": 1.15, - "learning_rate": 1.2726027397260275e-05, - "loss": 0.0148, - "step": 12610 - }, - { - "epoch": 1.15, - "learning_rate": 1.2712328767123288e-05, - "loss": 0.054, - "step": 12620 - }, - { - "epoch": 1.15, - "learning_rate": 1.2698630136986301e-05, - "loss": 0.0121, - "step": 12630 - }, - { - "epoch": 1.15, - "learning_rate": 1.2684931506849316e-05, - "loss": 0.0097, - "step": 12640 - }, - { - "epoch": 1.16, - "learning_rate": 1.2671232876712329e-05, - "loss": 0.0102, - "step": 12650 - }, - { - "epoch": 1.16, - "learning_rate": 1.2657534246575343e-05, - "loss": 0.0095, - "step": 12660 - }, - { - "epoch": 1.16, - "learning_rate": 1.2643835616438358e-05, - "loss": 0.009, - "step": 12670 - }, - { - "epoch": 1.16, - "learning_rate": 1.263013698630137e-05, - "loss": 0.0089, - "step": 12680 - }, - { - "epoch": 1.16, - "learning_rate": 1.2616438356164384e-05, - "loss": 0.0102, - "step": 12690 - }, - { - "epoch": 1.16, - "learning_rate": 1.2602739726027396e-05, - "loss": 0.0102, - "step": 12700 - }, - { - "epoch": 1.16, - "learning_rate": 1.2589041095890411e-05, - "loss": 0.0107, - "step": 12710 - }, - { - "epoch": 1.16, - "learning_rate": 1.2575342465753426e-05, - "loss": 0.0123, - "step": 12720 - }, - { - "epoch": 1.16, - "learning_rate": 1.2561643835616438e-05, - "loss": 0.0085, - "step": 12730 - }, - { - "epoch": 1.16, - "learning_rate": 1.2547945205479453e-05, - "loss": 0.0109, - "step": 12740 - }, - { - "epoch": 1.16, - "learning_rate": 1.2534246575342466e-05, - "loss": 0.0144, - "step": 12750 - }, - { - "epoch": 1.17, - "learning_rate": 1.2520547945205479e-05, - "loss": 0.0096, - "step": 12760 - }, - { - "epoch": 1.17, - "learning_rate": 1.2506849315068495e-05, - "loss": 0.0133, - "step": 12770 - }, - { - "epoch": 1.17, - "learning_rate": 1.2493150684931508e-05, - "loss": 0.0105, - "step": 12780 - }, - { - "epoch": 1.17, - "learning_rate": 1.247945205479452e-05, - "loss": 0.0078, - "step": 12790 - }, - { - "epoch": 1.17, - "learning_rate": 1.2465753424657534e-05, - "loss": 0.013, - "step": 12800 - }, - { - "epoch": 1.17, - "learning_rate": 1.2452054794520548e-05, - "loss": 0.0109, - "step": 12810 - }, - { - "epoch": 1.17, - "learning_rate": 1.2438356164383561e-05, - "loss": 0.0117, - "step": 12820 - }, - { - "epoch": 1.17, - "learning_rate": 1.2424657534246576e-05, - "loss": 0.008, - "step": 12830 - }, - { - "epoch": 1.17, - "learning_rate": 1.241095890410959e-05, - "loss": 0.0088, - "step": 12840 - }, - { - "epoch": 1.17, - "learning_rate": 1.2397260273972603e-05, - "loss": 0.0104, - "step": 12850 - }, - { - "epoch": 1.17, - "learning_rate": 1.2383561643835616e-05, - "loss": 0.0206, - "step": 12860 - }, - { - "epoch": 1.18, - "learning_rate": 1.236986301369863e-05, - "loss": 0.0091, - "step": 12870 - }, - { - "epoch": 1.18, - "learning_rate": 1.2356164383561645e-05, - "loss": 0.0093, - "step": 12880 - }, - { - "epoch": 1.18, - "learning_rate": 1.2342465753424658e-05, - "loss": 0.0087, - "step": 12890 - }, - { - "epoch": 1.18, - "learning_rate": 1.2328767123287671e-05, - "loss": 0.0206, - "step": 12900 - }, - { - "epoch": 1.18, - "learning_rate": 1.2315068493150685e-05, - "loss": 0.0114, - "step": 12910 - }, - { - "epoch": 1.18, - "learning_rate": 1.2301369863013698e-05, - "loss": 0.0118, - "step": 12920 - }, - { - "epoch": 1.18, - "learning_rate": 1.2287671232876713e-05, - "loss": 0.009, - "step": 12930 - }, - { - "epoch": 1.18, - "learning_rate": 1.2273972602739727e-05, - "loss": 0.0089, - "step": 12940 - }, - { - "epoch": 1.18, - "learning_rate": 1.226027397260274e-05, - "loss": 0.0085, - "step": 12950 - }, - { - "epoch": 1.18, - "learning_rate": 1.2246575342465753e-05, - "loss": 0.0087, - "step": 12960 - }, - { - "epoch": 1.18, - "learning_rate": 1.2232876712328768e-05, - "loss": 0.0122, - "step": 12970 - }, - { - "epoch": 1.19, - "learning_rate": 1.221917808219178e-05, - "loss": 0.0123, - "step": 12980 - }, - { - "epoch": 1.19, - "learning_rate": 1.2205479452054795e-05, - "loss": 0.0113, - "step": 12990 - }, - { - "epoch": 1.19, - "learning_rate": 1.2191780821917808e-05, - "loss": 0.0081, - "step": 13000 - }, - { - "epoch": 1.19, - "learning_rate": 1.2178082191780823e-05, - "loss": 0.0079, - "step": 13010 - }, - { - "epoch": 1.19, - "learning_rate": 1.2164383561643836e-05, - "loss": 0.0082, - "step": 13020 - }, - { - "epoch": 1.19, - "learning_rate": 1.2150684931506848e-05, - "loss": 0.0103, - "step": 13030 - }, - { - "epoch": 1.19, - "learning_rate": 1.2136986301369865e-05, - "loss": 0.0136, - "step": 13040 - }, - { - "epoch": 1.19, - "learning_rate": 1.2123287671232878e-05, - "loss": 0.0121, - "step": 13050 - }, - { - "epoch": 1.19, - "learning_rate": 1.210958904109589e-05, - "loss": 0.0114, - "step": 13060 - }, - { - "epoch": 1.19, - "learning_rate": 1.2095890410958905e-05, - "loss": 0.0144, - "step": 13070 - }, - { - "epoch": 1.19, - "learning_rate": 1.2082191780821918e-05, - "loss": 0.0135, - "step": 13080 - }, - { - "epoch": 1.2, - "learning_rate": 1.206849315068493e-05, - "loss": 0.01, - "step": 13090 - }, - { - "epoch": 1.2, - "learning_rate": 1.2054794520547945e-05, - "loss": 0.0111, - "step": 13100 - }, - { - "epoch": 1.2, - "learning_rate": 1.204109589041096e-05, - "loss": 0.0114, - "step": 13110 - }, - { - "epoch": 1.2, - "learning_rate": 1.2027397260273973e-05, - "loss": 0.0072, - "step": 13120 - }, - { - "epoch": 1.2, - "learning_rate": 1.2013698630136986e-05, - "loss": 0.0114, - "step": 13130 - }, - { - "epoch": 1.2, - "learning_rate": 1.2e-05, - "loss": 0.0118, - "step": 13140 - }, - { - "epoch": 1.2, - "learning_rate": 1.1986301369863015e-05, - "loss": 0.0102, - "step": 13150 - }, - { - "epoch": 1.2, - "learning_rate": 1.1972602739726028e-05, - "loss": 0.0127, - "step": 13160 - }, - { - "epoch": 1.2, - "learning_rate": 1.195890410958904e-05, - "loss": 0.0111, - "step": 13170 - }, - { - "epoch": 1.2, - "learning_rate": 1.1945205479452055e-05, - "loss": 0.0098, - "step": 13180 - }, - { - "epoch": 1.2, - "learning_rate": 1.1931506849315068e-05, - "loss": 0.0094, - "step": 13190 - }, - { - "epoch": 1.21, - "learning_rate": 1.1917808219178083e-05, - "loss": 0.0098, - "step": 13200 - }, - { - "epoch": 1.21, - "learning_rate": 1.1904109589041097e-05, - "loss": 0.0091, - "step": 13210 - }, - { - "epoch": 1.21, - "learning_rate": 1.189041095890411e-05, - "loss": 0.0127, - "step": 13220 - }, - { - "epoch": 1.21, - "learning_rate": 1.1876712328767123e-05, - "loss": 0.0106, - "step": 13230 - }, - { - "epoch": 1.21, - "learning_rate": 1.1863013698630138e-05, - "loss": 0.011, - "step": 13240 - }, - { - "epoch": 1.21, - "learning_rate": 1.184931506849315e-05, - "loss": 0.0136, - "step": 13250 - }, - { - "epoch": 1.21, - "learning_rate": 1.1835616438356165e-05, - "loss": 0.0111, - "step": 13260 - }, - { - "epoch": 1.21, - "learning_rate": 1.1821917808219178e-05, - "loss": 0.0166, - "step": 13270 - }, - { - "epoch": 1.21, - "learning_rate": 1.1808219178082192e-05, - "loss": 0.0113, - "step": 13280 - }, - { - "epoch": 1.21, - "learning_rate": 1.1794520547945205e-05, - "loss": 0.0204, - "step": 13290 - }, - { - "epoch": 1.21, - "learning_rate": 1.1780821917808218e-05, - "loss": 0.0113, - "step": 13300 - }, - { - "epoch": 1.22, - "learning_rate": 1.1767123287671234e-05, - "loss": 0.0068, - "step": 13310 - }, - { - "epoch": 1.22, - "learning_rate": 1.1753424657534247e-05, - "loss": 0.0096, - "step": 13320 - }, - { - "epoch": 1.22, - "learning_rate": 1.173972602739726e-05, - "loss": 0.0111, - "step": 13330 - }, - { - "epoch": 1.22, - "learning_rate": 1.1726027397260275e-05, - "loss": 0.0116, - "step": 13340 - }, - { - "epoch": 1.22, - "learning_rate": 1.1712328767123288e-05, - "loss": 0.0147, - "step": 13350 - }, - { - "epoch": 1.22, - "learning_rate": 1.16986301369863e-05, - "loss": 0.0125, - "step": 13360 - }, - { - "epoch": 1.22, - "learning_rate": 1.1684931506849315e-05, - "loss": 0.0139, - "step": 13370 - }, - { - "epoch": 1.22, - "learning_rate": 1.167123287671233e-05, - "loss": 0.0106, - "step": 13380 - }, - { - "epoch": 1.22, - "learning_rate": 1.1657534246575343e-05, - "loss": 0.011, - "step": 13390 - }, - { - "epoch": 1.22, - "learning_rate": 1.1643835616438355e-05, - "loss": 0.0101, - "step": 13400 - }, - { - "epoch": 1.22, - "learning_rate": 1.163013698630137e-05, - "loss": 0.0124, - "step": 13410 - }, - { - "epoch": 1.23, - "learning_rate": 1.1616438356164385e-05, - "loss": 0.0109, - "step": 13420 - }, - { - "epoch": 1.23, - "learning_rate": 1.1602739726027397e-05, - "loss": 0.0116, - "step": 13430 - }, - { - "epoch": 1.23, - "learning_rate": 1.1589041095890412e-05, - "loss": 0.0117, - "step": 13440 - }, - { - "epoch": 1.23, - "learning_rate": 1.1575342465753425e-05, - "loss": 0.0094, - "step": 13450 - }, - { - "epoch": 1.23, - "learning_rate": 1.1561643835616438e-05, - "loss": 0.0095, - "step": 13460 - }, - { - "epoch": 1.23, - "learning_rate": 1.154794520547945e-05, - "loss": 0.0114, - "step": 13470 - }, - { - "epoch": 1.23, - "learning_rate": 1.1534246575342467e-05, - "loss": 0.0136, - "step": 13480 - }, - { - "epoch": 1.23, - "learning_rate": 1.152054794520548e-05, - "loss": 0.0124, - "step": 13490 - }, - { - "epoch": 1.23, - "learning_rate": 1.1506849315068493e-05, - "loss": 0.0128, - "step": 13500 - }, - { - "epoch": 1.23, - "learning_rate": 1.1493150684931507e-05, - "loss": 0.0151, - "step": 13510 - }, - { - "epoch": 1.23, - "learning_rate": 1.147945205479452e-05, - "loss": 0.0372, - "step": 13520 - }, - { - "epoch": 1.24, - "learning_rate": 1.1465753424657535e-05, - "loss": 0.0104, - "step": 13530 - }, - { - "epoch": 1.24, - "learning_rate": 1.145205479452055e-05, - "loss": 0.012, - "step": 13540 - }, - { - "epoch": 1.24, - "learning_rate": 1.1438356164383562e-05, - "loss": 0.0123, - "step": 13550 - }, - { - "epoch": 1.24, - "learning_rate": 1.1424657534246575e-05, - "loss": 0.01, - "step": 13560 - }, - { - "epoch": 1.24, - "learning_rate": 1.1410958904109588e-05, - "loss": 0.0078, - "step": 13570 - }, - { - "epoch": 1.24, - "learning_rate": 1.1397260273972604e-05, - "loss": 0.0099, - "step": 13580 - }, - { - "epoch": 1.24, - "learning_rate": 1.1383561643835617e-05, - "loss": 0.0104, - "step": 13590 - }, - { - "epoch": 1.24, - "learning_rate": 1.136986301369863e-05, - "loss": 0.0089, - "step": 13600 - }, - { - "epoch": 1.24, - "learning_rate": 1.1356164383561645e-05, - "loss": 0.011, - "step": 13610 - }, - { - "epoch": 1.24, - "learning_rate": 1.1342465753424657e-05, - "loss": 0.0084, - "step": 13620 - }, - { - "epoch": 1.24, - "learning_rate": 1.132876712328767e-05, - "loss": 0.0147, - "step": 13630 - }, - { - "epoch": 1.25, - "learning_rate": 1.1315068493150687e-05, - "loss": 0.01, - "step": 13640 - }, - { - "epoch": 1.25, - "learning_rate": 1.13013698630137e-05, - "loss": 0.0104, - "step": 13650 - }, - { - "epoch": 1.25, - "learning_rate": 1.1287671232876712e-05, - "loss": 0.0085, - "step": 13660 - }, - { - "epoch": 1.25, - "learning_rate": 1.1273972602739725e-05, - "loss": 0.0077, - "step": 13670 - }, - { - "epoch": 1.25, - "learning_rate": 1.126027397260274e-05, - "loss": 0.0078, - "step": 13680 - }, - { - "epoch": 1.25, - "learning_rate": 1.1246575342465754e-05, - "loss": 0.0076, - "step": 13690 - }, - { - "epoch": 1.25, - "learning_rate": 1.1232876712328767e-05, - "loss": 0.0148, - "step": 13700 - }, - { - "epoch": 1.25, - "learning_rate": 1.1219178082191782e-05, - "loss": 0.0098, - "step": 13710 - }, - { - "epoch": 1.25, - "learning_rate": 1.1205479452054795e-05, - "loss": 0.0121, - "step": 13720 - }, - { - "epoch": 1.25, - "learning_rate": 1.1191780821917808e-05, - "loss": 0.0127, - "step": 13730 - }, - { - "epoch": 1.25, - "learning_rate": 1.1178082191780822e-05, - "loss": 0.0102, - "step": 13740 - }, - { - "epoch": 1.26, - "learning_rate": 1.1164383561643837e-05, - "loss": 0.0105, - "step": 13750 - }, - { - "epoch": 1.26, - "learning_rate": 1.115068493150685e-05, - "loss": 0.0082, - "step": 13760 - }, - { - "epoch": 1.26, - "learning_rate": 1.1136986301369862e-05, - "loss": 0.0101, - "step": 13770 - }, - { - "epoch": 1.26, - "learning_rate": 1.1123287671232877e-05, - "loss": 0.0088, - "step": 13780 - }, - { - "epoch": 1.26, - "learning_rate": 1.110958904109589e-05, - "loss": 0.0108, - "step": 13790 - }, - { - "epoch": 1.26, - "learning_rate": 1.1095890410958904e-05, - "loss": 0.0115, - "step": 13800 - }, - { - "epoch": 1.26, - "learning_rate": 1.1082191780821919e-05, - "loss": 0.0274, - "step": 13810 - }, - { - "epoch": 1.26, - "learning_rate": 1.1068493150684932e-05, - "loss": 0.0092, - "step": 13820 - }, - { - "epoch": 1.26, - "learning_rate": 1.1054794520547945e-05, - "loss": 0.0338, - "step": 13830 - }, - { - "epoch": 1.26, - "learning_rate": 1.104109589041096e-05, - "loss": 0.0109, - "step": 13840 - }, - { - "epoch": 1.26, - "learning_rate": 1.1027397260273972e-05, - "loss": 0.0121, - "step": 13850 - }, - { - "epoch": 1.27, - "learning_rate": 1.1013698630136987e-05, - "loss": 0.0129, - "step": 13860 - }, - { - "epoch": 1.27, - "learning_rate": 1.1e-05, - "loss": 0.0148, - "step": 13870 - }, - { - "epoch": 1.27, - "learning_rate": 1.0986301369863014e-05, - "loss": 0.0099, - "step": 13880 - }, - { - "epoch": 1.27, - "learning_rate": 1.0972602739726027e-05, - "loss": 0.0093, - "step": 13890 - }, - { - "epoch": 1.27, - "learning_rate": 1.095890410958904e-05, - "loss": 0.008, - "step": 13900 - }, - { - "epoch": 1.27, - "learning_rate": 1.0945205479452056e-05, - "loss": 0.0088, - "step": 13910 - }, - { - "epoch": 1.27, - "learning_rate": 1.093150684931507e-05, - "loss": 0.0137, - "step": 13920 - }, - { - "epoch": 1.27, - "learning_rate": 1.0917808219178082e-05, - "loss": 0.0106, - "step": 13930 - }, - { - "epoch": 1.27, - "learning_rate": 1.0904109589041097e-05, - "loss": 0.0078, - "step": 13940 - }, - { - "epoch": 1.27, - "learning_rate": 1.089041095890411e-05, - "loss": 0.0087, - "step": 13950 - }, - { - "epoch": 1.27, - "learning_rate": 1.0876712328767124e-05, - "loss": 0.011, - "step": 13960 - }, - { - "epoch": 1.28, - "learning_rate": 1.0863013698630137e-05, - "loss": 0.0088, - "step": 13970 - }, - { - "epoch": 1.28, - "learning_rate": 1.0849315068493152e-05, - "loss": 0.0123, - "step": 13980 - }, - { - "epoch": 1.28, - "learning_rate": 1.0835616438356164e-05, - "loss": 0.016, - "step": 13990 - }, - { - "epoch": 1.28, - "learning_rate": 1.0821917808219177e-05, - "loss": 0.0123, - "step": 14000 - }, - { - "epoch": 1.28, - "learning_rate": 1.0808219178082192e-05, - "loss": 0.0146, - "step": 14010 - }, - { - "epoch": 1.28, - "learning_rate": 1.0794520547945206e-05, - "loss": 0.0127, - "step": 14020 - }, - { - "epoch": 1.28, - "learning_rate": 1.078082191780822e-05, - "loss": 0.0151, - "step": 14030 - }, - { - "epoch": 1.28, - "learning_rate": 1.0767123287671234e-05, - "loss": 0.0094, - "step": 14040 - }, - { - "epoch": 1.28, - "learning_rate": 1.0753424657534247e-05, - "loss": 0.0123, - "step": 14050 - }, - { - "epoch": 1.28, - "learning_rate": 1.073972602739726e-05, - "loss": 0.0094, - "step": 14060 - }, - { - "epoch": 1.28, - "learning_rate": 1.0726027397260274e-05, - "loss": 0.0095, - "step": 14070 - }, - { - "epoch": 1.29, - "learning_rate": 1.0712328767123289e-05, - "loss": 0.0142, - "step": 14080 - }, - { - "epoch": 1.29, - "learning_rate": 1.0698630136986302e-05, - "loss": 0.0088, - "step": 14090 - }, - { - "epoch": 1.29, - "learning_rate": 1.0684931506849315e-05, - "loss": 0.0112, - "step": 14100 - }, - { - "epoch": 1.29, - "learning_rate": 1.0671232876712329e-05, - "loss": 0.0112, - "step": 14110 - }, - { - "epoch": 1.29, - "learning_rate": 1.0657534246575342e-05, - "loss": 0.0186, - "step": 14120 - }, - { - "epoch": 1.29, - "learning_rate": 1.0643835616438357e-05, - "loss": 0.0111, - "step": 14130 - }, - { - "epoch": 1.29, - "learning_rate": 1.0630136986301371e-05, - "loss": 0.0098, - "step": 14140 - }, - { - "epoch": 1.29, - "learning_rate": 1.0616438356164384e-05, - "loss": 0.0112, - "step": 14150 - }, - { - "epoch": 1.29, - "learning_rate": 1.0602739726027397e-05, - "loss": 0.0082, - "step": 14160 - }, - { - "epoch": 1.29, - "learning_rate": 1.058904109589041e-05, - "loss": 0.0104, - "step": 14170 - }, - { - "epoch": 1.29, - "learning_rate": 1.0575342465753426e-05, - "loss": 0.0087, - "step": 14180 - }, - { - "epoch": 1.3, - "learning_rate": 1.0561643835616439e-05, - "loss": 0.0102, - "step": 14190 - }, - { - "epoch": 1.3, - "learning_rate": 1.0547945205479452e-05, - "loss": 0.013, - "step": 14200 - }, - { - "epoch": 1.3, - "learning_rate": 1.0534246575342466e-05, - "loss": 0.0087, - "step": 14210 - }, - { - "epoch": 1.3, - "learning_rate": 1.052054794520548e-05, - "loss": 0.0093, - "step": 14220 - }, - { - "epoch": 1.3, - "learning_rate": 1.0506849315068494e-05, - "loss": 0.0132, - "step": 14230 - }, - { - "epoch": 1.3, - "learning_rate": 1.0493150684931508e-05, - "loss": 0.0127, - "step": 14240 - }, - { - "epoch": 1.3, - "learning_rate": 1.0479452054794521e-05, - "loss": 0.0109, - "step": 14250 - }, - { - "epoch": 1.3, - "learning_rate": 1.0465753424657534e-05, - "loss": 0.0077, - "step": 14260 - }, - { - "epoch": 1.3, - "learning_rate": 1.0452054794520547e-05, - "loss": 0.0109, - "step": 14270 - }, - { - "epoch": 1.3, - "learning_rate": 1.0438356164383562e-05, - "loss": 0.0123, - "step": 14280 - }, - { - "epoch": 1.31, - "learning_rate": 1.0424657534246576e-05, - "loss": 0.0076, - "step": 14290 - }, - { - "epoch": 1.31, - "learning_rate": 1.0410958904109589e-05, - "loss": 0.0107, - "step": 14300 - }, - { - "epoch": 1.31, - "learning_rate": 1.0397260273972604e-05, - "loss": 0.0101, - "step": 14310 - }, - { - "epoch": 1.31, - "learning_rate": 1.0383561643835617e-05, - "loss": 0.0082, - "step": 14320 - }, - { - "epoch": 1.31, - "learning_rate": 1.036986301369863e-05, - "loss": 0.009, - "step": 14330 - }, - { - "epoch": 1.31, - "learning_rate": 1.0356164383561646e-05, - "loss": 0.0104, - "step": 14340 - }, - { - "epoch": 1.31, - "learning_rate": 1.0342465753424659e-05, - "loss": 0.0091, - "step": 14350 - }, - { - "epoch": 1.31, - "learning_rate": 1.0328767123287671e-05, - "loss": 0.0121, - "step": 14360 - }, - { - "epoch": 1.31, - "learning_rate": 1.0315068493150684e-05, - "loss": 0.0097, - "step": 14370 - }, - { - "epoch": 1.31, - "learning_rate": 1.0301369863013699e-05, - "loss": 0.0087, - "step": 14380 - }, - { - "epoch": 1.31, - "learning_rate": 1.0287671232876712e-05, - "loss": 0.013, - "step": 14390 - }, - { - "epoch": 1.32, - "learning_rate": 1.0273972602739726e-05, - "loss": 0.0368, - "step": 14400 - }, - { - "epoch": 1.32, - "learning_rate": 1.0260273972602741e-05, - "loss": 0.0123, - "step": 14410 - }, - { - "epoch": 1.32, - "learning_rate": 1.0246575342465754e-05, - "loss": 0.0132, - "step": 14420 - }, - { - "epoch": 1.32, - "learning_rate": 1.0232876712328767e-05, - "loss": 0.0114, - "step": 14430 - }, - { - "epoch": 1.32, - "learning_rate": 1.0219178082191781e-05, - "loss": 0.0104, - "step": 14440 - }, - { - "epoch": 1.32, - "learning_rate": 1.0205479452054796e-05, - "loss": 0.0118, - "step": 14450 - }, - { - "epoch": 1.32, - "learning_rate": 1.0191780821917809e-05, - "loss": 0.0084, - "step": 14460 - }, - { - "epoch": 1.32, - "learning_rate": 1.0178082191780822e-05, - "loss": 0.0099, - "step": 14470 - }, - { - "epoch": 1.32, - "learning_rate": 1.0164383561643836e-05, - "loss": 0.0113, - "step": 14480 - }, - { - "epoch": 1.32, - "learning_rate": 1.0150684931506849e-05, - "loss": 0.0095, - "step": 14490 - }, - { - "epoch": 1.32, - "learning_rate": 1.0136986301369862e-05, - "loss": 0.0093, - "step": 14500 - }, - { - "epoch": 1.33, - "learning_rate": 1.0123287671232878e-05, - "loss": 0.0089, - "step": 14510 - }, - { - "epoch": 1.33, - "learning_rate": 1.0109589041095891e-05, - "loss": 0.0125, - "step": 14520 - }, - { - "epoch": 1.33, - "learning_rate": 1.0095890410958904e-05, - "loss": 0.0101, - "step": 14530 - }, - { - "epoch": 1.33, - "learning_rate": 1.0082191780821918e-05, - "loss": 0.0127, - "step": 14540 - }, - { - "epoch": 1.33, - "learning_rate": 1.0068493150684931e-05, - "loss": 0.014, - "step": 14550 - }, - { - "epoch": 1.33, - "learning_rate": 1.0054794520547946e-05, - "loss": 0.0102, - "step": 14560 - }, - { - "epoch": 1.33, - "learning_rate": 1.0041095890410959e-05, - "loss": 0.0127, - "step": 14570 - }, - { - "epoch": 1.33, - "learning_rate": 1.0027397260273973e-05, - "loss": 0.0121, - "step": 14580 - }, - { - "epoch": 1.33, - "learning_rate": 1.0013698630136986e-05, - "loss": 0.0084, - "step": 14590 - }, - { - "epoch": 1.33, - "learning_rate": 9.999999999999999e-06, - "loss": 0.0087, - "step": 14600 - }, - { - "epoch": 1.33, - "learning_rate": 9.986301369863015e-06, - "loss": 0.0115, - "step": 14610 - }, - { - "epoch": 1.34, - "learning_rate": 9.972602739726028e-06, - "loss": 0.032, - "step": 14620 - }, - { - "epoch": 1.34, - "learning_rate": 9.958904109589041e-06, - "loss": 0.0134, - "step": 14630 - }, - { - "epoch": 1.34, - "learning_rate": 9.945205479452056e-06, - "loss": 0.0124, - "step": 14640 - }, - { - "epoch": 1.34, - "learning_rate": 9.931506849315069e-06, - "loss": 0.0101, - "step": 14650 - }, - { - "epoch": 1.34, - "learning_rate": 9.917808219178081e-06, - "loss": 0.0116, - "step": 14660 - }, - { - "epoch": 1.34, - "learning_rate": 9.904109589041096e-06, - "loss": 0.01, - "step": 14670 - }, - { - "epoch": 1.34, - "learning_rate": 9.89041095890411e-06, - "loss": 0.0131, - "step": 14680 - }, - { - "epoch": 1.34, - "learning_rate": 9.876712328767123e-06, - "loss": 0.0097, - "step": 14690 - }, - { - "epoch": 1.34, - "learning_rate": 9.863013698630136e-06, - "loss": 0.0076, - "step": 14700 - }, - { - "epoch": 1.34, - "learning_rate": 9.849315068493151e-06, - "loss": 0.0137, - "step": 14710 - }, - { - "epoch": 1.34, - "learning_rate": 9.835616438356166e-06, - "loss": 0.0125, - "step": 14720 - }, - { - "epoch": 1.35, - "learning_rate": 9.821917808219178e-06, - "loss": 0.0092, - "step": 14730 - }, - { - "epoch": 1.35, - "learning_rate": 9.808219178082193e-06, - "loss": 0.0088, - "step": 14740 - }, - { - "epoch": 1.35, - "learning_rate": 9.794520547945206e-06, - "loss": 0.0119, - "step": 14750 - }, - { - "epoch": 1.35, - "learning_rate": 9.780821917808219e-06, - "loss": 0.0119, - "step": 14760 - }, - { - "epoch": 1.35, - "learning_rate": 9.767123287671232e-06, - "loss": 0.0103, - "step": 14770 - }, - { - "epoch": 1.35, - "learning_rate": 9.753424657534248e-06, - "loss": 0.0077, - "step": 14780 - }, - { - "epoch": 1.35, - "learning_rate": 9.73972602739726e-06, - "loss": 0.0088, - "step": 14790 - }, - { - "epoch": 1.35, - "learning_rate": 9.726027397260274e-06, - "loss": 0.0082, - "step": 14800 - }, - { - "epoch": 1.35, - "learning_rate": 9.712328767123288e-06, - "loss": 0.009, - "step": 14810 - }, - { - "epoch": 1.35, - "learning_rate": 9.698630136986301e-06, - "loss": 0.0102, - "step": 14820 - }, - { - "epoch": 1.35, - "learning_rate": 9.684931506849316e-06, - "loss": 0.0109, - "step": 14830 - }, - { - "epoch": 1.36, - "learning_rate": 9.67123287671233e-06, - "loss": 0.0114, - "step": 14840 - }, - { - "epoch": 1.36, - "learning_rate": 9.657534246575343e-06, - "loss": 0.0116, - "step": 14850 - }, - { - "epoch": 1.36, - "learning_rate": 9.643835616438356e-06, - "loss": 0.0117, - "step": 14860 - }, - { - "epoch": 1.36, - "learning_rate": 9.630136986301369e-06, - "loss": 0.0078, - "step": 14870 - }, - { - "epoch": 1.36, - "learning_rate": 9.616438356164383e-06, - "loss": 0.0089, - "step": 14880 - }, - { - "epoch": 1.36, - "learning_rate": 9.602739726027398e-06, - "loss": 0.0125, - "step": 14890 - }, - { - "epoch": 1.36, - "learning_rate": 9.589041095890411e-06, - "loss": 0.0123, - "step": 14900 - }, - { - "epoch": 1.36, - "learning_rate": 9.575342465753425e-06, - "loss": 0.0084, - "step": 14910 - }, - { - "epoch": 1.36, - "learning_rate": 9.561643835616438e-06, - "loss": 0.0092, - "step": 14920 - }, - { - "epoch": 1.36, - "learning_rate": 9.547945205479451e-06, - "loss": 0.0134, - "step": 14930 - }, - { - "epoch": 1.36, - "learning_rate": 9.534246575342467e-06, - "loss": 0.0078, - "step": 14940 - }, - { - "epoch": 1.37, - "learning_rate": 9.52054794520548e-06, - "loss": 0.013, - "step": 14950 - }, - { - "epoch": 1.37, - "learning_rate": 9.506849315068493e-06, - "loss": 0.0107, - "step": 14960 - }, - { - "epoch": 1.37, - "learning_rate": 9.493150684931506e-06, - "loss": 0.0114, - "step": 14970 - }, - { - "epoch": 1.37, - "learning_rate": 9.47945205479452e-06, - "loss": 0.0111, - "step": 14980 - }, - { - "epoch": 1.37, - "learning_rate": 9.465753424657535e-06, - "loss": 0.012, - "step": 14990 - }, - { - "epoch": 1.37, - "learning_rate": 9.452054794520548e-06, - "loss": 0.0106, - "step": 15000 - }, - { - "epoch": 1.37, - "learning_rate": 9.438356164383563e-06, - "loss": 0.0125, - "step": 15010 - }, - { - "epoch": 1.37, - "learning_rate": 9.424657534246576e-06, - "loss": 0.0519, - "step": 15020 - }, - { - "epoch": 1.37, - "learning_rate": 9.410958904109588e-06, - "loss": 0.011, - "step": 15030 - }, - { - "epoch": 1.37, - "learning_rate": 9.397260273972603e-06, - "loss": 0.0082, - "step": 15040 - }, - { - "epoch": 1.37, - "learning_rate": 9.383561643835618e-06, - "loss": 0.0491, - "step": 15050 - }, - { - "epoch": 1.38, - "learning_rate": 9.36986301369863e-06, - "loss": 0.0079, - "step": 15060 - }, - { - "epoch": 1.38, - "learning_rate": 9.356164383561643e-06, - "loss": 0.0137, - "step": 15070 - }, - { - "epoch": 1.38, - "learning_rate": 9.342465753424658e-06, - "loss": 0.0114, - "step": 15080 - }, - { - "epoch": 1.38, - "learning_rate": 9.32876712328767e-06, - "loss": 0.0071, - "step": 15090 - }, - { - "epoch": 1.38, - "learning_rate": 9.315068493150685e-06, - "loss": 0.0116, - "step": 15100 - }, - { - "epoch": 1.38, - "learning_rate": 9.3013698630137e-06, - "loss": 0.012, - "step": 15110 - }, - { - "epoch": 1.38, - "learning_rate": 9.287671232876713e-06, - "loss": 0.0104, - "step": 15120 - }, - { - "epoch": 1.38, - "learning_rate": 9.273972602739726e-06, - "loss": 0.0104, - "step": 15130 - }, - { - "epoch": 1.38, - "learning_rate": 9.26027397260274e-06, - "loss": 0.0085, - "step": 15140 - }, - { - "epoch": 1.38, - "learning_rate": 9.246575342465753e-06, - "loss": 0.0234, - "step": 15150 - }, - { - "epoch": 1.38, - "learning_rate": 9.232876712328768e-06, - "loss": 0.0097, - "step": 15160 - }, - { - "epoch": 1.39, - "learning_rate": 9.21917808219178e-06, - "loss": 0.0105, - "step": 15170 - }, - { - "epoch": 1.39, - "learning_rate": 9.205479452054795e-06, - "loss": 0.0109, - "step": 15180 - }, - { - "epoch": 1.39, - "learning_rate": 9.191780821917808e-06, - "loss": 0.0146, - "step": 15190 - }, - { - "epoch": 1.39, - "learning_rate": 9.178082191780821e-06, - "loss": 0.0149, - "step": 15200 - }, - { - "epoch": 1.39, - "learning_rate": 9.164383561643837e-06, - "loss": 0.0087, - "step": 15210 - }, - { - "epoch": 1.39, - "learning_rate": 9.15068493150685e-06, - "loss": 0.0078, - "step": 15220 - }, - { - "epoch": 1.39, - "learning_rate": 9.136986301369863e-06, - "loss": 0.0167, - "step": 15230 - }, - { - "epoch": 1.39, - "learning_rate": 9.123287671232878e-06, - "loss": 0.0121, - "step": 15240 - }, - { - "epoch": 1.39, - "learning_rate": 9.10958904109589e-06, - "loss": 0.0092, - "step": 15250 - }, - { - "epoch": 1.39, - "learning_rate": 9.095890410958905e-06, - "loss": 0.0118, - "step": 15260 - }, - { - "epoch": 1.39, - "learning_rate": 9.082191780821918e-06, - "loss": 0.0085, - "step": 15270 - }, - { - "epoch": 1.4, - "learning_rate": 9.068493150684932e-06, - "loss": 0.0111, - "step": 15280 - }, - { - "epoch": 1.4, - "learning_rate": 9.054794520547945e-06, - "loss": 0.0138, - "step": 15290 - }, - { - "epoch": 1.4, - "learning_rate": 9.041095890410958e-06, - "loss": 0.0068, - "step": 15300 - }, - { - "epoch": 1.4, - "learning_rate": 9.027397260273973e-06, - "loss": 0.0102, - "step": 15310 - }, - { - "epoch": 1.4, - "learning_rate": 9.013698630136987e-06, - "loss": 0.0101, - "step": 15320 - }, - { - "epoch": 1.4, - "learning_rate": 9e-06, - "loss": 0.0084, - "step": 15330 - }, - { - "epoch": 1.4, - "learning_rate": 8.986301369863013e-06, - "loss": 0.0085, - "step": 15340 - }, - { - "epoch": 1.4, - "learning_rate": 8.972602739726028e-06, - "loss": 0.0113, - "step": 15350 - }, - { - "epoch": 1.4, - "learning_rate": 8.95890410958904e-06, - "loss": 0.0099, - "step": 15360 - }, - { - "epoch": 1.4, - "learning_rate": 8.945205479452055e-06, - "loss": 0.0109, - "step": 15370 - }, - { - "epoch": 1.4, - "learning_rate": 8.93150684931507e-06, - "loss": 0.0143, - "step": 15380 - }, - { - "epoch": 1.41, - "learning_rate": 8.917808219178083e-06, - "loss": 0.01, - "step": 15390 - }, - { - "epoch": 1.41, - "learning_rate": 8.904109589041095e-06, - "loss": 0.007, - "step": 15400 - }, - { - "epoch": 1.41, - "learning_rate": 8.89041095890411e-06, - "loss": 0.0067, - "step": 15410 - }, - { - "epoch": 1.41, - "learning_rate": 8.876712328767123e-06, - "loss": 0.0109, - "step": 15420 - }, - { - "epoch": 1.41, - "learning_rate": 8.863013698630137e-06, - "loss": 0.0115, - "step": 15430 - }, - { - "epoch": 1.41, - "learning_rate": 8.84931506849315e-06, - "loss": 0.0112, - "step": 15440 - }, - { - "epoch": 1.41, - "learning_rate": 8.835616438356165e-06, - "loss": 0.0111, - "step": 15450 - }, - { - "epoch": 1.41, - "learning_rate": 8.821917808219178e-06, - "loss": 0.0115, - "step": 15460 - }, - { - "epoch": 1.41, - "learning_rate": 8.80821917808219e-06, - "loss": 0.0063, - "step": 15470 - }, - { - "epoch": 1.41, - "learning_rate": 8.794520547945207e-06, - "loss": 0.0122, - "step": 15480 - }, - { - "epoch": 1.41, - "learning_rate": 8.78082191780822e-06, - "loss": 0.0093, - "step": 15490 - }, - { - "epoch": 1.42, - "learning_rate": 8.767123287671233e-06, - "loss": 0.0095, - "step": 15500 - }, - { - "epoch": 1.42, - "learning_rate": 8.753424657534247e-06, - "loss": 0.0117, - "step": 15510 - }, - { - "epoch": 1.42, - "learning_rate": 8.73972602739726e-06, - "loss": 0.008, - "step": 15520 - }, - { - "epoch": 1.42, - "learning_rate": 8.726027397260273e-06, - "loss": 0.009, - "step": 15530 - }, - { - "epoch": 1.42, - "learning_rate": 8.712328767123288e-06, - "loss": 0.0115, - "step": 15540 - }, - { - "epoch": 1.42, - "learning_rate": 8.698630136986302e-06, - "loss": 0.0071, - "step": 15550 - }, - { - "epoch": 1.42, - "learning_rate": 8.684931506849315e-06, - "loss": 0.0103, - "step": 15560 - }, - { - "epoch": 1.42, - "learning_rate": 8.671232876712328e-06, - "loss": 0.0079, - "step": 15570 - }, - { - "epoch": 1.42, - "learning_rate": 8.657534246575343e-06, - "loss": 0.01, - "step": 15580 - }, - { - "epoch": 1.42, - "learning_rate": 8.643835616438357e-06, - "loss": 0.0073, - "step": 15590 - }, - { - "epoch": 1.42, - "learning_rate": 8.63013698630137e-06, - "loss": 0.0119, - "step": 15600 - }, - { - "epoch": 1.43, - "learning_rate": 8.616438356164385e-06, - "loss": 0.0181, - "step": 15610 - }, - { - "epoch": 1.43, - "learning_rate": 8.602739726027397e-06, - "loss": 0.0082, - "step": 15620 - }, - { - "epoch": 1.43, - "learning_rate": 8.58904109589041e-06, - "loss": 0.008, - "step": 15630 - }, - { - "epoch": 1.43, - "learning_rate": 8.575342465753425e-06, - "loss": 0.0082, - "step": 15640 - }, - { - "epoch": 1.43, - "learning_rate": 8.56164383561644e-06, - "loss": 0.0095, - "step": 15650 - }, - { - "epoch": 1.43, - "learning_rate": 8.547945205479452e-06, - "loss": 0.0088, - "step": 15660 - }, - { - "epoch": 1.43, - "learning_rate": 8.534246575342465e-06, - "loss": 0.0077, - "step": 15670 - }, - { - "epoch": 1.43, - "learning_rate": 8.52054794520548e-06, - "loss": 0.0082, - "step": 15680 - }, - { - "epoch": 1.43, - "learning_rate": 8.506849315068493e-06, - "loss": 0.011, - "step": 15690 - }, - { - "epoch": 1.43, - "learning_rate": 8.493150684931507e-06, - "loss": 0.0098, - "step": 15700 - }, - { - "epoch": 1.43, - "learning_rate": 8.479452054794522e-06, - "loss": 0.009, - "step": 15710 - }, - { - "epoch": 1.44, - "learning_rate": 8.465753424657535e-06, - "loss": 0.0112, - "step": 15720 - }, - { - "epoch": 1.44, - "learning_rate": 8.452054794520548e-06, - "loss": 0.0193, - "step": 15730 - }, - { - "epoch": 1.44, - "learning_rate": 8.43835616438356e-06, - "loss": 0.0472, - "step": 15740 - }, - { - "epoch": 1.44, - "learning_rate": 8.424657534246577e-06, - "loss": 0.0109, - "step": 15750 - }, - { - "epoch": 1.44, - "learning_rate": 8.41095890410959e-06, - "loss": 0.0103, - "step": 15760 - }, - { - "epoch": 1.44, - "learning_rate": 8.397260273972602e-06, - "loss": 0.0101, - "step": 15770 - }, - { - "epoch": 1.44, - "learning_rate": 8.383561643835617e-06, - "loss": 0.0087, - "step": 15780 - }, - { - "epoch": 1.44, - "learning_rate": 8.36986301369863e-06, - "loss": 0.0101, - "step": 15790 - }, - { - "epoch": 1.44, - "learning_rate": 8.356164383561643e-06, - "loss": 0.0078, - "step": 15800 - }, - { - "epoch": 1.44, - "learning_rate": 8.342465753424659e-06, - "loss": 0.009, - "step": 15810 - }, - { - "epoch": 1.44, - "learning_rate": 8.328767123287672e-06, - "loss": 0.01, - "step": 15820 - }, - { - "epoch": 1.45, - "learning_rate": 8.315068493150685e-06, - "loss": 0.0075, - "step": 15830 - }, - { - "epoch": 1.45, - "learning_rate": 8.301369863013698e-06, - "loss": 0.0098, - "step": 15840 - }, - { - "epoch": 1.45, - "learning_rate": 8.287671232876712e-06, - "loss": 0.0133, - "step": 15850 - }, - { - "epoch": 1.45, - "learning_rate": 8.273972602739727e-06, - "loss": 0.0109, - "step": 15860 - }, - { - "epoch": 1.45, - "learning_rate": 8.26027397260274e-06, - "loss": 0.0113, - "step": 15870 - }, - { - "epoch": 1.45, - "learning_rate": 8.246575342465754e-06, - "loss": 0.0102, - "step": 15880 - }, - { - "epoch": 1.45, - "learning_rate": 8.232876712328767e-06, - "loss": 0.0135, - "step": 15890 - }, - { - "epoch": 1.45, - "learning_rate": 8.21917808219178e-06, - "loss": 0.0085, - "step": 15900 - }, - { - "epoch": 1.45, - "learning_rate": 8.205479452054795e-06, - "loss": 0.0113, - "step": 15910 - }, - { - "epoch": 1.45, - "learning_rate": 8.19178082191781e-06, - "loss": 0.0097, - "step": 15920 - }, - { - "epoch": 1.45, - "learning_rate": 8.178082191780822e-06, - "loss": 0.0087, - "step": 15930 - }, - { - "epoch": 1.46, - "learning_rate": 8.164383561643835e-06, - "loss": 0.012, - "step": 15940 - }, - { - "epoch": 1.46, - "learning_rate": 8.15068493150685e-06, - "loss": 0.0118, - "step": 15950 - }, - { - "epoch": 1.46, - "learning_rate": 8.136986301369862e-06, - "loss": 0.0095, - "step": 15960 - }, - { - "epoch": 1.46, - "learning_rate": 8.123287671232877e-06, - "loss": 0.0123, - "step": 15970 - }, - { - "epoch": 1.46, - "learning_rate": 8.109589041095892e-06, - "loss": 0.0122, - "step": 15980 - }, - { - "epoch": 1.46, - "learning_rate": 8.095890410958904e-06, - "loss": 0.0231, - "step": 15990 - }, - { - "epoch": 1.46, - "learning_rate": 8.082191780821917e-06, - "loss": 0.0094, - "step": 16000 - }, - { - "epoch": 1.46, - "learning_rate": 8.068493150684932e-06, - "loss": 0.0081, - "step": 16010 - }, - { - "epoch": 1.46, - "learning_rate": 8.054794520547946e-06, - "loss": 0.0121, - "step": 16020 - }, - { - "epoch": 1.46, - "learning_rate": 8.04109589041096e-06, - "loss": 0.0094, - "step": 16030 - }, - { - "epoch": 1.46, - "learning_rate": 8.027397260273972e-06, - "loss": 0.0077, - "step": 16040 - }, - { - "epoch": 1.47, - "learning_rate": 8.013698630136987e-06, - "loss": 0.0085, - "step": 16050 - }, - { - "epoch": 1.47, - "learning_rate": 8e-06, - "loss": 0.0108, - "step": 16060 - }, - { - "epoch": 1.47, - "learning_rate": 7.986301369863013e-06, - "loss": 0.0079, - "step": 16070 - }, - { - "epoch": 1.47, - "learning_rate": 7.972602739726029e-06, - "loss": 0.0132, - "step": 16080 - }, - { - "epoch": 1.47, - "learning_rate": 7.958904109589042e-06, - "loss": 0.0092, - "step": 16090 - }, - { - "epoch": 1.47, - "learning_rate": 7.945205479452055e-06, - "loss": 0.0097, - "step": 16100 - }, - { - "epoch": 1.47, - "learning_rate": 7.931506849315069e-06, - "loss": 0.0111, - "step": 16110 - }, - { - "epoch": 1.47, - "learning_rate": 7.917808219178082e-06, - "loss": 0.0095, - "step": 16120 - }, - { - "epoch": 1.47, - "learning_rate": 7.904109589041097e-06, - "loss": 0.0074, - "step": 16130 - }, - { - "epoch": 1.47, - "learning_rate": 7.89041095890411e-06, - "loss": 0.0121, - "step": 16140 - }, - { - "epoch": 1.47, - "learning_rate": 7.876712328767124e-06, - "loss": 0.0105, - "step": 16150 - }, - { - "epoch": 1.48, - "learning_rate": 7.863013698630137e-06, - "loss": 0.0068, - "step": 16160 - }, - { - "epoch": 1.48, - "learning_rate": 7.84931506849315e-06, - "loss": 0.0098, - "step": 16170 - }, - { - "epoch": 1.48, - "learning_rate": 7.835616438356164e-06, - "loss": 0.0116, - "step": 16180 - }, - { - "epoch": 1.48, - "learning_rate": 7.821917808219179e-06, - "loss": 0.0102, - "step": 16190 - }, - { - "epoch": 1.48, - "learning_rate": 7.808219178082192e-06, - "loss": 0.0074, - "step": 16200 - }, - { - "epoch": 1.48, - "learning_rate": 7.794520547945206e-06, - "loss": 0.0114, - "step": 16210 - }, - { - "epoch": 1.48, - "learning_rate": 7.78082191780822e-06, - "loss": 0.0088, - "step": 16220 - }, - { - "epoch": 1.48, - "learning_rate": 7.767123287671232e-06, - "loss": 0.0091, - "step": 16230 - }, - { - "epoch": 1.48, - "learning_rate": 7.753424657534247e-06, - "loss": 0.0104, - "step": 16240 - }, - { - "epoch": 1.48, - "learning_rate": 7.739726027397261e-06, - "loss": 0.0082, - "step": 16250 - }, - { - "epoch": 1.48, - "learning_rate": 7.726027397260274e-06, - "loss": 0.012, - "step": 16260 - }, - { - "epoch": 1.49, - "learning_rate": 7.712328767123287e-06, - "loss": 0.0068, - "step": 16270 - }, - { - "epoch": 1.49, - "learning_rate": 7.698630136986302e-06, - "loss": 0.0137, - "step": 16280 - }, - { - "epoch": 1.49, - "learning_rate": 7.684931506849316e-06, - "loss": 0.0074, - "step": 16290 - }, - { - "epoch": 1.49, - "learning_rate": 7.671232876712329e-06, - "loss": 0.0079, - "step": 16300 - }, - { - "epoch": 1.49, - "learning_rate": 7.657534246575344e-06, - "loss": 0.0091, - "step": 16310 - }, - { - "epoch": 1.49, - "learning_rate": 7.643835616438356e-06, - "loss": 0.0083, - "step": 16320 - }, - { - "epoch": 1.49, - "learning_rate": 7.63013698630137e-06, - "loss": 0.016, - "step": 16330 - }, - { - "epoch": 1.49, - "learning_rate": 7.616438356164383e-06, - "loss": 0.0096, - "step": 16340 - }, - { - "epoch": 1.49, - "learning_rate": 7.602739726027398e-06, - "loss": 0.0074, - "step": 16350 - }, - { - "epoch": 1.49, - "learning_rate": 7.589041095890411e-06, - "loss": 0.0125, - "step": 16360 - }, - { - "epoch": 1.49, - "learning_rate": 7.575342465753424e-06, - "loss": 0.0428, - "step": 16370 - }, - { - "epoch": 1.5, - "learning_rate": 7.561643835616439e-06, - "loss": 0.0093, - "step": 16380 - }, - { - "epoch": 1.5, - "learning_rate": 7.5479452054794526e-06, - "loss": 0.0104, - "step": 16390 - }, - { - "epoch": 1.5, - "learning_rate": 7.5342465753424655e-06, - "loss": 0.0113, - "step": 16400 - }, - { - "epoch": 1.5, - "learning_rate": 7.52054794520548e-06, - "loss": 0.012, - "step": 16410 - }, - { - "epoch": 1.5, - "learning_rate": 7.506849315068494e-06, - "loss": 0.0147, - "step": 16420 - }, - { - "epoch": 1.5, - "learning_rate": 7.493150684931507e-06, - "loss": 0.013, - "step": 16430 - }, - { - "epoch": 1.5, - "learning_rate": 7.479452054794521e-06, - "loss": 0.0104, - "step": 16440 - }, - { - "epoch": 1.5, - "learning_rate": 7.465753424657534e-06, - "loss": 0.0087, - "step": 16450 - }, - { - "epoch": 1.5, - "learning_rate": 7.452054794520548e-06, - "loss": 0.0114, - "step": 16460 - }, - { - "epoch": 1.5, - "learning_rate": 7.438356164383562e-06, - "loss": 0.0111, - "step": 16470 - }, - { - "epoch": 1.51, - "learning_rate": 7.424657534246575e-06, - "loss": 0.0107, - "step": 16480 - }, - { - "epoch": 1.51, - "learning_rate": 7.410958904109589e-06, - "loss": 0.0079, - "step": 16490 - }, - { - "epoch": 1.51, - "learning_rate": 7.397260273972603e-06, - "loss": 0.0112, - "step": 16500 - }, - { - "epoch": 1.51, - "learning_rate": 7.3835616438356164e-06, - "loss": 0.0097, - "step": 16510 - }, - { - "epoch": 1.51, - "learning_rate": 7.36986301369863e-06, - "loss": 0.011, - "step": 16520 - }, - { - "epoch": 1.51, - "learning_rate": 7.356164383561644e-06, - "loss": 0.0075, - "step": 16530 - }, - { - "epoch": 1.51, - "learning_rate": 7.342465753424658e-06, - "loss": 0.01, - "step": 16540 - }, - { - "epoch": 1.51, - "learning_rate": 7.328767123287671e-06, - "loss": 0.0106, - "step": 16550 - }, - { - "epoch": 1.51, - "learning_rate": 7.315068493150685e-06, - "loss": 0.0123, - "step": 16560 - }, - { - "epoch": 1.51, - "learning_rate": 7.301369863013699e-06, - "loss": 0.0059, - "step": 16570 - }, - { - "epoch": 1.51, - "learning_rate": 7.2876712328767125e-06, - "loss": 0.0131, - "step": 16580 - }, - { - "epoch": 1.52, - "learning_rate": 7.273972602739726e-06, - "loss": 0.0129, - "step": 16590 - }, - { - "epoch": 1.52, - "learning_rate": 7.260273972602739e-06, - "loss": 0.009, - "step": 16600 - }, - { - "epoch": 1.52, - "learning_rate": 7.246575342465754e-06, - "loss": 0.0093, - "step": 16610 - }, - { - "epoch": 1.52, - "learning_rate": 7.232876712328767e-06, - "loss": 0.0133, - "step": 16620 - }, - { - "epoch": 1.52, - "learning_rate": 7.219178082191781e-06, - "loss": 0.0115, - "step": 16630 - }, - { - "epoch": 1.52, - "learning_rate": 7.205479452054795e-06, - "loss": 0.0079, - "step": 16640 - }, - { - "epoch": 1.52, - "learning_rate": 7.191780821917808e-06, - "loss": 0.0102, - "step": 16650 - }, - { - "epoch": 1.52, - "learning_rate": 7.178082191780822e-06, - "loss": 0.0138, - "step": 16660 - }, - { - "epoch": 1.52, - "learning_rate": 7.164383561643836e-06, - "loss": 0.0173, - "step": 16670 - }, - { - "epoch": 1.52, - "learning_rate": 7.150684931506849e-06, - "loss": 0.0098, - "step": 16680 - }, - { - "epoch": 1.52, - "learning_rate": 7.1369863013698635e-06, - "loss": 0.0113, - "step": 16690 - }, - { - "epoch": 1.53, - "learning_rate": 7.123287671232876e-06, - "loss": 0.0131, - "step": 16700 - }, - { - "epoch": 1.53, - "learning_rate": 7.109589041095891e-06, - "loss": 0.0064, - "step": 16710 - }, - { - "epoch": 1.53, - "learning_rate": 7.095890410958905e-06, - "loss": 0.0088, - "step": 16720 - }, - { - "epoch": 1.53, - "learning_rate": 7.0821917808219175e-06, - "loss": 0.0148, - "step": 16730 - }, - { - "epoch": 1.53, - "learning_rate": 7.068493150684932e-06, - "loss": 0.0102, - "step": 16740 - }, - { - "epoch": 1.53, - "learning_rate": 7.054794520547945e-06, - "loss": 0.0103, - "step": 16750 - }, - { - "epoch": 1.53, - "learning_rate": 7.041095890410959e-06, - "loss": 0.009, - "step": 16760 - }, - { - "epoch": 1.53, - "learning_rate": 7.027397260273973e-06, - "loss": 0.0099, - "step": 16770 - }, - { - "epoch": 1.53, - "learning_rate": 7.013698630136986e-06, - "loss": 0.0156, - "step": 16780 - }, - { - "epoch": 1.53, - "learning_rate": 7e-06, - "loss": 0.0108, - "step": 16790 - }, - { - "epoch": 1.53, - "learning_rate": 6.986301369863014e-06, - "loss": 0.0114, - "step": 16800 - }, - { - "epoch": 1.54, - "learning_rate": 6.972602739726027e-06, - "loss": 0.0119, - "step": 16810 - }, - { - "epoch": 1.54, - "learning_rate": 6.958904109589042e-06, - "loss": 0.0143, - "step": 16820 - }, - { - "epoch": 1.54, - "learning_rate": 6.945205479452055e-06, - "loss": 0.0096, - "step": 16830 - }, - { - "epoch": 1.54, - "learning_rate": 6.9315068493150685e-06, - "loss": 0.0102, - "step": 16840 - }, - { - "epoch": 1.54, - "learning_rate": 6.917808219178082e-06, - "loss": 0.0106, - "step": 16850 - }, - { - "epoch": 1.54, - "learning_rate": 6.904109589041096e-06, - "loss": 0.0128, - "step": 16860 - }, - { - "epoch": 1.54, - "learning_rate": 6.89041095890411e-06, - "loss": 0.015, - "step": 16870 - }, - { - "epoch": 1.54, - "learning_rate": 6.876712328767123e-06, - "loss": 0.0076, - "step": 16880 - }, - { - "epoch": 1.54, - "learning_rate": 6.863013698630137e-06, - "loss": 0.009, - "step": 16890 - }, - { - "epoch": 1.54, - "learning_rate": 6.849315068493151e-06, - "loss": 0.0086, - "step": 16900 - }, - { - "epoch": 1.54, - "learning_rate": 6.835616438356165e-06, - "loss": 0.0105, - "step": 16910 - }, - { - "epoch": 1.55, - "learning_rate": 6.821917808219178e-06, - "loss": 0.0091, - "step": 16920 - }, - { - "epoch": 1.55, - "learning_rate": 6.808219178082192e-06, - "loss": 0.0082, - "step": 16930 - }, - { - "epoch": 1.55, - "learning_rate": 6.794520547945206e-06, - "loss": 0.0079, - "step": 16940 - }, - { - "epoch": 1.55, - "learning_rate": 6.780821917808219e-06, - "loss": 0.0084, - "step": 16950 - }, - { - "epoch": 1.55, - "learning_rate": 6.767123287671233e-06, - "loss": 0.007, - "step": 16960 - }, - { - "epoch": 1.55, - "learning_rate": 6.753424657534247e-06, - "loss": 0.0076, - "step": 16970 - }, - { - "epoch": 1.55, - "learning_rate": 6.73972602739726e-06, - "loss": 0.0105, - "step": 16980 - }, - { - "epoch": 1.55, - "learning_rate": 6.726027397260274e-06, - "loss": 0.0107, - "step": 16990 - }, - { - "epoch": 1.55, - "learning_rate": 6.712328767123287e-06, - "loss": 0.0118, - "step": 17000 - }, - { - "epoch": 1.55, - "learning_rate": 6.698630136986302e-06, - "loss": 0.0093, - "step": 17010 - }, - { - "epoch": 1.55, - "learning_rate": 6.684931506849316e-06, - "loss": 0.0068, - "step": 17020 - }, - { - "epoch": 1.56, - "learning_rate": 6.6712328767123285e-06, - "loss": 0.01, - "step": 17030 - }, - { - "epoch": 1.56, - "learning_rate": 6.657534246575343e-06, - "loss": 0.011, - "step": 17040 - }, - { - "epoch": 1.56, - "learning_rate": 6.643835616438356e-06, - "loss": 0.0094, - "step": 17050 - }, - { - "epoch": 1.56, - "learning_rate": 6.63013698630137e-06, - "loss": 0.0086, - "step": 17060 - }, - { - "epoch": 1.56, - "learning_rate": 6.616438356164384e-06, - "loss": 0.0123, - "step": 17070 - }, - { - "epoch": 1.56, - "learning_rate": 6.602739726027397e-06, - "loss": 0.0095, - "step": 17080 - }, - { - "epoch": 1.56, - "learning_rate": 6.589041095890412e-06, - "loss": 0.0071, - "step": 17090 - }, - { - "epoch": 1.56, - "learning_rate": 6.5753424657534245e-06, - "loss": 0.0096, - "step": 17100 - }, - { - "epoch": 1.56, - "learning_rate": 6.561643835616438e-06, - "loss": 0.0061, - "step": 17110 - }, - { - "epoch": 1.56, - "learning_rate": 6.547945205479453e-06, - "loss": 0.0112, - "step": 17120 - }, - { - "epoch": 1.56, - "learning_rate": 6.534246575342466e-06, - "loss": 0.0079, - "step": 17130 - }, - { - "epoch": 1.57, - "learning_rate": 6.5205479452054794e-06, - "loss": 0.0055, - "step": 17140 - }, - { - "epoch": 1.57, - "learning_rate": 6.506849315068493e-06, - "loss": 0.0439, - "step": 17150 - }, - { - "epoch": 1.57, - "learning_rate": 6.493150684931507e-06, - "loss": 0.0113, - "step": 17160 - }, - { - "epoch": 1.57, - "learning_rate": 6.479452054794521e-06, - "loss": 0.0093, - "step": 17170 - }, - { - "epoch": 1.57, - "learning_rate": 6.465753424657534e-06, - "loss": 0.0096, - "step": 17180 - }, - { - "epoch": 1.57, - "learning_rate": 6.452054794520548e-06, - "loss": 0.0099, - "step": 17190 - }, - { - "epoch": 1.57, - "learning_rate": 6.438356164383562e-06, - "loss": 0.0088, - "step": 17200 - }, - { - "epoch": 1.57, - "learning_rate": 6.4246575342465755e-06, - "loss": 0.0096, - "step": 17210 - }, - { - "epoch": 1.57, - "learning_rate": 6.410958904109589e-06, - "loss": 0.019, - "step": 17220 - }, - { - "epoch": 1.57, - "learning_rate": 6.397260273972603e-06, - "loss": 0.0421, - "step": 17230 - }, - { - "epoch": 1.57, - "learning_rate": 6.383561643835617e-06, - "loss": 0.0132, - "step": 17240 - }, - { - "epoch": 1.58, - "learning_rate": 6.3698630136986296e-06, - "loss": 0.0087, - "step": 17250 - }, - { - "epoch": 1.58, - "learning_rate": 6.356164383561644e-06, - "loss": 0.0097, - "step": 17260 - }, - { - "epoch": 1.58, - "learning_rate": 6.342465753424658e-06, - "loss": 0.0089, - "step": 17270 - }, - { - "epoch": 1.58, - "learning_rate": 6.328767123287672e-06, - "loss": 0.0082, - "step": 17280 - }, - { - "epoch": 1.58, - "learning_rate": 6.315068493150685e-06, - "loss": 0.0087, - "step": 17290 - }, - { - "epoch": 1.58, - "learning_rate": 6.301369863013698e-06, - "loss": 0.0104, - "step": 17300 - }, - { - "epoch": 1.58, - "learning_rate": 6.287671232876713e-06, - "loss": 0.008, - "step": 17310 - }, - { - "epoch": 1.58, - "learning_rate": 6.2739726027397265e-06, - "loss": 0.0093, - "step": 17320 - }, - { - "epoch": 1.58, - "learning_rate": 6.260273972602739e-06, - "loss": 0.0101, - "step": 17330 - }, - { - "epoch": 1.58, - "learning_rate": 6.246575342465754e-06, - "loss": 0.0099, - "step": 17340 - }, - { - "epoch": 1.58, - "learning_rate": 6.232876712328767e-06, - "loss": 0.0102, - "step": 17350 - }, - { - "epoch": 1.59, - "learning_rate": 6.2191780821917805e-06, - "loss": 0.0116, - "step": 17360 - }, - { - "epoch": 1.59, - "learning_rate": 6.205479452054795e-06, - "loss": 0.0096, - "step": 17370 - }, - { - "epoch": 1.59, - "learning_rate": 6.191780821917808e-06, - "loss": 0.0116, - "step": 17380 - }, - { - "epoch": 1.59, - "learning_rate": 6.1780821917808226e-06, - "loss": 0.0077, - "step": 17390 - }, - { - "epoch": 1.59, - "learning_rate": 6.1643835616438354e-06, - "loss": 0.0087, - "step": 17400 - }, - { - "epoch": 1.59, - "learning_rate": 6.150684931506849e-06, - "loss": 0.0083, - "step": 17410 - }, - { - "epoch": 1.59, - "learning_rate": 6.136986301369864e-06, - "loss": 0.0096, - "step": 17420 - }, - { - "epoch": 1.59, - "learning_rate": 6.123287671232877e-06, - "loss": 0.0094, - "step": 17430 - }, - { - "epoch": 1.59, - "learning_rate": 6.10958904109589e-06, - "loss": 0.0114, - "step": 17440 - }, - { - "epoch": 1.59, - "learning_rate": 6.095890410958904e-06, - "loss": 0.0108, - "step": 17450 - }, - { - "epoch": 1.59, - "learning_rate": 6.082191780821918e-06, - "loss": 0.0081, - "step": 17460 - }, - { - "epoch": 1.6, - "learning_rate": 6.068493150684932e-06, - "loss": 0.0101, - "step": 17470 - }, - { - "epoch": 1.6, - "learning_rate": 6.054794520547945e-06, - "loss": 0.0084, - "step": 17480 - }, - { - "epoch": 1.6, - "learning_rate": 6.041095890410959e-06, - "loss": 0.0104, - "step": 17490 - }, - { - "epoch": 1.6, - "learning_rate": 6.027397260273973e-06, - "loss": 0.0136, - "step": 17500 - }, - { - "epoch": 1.6, - "learning_rate": 6.0136986301369864e-06, - "loss": 0.0108, - "step": 17510 - }, - { - "epoch": 1.6, - "learning_rate": 6e-06, - "loss": 0.0098, - "step": 17520 - }, - { - "epoch": 1.6, - "learning_rate": 5.986301369863014e-06, - "loss": 0.0101, - "step": 17530 - }, - { - "epoch": 1.6, - "learning_rate": 5.972602739726028e-06, - "loss": 0.0099, - "step": 17540 - }, - { - "epoch": 1.6, - "learning_rate": 5.958904109589041e-06, - "loss": 0.0091, - "step": 17550 - }, - { - "epoch": 1.6, - "learning_rate": 5.945205479452055e-06, - "loss": 0.0124, - "step": 17560 - }, - { - "epoch": 1.6, - "learning_rate": 5.931506849315069e-06, - "loss": 0.0085, - "step": 17570 - }, - { - "epoch": 1.61, - "learning_rate": 5.9178082191780825e-06, - "loss": 0.0097, - "step": 17580 - }, - { - "epoch": 1.61, - "learning_rate": 5.904109589041096e-06, - "loss": 0.0082, - "step": 17590 - }, - { - "epoch": 1.61, - "learning_rate": 5.890410958904109e-06, - "loss": 0.0108, - "step": 17600 - }, - { - "epoch": 1.61, - "learning_rate": 5.876712328767124e-06, - "loss": 0.0094, - "step": 17610 - }, - { - "epoch": 1.61, - "learning_rate": 5.863013698630137e-06, - "loss": 0.0105, - "step": 17620 - }, - { - "epoch": 1.61, - "learning_rate": 5.84931506849315e-06, - "loss": 0.0087, - "step": 17630 - }, - { - "epoch": 1.61, - "learning_rate": 5.835616438356165e-06, - "loss": 0.0089, - "step": 17640 - }, - { - "epoch": 1.61, - "learning_rate": 5.821917808219178e-06, - "loss": 0.01, - "step": 17650 - }, - { - "epoch": 1.61, - "learning_rate": 5.808219178082192e-06, - "loss": 0.0137, - "step": 17660 - }, - { - "epoch": 1.61, - "learning_rate": 5.794520547945206e-06, - "loss": 0.0109, - "step": 17670 - }, - { - "epoch": 1.61, - "learning_rate": 5.780821917808219e-06, - "loss": 0.0073, - "step": 17680 - }, - { - "epoch": 1.62, - "learning_rate": 5.7671232876712335e-06, - "loss": 0.0105, - "step": 17690 - }, - { - "epoch": 1.62, - "learning_rate": 5.753424657534246e-06, - "loss": 0.0109, - "step": 17700 - }, - { - "epoch": 1.62, - "learning_rate": 5.73972602739726e-06, - "loss": 0.0079, - "step": 17710 - }, - { - "epoch": 1.62, - "learning_rate": 5.726027397260275e-06, - "loss": 0.0093, - "step": 17720 - }, - { - "epoch": 1.62, - "learning_rate": 5.7123287671232875e-06, - "loss": 0.0116, - "step": 17730 - }, - { - "epoch": 1.62, - "learning_rate": 5.698630136986302e-06, - "loss": 0.0095, - "step": 17740 - }, - { - "epoch": 1.62, - "learning_rate": 5.684931506849315e-06, - "loss": 0.0094, - "step": 17750 - }, - { - "epoch": 1.62, - "learning_rate": 5.671232876712329e-06, - "loss": 0.0076, - "step": 17760 - }, - { - "epoch": 1.62, - "learning_rate": 5.657534246575343e-06, - "loss": 0.0059, - "step": 17770 - }, - { - "epoch": 1.62, - "learning_rate": 5.643835616438356e-06, - "loss": 0.0081, - "step": 17780 - }, - { - "epoch": 1.62, - "learning_rate": 5.63013698630137e-06, - "loss": 0.0072, - "step": 17790 - }, - { - "epoch": 1.63, - "learning_rate": 5.616438356164384e-06, - "loss": 0.0127, - "step": 17800 - }, - { - "epoch": 1.63, - "learning_rate": 5.602739726027397e-06, - "loss": 0.0091, - "step": 17810 - }, - { - "epoch": 1.63, - "learning_rate": 5.589041095890411e-06, - "loss": 0.0101, - "step": 17820 - }, - { - "epoch": 1.63, - "learning_rate": 5.575342465753425e-06, - "loss": 0.0094, - "step": 17830 - }, - { - "epoch": 1.63, - "learning_rate": 5.5616438356164385e-06, - "loss": 0.0159, - "step": 17840 - }, - { - "epoch": 1.63, - "learning_rate": 5.547945205479452e-06, - "loss": 0.011, - "step": 17850 - }, - { - "epoch": 1.63, - "learning_rate": 5.534246575342466e-06, - "loss": 0.0095, - "step": 17860 - }, - { - "epoch": 1.63, - "learning_rate": 5.52054794520548e-06, - "loss": 0.0174, - "step": 17870 - }, - { - "epoch": 1.63, - "learning_rate": 5.506849315068493e-06, - "loss": 0.0148, - "step": 17880 - }, - { - "epoch": 1.63, - "learning_rate": 5.493150684931507e-06, - "loss": 0.0103, - "step": 17890 - }, - { - "epoch": 1.63, - "learning_rate": 5.47945205479452e-06, - "loss": 0.0101, - "step": 17900 - }, - { - "epoch": 1.64, - "learning_rate": 5.465753424657535e-06, - "loss": 0.0097, - "step": 17910 - }, - { - "epoch": 1.64, - "learning_rate": 5.452054794520548e-06, - "loss": 0.0058, - "step": 17920 - }, - { - "epoch": 1.64, - "learning_rate": 5.438356164383562e-06, - "loss": 0.0108, - "step": 17930 - }, - { - "epoch": 1.64, - "learning_rate": 5.424657534246576e-06, - "loss": 0.0102, - "step": 17940 - }, - { - "epoch": 1.64, - "learning_rate": 5.410958904109589e-06, - "loss": 0.013, - "step": 17950 - }, - { - "epoch": 1.64, - "learning_rate": 5.397260273972603e-06, - "loss": 0.0074, - "step": 17960 - }, - { - "epoch": 1.64, - "learning_rate": 5.383561643835617e-06, - "loss": 0.0096, - "step": 17970 - }, - { - "epoch": 1.64, - "learning_rate": 5.36986301369863e-06, - "loss": 0.0091, - "step": 17980 - }, - { - "epoch": 1.64, - "learning_rate": 5.356164383561644e-06, - "loss": 0.0094, - "step": 17990 - }, - { - "epoch": 1.64, - "learning_rate": 5.342465753424657e-06, - "loss": 0.0105, - "step": 18000 - }, - { - "epoch": 1.64, - "learning_rate": 5.328767123287671e-06, - "loss": 0.0077, - "step": 18010 - }, - { - "epoch": 1.65, - "learning_rate": 5.3150684931506856e-06, - "loss": 0.0081, - "step": 18020 - }, - { - "epoch": 1.65, - "learning_rate": 5.3013698630136985e-06, - "loss": 0.0115, - "step": 18030 - }, - { - "epoch": 1.65, - "learning_rate": 5.287671232876713e-06, - "loss": 0.0103, - "step": 18040 - }, - { - "epoch": 1.65, - "learning_rate": 5.273972602739726e-06, - "loss": 0.0112, - "step": 18050 - }, - { - "epoch": 1.65, - "learning_rate": 5.26027397260274e-06, - "loss": 0.0078, - "step": 18060 - }, - { - "epoch": 1.65, - "learning_rate": 5.246575342465754e-06, - "loss": 0.0096, - "step": 18070 - }, - { - "epoch": 1.65, - "learning_rate": 5.232876712328767e-06, - "loss": 0.0095, - "step": 18080 - }, - { - "epoch": 1.65, - "learning_rate": 5.219178082191781e-06, - "loss": 0.0115, - "step": 18090 - }, - { - "epoch": 1.65, - "learning_rate": 5.2054794520547945e-06, - "loss": 0.0103, - "step": 18100 - }, - { - "epoch": 1.65, - "learning_rate": 5.191780821917808e-06, - "loss": 0.0088, - "step": 18110 - }, - { - "epoch": 1.65, - "learning_rate": 5.178082191780823e-06, - "loss": 0.0111, - "step": 18120 - }, - { - "epoch": 1.66, - "learning_rate": 5.164383561643836e-06, - "loss": 0.0169, - "step": 18130 - }, - { - "epoch": 1.66, - "learning_rate": 5.1506849315068494e-06, - "loss": 0.0137, - "step": 18140 - }, - { - "epoch": 1.66, - "learning_rate": 5.136986301369863e-06, - "loss": 0.0094, - "step": 18150 - }, - { - "epoch": 1.66, - "learning_rate": 5.123287671232877e-06, - "loss": 0.0096, - "step": 18160 - }, - { - "epoch": 1.66, - "learning_rate": 5.109589041095891e-06, - "loss": 0.0294, - "step": 18170 - }, - { - "epoch": 1.66, - "learning_rate": 5.095890410958904e-06, - "loss": 0.0107, - "step": 18180 - }, - { - "epoch": 1.66, - "learning_rate": 5.082191780821918e-06, - "loss": 0.0106, - "step": 18190 - }, - { - "epoch": 1.66, - "learning_rate": 5.068493150684931e-06, - "loss": 0.0074, - "step": 18200 - }, - { - "epoch": 1.66, - "learning_rate": 5.0547945205479455e-06, - "loss": 0.0101, - "step": 18210 - }, - { - "epoch": 1.66, - "learning_rate": 5.041095890410959e-06, - "loss": 0.0111, - "step": 18220 - }, - { - "epoch": 1.66, - "learning_rate": 5.027397260273973e-06, - "loss": 0.0076, - "step": 18230 - }, - { - "epoch": 1.67, - "learning_rate": 5.013698630136987e-06, - "loss": 0.0091, - "step": 18240 - }, - { - "epoch": 1.67, - "learning_rate": 4.9999999999999996e-06, - "loss": 0.014, - "step": 18250 - }, - { - "epoch": 1.67, - "learning_rate": 4.986301369863014e-06, - "loss": 0.0097, - "step": 18260 - }, - { - "epoch": 1.67, - "learning_rate": 4.972602739726028e-06, - "loss": 0.009, - "step": 18270 - }, - { - "epoch": 1.67, - "learning_rate": 4.958904109589041e-06, - "loss": 0.0096, - "step": 18280 - }, - { - "epoch": 1.67, - "learning_rate": 4.945205479452055e-06, - "loss": 0.0059, - "step": 18290 - }, - { - "epoch": 1.67, - "learning_rate": 4.931506849315068e-06, - "loss": 0.0113, - "step": 18300 - }, - { - "epoch": 1.67, - "learning_rate": 4.917808219178083e-06, - "loss": 0.0088, - "step": 18310 - }, - { - "epoch": 1.67, - "learning_rate": 4.9041095890410965e-06, - "loss": 0.0063, - "step": 18320 - }, - { - "epoch": 1.67, - "learning_rate": 4.890410958904109e-06, - "loss": 0.0063, - "step": 18330 - }, - { - "epoch": 1.67, - "learning_rate": 4.876712328767124e-06, - "loss": 0.0122, - "step": 18340 - }, - { - "epoch": 1.68, - "learning_rate": 4.863013698630137e-06, - "loss": 0.0107, - "step": 18350 - }, - { - "epoch": 1.68, - "learning_rate": 4.8493150684931505e-06, - "loss": 0.0083, - "step": 18360 - }, - { - "epoch": 1.68, - "learning_rate": 4.835616438356165e-06, - "loss": 0.0104, - "step": 18370 - }, - { - "epoch": 1.68, - "learning_rate": 4.821917808219178e-06, - "loss": 0.0119, - "step": 18380 - }, - { - "epoch": 1.68, - "learning_rate": 4.808219178082192e-06, - "loss": 0.0094, - "step": 18390 - }, - { - "epoch": 1.68, - "learning_rate": 4.7945205479452054e-06, - "loss": 0.0099, - "step": 18400 - }, - { - "epoch": 1.68, - "learning_rate": 4.780821917808219e-06, - "loss": 0.0153, - "step": 18410 - }, - { - "epoch": 1.68, - "learning_rate": 4.767123287671234e-06, - "loss": 0.0108, - "step": 18420 - }, - { - "epoch": 1.68, - "learning_rate": 4.753424657534247e-06, - "loss": 0.0069, - "step": 18430 - }, - { - "epoch": 1.68, - "learning_rate": 4.73972602739726e-06, - "loss": 0.0123, - "step": 18440 - }, - { - "epoch": 1.68, - "learning_rate": 4.726027397260274e-06, - "loss": 0.0117, - "step": 18450 - }, - { - "epoch": 1.69, - "learning_rate": 4.712328767123288e-06, - "loss": 0.0113, - "step": 18460 - }, - { - "epoch": 1.69, - "learning_rate": 4.6986301369863015e-06, - "loss": 0.0103, - "step": 18470 - }, - { - "epoch": 1.69, - "learning_rate": 4.684931506849315e-06, - "loss": 0.0078, - "step": 18480 - }, - { - "epoch": 1.69, - "learning_rate": 4.671232876712329e-06, - "loss": 0.0113, - "step": 18490 - }, - { - "epoch": 1.69, - "learning_rate": 4.657534246575343e-06, - "loss": 0.0069, - "step": 18500 - }, - { - "epoch": 1.69, - "learning_rate": 4.643835616438356e-06, - "loss": 0.0141, - "step": 18510 - }, - { - "epoch": 1.69, - "learning_rate": 4.63013698630137e-06, - "loss": 0.0099, - "step": 18520 - }, - { - "epoch": 1.69, - "learning_rate": 4.616438356164384e-06, - "loss": 0.0084, - "step": 18530 - }, - { - "epoch": 1.69, - "learning_rate": 4.602739726027398e-06, - "loss": 0.009, - "step": 18540 - }, - { - "epoch": 1.69, - "learning_rate": 4.5890410958904105e-06, - "loss": 0.0108, - "step": 18550 - }, - { - "epoch": 1.69, - "learning_rate": 4.575342465753425e-06, - "loss": 0.0157, - "step": 18560 - }, - { - "epoch": 1.7, - "learning_rate": 4.561643835616439e-06, - "loss": 0.0084, - "step": 18570 - }, - { - "epoch": 1.7, - "learning_rate": 4.5479452054794525e-06, - "loss": 0.0069, - "step": 18580 - }, - { - "epoch": 1.7, - "learning_rate": 4.534246575342466e-06, - "loss": 0.0083, - "step": 18590 - }, - { - "epoch": 1.7, - "learning_rate": 4.520547945205479e-06, - "loss": 0.0078, - "step": 18600 - }, - { - "epoch": 1.7, - "learning_rate": 4.506849315068494e-06, - "loss": 0.0101, - "step": 18610 - }, - { - "epoch": 1.7, - "learning_rate": 4.4931506849315066e-06, - "loss": 0.0196, - "step": 18620 - }, - { - "epoch": 1.7, - "learning_rate": 4.47945205479452e-06, - "loss": 0.0092, - "step": 18630 - }, - { - "epoch": 1.7, - "learning_rate": 4.465753424657535e-06, - "loss": 0.0102, - "step": 18640 - }, - { - "epoch": 1.7, - "learning_rate": 4.452054794520548e-06, - "loss": 0.0098, - "step": 18650 - }, - { - "epoch": 1.7, - "learning_rate": 4.4383561643835615e-06, - "loss": 0.014, - "step": 18660 - }, - { - "epoch": 1.71, - "learning_rate": 4.424657534246575e-06, - "loss": 0.0147, - "step": 18670 - }, - { - "epoch": 1.71, - "learning_rate": 4.410958904109589e-06, - "loss": 0.0088, - "step": 18680 - }, - { - "epoch": 1.71, - "learning_rate": 4.3972602739726035e-06, - "loss": 0.009, - "step": 18690 - }, - { - "epoch": 1.71, - "learning_rate": 4.383561643835616e-06, - "loss": 0.0101, - "step": 18700 - }, - { - "epoch": 1.71, - "learning_rate": 4.36986301369863e-06, - "loss": 0.0099, - "step": 18710 - }, - { - "epoch": 1.71, - "learning_rate": 4.356164383561644e-06, - "loss": 0.0086, - "step": 18720 - }, - { - "epoch": 1.71, - "learning_rate": 4.3424657534246575e-06, - "loss": 0.0097, - "step": 18730 - }, - { - "epoch": 1.71, - "learning_rate": 4.328767123287671e-06, - "loss": 0.0051, - "step": 18740 - }, - { - "epoch": 1.71, - "learning_rate": 4.315068493150685e-06, - "loss": 0.0091, - "step": 18750 - }, - { - "epoch": 1.71, - "learning_rate": 4.301369863013699e-06, - "loss": 0.0422, - "step": 18760 - }, - { - "epoch": 1.71, - "learning_rate": 4.2876712328767124e-06, - "loss": 0.0119, - "step": 18770 - }, - { - "epoch": 1.72, - "learning_rate": 4.273972602739726e-06, - "loss": 0.0122, - "step": 18780 - }, - { - "epoch": 1.72, - "learning_rate": 4.26027397260274e-06, - "loss": 0.0084, - "step": 18790 - }, - { - "epoch": 1.72, - "learning_rate": 4.246575342465754e-06, - "loss": 0.0077, - "step": 18800 - }, - { - "epoch": 1.72, - "learning_rate": 4.232876712328767e-06, - "loss": 0.0101, - "step": 18810 - }, - { - "epoch": 1.72, - "learning_rate": 4.21917808219178e-06, - "loss": 0.009, - "step": 18820 - }, - { - "epoch": 1.72, - "learning_rate": 4.205479452054795e-06, - "loss": 0.0083, - "step": 18830 - }, - { - "epoch": 1.72, - "learning_rate": 4.1917808219178085e-06, - "loss": 0.0101, - "step": 18840 - }, - { - "epoch": 1.72, - "learning_rate": 4.178082191780821e-06, - "loss": 0.0075, - "step": 18850 - }, - { - "epoch": 1.72, - "learning_rate": 4.164383561643836e-06, - "loss": 0.0091, - "step": 18860 - }, - { - "epoch": 1.72, - "learning_rate": 4.150684931506849e-06, - "loss": 0.0067, - "step": 18870 - }, - { - "epoch": 1.72, - "learning_rate": 4.136986301369863e-06, - "loss": 0.009, - "step": 18880 - }, - { - "epoch": 1.73, - "learning_rate": 4.123287671232877e-06, - "loss": 0.0066, - "step": 18890 - }, - { - "epoch": 1.73, - "learning_rate": 4.10958904109589e-06, - "loss": 0.0081, - "step": 18900 - }, - { - "epoch": 1.73, - "learning_rate": 4.095890410958905e-06, - "loss": 0.0087, - "step": 18910 - }, - { - "epoch": 1.73, - "learning_rate": 4.0821917808219175e-06, - "loss": 0.0102, - "step": 18920 - }, - { - "epoch": 1.73, - "learning_rate": 4.068493150684931e-06, - "loss": 0.0136, - "step": 18930 - }, - { - "epoch": 1.73, - "learning_rate": 4.054794520547946e-06, - "loss": 0.0092, - "step": 18940 - }, - { - "epoch": 1.73, - "learning_rate": 4.041095890410959e-06, - "loss": 0.0116, - "step": 18950 - }, - { - "epoch": 1.73, - "learning_rate": 4.027397260273973e-06, - "loss": 0.0363, - "step": 18960 - }, - { - "epoch": 1.73, - "learning_rate": 4.013698630136986e-06, - "loss": 0.0117, - "step": 18970 - }, - { - "epoch": 1.73, - "learning_rate": 4e-06, - "loss": 0.0069, - "step": 18980 - }, - { - "epoch": 1.73, - "learning_rate": 3.986301369863014e-06, - "loss": 0.0071, - "step": 18990 - }, - { - "epoch": 1.74, - "learning_rate": 3.972602739726027e-06, - "loss": 0.0099, - "step": 19000 - }, - { - "epoch": 1.74, - "learning_rate": 3.958904109589041e-06, - "loss": 0.0095, - "step": 19010 - }, - { - "epoch": 1.74, - "learning_rate": 3.945205479452055e-06, - "loss": 0.008, - "step": 19020 - }, - { - "epoch": 1.74, - "learning_rate": 3.9315068493150684e-06, - "loss": 0.008, - "step": 19030 - }, - { - "epoch": 1.74, - "learning_rate": 3.917808219178082e-06, - "loss": 0.0143, - "step": 19040 - }, - { - "epoch": 1.74, - "learning_rate": 3.904109589041096e-06, - "loss": 0.0089, - "step": 19050 - }, - { - "epoch": 1.74, - "learning_rate": 3.89041095890411e-06, - "loss": 0.0112, - "step": 19060 - }, - { - "epoch": 1.74, - "learning_rate": 3.876712328767123e-06, - "loss": 0.0083, - "step": 19070 - }, - { - "epoch": 1.74, - "learning_rate": 3.863013698630137e-06, - "loss": 0.0142, - "step": 19080 - }, - { - "epoch": 1.74, - "learning_rate": 3.849315068493151e-06, - "loss": 0.011, - "step": 19090 - }, - { - "epoch": 1.74, - "learning_rate": 3.8356164383561645e-06, - "loss": 0.0089, - "step": 19100 - }, - { - "epoch": 1.75, - "learning_rate": 3.821917808219178e-06, - "loss": 0.0072, - "step": 19110 - }, - { - "epoch": 1.75, - "learning_rate": 3.8082191780821916e-06, - "loss": 0.0076, - "step": 19120 - }, - { - "epoch": 1.75, - "learning_rate": 3.7945205479452057e-06, - "loss": 0.0114, - "step": 19130 - }, - { - "epoch": 1.75, - "learning_rate": 3.7808219178082194e-06, - "loss": 0.0103, - "step": 19140 - }, - { - "epoch": 1.75, - "learning_rate": 3.7671232876712327e-06, - "loss": 0.0122, - "step": 19150 - }, - { - "epoch": 1.75, - "learning_rate": 3.753424657534247e-06, - "loss": 0.0125, - "step": 19160 - }, - { - "epoch": 1.75, - "learning_rate": 3.7397260273972606e-06, - "loss": 0.0086, - "step": 19170 - }, - { - "epoch": 1.75, - "learning_rate": 3.726027397260274e-06, - "loss": 0.0142, - "step": 19180 - }, - { - "epoch": 1.75, - "learning_rate": 3.7123287671232876e-06, - "loss": 0.0283, - "step": 19190 - }, - { - "epoch": 1.75, - "learning_rate": 3.6986301369863014e-06, - "loss": 0.0076, - "step": 19200 - }, - { - "epoch": 1.75, - "learning_rate": 3.684931506849315e-06, - "loss": 0.0098, - "step": 19210 - }, - { - "epoch": 1.76, - "learning_rate": 3.671232876712329e-06, - "loss": 0.0116, - "step": 19220 - }, - { - "epoch": 1.76, - "learning_rate": 3.6575342465753425e-06, - "loss": 0.0109, - "step": 19230 - }, - { - "epoch": 1.76, - "learning_rate": 3.6438356164383563e-06, - "loss": 0.0138, - "step": 19240 - }, - { - "epoch": 1.76, - "learning_rate": 3.6301369863013696e-06, - "loss": 0.0063, - "step": 19250 - }, - { - "epoch": 1.76, - "learning_rate": 3.6164383561643837e-06, - "loss": 0.0103, - "step": 19260 - }, - { - "epoch": 1.76, - "learning_rate": 3.6027397260273974e-06, - "loss": 0.0097, - "step": 19270 - }, - { - "epoch": 1.76, - "learning_rate": 3.589041095890411e-06, - "loss": 0.0087, - "step": 19280 - }, - { - "epoch": 1.76, - "learning_rate": 3.5753424657534245e-06, - "loss": 0.0115, - "step": 19290 - }, - { - "epoch": 1.76, - "learning_rate": 3.561643835616438e-06, - "loss": 0.0099, - "step": 19300 - }, - { - "epoch": 1.76, - "learning_rate": 3.5479452054794523e-06, - "loss": 0.0102, - "step": 19310 - }, - { - "epoch": 1.76, - "learning_rate": 3.534246575342466e-06, - "loss": 0.0146, - "step": 19320 - }, - { - "epoch": 1.77, - "learning_rate": 3.5205479452054794e-06, - "loss": 0.0109, - "step": 19330 - }, - { - "epoch": 1.77, - "learning_rate": 3.506849315068493e-06, - "loss": 0.0083, - "step": 19340 - }, - { - "epoch": 1.77, - "learning_rate": 3.493150684931507e-06, - "loss": 0.0098, - "step": 19350 - }, - { - "epoch": 1.77, - "learning_rate": 3.479452054794521e-06, - "loss": 0.0072, - "step": 19360 - }, - { - "epoch": 1.77, - "learning_rate": 3.4657534246575343e-06, - "loss": 0.0111, - "step": 19370 - }, - { - "epoch": 1.77, - "learning_rate": 3.452054794520548e-06, - "loss": 0.0303, - "step": 19380 - }, - { - "epoch": 1.77, - "learning_rate": 3.4383561643835617e-06, - "loss": 0.0089, - "step": 19390 - }, - { - "epoch": 1.77, - "learning_rate": 3.4246575342465754e-06, - "loss": 0.0107, - "step": 19400 - }, - { - "epoch": 1.77, - "learning_rate": 3.410958904109589e-06, - "loss": 0.0099, - "step": 19410 - }, - { - "epoch": 1.77, - "learning_rate": 3.397260273972603e-06, - "loss": 0.008, - "step": 19420 - }, - { - "epoch": 1.77, - "learning_rate": 3.3835616438356166e-06, - "loss": 0.0095, - "step": 19430 - }, - { - "epoch": 1.78, - "learning_rate": 3.36986301369863e-06, - "loss": 0.0097, - "step": 19440 - }, - { - "epoch": 1.78, - "learning_rate": 3.3561643835616436e-06, - "loss": 0.0105, - "step": 19450 - }, - { - "epoch": 1.78, - "learning_rate": 3.342465753424658e-06, - "loss": 0.0097, - "step": 19460 - }, - { - "epoch": 1.78, - "learning_rate": 3.3287671232876715e-06, - "loss": 0.0104, - "step": 19470 - }, - { - "epoch": 1.78, - "learning_rate": 3.315068493150685e-06, - "loss": 0.0103, - "step": 19480 - }, - { - "epoch": 1.78, - "learning_rate": 3.3013698630136985e-06, - "loss": 0.0135, - "step": 19490 - }, - { - "epoch": 1.78, - "learning_rate": 3.2876712328767123e-06, - "loss": 0.0085, - "step": 19500 - }, - { - "epoch": 1.78, - "learning_rate": 3.2739726027397264e-06, - "loss": 0.0079, - "step": 19510 - }, - { - "epoch": 1.78, - "learning_rate": 3.2602739726027397e-06, - "loss": 0.0079, - "step": 19520 - }, - { - "epoch": 1.78, - "learning_rate": 3.2465753424657534e-06, - "loss": 0.009, - "step": 19530 - }, - { - "epoch": 1.78, - "learning_rate": 3.232876712328767e-06, - "loss": 0.0114, - "step": 19540 - }, - { - "epoch": 1.79, - "learning_rate": 3.219178082191781e-06, - "loss": 0.0106, - "step": 19550 - }, - { - "epoch": 1.79, - "learning_rate": 3.2054794520547946e-06, - "loss": 0.0105, - "step": 19560 - }, - { - "epoch": 1.79, - "learning_rate": 3.1917808219178083e-06, - "loss": 0.0084, - "step": 19570 - }, - { - "epoch": 1.79, - "learning_rate": 3.178082191780822e-06, - "loss": 0.0077, - "step": 19580 - }, - { - "epoch": 1.79, - "learning_rate": 3.164383561643836e-06, - "loss": 0.0106, - "step": 19590 - }, - { - "epoch": 1.79, - "learning_rate": 3.150684931506849e-06, - "loss": 0.0112, - "step": 19600 - }, - { - "epoch": 1.79, - "learning_rate": 3.1369863013698632e-06, - "loss": 0.007, - "step": 19610 - }, - { - "epoch": 1.79, - "learning_rate": 3.123287671232877e-06, - "loss": 0.0084, - "step": 19620 - }, - { - "epoch": 1.79, - "learning_rate": 3.1095890410958903e-06, - "loss": 0.0069, - "step": 19630 - }, - { - "epoch": 1.79, - "learning_rate": 3.095890410958904e-06, - "loss": 0.0098, - "step": 19640 - }, - { - "epoch": 1.79, - "learning_rate": 3.0821917808219177e-06, - "loss": 0.0102, - "step": 19650 - }, - { - "epoch": 1.8, - "learning_rate": 3.068493150684932e-06, - "loss": 0.0105, - "step": 19660 - }, - { - "epoch": 1.8, - "learning_rate": 3.054794520547945e-06, - "loss": 0.0108, - "step": 19670 - }, - { - "epoch": 1.8, - "learning_rate": 3.041095890410959e-06, - "loss": 0.0095, - "step": 19680 - }, - { - "epoch": 1.8, - "learning_rate": 3.0273972602739726e-06, - "loss": 0.0092, - "step": 19690 - }, - { - "epoch": 1.8, - "learning_rate": 3.0136986301369864e-06, - "loss": 0.0107, - "step": 19700 - }, - { - "epoch": 1.8, - "learning_rate": 3e-06, - "loss": 0.0086, - "step": 19710 - }, - { - "epoch": 1.8, - "learning_rate": 2.986301369863014e-06, - "loss": 0.0099, - "step": 19720 - }, - { - "epoch": 1.8, - "learning_rate": 2.9726027397260275e-06, - "loss": 0.0074, - "step": 19730 - }, - { - "epoch": 1.8, - "learning_rate": 2.9589041095890413e-06, - "loss": 0.0085, - "step": 19740 - }, - { - "epoch": 1.8, - "learning_rate": 2.9452054794520546e-06, - "loss": 0.009, - "step": 19750 - }, - { - "epoch": 1.8, - "learning_rate": 2.9315068493150687e-06, - "loss": 0.0084, - "step": 19760 - }, - { - "epoch": 1.81, - "learning_rate": 2.9178082191780824e-06, - "loss": 0.0093, - "step": 19770 - }, - { - "epoch": 1.81, - "learning_rate": 2.904109589041096e-06, - "loss": 0.0065, - "step": 19780 - }, - { - "epoch": 1.81, - "learning_rate": 2.8904109589041095e-06, - "loss": 0.0085, - "step": 19790 - }, - { - "epoch": 1.81, - "learning_rate": 2.876712328767123e-06, - "loss": 0.0103, - "step": 19800 - }, - { - "epoch": 1.81, - "learning_rate": 2.8630136986301373e-06, - "loss": 0.0117, - "step": 19810 - }, - { - "epoch": 1.81, - "learning_rate": 2.849315068493151e-06, - "loss": 0.0122, - "step": 19820 - }, - { - "epoch": 1.81, - "learning_rate": 2.8356164383561644e-06, - "loss": 0.0091, - "step": 19830 - }, - { - "epoch": 1.81, - "learning_rate": 2.821917808219178e-06, - "loss": 0.0087, - "step": 19840 - }, - { - "epoch": 1.81, - "learning_rate": 2.808219178082192e-06, - "loss": 0.0085, - "step": 19850 - }, - { - "epoch": 1.81, - "learning_rate": 2.7945205479452055e-06, - "loss": 0.0091, - "step": 19860 - }, - { - "epoch": 1.81, - "learning_rate": 2.7808219178082193e-06, - "loss": 0.0071, - "step": 19870 - }, - { - "epoch": 1.82, - "learning_rate": 2.767123287671233e-06, - "loss": 0.0086, - "step": 19880 - }, - { - "epoch": 1.82, - "learning_rate": 2.7534246575342467e-06, - "loss": 0.009, - "step": 19890 - }, - { - "epoch": 1.82, - "learning_rate": 2.73972602739726e-06, - "loss": 0.0079, - "step": 19900 - }, - { - "epoch": 1.82, - "learning_rate": 2.726027397260274e-06, - "loss": 0.0101, - "step": 19910 - }, - { - "epoch": 1.82, - "learning_rate": 2.712328767123288e-06, - "loss": 0.0116, - "step": 19920 - }, - { - "epoch": 1.82, - "learning_rate": 2.6986301369863016e-06, - "loss": 0.01, - "step": 19930 - }, - { - "epoch": 1.82, - "learning_rate": 2.684931506849315e-06, - "loss": 0.0116, - "step": 19940 - }, - { - "epoch": 1.82, - "learning_rate": 2.6712328767123286e-06, - "loss": 0.0079, - "step": 19950 - }, - { - "epoch": 1.82, - "learning_rate": 2.6575342465753428e-06, - "loss": 0.009, - "step": 19960 - }, - { - "epoch": 1.82, - "learning_rate": 2.6438356164383565e-06, - "loss": 0.0095, - "step": 19970 - }, - { - "epoch": 1.82, - "learning_rate": 2.63013698630137e-06, - "loss": 0.0101, - "step": 19980 - }, - { - "epoch": 1.83, - "learning_rate": 2.6164383561643835e-06, - "loss": 0.0098, - "step": 19990 - }, - { - "epoch": 1.83, - "learning_rate": 2.6027397260273973e-06, - "loss": 0.0097, - "step": 20000 - }, - { - "epoch": 1.83, - "learning_rate": 2.5890410958904114e-06, - "loss": 0.0088, - "step": 20010 - }, - { - "epoch": 1.83, - "learning_rate": 2.5753424657534247e-06, - "loss": 0.0089, - "step": 20020 - }, - { - "epoch": 1.83, - "learning_rate": 2.5616438356164384e-06, - "loss": 0.012, - "step": 20030 - }, - { - "epoch": 1.83, - "learning_rate": 2.547945205479452e-06, - "loss": 0.0093, - "step": 20040 - }, - { - "epoch": 1.83, - "learning_rate": 2.5342465753424655e-06, - "loss": 0.0373, - "step": 20050 - }, - { - "epoch": 1.83, - "learning_rate": 2.5205479452054796e-06, - "loss": 0.0088, - "step": 20060 - }, - { - "epoch": 1.83, - "learning_rate": 2.5068493150684933e-06, - "loss": 0.0082, - "step": 20070 - }, - { - "epoch": 1.83, - "learning_rate": 2.493150684931507e-06, - "loss": 0.0084, - "step": 20080 - }, - { - "epoch": 1.83, - "learning_rate": 2.4794520547945204e-06, - "loss": 0.0088, - "step": 20090 - }, - { - "epoch": 1.84, - "learning_rate": 2.465753424657534e-06, - "loss": 0.0072, - "step": 20100 - }, - { - "epoch": 1.84, - "learning_rate": 2.4520547945205482e-06, - "loss": 0.0086, - "step": 20110 - }, - { - "epoch": 1.84, - "learning_rate": 2.438356164383562e-06, - "loss": 0.0061, - "step": 20120 - }, - { - "epoch": 1.84, - "learning_rate": 2.4246575342465753e-06, - "loss": 0.0094, - "step": 20130 - }, - { - "epoch": 1.84, - "learning_rate": 2.410958904109589e-06, - "loss": 0.0075, - "step": 20140 - }, - { - "epoch": 1.84, - "learning_rate": 2.3972602739726027e-06, - "loss": 0.0081, - "step": 20150 - }, - { - "epoch": 1.84, - "learning_rate": 2.383561643835617e-06, - "loss": 0.0101, - "step": 20160 - }, - { - "epoch": 1.84, - "learning_rate": 2.36986301369863e-06, - "loss": 0.0105, - "step": 20170 - }, - { - "epoch": 1.84, - "learning_rate": 2.356164383561644e-06, - "loss": 0.0111, - "step": 20180 - }, - { - "epoch": 1.84, - "learning_rate": 2.3424657534246576e-06, - "loss": 0.0086, - "step": 20190 - }, - { - "epoch": 1.84, - "learning_rate": 2.3287671232876713e-06, - "loss": 0.0082, - "step": 20200 - }, - { - "epoch": 1.85, - "learning_rate": 2.315068493150685e-06, - "loss": 0.0105, - "step": 20210 - }, - { - "epoch": 1.85, - "learning_rate": 2.301369863013699e-06, - "loss": 0.0054, - "step": 20220 - }, - { - "epoch": 1.85, - "learning_rate": 2.2876712328767125e-06, - "loss": 0.0112, - "step": 20230 - }, - { - "epoch": 1.85, - "learning_rate": 2.2739726027397262e-06, - "loss": 0.0097, - "step": 20240 - }, - { - "epoch": 1.85, - "learning_rate": 2.2602739726027396e-06, - "loss": 0.0113, - "step": 20250 - }, - { - "epoch": 1.85, - "learning_rate": 2.2465753424657533e-06, - "loss": 0.0086, - "step": 20260 - }, - { - "epoch": 1.85, - "learning_rate": 2.2328767123287674e-06, - "loss": 0.0165, - "step": 20270 - }, - { - "epoch": 1.85, - "learning_rate": 2.2191780821917807e-06, - "loss": 0.0104, - "step": 20280 - }, - { - "epoch": 1.85, - "learning_rate": 2.2054794520547945e-06, - "loss": 0.0094, - "step": 20290 - }, - { - "epoch": 1.85, - "learning_rate": 2.191780821917808e-06, - "loss": 0.008, - "step": 20300 - }, - { - "epoch": 1.85, - "learning_rate": 2.178082191780822e-06, - "loss": 0.0105, - "step": 20310 - }, - { - "epoch": 1.86, - "learning_rate": 2.1643835616438356e-06, - "loss": 0.011, - "step": 20320 - }, - { - "epoch": 1.86, - "learning_rate": 2.1506849315068494e-06, - "loss": 0.0073, - "step": 20330 - }, - { - "epoch": 1.86, - "learning_rate": 2.136986301369863e-06, - "loss": 0.0085, - "step": 20340 - }, - { - "epoch": 1.86, - "learning_rate": 2.123287671232877e-06, - "loss": 0.0063, - "step": 20350 - }, - { - "epoch": 1.86, - "learning_rate": 2.10958904109589e-06, - "loss": 0.0071, - "step": 20360 - }, - { - "epoch": 1.86, - "learning_rate": 2.0958904109589043e-06, - "loss": 0.008, - "step": 20370 - }, - { - "epoch": 1.86, - "learning_rate": 2.082191780821918e-06, - "loss": 0.0101, - "step": 20380 - }, - { - "epoch": 1.86, - "learning_rate": 2.0684931506849317e-06, - "loss": 0.0149, - "step": 20390 - }, - { - "epoch": 1.86, - "learning_rate": 2.054794520547945e-06, - "loss": 0.0098, - "step": 20400 - }, - { - "epoch": 1.86, - "learning_rate": 2.0410958904109587e-06, - "loss": 0.0108, - "step": 20410 - }, - { - "epoch": 1.86, - "learning_rate": 2.027397260273973e-06, - "loss": 0.0078, - "step": 20420 - }, - { - "epoch": 1.87, - "learning_rate": 2.0136986301369866e-06, - "loss": 0.0081, - "step": 20430 - }, - { - "epoch": 1.87, - "learning_rate": 2e-06, - "loss": 0.0174, - "step": 20440 - }, - { - "epoch": 1.87, - "learning_rate": 1.9863013698630136e-06, - "loss": 0.0078, - "step": 20450 - }, - { - "epoch": 1.87, - "learning_rate": 1.9726027397260274e-06, - "loss": 0.0124, - "step": 20460 - }, - { - "epoch": 1.87, - "learning_rate": 1.958904109589041e-06, - "loss": 0.0091, - "step": 20470 - }, - { - "epoch": 1.87, - "learning_rate": 1.945205479452055e-06, - "loss": 0.0103, - "step": 20480 - }, - { - "epoch": 1.87, - "learning_rate": 1.9315068493150685e-06, - "loss": 0.0076, - "step": 20490 - }, - { - "epoch": 1.87, - "learning_rate": 1.9178082191780823e-06, - "loss": 0.0134, - "step": 20500 - }, - { - "epoch": 1.87, - "learning_rate": 1.9041095890410958e-06, - "loss": 0.0082, - "step": 20510 - }, - { - "epoch": 1.87, - "learning_rate": 1.8904109589041097e-06, - "loss": 0.0125, - "step": 20520 - }, - { - "epoch": 1.87, - "learning_rate": 1.8767123287671234e-06, - "loss": 0.0078, - "step": 20530 - }, - { - "epoch": 1.88, - "learning_rate": 1.863013698630137e-06, - "loss": 0.0101, - "step": 20540 - }, - { - "epoch": 1.88, - "learning_rate": 1.8493150684931507e-06, - "loss": 0.0102, - "step": 20550 - }, - { - "epoch": 1.88, - "learning_rate": 1.8356164383561644e-06, - "loss": 0.0075, - "step": 20560 - }, - { - "epoch": 1.88, - "learning_rate": 1.8219178082191781e-06, - "loss": 0.0096, - "step": 20570 - }, - { - "epoch": 1.88, - "learning_rate": 1.8082191780821919e-06, - "loss": 0.0073, - "step": 20580 - }, - { - "epoch": 1.88, - "learning_rate": 1.7945205479452056e-06, - "loss": 0.0074, - "step": 20590 - }, - { - "epoch": 1.88, - "learning_rate": 1.780821917808219e-06, - "loss": 0.0104, - "step": 20600 - }, - { - "epoch": 1.88, - "learning_rate": 1.767123287671233e-06, - "loss": 0.008, - "step": 20610 - }, - { - "epoch": 1.88, - "learning_rate": 1.7534246575342465e-06, - "loss": 0.0093, - "step": 20620 - }, - { - "epoch": 1.88, - "learning_rate": 1.7397260273972605e-06, - "loss": 0.0096, - "step": 20630 - }, - { - "epoch": 1.88, - "learning_rate": 1.726027397260274e-06, - "loss": 0.0133, - "step": 20640 - }, - { - "epoch": 1.89, - "learning_rate": 1.7123287671232877e-06, - "loss": 0.0105, - "step": 20650 - }, - { - "epoch": 1.89, - "learning_rate": 1.6986301369863014e-06, - "loss": 0.0077, - "step": 20660 - }, - { - "epoch": 1.89, - "learning_rate": 1.684931506849315e-06, - "loss": 0.0098, - "step": 20670 - }, - { - "epoch": 1.89, - "learning_rate": 1.671232876712329e-06, - "loss": 0.0119, - "step": 20680 - }, - { - "epoch": 1.89, - "learning_rate": 1.6575342465753424e-06, - "loss": 0.0096, - "step": 20690 - }, - { - "epoch": 1.89, - "learning_rate": 1.6438356164383561e-06, - "loss": 0.0095, - "step": 20700 - }, - { - "epoch": 1.89, - "learning_rate": 1.6301369863013699e-06, - "loss": 0.0101, - "step": 20710 - }, - { - "epoch": 1.89, - "learning_rate": 1.6164383561643836e-06, - "loss": 0.01, - "step": 20720 - }, - { - "epoch": 1.89, - "learning_rate": 1.6027397260273973e-06, - "loss": 0.01, - "step": 20730 - }, - { - "epoch": 1.89, - "learning_rate": 1.589041095890411e-06, - "loss": 0.0072, - "step": 20740 - }, - { - "epoch": 1.89, - "learning_rate": 1.5753424657534245e-06, - "loss": 0.0144, - "step": 20750 - }, - { - "epoch": 1.9, - "learning_rate": 1.5616438356164385e-06, - "loss": 0.0093, - "step": 20760 - }, - { - "epoch": 1.9, - "learning_rate": 1.547945205479452e-06, - "loss": 0.0124, - "step": 20770 - }, - { - "epoch": 1.9, - "learning_rate": 1.534246575342466e-06, - "loss": 0.0118, - "step": 20780 - }, - { - "epoch": 1.9, - "learning_rate": 1.5205479452054794e-06, - "loss": 0.0111, - "step": 20790 - }, - { - "epoch": 1.9, - "learning_rate": 1.5068493150684932e-06, - "loss": 0.0069, - "step": 20800 - }, - { - "epoch": 1.9, - "learning_rate": 1.493150684931507e-06, - "loss": 0.0074, - "step": 20810 - }, - { - "epoch": 1.9, - "learning_rate": 1.4794520547945206e-06, - "loss": 0.032, - "step": 20820 - }, - { - "epoch": 1.9, - "learning_rate": 1.4657534246575344e-06, - "loss": 0.0077, - "step": 20830 - }, - { - "epoch": 1.9, - "learning_rate": 1.452054794520548e-06, - "loss": 0.0084, - "step": 20840 - }, - { - "epoch": 1.9, - "learning_rate": 1.4383561643835616e-06, - "loss": 0.0131, - "step": 20850 - }, - { - "epoch": 1.91, - "learning_rate": 1.4246575342465755e-06, - "loss": 0.0225, - "step": 20860 - }, - { - "epoch": 1.91, - "learning_rate": 1.410958904109589e-06, - "loss": 0.0125, - "step": 20870 - }, - { - "epoch": 1.91, - "learning_rate": 1.3972602739726028e-06, - "loss": 0.0063, - "step": 20880 - }, - { - "epoch": 1.91, - "learning_rate": 1.3835616438356165e-06, - "loss": 0.0457, - "step": 20890 - }, - { - "epoch": 1.91, - "learning_rate": 1.36986301369863e-06, - "loss": 0.0098, - "step": 20900 - }, - { - "epoch": 1.91, - "learning_rate": 1.356164383561644e-06, - "loss": 0.0082, - "step": 20910 - }, - { - "epoch": 1.91, - "learning_rate": 1.3424657534246575e-06, - "loss": 0.0431, - "step": 20920 - }, - { - "epoch": 1.91, - "learning_rate": 1.3287671232876714e-06, - "loss": 0.0072, - "step": 20930 - }, - { - "epoch": 1.91, - "learning_rate": 1.315068493150685e-06, - "loss": 0.0081, - "step": 20940 - }, - { - "epoch": 1.91, - "learning_rate": 1.3013698630136986e-06, - "loss": 0.0129, - "step": 20950 - }, - { - "epoch": 1.91, - "learning_rate": 1.2876712328767124e-06, - "loss": 0.0078, - "step": 20960 - }, - { - "epoch": 1.92, - "learning_rate": 1.273972602739726e-06, - "loss": 0.0101, - "step": 20970 - }, - { - "epoch": 1.92, - "learning_rate": 1.2602739726027398e-06, - "loss": 0.0123, - "step": 20980 - }, - { - "epoch": 1.92, - "learning_rate": 1.2465753424657535e-06, - "loss": 0.0074, - "step": 20990 - }, - { - "epoch": 1.92, - "learning_rate": 1.232876712328767e-06, - "loss": 0.0115, - "step": 21000 - }, - { - "epoch": 1.92, - "learning_rate": 1.219178082191781e-06, - "loss": 0.0066, - "step": 21010 - }, - { - "epoch": 1.92, - "learning_rate": 1.2054794520547945e-06, - "loss": 0.0111, - "step": 21020 - }, - { - "epoch": 1.92, - "learning_rate": 1.1917808219178084e-06, - "loss": 0.01, - "step": 21030 - }, - { - "epoch": 1.92, - "learning_rate": 1.178082191780822e-06, - "loss": 0.0096, - "step": 21040 - }, - { - "epoch": 1.92, - "learning_rate": 1.1643835616438357e-06, - "loss": 0.0095, - "step": 21050 - }, - { - "epoch": 1.92, - "learning_rate": 1.1506849315068494e-06, - "loss": 0.0078, - "step": 21060 - }, - { - "epoch": 1.92, - "learning_rate": 1.1369863013698631e-06, - "loss": 0.0217, - "step": 21070 - }, - { - "epoch": 1.93, - "learning_rate": 1.1232876712328766e-06, - "loss": 0.0111, - "step": 21080 - }, - { - "epoch": 1.93, - "learning_rate": 1.1095890410958904e-06, - "loss": 0.0087, - "step": 21090 - }, - { - "epoch": 1.93, - "learning_rate": 1.095890410958904e-06, - "loss": 0.0068, - "step": 21100 - }, - { - "epoch": 1.93, - "learning_rate": 1.0821917808219178e-06, - "loss": 0.0162, - "step": 21110 - }, - { - "epoch": 1.93, - "learning_rate": 1.0684931506849315e-06, - "loss": 0.0101, - "step": 21120 - }, - { - "epoch": 1.93, - "learning_rate": 1.054794520547945e-06, - "loss": 0.0082, - "step": 21130 - }, - { - "epoch": 1.93, - "learning_rate": 1.041095890410959e-06, - "loss": 0.0079, - "step": 21140 - }, - { - "epoch": 1.93, - "learning_rate": 1.0273972602739725e-06, - "loss": 0.0129, - "step": 21150 - }, - { - "epoch": 1.93, - "learning_rate": 1.0136986301369864e-06, - "loss": 0.0105, - "step": 21160 - }, - { - "epoch": 1.93, - "learning_rate": 1e-06, - "loss": 0.0092, - "step": 21170 - }, - { - "epoch": 1.93, - "learning_rate": 9.863013698630137e-07, - "loss": 0.0097, - "step": 21180 - }, - { - "epoch": 1.94, - "learning_rate": 9.726027397260274e-07, - "loss": 0.0091, - "step": 21190 - }, - { - "epoch": 1.94, - "learning_rate": 9.589041095890411e-07, - "loss": 0.0127, - "step": 21200 - }, - { - "epoch": 1.94, - "learning_rate": 9.452054794520549e-07, - "loss": 0.0135, - "step": 21210 - }, - { - "epoch": 1.94, - "learning_rate": 9.315068493150685e-07, - "loss": 0.006, - "step": 21220 - }, - { - "epoch": 1.94, - "learning_rate": 9.178082191780822e-07, - "loss": 0.0101, - "step": 21230 + "epoch": 1.3, + "learning_rate": 1.0547945205479452e-05, + "loss": 0.0761, + "step": 7100 }, { - "epoch": 1.94, - "learning_rate": 9.041095890410959e-07, - "loss": 0.0059, - "step": 21240 + "epoch": 1.32, + "learning_rate": 1.0273972602739726e-05, + "loss": 0.0739, + "step": 7200 }, { - "epoch": 1.94, - "learning_rate": 8.904109589041095e-07, - "loss": 0.0091, - "step": 21250 + "epoch": 1.33, + "learning_rate": 9.999999999999999e-06, + "loss": 0.0763, + "step": 7300 }, { - "epoch": 1.94, - "learning_rate": 8.767123287671233e-07, - "loss": 0.015, - "step": 21260 + "epoch": 1.35, + "learning_rate": 9.726027397260274e-06, + "loss": 0.0796, + "step": 7400 }, { - "epoch": 1.94, - "learning_rate": 8.63013698630137e-07, - "loss": 0.0068, - "step": 21270 + "epoch": 1.37, + "learning_rate": 9.452054794520548e-06, + "loss": 0.08, + "step": 7500 }, { - "epoch": 1.94, - "learning_rate": 8.493150684931507e-07, - "loss": 0.0055, - "step": 21280 + "epoch": 1.39, + "learning_rate": 9.178082191780821e-06, + "loss": 0.0768, + "step": 7600 }, { - "epoch": 1.94, - "learning_rate": 8.356164383561644e-07, - "loss": 0.0102, - "step": 21290 + "epoch": 1.41, + "learning_rate": 8.904109589041095e-06, + "loss": 0.0766, + "step": 7700 }, { - "epoch": 1.95, - "learning_rate": 8.219178082191781e-07, - "loss": 0.009, - "step": 21300 + "epoch": 1.42, + "learning_rate": 8.63013698630137e-06, + "loss": 0.0739, + "step": 7800 }, { - "epoch": 1.95, - "learning_rate": 8.082191780821918e-07, - "loss": 0.0106, - "step": 21310 + "epoch": 1.44, + "learning_rate": 8.356164383561643e-06, + "loss": 0.0718, + "step": 7900 }, { - "epoch": 1.95, - "learning_rate": 7.945205479452055e-07, - "loss": 0.0092, - "step": 21320 + "epoch": 1.46, + "learning_rate": 8.082191780821917e-06, + "loss": 0.0737, + "step": 8000 }, { - "epoch": 1.95, - "learning_rate": 7.808219178082192e-07, - "loss": 0.009, - "step": 21330 + "epoch": 1.48, + "learning_rate": 7.808219178082192e-06, + "loss": 0.0703, + "step": 8100 }, { - "epoch": 1.95, - "learning_rate": 7.67123287671233e-07, - "loss": 0.0098, - "step": 21340 + "epoch": 1.5, + "learning_rate": 7.5342465753424655e-06, + "loss": 0.0723, + "step": 8200 }, { - "epoch": 1.95, - "learning_rate": 7.534246575342466e-07, - "loss": 0.0282, - "step": 21350 + "epoch": 1.52, + "learning_rate": 7.260273972602739e-06, + "loss": 0.0747, + "step": 8300 }, { - "epoch": 1.95, - "learning_rate": 7.397260273972603e-07, - "loss": 0.0064, - "step": 21360 + "epoch": 1.53, + "learning_rate": 6.986301369863014e-06, + "loss": 0.0787, + "step": 8400 }, { - "epoch": 1.95, - "learning_rate": 7.26027397260274e-07, - "loss": 0.0096, - "step": 21370 + "epoch": 1.55, + "learning_rate": 6.712328767123287e-06, + "loss": 0.0704, + "step": 8500 }, { - "epoch": 1.95, - "learning_rate": 7.123287671232878e-07, - "loss": 0.0105, - "step": 21380 + "epoch": 1.57, + "learning_rate": 6.438356164383562e-06, + "loss": 0.0703, + "step": 8600 }, { - "epoch": 1.95, - "learning_rate": 6.986301369863014e-07, - "loss": 0.0087, - "step": 21390 + "epoch": 1.59, + "learning_rate": 6.1643835616438354e-06, + "loss": 0.0688, + "step": 8700 }, { - "epoch": 1.95, - "learning_rate": 6.84931506849315e-07, - "loss": 0.0097, - "step": 21400 + "epoch": 1.61, + "learning_rate": 5.890410958904109e-06, + "loss": 0.0762, + "step": 8800 }, { - "epoch": 1.96, - "learning_rate": 6.712328767123287e-07, - "loss": 0.0064, - "step": 21410 + "epoch": 1.63, + "learning_rate": 5.616438356164384e-06, + "loss": 0.0723, + "step": 8900 }, { - "epoch": 1.96, - "learning_rate": 6.575342465753425e-07, - "loss": 0.008, - "step": 21420 + "epoch": 1.64, + "learning_rate": 5.342465753424657e-06, + "loss": 0.073, + "step": 9000 }, { - "epoch": 1.96, - "learning_rate": 6.438356164383562e-07, - "loss": 0.0094, - "step": 21430 + "epoch": 1.66, + "learning_rate": 5.068493150684931e-06, + "loss": 0.0725, + "step": 9100 }, { - "epoch": 1.96, - "learning_rate": 6.301369863013699e-07, - "loss": 0.0081, - "step": 21440 + "epoch": 1.68, + "learning_rate": 4.7945205479452054e-06, + "loss": 0.0719, + "step": 9200 }, { - "epoch": 1.96, - "learning_rate": 6.164383561643835e-07, - "loss": 0.0086, - "step": 21450 + "epoch": 1.7, + "learning_rate": 4.520547945205479e-06, + "loss": 0.0674, + "step": 9300 }, { - "epoch": 1.96, - "learning_rate": 6.027397260273972e-07, - "loss": 0.0068, - "step": 21460 + "epoch": 1.72, + "learning_rate": 4.246575342465754e-06, + "loss": 0.0667, + "step": 9400 }, { - "epoch": 1.96, - "learning_rate": 5.89041095890411e-07, - "loss": 0.0108, - "step": 21470 + "epoch": 1.74, + "learning_rate": 3.972602739726027e-06, + "loss": 0.0722, + "step": 9500 }, { - "epoch": 1.96, - "learning_rate": 5.753424657534247e-07, - "loss": 0.0081, - "step": 21480 + "epoch": 1.75, + "learning_rate": 3.6986301369863014e-06, + "loss": 0.071, + "step": 9600 }, { - "epoch": 1.96, - "learning_rate": 5.616438356164383e-07, - "loss": 0.0075, - "step": 21490 + "epoch": 1.77, + "learning_rate": 3.4246575342465754e-06, + "loss": 0.074, + "step": 9700 }, { - "epoch": 1.96, - "learning_rate": 5.47945205479452e-07, - "loss": 0.0078, - "step": 21500 + "epoch": 1.79, + "learning_rate": 3.150684931506849e-06, + "loss": 0.0696, + "step": 9800 }, { - "epoch": 1.96, - "learning_rate": 5.342465753424658e-07, - "loss": 0.0126, - "step": 21510 + "epoch": 1.81, + "learning_rate": 2.876712328767123e-06, + "loss": 0.0699, + "step": 9900 }, { - "epoch": 1.97, - "learning_rate": 5.205479452054795e-07, - "loss": 0.0089, - "step": 21520 + "epoch": 1.83, + "learning_rate": 2.6027397260273973e-06, + "loss": 0.0734, + "step": 10000 }, { - "epoch": 1.97, - "learning_rate": 5.068493150684932e-07, - "loss": 0.0089, - "step": 21530 + "epoch": 1.84, + "learning_rate": 2.3287671232876713e-06, + "loss": 0.0687, + "step": 10100 }, { - "epoch": 1.97, - "learning_rate": 4.931506849315068e-07, - "loss": 0.0097, - "step": 21540 + "epoch": 1.86, + "learning_rate": 2.054794520547945e-06, + "loss": 0.0729, + "step": 10200 }, { - "epoch": 1.97, - "learning_rate": 4.794520547945206e-07, - "loss": 0.0123, - "step": 21550 + "epoch": 1.88, + "learning_rate": 1.780821917808219e-06, + "loss": 0.0703, + "step": 10300 }, { - "epoch": 1.97, - "learning_rate": 4.6575342465753424e-07, - "loss": 0.0123, - "step": 21560 + "epoch": 1.9, + "learning_rate": 1.5068493150684932e-06, + "loss": 0.0734, + "step": 10400 }, { - "epoch": 1.97, - "learning_rate": 4.5205479452054796e-07, - "loss": 0.0079, - "step": 21570 + "epoch": 1.92, + "learning_rate": 1.232876712328767e-06, + "loss": 0.0728, + "step": 10500 }, { - "epoch": 1.97, - "learning_rate": 4.3835616438356164e-07, - "loss": 0.0096, - "step": 21580 + "epoch": 1.94, + "learning_rate": 9.589041095890411e-07, + "loss": 0.073, + "step": 10600 }, { - "epoch": 1.97, - "learning_rate": 4.2465753424657536e-07, - "loss": 0.009, - "step": 21590 + "epoch": 1.95, + "learning_rate": 6.84931506849315e-07, + "loss": 0.0694, + "step": 10700 }, { "epoch": 1.97, "learning_rate": 4.1095890410958903e-07, - "loss": 0.009, - "step": 21600 - }, - { - "epoch": 1.97, - "learning_rate": 3.9726027397260276e-07, - "loss": 0.0094, - "step": 21610 - }, - { - "epoch": 1.97, - "learning_rate": 3.835616438356165e-07, - "loss": 0.0106, - "step": 21620 - }, - { - "epoch": 1.98, - "learning_rate": 3.6986301369863016e-07, - "loss": 0.0104, - "step": 21630 - }, - { - "epoch": 1.98, - "learning_rate": 3.561643835616439e-07, - "loss": 0.0103, - "step": 21640 - }, - { - "epoch": 1.98, - "learning_rate": 3.424657534246575e-07, - "loss": 0.0075, - "step": 21650 - }, - { - "epoch": 1.98, - "learning_rate": 3.2876712328767123e-07, - "loss": 0.0082, - "step": 21660 - }, - { - "epoch": 1.98, - "learning_rate": 3.1506849315068495e-07, - "loss": 0.01, - "step": 21670 - }, - { - "epoch": 1.98, - "learning_rate": 3.013698630136986e-07, - "loss": 0.0103, - "step": 21680 - }, - { - "epoch": 1.98, - "learning_rate": 2.8767123287671235e-07, - "loss": 0.0108, - "step": 21690 - }, - { - "epoch": 1.98, - "learning_rate": 2.73972602739726e-07, - "loss": 0.0456, - "step": 21700 - }, - { - "epoch": 1.98, - "learning_rate": 2.6027397260273975e-07, - "loss": 0.0095, - "step": 21710 - }, - { - "epoch": 1.98, - "learning_rate": 2.465753424657534e-07, - "loss": 0.0128, - "step": 21720 - }, - { - "epoch": 1.98, - "learning_rate": 2.3287671232876712e-07, - "loss": 0.0059, - "step": 21730 - }, - { - "epoch": 1.99, - "learning_rate": 2.1917808219178082e-07, - "loss": 0.0128, - "step": 21740 - }, - { - "epoch": 1.99, - "learning_rate": 2.0547945205479452e-07, - "loss": 0.0078, - "step": 21750 - }, - { - "epoch": 1.99, - "learning_rate": 1.9178082191780824e-07, - "loss": 0.0097, - "step": 21760 - }, - { - "epoch": 1.99, - "learning_rate": 1.7808219178082194e-07, - "loss": 0.0063, - "step": 21770 - }, - { - "epoch": 1.99, - "learning_rate": 1.6438356164383561e-07, - "loss": 0.0079, - "step": 21780 - }, - { - "epoch": 1.99, - "learning_rate": 1.506849315068493e-07, - "loss": 0.0076, - "step": 21790 + "loss": 0.0675, + "step": 10800 }, { "epoch": 1.99, "learning_rate": 1.36986301369863e-07, - "loss": 0.0083, - "step": 21800 - }, - { - "epoch": 1.99, - "learning_rate": 1.232876712328767e-07, - "loss": 0.0095, - "step": 21810 - }, - { - "epoch": 1.99, - "learning_rate": 1.0958904109589041e-07, - "loss": 0.007, - "step": 21820 - }, - { - "epoch": 1.99, - "learning_rate": 9.589041095890412e-08, - "loss": 0.0107, - "step": 21830 - }, - { - "epoch": 1.99, - "learning_rate": 8.219178082191781e-08, - "loss": 0.0094, - "step": 21840 - }, - { - "epoch": 2.0, - "learning_rate": 6.84931506849315e-08, - "loss": 0.0136, - "step": 21850 - }, - { - "epoch": 2.0, - "learning_rate": 5.4794520547945204e-08, - "loss": 0.0099, - "step": 21860 - }, - { - "epoch": 2.0, - "learning_rate": 4.1095890410958903e-08, - "loss": 0.0117, - "step": 21870 - }, - { - "epoch": 2.0, - "learning_rate": 2.7397260273972602e-08, - "loss": 0.0086, - "step": 21880 - }, - { - "epoch": 2.0, - "learning_rate": 1.3698630136986301e-08, - "loss": 0.0069, - "step": 21890 - }, - { - "epoch": 2.0, - "learning_rate": 0.0, - "loss": 0.0107, - "step": 21900 + "loss": 0.0709, + "step": 10900 }, { "epoch": 2.0, - "eval_exact_match": 61.45695364238411, - "eval_f1": 77.61128056734677, - "step": 21900 + "eval_exact_match": 67.9848628192999, + "eval_f1": 79.76062168781495, + "step": 10950 }, { "epoch": 2.0, - "step": 21900, + "step": 10950, "total_flos": 5.341230400536576e+16, - "train_loss": 0.01533625880400884, - "train_runtime": 28810.6986, - "train_samples_per_second": 6.081, - "train_steps_per_second": 0.76 + "train_loss": 0.09864264518702956, + "train_runtime": 9855.5808, + "train_samples_per_second": 17.777, + "train_steps_per_second": 1.111 } ], - "max_steps": 21900, + "max_steps": 10950, "num_train_epochs": 2, "total_flos": 5.341230400536576e+16, "trial_name": null,