{ "best_metric": null, "best_model_checkpoint": null, "epoch": 2.4615384615384617, "eval_steps": 500, "global_step": 8000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 4.999987022227664e-05, "loss": 2.3071, "step": 10 }, { "epoch": 0.01, "learning_rate": 4.9999480890453916e-05, "loss": 2.3499, "step": 20 }, { "epoch": 0.01, "learning_rate": 4.999883200857397e-05, "loss": 2.3748, "step": 30 }, { "epoch": 0.01, "learning_rate": 4.999792358337363e-05, "loss": 2.3309, "step": 40 }, { "epoch": 0.02, "learning_rate": 4.999675562428437e-05, "loss": 2.2385, "step": 50 }, { "epoch": 0.02, "learning_rate": 4.999532814343219e-05, "loss": 2.2679, "step": 60 }, { "epoch": 0.02, "learning_rate": 4.999364115563751e-05, "loss": 2.1055, "step": 70 }, { "epoch": 0.02, "learning_rate": 4.9991694678415e-05, "loss": 2.1972, "step": 80 }, { "epoch": 0.03, "learning_rate": 4.998948873197342e-05, "loss": 2.1413, "step": 90 }, { "epoch": 0.03, "learning_rate": 4.9987023339215374e-05, "loss": 2.0432, "step": 100 }, { "epoch": 0.03, "learning_rate": 4.998429852573712e-05, "loss": 2.1598, "step": 110 }, { "epoch": 0.04, "learning_rate": 4.998131431982826e-05, "loss": 2.1133, "step": 120 }, { "epoch": 0.04, "learning_rate": 4.997807075247146e-05, "loss": 2.1489, "step": 130 }, { "epoch": 0.04, "learning_rate": 4.9974567857342155e-05, "loss": 2.1323, "step": 140 }, { "epoch": 0.05, "learning_rate": 4.997080567080817e-05, "loss": 2.206, "step": 150 }, { "epoch": 0.05, "learning_rate": 4.996678423192933e-05, "loss": 2.1427, "step": 160 }, { "epoch": 0.05, "learning_rate": 4.99625035824571e-05, "loss": 2.019, "step": 170 }, { "epoch": 0.06, "learning_rate": 4.99579637668341e-05, "loss": 2.0283, "step": 180 }, { "epoch": 0.06, "learning_rate": 4.995316483219372e-05, "loss": 2.1083, "step": 190 }, { "epoch": 0.06, "learning_rate": 4.994810682835951e-05, "loss": 2.0178, "step": 200 }, { "epoch": 0.06, "learning_rate": 4.994278980784478e-05, "loss": 2.1224, "step": 210 }, { "epoch": 0.07, "learning_rate": 4.993721382585199e-05, "loss": 2.035, "step": 220 }, { "epoch": 0.07, "learning_rate": 4.9931378940272214e-05, "loss": 2.0377, "step": 230 }, { "epoch": 0.07, "learning_rate": 4.992528521168449e-05, "loss": 2.0102, "step": 240 }, { "epoch": 0.08, "learning_rate": 4.9918932703355256e-05, "loss": 2.0017, "step": 250 }, { "epoch": 0.08, "learning_rate": 4.991232148123761e-05, "loss": 2.0592, "step": 260 }, { "epoch": 0.08, "learning_rate": 4.9905451613970725e-05, "loss": 2.0303, "step": 270 }, { "epoch": 0.09, "learning_rate": 4.989832317287904e-05, "loss": 2.0704, "step": 280 }, { "epoch": 0.09, "learning_rate": 4.98909362319716e-05, "loss": 2.02, "step": 290 }, { "epoch": 0.09, "learning_rate": 4.988329086794122e-05, "loss": 2.0254, "step": 300 }, { "epoch": 0.1, "learning_rate": 4.9875387160163744e-05, "loss": 1.9905, "step": 310 }, { "epoch": 0.1, "learning_rate": 4.986722519069719e-05, "loss": 2.0126, "step": 320 }, { "epoch": 0.1, "learning_rate": 4.9858805044280895e-05, "loss": 1.997, "step": 330 }, { "epoch": 0.1, "learning_rate": 4.9850126808334665e-05, "loss": 1.9879, "step": 340 }, { "epoch": 0.11, "learning_rate": 4.984119057295783e-05, "loss": 1.9946, "step": 350 }, { "epoch": 0.11, "learning_rate": 4.9831996430928326e-05, "loss": 2.0136, "step": 360 }, { "epoch": 0.11, "learning_rate": 4.982254447770175e-05, "loss": 1.9844, "step": 370 }, { "epoch": 0.12, "learning_rate": 4.981283481141034e-05, "loss": 1.9043, "step": 380 }, { "epoch": 0.12, "learning_rate": 4.980286753286195e-05, "loss": 2.0086, "step": 390 }, { "epoch": 0.12, "learning_rate": 4.979264274553905e-05, "loss": 2.0385, "step": 400 }, { "epoch": 0.13, "learning_rate": 4.978216055559761e-05, "loss": 1.9869, "step": 410 }, { "epoch": 0.13, "learning_rate": 4.977142107186602e-05, "loss": 2.0016, "step": 420 }, { "epoch": 0.13, "learning_rate": 4.976042440584392e-05, "loss": 1.9087, "step": 430 }, { "epoch": 0.14, "learning_rate": 4.97491706717011e-05, "loss": 1.9727, "step": 440 }, { "epoch": 0.14, "learning_rate": 4.973765998627628e-05, "loss": 1.9395, "step": 450 }, { "epoch": 0.14, "learning_rate": 4.9725892469075905e-05, "loss": 1.9599, "step": 460 }, { "epoch": 0.14, "learning_rate": 4.97138682422729e-05, "loss": 1.9838, "step": 470 }, { "epoch": 0.15, "learning_rate": 4.9701587430705415e-05, "loss": 1.9527, "step": 480 }, { "epoch": 0.15, "learning_rate": 4.9689050161875506e-05, "loss": 1.9233, "step": 490 }, { "epoch": 0.15, "learning_rate": 4.967625656594782e-05, "loss": 1.92, "step": 500 }, { "epoch": 0.16, "learning_rate": 4.966320677574827e-05, "loss": 1.9433, "step": 510 }, { "epoch": 0.16, "learning_rate": 4.964990092676263e-05, "loss": 1.9149, "step": 520 }, { "epoch": 0.16, "learning_rate": 4.963633915713509e-05, "loss": 1.9289, "step": 530 }, { "epoch": 0.17, "learning_rate": 4.962252160766693e-05, "loss": 1.9178, "step": 540 }, { "epoch": 0.17, "learning_rate": 4.960844842181494e-05, "loss": 2.0173, "step": 550 }, { "epoch": 0.17, "learning_rate": 4.9594119745690014e-05, "loss": 1.9719, "step": 560 }, { "epoch": 0.18, "learning_rate": 4.957953572805558e-05, "loss": 2.0085, "step": 570 }, { "epoch": 0.18, "learning_rate": 4.956469652032609e-05, "loss": 2.0053, "step": 580 }, { "epoch": 0.18, "learning_rate": 4.9549602276565435e-05, "loss": 1.921, "step": 590 }, { "epoch": 0.18, "learning_rate": 4.953425315348534e-05, "loss": 1.9189, "step": 600 }, { "epoch": 0.19, "learning_rate": 4.951864931044374e-05, "loss": 1.8703, "step": 610 }, { "epoch": 0.19, "learning_rate": 4.950279090944313e-05, "loss": 1.9624, "step": 620 }, { "epoch": 0.19, "learning_rate": 4.94866781151289e-05, "loss": 1.9079, "step": 630 }, { "epoch": 0.2, "learning_rate": 4.947031109478758e-05, "loss": 1.9622, "step": 640 }, { "epoch": 0.2, "learning_rate": 4.9453690018345144e-05, "loss": 1.9186, "step": 650 }, { "epoch": 0.2, "learning_rate": 4.943681505836523e-05, "loss": 2.0174, "step": 660 }, { "epoch": 0.21, "learning_rate": 4.9419686390047334e-05, "loss": 1.9147, "step": 670 }, { "epoch": 0.21, "learning_rate": 4.9402304191225044e-05, "loss": 1.8727, "step": 680 }, { "epoch": 0.21, "learning_rate": 4.9384668642364126e-05, "loss": 1.8022, "step": 690 }, { "epoch": 0.22, "learning_rate": 4.93667799265607e-05, "loss": 1.9235, "step": 700 }, { "epoch": 0.22, "learning_rate": 4.934863822953929e-05, "loss": 1.8645, "step": 710 }, { "epoch": 0.22, "learning_rate": 4.9330243739650964e-05, "loss": 1.9773, "step": 720 }, { "epoch": 0.22, "learning_rate": 4.9311596647871317e-05, "loss": 1.885, "step": 730 }, { "epoch": 0.23, "learning_rate": 4.929269714779852e-05, "loss": 1.9021, "step": 740 }, { "epoch": 0.23, "learning_rate": 4.92735454356513e-05, "loss": 1.8347, "step": 750 }, { "epoch": 0.23, "learning_rate": 4.925414171026691e-05, "loss": 1.9182, "step": 760 }, { "epoch": 0.24, "learning_rate": 4.923448617309905e-05, "loss": 1.8929, "step": 770 }, { "epoch": 0.24, "learning_rate": 4.9214579028215776e-05, "loss": 1.8939, "step": 780 }, { "epoch": 0.24, "learning_rate": 4.919442048229743e-05, "loss": 1.9492, "step": 790 }, { "epoch": 0.25, "learning_rate": 4.917401074463441e-05, "loss": 1.933, "step": 800 }, { "epoch": 0.25, "learning_rate": 4.9153350027125064e-05, "loss": 1.9427, "step": 810 }, { "epoch": 0.25, "learning_rate": 4.913243854427346e-05, "loss": 1.9193, "step": 820 }, { "epoch": 0.26, "learning_rate": 4.911127651318717e-05, "loss": 1.8606, "step": 830 }, { "epoch": 0.26, "learning_rate": 4.9089864153575016e-05, "loss": 1.8983, "step": 840 }, { "epoch": 0.26, "learning_rate": 4.906820168774477e-05, "loss": 1.9328, "step": 850 }, { "epoch": 0.26, "learning_rate": 4.904628934060088e-05, "loss": 1.9659, "step": 860 }, { "epoch": 0.27, "learning_rate": 4.902412733964211e-05, "loss": 1.8672, "step": 870 }, { "epoch": 0.27, "learning_rate": 4.900171591495918e-05, "loss": 1.982, "step": 880 }, { "epoch": 0.27, "learning_rate": 4.8979055299232376e-05, "loss": 1.9288, "step": 890 }, { "epoch": 0.28, "learning_rate": 4.8956145727729156e-05, "loss": 1.8972, "step": 900 }, { "epoch": 0.28, "learning_rate": 4.893298743830168e-05, "loss": 1.9329, "step": 910 }, { "epoch": 0.28, "learning_rate": 4.890958067138436e-05, "loss": 1.9741, "step": 920 }, { "epoch": 0.29, "learning_rate": 4.8885925669991346e-05, "loss": 1.9489, "step": 930 }, { "epoch": 0.29, "learning_rate": 4.886202267971401e-05, "loss": 1.8751, "step": 940 }, { "epoch": 0.29, "learning_rate": 4.883787194871841e-05, "loss": 1.8534, "step": 950 }, { "epoch": 0.3, "learning_rate": 4.88134737277427e-05, "loss": 1.909, "step": 960 }, { "epoch": 0.3, "learning_rate": 4.878882827009452e-05, "loss": 2.0065, "step": 970 }, { "epoch": 0.3, "learning_rate": 4.8763935831648374e-05, "loss": 1.9246, "step": 980 }, { "epoch": 0.3, "learning_rate": 4.8738796670843004e-05, "loss": 1.9432, "step": 990 }, { "epoch": 0.31, "learning_rate": 4.8713411048678635e-05, "loss": 1.8779, "step": 1000 }, { "epoch": 0.31, "learning_rate": 4.868777922871434e-05, "loss": 1.8934, "step": 1010 }, { "epoch": 0.31, "learning_rate": 4.8661901477065244e-05, "loss": 1.971, "step": 1020 }, { "epoch": 0.32, "learning_rate": 4.863577806239982e-05, "loss": 1.89, "step": 1030 }, { "epoch": 0.32, "learning_rate": 4.860940925593703e-05, "loss": 1.8893, "step": 1040 }, { "epoch": 0.32, "learning_rate": 4.858279533144358e-05, "loss": 1.9838, "step": 1050 }, { "epoch": 0.33, "learning_rate": 4.855593656523103e-05, "loss": 1.8691, "step": 1060 }, { "epoch": 0.33, "learning_rate": 4.852883323615295e-05, "loss": 1.8883, "step": 1070 }, { "epoch": 0.33, "learning_rate": 4.8501485625601996e-05, "loss": 1.8107, "step": 1080 }, { "epoch": 0.34, "learning_rate": 4.847389401750701e-05, "loss": 1.9194, "step": 1090 }, { "epoch": 0.34, "learning_rate": 4.8446058698330115e-05, "loss": 1.8935, "step": 1100 }, { "epoch": 0.34, "learning_rate": 4.8417979957063624e-05, "loss": 1.8327, "step": 1110 }, { "epoch": 0.34, "learning_rate": 4.838965808522716e-05, "loss": 1.9377, "step": 1120 }, { "epoch": 0.35, "learning_rate": 4.836109337686457e-05, "loss": 1.855, "step": 1130 }, { "epoch": 0.35, "learning_rate": 4.833228612854087e-05, "loss": 1.9036, "step": 1140 }, { "epoch": 0.35, "learning_rate": 4.83032366393392e-05, "loss": 1.7476, "step": 1150 }, { "epoch": 0.36, "learning_rate": 4.827394521085767e-05, "loss": 1.909, "step": 1160 }, { "epoch": 0.36, "learning_rate": 4.8244412147206284e-05, "loss": 1.8406, "step": 1170 }, { "epoch": 0.36, "learning_rate": 4.8214637755003745e-05, "loss": 1.8613, "step": 1180 }, { "epoch": 0.37, "learning_rate": 4.8184622343374275e-05, "loss": 1.792, "step": 1190 }, { "epoch": 0.37, "learning_rate": 4.815436622394441e-05, "loss": 1.8728, "step": 1200 }, { "epoch": 0.37, "learning_rate": 4.812386971083979e-05, "loss": 1.8598, "step": 1210 }, { "epoch": 0.38, "learning_rate": 4.809313312068185e-05, "loss": 1.8852, "step": 1220 }, { "epoch": 0.38, "learning_rate": 4.806215677258456e-05, "loss": 1.8597, "step": 1230 }, { "epoch": 0.38, "learning_rate": 4.803094098815112e-05, "loss": 1.8456, "step": 1240 }, { "epoch": 0.38, "learning_rate": 4.799948609147061e-05, "loss": 1.8788, "step": 1250 }, { "epoch": 0.39, "learning_rate": 4.7967792409114606e-05, "loss": 1.8523, "step": 1260 }, { "epoch": 0.39, "learning_rate": 4.7935860270133844e-05, "loss": 1.8789, "step": 1270 }, { "epoch": 0.39, "learning_rate": 4.7903690006054726e-05, "loss": 1.8876, "step": 1280 }, { "epoch": 0.4, "learning_rate": 4.787128195087596e-05, "loss": 1.8684, "step": 1290 }, { "epoch": 0.4, "learning_rate": 4.783863644106502e-05, "loss": 1.8249, "step": 1300 }, { "epoch": 0.4, "learning_rate": 4.780575381555472e-05, "loss": 1.9299, "step": 1310 }, { "epoch": 0.41, "learning_rate": 4.7772634415739624e-05, "loss": 1.8618, "step": 1320 }, { "epoch": 0.41, "learning_rate": 4.7739278585472573e-05, "loss": 1.9543, "step": 1330 }, { "epoch": 0.41, "learning_rate": 4.7705686671061054e-05, "loss": 1.8329, "step": 1340 }, { "epoch": 0.42, "learning_rate": 4.767185902126364e-05, "loss": 1.9283, "step": 1350 }, { "epoch": 0.42, "learning_rate": 4.763779598728636e-05, "loss": 1.9151, "step": 1360 }, { "epoch": 0.42, "learning_rate": 4.760349792277906e-05, "loss": 1.9361, "step": 1370 }, { "epoch": 0.42, "learning_rate": 4.7568965183831726e-05, "loss": 1.9205, "step": 1380 }, { "epoch": 0.43, "learning_rate": 4.753419812897076e-05, "loss": 1.8365, "step": 1390 }, { "epoch": 0.43, "learning_rate": 4.74991971191553e-05, "loss": 1.8918, "step": 1400 }, { "epoch": 0.43, "learning_rate": 4.7463962517773474e-05, "loss": 1.7551, "step": 1410 }, { "epoch": 0.44, "learning_rate": 4.742849469063858e-05, "loss": 1.892, "step": 1420 }, { "epoch": 0.44, "learning_rate": 4.7392794005985326e-05, "loss": 1.8103, "step": 1430 }, { "epoch": 0.44, "learning_rate": 4.735686083446599e-05, "loss": 1.8027, "step": 1440 }, { "epoch": 0.45, "learning_rate": 4.732069554914661e-05, "loss": 1.8752, "step": 1450 }, { "epoch": 0.45, "learning_rate": 4.728429852550303e-05, "loss": 1.9409, "step": 1460 }, { "epoch": 0.45, "learning_rate": 4.72476701414171e-05, "loss": 1.9495, "step": 1470 }, { "epoch": 0.46, "learning_rate": 4.721081077717267e-05, "loss": 1.8533, "step": 1480 }, { "epoch": 0.46, "learning_rate": 4.7173720815451697e-05, "loss": 1.9093, "step": 1490 }, { "epoch": 0.46, "learning_rate": 4.713640064133025e-05, "loss": 1.8799, "step": 1500 }, { "epoch": 0.46, "learning_rate": 4.7098850642274496e-05, "loss": 1.9039, "step": 1510 }, { "epoch": 0.47, "learning_rate": 4.706107120813671e-05, "loss": 1.9175, "step": 1520 }, { "epoch": 0.47, "learning_rate": 4.702306273115122e-05, "loss": 1.8675, "step": 1530 }, { "epoch": 0.47, "learning_rate": 4.69848256059303e-05, "loss": 1.8481, "step": 1540 }, { "epoch": 0.48, "learning_rate": 4.694636022946012e-05, "loss": 1.8857, "step": 1550 }, { "epoch": 0.48, "learning_rate": 4.690766700109659e-05, "loss": 1.8976, "step": 1560 }, { "epoch": 0.48, "learning_rate": 4.686874632256125e-05, "loss": 1.8069, "step": 1570 }, { "epoch": 0.49, "learning_rate": 4.682959859793706e-05, "loss": 1.8263, "step": 1580 }, { "epoch": 0.49, "learning_rate": 4.679022423366424e-05, "loss": 1.8853, "step": 1590 }, { "epoch": 0.49, "learning_rate": 4.6750623638535986e-05, "loss": 1.8682, "step": 1600 }, { "epoch": 0.5, "learning_rate": 4.6710797223694325e-05, "loss": 1.8553, "step": 1610 }, { "epoch": 0.5, "learning_rate": 4.667074540262577e-05, "loss": 1.867, "step": 1620 }, { "epoch": 0.5, "learning_rate": 4.663046859115706e-05, "loss": 1.9481, "step": 1630 }, { "epoch": 0.5, "learning_rate": 4.658996720745081e-05, "loss": 1.8909, "step": 1640 }, { "epoch": 0.51, "learning_rate": 4.654924167200123e-05, "loss": 1.9205, "step": 1650 }, { "epoch": 0.51, "learning_rate": 4.6508292407629694e-05, "loss": 1.8306, "step": 1660 }, { "epoch": 0.51, "learning_rate": 4.6467119839480374e-05, "loss": 1.8504, "step": 1670 }, { "epoch": 0.52, "learning_rate": 4.6425724395015865e-05, "loss": 1.7526, "step": 1680 }, { "epoch": 0.52, "learning_rate": 4.638410650401267e-05, "loss": 1.9188, "step": 1690 }, { "epoch": 0.52, "learning_rate": 4.6342266598556814e-05, "loss": 1.8914, "step": 1700 }, { "epoch": 0.53, "learning_rate": 4.63002051130393e-05, "loss": 1.8979, "step": 1710 }, { "epoch": 0.53, "learning_rate": 4.625792248415164e-05, "loss": 1.8876, "step": 1720 }, { "epoch": 0.53, "learning_rate": 4.62154191508813e-05, "loss": 1.925, "step": 1730 }, { "epoch": 0.54, "learning_rate": 4.617269555450715e-05, "loss": 1.9125, "step": 1740 }, { "epoch": 0.54, "learning_rate": 4.6129752138594874e-05, "loss": 1.8517, "step": 1750 }, { "epoch": 0.54, "learning_rate": 4.608658934899236e-05, "loss": 1.8005, "step": 1760 }, { "epoch": 0.54, "learning_rate": 4.604320763382511e-05, "loss": 1.8604, "step": 1770 }, { "epoch": 0.55, "learning_rate": 4.5999607443491535e-05, "loss": 1.8665, "step": 1780 }, { "epoch": 0.55, "learning_rate": 4.595578923065831e-05, "loss": 1.9015, "step": 1790 }, { "epoch": 0.55, "learning_rate": 4.5911753450255665e-05, "loss": 1.879, "step": 1800 }, { "epoch": 0.56, "learning_rate": 4.5867500559472676e-05, "loss": 1.8441, "step": 1810 }, { "epoch": 0.56, "learning_rate": 4.5823031017752485e-05, "loss": 1.8003, "step": 1820 }, { "epoch": 0.56, "learning_rate": 4.577834528678757e-05, "loss": 1.8833, "step": 1830 }, { "epoch": 0.57, "learning_rate": 4.573344383051492e-05, "loss": 1.8897, "step": 1840 }, { "epoch": 0.57, "learning_rate": 4.568832711511125e-05, "loss": 1.7957, "step": 1850 }, { "epoch": 0.57, "learning_rate": 4.56429956089881e-05, "loss": 1.8376, "step": 1860 }, { "epoch": 0.58, "learning_rate": 4.5597449782787076e-05, "loss": 1.8961, "step": 1870 }, { "epoch": 0.58, "learning_rate": 4.555169010937485e-05, "loss": 1.7671, "step": 1880 }, { "epoch": 0.58, "learning_rate": 4.5505717063838326e-05, "loss": 1.8202, "step": 1890 }, { "epoch": 0.58, "learning_rate": 4.5459531123479676e-05, "loss": 1.8546, "step": 1900 }, { "epoch": 0.59, "learning_rate": 4.54131327678114e-05, "loss": 1.7202, "step": 1910 }, { "epoch": 0.59, "learning_rate": 4.5366522478551335e-05, "loss": 1.771, "step": 1920 }, { "epoch": 0.59, "learning_rate": 4.531970073961764e-05, "loss": 1.8087, "step": 1930 }, { "epoch": 0.6, "learning_rate": 4.5272668037123845e-05, "loss": 1.7593, "step": 1940 }, { "epoch": 0.6, "learning_rate": 4.522542485937369e-05, "loss": 1.89, "step": 1950 }, { "epoch": 0.6, "learning_rate": 4.517797169685614e-05, "loss": 1.8375, "step": 1960 }, { "epoch": 0.61, "learning_rate": 4.5130309042240275e-05, "loss": 1.8607, "step": 1970 }, { "epoch": 0.61, "learning_rate": 4.508243739037016e-05, "loss": 1.8552, "step": 1980 }, { "epoch": 0.61, "learning_rate": 4.503435723825971e-05, "loss": 1.8239, "step": 1990 }, { "epoch": 0.62, "learning_rate": 4.498606908508754e-05, "loss": 1.8484, "step": 2000 }, { "epoch": 0.62, "learning_rate": 4.4937573432191766e-05, "loss": 1.8745, "step": 2010 }, { "epoch": 0.62, "learning_rate": 4.488887078306484e-05, "loss": 1.8425, "step": 2020 }, { "epoch": 0.62, "learning_rate": 4.4839961643348274e-05, "loss": 1.8812, "step": 2030 }, { "epoch": 0.63, "learning_rate": 4.47908465208274e-05, "loss": 1.8976, "step": 2040 }, { "epoch": 0.63, "learning_rate": 4.4741525925426125e-05, "loss": 1.8705, "step": 2050 }, { "epoch": 0.63, "learning_rate": 4.469200036920162e-05, "loss": 1.8139, "step": 2060 }, { "epoch": 0.64, "learning_rate": 4.464227036633901e-05, "loss": 1.8043, "step": 2070 }, { "epoch": 0.64, "learning_rate": 4.4592336433146e-05, "loss": 1.843, "step": 2080 }, { "epoch": 0.64, "learning_rate": 4.454219908804757e-05, "loss": 1.912, "step": 2090 }, { "epoch": 0.65, "learning_rate": 4.449185885158056e-05, "loss": 1.8339, "step": 2100 }, { "epoch": 0.65, "learning_rate": 4.444131624638828e-05, "loss": 1.8523, "step": 2110 }, { "epoch": 0.65, "learning_rate": 4.4390571797215066e-05, "loss": 1.8037, "step": 2120 }, { "epoch": 0.66, "learning_rate": 4.433962603090083e-05, "loss": 1.8474, "step": 2130 }, { "epoch": 0.66, "learning_rate": 4.428847947637563e-05, "loss": 1.8136, "step": 2140 }, { "epoch": 0.66, "learning_rate": 4.4237132664654154e-05, "loss": 1.7571, "step": 2150 }, { "epoch": 0.66, "learning_rate": 4.418558612883016e-05, "loss": 1.8639, "step": 2160 }, { "epoch": 0.67, "learning_rate": 4.413384040407103e-05, "loss": 1.8588, "step": 2170 }, { "epoch": 0.67, "learning_rate": 4.408189602761214e-05, "loss": 1.8709, "step": 2180 }, { "epoch": 0.67, "learning_rate": 4.4029753538751336e-05, "loss": 1.8106, "step": 2190 }, { "epoch": 0.68, "learning_rate": 4.397741347884329e-05, "loss": 1.8488, "step": 2200 }, { "epoch": 0.68, "learning_rate": 4.3924876391293915e-05, "loss": 1.7862, "step": 2210 }, { "epoch": 0.68, "learning_rate": 4.387214282155469e-05, "loss": 1.8279, "step": 2220 }, { "epoch": 0.69, "learning_rate": 4.381921331711704e-05, "loss": 1.806, "step": 2230 }, { "epoch": 0.69, "learning_rate": 4.376608842750658e-05, "loss": 1.8176, "step": 2240 }, { "epoch": 0.69, "learning_rate": 4.371276870427753e-05, "loss": 1.7647, "step": 2250 }, { "epoch": 0.7, "learning_rate": 4.3659254701006844e-05, "loss": 1.7559, "step": 2260 }, { "epoch": 0.7, "learning_rate": 4.360554697328858e-05, "loss": 1.7921, "step": 2270 }, { "epoch": 0.7, "learning_rate": 4.355164607872806e-05, "loss": 1.7913, "step": 2280 }, { "epoch": 0.7, "learning_rate": 4.3497552576936106e-05, "loss": 1.7821, "step": 2290 }, { "epoch": 0.71, "learning_rate": 4.344326702952326e-05, "loss": 1.7978, "step": 2300 }, { "epoch": 0.71, "learning_rate": 4.3388790000093885e-05, "loss": 1.8159, "step": 2310 }, { "epoch": 0.71, "learning_rate": 4.3334122054240374e-05, "loss": 1.8286, "step": 2320 }, { "epoch": 0.72, "learning_rate": 4.327926375953725e-05, "loss": 1.8697, "step": 2330 }, { "epoch": 0.72, "learning_rate": 4.3224215685535294e-05, "loss": 1.8097, "step": 2340 }, { "epoch": 0.72, "learning_rate": 4.316897840375558e-05, "loss": 1.8376, "step": 2350 }, { "epoch": 0.73, "learning_rate": 4.3113552487683626e-05, "loss": 1.8426, "step": 2360 }, { "epoch": 0.73, "learning_rate": 4.305793851276335e-05, "loss": 1.829, "step": 2370 }, { "epoch": 0.73, "learning_rate": 4.3002137056391165e-05, "loss": 1.7839, "step": 2380 }, { "epoch": 0.74, "learning_rate": 4.294614869790994e-05, "loss": 1.7525, "step": 2390 }, { "epoch": 0.74, "learning_rate": 4.288997401860303e-05, "loss": 1.8412, "step": 2400 }, { "epoch": 0.74, "learning_rate": 4.283361360168817e-05, "loss": 1.8285, "step": 2410 }, { "epoch": 0.74, "learning_rate": 4.277706803231151e-05, "loss": 1.8334, "step": 2420 }, { "epoch": 0.75, "learning_rate": 4.272033789754146e-05, "loss": 1.7826, "step": 2430 }, { "epoch": 0.75, "learning_rate": 4.266342378636263e-05, "loss": 1.7866, "step": 2440 }, { "epoch": 0.75, "learning_rate": 4.260632628966974e-05, "loss": 1.8037, "step": 2450 }, { "epoch": 0.76, "learning_rate": 4.254904600026143e-05, "loss": 1.7932, "step": 2460 }, { "epoch": 0.76, "learning_rate": 4.249158351283414e-05, "loss": 1.8014, "step": 2470 }, { "epoch": 0.76, "learning_rate": 4.2433939423975945e-05, "loss": 1.7316, "step": 2480 }, { "epoch": 0.77, "learning_rate": 4.237611433216032e-05, "loss": 1.8431, "step": 2490 }, { "epoch": 0.77, "learning_rate": 4.231810883773999e-05, "loss": 1.8427, "step": 2500 }, { "epoch": 0.77, "learning_rate": 4.225992354294061e-05, "loss": 1.8151, "step": 2510 }, { "epoch": 0.78, "learning_rate": 4.2201559051854605e-05, "loss": 1.8124, "step": 2520 }, { "epoch": 0.78, "learning_rate": 4.2143015970434827e-05, "loss": 1.7989, "step": 2530 }, { "epoch": 0.78, "learning_rate": 4.2084294906488314e-05, "loss": 1.811, "step": 2540 }, { "epoch": 0.78, "learning_rate": 4.202539646966993e-05, "loss": 1.823, "step": 2550 }, { "epoch": 0.79, "learning_rate": 4.196632127147609e-05, "loss": 1.8881, "step": 2560 }, { "epoch": 0.79, "learning_rate": 4.190706992523836e-05, "loss": 1.8345, "step": 2570 }, { "epoch": 0.79, "learning_rate": 4.184764304611715e-05, "loss": 1.7935, "step": 2580 }, { "epoch": 0.8, "learning_rate": 4.178804125109523e-05, "loss": 1.8584, "step": 2590 }, { "epoch": 0.8, "learning_rate": 4.172826515897146e-05, "loss": 1.8177, "step": 2600 }, { "epoch": 0.8, "learning_rate": 4.166831539035423e-05, "loss": 1.7585, "step": 2610 }, { "epoch": 0.81, "learning_rate": 4.160819256765509e-05, "loss": 1.787, "step": 2620 }, { "epoch": 0.81, "learning_rate": 4.15478973150823e-05, "loss": 1.7788, "step": 2630 }, { "epoch": 0.81, "learning_rate": 4.1487430258634315e-05, "loss": 1.8552, "step": 2640 }, { "epoch": 0.82, "learning_rate": 4.1426792026093276e-05, "loss": 1.784, "step": 2650 }, { "epoch": 0.82, "learning_rate": 4.1365983247018514e-05, "loss": 1.7272, "step": 2660 }, { "epoch": 0.82, "learning_rate": 4.130500455274005e-05, "loss": 1.8041, "step": 2670 }, { "epoch": 0.82, "learning_rate": 4.124385657635196e-05, "loss": 1.8059, "step": 2680 }, { "epoch": 0.83, "learning_rate": 4.1182539952705844e-05, "loss": 1.7124, "step": 2690 }, { "epoch": 0.83, "learning_rate": 4.1121055318404264e-05, "loss": 1.8392, "step": 2700 }, { "epoch": 0.83, "learning_rate": 4.1059403311794085e-05, "loss": 1.8285, "step": 2710 }, { "epoch": 0.84, "learning_rate": 4.099758457295988e-05, "loss": 1.7987, "step": 2720 }, { "epoch": 0.84, "learning_rate": 4.093559974371725e-05, "loss": 1.8111, "step": 2730 }, { "epoch": 0.84, "learning_rate": 4.087344946760619e-05, "loss": 1.9137, "step": 2740 }, { "epoch": 0.85, "learning_rate": 4.0811134389884433e-05, "loss": 1.7893, "step": 2750 }, { "epoch": 0.85, "learning_rate": 4.074865515752068e-05, "loss": 1.7505, "step": 2760 }, { "epoch": 0.85, "learning_rate": 4.068601241918792e-05, "loss": 1.753, "step": 2770 }, { "epoch": 0.86, "learning_rate": 4.062320682525673e-05, "loss": 1.8223, "step": 2780 }, { "epoch": 0.86, "learning_rate": 4.056023902778846e-05, "loss": 1.7569, "step": 2790 }, { "epoch": 0.86, "learning_rate": 4.049710968052851e-05, "loss": 1.7899, "step": 2800 }, { "epoch": 0.86, "learning_rate": 4.04338194388995e-05, "loss": 1.7871, "step": 2810 }, { "epoch": 0.87, "learning_rate": 4.037036895999453e-05, "loss": 1.7877, "step": 2820 }, { "epoch": 0.87, "learning_rate": 4.0306758902570286e-05, "loss": 1.8113, "step": 2830 }, { "epoch": 0.87, "learning_rate": 4.0242989927040234e-05, "loss": 1.7965, "step": 2840 }, { "epoch": 0.88, "learning_rate": 4.0179062695467784e-05, "loss": 1.7561, "step": 2850 }, { "epoch": 0.88, "learning_rate": 4.011497787155938e-05, "loss": 1.8149, "step": 2860 }, { "epoch": 0.88, "learning_rate": 4.005073612065761e-05, "loss": 1.8009, "step": 2870 }, { "epoch": 0.89, "learning_rate": 3.998633810973435e-05, "loss": 1.8362, "step": 2880 }, { "epoch": 0.89, "learning_rate": 3.992178450738378e-05, "loss": 1.7808, "step": 2890 }, { "epoch": 0.89, "learning_rate": 3.985707598381544e-05, "loss": 1.7197, "step": 2900 }, { "epoch": 0.9, "learning_rate": 3.979221321084734e-05, "loss": 1.8189, "step": 2910 }, { "epoch": 0.9, "learning_rate": 3.972719686189891e-05, "loss": 1.8474, "step": 2920 }, { "epoch": 0.9, "learning_rate": 3.9662027611984064e-05, "loss": 1.7893, "step": 2930 }, { "epoch": 0.9, "learning_rate": 3.959670613770414e-05, "loss": 1.7774, "step": 2940 }, { "epoch": 0.91, "learning_rate": 3.9531233117240916e-05, "loss": 1.7849, "step": 2950 }, { "epoch": 0.91, "learning_rate": 3.946560923034957e-05, "loss": 1.8474, "step": 2960 }, { "epoch": 0.91, "learning_rate": 3.939983515835157e-05, "loss": 1.7107, "step": 2970 }, { "epoch": 0.92, "learning_rate": 3.933391158412767e-05, "loss": 1.8542, "step": 2980 }, { "epoch": 0.92, "learning_rate": 3.92678391921108e-05, "loss": 1.8095, "step": 2990 }, { "epoch": 0.92, "learning_rate": 3.920161866827889e-05, "loss": 1.7845, "step": 3000 }, { "epoch": 0.93, "learning_rate": 3.913525070014789e-05, "loss": 1.7703, "step": 3010 }, { "epoch": 0.93, "learning_rate": 3.9068735976764476e-05, "loss": 1.8042, "step": 3020 }, { "epoch": 0.93, "learning_rate": 3.900207518869901e-05, "loss": 1.7716, "step": 3030 }, { "epoch": 0.94, "learning_rate": 3.8935269028038297e-05, "loss": 1.8361, "step": 3040 }, { "epoch": 0.94, "learning_rate": 3.886831818837847e-05, "loss": 1.7714, "step": 3050 }, { "epoch": 0.94, "learning_rate": 3.880122336481774e-05, "loss": 1.7702, "step": 3060 }, { "epoch": 0.94, "learning_rate": 3.873398525394917e-05, "loss": 1.737, "step": 3070 }, { "epoch": 0.95, "learning_rate": 3.866660455385347e-05, "loss": 1.7988, "step": 3080 }, { "epoch": 0.95, "learning_rate": 3.8599081964091765e-05, "loss": 1.7323, "step": 3090 }, { "epoch": 0.95, "learning_rate": 3.853141818569829e-05, "loss": 1.7876, "step": 3100 }, { "epoch": 0.96, "learning_rate": 3.846361392117312e-05, "loss": 1.8245, "step": 3110 }, { "epoch": 0.96, "learning_rate": 3.8395669874474915e-05, "loss": 1.7646, "step": 3120 }, { "epoch": 0.96, "learning_rate": 3.832758675101358e-05, "loss": 1.6854, "step": 3130 }, { "epoch": 0.97, "learning_rate": 3.8259365257642905e-05, "loss": 1.8589, "step": 3140 }, { "epoch": 0.97, "learning_rate": 3.819100610265332e-05, "loss": 1.8388, "step": 3150 }, { "epoch": 0.97, "learning_rate": 3.812250999576446e-05, "loss": 1.834, "step": 3160 }, { "epoch": 0.98, "learning_rate": 3.805387764811784e-05, "loss": 1.7583, "step": 3170 }, { "epoch": 0.98, "learning_rate": 3.7985109772269436e-05, "loss": 1.6828, "step": 3180 }, { "epoch": 0.98, "learning_rate": 3.791620708218231e-05, "loss": 1.9199, "step": 3190 }, { "epoch": 0.98, "learning_rate": 3.784717029321922e-05, "loss": 1.8368, "step": 3200 }, { "epoch": 0.99, "learning_rate": 3.777800012213514e-05, "loss": 1.8072, "step": 3210 }, { "epoch": 0.99, "learning_rate": 3.770869728706985e-05, "loss": 1.7823, "step": 3220 }, { "epoch": 0.99, "learning_rate": 3.7639262507540495e-05, "loss": 1.7866, "step": 3230 }, { "epoch": 1.0, "learning_rate": 3.756969650443408e-05, "loss": 1.8173, "step": 3240 }, { "epoch": 1.0, "learning_rate": 3.7500000000000003e-05, "loss": 1.7384, "step": 3250 }, { "epoch": 1.0, "learning_rate": 3.743017371784256e-05, "loss": 1.7833, "step": 3260 }, { "epoch": 1.01, "learning_rate": 3.7360218382913425e-05, "loss": 1.8096, "step": 3270 }, { "epoch": 1.01, "learning_rate": 3.7290134721504125e-05, "loss": 1.8112, "step": 3280 }, { "epoch": 1.01, "learning_rate": 3.7219923461238505e-05, "loss": 1.7851, "step": 3290 }, { "epoch": 1.02, "learning_rate": 3.714958533106515e-05, "loss": 1.8376, "step": 3300 }, { "epoch": 1.02, "learning_rate": 3.707912106124988e-05, "loss": 1.7189, "step": 3310 }, { "epoch": 1.02, "learning_rate": 3.7008531383368075e-05, "loss": 1.7853, "step": 3320 }, { "epoch": 1.02, "learning_rate": 3.693781703029716e-05, "loss": 1.8215, "step": 3330 }, { "epoch": 1.03, "learning_rate": 3.6866978736208957e-05, "loss": 1.7875, "step": 3340 }, { "epoch": 1.03, "learning_rate": 3.679601723656205e-05, "loss": 1.8367, "step": 3350 }, { "epoch": 1.03, "learning_rate": 3.672493326809422e-05, "loss": 1.766, "step": 3360 }, { "epoch": 1.04, "learning_rate": 3.665372756881468e-05, "loss": 1.774, "step": 3370 }, { "epoch": 1.04, "learning_rate": 3.6582400877996546e-05, "loss": 1.7966, "step": 3380 }, { "epoch": 1.04, "learning_rate": 3.651095393616904e-05, "loss": 1.7605, "step": 3390 }, { "epoch": 1.05, "learning_rate": 3.643938748510989e-05, "loss": 1.7661, "step": 3400 }, { "epoch": 1.05, "learning_rate": 3.6367702267837564e-05, "loss": 1.8226, "step": 3410 }, { "epoch": 1.05, "learning_rate": 3.629589902860363e-05, "loss": 1.6693, "step": 3420 }, { "epoch": 1.06, "learning_rate": 3.6223978512884945e-05, "loss": 1.8266, "step": 3430 }, { "epoch": 1.06, "learning_rate": 3.615194146737598e-05, "loss": 1.8096, "step": 3440 }, { "epoch": 1.06, "learning_rate": 3.6079788639981036e-05, "loss": 1.8243, "step": 3450 }, { "epoch": 1.06, "learning_rate": 3.600752077980648e-05, "loss": 1.7899, "step": 3460 }, { "epoch": 1.07, "learning_rate": 3.5935138637153e-05, "loss": 1.8055, "step": 3470 }, { "epoch": 1.07, "learning_rate": 3.586264296350775e-05, "loss": 1.7409, "step": 3480 }, { "epoch": 1.07, "learning_rate": 3.579003451153661e-05, "loss": 1.6908, "step": 3490 }, { "epoch": 1.08, "learning_rate": 3.5717314035076355e-05, "loss": 1.8248, "step": 3500 }, { "epoch": 1.08, "learning_rate": 3.564448228912682e-05, "loss": 1.7718, "step": 3510 }, { "epoch": 1.08, "learning_rate": 3.557154002984305e-05, "loss": 1.7122, "step": 3520 }, { "epoch": 1.09, "learning_rate": 3.549848801452747e-05, "loss": 1.696, "step": 3530 }, { "epoch": 1.09, "learning_rate": 3.5425327001622036e-05, "loss": 1.7565, "step": 3540 }, { "epoch": 1.09, "learning_rate": 3.53520577507003e-05, "loss": 1.7459, "step": 3550 }, { "epoch": 1.1, "learning_rate": 3.527868102245961e-05, "loss": 1.7749, "step": 3560 }, { "epoch": 1.1, "learning_rate": 3.520519757871313e-05, "loss": 1.774, "step": 3570 }, { "epoch": 1.1, "learning_rate": 3.5131608182381985e-05, "loss": 1.7995, "step": 3580 }, { "epoch": 1.1, "learning_rate": 3.505791359748733e-05, "loss": 1.8139, "step": 3590 }, { "epoch": 1.11, "learning_rate": 3.498411458914238e-05, "loss": 1.803, "step": 3600 }, { "epoch": 1.11, "learning_rate": 3.491021192354455e-05, "loss": 1.8434, "step": 3610 }, { "epoch": 1.11, "learning_rate": 3.48362063679674e-05, "loss": 1.6962, "step": 3620 }, { "epoch": 1.12, "learning_rate": 3.4762098690752724e-05, "loss": 1.8522, "step": 3630 }, { "epoch": 1.12, "learning_rate": 3.4687889661302576e-05, "loss": 1.8024, "step": 3640 }, { "epoch": 1.12, "learning_rate": 3.461358005007128e-05, "loss": 1.7276, "step": 3650 }, { "epoch": 1.13, "learning_rate": 3.453917062855738e-05, "loss": 1.8394, "step": 3660 }, { "epoch": 1.13, "learning_rate": 3.4464662169295735e-05, "loss": 1.8862, "step": 3670 }, { "epoch": 1.13, "learning_rate": 3.439005544584939e-05, "loss": 1.8092, "step": 3680 }, { "epoch": 1.14, "learning_rate": 3.43153512328016e-05, "loss": 1.8219, "step": 3690 }, { "epoch": 1.14, "learning_rate": 3.424055030574778e-05, "loss": 1.7947, "step": 3700 }, { "epoch": 1.14, "learning_rate": 3.4165653441287444e-05, "loss": 1.7079, "step": 3710 }, { "epoch": 1.14, "learning_rate": 3.4090661417016175e-05, "loss": 1.8115, "step": 3720 }, { "epoch": 1.15, "learning_rate": 3.40155750115175e-05, "loss": 1.7297, "step": 3730 }, { "epoch": 1.15, "learning_rate": 3.394039500435483e-05, "loss": 1.8022, "step": 3740 }, { "epoch": 1.15, "learning_rate": 3.386512217606339e-05, "loss": 1.7899, "step": 3750 }, { "epoch": 1.16, "learning_rate": 3.378975730814208e-05, "loss": 1.8002, "step": 3760 }, { "epoch": 1.16, "learning_rate": 3.3714301183045385e-05, "loss": 1.7864, "step": 3770 }, { "epoch": 1.16, "learning_rate": 3.363875458417522e-05, "loss": 1.7376, "step": 3780 }, { "epoch": 1.17, "learning_rate": 3.3563118295872843e-05, "loss": 1.8043, "step": 3790 }, { "epoch": 1.17, "learning_rate": 3.348739310341068e-05, "loss": 1.8033, "step": 3800 }, { "epoch": 1.17, "learning_rate": 3.3411579792984174e-05, "loss": 1.7982, "step": 3810 }, { "epoch": 1.18, "learning_rate": 3.333567915170364e-05, "loss": 1.7382, "step": 3820 }, { "epoch": 1.18, "learning_rate": 3.325969196758605e-05, "loss": 1.7766, "step": 3830 }, { "epoch": 1.18, "learning_rate": 3.318361902954692e-05, "loss": 1.7832, "step": 3840 }, { "epoch": 1.18, "learning_rate": 3.310746112739207e-05, "loss": 1.7566, "step": 3850 }, { "epoch": 1.19, "learning_rate": 3.303121905180942e-05, "loss": 1.7124, "step": 3860 }, { "epoch": 1.19, "learning_rate": 3.295489359436083e-05, "loss": 1.7657, "step": 3870 }, { "epoch": 1.19, "learning_rate": 3.2878485547473816e-05, "loss": 1.7219, "step": 3880 }, { "epoch": 1.2, "learning_rate": 3.280199570443337e-05, "loss": 1.713, "step": 3890 }, { "epoch": 1.2, "learning_rate": 3.272542485937369e-05, "loss": 1.7764, "step": 3900 }, { "epoch": 1.2, "learning_rate": 3.264877380726999e-05, "loss": 1.7705, "step": 3910 }, { "epoch": 1.21, "learning_rate": 3.257204334393019e-05, "loss": 1.7792, "step": 3920 }, { "epoch": 1.21, "learning_rate": 3.249523426598669e-05, "loss": 1.6632, "step": 3930 }, { "epoch": 1.21, "learning_rate": 3.241834737088804e-05, "loss": 1.7113, "step": 3940 }, { "epoch": 1.22, "learning_rate": 3.234138345689077e-05, "loss": 1.8279, "step": 3950 }, { "epoch": 1.22, "learning_rate": 3.226434332305098e-05, "loss": 1.7193, "step": 3960 }, { "epoch": 1.22, "learning_rate": 3.218722776921614e-05, "loss": 1.8368, "step": 3970 }, { "epoch": 1.22, "learning_rate": 3.211003759601671e-05, "loss": 1.804, "step": 3980 }, { "epoch": 1.23, "learning_rate": 3.2032773604857915e-05, "loss": 1.6929, "step": 3990 }, { "epoch": 1.23, "learning_rate": 3.195543659791132e-05, "loss": 1.719, "step": 4000 }, { "epoch": 1.23, "learning_rate": 3.187802737810659e-05, "loss": 1.797, "step": 4010 }, { "epoch": 1.24, "learning_rate": 3.180054674912311e-05, "loss": 1.8046, "step": 4020 }, { "epoch": 1.24, "learning_rate": 3.172299551538164e-05, "loss": 1.7679, "step": 4030 }, { "epoch": 1.24, "learning_rate": 3.164537448203601e-05, "loss": 1.7048, "step": 4040 }, { "epoch": 1.25, "learning_rate": 3.1567684454964675e-05, "loss": 1.7146, "step": 4050 }, { "epoch": 1.25, "learning_rate": 3.1489926240762437e-05, "loss": 1.742, "step": 4060 }, { "epoch": 1.25, "learning_rate": 3.141210064673201e-05, "loss": 1.8007, "step": 4070 }, { "epoch": 1.26, "learning_rate": 3.133420848087566e-05, "loss": 1.7478, "step": 4080 }, { "epoch": 1.26, "learning_rate": 3.125625055188684e-05, "loss": 1.7919, "step": 4090 }, { "epoch": 1.26, "learning_rate": 3.1178227669141744e-05, "loss": 1.8328, "step": 4100 }, { "epoch": 1.26, "learning_rate": 3.110014064269094e-05, "loss": 1.8401, "step": 4110 }, { "epoch": 1.27, "learning_rate": 3.102199028325095e-05, "loss": 1.7198, "step": 4120 }, { "epoch": 1.27, "learning_rate": 3.094377740219583e-05, "loss": 1.733, "step": 4130 }, { "epoch": 1.27, "learning_rate": 3.0865502811548755e-05, "loss": 1.7315, "step": 4140 }, { "epoch": 1.28, "learning_rate": 3.078716732397358e-05, "loss": 1.6147, "step": 4150 }, { "epoch": 1.28, "learning_rate": 3.0708771752766394e-05, "loss": 1.8375, "step": 4160 }, { "epoch": 1.28, "learning_rate": 3.063031691184711e-05, "loss": 1.7714, "step": 4170 }, { "epoch": 1.29, "learning_rate": 3.0551803615750976e-05, "loss": 1.7458, "step": 4180 }, { "epoch": 1.29, "learning_rate": 3.0473232679620127e-05, "loss": 1.6214, "step": 4190 }, { "epoch": 1.29, "learning_rate": 3.0394604919195156e-05, "loss": 1.7469, "step": 4200 }, { "epoch": 1.3, "learning_rate": 3.0315921150806593e-05, "loss": 1.8128, "step": 4210 }, { "epoch": 1.3, "learning_rate": 3.0237182191366464e-05, "loss": 1.7645, "step": 4220 }, { "epoch": 1.3, "learning_rate": 3.015838885835981e-05, "loss": 1.7573, "step": 4230 }, { "epoch": 1.3, "learning_rate": 3.007954196983616e-05, "loss": 1.7605, "step": 4240 }, { "epoch": 1.31, "learning_rate": 3.0000642344401113e-05, "loss": 1.822, "step": 4250 }, { "epoch": 1.31, "learning_rate": 2.9921690801207758e-05, "loss": 1.8095, "step": 4260 }, { "epoch": 1.31, "learning_rate": 2.9842688159948222e-05, "loss": 1.7215, "step": 4270 }, { "epoch": 1.32, "learning_rate": 2.976363524084512e-05, "loss": 1.7886, "step": 4280 }, { "epoch": 1.32, "learning_rate": 2.9684532864643122e-05, "loss": 1.8241, "step": 4290 }, { "epoch": 1.32, "learning_rate": 2.960538185260029e-05, "loss": 1.7912, "step": 4300 }, { "epoch": 1.33, "learning_rate": 2.9526183026479702e-05, "loss": 1.7854, "step": 4310 }, { "epoch": 1.33, "learning_rate": 2.9446937208540808e-05, "loss": 1.8215, "step": 4320 }, { "epoch": 1.33, "learning_rate": 2.936764522153097e-05, "loss": 1.6265, "step": 4330 }, { "epoch": 1.34, "learning_rate": 2.9288307888676863e-05, "loss": 1.7179, "step": 4340 }, { "epoch": 1.34, "learning_rate": 2.920892603367596e-05, "loss": 1.7557, "step": 4350 }, { "epoch": 1.34, "learning_rate": 2.9129500480687992e-05, "loss": 1.8139, "step": 4360 }, { "epoch": 1.34, "learning_rate": 2.905003205432633e-05, "loss": 1.7388, "step": 4370 }, { "epoch": 1.35, "learning_rate": 2.897052157964952e-05, "loss": 1.7487, "step": 4380 }, { "epoch": 1.35, "learning_rate": 2.889096988215259e-05, "loss": 1.7834, "step": 4390 }, { "epoch": 1.35, "learning_rate": 2.8811377787758636e-05, "loss": 1.7562, "step": 4400 }, { "epoch": 1.36, "learning_rate": 2.8731746122810105e-05, "loss": 1.8293, "step": 4410 }, { "epoch": 1.36, "learning_rate": 2.8652075714060295e-05, "loss": 1.8106, "step": 4420 }, { "epoch": 1.36, "learning_rate": 2.8572367388664738e-05, "loss": 1.6614, "step": 4430 }, { "epoch": 1.37, "learning_rate": 2.849262197417265e-05, "loss": 1.7935, "step": 4440 }, { "epoch": 1.37, "learning_rate": 2.8412840298518296e-05, "loss": 1.6841, "step": 4450 }, { "epoch": 1.37, "learning_rate": 2.8333023190012398e-05, "loss": 1.7412, "step": 4460 }, { "epoch": 1.38, "learning_rate": 2.8253171477333586e-05, "loss": 1.7687, "step": 4470 }, { "epoch": 1.38, "learning_rate": 2.817328598951972e-05, "loss": 1.7405, "step": 4480 }, { "epoch": 1.38, "learning_rate": 2.810136085822931e-05, "loss": 1.7526, "step": 4490 }, { "epoch": 1.38, "learning_rate": 2.8021413482907176e-05, "loss": 1.7581, "step": 4500 }, { "epoch": 1.39, "learning_rate": 2.794143473861198e-05, "loss": 1.7375, "step": 4510 }, { "epoch": 1.39, "learning_rate": 2.786142545570049e-05, "loss": 1.7513, "step": 4520 }, { "epoch": 1.39, "learning_rate": 2.7781386464846497e-05, "loss": 1.789, "step": 4530 }, { "epoch": 1.4, "learning_rate": 2.7701318597032248e-05, "loss": 1.78, "step": 4540 }, { "epoch": 1.4, "learning_rate": 2.7621222683539792e-05, "loss": 1.7686, "step": 4550 }, { "epoch": 1.4, "learning_rate": 2.754109955594235e-05, "loss": 1.7522, "step": 4560 }, { "epoch": 1.41, "learning_rate": 2.7460950046095696e-05, "loss": 1.8161, "step": 4570 }, { "epoch": 1.41, "learning_rate": 2.738077498612949e-05, "loss": 1.7934, "step": 4580 }, { "epoch": 1.41, "learning_rate": 2.7300575208438683e-05, "loss": 1.6656, "step": 4590 }, { "epoch": 1.42, "learning_rate": 2.7220351545674834e-05, "loss": 1.7707, "step": 4600 }, { "epoch": 1.42, "learning_rate": 2.7140104830737496e-05, "loss": 1.785, "step": 4610 }, { "epoch": 1.42, "learning_rate": 2.705983589676554e-05, "loss": 1.7058, "step": 4620 }, { "epoch": 1.42, "learning_rate": 2.6979545577128522e-05, "loss": 1.7567, "step": 4630 }, { "epoch": 1.43, "learning_rate": 2.6899234705418052e-05, "loss": 1.7514, "step": 4640 }, { "epoch": 1.43, "learning_rate": 2.681890411543908e-05, "loss": 1.8049, "step": 4650 }, { "epoch": 1.43, "learning_rate": 2.6738554641201298e-05, "loss": 1.7631, "step": 4660 }, { "epoch": 1.44, "learning_rate": 2.6658187116910455e-05, "loss": 1.7324, "step": 4670 }, { "epoch": 1.44, "learning_rate": 2.6577802376959698e-05, "loss": 1.7128, "step": 4680 }, { "epoch": 1.44, "learning_rate": 2.64974012559209e-05, "loss": 1.7512, "step": 4690 }, { "epoch": 1.45, "learning_rate": 2.641698458853603e-05, "loss": 1.7625, "step": 4700 }, { "epoch": 1.45, "learning_rate": 2.6336553209708447e-05, "loss": 1.7298, "step": 4710 }, { "epoch": 1.45, "learning_rate": 2.6256107954494242e-05, "loss": 1.8079, "step": 4720 }, { "epoch": 1.46, "learning_rate": 2.6175649658093586e-05, "loss": 1.7074, "step": 4730 }, { "epoch": 1.46, "learning_rate": 2.609517915584204e-05, "loss": 1.7502, "step": 4740 }, { "epoch": 1.46, "learning_rate": 2.6014697283201907e-05, "loss": 1.6578, "step": 4750 }, { "epoch": 1.46, "learning_rate": 2.5934204875753494e-05, "loss": 1.7288, "step": 4760 }, { "epoch": 1.47, "learning_rate": 2.5853702769186528e-05, "loss": 1.7633, "step": 4770 }, { "epoch": 1.47, "learning_rate": 2.5773191799291417e-05, "loss": 1.7406, "step": 4780 }, { "epoch": 1.47, "learning_rate": 2.569267280195059e-05, "loss": 1.7973, "step": 4790 }, { "epoch": 1.48, "learning_rate": 2.5612146613129828e-05, "loss": 1.7675, "step": 4800 }, { "epoch": 1.48, "learning_rate": 2.553161406886955e-05, "loss": 1.7535, "step": 4810 }, { "epoch": 1.48, "learning_rate": 2.5451076005276197e-05, "loss": 1.6998, "step": 4820 }, { "epoch": 1.49, "learning_rate": 2.537053325851348e-05, "loss": 1.7722, "step": 4830 }, { "epoch": 1.49, "learning_rate": 2.5289986664793743e-05, "loss": 1.7413, "step": 4840 }, { "epoch": 1.49, "learning_rate": 2.520943706036927e-05, "loss": 1.7309, "step": 4850 }, { "epoch": 1.5, "learning_rate": 2.5128885281523606e-05, "loss": 1.761, "step": 4860 }, { "epoch": 1.5, "learning_rate": 2.5048332164562872e-05, "loss": 1.671, "step": 4870 }, { "epoch": 1.5, "learning_rate": 2.4967778545807074e-05, "loss": 1.7386, "step": 4880 }, { "epoch": 1.5, "learning_rate": 2.4887225261581436e-05, "loss": 1.808, "step": 4890 }, { "epoch": 1.51, "learning_rate": 2.4806673148207693e-05, "loss": 1.8183, "step": 4900 }, { "epoch": 1.51, "learning_rate": 2.4726123041995463e-05, "loss": 1.7582, "step": 4910 }, { "epoch": 1.51, "learning_rate": 2.4645575779233464e-05, "loss": 1.7412, "step": 4920 }, { "epoch": 1.52, "learning_rate": 2.4565032196180952e-05, "loss": 1.8324, "step": 4930 }, { "epoch": 1.52, "learning_rate": 2.4484493129058944e-05, "loss": 1.7412, "step": 4940 }, { "epoch": 1.52, "learning_rate": 2.4403959414041583e-05, "loss": 1.7282, "step": 4950 }, { "epoch": 1.53, "learning_rate": 2.4323431887247446e-05, "loss": 1.7489, "step": 4960 }, { "epoch": 1.53, "learning_rate": 2.424291138473085e-05, "loss": 1.7148, "step": 4970 }, { "epoch": 1.53, "learning_rate": 2.4162398742473214e-05, "loss": 1.7889, "step": 4980 }, { "epoch": 1.54, "learning_rate": 2.4089944775835744e-05, "loss": 1.703, "step": 4990 }, { "epoch": 1.54, "learning_rate": 2.4009449370899602e-05, "loss": 1.8049, "step": 5000 }, { "epoch": 1.54, "learning_rate": 2.39289642500759e-05, "loss": 1.7172, "step": 5010 }, { "epoch": 1.54, "learning_rate": 2.384849024897869e-05, "loss": 1.7325, "step": 5020 }, { "epoch": 1.55, "learning_rate": 2.3768028203106612e-05, "loss": 1.7595, "step": 5030 }, { "epoch": 1.55, "learning_rate": 2.3687578947834123e-05, "loss": 1.8591, "step": 5040 }, { "epoch": 1.55, "learning_rate": 2.3607143318402945e-05, "loss": 1.7599, "step": 5050 }, { "epoch": 1.56, "learning_rate": 2.3526722149913288e-05, "loss": 1.6924, "step": 5060 }, { "epoch": 1.56, "learning_rate": 2.3446316277315256e-05, "loss": 1.7625, "step": 5070 }, { "epoch": 1.56, "learning_rate": 2.3365926535400135e-05, "loss": 1.6716, "step": 5080 }, { "epoch": 1.57, "learning_rate": 2.328555375879173e-05, "loss": 1.8215, "step": 5090 }, { "epoch": 1.57, "learning_rate": 2.3205198781937727e-05, "loss": 1.7256, "step": 5100 }, { "epoch": 1.57, "learning_rate": 2.3124862439101015e-05, "loss": 1.8048, "step": 5110 }, { "epoch": 1.58, "learning_rate": 2.3044545564350984e-05, "loss": 1.7619, "step": 5120 }, { "epoch": 1.58, "learning_rate": 2.2964248991554936e-05, "loss": 1.7051, "step": 5130 }, { "epoch": 1.58, "learning_rate": 2.2883973554369392e-05, "loss": 1.7944, "step": 5140 }, { "epoch": 1.58, "learning_rate": 2.280372008623142e-05, "loss": 1.747, "step": 5150 }, { "epoch": 1.59, "learning_rate": 2.272348942035002e-05, "loss": 1.7031, "step": 5160 }, { "epoch": 1.59, "learning_rate": 2.2643282389697438e-05, "loss": 1.7668, "step": 5170 }, { "epoch": 1.59, "learning_rate": 2.2563099827000535e-05, "loss": 1.7229, "step": 5180 }, { "epoch": 1.6, "learning_rate": 2.248294256473217e-05, "loss": 1.766, "step": 5190 }, { "epoch": 1.6, "learning_rate": 2.240281143510247e-05, "loss": 1.6844, "step": 5200 }, { "epoch": 1.6, "learning_rate": 2.2322707270050307e-05, "loss": 1.7874, "step": 5210 }, { "epoch": 1.61, "learning_rate": 2.2242630901234567e-05, "loss": 1.7173, "step": 5220 }, { "epoch": 1.61, "learning_rate": 2.2162583160025554e-05, "loss": 1.781, "step": 5230 }, { "epoch": 1.61, "learning_rate": 2.2082564877496368e-05, "loss": 1.6897, "step": 5240 }, { "epoch": 1.62, "learning_rate": 2.200257688441423e-05, "loss": 1.7488, "step": 5250 }, { "epoch": 1.62, "learning_rate": 2.192262001123193e-05, "loss": 1.7322, "step": 5260 }, { "epoch": 1.62, "learning_rate": 2.1842695088079144e-05, "loss": 1.693, "step": 5270 }, { "epoch": 1.62, "learning_rate": 2.176280294475383e-05, "loss": 1.7459, "step": 5280 }, { "epoch": 1.63, "learning_rate": 2.1682944410713624e-05, "loss": 1.7782, "step": 5290 }, { "epoch": 1.63, "learning_rate": 2.1603120315067245e-05, "loss": 1.7702, "step": 5300 }, { "epoch": 1.63, "learning_rate": 2.152333148656582e-05, "loss": 1.7096, "step": 5310 }, { "epoch": 1.64, "learning_rate": 2.1443578753594363e-05, "loss": 1.7202, "step": 5320 }, { "epoch": 1.64, "learning_rate": 2.1363862944163127e-05, "loss": 1.8285, "step": 5330 }, { "epoch": 1.64, "learning_rate": 2.1284184885898995e-05, "loss": 1.7059, "step": 5340 }, { "epoch": 1.65, "learning_rate": 2.1204545406036956e-05, "loss": 1.7296, "step": 5350 }, { "epoch": 1.65, "learning_rate": 2.112494533141141e-05, "loss": 1.7854, "step": 5360 }, { "epoch": 1.65, "learning_rate": 2.1045385488447698e-05, "loss": 1.7604, "step": 5370 }, { "epoch": 1.66, "learning_rate": 2.0965866703153436e-05, "loss": 1.7031, "step": 5380 }, { "epoch": 1.66, "learning_rate": 2.0886389801109965e-05, "loss": 1.7645, "step": 5390 }, { "epoch": 1.66, "learning_rate": 2.0806955607463813e-05, "loss": 1.7921, "step": 5400 }, { "epoch": 1.66, "learning_rate": 2.0727564946918087e-05, "loss": 1.7339, "step": 5410 }, { "epoch": 1.67, "learning_rate": 2.064821864372391e-05, "loss": 1.6605, "step": 5420 }, { "epoch": 1.67, "learning_rate": 2.05689175216719e-05, "loss": 1.7728, "step": 5430 }, { "epoch": 1.67, "learning_rate": 2.0489662404083567e-05, "loss": 1.7612, "step": 5440 }, { "epoch": 1.68, "learning_rate": 2.0410454113802815e-05, "loss": 1.7345, "step": 5450 }, { "epoch": 1.68, "learning_rate": 2.0331293473187387e-05, "loss": 1.6611, "step": 5460 }, { "epoch": 1.68, "learning_rate": 2.0252181304100285e-05, "loss": 1.6373, "step": 5470 }, { "epoch": 1.69, "learning_rate": 2.0173118427901286e-05, "loss": 1.7316, "step": 5480 }, { "epoch": 1.69, "learning_rate": 2.0094105665438402e-05, "loss": 1.8334, "step": 5490 }, { "epoch": 1.69, "learning_rate": 2.0015143837039336e-05, "loss": 1.6986, "step": 5500 }, { "epoch": 1.7, "learning_rate": 1.9936233762503004e-05, "loss": 1.717, "step": 5510 }, { "epoch": 1.7, "learning_rate": 1.9857376261090996e-05, "loss": 1.7701, "step": 5520 }, { "epoch": 1.7, "learning_rate": 1.977857215151906e-05, "loss": 1.7367, "step": 5530 }, { "epoch": 1.7, "learning_rate": 1.969982225194864e-05, "loss": 1.6651, "step": 5540 }, { "epoch": 1.71, "learning_rate": 1.9621127379978338e-05, "loss": 1.6955, "step": 5550 }, { "epoch": 1.71, "learning_rate": 1.9542488352635474e-05, "loss": 1.7073, "step": 5560 }, { "epoch": 1.71, "learning_rate": 1.946390598636756e-05, "loss": 1.7672, "step": 5570 }, { "epoch": 1.72, "learning_rate": 1.938538109703384e-05, "loss": 1.7534, "step": 5580 }, { "epoch": 1.72, "learning_rate": 1.9306914499896814e-05, "loss": 1.7723, "step": 5590 }, { "epoch": 1.72, "learning_rate": 1.9228507009613817e-05, "loss": 1.7417, "step": 5600 }, { "epoch": 1.73, "learning_rate": 1.915015944022846e-05, "loss": 1.8151, "step": 5610 }, { "epoch": 1.73, "learning_rate": 1.9071872605162302e-05, "loss": 1.7074, "step": 5620 }, { "epoch": 1.73, "learning_rate": 1.899364731720632e-05, "loss": 1.7825, "step": 5630 }, { "epoch": 1.74, "learning_rate": 1.8915484388512483e-05, "loss": 1.7693, "step": 5640 }, { "epoch": 1.74, "learning_rate": 1.8837384630585363e-05, "loss": 1.6949, "step": 5650 }, { "epoch": 1.74, "learning_rate": 1.8759348854273642e-05, "loss": 1.7651, "step": 5660 }, { "epoch": 1.74, "learning_rate": 1.868137786976177e-05, "loss": 1.6945, "step": 5670 }, { "epoch": 1.75, "learning_rate": 1.8603472486561487e-05, "loss": 1.7426, "step": 5680 }, { "epoch": 1.75, "learning_rate": 1.8525633513503453e-05, "loss": 1.7186, "step": 5690 }, { "epoch": 1.75, "learning_rate": 1.8447861758728848e-05, "loss": 1.6421, "step": 5700 }, { "epoch": 1.76, "learning_rate": 1.8370158029680977e-05, "loss": 1.747, "step": 5710 }, { "epoch": 1.76, "learning_rate": 1.829252313309688e-05, "loss": 1.8357, "step": 5720 }, { "epoch": 1.76, "learning_rate": 1.8214957874998973e-05, "loss": 1.7161, "step": 5730 }, { "epoch": 1.77, "learning_rate": 1.8137463060686657e-05, "loss": 1.6681, "step": 5740 }, { "epoch": 1.77, "learning_rate": 1.8060039494727977e-05, "loss": 1.8062, "step": 5750 }, { "epoch": 1.77, "learning_rate": 1.798268798095128e-05, "loss": 1.7861, "step": 5760 }, { "epoch": 1.78, "learning_rate": 1.7905409322436804e-05, "loss": 1.7414, "step": 5770 }, { "epoch": 1.78, "learning_rate": 1.7828204321508446e-05, "loss": 1.7865, "step": 5780 }, { "epoch": 1.78, "learning_rate": 1.775107377972534e-05, "loss": 1.7454, "step": 5790 }, { "epoch": 1.78, "learning_rate": 1.767401849787357e-05, "loss": 1.6788, "step": 5800 }, { "epoch": 1.79, "learning_rate": 1.7597039275957853e-05, "loss": 1.7966, "step": 5810 }, { "epoch": 1.79, "learning_rate": 1.7520136913193265e-05, "loss": 1.7801, "step": 5820 }, { "epoch": 1.79, "learning_rate": 1.7443312207996864e-05, "loss": 1.7609, "step": 5830 }, { "epoch": 1.8, "learning_rate": 1.7366565957979503e-05, "loss": 1.6759, "step": 5840 }, { "epoch": 1.8, "learning_rate": 1.728989895993745e-05, "loss": 1.7015, "step": 5850 }, { "epoch": 1.8, "learning_rate": 1.7213312009844186e-05, "loss": 1.8142, "step": 5860 }, { "epoch": 1.81, "learning_rate": 1.7136805902842127e-05, "loss": 1.7001, "step": 5870 }, { "epoch": 1.81, "learning_rate": 1.7060381433234325e-05, "loss": 1.6603, "step": 5880 }, { "epoch": 1.81, "learning_rate": 1.6984039394476287e-05, "loss": 1.6609, "step": 5890 }, { "epoch": 1.82, "learning_rate": 1.6907780579167688e-05, "loss": 1.7669, "step": 5900 }, { "epoch": 1.82, "learning_rate": 1.6831605779044157e-05, "loss": 1.7942, "step": 5910 }, { "epoch": 1.82, "learning_rate": 1.675551578496907e-05, "loss": 1.7485, "step": 5920 }, { "epoch": 1.82, "learning_rate": 1.6679511386925337e-05, "loss": 1.8651, "step": 5930 }, { "epoch": 1.83, "learning_rate": 1.6603593374007153e-05, "loss": 1.7195, "step": 5940 }, { "epoch": 1.83, "learning_rate": 1.6527762534411888e-05, "loss": 1.8096, "step": 5950 }, { "epoch": 1.83, "learning_rate": 1.6452019655431828e-05, "loss": 1.7542, "step": 5960 }, { "epoch": 1.84, "learning_rate": 1.637636552344604e-05, "loss": 1.6707, "step": 5970 }, { "epoch": 1.84, "learning_rate": 1.6300800923912224e-05, "loss": 1.7567, "step": 5980 }, { "epoch": 1.84, "learning_rate": 1.62253266413585e-05, "loss": 1.6797, "step": 5990 }, { "epoch": 1.85, "learning_rate": 1.6149943459375312e-05, "loss": 1.771, "step": 6000 }, { "epoch": 1.85, "learning_rate": 1.6074652160607302e-05, "loss": 1.7342, "step": 6010 }, { "epoch": 1.85, "learning_rate": 1.5999453526745104e-05, "loss": 1.7548, "step": 6020 }, { "epoch": 1.86, "learning_rate": 1.592434833851734e-05, "loss": 1.7672, "step": 6030 }, { "epoch": 1.86, "learning_rate": 1.5849337375682435e-05, "loss": 1.7551, "step": 6040 }, { "epoch": 1.86, "learning_rate": 1.577442141702054e-05, "loss": 1.7222, "step": 6050 }, { "epoch": 1.86, "learning_rate": 1.5699601240325474e-05, "loss": 1.7753, "step": 6060 }, { "epoch": 1.87, "learning_rate": 1.5624877622396588e-05, "loss": 1.6938, "step": 6070 }, { "epoch": 1.87, "learning_rate": 1.5550251339030783e-05, "loss": 1.8221, "step": 6080 }, { "epoch": 1.87, "learning_rate": 1.5475723165014393e-05, "loss": 1.6762, "step": 6090 }, { "epoch": 1.88, "learning_rate": 1.5401293874115147e-05, "loss": 1.7778, "step": 6100 }, { "epoch": 1.88, "learning_rate": 1.532696423907416e-05, "loss": 1.691, "step": 6110 }, { "epoch": 1.88, "learning_rate": 1.5252735031597915e-05, "loss": 1.7743, "step": 6120 }, { "epoch": 1.89, "learning_rate": 1.5178607022350186e-05, "loss": 1.7326, "step": 6130 }, { "epoch": 1.89, "learning_rate": 1.5104580980944141e-05, "loss": 1.7494, "step": 6140 }, { "epoch": 1.89, "learning_rate": 1.5030657675934256e-05, "loss": 1.7394, "step": 6150 }, { "epoch": 1.9, "learning_rate": 1.4956837874808391e-05, "loss": 1.6943, "step": 6160 }, { "epoch": 1.9, "learning_rate": 1.4883122343979822e-05, "loss": 1.7178, "step": 6170 }, { "epoch": 1.9, "learning_rate": 1.4809511848779217e-05, "loss": 1.7173, "step": 6180 }, { "epoch": 1.9, "learning_rate": 1.4736007153446801e-05, "loss": 1.7598, "step": 6190 }, { "epoch": 1.91, "learning_rate": 1.466260902112433e-05, "loss": 1.6631, "step": 6200 }, { "epoch": 1.91, "learning_rate": 1.4589318213847197e-05, "loss": 1.6904, "step": 6210 }, { "epoch": 1.91, "learning_rate": 1.4516135492536539e-05, "loss": 1.7382, "step": 6220 }, { "epoch": 1.92, "learning_rate": 1.444306161699131e-05, "loss": 1.7938, "step": 6230 }, { "epoch": 1.92, "learning_rate": 1.4370097345880407e-05, "loss": 1.687, "step": 6240 }, { "epoch": 1.92, "learning_rate": 1.4297243436734797e-05, "loss": 1.7148, "step": 6250 }, { "epoch": 1.93, "learning_rate": 1.422450064593961e-05, "loss": 1.7315, "step": 6260 }, { "epoch": 1.93, "learning_rate": 1.4151869728726378e-05, "loss": 1.7163, "step": 6270 }, { "epoch": 1.93, "learning_rate": 1.4079351439165106e-05, "loss": 1.6939, "step": 6280 }, { "epoch": 1.94, "learning_rate": 1.4006946530156462e-05, "loss": 1.7193, "step": 6290 }, { "epoch": 1.94, "learning_rate": 1.3934655753423976e-05, "loss": 1.7822, "step": 6300 }, { "epoch": 1.94, "learning_rate": 1.386247985950628e-05, "loss": 1.7014, "step": 6310 }, { "epoch": 1.94, "learning_rate": 1.3790419597749199e-05, "loss": 1.6475, "step": 6320 }, { "epoch": 1.95, "learning_rate": 1.3718475716298073e-05, "loss": 1.773, "step": 6330 }, { "epoch": 1.95, "learning_rate": 1.3646648962089965e-05, "loss": 1.7656, "step": 6340 }, { "epoch": 1.95, "learning_rate": 1.3574940080845875e-05, "loss": 1.697, "step": 6350 }, { "epoch": 1.96, "learning_rate": 1.3503349817063047e-05, "loss": 1.7529, "step": 6360 }, { "epoch": 1.96, "learning_rate": 1.3431878914007167e-05, "loss": 1.685, "step": 6370 }, { "epoch": 1.96, "learning_rate": 1.3360528113704751e-05, "loss": 1.7564, "step": 6380 }, { "epoch": 1.97, "learning_rate": 1.3289298156935348e-05, "loss": 1.7169, "step": 6390 }, { "epoch": 1.97, "learning_rate": 1.321818978322387e-05, "loss": 1.7092, "step": 6400 }, { "epoch": 1.97, "learning_rate": 1.3147203730832963e-05, "loss": 1.72, "step": 6410 }, { "epoch": 1.98, "learning_rate": 1.3076340736755293e-05, "loss": 1.815, "step": 6420 }, { "epoch": 1.98, "learning_rate": 1.3005601536705889e-05, "loss": 1.7852, "step": 6430 }, { "epoch": 1.98, "learning_rate": 1.293498686511454e-05, "loss": 1.7391, "step": 6440 }, { "epoch": 1.98, "learning_rate": 1.2864497455118152e-05, "loss": 1.7131, "step": 6450 }, { "epoch": 1.99, "learning_rate": 1.2794134038553141e-05, "loss": 1.7932, "step": 6460 }, { "epoch": 1.99, "learning_rate": 1.2723897345947828e-05, "loss": 1.73, "step": 6470 }, { "epoch": 1.99, "learning_rate": 1.2653788106514852e-05, "loss": 1.7414, "step": 6480 }, { "epoch": 2.0, "learning_rate": 1.2583807048143617e-05, "loss": 1.7401, "step": 6490 }, { "epoch": 2.0, "learning_rate": 1.2513954897392727e-05, "loss": 1.7765, "step": 6500 }, { "epoch": 2.0, "learning_rate": 1.2444232379482398e-05, "loss": 1.6992, "step": 6510 }, { "epoch": 2.01, "learning_rate": 1.237464021828704e-05, "loss": 1.7785, "step": 6520 }, { "epoch": 2.01, "learning_rate": 1.2305179136327608e-05, "loss": 1.7377, "step": 6530 }, { "epoch": 2.01, "learning_rate": 1.2235849854764194e-05, "loss": 1.7176, "step": 6540 }, { "epoch": 2.02, "learning_rate": 1.2166653093388506e-05, "loss": 1.6861, "step": 6550 }, { "epoch": 2.02, "learning_rate": 1.2097589570616394e-05, "loss": 1.7549, "step": 6560 }, { "epoch": 2.02, "learning_rate": 1.2028660003480399e-05, "loss": 1.7556, "step": 6570 }, { "epoch": 2.02, "learning_rate": 1.1959865107622307e-05, "loss": 1.6384, "step": 6580 }, { "epoch": 2.03, "learning_rate": 1.1891205597285712e-05, "loss": 1.6467, "step": 6590 }, { "epoch": 2.03, "learning_rate": 1.1822682185308612e-05, "loss": 1.7592, "step": 6600 }, { "epoch": 2.03, "learning_rate": 1.1754295583116004e-05, "loss": 1.6615, "step": 6610 }, { "epoch": 2.04, "learning_rate": 1.168604650071247e-05, "loss": 1.6877, "step": 6620 }, { "epoch": 2.04, "learning_rate": 1.1617935646674885e-05, "loss": 1.6958, "step": 6630 }, { "epoch": 2.04, "learning_rate": 1.154996372814495e-05, "loss": 1.7402, "step": 6640 }, { "epoch": 2.05, "learning_rate": 1.1482131450821937e-05, "loss": 1.6788, "step": 6650 }, { "epoch": 2.05, "learning_rate": 1.1414439518955334e-05, "loss": 1.7162, "step": 6660 }, { "epoch": 2.05, "learning_rate": 1.1346888635337522e-05, "loss": 1.7055, "step": 6670 }, { "epoch": 2.06, "learning_rate": 1.1279479501296492e-05, "loss": 1.7428, "step": 6680 }, { "epoch": 2.06, "learning_rate": 1.1212212816688558e-05, "loss": 1.694, "step": 6690 }, { "epoch": 2.06, "learning_rate": 1.1145089279891102e-05, "loss": 1.7136, "step": 6700 }, { "epoch": 2.06, "learning_rate": 1.107810958779531e-05, "loss": 1.7091, "step": 6710 }, { "epoch": 2.07, "learning_rate": 1.101127443579891e-05, "loss": 1.6638, "step": 6720 }, { "epoch": 2.07, "learning_rate": 1.0944584517799045e-05, "loss": 1.7634, "step": 6730 }, { "epoch": 2.07, "learning_rate": 1.0878040526184965e-05, "loss": 1.684, "step": 6740 }, { "epoch": 2.08, "learning_rate": 1.081164315183088e-05, "loss": 1.7748, "step": 6750 }, { "epoch": 2.08, "learning_rate": 1.0745393084088789e-05, "loss": 1.674, "step": 6760 }, { "epoch": 2.08, "learning_rate": 1.0679291010781362e-05, "loss": 1.725, "step": 6770 }, { "epoch": 2.09, "learning_rate": 1.0613337618194691e-05, "loss": 1.7543, "step": 6780 }, { "epoch": 2.09, "learning_rate": 1.0547533591071285e-05, "loss": 1.7589, "step": 6790 }, { "epoch": 2.09, "learning_rate": 1.0481879612602882e-05, "loss": 1.6569, "step": 6800 }, { "epoch": 2.1, "learning_rate": 1.0416376364423396e-05, "loss": 1.7447, "step": 6810 }, { "epoch": 2.1, "learning_rate": 1.035102452660183e-05, "loss": 1.6936, "step": 6820 }, { "epoch": 2.1, "learning_rate": 1.0285824777635172e-05, "loss": 1.7011, "step": 6830 }, { "epoch": 2.1, "learning_rate": 1.022077779444145e-05, "loss": 1.7522, "step": 6840 }, { "epoch": 2.11, "learning_rate": 1.0155884252352616e-05, "loss": 1.6488, "step": 6850 }, { "epoch": 2.11, "learning_rate": 1.009114482510754e-05, "loss": 1.7552, "step": 6860 }, { "epoch": 2.11, "learning_rate": 1.0026560184845066e-05, "loss": 1.6861, "step": 6870 }, { "epoch": 2.12, "learning_rate": 9.962131002097022e-06, "loss": 1.7417, "step": 6880 }, { "epoch": 2.12, "learning_rate": 9.897857945781196e-06, "loss": 1.741, "step": 6890 }, { "epoch": 2.12, "learning_rate": 9.833741683194475e-06, "loss": 1.7206, "step": 6900 }, { "epoch": 2.13, "learning_rate": 9.76978288000586e-06, "loss": 1.7312, "step": 6910 }, { "epoch": 2.13, "learning_rate": 9.70598220024958e-06, "loss": 1.7613, "step": 6920 }, { "epoch": 2.13, "learning_rate": 9.642340306318203e-06, "loss": 1.6891, "step": 6930 }, { "epoch": 2.14, "learning_rate": 9.578857858955715e-06, "loss": 1.7404, "step": 6940 }, { "epoch": 2.14, "learning_rate": 9.515535517250737e-06, "loss": 1.7611, "step": 6950 }, { "epoch": 2.14, "learning_rate": 9.452373938629619e-06, "loss": 1.6799, "step": 6960 }, { "epoch": 2.14, "learning_rate": 9.389373778849612e-06, "loss": 1.7458, "step": 6970 }, { "epoch": 2.15, "learning_rate": 9.3265356919921e-06, "loss": 1.6968, "step": 6980 }, { "epoch": 2.15, "learning_rate": 9.26386033045582e-06, "loss": 1.768, "step": 6990 }, { "epoch": 2.15, "learning_rate": 9.201348344950001e-06, "loss": 1.7553, "step": 7000 }, { "epoch": 2.16, "learning_rate": 9.13900038448771e-06, "loss": 1.7224, "step": 7010 }, { "epoch": 2.16, "learning_rate": 9.07681709637905e-06, "loss": 1.7873, "step": 7020 }, { "epoch": 2.16, "learning_rate": 9.014799126224471e-06, "loss": 1.7164, "step": 7030 }, { "epoch": 2.17, "learning_rate": 8.952947117908047e-06, "loss": 1.6788, "step": 7040 }, { "epoch": 2.17, "learning_rate": 8.891261713590807e-06, "loss": 1.7079, "step": 7050 }, { "epoch": 2.17, "learning_rate": 8.829743553704056e-06, "loss": 1.7201, "step": 7060 }, { "epoch": 2.18, "learning_rate": 8.768393276942743e-06, "loss": 1.711, "step": 7070 }, { "epoch": 2.18, "learning_rate": 8.707211520258782e-06, "loss": 1.7318, "step": 7080 }, { "epoch": 2.18, "learning_rate": 8.646198918854526e-06, "loss": 1.7381, "step": 7090 }, { "epoch": 2.18, "learning_rate": 8.585356106176094e-06, "loss": 1.7835, "step": 7100 }, { "epoch": 2.19, "learning_rate": 8.524683713906805e-06, "loss": 1.6947, "step": 7110 }, { "epoch": 2.19, "learning_rate": 8.464182371960668e-06, "loss": 1.7613, "step": 7120 }, { "epoch": 2.19, "learning_rate": 8.403852708475792e-06, "loss": 1.7626, "step": 7130 }, { "epoch": 2.2, "learning_rate": 8.34369534980789e-06, "loss": 1.69, "step": 7140 }, { "epoch": 2.2, "learning_rate": 8.283710920523763e-06, "loss": 1.6924, "step": 7150 }, { "epoch": 2.2, "learning_rate": 8.223900043394825e-06, "loss": 1.7192, "step": 7160 }, { "epoch": 2.21, "learning_rate": 8.164263339390635e-06, "loss": 1.6862, "step": 7170 }, { "epoch": 2.21, "learning_rate": 8.104801427672456e-06, "loss": 1.661, "step": 7180 }, { "epoch": 2.21, "learning_rate": 8.045514925586784e-06, "loss": 1.7663, "step": 7190 }, { "epoch": 2.22, "learning_rate": 7.986404448659023e-06, "loss": 1.778, "step": 7200 }, { "epoch": 2.22, "learning_rate": 7.927470610587028e-06, "loss": 1.7301, "step": 7210 }, { "epoch": 2.22, "learning_rate": 7.868714023234727e-06, "loss": 1.6498, "step": 7220 }, { "epoch": 2.22, "learning_rate": 7.810135296625818e-06, "loss": 1.6531, "step": 7230 }, { "epoch": 2.23, "learning_rate": 7.751735038937405e-06, "loss": 1.7291, "step": 7240 }, { "epoch": 2.23, "learning_rate": 7.693513856493684e-06, "loss": 1.739, "step": 7250 }, { "epoch": 2.23, "learning_rate": 7.63547235375966e-06, "loss": 1.8113, "step": 7260 }, { "epoch": 2.24, "learning_rate": 7.577611133334858e-06, "loss": 1.763, "step": 7270 }, { "epoch": 2.24, "learning_rate": 7.519930795947072e-06, "loss": 1.7555, "step": 7280 }, { "epoch": 2.24, "learning_rate": 7.462431940446135e-06, "loss": 1.6683, "step": 7290 }, { "epoch": 2.25, "learning_rate": 7.40511516379769e-06, "loss": 1.6882, "step": 7300 }, { "epoch": 2.25, "learning_rate": 7.347981061077011e-06, "loss": 1.623, "step": 7310 }, { "epoch": 2.25, "learning_rate": 7.291030225462781e-06, "loss": 1.7501, "step": 7320 }, { "epoch": 2.26, "learning_rate": 7.2342632482309825e-06, "loss": 1.7275, "step": 7330 }, { "epoch": 2.26, "learning_rate": 7.177680718748767e-06, "loss": 1.6834, "step": 7340 }, { "epoch": 2.26, "learning_rate": 7.1212832244682585e-06, "loss": 1.7849, "step": 7350 }, { "epoch": 2.26, "learning_rate": 7.065071350920538e-06, "loss": 1.7107, "step": 7360 }, { "epoch": 2.27, "learning_rate": 7.009045681709522e-06, "loss": 1.727, "step": 7370 }, { "epoch": 2.27, "learning_rate": 6.953206798505918e-06, "loss": 1.611, "step": 7380 }, { "epoch": 2.27, "learning_rate": 6.8975552810411765e-06, "loss": 1.7685, "step": 7390 }, { "epoch": 2.28, "learning_rate": 6.842091707101473e-06, "loss": 1.8104, "step": 7400 }, { "epoch": 2.28, "learning_rate": 6.786816652521719e-06, "loss": 1.7068, "step": 7410 }, { "epoch": 2.28, "learning_rate": 6.73173069117958e-06, "loss": 1.7325, "step": 7420 }, { "epoch": 2.29, "learning_rate": 6.676834394989495e-06, "loss": 1.6711, "step": 7430 }, { "epoch": 2.29, "learning_rate": 6.622128333896768e-06, "loss": 1.6467, "step": 7440 }, { "epoch": 2.29, "learning_rate": 6.56761307587167e-06, "loss": 1.6948, "step": 7450 }, { "epoch": 2.3, "learning_rate": 6.513289186903463e-06, "loss": 1.6952, "step": 7460 }, { "epoch": 2.3, "learning_rate": 6.459157230994603e-06, "loss": 1.6495, "step": 7470 }, { "epoch": 2.3, "learning_rate": 6.405217770154853e-06, "loss": 1.7286, "step": 7480 }, { "epoch": 2.3, "learning_rate": 6.3514713643954475e-06, "loss": 1.7481, "step": 7490 }, { "epoch": 2.31, "learning_rate": 6.297918571723288e-06, "loss": 1.74, "step": 7500 }, { "epoch": 2.31, "learning_rate": 6.244559948135109e-06, "loss": 1.7402, "step": 7510 }, { "epoch": 2.31, "learning_rate": 6.191396047611794e-06, "loss": 1.7179, "step": 7520 }, { "epoch": 2.32, "learning_rate": 6.138427422112539e-06, "loss": 1.7242, "step": 7530 }, { "epoch": 2.32, "learning_rate": 6.085654621569137e-06, "loss": 1.7131, "step": 7540 }, { "epoch": 2.32, "learning_rate": 6.0330781938803034e-06, "loss": 1.6449, "step": 7550 }, { "epoch": 2.33, "learning_rate": 5.980698684905989e-06, "loss": 1.7099, "step": 7560 }, { "epoch": 2.33, "learning_rate": 5.928516638461639e-06, "loss": 1.7227, "step": 7570 }, { "epoch": 2.33, "learning_rate": 5.876532596312645e-06, "loss": 1.6573, "step": 7580 }, { "epoch": 2.34, "learning_rate": 5.824747098168651e-06, "loss": 1.634, "step": 7590 }, { "epoch": 2.34, "learning_rate": 5.773160681677983e-06, "loss": 1.5946, "step": 7600 }, { "epoch": 2.34, "learning_rate": 5.721773882422057e-06, "loss": 1.6724, "step": 7610 }, { "epoch": 2.34, "learning_rate": 5.6705872339098186e-06, "loss": 1.7526, "step": 7620 }, { "epoch": 2.35, "learning_rate": 5.6196012675722055e-06, "loss": 1.6617, "step": 7630 }, { "epoch": 2.35, "learning_rate": 5.568816512756633e-06, "loss": 1.7609, "step": 7640 }, { "epoch": 2.35, "learning_rate": 5.5182334967214725e-06, "loss": 1.677, "step": 7650 }, { "epoch": 2.36, "learning_rate": 5.467852744630633e-06, "loss": 1.6871, "step": 7660 }, { "epoch": 2.36, "learning_rate": 5.417674779548062e-06, "loss": 1.6679, "step": 7670 }, { "epoch": 2.36, "learning_rate": 5.367700122432315e-06, "loss": 1.7013, "step": 7680 }, { "epoch": 2.37, "learning_rate": 5.317929292131163e-06, "loss": 1.74, "step": 7690 }, { "epoch": 2.37, "learning_rate": 5.268362805376237e-06, "loss": 1.7601, "step": 7700 }, { "epoch": 2.37, "learning_rate": 5.219001176777574e-06, "loss": 1.7004, "step": 7710 }, { "epoch": 2.38, "learning_rate": 5.16984491881837e-06, "loss": 1.7618, "step": 7720 }, { "epoch": 2.38, "learning_rate": 5.120894541849599e-06, "loss": 1.7195, "step": 7730 }, { "epoch": 2.38, "learning_rate": 5.072150554084745e-06, "loss": 1.7452, "step": 7740 }, { "epoch": 2.38, "learning_rate": 5.023613461594512e-06, "loss": 1.6854, "step": 7750 }, { "epoch": 2.39, "learning_rate": 4.9752837683015505e-06, "loss": 1.7028, "step": 7760 }, { "epoch": 2.39, "learning_rate": 4.927161975975284e-06, "loss": 1.7031, "step": 7770 }, { "epoch": 2.39, "learning_rate": 4.879248584226645e-06, "loss": 1.7488, "step": 7780 }, { "epoch": 2.4, "learning_rate": 4.831544090502896e-06, "loss": 1.686, "step": 7790 }, { "epoch": 2.4, "learning_rate": 4.784048990082484e-06, "loss": 1.7036, "step": 7800 }, { "epoch": 2.4, "learning_rate": 4.736763776069897e-06, "loss": 1.6534, "step": 7810 }, { "epoch": 2.41, "learning_rate": 4.689688939390521e-06, "loss": 1.6286, "step": 7820 }, { "epoch": 2.41, "learning_rate": 4.642824968785572e-06, "loss": 1.6757, "step": 7830 }, { "epoch": 2.41, "learning_rate": 4.596172350807004e-06, "loss": 1.7313, "step": 7840 }, { "epoch": 2.42, "learning_rate": 4.549731569812457e-06, "loss": 1.7432, "step": 7850 }, { "epoch": 2.42, "learning_rate": 4.5035031079602445e-06, "loss": 1.7304, "step": 7860 }, { "epoch": 2.42, "learning_rate": 4.457487445204311e-06, "loss": 1.772, "step": 7870 }, { "epoch": 2.42, "learning_rate": 4.411685059289314e-06, "loss": 1.6358, "step": 7880 }, { "epoch": 2.43, "learning_rate": 4.366096425745597e-06, "loss": 1.7221, "step": 7890 }, { "epoch": 2.43, "learning_rate": 4.320722017884274e-06, "loss": 1.6839, "step": 7900 }, { "epoch": 2.43, "learning_rate": 4.275562306792352e-06, "loss": 1.6858, "step": 7910 }, { "epoch": 2.44, "learning_rate": 4.2306177613277765e-06, "loss": 1.7469, "step": 7920 }, { "epoch": 2.44, "learning_rate": 4.185888848114614e-06, "loss": 1.6618, "step": 7930 }, { "epoch": 2.44, "learning_rate": 4.141376031538186e-06, "loss": 1.7702, "step": 7940 }, { "epoch": 2.45, "learning_rate": 4.097079773740256e-06, "loss": 1.7706, "step": 7950 }, { "epoch": 2.45, "learning_rate": 4.053000534614218e-06, "loss": 1.7161, "step": 7960 }, { "epoch": 2.45, "learning_rate": 4.0091387718003415e-06, "loss": 1.6853, "step": 7970 }, { "epoch": 2.46, "learning_rate": 3.9654949406809995e-06, "loss": 1.6551, "step": 7980 }, { "epoch": 2.46, "learning_rate": 3.922069494375963e-06, "loss": 1.7098, "step": 7990 }, { "epoch": 2.46, "learning_rate": 3.878862883737666e-06, "loss": 1.7272, "step": 8000 } ], "logging_steps": 10, "max_steps": 9750, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 1000, "total_flos": 2.735289189695226e+18, "train_batch_size": 4, "trial_name": null, "trial_params": null }