{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.923594785210515, "global_step": 6000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 2.6666666666666667e-08, "loss": 0.6993, "step": 10 }, { "epoch": 0.01, "learning_rate": 5.3333333333333334e-08, "loss": 0.6971, "step": 20 }, { "epoch": 0.01, "learning_rate": 8.666666666666666e-08, "loss": 0.7043, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.2e-07, "loss": 0.6975, "step": 40 }, { "epoch": 0.02, "learning_rate": 1.533333333333333e-07, "loss": 0.699, "step": 50 }, { "epoch": 0.02, "learning_rate": 1.8666666666666667e-07, "loss": 0.7028, "step": 60 }, { "epoch": 0.02, "learning_rate": 2.1666666666666667e-07, "loss": 0.7, "step": 70 }, { "epoch": 0.03, "learning_rate": 2.5e-07, "loss": 0.6964, "step": 80 }, { "epoch": 0.03, "learning_rate": 2.833333333333333e-07, "loss": 0.7013, "step": 90 }, { "epoch": 0.03, "learning_rate": 3.166666666666666e-07, "loss": 0.699, "step": 100 }, { "epoch": 0.04, "learning_rate": 3.4666666666666665e-07, "loss": 0.7008, "step": 110 }, { "epoch": 0.04, "learning_rate": 3.7999999999999996e-07, "loss": 0.6943, "step": 120 }, { "epoch": 0.04, "learning_rate": 4.1333333333333333e-07, "loss": 0.7018, "step": 130 }, { "epoch": 0.04, "learning_rate": 4.4666666666666664e-07, "loss": 0.6988, "step": 140 }, { "epoch": 0.05, "learning_rate": 4.8e-07, "loss": 0.6998, "step": 150 }, { "epoch": 0.05, "learning_rate": 5.133333333333333e-07, "loss": 0.696, "step": 160 }, { "epoch": 0.05, "learning_rate": 5.466666666666666e-07, "loss": 0.6939, "step": 170 }, { "epoch": 0.06, "learning_rate": 5.766666666666666e-07, "loss": 0.6959, "step": 180 }, { "epoch": 0.06, "learning_rate": 6.1e-07, "loss": 0.7006, "step": 190 }, { "epoch": 0.06, "learning_rate": 6.433333333333332e-07, "loss": 0.6958, "step": 200 }, { "epoch": 0.07, "learning_rate": 6.766666666666666e-07, "loss": 0.6957, "step": 210 }, { "epoch": 0.07, "learning_rate": 7.1e-07, "loss": 0.6925, "step": 220 }, { "epoch": 0.07, "learning_rate": 7.433333333333332e-07, "loss": 0.6927, "step": 230 }, { "epoch": 0.08, "learning_rate": 7.766666666666666e-07, "loss": 0.697, "step": 240 }, { "epoch": 0.08, "learning_rate": 8.1e-07, "loss": 0.692, "step": 250 }, { "epoch": 0.08, "learning_rate": 8.433333333333333e-07, "loss": 0.6998, "step": 260 }, { "epoch": 0.09, "learning_rate": 8.766666666666667e-07, "loss": 0.6905, "step": 270 }, { "epoch": 0.09, "learning_rate": 9.1e-07, "loss": 0.6925, "step": 280 }, { "epoch": 0.09, "learning_rate": 9.433333333333333e-07, "loss": 0.699, "step": 290 }, { "epoch": 0.1, "learning_rate": 9.766666666666667e-07, "loss": 0.6998, "step": 300 }, { "epoch": 0.1, "learning_rate": 1.0099999999999999e-06, "loss": 0.6948, "step": 310 }, { "epoch": 0.1, "learning_rate": 1.0433333333333332e-06, "loss": 0.7013, "step": 320 }, { "epoch": 0.11, "learning_rate": 1.0766666666666666e-06, "loss": 0.6939, "step": 330 }, { "epoch": 0.11, "learning_rate": 1.11e-06, "loss": 0.6937, "step": 340 }, { "epoch": 0.11, "learning_rate": 1.1433333333333333e-06, "loss": 0.6921, "step": 350 }, { "epoch": 0.12, "learning_rate": 1.1766666666666667e-06, "loss": 0.6917, "step": 360 }, { "epoch": 0.12, "learning_rate": 1.2099999999999998e-06, "loss": 0.6898, "step": 370 }, { "epoch": 0.12, "learning_rate": 1.2433333333333334e-06, "loss": 0.6929, "step": 380 }, { "epoch": 0.13, "learning_rate": 1.2766666666666666e-06, "loss": 0.6988, "step": 390 }, { "epoch": 0.13, "learning_rate": 1.31e-06, "loss": 0.6946, "step": 400 }, { "epoch": 0.13, "learning_rate": 1.3433333333333331e-06, "loss": 0.6919, "step": 410 }, { "epoch": 0.13, "learning_rate": 1.3766666666666667e-06, "loss": 0.6921, "step": 420 }, { "epoch": 0.14, "learning_rate": 1.4099999999999998e-06, "loss": 0.6888, "step": 430 }, { "epoch": 0.14, "learning_rate": 1.4433333333333332e-06, "loss": 0.6912, "step": 440 }, { "epoch": 0.14, "learning_rate": 1.4766666666666666e-06, "loss": 0.6996, "step": 450 }, { "epoch": 0.15, "learning_rate": 1.51e-06, "loss": 0.6878, "step": 460 }, { "epoch": 0.15, "learning_rate": 1.543333333333333e-06, "loss": 0.6894, "step": 470 }, { "epoch": 0.15, "learning_rate": 1.5766666666666667e-06, "loss": 0.6808, "step": 480 }, { "epoch": 0.16, "learning_rate": 1.6066666666666666e-06, "loss": 0.6859, "step": 490 }, { "epoch": 0.16, "learning_rate": 1.6399999999999998e-06, "loss": 0.6766, "step": 500 }, { "epoch": 0.16, "learning_rate": 1.6733333333333333e-06, "loss": 0.6733, "step": 510 }, { "epoch": 0.17, "learning_rate": 1.7066666666666667e-06, "loss": 0.669, "step": 520 }, { "epoch": 0.17, "learning_rate": 1.7399999999999999e-06, "loss": 0.6586, "step": 530 }, { "epoch": 0.17, "learning_rate": 1.7733333333333334e-06, "loss": 0.6582, "step": 540 }, { "epoch": 0.18, "learning_rate": 1.8066666666666666e-06, "loss": 0.6755, "step": 550 }, { "epoch": 0.18, "learning_rate": 1.84e-06, "loss": 0.6702, "step": 560 }, { "epoch": 0.18, "learning_rate": 1.8733333333333331e-06, "loss": 0.6716, "step": 570 }, { "epoch": 0.19, "learning_rate": 1.9066666666666667e-06, "loss": 0.6565, "step": 580 }, { "epoch": 0.19, "learning_rate": 1.94e-06, "loss": 0.6626, "step": 590 }, { "epoch": 0.19, "learning_rate": 1.9733333333333332e-06, "loss": 0.6413, "step": 600 }, { "epoch": 0.2, "learning_rate": 1.999999379017226e-06, "loss": 0.6472, "step": 610 }, { "epoch": 0.2, "learning_rate": 1.9999776447011173e-06, "loss": 0.6428, "step": 620 }, { "epoch": 0.2, "learning_rate": 1.9999248620175453e-06, "loss": 0.66, "step": 630 }, { "epoch": 0.21, "learning_rate": 1.9998508125827426e-06, "loss": 0.6937, "step": 640 }, { "epoch": 0.21, "learning_rate": 1.999739043312754e-06, "loss": 0.6568, "step": 650 }, { "epoch": 0.21, "learning_rate": 1.9995962330836514e-06, "loss": 0.6446, "step": 660 }, { "epoch": 0.21, "learning_rate": 1.9994223863295585e-06, "loss": 0.6401, "step": 670 }, { "epoch": 0.22, "learning_rate": 1.9992175084482535e-06, "loss": 0.6356, "step": 680 }, { "epoch": 0.22, "learning_rate": 1.998981605801002e-06, "loss": 0.659, "step": 690 }, { "epoch": 0.22, "learning_rate": 1.9987146857123613e-06, "loss": 0.6346, "step": 700 }, { "epoch": 0.23, "learning_rate": 1.9984167564699483e-06, "loss": 0.6383, "step": 710 }, { "epoch": 0.23, "learning_rate": 1.9980878273241866e-06, "loss": 0.6394, "step": 720 }, { "epoch": 0.23, "learning_rate": 1.997727908488018e-06, "loss": 0.6344, "step": 730 }, { "epoch": 0.24, "learning_rate": 1.997337011136584e-06, "loss": 0.6226, "step": 740 }, { "epoch": 0.24, "learning_rate": 1.9969151474068815e-06, "loss": 0.6352, "step": 750 }, { "epoch": 0.24, "learning_rate": 1.996462330397382e-06, "loss": 0.6374, "step": 760 }, { "epoch": 0.25, "learning_rate": 1.99597857416763e-06, "loss": 0.6257, "step": 770 }, { "epoch": 0.25, "learning_rate": 1.9954638937378017e-06, "loss": 0.6307, "step": 780 }, { "epoch": 0.25, "learning_rate": 1.9949183050882417e-06, "loss": 0.6394, "step": 790 }, { "epoch": 0.26, "learning_rate": 1.9943418251589657e-06, "loss": 0.6472, "step": 800 }, { "epoch": 0.26, "learning_rate": 1.9937344718491347e-06, "loss": 0.6443, "step": 810 }, { "epoch": 0.26, "learning_rate": 1.9930962640164985e-06, "loss": 0.6225, "step": 820 }, { "epoch": 0.27, "learning_rate": 1.9924272214768115e-06, "loss": 0.6337, "step": 830 }, { "epoch": 0.27, "learning_rate": 1.991727365003217e-06, "loss": 0.6136, "step": 840 }, { "epoch": 0.27, "learning_rate": 1.9909967163256016e-06, "loss": 0.6309, "step": 850 }, { "epoch": 0.28, "learning_rate": 1.9902352981299204e-06, "loss": 0.6311, "step": 860 }, { "epoch": 0.28, "learning_rate": 1.9894431340574953e-06, "loss": 0.6111, "step": 870 }, { "epoch": 0.28, "learning_rate": 1.9886202487042762e-06, "loss": 0.6334, "step": 880 }, { "epoch": 0.29, "learning_rate": 1.9877666676200812e-06, "loss": 0.6193, "step": 890 }, { "epoch": 0.29, "learning_rate": 1.9868824173078025e-06, "loss": 0.6369, "step": 900 }, { "epoch": 0.29, "learning_rate": 1.9859675252225824e-06, "loss": 0.6283, "step": 910 }, { "epoch": 0.29, "learning_rate": 1.985022019770961e-06, "loss": 0.5903, "step": 920 }, { "epoch": 0.3, "learning_rate": 1.9840459303099958e-06, "loss": 0.6118, "step": 930 }, { "epoch": 0.3, "learning_rate": 1.983039287146348e-06, "loss": 0.6102, "step": 940 }, { "epoch": 0.3, "learning_rate": 1.982002121535344e-06, "loss": 0.6058, "step": 950 }, { "epoch": 0.31, "learning_rate": 1.9809344656800026e-06, "loss": 0.6103, "step": 960 }, { "epoch": 0.31, "learning_rate": 1.9798363527300355e-06, "loss": 0.6137, "step": 970 }, { "epoch": 0.31, "learning_rate": 1.97870781678082e-06, "loss": 0.6079, "step": 980 }, { "epoch": 0.32, "learning_rate": 1.9775488928723377e-06, "loss": 0.5953, "step": 990 }, { "epoch": 0.32, "learning_rate": 1.976359616988089e-06, "loss": 0.5808, "step": 1000 }, { "epoch": 0.32, "learning_rate": 1.975140026053974e-06, "loss": 0.604, "step": 1010 }, { "epoch": 0.33, "learning_rate": 1.9738901579371454e-06, "loss": 0.6013, "step": 1020 }, { "epoch": 0.33, "learning_rate": 1.9726100514448373e-06, "loss": 0.5625, "step": 1030 }, { "epoch": 0.33, "learning_rate": 1.9712997463231535e-06, "loss": 0.5886, "step": 1040 }, { "epoch": 0.34, "learning_rate": 1.9699592832558395e-06, "loss": 0.5952, "step": 1050 }, { "epoch": 0.34, "learning_rate": 1.968588703863015e-06, "loss": 0.5767, "step": 1060 }, { "epoch": 0.34, "learning_rate": 1.967188050699884e-06, "loss": 0.5858, "step": 1070 }, { "epoch": 0.35, "learning_rate": 1.9657573672554125e-06, "loss": 0.5786, "step": 1080 }, { "epoch": 0.35, "learning_rate": 1.9642966979509797e-06, "loss": 0.5701, "step": 1090 }, { "epoch": 0.35, "learning_rate": 1.962806088138996e-06, "loss": 0.5874, "step": 1100 }, { "epoch": 0.36, "learning_rate": 1.9612855841014974e-06, "loss": 0.603, "step": 1110 }, { "epoch": 0.36, "learning_rate": 1.959735233048708e-06, "loss": 0.5846, "step": 1120 }, { "epoch": 0.36, "learning_rate": 1.9581550831175724e-06, "loss": 0.5769, "step": 1130 }, { "epoch": 0.37, "learning_rate": 1.956545183370263e-06, "loss": 0.5643, "step": 1140 }, { "epoch": 0.37, "learning_rate": 1.954905583792657e-06, "loss": 0.59, "step": 1150 }, { "epoch": 0.37, "learning_rate": 1.953236335292782e-06, "loss": 0.5629, "step": 1160 }, { "epoch": 0.38, "learning_rate": 1.9515374896992374e-06, "loss": 0.5685, "step": 1170 }, { "epoch": 0.38, "learning_rate": 1.949809099759585e-06, "loss": 0.5782, "step": 1180 }, { "epoch": 0.38, "learning_rate": 1.9480512191387103e-06, "loss": 0.578, "step": 1190 }, { "epoch": 0.38, "learning_rate": 1.946263902417157e-06, "loss": 0.5844, "step": 1200 }, { "epoch": 0.39, "learning_rate": 1.944447205089432e-06, "loss": 0.5617, "step": 1210 }, { "epoch": 0.39, "learning_rate": 1.942601183562282e-06, "loss": 0.5909, "step": 1220 }, { "epoch": 0.39, "learning_rate": 1.940725895152944e-06, "loss": 0.5597, "step": 1230 }, { "epoch": 0.4, "learning_rate": 1.938821398087363e-06, "loss": 0.5374, "step": 1240 }, { "epoch": 0.4, "learning_rate": 1.9368877514983846e-06, "loss": 0.5788, "step": 1250 }, { "epoch": 0.4, "learning_rate": 1.9349250154239227e-06, "loss": 0.5733, "step": 1260 }, { "epoch": 0.41, "learning_rate": 1.932933250805089e-06, "loss": 0.5579, "step": 1270 }, { "epoch": 0.41, "learning_rate": 1.9309125194843065e-06, "loss": 0.5536, "step": 1280 }, { "epoch": 0.41, "learning_rate": 1.9290691466029904e-06, "loss": 0.5624, "step": 1290 }, { "epoch": 0.42, "learning_rate": 1.9269935521439704e-06, "loss": 0.5594, "step": 1300 }, { "epoch": 0.42, "learning_rate": 1.924889175405071e-06, "loss": 0.6108, "step": 1310 }, { "epoch": 0.42, "learning_rate": 1.9227560817252142e-06, "loss": 0.5796, "step": 1320 }, { "epoch": 0.43, "learning_rate": 1.9205943373349583e-06, "loss": 0.5579, "step": 1330 }, { "epoch": 0.43, "learning_rate": 1.9184040093544373e-06, "loss": 0.5476, "step": 1340 }, { "epoch": 0.43, "learning_rate": 1.9161851657912802e-06, "loss": 0.551, "step": 1350 }, { "epoch": 0.44, "learning_rate": 1.9139378755384964e-06, "loss": 0.5362, "step": 1360 }, { "epoch": 0.44, "learning_rate": 1.9116622083723404e-06, "loss": 0.5626, "step": 1370 }, { "epoch": 0.44, "learning_rate": 1.909358234950141e-06, "loss": 0.5954, "step": 1380 }, { "epoch": 0.45, "learning_rate": 1.9070260268081125e-06, "loss": 0.5523, "step": 1390 }, { "epoch": 0.45, "learning_rate": 1.9046656563591275e-06, "loss": 0.5732, "step": 1400 }, { "epoch": 0.45, "learning_rate": 1.9022771968904741e-06, "loss": 0.5865, "step": 1410 }, { "epoch": 0.46, "learning_rate": 1.8998607225615772e-06, "loss": 0.5725, "step": 1420 }, { "epoch": 0.46, "learning_rate": 1.8974163084016972e-06, "loss": 0.5925, "step": 1430 }, { "epoch": 0.46, "learning_rate": 1.8949440303076001e-06, "loss": 0.5535, "step": 1440 }, { "epoch": 0.46, "learning_rate": 1.8924439650412001e-06, "loss": 0.5363, "step": 1450 }, { "epoch": 0.47, "learning_rate": 1.8899161902271778e-06, "loss": 0.5486, "step": 1460 }, { "epoch": 0.47, "learning_rate": 1.8873607843505695e-06, "loss": 0.5412, "step": 1470 }, { "epoch": 0.47, "learning_rate": 1.884777826754329e-06, "loss": 0.577, "step": 1480 }, { "epoch": 0.48, "learning_rate": 1.8821673976368662e-06, "loss": 0.5408, "step": 1490 }, { "epoch": 0.48, "learning_rate": 1.8795295780495551e-06, "loss": 0.5521, "step": 1500 }, { "epoch": 0.48, "learning_rate": 1.876864449894219e-06, "loss": 0.5774, "step": 1510 }, { "epoch": 0.49, "learning_rate": 1.8741720959205854e-06, "loss": 0.5678, "step": 1520 }, { "epoch": 0.49, "learning_rate": 1.8714525997237192e-06, "loss": 0.5533, "step": 1530 }, { "epoch": 0.49, "learning_rate": 1.8687060457414253e-06, "loss": 0.5541, "step": 1540 }, { "epoch": 0.5, "learning_rate": 1.8659325192516267e-06, "loss": 0.5594, "step": 1550 }, { "epoch": 0.5, "learning_rate": 1.8631321063697189e-06, "loss": 0.5485, "step": 1560 }, { "epoch": 0.5, "learning_rate": 1.8603048940458935e-06, "loss": 0.5777, "step": 1570 }, { "epoch": 0.51, "learning_rate": 1.8574509700624403e-06, "loss": 0.5569, "step": 1580 }, { "epoch": 0.51, "learning_rate": 1.8545704230310206e-06, "loss": 0.5611, "step": 1590 }, { "epoch": 0.51, "learning_rate": 1.8516633423899166e-06, "loss": 0.5451, "step": 1600 }, { "epoch": 0.52, "learning_rate": 1.8487298184012543e-06, "loss": 0.5906, "step": 1610 }, { "epoch": 0.52, "learning_rate": 1.8457699421482003e-06, "loss": 0.555, "step": 1620 }, { "epoch": 0.52, "learning_rate": 1.8427838055321349e-06, "loss": 0.5435, "step": 1630 }, { "epoch": 0.53, "learning_rate": 1.8397715012697976e-06, "loss": 0.5789, "step": 1640 }, { "epoch": 0.53, "learning_rate": 1.836733122890409e-06, "loss": 0.56, "step": 1650 }, { "epoch": 0.53, "learning_rate": 1.8336687647327666e-06, "loss": 0.545, "step": 1660 }, { "epoch": 0.54, "learning_rate": 1.8305785219423155e-06, "loss": 0.5888, "step": 1670 }, { "epoch": 0.54, "learning_rate": 1.827462490468194e-06, "loss": 0.5551, "step": 1680 }, { "epoch": 0.54, "learning_rate": 1.824320767060256e-06, "loss": 0.5491, "step": 1690 }, { "epoch": 0.54, "learning_rate": 1.821153449266064e-06, "loss": 0.5504, "step": 1700 }, { "epoch": 0.55, "learning_rate": 1.817960635427864e-06, "loss": 0.5411, "step": 1710 }, { "epoch": 0.55, "learning_rate": 1.8150653857736957e-06, "loss": 0.5527, "step": 1720 }, { "epoch": 0.55, "learning_rate": 1.8118244032175022e-06, "loss": 0.5753, "step": 1730 }, { "epoch": 0.56, "learning_rate": 1.8085582142754256e-06, "loss": 0.5519, "step": 1740 }, { "epoch": 0.56, "learning_rate": 1.8052669203595676e-06, "loss": 0.5601, "step": 1750 }, { "epoch": 0.56, "learning_rate": 1.8019506236615156e-06, "loss": 0.5657, "step": 1760 }, { "epoch": 0.57, "learning_rate": 1.79860942714917e-06, "loss": 0.5304, "step": 1770 }, { "epoch": 0.57, "learning_rate": 1.7952434345635467e-06, "loss": 0.5566, "step": 1780 }, { "epoch": 0.57, "learning_rate": 1.7918527504155572e-06, "loss": 0.5712, "step": 1790 }, { "epoch": 0.58, "learning_rate": 1.7884374799827624e-06, "loss": 0.541, "step": 1800 }, { "epoch": 0.58, "learning_rate": 1.7849977293061041e-06, "loss": 0.5325, "step": 1810 }, { "epoch": 0.58, "learning_rate": 1.7815336051866126e-06, "loss": 0.5566, "step": 1820 }, { "epoch": 0.59, "learning_rate": 1.7780452151820913e-06, "loss": 0.5507, "step": 1830 }, { "epoch": 0.59, "learning_rate": 1.7745326676037764e-06, "loss": 0.5756, "step": 1840 }, { "epoch": 0.59, "learning_rate": 1.7709960715129736e-06, "loss": 0.545, "step": 1850 }, { "epoch": 0.6, "learning_rate": 1.7674355367176728e-06, "loss": 0.535, "step": 1860 }, { "epoch": 0.6, "learning_rate": 1.7638511737691383e-06, "loss": 0.5461, "step": 1870 }, { "epoch": 0.6, "learning_rate": 1.7602430939584756e-06, "loss": 0.5583, "step": 1880 }, { "epoch": 0.61, "learning_rate": 1.7566114093131773e-06, "loss": 0.5313, "step": 1890 }, { "epoch": 0.61, "learning_rate": 1.7529562325936433e-06, "loss": 0.5618, "step": 1900 }, { "epoch": 0.61, "learning_rate": 1.749277677289681e-06, "loss": 0.5486, "step": 1910 }, { "epoch": 0.62, "learning_rate": 1.7455758576169797e-06, "loss": 0.5512, "step": 1920 }, { "epoch": 0.62, "learning_rate": 1.7418508885135672e-06, "loss": 0.531, "step": 1930 }, { "epoch": 0.62, "learning_rate": 1.7381028856362383e-06, "loss": 0.546, "step": 1940 }, { "epoch": 0.63, "learning_rate": 1.734331965356965e-06, "loss": 0.5604, "step": 1950 }, { "epoch": 0.63, "learning_rate": 1.7305382447592835e-06, "loss": 0.5492, "step": 1960 }, { "epoch": 0.63, "learning_rate": 1.7267218416346581e-06, "loss": 0.5331, "step": 1970 }, { "epoch": 0.63, "learning_rate": 1.722882874478824e-06, "loss": 0.5516, "step": 1980 }, { "epoch": 0.64, "learning_rate": 1.7190214624881096e-06, "loss": 0.551, "step": 1990 }, { "epoch": 0.64, "learning_rate": 1.7151377255557329e-06, "loss": 0.5769, "step": 2000 }, { "epoch": 0.64, "learning_rate": 1.7112317842680812e-06, "loss": 0.555, "step": 2010 }, { "epoch": 0.65, "learning_rate": 1.7073037599009666e-06, "loss": 0.5256, "step": 2020 }, { "epoch": 0.65, "learning_rate": 1.7033537744158588e-06, "loss": 0.5507, "step": 2030 }, { "epoch": 0.65, "learning_rate": 1.6993819504561003e-06, "loss": 0.5588, "step": 2040 }, { "epoch": 0.66, "learning_rate": 1.6953884113430984e-06, "loss": 0.5347, "step": 2050 }, { "epoch": 0.66, "learning_rate": 1.6913732810724946e-06, "loss": 0.5565, "step": 2060 }, { "epoch": 0.66, "learning_rate": 1.687336684310316e-06, "loss": 0.5568, "step": 2070 }, { "epoch": 0.67, "learning_rate": 1.683278746389104e-06, "loss": 0.543, "step": 2080 }, { "epoch": 0.67, "learning_rate": 1.6791995933040228e-06, "loss": 0.5402, "step": 2090 }, { "epoch": 0.67, "learning_rate": 1.675099351708948e-06, "loss": 0.5536, "step": 2100 }, { "epoch": 0.68, "learning_rate": 1.670978148912533e-06, "loss": 0.5522, "step": 2110 }, { "epoch": 0.68, "learning_rate": 1.6668361128742574e-06, "loss": 0.5611, "step": 2120 }, { "epoch": 0.68, "learning_rate": 1.6626733722004533e-06, "loss": 0.5365, "step": 2130 }, { "epoch": 0.69, "learning_rate": 1.6584900561403119e-06, "loss": 0.5366, "step": 2140 }, { "epoch": 0.69, "learning_rate": 1.6542862945818716e-06, "loss": 0.5411, "step": 2150 }, { "epoch": 0.69, "learning_rate": 1.6500622180479832e-06, "loss": 0.5077, "step": 2160 }, { "epoch": 0.7, "learning_rate": 1.64581795769226e-06, "loss": 0.5425, "step": 2170 }, { "epoch": 0.7, "learning_rate": 1.6415536452950023e-06, "loss": 0.5599, "step": 2180 }, { "epoch": 0.7, "learning_rate": 1.6372694132591085e-06, "loss": 0.5502, "step": 2190 }, { "epoch": 0.71, "learning_rate": 1.6329653946059633e-06, "loss": 0.5308, "step": 2200 }, { "epoch": 0.71, "learning_rate": 1.6286417229713061e-06, "loss": 0.522, "step": 2210 }, { "epoch": 0.71, "learning_rate": 1.6242985326010845e-06, "loss": 0.5391, "step": 2220 }, { "epoch": 0.71, "learning_rate": 1.6199359583472832e-06, "loss": 0.5216, "step": 2230 }, { "epoch": 0.72, "learning_rate": 1.6155541356637393e-06, "loss": 0.5354, "step": 2240 }, { "epoch": 0.72, "learning_rate": 1.6111532006019346e-06, "loss": 0.5279, "step": 2250 }, { "epoch": 0.72, "learning_rate": 1.6067332898067732e-06, "loss": 0.5437, "step": 2260 }, { "epoch": 0.73, "learning_rate": 1.6022945405123372e-06, "loss": 0.5246, "step": 2270 }, { "epoch": 0.73, "learning_rate": 1.5978370905376271e-06, "loss": 0.5445, "step": 2280 }, { "epoch": 0.73, "learning_rate": 1.5933610782822815e-06, "loss": 0.5355, "step": 2290 }, { "epoch": 0.74, "learning_rate": 1.5888666427222805e-06, "loss": 0.5239, "step": 2300 }, { "epoch": 0.74, "learning_rate": 1.5843539234056306e-06, "loss": 0.5275, "step": 2310 }, { "epoch": 0.74, "learning_rate": 1.5798230604480316e-06, "loss": 0.5415, "step": 2320 }, { "epoch": 0.75, "learning_rate": 1.5752741945285263e-06, "loss": 0.5405, "step": 2330 }, { "epoch": 0.75, "learning_rate": 1.5707074668851332e-06, "loss": 0.5397, "step": 2340 }, { "epoch": 0.75, "learning_rate": 1.56612301931046e-06, "loss": 0.5301, "step": 2350 }, { "epoch": 0.76, "learning_rate": 1.5615209941473013e-06, "loss": 0.5678, "step": 2360 }, { "epoch": 0.76, "learning_rate": 1.5569015342842201e-06, "loss": 0.5284, "step": 2370 }, { "epoch": 0.76, "learning_rate": 1.5522647831511102e-06, "loss": 0.529, "step": 2380 }, { "epoch": 0.77, "learning_rate": 1.5476108847147428e-06, "loss": 0.5362, "step": 2390 }, { "epoch": 0.77, "learning_rate": 1.5429399834742971e-06, "loss": 0.5337, "step": 2400 }, { "epoch": 0.77, "learning_rate": 1.5382522244568733e-06, "loss": 0.5356, "step": 2410 }, { "epoch": 0.78, "learning_rate": 1.5335477532129902e-06, "loss": 0.5174, "step": 2420 }, { "epoch": 0.78, "learning_rate": 1.5288267158120642e-06, "loss": 0.5387, "step": 2430 }, { "epoch": 0.78, "learning_rate": 1.5240892588378773e-06, "loss": 0.5443, "step": 2440 }, { "epoch": 0.79, "learning_rate": 1.5193355293840217e-06, "loss": 0.534, "step": 2450 }, { "epoch": 0.79, "learning_rate": 1.5145656750493369e-06, "loss": 0.549, "step": 2460 }, { "epoch": 0.79, "learning_rate": 1.5097798439333236e-06, "loss": 0.5442, "step": 2470 }, { "epoch": 0.8, "learning_rate": 1.5049781846315463e-06, "loss": 0.5272, "step": 2480 }, { "epoch": 0.8, "learning_rate": 1.500160846231021e-06, "loss": 0.5324, "step": 2490 }, { "epoch": 0.8, "learning_rate": 1.4953279783055838e-06, "loss": 0.5234, "step": 2500 }, { "epoch": 0.8, "learning_rate": 1.490479730911249e-06, "loss": 0.5714, "step": 2510 }, { "epoch": 0.81, "learning_rate": 1.4856162545815477e-06, "loss": 0.523, "step": 2520 }, { "epoch": 0.81, "learning_rate": 1.4807377003228568e-06, "loss": 0.5223, "step": 2530 }, { "epoch": 0.81, "learning_rate": 1.4758442196097071e-06, "loss": 0.5329, "step": 2540 }, { "epoch": 0.82, "learning_rate": 1.4709359643800834e-06, "loss": 0.5243, "step": 2550 }, { "epoch": 0.82, "learning_rate": 1.4660130870307036e-06, "loss": 0.5073, "step": 2560 }, { "epoch": 0.82, "learning_rate": 1.4610757404122905e-06, "loss": 0.5108, "step": 2570 }, { "epoch": 0.83, "learning_rate": 1.4561240778248228e-06, "loss": 0.5287, "step": 2580 }, { "epoch": 0.83, "learning_rate": 1.4511582530127771e-06, "loss": 0.5575, "step": 2590 }, { "epoch": 0.83, "learning_rate": 1.4466770294041548e-06, "loss": 0.5339, "step": 2600 }, { "epoch": 0.84, "learning_rate": 1.4416847215024572e-06, "loss": 0.5336, "step": 2610 }, { "epoch": 0.84, "learning_rate": 1.4366786997046453e-06, "loss": 0.5224, "step": 2620 }, { "epoch": 0.84, "learning_rate": 1.4316591194429983e-06, "loss": 0.5284, "step": 2630 }, { "epoch": 0.85, "learning_rate": 1.426626136570772e-06, "loss": 0.5269, "step": 2640 }, { "epoch": 0.85, "learning_rate": 1.4220851219018589e-06, "loss": 0.5295, "step": 2650 }, { "epoch": 0.85, "learning_rate": 1.4170271049317458e-06, "loss": 0.5314, "step": 2660 }, { "epoch": 0.86, "learning_rate": 1.4119561396613733e-06, "loss": 0.5217, "step": 2670 }, { "epoch": 0.86, "learning_rate": 1.4068723835394544e-06, "loss": 0.5304, "step": 2680 }, { "epoch": 0.86, "learning_rate": 1.4017759944118463e-06, "loss": 0.5403, "step": 2690 }, { "epoch": 0.87, "learning_rate": 1.3966671305166487e-06, "loss": 0.5166, "step": 2700 }, { "epoch": 0.87, "learning_rate": 1.391545950479291e-06, "loss": 0.5413, "step": 2710 }, { "epoch": 0.87, "learning_rate": 1.3864126133076077e-06, "loss": 0.5279, "step": 2720 }, { "epoch": 0.88, "learning_rate": 1.3812672783869008e-06, "loss": 0.5086, "step": 2730 }, { "epoch": 0.88, "learning_rate": 1.376110105474991e-06, "loss": 0.5144, "step": 2740 }, { "epoch": 0.88, "learning_rate": 1.3709412546972574e-06, "loss": 0.5129, "step": 2750 }, { "epoch": 0.88, "learning_rate": 1.3657608865416665e-06, "loss": 0.5185, "step": 2760 }, { "epoch": 0.89, "learning_rate": 1.3605691618537876e-06, "loss": 0.5323, "step": 2770 }, { "epoch": 0.89, "learning_rate": 1.3553662418318008e-06, "loss": 0.5266, "step": 2780 }, { "epoch": 0.89, "learning_rate": 1.35015228802149e-06, "loss": 0.5372, "step": 2790 }, { "epoch": 0.9, "learning_rate": 1.3449274623112278e-06, "loss": 0.5124, "step": 2800 }, { "epoch": 0.9, "learning_rate": 1.3396919269269494e-06, "loss": 0.5253, "step": 2810 }, { "epoch": 0.9, "learning_rate": 1.3344458444271159e-06, "loss": 0.5253, "step": 2820 }, { "epoch": 0.91, "learning_rate": 1.3291893776976655e-06, "loss": 0.5158, "step": 2830 }, { "epoch": 0.91, "learning_rate": 1.3239226899469572e-06, "loss": 0.5208, "step": 2840 }, { "epoch": 0.91, "learning_rate": 1.3186459447007035e-06, "loss": 0.5293, "step": 2850 }, { "epoch": 0.92, "learning_rate": 1.313359305796892e-06, "loss": 0.5342, "step": 2860 }, { "epoch": 0.92, "learning_rate": 1.3080629373807002e-06, "loss": 0.5184, "step": 2870 }, { "epoch": 0.92, "learning_rate": 1.3027570038993972e-06, "loss": 0.543, "step": 2880 }, { "epoch": 0.93, "learning_rate": 1.2974416700972377e-06, "loss": 0.5352, "step": 2890 }, { "epoch": 0.93, "learning_rate": 1.2921171010103486e-06, "loss": 0.51, "step": 2900 }, { "epoch": 0.93, "learning_rate": 1.2867834619616036e-06, "loss": 0.5405, "step": 2910 }, { "epoch": 0.94, "learning_rate": 1.2814409185554894e-06, "loss": 0.5379, "step": 2920 }, { "epoch": 0.94, "learning_rate": 1.2760896366729656e-06, "loss": 0.5133, "step": 2930 }, { "epoch": 0.94, "learning_rate": 1.2707297824663124e-06, "loss": 0.5352, "step": 2940 }, { "epoch": 0.95, "learning_rate": 1.2653615223539738e-06, "loss": 0.5147, "step": 2950 }, { "epoch": 0.95, "learning_rate": 1.2599850230153876e-06, "loss": 0.5145, "step": 2960 }, { "epoch": 0.95, "learning_rate": 1.2546004513858138e-06, "loss": 0.5347, "step": 2970 }, { "epoch": 0.96, "learning_rate": 1.249207974651148e-06, "loss": 0.5222, "step": 2980 }, { "epoch": 0.96, "learning_rate": 1.2438077602427326e-06, "loss": 0.5518, "step": 2990 }, { "epoch": 0.96, "learning_rate": 1.238399975832157e-06, "loss": 0.5474, "step": 3000 }, { "epoch": 0.96, "learning_rate": 1.2329847893260523e-06, "loss": 0.5284, "step": 3010 }, { "epoch": 0.97, "learning_rate": 1.2275623688608783e-06, "loss": 0.5303, "step": 3020 }, { "epoch": 0.97, "learning_rate": 1.222132882797701e-06, "loss": 0.5255, "step": 3030 }, { "epoch": 0.97, "learning_rate": 1.2166964997169685e-06, "loss": 0.5114, "step": 3040 }, { "epoch": 0.98, "learning_rate": 1.2112533884132734e-06, "loss": 0.5186, "step": 3050 }, { "epoch": 0.98, "learning_rate": 1.2058037178901135e-06, "loss": 0.5251, "step": 3060 }, { "epoch": 0.98, "learning_rate": 1.2003476573546446e-06, "loss": 0.5334, "step": 3070 }, { "epoch": 0.99, "learning_rate": 1.194885376212426e-06, "loss": 0.5552, "step": 3080 }, { "epoch": 0.99, "learning_rate": 1.1894170440621618e-06, "loss": 0.5399, "step": 3090 }, { "epoch": 0.99, "learning_rate": 1.1839428306904332e-06, "loss": 0.53, "step": 3100 }, { "epoch": 1.0, "learning_rate": 1.1784629060664283e-06, "loss": 0.5026, "step": 3110 }, { "epoch": 1.0, "learning_rate": 1.172977440336664e-06, "loss": 0.5273, "step": 3120 }, { "epoch": 1.0, "learning_rate": 1.167486603819704e-06, "loss": 0.5394, "step": 3130 }, { "epoch": 1.01, "learning_rate": 1.1619905670008687e-06, "loss": 0.5064, "step": 3140 }, { "epoch": 1.01, "learning_rate": 1.1564895005269444e-06, "loss": 0.5017, "step": 3150 }, { "epoch": 1.01, "learning_rate": 1.1509835752008826e-06, "loss": 0.4899, "step": 3160 }, { "epoch": 1.02, "learning_rate": 1.1454729619764979e-06, "loss": 0.5225, "step": 3170 }, { "epoch": 1.02, "learning_rate": 1.1399578319531589e-06, "loss": 0.5183, "step": 3180 }, { "epoch": 1.02, "learning_rate": 1.1344383563704783e-06, "loss": 0.5085, "step": 3190 }, { "epoch": 1.03, "learning_rate": 1.1289147066029923e-06, "loss": 0.5064, "step": 3200 }, { "epoch": 1.03, "learning_rate": 1.1233870541548433e-06, "loss": 0.5008, "step": 3210 }, { "epoch": 1.03, "learning_rate": 1.1178555706544523e-06, "loss": 0.5301, "step": 3220 }, { "epoch": 1.04, "learning_rate": 1.1123204278491916e-06, "loss": 0.5334, "step": 3230 }, { "epoch": 1.04, "learning_rate": 1.1067817976000503e-06, "loss": 0.5323, "step": 3240 }, { "epoch": 1.04, "learning_rate": 1.1012398518763e-06, "loss": 0.4997, "step": 3250 }, { "epoch": 1.05, "learning_rate": 1.0956947627501553e-06, "loss": 0.5, "step": 3260 }, { "epoch": 1.05, "learning_rate": 1.0901467023914293e-06, "loss": 0.5257, "step": 3270 }, { "epoch": 1.05, "learning_rate": 1.0845958430621904e-06, "loss": 0.5226, "step": 3280 }, { "epoch": 1.05, "learning_rate": 1.0790423571114113e-06, "loss": 0.5253, "step": 3290 }, { "epoch": 1.06, "learning_rate": 1.0734864169696192e-06, "loss": 0.5202, "step": 3300 }, { "epoch": 1.06, "learning_rate": 1.0679281951435417e-06, "loss": 0.5151, "step": 3310 }, { "epoch": 1.06, "learning_rate": 1.0623678642107506e-06, "loss": 0.5196, "step": 3320 }, { "epoch": 1.07, "learning_rate": 1.0568055968143036e-06, "loss": 0.5205, "step": 3330 }, { "epoch": 1.07, "learning_rate": 1.0512415656573828e-06, "loss": 0.4953, "step": 3340 }, { "epoch": 1.07, "learning_rate": 1.0456759434979352e-06, "loss": 0.4759, "step": 3350 }, { "epoch": 1.08, "learning_rate": 1.0401089031433055e-06, "loss": 0.5406, "step": 3360 }, { "epoch": 1.08, "learning_rate": 1.034540617444872e-06, "loss": 0.4961, "step": 3370 }, { "epoch": 1.08, "learning_rate": 1.0289712592926808e-06, "loss": 0.5322, "step": 3380 }, { "epoch": 1.09, "learning_rate": 1.023401001610076e-06, "loss": 0.522, "step": 3390 }, { "epoch": 1.09, "learning_rate": 1.0178300173483312e-06, "loss": 0.5249, "step": 3400 }, { "epoch": 1.09, "learning_rate": 1.01225847948128e-06, "loss": 0.528, "step": 3410 }, { "epoch": 1.1, "learning_rate": 1.0066865609999446e-06, "loss": 0.5108, "step": 3420 }, { "epoch": 1.1, "learning_rate": 1.001114434907165e-06, "loss": 0.4848, "step": 3430 }, { "epoch": 1.1, "learning_rate": 9.955422742122276e-07, "loss": 0.5241, "step": 3440 }, { "epoch": 1.11, "learning_rate": 9.899702519254922e-07, "loss": 0.5059, "step": 3450 }, { "epoch": 1.11, "learning_rate": 9.843985410530223e-07, "loss": 0.5002, "step": 3460 }, { "epoch": 1.11, "learning_rate": 9.788273145912125e-07, "loss": 0.5022, "step": 3470 }, { "epoch": 1.12, "learning_rate": 9.73256745521415e-07, "loss": 0.5063, "step": 3480 }, { "epoch": 1.12, "learning_rate": 9.676870068045726e-07, "loss": 0.5293, "step": 3490 }, { "epoch": 1.12, "learning_rate": 9.621182713758458e-07, "loss": 0.5186, "step": 3500 }, { "epoch": 1.13, "learning_rate": 9.565507121392435e-07, "loss": 0.5322, "step": 3510 }, { "epoch": 1.13, "learning_rate": 9.509845019622556e-07, "loss": 0.5253, "step": 3520 }, { "epoch": 1.13, "learning_rate": 9.45419813670484e-07, "loss": 0.5173, "step": 3530 }, { "epoch": 1.13, "learning_rate": 9.398568200422788e-07, "loss": 0.5187, "step": 3540 }, { "epoch": 1.14, "learning_rate": 9.34295693803371e-07, "loss": 0.5233, "step": 3550 }, { "epoch": 1.14, "learning_rate": 9.287366076215121e-07, "loss": 0.5182, "step": 3560 }, { "epoch": 1.14, "learning_rate": 9.231797341011111e-07, "loss": 0.5295, "step": 3570 }, { "epoch": 1.15, "learning_rate": 9.176252457778768e-07, "loss": 0.5181, "step": 3580 }, { "epoch": 1.15, "learning_rate": 9.120733151134578e-07, "loss": 0.5262, "step": 3590 }, { "epoch": 1.15, "learning_rate": 9.065241144900921e-07, "loss": 0.5431, "step": 3600 }, { "epoch": 1.16, "learning_rate": 9.009778162052516e-07, "loss": 0.4948, "step": 3610 }, { "epoch": 1.16, "learning_rate": 8.954345924662926e-07, "loss": 0.517, "step": 3620 }, { "epoch": 1.16, "learning_rate": 8.898946153851106e-07, "loss": 0.5251, "step": 3630 }, { "epoch": 1.17, "learning_rate": 8.843580569727952e-07, "loss": 0.5089, "step": 3640 }, { "epoch": 1.17, "learning_rate": 8.788250891342897e-07, "loss": 0.5261, "step": 3650 }, { "epoch": 1.17, "learning_rate": 8.732958836630528e-07, "loss": 0.4978, "step": 3660 }, { "epoch": 1.18, "learning_rate": 8.677706122357257e-07, "loss": 0.4982, "step": 3670 }, { "epoch": 1.18, "learning_rate": 8.622494464068014e-07, "loss": 0.5143, "step": 3680 }, { "epoch": 1.18, "learning_rate": 8.567325576032973e-07, "loss": 0.4988, "step": 3690 }, { "epoch": 1.19, "learning_rate": 8.512201171194338e-07, "loss": 0.4856, "step": 3700 }, { "epoch": 1.19, "learning_rate": 8.457122961113148e-07, "loss": 0.4995, "step": 3710 }, { "epoch": 1.19, "learning_rate": 8.402092655916145e-07, "loss": 0.5174, "step": 3720 }, { "epoch": 1.2, "learning_rate": 8.347111964242652e-07, "loss": 0.5359, "step": 3730 }, { "epoch": 1.2, "learning_rate": 8.292182593191559e-07, "loss": 0.5072, "step": 3740 }, { "epoch": 1.2, "learning_rate": 8.237306248268292e-07, "loss": 0.5049, "step": 3750 }, { "epoch": 1.21, "learning_rate": 8.182484633331854e-07, "loss": 0.5217, "step": 3760 }, { "epoch": 1.21, "learning_rate": 8.127719450541949e-07, "loss": 0.5176, "step": 3770 }, { "epoch": 1.21, "learning_rate": 8.073012400306107e-07, "loss": 0.49, "step": 3780 }, { "epoch": 1.22, "learning_rate": 8.018365181226905e-07, "loss": 0.5044, "step": 3790 }, { "epoch": 1.22, "learning_rate": 7.963779490049206e-07, "loss": 0.5039, "step": 3800 }, { "epoch": 1.22, "learning_rate": 7.909257021607501e-07, "loss": 0.5094, "step": 3810 }, { "epoch": 1.22, "learning_rate": 7.854799468773269e-07, "loss": 0.5336, "step": 3820 }, { "epoch": 1.23, "learning_rate": 7.800408522402417e-07, "loss": 0.5073, "step": 3830 }, { "epoch": 1.23, "learning_rate": 7.746085871282792e-07, "loss": 0.5266, "step": 3840 }, { "epoch": 1.23, "learning_rate": 7.691833202081737e-07, "loss": 0.5199, "step": 3850 }, { "epoch": 1.24, "learning_rate": 7.637652199293712e-07, "loss": 0.474, "step": 3860 }, { "epoch": 1.24, "learning_rate": 7.583544545188009e-07, "loss": 0.5348, "step": 3870 }, { "epoch": 1.24, "learning_rate": 7.529511919756514e-07, "loss": 0.4942, "step": 3880 }, { "epoch": 1.25, "learning_rate": 7.475556000661546e-07, "loss": 0.4997, "step": 3890 }, { "epoch": 1.25, "learning_rate": 7.421678463183749e-07, "loss": 0.506, "step": 3900 }, { "epoch": 1.25, "learning_rate": 7.36788098017011e-07, "loss": 0.5053, "step": 3910 }, { "epoch": 1.26, "learning_rate": 7.314165221981987e-07, "loss": 0.5237, "step": 3920 }, { "epoch": 1.26, "learning_rate": 7.260532856443265e-07, "loss": 0.5333, "step": 3930 }, { "epoch": 1.26, "learning_rate": 7.206985548788565e-07, "loss": 0.5091, "step": 3940 }, { "epoch": 1.27, "learning_rate": 7.153524961611539e-07, "loss": 0.5137, "step": 3950 }, { "epoch": 1.27, "learning_rate": 7.100152754813251e-07, "loss": 0.5018, "step": 3960 }, { "epoch": 1.27, "learning_rate": 7.046870585550633e-07, "loss": 0.5225, "step": 3970 }, { "epoch": 1.28, "learning_rate": 6.993680108185041e-07, "loss": 0.4929, "step": 3980 }, { "epoch": 1.28, "learning_rate": 6.940582974230884e-07, "loss": 0.4791, "step": 3990 }, { "epoch": 1.28, "learning_rate": 6.88758083230434e-07, "loss": 0.5104, "step": 4000 }, { "epoch": 1.29, "learning_rate": 6.834675328072179e-07, "loss": 0.5057, "step": 4010 }, { "epoch": 1.29, "learning_rate": 6.781868104200659e-07, "loss": 0.528, "step": 4020 }, { "epoch": 1.29, "learning_rate": 6.72916080030453e-07, "loss": 0.4958, "step": 4030 }, { "epoch": 1.3, "learning_rate": 6.676555052896107e-07, "loss": 0.5154, "step": 4040 }, { "epoch": 1.3, "learning_rate": 6.624052495334486e-07, "loss": 0.5279, "step": 4050 }, { "epoch": 1.3, "learning_rate": 6.571654757774811e-07, "loss": 0.5204, "step": 4060 }, { "epoch": 1.3, "learning_rate": 6.51936346711766e-07, "loss": 0.5288, "step": 4070 }, { "epoch": 1.31, "learning_rate": 6.467180246958535e-07, "loss": 0.5281, "step": 4080 }, { "epoch": 1.31, "learning_rate": 6.415106717537452e-07, "loss": 0.5227, "step": 4090 }, { "epoch": 1.31, "learning_rate": 6.363144495688639e-07, "loss": 0.5171, "step": 4100 }, { "epoch": 1.32, "learning_rate": 6.311295194790311e-07, "loss": 0.5023, "step": 4110 }, { "epoch": 1.32, "learning_rate": 6.259560424714611e-07, "loss": 0.5009, "step": 4120 }, { "epoch": 1.32, "learning_rate": 6.207941791777602e-07, "loss": 0.5284, "step": 4130 }, { "epoch": 1.33, "learning_rate": 6.156440898689391e-07, "loss": 0.4981, "step": 4140 }, { "epoch": 1.33, "learning_rate": 6.10505934450438e-07, "loss": 0.4968, "step": 4150 }, { "epoch": 1.33, "learning_rate": 6.053798724571613e-07, "loss": 0.5002, "step": 4160 }, { "epoch": 1.34, "learning_rate": 6.002660630485229e-07, "loss": 0.5072, "step": 4170 }, { "epoch": 1.34, "learning_rate": 5.951646650035063e-07, "loss": 0.4958, "step": 4180 }, { "epoch": 1.34, "learning_rate": 5.900758367157333e-07, "loss": 0.4906, "step": 4190 }, { "epoch": 1.35, "learning_rate": 5.849997361885471e-07, "loss": 0.5071, "step": 4200 }, { "epoch": 1.35, "learning_rate": 5.799365210301052e-07, "loss": 0.5118, "step": 4210 }, { "epoch": 1.35, "learning_rate": 5.748863484484873e-07, "loss": 0.4934, "step": 4220 }, { "epoch": 1.36, "learning_rate": 5.698493752468125e-07, "loss": 0.4942, "step": 4230 }, { "epoch": 1.36, "learning_rate": 5.648257578183717e-07, "loss": 0.5035, "step": 4240 }, { "epoch": 1.36, "learning_rate": 5.598156521417718e-07, "loss": 0.4959, "step": 4250 }, { "epoch": 1.37, "learning_rate": 5.548192137760929e-07, "loss": 0.5081, "step": 4260 }, { "epoch": 1.37, "learning_rate": 5.49836597856057e-07, "loss": 0.5056, "step": 4270 }, { "epoch": 1.37, "learning_rate": 5.448679590872127e-07, "loss": 0.5279, "step": 4280 }, { "epoch": 1.38, "learning_rate": 5.399134517411314e-07, "loss": 0.5104, "step": 4290 }, { "epoch": 1.38, "learning_rate": 5.349732296506164e-07, "loss": 0.541, "step": 4300 }, { "epoch": 1.38, "learning_rate": 5.300474462049278e-07, "loss": 0.4947, "step": 4310 }, { "epoch": 1.39, "learning_rate": 5.251362543450186e-07, "loss": 0.5006, "step": 4320 }, { "epoch": 1.39, "learning_rate": 5.20239806558788e-07, "loss": 0.5475, "step": 4330 }, { "epoch": 1.39, "learning_rate": 5.153582548763447e-07, "loss": 0.4827, "step": 4340 }, { "epoch": 1.39, "learning_rate": 5.104917508652871e-07, "loss": 0.496, "step": 4350 }, { "epoch": 1.4, "learning_rate": 5.056404456259975e-07, "loss": 0.5047, "step": 4360 }, { "epoch": 1.4, "learning_rate": 5.008044897869523e-07, "loss": 0.5015, "step": 4370 }, { "epoch": 1.4, "learning_rate": 4.959840335000407e-07, "loss": 0.5439, "step": 4380 }, { "epoch": 1.41, "learning_rate": 4.911792264359078e-07, "loss": 0.4873, "step": 4390 }, { "epoch": 1.41, "learning_rate": 4.863902177793031e-07, "loss": 0.5053, "step": 4400 }, { "epoch": 1.41, "learning_rate": 4.816171562244522e-07, "loss": 0.4916, "step": 4410 }, { "epoch": 1.42, "learning_rate": 4.768601899704368e-07, "loss": 0.5219, "step": 4420 }, { "epoch": 1.42, "learning_rate": 4.721194667165945e-07, "loss": 0.5217, "step": 4430 }, { "epoch": 1.42, "learning_rate": 4.673951336579348e-07, "loss": 0.5125, "step": 4440 }, { "epoch": 1.43, "learning_rate": 4.626873374805642e-07, "loss": 0.503, "step": 4450 }, { "epoch": 1.43, "learning_rate": 4.5799622435713727e-07, "loss": 0.4909, "step": 4460 }, { "epoch": 1.43, "learning_rate": 4.533219399423137e-07, "loss": 0.5168, "step": 4470 }, { "epoch": 1.44, "learning_rate": 4.4866462936823837e-07, "loss": 0.5233, "step": 4480 }, { "epoch": 1.44, "learning_rate": 4.4402443724003334e-07, "loss": 0.5158, "step": 4490 }, { "epoch": 1.44, "learning_rate": 4.394015076313107e-07, "loss": 0.5162, "step": 4500 }, { "epoch": 1.45, "learning_rate": 4.347959840796961e-07, "loss": 0.4965, "step": 4510 }, { "epoch": 1.45, "learning_rate": 4.302080095823739e-07, "loss": 0.475, "step": 4520 }, { "epoch": 1.45, "learning_rate": 4.256377265916463e-07, "loss": 0.5333, "step": 4530 }, { "epoch": 1.46, "learning_rate": 4.210852770105121e-07, "loss": 0.514, "step": 4540 }, { "epoch": 1.46, "learning_rate": 4.165508021882583e-07, "loss": 0.5091, "step": 4550 }, { "epoch": 1.46, "learning_rate": 4.120344429160726e-07, "loss": 0.4908, "step": 4560 }, { "epoch": 1.47, "learning_rate": 4.0753633942267263e-07, "loss": 0.5042, "step": 4570 }, { "epoch": 1.47, "learning_rate": 4.030566313699505e-07, "loss": 0.5021, "step": 4580 }, { "epoch": 1.47, "learning_rate": 3.9859545784863735e-07, "loss": 0.5313, "step": 4590 }, { "epoch": 1.47, "learning_rate": 3.9415295737398416e-07, "loss": 0.5158, "step": 4600 }, { "epoch": 1.48, "learning_rate": 3.8972926788146285e-07, "loss": 0.5122, "step": 4610 }, { "epoch": 1.48, "learning_rate": 3.853245267224794e-07, "loss": 0.5112, "step": 4620 }, { "epoch": 1.48, "learning_rate": 3.809388706601142e-07, "loss": 0.5023, "step": 4630 }, { "epoch": 1.49, "learning_rate": 3.7657243586487197e-07, "loss": 0.4939, "step": 4640 }, { "epoch": 1.49, "learning_rate": 3.7222535791045595e-07, "loss": 0.5016, "step": 4650 }, { "epoch": 1.49, "learning_rate": 3.6789777176955703e-07, "loss": 0.5089, "step": 4660 }, { "epoch": 1.5, "learning_rate": 3.6358981180966375e-07, "loss": 0.4797, "step": 4670 }, { "epoch": 1.5, "learning_rate": 3.593016117888901e-07, "loss": 0.5253, "step": 4680 }, { "epoch": 1.5, "learning_rate": 3.5503330485182217e-07, "loss": 0.4959, "step": 4690 }, { "epoch": 1.51, "learning_rate": 3.5120894674136217e-07, "loss": 0.507, "step": 4700 }, { "epoch": 1.51, "learning_rate": 3.469788012612012e-07, "loss": 0.4975, "step": 4710 }, { "epoch": 1.51, "learning_rate": 3.427689314764486e-07, "loss": 0.5109, "step": 4720 }, { "epoch": 1.52, "learning_rate": 3.385794680996115e-07, "loss": 0.5059, "step": 4730 }, { "epoch": 1.52, "learning_rate": 3.344105412095959e-07, "loss": 0.507, "step": 4740 }, { "epoch": 1.52, "learning_rate": 3.302622802476694e-07, "loss": 0.4812, "step": 4750 }, { "epoch": 1.53, "learning_rate": 3.26134814013442e-07, "loss": 0.4863, "step": 4760 }, { "epoch": 1.53, "learning_rate": 3.2202827066086734e-07, "loss": 0.4808, "step": 4770 }, { "epoch": 1.53, "learning_rate": 3.179427776942626e-07, "loss": 0.5236, "step": 4780 }, { "epoch": 1.54, "learning_rate": 3.1387846196435043e-07, "loss": 0.473, "step": 4790 }, { "epoch": 1.54, "learning_rate": 3.098354496643198e-07, "loss": 0.5148, "step": 4800 }, { "epoch": 1.54, "learning_rate": 3.0581386632590943e-07, "loss": 0.4956, "step": 4810 }, { "epoch": 1.55, "learning_rate": 3.0181383681550756e-07, "loss": 0.4898, "step": 4820 }, { "epoch": 1.55, "learning_rate": 2.9783548533027636e-07, "loss": 0.4821, "step": 4830 }, { "epoch": 1.55, "learning_rate": 2.9387893539429673e-07, "loss": 0.507, "step": 4840 }, { "epoch": 1.55, "learning_rate": 2.899443098547311e-07, "loss": 0.5288, "step": 4850 }, { "epoch": 1.56, "learning_rate": 2.860317308780098e-07, "loss": 0.5102, "step": 4860 }, { "epoch": 1.56, "learning_rate": 2.821413199460384e-07, "loss": 0.5139, "step": 4870 }, { "epoch": 1.56, "learning_rate": 2.782731978524263e-07, "loss": 0.5116, "step": 4880 }, { "epoch": 1.57, "learning_rate": 2.7442748469873334e-07, "loss": 0.5268, "step": 4890 }, { "epoch": 1.57, "learning_rate": 2.7060429989074486e-07, "loss": 0.5082, "step": 4900 }, { "epoch": 1.57, "learning_rate": 2.6680376213476074e-07, "loss": 0.5057, "step": 4910 }, { "epoch": 1.58, "learning_rate": 2.6302598943391285e-07, "loss": 0.5126, "step": 4920 }, { "epoch": 1.58, "learning_rate": 2.592710990844972e-07, "loss": 0.4905, "step": 4930 }, { "epoch": 1.58, "learning_rate": 2.5553920767233617e-07, "loss": 0.4916, "step": 4940 }, { "epoch": 1.59, "learning_rate": 2.518304310691558e-07, "loss": 0.4932, "step": 4950 }, { "epoch": 1.59, "learning_rate": 2.48144884428989e-07, "loss": 0.4999, "step": 4960 }, { "epoch": 1.59, "learning_rate": 2.444826821846007e-07, "loss": 0.4867, "step": 4970 }, { "epoch": 1.6, "learning_rate": 2.4084393804393345e-07, "loss": 0.5218, "step": 4980 }, { "epoch": 1.6, "learning_rate": 2.3722876498657906e-07, "loss": 0.4948, "step": 4990 }, { "epoch": 1.6, "learning_rate": 2.336372752602671e-07, "loss": 0.5039, "step": 5000 }, { "epoch": 1.61, "learning_rate": 2.300695803773841e-07, "loss": 0.4744, "step": 5010 }, { "epoch": 1.61, "learning_rate": 2.2652579111150806e-07, "loss": 0.4962, "step": 5020 }, { "epoch": 1.61, "learning_rate": 2.2300601749396986e-07, "loss": 0.5068, "step": 5030 }, { "epoch": 1.62, "learning_rate": 2.1951036881043704e-07, "loss": 0.5082, "step": 5040 }, { "epoch": 1.62, "learning_rate": 2.1603895359752145e-07, "loss": 0.4763, "step": 5050 }, { "epoch": 1.62, "learning_rate": 2.125918796394076e-07, "loss": 0.503, "step": 5060 }, { "epoch": 1.63, "learning_rate": 2.0916925396450692e-07, "loss": 0.5045, "step": 5070 }, { "epoch": 1.63, "learning_rate": 2.0577118284213447e-07, "loss": 0.5247, "step": 5080 }, { "epoch": 1.63, "learning_rate": 2.0239777177921025e-07, "loss": 0.4982, "step": 5090 }, { "epoch": 1.64, "learning_rate": 1.9904912551698193e-07, "loss": 0.5274, "step": 5100 }, { "epoch": 1.64, "learning_rate": 1.957253480277733e-07, "loss": 0.479, "step": 5110 }, { "epoch": 1.64, "learning_rate": 1.9242654251175683e-07, "loss": 0.4976, "step": 5120 }, { "epoch": 1.64, "learning_rate": 1.891528113937473e-07, "loss": 0.5086, "step": 5130 }, { "epoch": 1.65, "learning_rate": 1.8590425632002426e-07, "loss": 0.509, "step": 5140 }, { "epoch": 1.65, "learning_rate": 1.8268097815517402e-07, "loss": 0.5139, "step": 5150 }, { "epoch": 1.65, "learning_rate": 1.7948307697895937e-07, "loss": 0.5068, "step": 5160 }, { "epoch": 1.66, "learning_rate": 1.7631065208320995e-07, "loss": 0.5124, "step": 5170 }, { "epoch": 1.66, "learning_rate": 1.7316380196874236e-07, "loss": 0.5039, "step": 5180 }, { "epoch": 1.66, "learning_rate": 1.7004262434229922e-07, "loss": 0.5173, "step": 5190 }, { "epoch": 1.67, "learning_rate": 1.6694721611351703e-07, "loss": 0.5033, "step": 5200 }, { "epoch": 1.67, "learning_rate": 1.6387767339191615e-07, "loss": 0.5017, "step": 5210 }, { "epoch": 1.67, "learning_rate": 1.6083409148391803e-07, "loss": 0.5063, "step": 5220 }, { "epoch": 1.68, "learning_rate": 1.5781656488988447e-07, "loss": 0.509, "step": 5230 }, { "epoch": 1.68, "learning_rate": 1.5482518730118443e-07, "loss": 0.5369, "step": 5240 }, { "epoch": 1.68, "learning_rate": 1.518600515972851e-07, "loss": 0.5068, "step": 5250 }, { "epoch": 1.69, "learning_rate": 1.4892124984286736e-07, "loss": 0.5026, "step": 5260 }, { "epoch": 1.69, "learning_rate": 1.460088732849678e-07, "loss": 0.4871, "step": 5270 }, { "epoch": 1.69, "learning_rate": 1.4312301235014534e-07, "loss": 0.5198, "step": 5280 }, { "epoch": 1.7, "learning_rate": 1.4026375664167434e-07, "loss": 0.4999, "step": 5290 }, { "epoch": 1.7, "learning_rate": 1.3743119493676102e-07, "loss": 0.4855, "step": 5300 }, { "epoch": 1.7, "learning_rate": 1.3462541518378823e-07, "loss": 0.5142, "step": 5310 }, { "epoch": 1.71, "learning_rate": 1.318465044995841e-07, "loss": 0.5197, "step": 5320 }, { "epoch": 1.71, "learning_rate": 1.290945491667178e-07, "loss": 0.5181, "step": 5330 }, { "epoch": 1.71, "learning_rate": 1.2636963463081996e-07, "loss": 0.5227, "step": 5340 }, { "epoch": 1.72, "learning_rate": 1.2367184549792963e-07, "loss": 0.5132, "step": 5350 }, { "epoch": 1.72, "learning_rate": 1.2100126553186784e-07, "loss": 0.5024, "step": 5360 }, { "epoch": 1.72, "learning_rate": 1.1835797765163601e-07, "loss": 0.5134, "step": 5370 }, { "epoch": 1.72, "learning_rate": 1.1574206392884279e-07, "loss": 0.4833, "step": 5380 }, { "epoch": 1.73, "learning_rate": 1.1315360558515408e-07, "loss": 0.496, "step": 5390 }, { "epoch": 1.73, "learning_rate": 1.1059268298977286e-07, "loss": 0.5051, "step": 5400 }, { "epoch": 1.73, "learning_rate": 1.0805937565694189e-07, "loss": 0.5138, "step": 5410 }, { "epoch": 1.74, "learning_rate": 1.055537622434769e-07, "loss": 0.5038, "step": 5420 }, { "epoch": 1.74, "learning_rate": 1.0307592054632297e-07, "loss": 0.4939, "step": 5430 }, { "epoch": 1.74, "learning_rate": 1.0062592750013944e-07, "loss": 0.498, "step": 5440 }, { "epoch": 1.75, "learning_rate": 9.820385917491114e-08, "loss": 0.4849, "step": 5450 }, { "epoch": 1.75, "learning_rate": 9.580979077358708e-08, "loss": 0.5012, "step": 5460 }, { "epoch": 1.75, "learning_rate": 9.344379662974412e-08, "loss": 0.509, "step": 5470 }, { "epoch": 1.76, "learning_rate": 9.110595020528e-08, "loss": 0.5329, "step": 5480 }, { "epoch": 1.76, "learning_rate": 8.879632408813275e-08, "loss": 0.5232, "step": 5490 }, { "epoch": 1.76, "learning_rate": 8.65149899900256e-08, "loss": 0.5319, "step": 5500 }, { "epoch": 1.77, "learning_rate": 8.426201874424154e-08, "loss": 0.4951, "step": 5510 }, { "epoch": 1.77, "learning_rate": 8.203748030342328e-08, "loss": 0.5111, "step": 5520 }, { "epoch": 1.77, "learning_rate": 7.98414437374021e-08, "loss": 0.4982, "step": 5530 }, { "epoch": 1.78, "learning_rate": 7.767397723105252e-08, "loss": 0.4836, "step": 5540 }, { "epoch": 1.78, "learning_rate": 7.553514808217543e-08, "loss": 0.4937, "step": 5550 }, { "epoch": 1.78, "learning_rate": 7.342502269940876e-08, "loss": 0.4974, "step": 5560 }, { "epoch": 1.79, "learning_rate": 7.134366660016578e-08, "loss": 0.4974, "step": 5570 }, { "epoch": 1.79, "learning_rate": 6.929114440859973e-08, "loss": 0.5046, "step": 5580 }, { "epoch": 1.79, "learning_rate": 6.726751985359869e-08, "loss": 0.5193, "step": 5590 }, { "epoch": 1.8, "learning_rate": 6.527285576680608e-08, "loss": 0.5089, "step": 5600 }, { "epoch": 1.8, "learning_rate": 6.330721408067008e-08, "loss": 0.4935, "step": 5610 }, { "epoch": 1.8, "learning_rate": 6.137065582652034e-08, "loss": 0.5122, "step": 5620 }, { "epoch": 1.8, "learning_rate": 5.9463241132673313e-08, "loss": 0.4797, "step": 5630 }, { "epoch": 1.81, "learning_rate": 5.7585029222565254e-08, "loss": 0.5062, "step": 5640 }, { "epoch": 1.81, "learning_rate": 5.573607841291328e-08, "loss": 0.489, "step": 5650 }, { "epoch": 1.81, "learning_rate": 5.391644611190527e-08, "loss": 0.5044, "step": 5660 }, { "epoch": 1.82, "learning_rate": 5.2126188817416374e-08, "loss": 0.5254, "step": 5670 }, { "epoch": 1.82, "learning_rate": 5.036536211525555e-08, "loss": 0.5024, "step": 5680 }, { "epoch": 1.82, "learning_rate": 4.863402067743949e-08, "loss": 0.5075, "step": 5690 }, { "epoch": 1.83, "learning_rate": 4.6932218260495205e-08, "loss": 0.4883, "step": 5700 }, { "epoch": 1.83, "learning_rate": 4.5260007703790685e-08, "loss": 0.4976, "step": 5710 }, { "epoch": 1.83, "learning_rate": 4.361744092789443e-08, "loss": 0.4856, "step": 5720 }, { "epoch": 1.84, "learning_rate": 4.200456893296323e-08, "loss": 0.5119, "step": 5730 }, { "epoch": 1.84, "learning_rate": 4.042144179715945e-08, "loss": 0.5115, "step": 5740 }, { "epoch": 1.84, "learning_rate": 3.8868108675094605e-08, "loss": 0.5033, "step": 5750 }, { "epoch": 1.85, "learning_rate": 3.734461779630438e-08, "loss": 0.5275, "step": 5760 }, { "epoch": 1.85, "learning_rate": 3.5851016463751326e-08, "loss": 0.4971, "step": 5770 }, { "epoch": 1.85, "learning_rate": 3.4387351052354794e-08, "loss": 0.4897, "step": 5780 }, { "epoch": 1.86, "learning_rate": 3.295366700755253e-08, "loss": 0.5228, "step": 5790 }, { "epoch": 1.86, "learning_rate": 3.155000884388837e-08, "loss": 0.4984, "step": 5800 }, { "epoch": 1.86, "learning_rate": 3.017642014363153e-08, "loss": 0.5088, "step": 5810 }, { "epoch": 1.87, "learning_rate": 2.883294355542165e-08, "loss": 0.5066, "step": 5820 }, { "epoch": 1.87, "learning_rate": 2.7519620792946342e-08, "loss": 0.5143, "step": 5830 }, { "epoch": 1.87, "learning_rate": 2.6236492633644803e-08, "loss": 0.4984, "step": 5840 }, { "epoch": 1.88, "learning_rate": 2.4983598917442506e-08, "loss": 0.4915, "step": 5850 }, { "epoch": 1.88, "learning_rate": 2.376097854551329e-08, "loss": 0.5076, "step": 5860 }, { "epoch": 1.88, "learning_rate": 2.2568669479072543e-08, "loss": 0.4881, "step": 5870 }, { "epoch": 1.89, "learning_rate": 2.140670873819772e-08, "loss": 0.4999, "step": 5880 }, { "epoch": 1.89, "learning_rate": 2.0275132400679572e-08, "loss": 0.4984, "step": 5890 }, { "epoch": 1.89, "learning_rate": 1.9173975600901394e-08, "loss": 0.4792, "step": 5900 }, { "epoch": 1.89, "learning_rate": 1.8103272528748436e-08, "loss": 0.4904, "step": 5910 }, { "epoch": 1.9, "learning_rate": 1.7063056428546552e-08, "loss": 0.4989, "step": 5920 }, { "epoch": 1.9, "learning_rate": 1.6053359598029226e-08, "loss": 0.5054, "step": 5930 }, { "epoch": 1.9, "learning_rate": 1.50742133873355e-08, "loss": 0.4859, "step": 5940 }, { "epoch": 1.91, "learning_rate": 1.4125648198036633e-08, "loss": 0.4821, "step": 5950 }, { "epoch": 1.91, "learning_rate": 1.3207693482191307e-08, "loss": 0.4995, "step": 5960 }, { "epoch": 1.91, "learning_rate": 1.232037774143213e-08, "loss": 0.5035, "step": 5970 }, { "epoch": 1.92, "learning_rate": 1.146372852608024e-08, "loss": 0.5021, "step": 5980 }, { "epoch": 1.92, "learning_rate": 1.063777243428965e-08, "loss": 0.4814, "step": 5990 }, { "epoch": 1.92, "learning_rate": 9.842535111222238e-09, "loss": 0.4901, "step": 6000 } ], "max_steps": 6238, "num_train_epochs": 2, "total_flos": 2.0920322636916982e+18, "trial_name": null, "trial_params": null }