{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 500, "global_step": 6477, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.0000000000000002e-06, "loss": 0.5967, "step": 10 }, { "epoch": 0.01, "learning_rate": 2.0000000000000003e-06, "loss": 0.5579, "step": 20 }, { "epoch": 0.01, "learning_rate": 3e-06, "loss": 0.5271, "step": 30 }, { "epoch": 0.02, "learning_rate": 4.000000000000001e-06, "loss": 0.4947, "step": 40 }, { "epoch": 0.02, "learning_rate": 5e-06, "loss": 0.4531, "step": 50 }, { "epoch": 0.03, "learning_rate": 6e-06, "loss": 0.4091, "step": 60 }, { "epoch": 0.03, "learning_rate": 7e-06, "loss": 0.3732, "step": 70 }, { "epoch": 0.04, "learning_rate": 8.000000000000001e-06, "loss": 0.3181, "step": 80 }, { "epoch": 0.04, "learning_rate": 9e-06, "loss": 0.2763, "step": 90 }, { "epoch": 0.05, "learning_rate": 1e-05, "loss": 0.2419, "step": 100 }, { "epoch": 0.05, "learning_rate": 1.1000000000000001e-05, "loss": 0.2118, "step": 110 }, { "epoch": 0.06, "learning_rate": 1.2e-05, "loss": 0.1898, "step": 120 }, { "epoch": 0.06, "learning_rate": 1.3000000000000001e-05, "loss": 0.1659, "step": 130 }, { "epoch": 0.06, "learning_rate": 1.4e-05, "loss": 0.1456, "step": 140 }, { "epoch": 0.07, "learning_rate": 1.5000000000000002e-05, "loss": 0.1309, "step": 150 }, { "epoch": 0.07, "learning_rate": 1.6000000000000003e-05, "loss": 0.1123, "step": 160 }, { "epoch": 0.08, "learning_rate": 1.7e-05, "loss": 0.1133, "step": 170 }, { "epoch": 0.08, "learning_rate": 1.8e-05, "loss": 0.1069, "step": 180 }, { "epoch": 0.09, "learning_rate": 1.9e-05, "loss": 0.0919, "step": 190 }, { "epoch": 0.09, "learning_rate": 2e-05, "loss": 0.0874, "step": 200 }, { "epoch": 0.1, "learning_rate": 1.9999874753792008e-05, "loss": 0.0889, "step": 210 }, { "epoch": 0.1, "learning_rate": 1.999949901830535e-05, "loss": 0.0802, "step": 220 }, { "epoch": 0.11, "learning_rate": 1.9998872802951915e-05, "loss": 0.0804, "step": 230 }, { "epoch": 0.11, "learning_rate": 1.999799612341792e-05, "loss": 0.0744, "step": 240 }, { "epoch": 0.12, "learning_rate": 1.999686900166353e-05, "loss": 0.0751, "step": 250 }, { "epoch": 0.12, "learning_rate": 1.9995491465922284e-05, "loss": 0.0746, "step": 260 }, { "epoch": 0.13, "learning_rate": 1.9993863550700413e-05, "loss": 0.0721, "step": 270 }, { "epoch": 0.13, "learning_rate": 1.999198529677595e-05, "loss": 0.0714, "step": 280 }, { "epoch": 0.13, "learning_rate": 1.9989856751197742e-05, "loss": 0.0658, "step": 290 }, { "epoch": 0.14, "learning_rate": 1.998747796728423e-05, "loss": 0.069, "step": 300 }, { "epoch": 0.14, "learning_rate": 1.998484900462216e-05, "loss": 0.0647, "step": 310 }, { "epoch": 0.15, "learning_rate": 1.9981969929065044e-05, "loss": 0.0712, "step": 320 }, { "epoch": 0.15, "learning_rate": 1.997884081273154e-05, "loss": 0.0592, "step": 330 }, { "epoch": 0.16, "learning_rate": 1.9975461734003646e-05, "loss": 0.0723, "step": 340 }, { "epoch": 0.16, "learning_rate": 1.9971832777524717e-05, "loss": 0.0636, "step": 350 }, { "epoch": 0.17, "learning_rate": 1.996795403419736e-05, "loss": 0.062, "step": 360 }, { "epoch": 0.17, "learning_rate": 1.9963825601181152e-05, "loss": 0.0629, "step": 370 }, { "epoch": 0.18, "learning_rate": 1.9959447581890218e-05, "loss": 0.0577, "step": 380 }, { "epoch": 0.18, "learning_rate": 1.995482008599061e-05, "loss": 0.0611, "step": 390 }, { "epoch": 0.19, "learning_rate": 1.9949943229397597e-05, "loss": 0.0657, "step": 400 }, { "epoch": 0.19, "learning_rate": 1.994481713427274e-05, "loss": 0.0542, "step": 410 }, { "epoch": 0.19, "learning_rate": 1.9939441929020823e-05, "loss": 0.0522, "step": 420 }, { "epoch": 0.2, "learning_rate": 1.993381774828668e-05, "loss": 0.0562, "step": 430 }, { "epoch": 0.2, "learning_rate": 1.992794473295175e-05, "loss": 0.0608, "step": 440 }, { "epoch": 0.21, "learning_rate": 1.9921823030130635e-05, "loss": 0.0653, "step": 450 }, { "epoch": 0.21, "learning_rate": 1.9915452793167335e-05, "loss": 0.0623, "step": 460 }, { "epoch": 0.22, "learning_rate": 1.9908834181631455e-05, "loss": 0.0593, "step": 470 }, { "epoch": 0.22, "learning_rate": 1.99019673613142e-05, "loss": 0.0534, "step": 480 }, { "epoch": 0.23, "learning_rate": 1.9894852504224206e-05, "loss": 0.0536, "step": 490 }, { "epoch": 0.23, "learning_rate": 1.988748978858325e-05, "loss": 0.058, "step": 500 }, { "epoch": 0.24, "learning_rate": 1.9879879398821775e-05, "loss": 0.0573, "step": 510 }, { "epoch": 0.24, "learning_rate": 1.9872021525574274e-05, "loss": 0.0589, "step": 520 }, { "epoch": 0.25, "learning_rate": 1.9863916365674506e-05, "loss": 0.0589, "step": 530 }, { "epoch": 0.25, "learning_rate": 1.985556412215059e-05, "loss": 0.0604, "step": 540 }, { "epoch": 0.25, "learning_rate": 1.9846965004219883e-05, "loss": 0.0512, "step": 550 }, { "epoch": 0.26, "learning_rate": 1.983811922728377e-05, "loss": 0.0571, "step": 560 }, { "epoch": 0.26, "learning_rate": 1.9829027012922257e-05, "loss": 0.0599, "step": 570 }, { "epoch": 0.27, "learning_rate": 1.981968858888841e-05, "loss": 0.0493, "step": 580 }, { "epoch": 0.27, "learning_rate": 1.9810104189102683e-05, "loss": 0.0483, "step": 590 }, { "epoch": 0.28, "learning_rate": 1.9800274053647014e-05, "loss": 0.0478, "step": 600 }, { "epoch": 0.28, "learning_rate": 1.979019842875884e-05, "loss": 0.0542, "step": 610 }, { "epoch": 0.29, "learning_rate": 1.977987756682492e-05, "loss": 0.0494, "step": 620 }, { "epoch": 0.29, "learning_rate": 1.9769311726375026e-05, "loss": 0.0503, "step": 630 }, { "epoch": 0.3, "learning_rate": 1.975850117207544e-05, "loss": 0.0549, "step": 640 }, { "epoch": 0.3, "learning_rate": 1.9747446174722357e-05, "loss": 0.0442, "step": 650 }, { "epoch": 0.31, "learning_rate": 1.9736147011235073e-05, "loss": 0.0498, "step": 660 }, { "epoch": 0.31, "learning_rate": 1.972460396464906e-05, "loss": 0.0528, "step": 670 }, { "epoch": 0.31, "learning_rate": 1.9712817324108885e-05, "loss": 0.0552, "step": 680 }, { "epoch": 0.32, "learning_rate": 1.9700787384860957e-05, "loss": 0.0508, "step": 690 }, { "epoch": 0.32, "learning_rate": 1.9688514448246124e-05, "loss": 0.0555, "step": 700 }, { "epoch": 0.33, "learning_rate": 1.9675998821692145e-05, "loss": 0.0523, "step": 710 }, { "epoch": 0.33, "learning_rate": 1.9663240818705968e-05, "loss": 0.0513, "step": 720 }, { "epoch": 0.34, "learning_rate": 1.96502407588659e-05, "loss": 0.0502, "step": 730 }, { "epoch": 0.34, "learning_rate": 1.9636998967813577e-05, "loss": 0.0509, "step": 740 }, { "epoch": 0.35, "learning_rate": 1.9623515777245822e-05, "loss": 0.0526, "step": 750 }, { "epoch": 0.35, "learning_rate": 1.9609791524906333e-05, "loss": 0.0513, "step": 760 }, { "epoch": 0.36, "learning_rate": 1.959582655457722e-05, "loss": 0.0508, "step": 770 }, { "epoch": 0.36, "learning_rate": 1.9581621216070403e-05, "loss": 0.0429, "step": 780 }, { "epoch": 0.37, "learning_rate": 1.956717586521884e-05, "loss": 0.0491, "step": 790 }, { "epoch": 0.37, "learning_rate": 1.9552490863867607e-05, "loss": 0.05, "step": 800 }, { "epoch": 0.38, "learning_rate": 1.953756657986486e-05, "loss": 0.0463, "step": 810 }, { "epoch": 0.38, "learning_rate": 1.9522403387052586e-05, "loss": 0.0529, "step": 820 }, { "epoch": 0.38, "learning_rate": 1.9507001665257265e-05, "loss": 0.048, "step": 830 }, { "epoch": 0.39, "learning_rate": 1.9491361800280355e-05, "loss": 0.0508, "step": 840 }, { "epoch": 0.39, "learning_rate": 1.947548418388861e-05, "loss": 0.0476, "step": 850 }, { "epoch": 0.4, "learning_rate": 1.9459369213804276e-05, "loss": 0.0393, "step": 860 }, { "epoch": 0.4, "learning_rate": 1.9443017293695132e-05, "loss": 0.0549, "step": 870 }, { "epoch": 0.41, "learning_rate": 1.9426428833164376e-05, "loss": 0.0515, "step": 880 }, { "epoch": 0.41, "learning_rate": 1.9409604247740368e-05, "loss": 0.0488, "step": 890 }, { "epoch": 0.42, "learning_rate": 1.9392543958866208e-05, "loss": 0.0522, "step": 900 }, { "epoch": 0.42, "learning_rate": 1.9375248393889198e-05, "loss": 0.0445, "step": 910 }, { "epoch": 0.43, "learning_rate": 1.9357717986050115e-05, "loss": 0.0499, "step": 920 }, { "epoch": 0.43, "learning_rate": 1.9339953174472388e-05, "loss": 0.0467, "step": 930 }, { "epoch": 0.44, "learning_rate": 1.9321954404151074e-05, "loss": 0.0409, "step": 940 }, { "epoch": 0.44, "learning_rate": 1.930372212594172e-05, "loss": 0.0465, "step": 950 }, { "epoch": 0.44, "learning_rate": 1.928525679654906e-05, "loss": 0.0477, "step": 960 }, { "epoch": 0.45, "learning_rate": 1.92665588785156e-05, "loss": 0.0503, "step": 970 }, { "epoch": 0.45, "learning_rate": 1.9247628840210004e-05, "loss": 0.0445, "step": 980 }, { "epoch": 0.46, "learning_rate": 1.922846715581537e-05, "loss": 0.0511, "step": 990 }, { "epoch": 0.46, "learning_rate": 1.9209074305317367e-05, "loss": 0.0438, "step": 1000 }, { "epoch": 0.47, "learning_rate": 1.9189450774492188e-05, "loss": 0.0407, "step": 1010 }, { "epoch": 0.47, "learning_rate": 1.9169597054894398e-05, "loss": 0.0428, "step": 1020 }, { "epoch": 0.48, "learning_rate": 1.9149513643844616e-05, "loss": 0.0471, "step": 1030 }, { "epoch": 0.48, "learning_rate": 1.912920104441706e-05, "loss": 0.0432, "step": 1040 }, { "epoch": 0.49, "learning_rate": 1.9108659765426937e-05, "loss": 0.0441, "step": 1050 }, { "epoch": 0.49, "learning_rate": 1.9087890321417707e-05, "loss": 0.0506, "step": 1060 }, { "epoch": 0.5, "learning_rate": 1.9066893232648197e-05, "loss": 0.0439, "step": 1070 }, { "epoch": 0.5, "learning_rate": 1.904566902507955e-05, "loss": 0.0423, "step": 1080 }, { "epoch": 0.5, "learning_rate": 1.902421823036207e-05, "loss": 0.047, "step": 1090 }, { "epoch": 0.51, "learning_rate": 1.90025413858219e-05, "loss": 0.0442, "step": 1100 }, { "epoch": 0.51, "learning_rate": 1.8980639034447552e-05, "loss": 0.0447, "step": 1110 }, { "epoch": 0.52, "learning_rate": 1.895851172487632e-05, "loss": 0.0464, "step": 1120 }, { "epoch": 0.52, "learning_rate": 1.8936160011380525e-05, "loss": 0.0458, "step": 1130 }, { "epoch": 0.53, "learning_rate": 1.891358445385364e-05, "loss": 0.0457, "step": 1140 }, { "epoch": 0.53, "learning_rate": 1.889078561779626e-05, "loss": 0.0478, "step": 1150 }, { "epoch": 0.54, "learning_rate": 1.8867764074301934e-05, "loss": 0.0499, "step": 1160 }, { "epoch": 0.54, "learning_rate": 1.8844520400042873e-05, "loss": 0.0517, "step": 1170 }, { "epoch": 0.55, "learning_rate": 1.8821055177255483e-05, "loss": 0.0434, "step": 1180 }, { "epoch": 0.55, "learning_rate": 1.879736899372581e-05, "loss": 0.0391, "step": 1190 }, { "epoch": 0.56, "learning_rate": 1.8773462442774772e-05, "loss": 0.0412, "step": 1200 }, { "epoch": 0.56, "learning_rate": 1.874933612324335e-05, "loss": 0.0487, "step": 1210 }, { "epoch": 0.57, "learning_rate": 1.8724990639477546e-05, "loss": 0.0429, "step": 1220 }, { "epoch": 0.57, "learning_rate": 1.8700426601313266e-05, "loss": 0.0427, "step": 1230 }, { "epoch": 0.57, "learning_rate": 1.867564462406104e-05, "loss": 0.0468, "step": 1240 }, { "epoch": 0.58, "learning_rate": 1.86506453284906e-05, "loss": 0.0417, "step": 1250 }, { "epoch": 0.58, "learning_rate": 1.862542934081534e-05, "loss": 0.0399, "step": 1260 }, { "epoch": 0.59, "learning_rate": 1.859999729267663e-05, "loss": 0.046, "step": 1270 }, { "epoch": 0.59, "learning_rate": 1.8574349821127986e-05, "loss": 0.0409, "step": 1280 }, { "epoch": 0.6, "learning_rate": 1.854848756861912e-05, "loss": 0.048, "step": 1290 }, { "epoch": 0.6, "learning_rate": 1.8522411182979846e-05, "loss": 0.0411, "step": 1300 }, { "epoch": 0.61, "learning_rate": 1.8496121317403843e-05, "loss": 0.04, "step": 1310 }, { "epoch": 0.61, "learning_rate": 1.8469618630432307e-05, "loss": 0.0455, "step": 1320 }, { "epoch": 0.62, "learning_rate": 1.844290378593745e-05, "loss": 0.0452, "step": 1330 }, { "epoch": 0.62, "learning_rate": 1.8415977453105864e-05, "loss": 0.0378, "step": 1340 }, { "epoch": 0.63, "learning_rate": 1.8388840306421763e-05, "loss": 0.0396, "step": 1350 }, { "epoch": 0.63, "learning_rate": 1.8361493025650097e-05, "loss": 0.0434, "step": 1360 }, { "epoch": 0.63, "learning_rate": 1.8333936295819504e-05, "loss": 0.0402, "step": 1370 }, { "epoch": 0.64, "learning_rate": 1.8306170807205162e-05, "loss": 0.0402, "step": 1380 }, { "epoch": 0.64, "learning_rate": 1.8278197255311515e-05, "loss": 0.0377, "step": 1390 }, { "epoch": 0.65, "learning_rate": 1.825001634085482e-05, "loss": 0.0389, "step": 1400 }, { "epoch": 0.65, "learning_rate": 1.8221628769745604e-05, "loss": 0.0442, "step": 1410 }, { "epoch": 0.66, "learning_rate": 1.8193035253071004e-05, "loss": 0.0428, "step": 1420 }, { "epoch": 0.66, "learning_rate": 1.8164236507076917e-05, "loss": 0.0398, "step": 1430 }, { "epoch": 0.67, "learning_rate": 1.8135233253150097e-05, "loss": 0.0383, "step": 1440 }, { "epoch": 0.67, "learning_rate": 1.810602621780006e-05, "loss": 0.0405, "step": 1450 }, { "epoch": 0.68, "learning_rate": 1.807661613264088e-05, "loss": 0.0379, "step": 1460 }, { "epoch": 0.68, "learning_rate": 1.80470037343729e-05, "loss": 0.0473, "step": 1470 }, { "epoch": 0.69, "learning_rate": 1.8017189764764226e-05, "loss": 0.0398, "step": 1480 }, { "epoch": 0.69, "learning_rate": 1.7987174970632197e-05, "loss": 0.0406, "step": 1490 }, { "epoch": 0.69, "learning_rate": 1.795696010382463e-05, "loss": 0.0406, "step": 1500 }, { "epoch": 0.7, "learning_rate": 1.7926545921201038e-05, "loss": 0.0444, "step": 1510 }, { "epoch": 0.7, "learning_rate": 1.789593318461362e-05, "loss": 0.0461, "step": 1520 }, { "epoch": 0.71, "learning_rate": 1.7865122660888207e-05, "loss": 0.0471, "step": 1530 }, { "epoch": 0.71, "learning_rate": 1.7834115121805063e-05, "loss": 0.0424, "step": 1540 }, { "epoch": 0.72, "learning_rate": 1.780291134407952e-05, "loss": 0.0488, "step": 1550 }, { "epoch": 0.72, "learning_rate": 1.7771512109342545e-05, "loss": 0.0412, "step": 1560 }, { "epoch": 0.73, "learning_rate": 1.7739918204121153e-05, "loss": 0.0409, "step": 1570 }, { "epoch": 0.73, "learning_rate": 1.7708130419818713e-05, "loss": 0.0421, "step": 1580 }, { "epoch": 0.74, "learning_rate": 1.767614955269511e-05, "loss": 0.039, "step": 1590 }, { "epoch": 0.74, "learning_rate": 1.7643976403846817e-05, "loss": 0.0422, "step": 1600 }, { "epoch": 0.75, "learning_rate": 1.7611611779186804e-05, "loss": 0.0393, "step": 1610 }, { "epoch": 0.75, "learning_rate": 1.757905648942438e-05, "loss": 0.0426, "step": 1620 }, { "epoch": 0.75, "learning_rate": 1.7546311350044865e-05, "loss": 0.0417, "step": 1630 }, { "epoch": 0.76, "learning_rate": 1.751337718128916e-05, "loss": 0.0433, "step": 1640 }, { "epoch": 0.76, "learning_rate": 1.748025480813322e-05, "loss": 0.0412, "step": 1650 }, { "epoch": 0.77, "learning_rate": 1.7446945060267367e-05, "loss": 0.0446, "step": 1660 }, { "epoch": 0.77, "learning_rate": 1.7413448772075532e-05, "loss": 0.0423, "step": 1670 }, { "epoch": 0.78, "learning_rate": 1.7379766782614323e-05, "loss": 0.0396, "step": 1680 }, { "epoch": 0.78, "learning_rate": 1.734589993559203e-05, "loss": 0.0391, "step": 1690 }, { "epoch": 0.79, "learning_rate": 1.7311849079347495e-05, "loss": 0.0376, "step": 1700 }, { "epoch": 0.79, "learning_rate": 1.7277615066828836e-05, "loss": 0.0405, "step": 1710 }, { "epoch": 0.8, "learning_rate": 1.7243198755572104e-05, "loss": 0.0424, "step": 1720 }, { "epoch": 0.8, "learning_rate": 1.7208601007679796e-05, "loss": 0.0406, "step": 1730 }, { "epoch": 0.81, "learning_rate": 1.7173822689799256e-05, "loss": 0.0382, "step": 1740 }, { "epoch": 0.81, "learning_rate": 1.7138864673100974e-05, "loss": 0.0402, "step": 1750 }, { "epoch": 0.82, "learning_rate": 1.7103727833256757e-05, "loss": 0.0453, "step": 1760 }, { "epoch": 0.82, "learning_rate": 1.706841305041779e-05, "loss": 0.0404, "step": 1770 }, { "epoch": 0.82, "learning_rate": 1.7032921209192603e-05, "loss": 0.0445, "step": 1780 }, { "epoch": 0.83, "learning_rate": 1.6997253198624905e-05, "loss": 0.0424, "step": 1790 }, { "epoch": 0.83, "learning_rate": 1.69614099121713e-05, "loss": 0.0378, "step": 1800 }, { "epoch": 0.84, "learning_rate": 1.6925392247678943e-05, "loss": 0.0447, "step": 1810 }, { "epoch": 0.84, "learning_rate": 1.688920110736301e-05, "loss": 0.0413, "step": 1820 }, { "epoch": 0.85, "learning_rate": 1.6852837397784108e-05, "loss": 0.0385, "step": 1830 }, { "epoch": 0.85, "learning_rate": 1.6816302029825594e-05, "loss": 0.0389, "step": 1840 }, { "epoch": 0.86, "learning_rate": 1.677959591867073e-05, "loss": 0.0393, "step": 1850 }, { "epoch": 0.86, "learning_rate": 1.6742719983779752e-05, "loss": 0.0401, "step": 1860 }, { "epoch": 0.87, "learning_rate": 1.670567514886687e-05, "loss": 0.0409, "step": 1870 }, { "epoch": 0.87, "learning_rate": 1.6668462341877098e-05, "loss": 0.0421, "step": 1880 }, { "epoch": 0.88, "learning_rate": 1.6631082494963037e-05, "loss": 0.0415, "step": 1890 }, { "epoch": 0.88, "learning_rate": 1.6593536544461494e-05, "loss": 0.0421, "step": 1900 }, { "epoch": 0.88, "learning_rate": 1.655582543087006e-05, "loss": 0.0419, "step": 1910 }, { "epoch": 0.89, "learning_rate": 1.651795009882353e-05, "loss": 0.0451, "step": 1920 }, { "epoch": 0.89, "learning_rate": 1.6479911497070245e-05, "loss": 0.0422, "step": 1930 }, { "epoch": 0.9, "learning_rate": 1.644171057844833e-05, "loss": 0.0438, "step": 1940 }, { "epoch": 0.9, "learning_rate": 1.6403348299861827e-05, "loss": 0.0396, "step": 1950 }, { "epoch": 0.91, "learning_rate": 1.636482562225672e-05, "loss": 0.0377, "step": 1960 }, { "epoch": 0.91, "learning_rate": 1.6326143510596865e-05, "loss": 0.0362, "step": 1970 }, { "epoch": 0.92, "learning_rate": 1.6287302933839826e-05, "loss": 0.0442, "step": 1980 }, { "epoch": 0.92, "learning_rate": 1.6248304864912592e-05, "loss": 0.0397, "step": 1990 }, { "epoch": 0.93, "learning_rate": 1.6209150280687217e-05, "loss": 0.0389, "step": 2000 }, { "epoch": 0.93, "learning_rate": 1.6169840161956333e-05, "loss": 0.0382, "step": 2010 }, { "epoch": 0.94, "learning_rate": 1.6130375493408613e-05, "loss": 0.0435, "step": 2020 }, { "epoch": 0.94, "learning_rate": 1.609075726360406e-05, "loss": 0.0399, "step": 2030 }, { "epoch": 0.94, "learning_rate": 1.6050986464949298e-05, "loss": 0.042, "step": 2040 }, { "epoch": 0.95, "learning_rate": 1.601106409367266e-05, "loss": 0.0394, "step": 2050 }, { "epoch": 0.95, "learning_rate": 1.5970991149799276e-05, "loss": 0.0399, "step": 2060 }, { "epoch": 0.96, "learning_rate": 1.5930768637125995e-05, "loss": 0.0424, "step": 2070 }, { "epoch": 0.96, "learning_rate": 1.5890397563196257e-05, "loss": 0.0416, "step": 2080 }, { "epoch": 0.97, "learning_rate": 1.5849878939274844e-05, "loss": 0.0426, "step": 2090 }, { "epoch": 0.97, "learning_rate": 1.580921378032256e-05, "loss": 0.0376, "step": 2100 }, { "epoch": 0.98, "learning_rate": 1.5768403104970792e-05, "loss": 0.0433, "step": 2110 }, { "epoch": 0.98, "learning_rate": 1.572744793549601e-05, "loss": 0.038, "step": 2120 }, { "epoch": 0.99, "learning_rate": 1.5686349297794148e-05, "loss": 0.0412, "step": 2130 }, { "epoch": 0.99, "learning_rate": 1.564510822135491e-05, "loss": 0.0383, "step": 2140 }, { "epoch": 1.0, "learning_rate": 1.5603725739235983e-05, "loss": 0.042, "step": 2150 }, { "epoch": 1.0, "eval_em": 0.6256, "eval_em_ic": 0.9328, "eval_f1": 0.8165, "eval_gen_len": 39.4971, "eval_levinstein_ratio": 0.9954, "eval_loss": 0.050038907676935196, "eval_precision": 0.9175, "eval_recall": 0.8452, "eval_runtime": 170.0158, "eval_samples_per_second": 10.07, "eval_steps_per_second": 0.159, "step": 2159 }, { "epoch": 1.0, "learning_rate": 1.556220288803716e-05, "loss": 0.0385, "step": 2160 }, { "epoch": 1.01, "learning_rate": 1.5520540707874373e-05, "loss": 0.0375, "step": 2170 }, { "epoch": 1.01, "learning_rate": 1.5478740242353644e-05, "loss": 0.0375, "step": 2180 }, { "epoch": 1.01, "learning_rate": 1.543680253854492e-05, "loss": 0.0374, "step": 2190 }, { "epoch": 1.02, "learning_rate": 1.5394728646955884e-05, "loss": 0.0407, "step": 2200 }, { "epoch": 1.02, "learning_rate": 1.535251962150562e-05, "loss": 0.0422, "step": 2210 }, { "epoch": 1.03, "learning_rate": 1.5310176519498187e-05, "loss": 0.0356, "step": 2220 }, { "epoch": 1.03, "learning_rate": 1.5267700401596185e-05, "loss": 0.0408, "step": 2230 }, { "epoch": 1.04, "learning_rate": 1.522509233179415e-05, "loss": 0.0383, "step": 2240 }, { "epoch": 1.04, "learning_rate": 1.5182353377391925e-05, "loss": 0.0386, "step": 2250 }, { "epoch": 1.05, "learning_rate": 1.5139484608967901e-05, "loss": 0.0381, "step": 2260 }, { "epoch": 1.05, "learning_rate": 1.5096487100352208e-05, "loss": 0.0404, "step": 2270 }, { "epoch": 1.06, "learning_rate": 1.5053361928599837e-05, "loss": 0.0364, "step": 2280 }, { "epoch": 1.06, "learning_rate": 1.5010110173963633e-05, "loss": 0.0388, "step": 2290 }, { "epoch": 1.07, "learning_rate": 1.496673291986724e-05, "loss": 0.0405, "step": 2300 }, { "epoch": 1.07, "learning_rate": 1.4923231252877983e-05, "loss": 0.0353, "step": 2310 }, { "epoch": 1.07, "learning_rate": 1.4879606262679625e-05, "loss": 0.0388, "step": 2320 }, { "epoch": 1.08, "learning_rate": 1.4835859042045088e-05, "loss": 0.0404, "step": 2330 }, { "epoch": 1.08, "learning_rate": 1.4791990686809068e-05, "loss": 0.0373, "step": 2340 }, { "epoch": 1.09, "learning_rate": 1.4748002295840593e-05, "loss": 0.0382, "step": 2350 }, { "epoch": 1.09, "learning_rate": 1.4703894971015503e-05, "loss": 0.0347, "step": 2360 }, { "epoch": 1.1, "learning_rate": 1.4659669817188824e-05, "loss": 0.0388, "step": 2370 }, { "epoch": 1.1, "learning_rate": 1.4615327942167125e-05, "loss": 0.0342, "step": 2380 }, { "epoch": 1.11, "learning_rate": 1.4570870456680745e-05, "loss": 0.0389, "step": 2390 }, { "epoch": 1.11, "learning_rate": 1.452629847435598e-05, "loss": 0.0373, "step": 2400 }, { "epoch": 1.12, "learning_rate": 1.4481613111687182e-05, "loss": 0.0348, "step": 2410 }, { "epoch": 1.12, "learning_rate": 1.4436815488008798e-05, "loss": 0.0418, "step": 2420 }, { "epoch": 1.13, "learning_rate": 1.4391906725467324e-05, "loss": 0.0362, "step": 2430 }, { "epoch": 1.13, "learning_rate": 1.4346887948993206e-05, "loss": 0.0426, "step": 2440 }, { "epoch": 1.13, "learning_rate": 1.430176028627265e-05, "loss": 0.0379, "step": 2450 }, { "epoch": 1.14, "learning_rate": 1.4256524867719386e-05, "loss": 0.0352, "step": 2460 }, { "epoch": 1.14, "learning_rate": 1.4211182826446334e-05, "loss": 0.0356, "step": 2470 }, { "epoch": 1.15, "learning_rate": 1.4165735298237251e-05, "loss": 0.0369, "step": 2480 }, { "epoch": 1.15, "learning_rate": 1.412018342151824e-05, "loss": 0.0333, "step": 2490 }, { "epoch": 1.16, "learning_rate": 1.4074528337329276e-05, "loss": 0.0333, "step": 2500 }, { "epoch": 1.16, "learning_rate": 1.4028771189295593e-05, "loss": 0.04, "step": 2510 }, { "epoch": 1.17, "learning_rate": 1.3982913123599038e-05, "loss": 0.0372, "step": 2520 }, { "epoch": 1.17, "learning_rate": 1.3936955288949383e-05, "loss": 0.0405, "step": 2530 }, { "epoch": 1.18, "learning_rate": 1.3890898836555532e-05, "loss": 0.0368, "step": 2540 }, { "epoch": 1.18, "learning_rate": 1.3844744920096691e-05, "loss": 0.0378, "step": 2550 }, { "epoch": 1.19, "learning_rate": 1.3798494695693456e-05, "loss": 0.0357, "step": 2560 }, { "epoch": 1.19, "learning_rate": 1.375214932187888e-05, "loss": 0.0343, "step": 2570 }, { "epoch": 1.19, "learning_rate": 1.3705709959569422e-05, "loss": 0.0322, "step": 2580 }, { "epoch": 1.2, "learning_rate": 1.3659177772035896e-05, "loss": 0.041, "step": 2590 }, { "epoch": 1.2, "learning_rate": 1.3612553924874305e-05, "loss": 0.0351, "step": 2600 }, { "epoch": 1.21, "learning_rate": 1.356583958597666e-05, "loss": 0.0372, "step": 2610 }, { "epoch": 1.21, "learning_rate": 1.3519035925501722e-05, "loss": 0.0375, "step": 2620 }, { "epoch": 1.22, "learning_rate": 1.3472144115845694e-05, "loss": 0.0344, "step": 2630 }, { "epoch": 1.22, "learning_rate": 1.3425165331612839e-05, "loss": 0.0388, "step": 2640 }, { "epoch": 1.23, "learning_rate": 1.337810074958608e-05, "loss": 0.0429, "step": 2650 }, { "epoch": 1.23, "learning_rate": 1.33309515486975e-05, "loss": 0.0351, "step": 2660 }, { "epoch": 1.24, "learning_rate": 1.3283718909998822e-05, "loss": 0.0412, "step": 2670 }, { "epoch": 1.24, "learning_rate": 1.323640401663182e-05, "loss": 0.0411, "step": 2680 }, { "epoch": 1.25, "learning_rate": 1.31890080537987e-05, "loss": 0.0376, "step": 2690 }, { "epoch": 1.25, "learning_rate": 1.3141532208732381e-05, "loss": 0.0348, "step": 2700 }, { "epoch": 1.26, "learning_rate": 1.3093977670666772e-05, "loss": 0.0411, "step": 2710 }, { "epoch": 1.26, "learning_rate": 1.304634563080699e-05, "loss": 0.0359, "step": 2720 }, { "epoch": 1.26, "learning_rate": 1.2998637282299507e-05, "loss": 0.0414, "step": 2730 }, { "epoch": 1.27, "learning_rate": 1.2950853820202274e-05, "loss": 0.0377, "step": 2740 }, { "epoch": 1.27, "learning_rate": 1.2902996441454775e-05, "loss": 0.035, "step": 2750 }, { "epoch": 1.28, "learning_rate": 1.2855066344848055e-05, "loss": 0.0352, "step": 2760 }, { "epoch": 1.28, "learning_rate": 1.2807064730994685e-05, "loss": 0.0371, "step": 2770 }, { "epoch": 1.29, "learning_rate": 1.275899280229868e-05, "loss": 0.0371, "step": 2780 }, { "epoch": 1.29, "learning_rate": 1.2710851762925403e-05, "loss": 0.0357, "step": 2790 }, { "epoch": 1.3, "learning_rate": 1.2662642818771377e-05, "loss": 0.0382, "step": 2800 }, { "epoch": 1.3, "learning_rate": 1.2614367177434094e-05, "loss": 0.0403, "step": 2810 }, { "epoch": 1.31, "learning_rate": 1.2566026048181756e-05, "loss": 0.0365, "step": 2820 }, { "epoch": 1.31, "learning_rate": 1.2517620641922986e-05, "loss": 0.0406, "step": 2830 }, { "epoch": 1.32, "learning_rate": 1.2469152171176503e-05, "loss": 0.0445, "step": 2840 }, { "epoch": 1.32, "learning_rate": 1.2420621850040741e-05, "loss": 0.0375, "step": 2850 }, { "epoch": 1.32, "learning_rate": 1.2372030894163437e-05, "loss": 0.0401, "step": 2860 }, { "epoch": 1.33, "learning_rate": 1.2323380520711185e-05, "loss": 0.0345, "step": 2870 }, { "epoch": 1.33, "learning_rate": 1.2274671948338947e-05, "loss": 0.0341, "step": 2880 }, { "epoch": 1.34, "learning_rate": 1.2225906397159518e-05, "loss": 0.0367, "step": 2890 }, { "epoch": 1.34, "learning_rate": 1.2177085088712965e-05, "loss": 0.0385, "step": 2900 }, { "epoch": 1.35, "learning_rate": 1.2128209245936045e-05, "loss": 0.0376, "step": 2910 }, { "epoch": 1.35, "learning_rate": 1.2079280093131554e-05, "loss": 0.0334, "step": 2920 }, { "epoch": 1.36, "learning_rate": 1.2030298855937654e-05, "loss": 0.0353, "step": 2930 }, { "epoch": 1.36, "learning_rate": 1.1981266761297196e-05, "loss": 0.0368, "step": 2940 }, { "epoch": 1.37, "learning_rate": 1.1932185037426964e-05, "loss": 0.0405, "step": 2950 }, { "epoch": 1.37, "learning_rate": 1.1883054913786915e-05, "loss": 0.0341, "step": 2960 }, { "epoch": 1.38, "learning_rate": 1.1833877621049386e-05, "loss": 0.0387, "step": 2970 }, { "epoch": 1.38, "learning_rate": 1.1784654391068264e-05, "loss": 0.0354, "step": 2980 }, { "epoch": 1.38, "learning_rate": 1.1735386456848134e-05, "loss": 0.0399, "step": 2990 }, { "epoch": 1.39, "learning_rate": 1.1686075052513378e-05, "loss": 0.0326, "step": 3000 }, { "epoch": 1.39, "learning_rate": 1.1636721413277274e-05, "loss": 0.0379, "step": 3010 }, { "epoch": 1.4, "learning_rate": 1.1587326775411061e-05, "loss": 0.0382, "step": 3020 }, { "epoch": 1.4, "learning_rate": 1.1537892376212956e-05, "loss": 0.0371, "step": 3030 }, { "epoch": 1.41, "learning_rate": 1.1488419453977166e-05, "loss": 0.0365, "step": 3040 }, { "epoch": 1.41, "learning_rate": 1.143890924796287e-05, "loss": 0.0361, "step": 3050 }, { "epoch": 1.42, "learning_rate": 1.1389362998363187e-05, "loss": 0.0366, "step": 3060 }, { "epoch": 1.42, "learning_rate": 1.133978194627409e-05, "loss": 0.0368, "step": 3070 }, { "epoch": 1.43, "learning_rate": 1.1290167333663326e-05, "loss": 0.0375, "step": 3080 }, { "epoch": 1.43, "learning_rate": 1.124052040333932e-05, "loss": 0.0338, "step": 3090 }, { "epoch": 1.44, "learning_rate": 1.1190842398920026e-05, "loss": 0.041, "step": 3100 }, { "epoch": 1.44, "learning_rate": 1.1141134564801775e-05, "loss": 0.034, "step": 3110 }, { "epoch": 1.45, "learning_rate": 1.109139814612811e-05, "loss": 0.038, "step": 3120 }, { "epoch": 1.45, "learning_rate": 1.1041634388758601e-05, "loss": 0.0397, "step": 3130 }, { "epoch": 1.45, "learning_rate": 1.0991844539237635e-05, "loss": 0.0326, "step": 3140 }, { "epoch": 1.46, "learning_rate": 1.0942029844763177e-05, "loss": 0.0377, "step": 3150 }, { "epoch": 1.46, "learning_rate": 1.0892191553155543e-05, "loss": 0.0361, "step": 3160 }, { "epoch": 1.47, "learning_rate": 1.0842330912826138e-05, "loss": 0.0343, "step": 3170 }, { "epoch": 1.47, "learning_rate": 1.0792449172746197e-05, "loss": 0.0384, "step": 3180 }, { "epoch": 1.48, "learning_rate": 1.0742547582415473e-05, "loss": 0.0336, "step": 3190 }, { "epoch": 1.48, "learning_rate": 1.0692627391830954e-05, "loss": 0.0339, "step": 3200 }, { "epoch": 1.49, "learning_rate": 1.0642689851455561e-05, "loss": 0.0369, "step": 3210 }, { "epoch": 1.49, "learning_rate": 1.0592736212186808e-05, "loss": 0.0338, "step": 3220 }, { "epoch": 1.5, "learning_rate": 1.0542767725325471e-05, "loss": 0.0346, "step": 3230 }, { "epoch": 1.5, "learning_rate": 1.0492785642544248e-05, "loss": 0.0349, "step": 3240 }, { "epoch": 1.51, "learning_rate": 1.0442791215856409e-05, "loss": 0.0402, "step": 3250 }, { "epoch": 1.51, "learning_rate": 1.0392785697584428e-05, "loss": 0.039, "step": 3260 }, { "epoch": 1.51, "learning_rate": 1.0342770340328609e-05, "loss": 0.0414, "step": 3270 }, { "epoch": 1.52, "learning_rate": 1.0292746396935727e-05, "loss": 0.0358, "step": 3280 }, { "epoch": 1.52, "learning_rate": 1.0242715120467616e-05, "loss": 0.0377, "step": 3290 }, { "epoch": 1.53, "learning_rate": 1.0192677764169817e-05, "loss": 0.0347, "step": 3300 }, { "epoch": 1.53, "learning_rate": 1.0142635581440148e-05, "loss": 0.0316, "step": 3310 }, { "epoch": 1.54, "learning_rate": 1.009258982579734e-05, "loss": 0.0404, "step": 3320 }, { "epoch": 1.54, "learning_rate": 1.004254175084962e-05, "loss": 0.0368, "step": 3330 }, { "epoch": 1.55, "learning_rate": 9.992492610263299e-06, "loss": 0.037, "step": 3340 }, { "epoch": 1.55, "learning_rate": 9.942443657731405e-06, "loss": 0.0338, "step": 3350 }, { "epoch": 1.56, "learning_rate": 9.892396146942228e-06, "loss": 0.0384, "step": 3360 }, { "epoch": 1.56, "learning_rate": 9.842351331547964e-06, "loss": 0.0398, "step": 3370 }, { "epoch": 1.57, "learning_rate": 9.792310465133284e-06, "loss": 0.0346, "step": 3380 }, { "epoch": 1.57, "learning_rate": 9.74227480118394e-06, "loss": 0.0342, "step": 3390 }, { "epoch": 1.57, "learning_rate": 9.692245593055368e-06, "loss": 0.039, "step": 3400 }, { "epoch": 1.58, "learning_rate": 9.64222409394129e-06, "loss": 0.0363, "step": 3410 }, { "epoch": 1.58, "learning_rate": 9.592211556842319e-06, "loss": 0.0359, "step": 3420 }, { "epoch": 1.59, "learning_rate": 9.542209234534581e-06, "loss": 0.0378, "step": 3430 }, { "epoch": 1.59, "learning_rate": 9.492218379538331e-06, "loss": 0.0317, "step": 3440 }, { "epoch": 1.6, "learning_rate": 9.442240244086573e-06, "loss": 0.0369, "step": 3450 }, { "epoch": 1.6, "learning_rate": 9.392276080093693e-06, "loss": 0.0387, "step": 3460 }, { "epoch": 1.61, "learning_rate": 9.342327139124107e-06, "loss": 0.0345, "step": 3470 }, { "epoch": 1.61, "learning_rate": 9.292394672360906e-06, "loss": 0.0343, "step": 3480 }, { "epoch": 1.62, "learning_rate": 9.242479930574515e-06, "loss": 0.0389, "step": 3490 }, { "epoch": 1.62, "learning_rate": 9.192584164091358e-06, "loss": 0.035, "step": 3500 }, { "epoch": 1.63, "learning_rate": 9.142708622762545e-06, "loss": 0.0357, "step": 3510 }, { "epoch": 1.63, "learning_rate": 9.092854555932561e-06, "loss": 0.035, "step": 3520 }, { "epoch": 1.64, "learning_rate": 9.043023212407971e-06, "loss": 0.0373, "step": 3530 }, { "epoch": 1.64, "learning_rate": 8.993215840426138e-06, "loss": 0.0334, "step": 3540 }, { "epoch": 1.64, "learning_rate": 8.94343368762396e-06, "loss": 0.0366, "step": 3550 }, { "epoch": 1.65, "learning_rate": 8.893678001006601e-06, "loss": 0.0377, "step": 3560 }, { "epoch": 1.65, "learning_rate": 8.843950026916283e-06, "loss": 0.0295, "step": 3570 }, { "epoch": 1.66, "learning_rate": 8.794251011001038e-06, "loss": 0.0358, "step": 3580 }, { "epoch": 1.66, "learning_rate": 8.744582198183528e-06, "loss": 0.0357, "step": 3590 }, { "epoch": 1.67, "learning_rate": 8.694944832629844e-06, "loss": 0.0328, "step": 3600 }, { "epoch": 1.67, "learning_rate": 8.645340157718346e-06, "loss": 0.0377, "step": 3610 }, { "epoch": 1.68, "learning_rate": 8.59576941600852e-06, "loss": 0.0342, "step": 3620 }, { "epoch": 1.68, "learning_rate": 8.546233849209856e-06, "loss": 0.0402, "step": 3630 }, { "epoch": 1.69, "learning_rate": 8.49673469815073e-06, "loss": 0.0378, "step": 3640 }, { "epoch": 1.69, "learning_rate": 8.447273202747337e-06, "loss": 0.0388, "step": 3650 }, { "epoch": 1.7, "learning_rate": 8.397850601972625e-06, "loss": 0.0372, "step": 3660 }, { "epoch": 1.7, "learning_rate": 8.34846813382526e-06, "loss": 0.0356, "step": 3670 }, { "epoch": 1.7, "learning_rate": 8.299127035298621e-06, "loss": 0.039, "step": 3680 }, { "epoch": 1.71, "learning_rate": 8.249828542349802e-06, "loss": 0.0329, "step": 3690 }, { "epoch": 1.71, "learning_rate": 8.200573889868666e-06, "loss": 0.036, "step": 3700 }, { "epoch": 1.72, "learning_rate": 8.151364311646902e-06, "loss": 0.0385, "step": 3710 }, { "epoch": 1.72, "learning_rate": 8.102201040347123e-06, "loss": 0.0361, "step": 3720 }, { "epoch": 1.73, "learning_rate": 8.053085307471989e-06, "loss": 0.0361, "step": 3730 }, { "epoch": 1.73, "learning_rate": 8.004018343333366e-06, "loss": 0.0309, "step": 3740 }, { "epoch": 1.74, "learning_rate": 7.955001377021484e-06, "loss": 0.0395, "step": 3750 }, { "epoch": 1.74, "learning_rate": 7.906035636374177e-06, "loss": 0.034, "step": 3760 }, { "epoch": 1.75, "learning_rate": 7.857122347946113e-06, "loss": 0.0354, "step": 3770 }, { "epoch": 1.75, "learning_rate": 7.808262736978074e-06, "loss": 0.0316, "step": 3780 }, { "epoch": 1.76, "learning_rate": 7.759458027366258e-06, "loss": 0.0374, "step": 3790 }, { "epoch": 1.76, "learning_rate": 7.710709441631624e-06, "loss": 0.0356, "step": 3800 }, { "epoch": 1.76, "learning_rate": 7.662018200889276e-06, "loss": 0.041, "step": 3810 }, { "epoch": 1.77, "learning_rate": 7.6133855248178715e-06, "loss": 0.0342, "step": 3820 }, { "epoch": 1.77, "learning_rate": 7.564812631629055e-06, "loss": 0.0334, "step": 3830 }, { "epoch": 1.78, "learning_rate": 7.516300738036972e-06, "loss": 0.0356, "step": 3840 }, { "epoch": 1.78, "learning_rate": 7.4678510592277576e-06, "loss": 0.0367, "step": 3850 }, { "epoch": 1.79, "learning_rate": 7.419464808829123e-06, "loss": 0.0311, "step": 3860 }, { "epoch": 1.79, "learning_rate": 7.3711431988799485e-06, "loss": 0.0398, "step": 3870 }, { "epoch": 1.8, "learning_rate": 7.322887439799913e-06, "loss": 0.0334, "step": 3880 }, { "epoch": 1.8, "learning_rate": 7.274698740359184e-06, "loss": 0.04, "step": 3890 }, { "epoch": 1.81, "learning_rate": 7.22657830764814e-06, "loss": 0.0352, "step": 3900 }, { "epoch": 1.81, "learning_rate": 7.17852734704712e-06, "loss": 0.0401, "step": 3910 }, { "epoch": 1.82, "learning_rate": 7.1305470621962495e-06, "loss": 0.0356, "step": 3920 }, { "epoch": 1.82, "learning_rate": 7.082638654965277e-06, "loss": 0.0357, "step": 3930 }, { "epoch": 1.82, "learning_rate": 7.034803325423465e-06, "loss": 0.0391, "step": 3940 }, { "epoch": 1.83, "learning_rate": 6.987042271809541e-06, "loss": 0.0334, "step": 3950 }, { "epoch": 1.83, "learning_rate": 6.9393566905016775e-06, "loss": 0.0317, "step": 3960 }, { "epoch": 1.84, "learning_rate": 6.891747775987521e-06, "loss": 0.0347, "step": 3970 }, { "epoch": 1.84, "learning_rate": 6.844216720834276e-06, "loss": 0.0321, "step": 3980 }, { "epoch": 1.85, "learning_rate": 6.796764715658821e-06, "loss": 0.0342, "step": 3990 }, { "epoch": 1.85, "learning_rate": 6.749392949097904e-06, "loss": 0.0385, "step": 4000 }, { "epoch": 1.86, "learning_rate": 6.702102607778347e-06, "loss": 0.0325, "step": 4010 }, { "epoch": 1.86, "learning_rate": 6.654894876287335e-06, "loss": 0.0337, "step": 4020 }, { "epoch": 1.87, "learning_rate": 6.607770937142743e-06, "loss": 0.0383, "step": 4030 }, { "epoch": 1.87, "learning_rate": 6.560731970763503e-06, "loss": 0.0366, "step": 4040 }, { "epoch": 1.88, "learning_rate": 6.51377915544005e-06, "loss": 0.0357, "step": 4050 }, { "epoch": 1.88, "learning_rate": 6.4669136673047985e-06, "loss": 0.0363, "step": 4060 }, { "epoch": 1.89, "learning_rate": 6.420136680302683e-06, "loss": 0.0401, "step": 4070 }, { "epoch": 1.89, "learning_rate": 6.373449366161755e-06, "loss": 0.0345, "step": 4080 }, { "epoch": 1.89, "learning_rate": 6.326852894363824e-06, "loss": 0.04, "step": 4090 }, { "epoch": 1.9, "learning_rate": 6.280348432115169e-06, "loss": 0.0338, "step": 4100 }, { "epoch": 1.9, "learning_rate": 6.233937144317302e-06, "loss": 0.0384, "step": 4110 }, { "epoch": 1.91, "learning_rate": 6.1876201935377835e-06, "loss": 0.038, "step": 4120 }, { "epoch": 1.91, "learning_rate": 6.1413987399811055e-06, "loss": 0.0406, "step": 4130 }, { "epoch": 1.92, "learning_rate": 6.0952739414596205e-06, "loss": 0.0334, "step": 4140 }, { "epoch": 1.92, "learning_rate": 6.049246953364551e-06, "loss": 0.0343, "step": 4150 }, { "epoch": 1.93, "learning_rate": 6.003318928637045e-06, "loss": 0.0329, "step": 4160 }, { "epoch": 1.93, "learning_rate": 5.957491017739292e-06, "loss": 0.0405, "step": 4170 }, { "epoch": 1.94, "learning_rate": 5.911764368625697e-06, "loss": 0.0334, "step": 4180 }, { "epoch": 1.94, "learning_rate": 5.866140126714146e-06, "loss": 0.0414, "step": 4190 }, { "epoch": 1.95, "learning_rate": 5.820619434857295e-06, "loss": 0.035, "step": 4200 }, { "epoch": 1.95, "learning_rate": 5.775203433313956e-06, "loss": 0.0366, "step": 4210 }, { "epoch": 1.95, "learning_rate": 5.729893259720523e-06, "loss": 0.0312, "step": 4220 }, { "epoch": 1.96, "learning_rate": 5.684690049062478e-06, "loss": 0.0366, "step": 4230 }, { "epoch": 1.96, "learning_rate": 5.639594933645969e-06, "loss": 0.0341, "step": 4240 }, { "epoch": 1.97, "learning_rate": 5.594609043069431e-06, "loss": 0.0318, "step": 4250 }, { "epoch": 1.97, "learning_rate": 5.549733504195314e-06, "loss": 0.037, "step": 4260 }, { "epoch": 1.98, "learning_rate": 5.50496944112183e-06, "loss": 0.0347, "step": 4270 }, { "epoch": 1.98, "learning_rate": 5.460317975154805e-06, "loss": 0.038, "step": 4280 }, { "epoch": 1.99, "learning_rate": 5.415780224779605e-06, "loss": 0.0318, "step": 4290 }, { "epoch": 1.99, "learning_rate": 5.371357305633098e-06, "loss": 0.0336, "step": 4300 }, { "epoch": 2.0, "learning_rate": 5.327050330475712e-06, "loss": 0.0342, "step": 4310 }, { "epoch": 2.0, "eval_em": 0.6361, "eval_em_ic": 0.9334, "eval_f1": 0.8229, "eval_gen_len": 39.5158, "eval_levinstein_ratio": 0.9955, "eval_loss": 0.049182891845703125, "eval_precision": 0.9159, "eval_recall": 0.8546, "eval_runtime": 165.2605, "eval_samples_per_second": 10.359, "eval_steps_per_second": 0.163, "step": 4318 }, { "epoch": 2.0, "learning_rate": 5.282860409163584e-06, "loss": 0.0371, "step": 4320 }, { "epoch": 2.01, "learning_rate": 5.238788648620714e-06, "loss": 0.0321, "step": 4330 }, { "epoch": 2.01, "learning_rate": 5.194836152811292e-06, "loss": 0.0366, "step": 4340 }, { "epoch": 2.01, "learning_rate": 5.1510040227120005e-06, "loss": 0.0413, "step": 4350 }, { "epoch": 2.02, "learning_rate": 5.107293356284451e-06, "loss": 0.0342, "step": 4360 }, { "epoch": 2.02, "learning_rate": 5.063705248447697e-06, "loss": 0.0354, "step": 4370 }, { "epoch": 2.03, "learning_rate": 5.0202407910507755e-06, "loss": 0.0337, "step": 4380 }, { "epoch": 2.03, "learning_rate": 4.976901072845379e-06, "loss": 0.0323, "step": 4390 }, { "epoch": 2.04, "learning_rate": 4.933687179458588e-06, "loss": 0.0353, "step": 4400 }, { "epoch": 2.04, "learning_rate": 4.890600193365654e-06, "loss": 0.0328, "step": 4410 }, { "epoch": 2.05, "learning_rate": 4.847641193862901e-06, "loss": 0.032, "step": 4420 }, { "epoch": 2.05, "learning_rate": 4.804811257040685e-06, "loss": 0.0345, "step": 4430 }, { "epoch": 2.06, "learning_rate": 4.7621114557564405e-06, "loss": 0.0339, "step": 4440 }, { "epoch": 2.06, "learning_rate": 4.719542859607812e-06, "loss": 0.0344, "step": 4450 }, { "epoch": 2.07, "learning_rate": 4.677106534905844e-06, "loss": 0.0432, "step": 4460 }, { "epoch": 2.07, "learning_rate": 4.6348035446482845e-06, "loss": 0.0331, "step": 4470 }, { "epoch": 2.08, "learning_rate": 4.592634948492965e-06, "loss": 0.0346, "step": 4480 }, { "epoch": 2.08, "learning_rate": 4.550601802731231e-06, "loss": 0.0345, "step": 4490 }, { "epoch": 2.08, "learning_rate": 4.508705160261506e-06, "loss": 0.0302, "step": 4500 }, { "epoch": 2.09, "learning_rate": 4.466946070562915e-06, "loss": 0.0368, "step": 4510 }, { "epoch": 2.09, "learning_rate": 4.425325579668982e-06, "loss": 0.0356, "step": 4520 }, { "epoch": 2.1, "learning_rate": 4.383844730141438e-06, "loss": 0.0343, "step": 4530 }, { "epoch": 2.1, "learning_rate": 4.342504561044105e-06, "loss": 0.0391, "step": 4540 }, { "epoch": 2.11, "learning_rate": 4.301306107916862e-06, "loss": 0.0377, "step": 4550 }, { "epoch": 2.11, "learning_rate": 4.260250402749725e-06, "loss": 0.0338, "step": 4560 }, { "epoch": 2.12, "learning_rate": 4.219338473956964e-06, "loss": 0.0339, "step": 4570 }, { "epoch": 2.12, "learning_rate": 4.178571346351366e-06, "loss": 0.0323, "step": 4580 }, { "epoch": 2.13, "learning_rate": 4.137950041118568e-06, "loss": 0.0323, "step": 4590 }, { "epoch": 2.13, "learning_rate": 4.097475575791454e-06, "loss": 0.0366, "step": 4600 }, { "epoch": 2.14, "learning_rate": 4.057148964224684e-06, "loss": 0.0322, "step": 4610 }, { "epoch": 2.14, "learning_rate": 4.016971216569294e-06, "loss": 0.0348, "step": 4620 }, { "epoch": 2.14, "learning_rate": 3.976943339247391e-06, "loss": 0.0343, "step": 4630 }, { "epoch": 2.15, "learning_rate": 3.937066334926949e-06, "loss": 0.0346, "step": 4640 }, { "epoch": 2.15, "learning_rate": 3.897341202496679e-06, "loss": 0.0357, "step": 4650 }, { "epoch": 2.16, "learning_rate": 3.857768937041021e-06, "loss": 0.0295, "step": 4660 }, { "epoch": 2.16, "learning_rate": 3.818350529815218e-06, "loss": 0.0347, "step": 4670 }, { "epoch": 2.17, "learning_rate": 3.779086968220472e-06, "loss": 0.0307, "step": 4680 }, { "epoch": 2.17, "learning_rate": 3.7399792357792207e-06, "loss": 0.033, "step": 4690 }, { "epoch": 2.18, "learning_rate": 3.7010283121105094e-06, "loss": 0.0341, "step": 4700 }, { "epoch": 2.18, "learning_rate": 3.662235172905435e-06, "loss": 0.0327, "step": 4710 }, { "epoch": 2.19, "learning_rate": 3.623600789902704e-06, "loss": 0.0319, "step": 4720 }, { "epoch": 2.19, "learning_rate": 3.5851261308643204e-06, "loss": 0.0298, "step": 4730 }, { "epoch": 2.2, "learning_rate": 3.5468121595513073e-06, "loss": 0.0332, "step": 4740 }, { "epoch": 2.2, "learning_rate": 3.508659835699595e-06, "loss": 0.0364, "step": 4750 }, { "epoch": 2.2, "learning_rate": 3.4706701149959577e-06, "loss": 0.0363, "step": 4760 }, { "epoch": 2.21, "learning_rate": 3.4328439490540844e-06, "loss": 0.0384, "step": 4770 }, { "epoch": 2.21, "learning_rate": 3.3951822853907525e-06, "loss": 0.0327, "step": 4780 }, { "epoch": 2.22, "learning_rate": 3.3576860674020683e-06, "loss": 0.0317, "step": 4790 }, { "epoch": 2.22, "learning_rate": 3.3203562343398545e-06, "loss": 0.0347, "step": 4800 }, { "epoch": 2.23, "learning_rate": 3.283193721288127e-06, "loss": 0.0336, "step": 4810 }, { "epoch": 2.23, "learning_rate": 3.246199459139643e-06, "loss": 0.0323, "step": 4820 }, { "epoch": 2.24, "learning_rate": 3.2093743745726147e-06, "loss": 0.0317, "step": 4830 }, { "epoch": 2.24, "learning_rate": 3.1727193900274866e-06, "loss": 0.0361, "step": 4840 }, { "epoch": 2.25, "learning_rate": 3.136235423683819e-06, "loss": 0.0359, "step": 4850 }, { "epoch": 2.25, "learning_rate": 3.0999233894373026e-06, "loss": 0.0336, "step": 4860 }, { "epoch": 2.26, "learning_rate": 3.0637841968768532e-06, "loss": 0.0375, "step": 4870 }, { "epoch": 2.26, "learning_rate": 3.027818751261836e-06, "loss": 0.0353, "step": 4880 }, { "epoch": 2.26, "learning_rate": 2.992027953499391e-06, "loss": 0.032, "step": 4890 }, { "epoch": 2.27, "learning_rate": 2.95641270012186e-06, "loss": 0.0306, "step": 4900 }, { "epoch": 2.27, "learning_rate": 2.92097388326432e-06, "loss": 0.0345, "step": 4910 }, { "epoch": 2.28, "learning_rate": 2.8857123906422644e-06, "loss": 0.0349, "step": 4920 }, { "epoch": 2.28, "learning_rate": 2.8506291055293358e-06, "loss": 0.0361, "step": 4930 }, { "epoch": 2.29, "learning_rate": 2.8157249067352254e-06, "loss": 0.0375, "step": 4940 }, { "epoch": 2.29, "learning_rate": 2.781000668583638e-06, "loss": 0.0402, "step": 4950 }, { "epoch": 2.3, "learning_rate": 2.7464572608904005e-06, "loss": 0.0361, "step": 4960 }, { "epoch": 2.3, "learning_rate": 2.7120955489416865e-06, "loss": 0.0325, "step": 4970 }, { "epoch": 2.31, "learning_rate": 2.677916393472315e-06, "loss": 0.0334, "step": 4980 }, { "epoch": 2.31, "learning_rate": 2.6439206506442083e-06, "loss": 0.0337, "step": 4990 }, { "epoch": 2.32, "learning_rate": 2.610109172024948e-06, "loss": 0.0369, "step": 5000 }, { "epoch": 2.32, "learning_rate": 2.5764828045664238e-06, "loss": 0.0349, "step": 5010 }, { "epoch": 2.33, "learning_rate": 2.5430423905836365e-06, "loss": 0.0356, "step": 5020 }, { "epoch": 2.33, "learning_rate": 2.5097887677336043e-06, "loss": 0.0342, "step": 5030 }, { "epoch": 2.33, "learning_rate": 2.4767227689943506e-06, "loss": 0.0329, "step": 5040 }, { "epoch": 2.34, "learning_rate": 2.4438452226440734e-06, "loss": 0.0306, "step": 5050 }, { "epoch": 2.34, "learning_rate": 2.411156952240373e-06, "loss": 0.0314, "step": 5060 }, { "epoch": 2.35, "learning_rate": 2.3786587765996283e-06, "loss": 0.035, "step": 5070 }, { "epoch": 2.35, "learning_rate": 2.346351509776498e-06, "loss": 0.0355, "step": 5080 }, { "epoch": 2.36, "learning_rate": 2.3142359610435105e-06, "loss": 0.0374, "step": 5090 }, { "epoch": 2.36, "learning_rate": 2.282312934870806e-06, "loss": 0.0338, "step": 5100 }, { "epoch": 2.37, "learning_rate": 2.2505832309059806e-06, "loss": 0.0327, "step": 5110 }, { "epoch": 2.37, "learning_rate": 2.2190476439540543e-06, "loss": 0.0347, "step": 5120 }, { "epoch": 2.38, "learning_rate": 2.1877069639575587e-06, "loss": 0.0383, "step": 5130 }, { "epoch": 2.38, "learning_rate": 2.1565619759767675e-06, "loss": 0.0359, "step": 5140 }, { "epoch": 2.39, "learning_rate": 2.1256134601700018e-06, "loss": 0.0288, "step": 5150 }, { "epoch": 2.39, "learning_rate": 2.0948621917741163e-06, "loss": 0.0336, "step": 5160 }, { "epoch": 2.39, "learning_rate": 2.0643089410850624e-06, "loss": 0.0304, "step": 5170 }, { "epoch": 2.4, "learning_rate": 2.033954473438593e-06, "loss": 0.0334, "step": 5180 }, { "epoch": 2.4, "learning_rate": 2.00379954919111e-06, "loss": 0.0347, "step": 5190 }, { "epoch": 2.41, "learning_rate": 1.9738449237005895e-06, "loss": 0.034, "step": 5200 }, { "epoch": 2.41, "learning_rate": 1.9440913473076815e-06, "loss": 0.0302, "step": 5210 }, { "epoch": 2.42, "learning_rate": 1.914539565316915e-06, "loss": 0.0383, "step": 5220 }, { "epoch": 2.42, "learning_rate": 1.885190317978015e-06, "loss": 0.0357, "step": 5230 }, { "epoch": 2.43, "learning_rate": 1.8560443404673644e-06, "loss": 0.0339, "step": 5240 }, { "epoch": 2.43, "learning_rate": 1.8271023628696027e-06, "loss": 0.0318, "step": 5250 }, { "epoch": 2.44, "learning_rate": 1.7983651101593126e-06, "loss": 0.0341, "step": 5260 }, { "epoch": 2.44, "learning_rate": 1.769833302182885e-06, "loss": 0.0316, "step": 5270 }, { "epoch": 2.45, "learning_rate": 1.7415076536404697e-06, "loss": 0.0299, "step": 5280 }, { "epoch": 2.45, "learning_rate": 1.7133888740680803e-06, "loss": 0.0302, "step": 5290 }, { "epoch": 2.45, "learning_rate": 1.6854776678198192e-06, "loss": 0.0326, "step": 5300 }, { "epoch": 2.46, "learning_rate": 1.6577747340502348e-06, "loss": 0.0323, "step": 5310 }, { "epoch": 2.46, "learning_rate": 1.6302807666968068e-06, "loss": 0.0342, "step": 5320 }, { "epoch": 2.47, "learning_rate": 1.6029964544625698e-06, "loss": 0.0333, "step": 5330 }, { "epoch": 2.47, "learning_rate": 1.5759224807988504e-06, "loss": 0.0367, "step": 5340 }, { "epoch": 2.48, "learning_rate": 1.5490595238881544e-06, "loss": 0.0349, "step": 5350 }, { "epoch": 2.48, "learning_rate": 1.5224082566271835e-06, "loss": 0.0363, "step": 5360 }, { "epoch": 2.49, "learning_rate": 1.4959693466099657e-06, "loss": 0.0301, "step": 5370 }, { "epoch": 2.49, "learning_rate": 1.4697434561111512e-06, "loss": 0.0379, "step": 5380 }, { "epoch": 2.5, "learning_rate": 1.443731242069405e-06, "loss": 0.0371, "step": 5390 }, { "epoch": 2.5, "learning_rate": 1.4179333560709563e-06, "loss": 0.0378, "step": 5400 }, { "epoch": 2.51, "learning_rate": 1.3923504443332914e-06, "loss": 0.0327, "step": 5410 }, { "epoch": 2.51, "learning_rate": 1.3669831476889429e-06, "loss": 0.0318, "step": 5420 }, { "epoch": 2.52, "learning_rate": 1.3418321015694525e-06, "loss": 0.0354, "step": 5430 }, { "epoch": 2.52, "learning_rate": 1.316897935989454e-06, "loss": 0.0371, "step": 5440 }, { "epoch": 2.52, "learning_rate": 1.2921812755308839e-06, "loss": 0.0363, "step": 5450 }, { "epoch": 2.53, "learning_rate": 1.2676827393273405e-06, "loss": 0.0426, "step": 5460 }, { "epoch": 2.53, "learning_rate": 1.2434029410485771e-06, "loss": 0.0378, "step": 5470 }, { "epoch": 2.54, "learning_rate": 1.219342488885128e-06, "loss": 0.0297, "step": 5480 }, { "epoch": 2.54, "learning_rate": 1.1955019855330708e-06, "loss": 0.0341, "step": 5490 }, { "epoch": 2.55, "learning_rate": 1.1718820281789323e-06, "loss": 0.0327, "step": 5500 }, { "epoch": 2.55, "learning_rate": 1.148483208484732e-06, "loss": 0.0321, "step": 5510 }, { "epoch": 2.56, "learning_rate": 1.125306112573159e-06, "loss": 0.0313, "step": 5520 }, { "epoch": 2.56, "learning_rate": 1.1023513210128878e-06, "loss": 0.0324, "step": 5530 }, { "epoch": 2.57, "learning_rate": 1.0796194088040358e-06, "loss": 0.0335, "step": 5540 }, { "epoch": 2.57, "learning_rate": 1.0571109453637674e-06, "loss": 0.0393, "step": 5550 }, { "epoch": 2.58, "learning_rate": 1.0348264945120202e-06, "loss": 0.0331, "step": 5560 }, { "epoch": 2.58, "learning_rate": 1.0127666144573845e-06, "loss": 0.0295, "step": 5570 }, { "epoch": 2.58, "learning_rate": 9.909318577831317e-07, "loss": 0.0347, "step": 5580 }, { "epoch": 2.59, "learning_rate": 9.693227714333487e-07, "loss": 0.036, "step": 5590 }, { "epoch": 2.59, "learning_rate": 9.47939896699267e-07, "loss": 0.0334, "step": 5600 }, { "epoch": 2.6, "learning_rate": 9.267837692056791e-07, "loss": 0.0334, "step": 5610 }, { "epoch": 2.6, "learning_rate": 9.05854918897533e-07, "loss": 0.0296, "step": 5620 }, { "epoch": 2.61, "learning_rate": 8.851538700266593e-07, "loss": 0.0309, "step": 5630 }, { "epoch": 2.61, "learning_rate": 8.646811411386313e-07, "loss": 0.0326, "step": 5640 }, { "epoch": 2.62, "learning_rate": 8.444372450597782e-07, "loss": 0.032, "step": 5650 }, { "epoch": 2.62, "learning_rate": 8.244226888843487e-07, "loss": 0.0323, "step": 5660 }, { "epoch": 2.63, "learning_rate": 8.046379739617915e-07, "loss": 0.0322, "step": 5670 }, { "epoch": 2.63, "learning_rate": 7.850835958842174e-07, "loss": 0.0349, "step": 5680 }, { "epoch": 2.64, "learning_rate": 7.657600444739588e-07, "loss": 0.0332, "step": 5690 }, { "epoch": 2.64, "learning_rate": 7.466678037713271e-07, "loss": 0.0343, "step": 5700 }, { "epoch": 2.64, "learning_rate": 7.278073520224737e-07, "loss": 0.0356, "step": 5710 }, { "epoch": 2.65, "learning_rate": 7.091791616674093e-07, "loss": 0.0376, "step": 5720 }, { "epoch": 2.65, "learning_rate": 6.907836993281736e-07, "loss": 0.0349, "step": 5730 }, { "epoch": 2.66, "learning_rate": 6.726214257971509e-07, "loss": 0.0335, "step": 5740 }, { "epoch": 2.66, "learning_rate": 6.546927960255167e-07, "loss": 0.0327, "step": 5750 }, { "epoch": 2.67, "learning_rate": 6.369982591118473e-07, "loss": 0.0299, "step": 5760 }, { "epoch": 2.67, "learning_rate": 6.195382582908793e-07, "loss": 0.0356, "step": 5770 }, { "epoch": 2.68, "learning_rate": 6.023132309223834e-07, "loss": 0.0355, "step": 5780 }, { "epoch": 2.68, "learning_rate": 5.853236084802372e-07, "loss": 0.0346, "step": 5790 }, { "epoch": 2.69, "learning_rate": 5.685698165415965e-07, "loss": 0.0329, "step": 5800 }, { "epoch": 2.69, "learning_rate": 5.520522747762424e-07, "loss": 0.0301, "step": 5810 }, { "epoch": 2.7, "learning_rate": 5.357713969360712e-07, "loss": 0.0327, "step": 5820 }, { "epoch": 2.7, "learning_rate": 5.197275908447241e-07, "loss": 0.0313, "step": 5830 }, { "epoch": 2.7, "learning_rate": 5.039212583873743e-07, "loss": 0.0315, "step": 5840 }, { "epoch": 2.71, "learning_rate": 4.883527955006662e-07, "loss": 0.0323, "step": 5850 }, { "epoch": 2.71, "learning_rate": 4.7302259216278625e-07, "loss": 0.0333, "step": 5860 }, { "epoch": 2.72, "learning_rate": 4.5793103238370053e-07, "loss": 0.0333, "step": 5870 }, { "epoch": 2.72, "learning_rate": 4.430784941955368e-07, "loss": 0.0291, "step": 5880 }, { "epoch": 2.73, "learning_rate": 4.2846534964311124e-07, "loss": 0.0323, "step": 5890 }, { "epoch": 2.73, "learning_rate": 4.1409196477461444e-07, "loss": 0.0353, "step": 5900 }, { "epoch": 2.74, "learning_rate": 3.999586996324367e-07, "loss": 0.0355, "step": 5910 }, { "epoch": 2.74, "learning_rate": 3.860659082441476e-07, "loss": 0.0345, "step": 5920 }, { "epoch": 2.75, "learning_rate": 3.724139386136394e-07, "loss": 0.0336, "step": 5930 }, { "epoch": 2.75, "learning_rate": 3.5900313271239664e-07, "loss": 0.0316, "step": 5940 }, { "epoch": 2.76, "learning_rate": 3.4583382647093267e-07, "loss": 0.0322, "step": 5950 }, { "epoch": 2.76, "learning_rate": 3.329063497703866e-07, "loss": 0.0333, "step": 5960 }, { "epoch": 2.77, "learning_rate": 3.2022102643424315e-07, "loss": 0.0344, "step": 5970 }, { "epoch": 2.77, "learning_rate": 3.0777817422023017e-07, "loss": 0.0362, "step": 5980 }, { "epoch": 2.77, "learning_rate": 2.955781048123607e-07, "loss": 0.0352, "step": 5990 }, { "epoch": 2.78, "learning_rate": 2.836211238131192e-07, "loss": 0.0336, "step": 6000 }, { "epoch": 2.78, "learning_rate": 2.7190753073581434e-07, "loss": 0.0338, "step": 6010 }, { "epoch": 2.79, "learning_rate": 2.604376189970659e-07, "loss": 0.0339, "step": 6020 }, { "epoch": 2.79, "learning_rate": 2.492116759094654e-07, "loss": 0.0375, "step": 6030 }, { "epoch": 2.8, "learning_rate": 2.382299826743739e-07, "loss": 0.0344, "step": 6040 }, { "epoch": 2.8, "learning_rate": 2.2749281437487891e-07, "loss": 0.033, "step": 6050 }, { "epoch": 2.81, "learning_rate": 2.1700043996890074e-07, "loss": 0.0348, "step": 6060 }, { "epoch": 2.81, "learning_rate": 2.067531222824648e-07, "loss": 0.0373, "step": 6070 }, { "epoch": 2.82, "learning_rate": 1.9675111800310341e-07, "loss": 0.0363, "step": 6080 }, { "epoch": 2.82, "learning_rate": 1.8699467767343993e-07, "loss": 0.0344, "step": 6090 }, { "epoch": 2.83, "learning_rate": 1.7748404568490696e-07, "loss": 0.0339, "step": 6100 }, { "epoch": 2.83, "learning_rate": 1.6821946027161917e-07, "loss": 0.0362, "step": 6110 }, { "epoch": 2.83, "learning_rate": 1.5920115350441913e-07, "loss": 0.0298, "step": 6120 }, { "epoch": 2.84, "learning_rate": 1.5042935128504965e-07, "loss": 0.0329, "step": 6130 }, { "epoch": 2.84, "learning_rate": 1.4190427334050294e-07, "loss": 0.0341, "step": 6140 }, { "epoch": 2.85, "learning_rate": 1.3362613321751815e-07, "loss": 0.0352, "step": 6150 }, { "epoch": 2.85, "learning_rate": 1.2559513827722582e-07, "loss": 0.0314, "step": 6160 }, { "epoch": 2.86, "learning_rate": 1.1781148968995959e-07, "loss": 0.0318, "step": 6170 }, { "epoch": 2.86, "learning_rate": 1.1027538243021274e-07, "loss": 0.0319, "step": 6180 }, { "epoch": 2.87, "learning_rate": 1.0298700527175521e-07, "loss": 0.0351, "step": 6190 }, { "epoch": 2.87, "learning_rate": 9.594654078290966e-08, "loss": 0.0388, "step": 6200 }, { "epoch": 2.88, "learning_rate": 8.915416532197186e-08, "loss": 0.0333, "step": 6210 }, { "epoch": 2.88, "learning_rate": 8.261004903279413e-08, "loss": 0.0338, "step": 6220 }, { "epoch": 2.89, "learning_rate": 7.631435584052772e-08, "loss": 0.0313, "step": 6230 }, { "epoch": 2.89, "learning_rate": 7.026724344751268e-08, "loss": 0.036, "step": 6240 }, { "epoch": 2.89, "learning_rate": 6.446886332932667e-08, "loss": 0.0309, "step": 6250 }, { "epoch": 2.9, "learning_rate": 5.89193607309968e-08, "loss": 0.0321, "step": 6260 }, { "epoch": 2.9, "learning_rate": 5.3618874663351475e-08, "loss": 0.035, "step": 6270 }, { "epoch": 2.91, "learning_rate": 4.856753789954982e-08, "loss": 0.0327, "step": 6280 }, { "epoch": 2.91, "learning_rate": 4.376547697174438e-08, "loss": 0.0347, "step": 6290 }, { "epoch": 2.92, "learning_rate": 3.921281216791917e-08, "loss": 0.0353, "step": 6300 }, { "epoch": 2.92, "learning_rate": 3.490965752887765e-08, "loss": 0.0337, "step": 6310 }, { "epoch": 2.93, "learning_rate": 3.0856120845377256e-08, "loss": 0.0335, "step": 6320 }, { "epoch": 2.93, "learning_rate": 2.7052303655438204e-08, "loss": 0.031, "step": 6330 }, { "epoch": 2.94, "learning_rate": 2.349830124179775e-08, "loss": 0.0358, "step": 6340 }, { "epoch": 2.94, "learning_rate": 2.0194202629519876e-08, "loss": 0.0322, "step": 6350 }, { "epoch": 2.95, "learning_rate": 1.714009058376931e-08, "loss": 0.0308, "step": 6360 }, { "epoch": 2.95, "learning_rate": 1.4336041607736495e-08, "loss": 0.0344, "step": 6370 }, { "epoch": 2.96, "learning_rate": 1.1782125940721367e-08, "loss": 0.035, "step": 6380 }, { "epoch": 2.96, "learning_rate": 9.478407556375857e-09, "loss": 0.0334, "step": 6390 }, { "epoch": 2.96, "learning_rate": 7.424944161096293e-09, "loss": 0.0306, "step": 6400 }, { "epoch": 2.97, "learning_rate": 5.621787192583439e-09, "loss": 0.0331, "step": 6410 }, { "epoch": 2.97, "learning_rate": 4.06898181855353e-09, "loss": 0.0351, "step": 6420 }, { "epoch": 2.98, "learning_rate": 2.7665669356025105e-09, "loss": 0.0315, "step": 6430 }, { "epoch": 2.98, "learning_rate": 1.7145751682368094e-09, "loss": 0.0329, "step": 6440 }, { "epoch": 2.99, "learning_rate": 9.130328680495571e-10, "loss": 0.0354, "step": 6450 }, { "epoch": 2.99, "learning_rate": 3.6196011306999237e-10, "loss": 0.0296, "step": 6460 }, { "epoch": 3.0, "learning_rate": 6.137070725165117e-11, "loss": 0.0346, "step": 6470 }, { "epoch": 3.0, "eval_em": 0.6384, "eval_em_ic": 0.9334, "eval_f1": 0.8248, "eval_gen_len": 39.5298, "eval_levinstein_ratio": 0.9955, "eval_loss": 0.04924384132027626, "eval_precision": 0.9141, "eval_recall": 0.8575, "eval_runtime": 166.4317, "eval_samples_per_second": 10.287, "eval_steps_per_second": 0.162, "step": 6477 }, { "epoch": 3.0, "step": 6477, "total_flos": 4.332530652023808e+16, "train_loss": 0.04666397547913275, "train_runtime": 1768.7071, "train_samples_per_second": 234.31, "train_steps_per_second": 3.662 } ], "logging_steps": 10, "max_steps": 6477, "num_train_epochs": 3, "save_steps": 500, "total_flos": 4.332530652023808e+16, "trial_name": null, "trial_params": null }