{ "best_metric": null, "best_model_checkpoint": null, "epoch": 177.5147928994083, "eval_steps": 500, "global_step": 30000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.05917159763313609, "grad_norm": 153.8770294189453, "learning_rate": 0.0001, "loss": 20.2214, "step": 10 }, { "epoch": 0.11834319526627218, "grad_norm": 74.8006820678711, "learning_rate": 0.0001, "loss": 5.0049, "step": 20 }, { "epoch": 0.17751479289940827, "grad_norm": 55.02305221557617, "learning_rate": 0.0001, "loss": 3.7767, "step": 30 }, { "epoch": 0.23668639053254437, "grad_norm": 58.42805480957031, "learning_rate": 0.0001, "loss": 3.3392, "step": 40 }, { "epoch": 0.2958579881656805, "grad_norm": 53.602413177490234, "learning_rate": 0.0001, "loss": 2.9888, "step": 50 }, { "epoch": 0.35502958579881655, "grad_norm": 49.62657165527344, "learning_rate": 0.0001, "loss": 2.7634, "step": 60 }, { "epoch": 0.41420118343195267, "grad_norm": 47.2798957824707, "learning_rate": 0.0001, "loss": 2.5771, "step": 70 }, { "epoch": 0.47337278106508873, "grad_norm": 42.863426208496094, "learning_rate": 0.0001, "loss": 2.3452, "step": 80 }, { "epoch": 0.5325443786982249, "grad_norm": 40.8491096496582, "learning_rate": 0.0001, "loss": 2.2521, "step": 90 }, { "epoch": 0.591715976331361, "grad_norm": 44.430477142333984, "learning_rate": 0.0001, "loss": 2.1279, "step": 100 }, { "epoch": 0.650887573964497, "grad_norm": 42.121849060058594, "learning_rate": 0.0001, "loss": 1.9346, "step": 110 }, { "epoch": 0.7100591715976331, "grad_norm": 43.765262603759766, "learning_rate": 0.0001, "loss": 1.8565, "step": 120 }, { "epoch": 0.7692307692307693, "grad_norm": 42.57600784301758, "learning_rate": 0.0001, "loss": 1.7533, "step": 130 }, { "epoch": 0.8284023668639053, "grad_norm": 35.68867111206055, "learning_rate": 0.0001, "loss": 1.6252, "step": 140 }, { "epoch": 0.8875739644970414, "grad_norm": 45.210853576660156, "learning_rate": 0.0001, "loss": 1.548, "step": 150 }, { "epoch": 0.9467455621301775, "grad_norm": 38.14447021484375, "learning_rate": 0.0001, "loss": 1.5014, "step": 160 }, { "epoch": 1.0059171597633136, "grad_norm": 40.30084228515625, "learning_rate": 0.0001, "loss": 1.4068, "step": 170 }, { "epoch": 1.0650887573964498, "grad_norm": 39.272117614746094, "learning_rate": 0.0001, "loss": 1.3457, "step": 180 }, { "epoch": 1.1242603550295858, "grad_norm": 31.490188598632812, "learning_rate": 0.0001, "loss": 1.2355, "step": 190 }, { "epoch": 1.183431952662722, "grad_norm": 31.656192779541016, "learning_rate": 0.0001, "loss": 1.1473, "step": 200 }, { "epoch": 1.242603550295858, "grad_norm": 36.3016357421875, "learning_rate": 0.0001, "loss": 1.1064, "step": 210 }, { "epoch": 1.301775147928994, "grad_norm": 38.86666488647461, "learning_rate": 0.0001, "loss": 1.046, "step": 220 }, { "epoch": 1.3609467455621302, "grad_norm": 39.29147720336914, "learning_rate": 0.0001, "loss": 1.0058, "step": 230 }, { "epoch": 1.4201183431952662, "grad_norm": 33.329917907714844, "learning_rate": 0.0001, "loss": 0.9461, "step": 240 }, { "epoch": 1.4792899408284024, "grad_norm": 34.38193130493164, "learning_rate": 0.0001, "loss": 0.8668, "step": 250 }, { "epoch": 1.5384615384615383, "grad_norm": 31.653987884521484, "learning_rate": 0.0001, "loss": 0.823, "step": 260 }, { "epoch": 1.5976331360946747, "grad_norm": 31.058713912963867, "learning_rate": 0.0001, "loss": 0.7936, "step": 270 }, { "epoch": 1.6568047337278107, "grad_norm": 31.71525764465332, "learning_rate": 0.0001, "loss": 0.7399, "step": 280 }, { "epoch": 1.7159763313609466, "grad_norm": 28.538848876953125, "learning_rate": 0.0001, "loss": 0.6946, "step": 290 }, { "epoch": 1.7751479289940828, "grad_norm": 26.211162567138672, "learning_rate": 0.0001, "loss": 0.6672, "step": 300 }, { "epoch": 1.834319526627219, "grad_norm": 26.915746688842773, "learning_rate": 0.0001, "loss": 0.6573, "step": 310 }, { "epoch": 1.893491124260355, "grad_norm": 26.777111053466797, "learning_rate": 0.0001, "loss": 0.6246, "step": 320 }, { "epoch": 1.952662721893491, "grad_norm": 22.7823429107666, "learning_rate": 0.0001, "loss": 0.5841, "step": 330 }, { "epoch": 2.0118343195266273, "grad_norm": 26.74056053161621, "learning_rate": 0.0001, "loss": 0.5517, "step": 340 }, { "epoch": 2.0710059171597632, "grad_norm": 23.421667098999023, "learning_rate": 0.0001, "loss": 0.5297, "step": 350 }, { "epoch": 2.1301775147928996, "grad_norm": 23.535478591918945, "learning_rate": 0.0001, "loss": 0.5149, "step": 360 }, { "epoch": 2.1893491124260356, "grad_norm": 18.5365047454834, "learning_rate": 0.0001, "loss": 0.4807, "step": 370 }, { "epoch": 2.2485207100591715, "grad_norm": 22.976980209350586, "learning_rate": 0.0001, "loss": 0.4663, "step": 380 }, { "epoch": 2.3076923076923075, "grad_norm": 21.123947143554688, "learning_rate": 0.0001, "loss": 0.4509, "step": 390 }, { "epoch": 2.366863905325444, "grad_norm": 24.507892608642578, "learning_rate": 0.0001, "loss": 0.4398, "step": 400 }, { "epoch": 2.42603550295858, "grad_norm": 22.395057678222656, "learning_rate": 0.0001, "loss": 0.422, "step": 410 }, { "epoch": 2.485207100591716, "grad_norm": 21.968156814575195, "learning_rate": 0.0001, "loss": 0.4093, "step": 420 }, { "epoch": 2.544378698224852, "grad_norm": 18.98883628845215, "learning_rate": 0.0001, "loss": 0.3963, "step": 430 }, { "epoch": 2.603550295857988, "grad_norm": 19.853185653686523, "learning_rate": 0.0001, "loss": 0.3803, "step": 440 }, { "epoch": 2.662721893491124, "grad_norm": 19.97403335571289, "learning_rate": 0.0001, "loss": 0.3701, "step": 450 }, { "epoch": 2.7218934911242605, "grad_norm": 19.338214874267578, "learning_rate": 0.0001, "loss": 0.3528, "step": 460 }, { "epoch": 2.7810650887573964, "grad_norm": 20.44488525390625, "learning_rate": 0.0001, "loss": 0.3524, "step": 470 }, { "epoch": 2.8402366863905324, "grad_norm": 16.130311965942383, "learning_rate": 0.0001, "loss": 0.3377, "step": 480 }, { "epoch": 2.899408284023669, "grad_norm": 19.017309188842773, "learning_rate": 0.0001, "loss": 0.3298, "step": 490 }, { "epoch": 2.9585798816568047, "grad_norm": 17.657939910888672, "learning_rate": 0.0001, "loss": 0.3189, "step": 500 }, { "epoch": 3.0177514792899407, "grad_norm": 15.550881385803223, "learning_rate": 0.0001, "loss": 0.3234, "step": 510 }, { "epoch": 3.076923076923077, "grad_norm": 16.31296157836914, "learning_rate": 0.0001, "loss": 0.3118, "step": 520 }, { "epoch": 3.136094674556213, "grad_norm": 17.804370880126953, "learning_rate": 0.0001, "loss": 0.2931, "step": 530 }, { "epoch": 3.195266272189349, "grad_norm": 14.263854026794434, "learning_rate": 0.0001, "loss": 0.2863, "step": 540 }, { "epoch": 3.2544378698224854, "grad_norm": 15.186551094055176, "learning_rate": 0.0001, "loss": 0.2806, "step": 550 }, { "epoch": 3.3136094674556213, "grad_norm": 15.333024978637695, "learning_rate": 0.0001, "loss": 0.2789, "step": 560 }, { "epoch": 3.3727810650887573, "grad_norm": 15.714407920837402, "learning_rate": 0.0001, "loss": 0.2811, "step": 570 }, { "epoch": 3.4319526627218933, "grad_norm": 15.480822563171387, "learning_rate": 0.0001, "loss": 0.2698, "step": 580 }, { "epoch": 3.4911242603550297, "grad_norm": 14.841946601867676, "learning_rate": 0.0001, "loss": 0.2723, "step": 590 }, { "epoch": 3.5502958579881656, "grad_norm": 15.96535587310791, "learning_rate": 0.0001, "loss": 0.2577, "step": 600 }, { "epoch": 3.609467455621302, "grad_norm": 14.859570503234863, "learning_rate": 0.0001, "loss": 0.2549, "step": 610 }, { "epoch": 3.668639053254438, "grad_norm": 13.688532829284668, "learning_rate": 0.0001, "loss": 0.2522, "step": 620 }, { "epoch": 3.727810650887574, "grad_norm": 13.702826499938965, "learning_rate": 0.0001, "loss": 0.2406, "step": 630 }, { "epoch": 3.78698224852071, "grad_norm": 14.145650863647461, "learning_rate": 0.0001, "loss": 0.2481, "step": 640 }, { "epoch": 3.8461538461538463, "grad_norm": 15.478382110595703, "learning_rate": 0.0001, "loss": 0.2487, "step": 650 }, { "epoch": 3.905325443786982, "grad_norm": 14.090296745300293, "learning_rate": 0.0001, "loss": 0.2439, "step": 660 }, { "epoch": 3.9644970414201186, "grad_norm": 12.591463088989258, "learning_rate": 0.0001, "loss": 0.2365, "step": 670 }, { "epoch": 4.023668639053255, "grad_norm": 13.442497253417969, "learning_rate": 0.0001, "loss": 0.2256, "step": 680 }, { "epoch": 4.0828402366863905, "grad_norm": 13.437725067138672, "learning_rate": 0.0001, "loss": 0.2298, "step": 690 }, { "epoch": 4.1420118343195265, "grad_norm": 14.193412780761719, "learning_rate": 0.0001, "loss": 0.2298, "step": 700 }, { "epoch": 4.201183431952662, "grad_norm": 13.714064598083496, "learning_rate": 0.0001, "loss": 0.2287, "step": 710 }, { "epoch": 4.260355029585799, "grad_norm": 14.350592613220215, "learning_rate": 0.0001, "loss": 0.2156, "step": 720 }, { "epoch": 4.319526627218935, "grad_norm": 13.923611640930176, "learning_rate": 0.0001, "loss": 0.2076, "step": 730 }, { "epoch": 4.378698224852071, "grad_norm": 11.358366012573242, "learning_rate": 0.0001, "loss": 0.2075, "step": 740 }, { "epoch": 4.437869822485207, "grad_norm": 15.445727348327637, "learning_rate": 0.0001, "loss": 0.2197, "step": 750 }, { "epoch": 4.497041420118343, "grad_norm": 14.4073486328125, "learning_rate": 0.0001, "loss": 0.2127, "step": 760 }, { "epoch": 4.556213017751479, "grad_norm": 12.812962532043457, "learning_rate": 0.0001, "loss": 0.2084, "step": 770 }, { "epoch": 4.615384615384615, "grad_norm": 12.11327075958252, "learning_rate": 0.0001, "loss": 0.2126, "step": 780 }, { "epoch": 4.674556213017752, "grad_norm": 13.59167766571045, "learning_rate": 0.0001, "loss": 0.2154, "step": 790 }, { "epoch": 4.733727810650888, "grad_norm": 13.989721298217773, "learning_rate": 0.0001, "loss": 0.199, "step": 800 }, { "epoch": 4.792899408284024, "grad_norm": 10.902786254882812, "learning_rate": 0.0001, "loss": 0.1989, "step": 810 }, { "epoch": 4.85207100591716, "grad_norm": 10.438324928283691, "learning_rate": 0.0001, "loss": 0.2015, "step": 820 }, { "epoch": 4.911242603550296, "grad_norm": 10.875935554504395, "learning_rate": 0.0001, "loss": 0.1894, "step": 830 }, { "epoch": 4.970414201183432, "grad_norm": 11.438563346862793, "learning_rate": 0.0001, "loss": 0.1966, "step": 840 }, { "epoch": 5.029585798816568, "grad_norm": 13.261100769042969, "learning_rate": 0.0001, "loss": 0.1862, "step": 850 }, { "epoch": 5.088757396449704, "grad_norm": 14.929048538208008, "learning_rate": 0.0001, "loss": 0.1891, "step": 860 }, { "epoch": 5.14792899408284, "grad_norm": 10.416974067687988, "learning_rate": 0.0001, "loss": 0.1857, "step": 870 }, { "epoch": 5.207100591715976, "grad_norm": 9.880386352539062, "learning_rate": 0.0001, "loss": 0.1889, "step": 880 }, { "epoch": 5.266272189349112, "grad_norm": 11.328648567199707, "learning_rate": 0.0001, "loss": 0.1878, "step": 890 }, { "epoch": 5.325443786982248, "grad_norm": 13.706796646118164, "learning_rate": 0.0001, "loss": 0.181, "step": 900 }, { "epoch": 5.384615384615385, "grad_norm": 11.462624549865723, "learning_rate": 0.0001, "loss": 0.172, "step": 910 }, { "epoch": 5.443786982248521, "grad_norm": 13.581514358520508, "learning_rate": 0.0001, "loss": 0.1781, "step": 920 }, { "epoch": 5.502958579881657, "grad_norm": 12.419181823730469, "learning_rate": 0.0001, "loss": 0.1767, "step": 930 }, { "epoch": 5.562130177514793, "grad_norm": 13.577881813049316, "learning_rate": 0.0001, "loss": 0.1768, "step": 940 }, { "epoch": 5.621301775147929, "grad_norm": 12.28160572052002, "learning_rate": 0.0001, "loss": 0.1707, "step": 950 }, { "epoch": 5.680473372781065, "grad_norm": 12.492323875427246, "learning_rate": 0.0001, "loss": 0.1715, "step": 960 }, { "epoch": 5.739644970414201, "grad_norm": 8.74298095703125, "learning_rate": 0.0001, "loss": 0.1732, "step": 970 }, { "epoch": 5.798816568047338, "grad_norm": 11.099392890930176, "learning_rate": 0.0001, "loss": 0.1682, "step": 980 }, { "epoch": 5.8579881656804735, "grad_norm": 9.262553215026855, "learning_rate": 0.0001, "loss": 0.1671, "step": 990 }, { "epoch": 5.9171597633136095, "grad_norm": 10.935064315795898, "learning_rate": 0.0001, "loss": 0.1725, "step": 1000 }, { "epoch": 5.976331360946745, "grad_norm": 8.737711906433105, "learning_rate": 0.0001, "loss": 0.1726, "step": 1010 }, { "epoch": 6.035502958579881, "grad_norm": 10.452316284179688, "learning_rate": 0.0001, "loss": 0.1673, "step": 1020 }, { "epoch": 6.094674556213017, "grad_norm": 9.871715545654297, "learning_rate": 0.0001, "loss": 0.1656, "step": 1030 }, { "epoch": 6.153846153846154, "grad_norm": 10.13133716583252, "learning_rate": 0.0001, "loss": 0.1596, "step": 1040 }, { "epoch": 6.21301775147929, "grad_norm": 9.236785888671875, "learning_rate": 0.0001, "loss": 0.1636, "step": 1050 }, { "epoch": 6.272189349112426, "grad_norm": 11.814075469970703, "learning_rate": 0.0001, "loss": 0.1609, "step": 1060 }, { "epoch": 6.331360946745562, "grad_norm": 10.119730949401855, "learning_rate": 0.0001, "loss": 0.1494, "step": 1070 }, { "epoch": 6.390532544378698, "grad_norm": 9.348946571350098, "learning_rate": 0.0001, "loss": 0.1529, "step": 1080 }, { "epoch": 6.449704142011834, "grad_norm": 10.565921783447266, "learning_rate": 0.0001, "loss": 0.1521, "step": 1090 }, { "epoch": 6.508875739644971, "grad_norm": 9.560077667236328, "learning_rate": 0.0001, "loss": 0.1538, "step": 1100 }, { "epoch": 6.568047337278107, "grad_norm": 10.612263679504395, "learning_rate": 0.0001, "loss": 0.1516, "step": 1110 }, { "epoch": 6.627218934911243, "grad_norm": 8.931841850280762, "learning_rate": 0.0001, "loss": 0.1469, "step": 1120 }, { "epoch": 6.686390532544379, "grad_norm": 10.13803768157959, "learning_rate": 0.0001, "loss": 0.1479, "step": 1130 }, { "epoch": 6.745562130177515, "grad_norm": 9.097755432128906, "learning_rate": 0.0001, "loss": 0.1512, "step": 1140 }, { "epoch": 6.804733727810651, "grad_norm": 9.387017250061035, "learning_rate": 0.0001, "loss": 0.1481, "step": 1150 }, { "epoch": 6.8639053254437865, "grad_norm": 10.371561050415039, "learning_rate": 0.0001, "loss": 0.1465, "step": 1160 }, { "epoch": 6.923076923076923, "grad_norm": 9.054409980773926, "learning_rate": 0.0001, "loss": 0.1421, "step": 1170 }, { "epoch": 6.982248520710059, "grad_norm": 10.072967529296875, "learning_rate": 0.0001, "loss": 0.1402, "step": 1180 }, { "epoch": 7.041420118343195, "grad_norm": 9.126667976379395, "learning_rate": 0.0001, "loss": 0.1437, "step": 1190 }, { "epoch": 7.100591715976331, "grad_norm": 9.495742797851562, "learning_rate": 0.0001, "loss": 0.1368, "step": 1200 }, { "epoch": 7.159763313609467, "grad_norm": 9.726602554321289, "learning_rate": 0.0001, "loss": 0.1375, "step": 1210 }, { "epoch": 7.218934911242604, "grad_norm": 9.073755264282227, "learning_rate": 0.0001, "loss": 0.141, "step": 1220 }, { "epoch": 7.27810650887574, "grad_norm": 10.156746864318848, "learning_rate": 0.0001, "loss": 0.1352, "step": 1230 }, { "epoch": 7.337278106508876, "grad_norm": 10.866272926330566, "learning_rate": 0.0001, "loss": 0.1353, "step": 1240 }, { "epoch": 7.396449704142012, "grad_norm": 8.42164134979248, "learning_rate": 0.0001, "loss": 0.1369, "step": 1250 }, { "epoch": 7.455621301775148, "grad_norm": 10.95656967163086, "learning_rate": 0.0001, "loss": 0.1457, "step": 1260 }, { "epoch": 7.514792899408284, "grad_norm": 10.434103012084961, "learning_rate": 0.0001, "loss": 0.1414, "step": 1270 }, { "epoch": 7.57396449704142, "grad_norm": 12.03824234008789, "learning_rate": 0.0001, "loss": 0.1382, "step": 1280 }, { "epoch": 7.633136094674557, "grad_norm": 8.857409477233887, "learning_rate": 0.0001, "loss": 0.1356, "step": 1290 }, { "epoch": 7.6923076923076925, "grad_norm": 9.715126037597656, "learning_rate": 0.0001, "loss": 0.1302, "step": 1300 }, { "epoch": 7.7514792899408285, "grad_norm": 10.48636245727539, "learning_rate": 0.0001, "loss": 0.1282, "step": 1310 }, { "epoch": 7.810650887573964, "grad_norm": 10.33277416229248, "learning_rate": 0.0001, "loss": 0.1305, "step": 1320 }, { "epoch": 7.8698224852071, "grad_norm": 8.459954261779785, "learning_rate": 0.0001, "loss": 0.132, "step": 1330 }, { "epoch": 7.928994082840236, "grad_norm": 8.733254432678223, "learning_rate": 0.0001, "loss": 0.1311, "step": 1340 }, { "epoch": 7.988165680473373, "grad_norm": 7.845705032348633, "learning_rate": 0.0001, "loss": 0.1302, "step": 1350 }, { "epoch": 8.04733727810651, "grad_norm": 8.295759201049805, "learning_rate": 0.0001, "loss": 0.1291, "step": 1360 }, { "epoch": 8.106508875739644, "grad_norm": 9.882453918457031, "learning_rate": 0.0001, "loss": 0.1268, "step": 1370 }, { "epoch": 8.165680473372781, "grad_norm": 9.334071159362793, "learning_rate": 0.0001, "loss": 0.1269, "step": 1380 }, { "epoch": 8.224852071005918, "grad_norm": 9.324673652648926, "learning_rate": 0.0001, "loss": 0.1255, "step": 1390 }, { "epoch": 8.284023668639053, "grad_norm": 9.478947639465332, "learning_rate": 0.0001, "loss": 0.1185, "step": 1400 }, { "epoch": 8.34319526627219, "grad_norm": 8.823881149291992, "learning_rate": 0.0001, "loss": 0.1213, "step": 1410 }, { "epoch": 8.402366863905325, "grad_norm": 8.3399076461792, "learning_rate": 0.0001, "loss": 0.1176, "step": 1420 }, { "epoch": 8.461538461538462, "grad_norm": 7.765944480895996, "learning_rate": 0.0001, "loss": 0.1191, "step": 1430 }, { "epoch": 8.520710059171599, "grad_norm": 7.110343933105469, "learning_rate": 0.0001, "loss": 0.1198, "step": 1440 }, { "epoch": 8.579881656804734, "grad_norm": 6.8289875984191895, "learning_rate": 0.0001, "loss": 0.1227, "step": 1450 }, { "epoch": 8.63905325443787, "grad_norm": 7.456328392028809, "learning_rate": 0.0001, "loss": 0.1162, "step": 1460 }, { "epoch": 8.698224852071005, "grad_norm": 9.243852615356445, "learning_rate": 0.0001, "loss": 0.1199, "step": 1470 }, { "epoch": 8.757396449704142, "grad_norm": 8.14656925201416, "learning_rate": 0.0001, "loss": 0.1207, "step": 1480 }, { "epoch": 8.816568047337277, "grad_norm": 7.939619541168213, "learning_rate": 0.0001, "loss": 0.1175, "step": 1490 }, { "epoch": 8.875739644970414, "grad_norm": 8.320149421691895, "learning_rate": 0.0001, "loss": 0.1203, "step": 1500 }, { "epoch": 8.934911242603551, "grad_norm": 10.517495155334473, "learning_rate": 0.0001, "loss": 0.1189, "step": 1510 }, { "epoch": 8.994082840236686, "grad_norm": 8.182599067687988, "learning_rate": 0.0001, "loss": 0.1171, "step": 1520 }, { "epoch": 9.053254437869823, "grad_norm": 9.076415061950684, "learning_rate": 0.0001, "loss": 0.1137, "step": 1530 }, { "epoch": 9.112426035502958, "grad_norm": 10.241826057434082, "learning_rate": 0.0001, "loss": 0.1166, "step": 1540 }, { "epoch": 9.171597633136095, "grad_norm": 7.812061786651611, "learning_rate": 0.0001, "loss": 0.115, "step": 1550 }, { "epoch": 9.23076923076923, "grad_norm": 8.304200172424316, "learning_rate": 0.0001, "loss": 0.1134, "step": 1560 }, { "epoch": 9.289940828402367, "grad_norm": 7.753726482391357, "learning_rate": 0.0001, "loss": 0.1092, "step": 1570 }, { "epoch": 9.349112426035504, "grad_norm": 7.462099075317383, "learning_rate": 0.0001, "loss": 0.1103, "step": 1580 }, { "epoch": 9.408284023668639, "grad_norm": 9.046765327453613, "learning_rate": 0.0001, "loss": 0.1085, "step": 1590 }, { "epoch": 9.467455621301776, "grad_norm": 8.030011177062988, "learning_rate": 0.0001, "loss": 0.1105, "step": 1600 }, { "epoch": 9.52662721893491, "grad_norm": 8.600719451904297, "learning_rate": 0.0001, "loss": 0.1094, "step": 1610 }, { "epoch": 9.585798816568047, "grad_norm": 7.835883140563965, "learning_rate": 0.0001, "loss": 0.1135, "step": 1620 }, { "epoch": 9.644970414201183, "grad_norm": 7.297766208648682, "learning_rate": 0.0001, "loss": 0.115, "step": 1630 }, { "epoch": 9.70414201183432, "grad_norm": 6.88030481338501, "learning_rate": 0.0001, "loss": 0.1077, "step": 1640 }, { "epoch": 9.763313609467456, "grad_norm": 8.758949279785156, "learning_rate": 0.0001, "loss": 0.1102, "step": 1650 }, { "epoch": 9.822485207100591, "grad_norm": 8.520991325378418, "learning_rate": 0.0001, "loss": 0.1049, "step": 1660 }, { "epoch": 9.881656804733728, "grad_norm": 9.739422798156738, "learning_rate": 0.0001, "loss": 0.1023, "step": 1670 }, { "epoch": 9.940828402366863, "grad_norm": 9.229175567626953, "learning_rate": 0.0001, "loss": 0.1024, "step": 1680 }, { "epoch": 10.0, "grad_norm": 9.158478736877441, "learning_rate": 0.0001, "loss": 0.1051, "step": 1690 }, { "epoch": 10.059171597633137, "grad_norm": 8.830097198486328, "learning_rate": 0.0001, "loss": 0.1018, "step": 1700 }, { "epoch": 10.118343195266272, "grad_norm": 8.569356918334961, "learning_rate": 0.0001, "loss": 0.1018, "step": 1710 }, { "epoch": 10.177514792899409, "grad_norm": 8.450271606445312, "learning_rate": 0.0001, "loss": 0.1015, "step": 1720 }, { "epoch": 10.236686390532544, "grad_norm": 7.130739212036133, "learning_rate": 0.0001, "loss": 0.1059, "step": 1730 }, { "epoch": 10.29585798816568, "grad_norm": 6.290536403656006, "learning_rate": 0.0001, "loss": 0.1016, "step": 1740 }, { "epoch": 10.355029585798816, "grad_norm": 6.820939540863037, "learning_rate": 0.0001, "loss": 0.1014, "step": 1750 }, { "epoch": 10.414201183431953, "grad_norm": 6.5812811851501465, "learning_rate": 0.0001, "loss": 0.1042, "step": 1760 }, { "epoch": 10.47337278106509, "grad_norm": 7.177812576293945, "learning_rate": 0.0001, "loss": 0.1067, "step": 1770 }, { "epoch": 10.532544378698224, "grad_norm": 7.701907634735107, "learning_rate": 0.0001, "loss": 0.1041, "step": 1780 }, { "epoch": 10.591715976331361, "grad_norm": 6.324485778808594, "learning_rate": 0.0001, "loss": 0.098, "step": 1790 }, { "epoch": 10.650887573964496, "grad_norm": 7.509298324584961, "learning_rate": 0.0001, "loss": 0.0997, "step": 1800 }, { "epoch": 10.710059171597633, "grad_norm": 6.726043224334717, "learning_rate": 0.0001, "loss": 0.103, "step": 1810 }, { "epoch": 10.76923076923077, "grad_norm": 7.341011047363281, "learning_rate": 0.0001, "loss": 0.0984, "step": 1820 }, { "epoch": 10.828402366863905, "grad_norm": 6.767984867095947, "learning_rate": 0.0001, "loss": 0.1007, "step": 1830 }, { "epoch": 10.887573964497042, "grad_norm": 8.20561695098877, "learning_rate": 0.0001, "loss": 0.1014, "step": 1840 }, { "epoch": 10.946745562130177, "grad_norm": 6.168294906616211, "learning_rate": 0.0001, "loss": 0.0986, "step": 1850 }, { "epoch": 11.005917159763314, "grad_norm": 7.324835777282715, "learning_rate": 0.0001, "loss": 0.0993, "step": 1860 }, { "epoch": 11.065088757396449, "grad_norm": 6.547602653503418, "learning_rate": 0.0001, "loss": 0.0949, "step": 1870 }, { "epoch": 11.124260355029586, "grad_norm": 7.943658828735352, "learning_rate": 0.0001, "loss": 0.0974, "step": 1880 }, { "epoch": 11.183431952662723, "grad_norm": 6.16728401184082, "learning_rate": 0.0001, "loss": 0.093, "step": 1890 }, { "epoch": 11.242603550295858, "grad_norm": 6.384585380554199, "learning_rate": 0.0001, "loss": 0.0956, "step": 1900 }, { "epoch": 11.301775147928995, "grad_norm": 6.243564605712891, "learning_rate": 0.0001, "loss": 0.0951, "step": 1910 }, { "epoch": 11.36094674556213, "grad_norm": 7.403732776641846, "learning_rate": 0.0001, "loss": 0.0913, "step": 1920 }, { "epoch": 11.420118343195266, "grad_norm": 7.579709053039551, "learning_rate": 0.0001, "loss": 0.0952, "step": 1930 }, { "epoch": 11.479289940828401, "grad_norm": 7.204126834869385, "learning_rate": 0.0001, "loss": 0.0908, "step": 1940 }, { "epoch": 11.538461538461538, "grad_norm": 7.589733600616455, "learning_rate": 0.0001, "loss": 0.0948, "step": 1950 }, { "epoch": 11.597633136094675, "grad_norm": 7.022636413574219, "learning_rate": 0.0001, "loss": 0.0923, "step": 1960 }, { "epoch": 11.65680473372781, "grad_norm": 7.036110877990723, "learning_rate": 0.0001, "loss": 0.0928, "step": 1970 }, { "epoch": 11.715976331360947, "grad_norm": 7.1176276206970215, "learning_rate": 0.0001, "loss": 0.0936, "step": 1980 }, { "epoch": 11.775147928994082, "grad_norm": 6.415647506713867, "learning_rate": 0.0001, "loss": 0.0948, "step": 1990 }, { "epoch": 11.834319526627219, "grad_norm": 6.902602195739746, "learning_rate": 0.0001, "loss": 0.0937, "step": 2000 }, { "epoch": 11.893491124260356, "grad_norm": 5.66604471206665, "learning_rate": 0.0001, "loss": 0.0891, "step": 2010 }, { "epoch": 11.95266272189349, "grad_norm": 6.13353967666626, "learning_rate": 0.0001, "loss": 0.0892, "step": 2020 }, { "epoch": 12.011834319526628, "grad_norm": 7.034632205963135, "learning_rate": 0.0001, "loss": 0.0905, "step": 2030 }, { "epoch": 12.071005917159763, "grad_norm": 6.073016166687012, "learning_rate": 0.0001, "loss": 0.0878, "step": 2040 }, { "epoch": 12.1301775147929, "grad_norm": 6.667945384979248, "learning_rate": 0.0001, "loss": 0.0879, "step": 2050 }, { "epoch": 12.189349112426035, "grad_norm": 6.973472595214844, "learning_rate": 0.0001, "loss": 0.0877, "step": 2060 }, { "epoch": 12.248520710059172, "grad_norm": 6.750099182128906, "learning_rate": 0.0001, "loss": 0.0921, "step": 2070 }, { "epoch": 12.307692307692308, "grad_norm": 6.361142635345459, "learning_rate": 0.0001, "loss": 0.0894, "step": 2080 }, { "epoch": 12.366863905325443, "grad_norm": 6.784110069274902, "learning_rate": 0.0001, "loss": 0.0941, "step": 2090 }, { "epoch": 12.42603550295858, "grad_norm": 5.801392078399658, "learning_rate": 0.0001, "loss": 0.0894, "step": 2100 }, { "epoch": 12.485207100591715, "grad_norm": 6.16151237487793, "learning_rate": 0.0001, "loss": 0.0912, "step": 2110 }, { "epoch": 12.544378698224852, "grad_norm": 5.845252513885498, "learning_rate": 0.0001, "loss": 0.0914, "step": 2120 }, { "epoch": 12.603550295857989, "grad_norm": 7.042261123657227, "learning_rate": 0.0001, "loss": 0.0922, "step": 2130 }, { "epoch": 12.662721893491124, "grad_norm": 7.15225076675415, "learning_rate": 0.0001, "loss": 0.0902, "step": 2140 }, { "epoch": 12.721893491124261, "grad_norm": 6.622318744659424, "learning_rate": 0.0001, "loss": 0.0931, "step": 2150 }, { "epoch": 12.781065088757396, "grad_norm": 7.196859836578369, "learning_rate": 0.0001, "loss": 0.089, "step": 2160 }, { "epoch": 12.840236686390533, "grad_norm": 6.348023891448975, "learning_rate": 0.0001, "loss": 0.0855, "step": 2170 }, { "epoch": 12.899408284023668, "grad_norm": 5.8107008934021, "learning_rate": 0.0001, "loss": 0.0863, "step": 2180 }, { "epoch": 12.958579881656805, "grad_norm": 5.817500591278076, "learning_rate": 0.0001, "loss": 0.0822, "step": 2190 }, { "epoch": 13.017751479289942, "grad_norm": 5.527827739715576, "learning_rate": 0.0001, "loss": 0.0826, "step": 2200 }, { "epoch": 13.076923076923077, "grad_norm": 7.51456880569458, "learning_rate": 0.0001, "loss": 0.0825, "step": 2210 }, { "epoch": 13.136094674556213, "grad_norm": 6.703935146331787, "learning_rate": 0.0001, "loss": 0.0827, "step": 2220 }, { "epoch": 13.195266272189349, "grad_norm": 6.039135932922363, "learning_rate": 0.0001, "loss": 0.0792, "step": 2230 }, { "epoch": 13.254437869822485, "grad_norm": 5.717588424682617, "learning_rate": 0.0001, "loss": 0.0769, "step": 2240 }, { "epoch": 13.31360946745562, "grad_norm": 6.256304740905762, "learning_rate": 0.0001, "loss": 0.0827, "step": 2250 }, { "epoch": 13.372781065088757, "grad_norm": 5.793753623962402, "learning_rate": 0.0001, "loss": 0.083, "step": 2260 }, { "epoch": 13.431952662721894, "grad_norm": 6.221829414367676, "learning_rate": 0.0001, "loss": 0.0781, "step": 2270 }, { "epoch": 13.49112426035503, "grad_norm": 6.572075366973877, "learning_rate": 0.0001, "loss": 0.0833, "step": 2280 }, { "epoch": 13.550295857988166, "grad_norm": 7.390249252319336, "learning_rate": 0.0001, "loss": 0.0833, "step": 2290 }, { "epoch": 13.609467455621301, "grad_norm": 7.071598052978516, "learning_rate": 0.0001, "loss": 0.0847, "step": 2300 }, { "epoch": 13.668639053254438, "grad_norm": 6.4986724853515625, "learning_rate": 0.0001, "loss": 0.0823, "step": 2310 }, { "epoch": 13.727810650887575, "grad_norm": 6.140120029449463, "learning_rate": 0.0001, "loss": 0.0834, "step": 2320 }, { "epoch": 13.78698224852071, "grad_norm": 7.085366725921631, "learning_rate": 0.0001, "loss": 0.0811, "step": 2330 }, { "epoch": 13.846153846153847, "grad_norm": 5.794652462005615, "learning_rate": 0.0001, "loss": 0.0798, "step": 2340 }, { "epoch": 13.905325443786982, "grad_norm": 6.52014684677124, "learning_rate": 0.0001, "loss": 0.0782, "step": 2350 }, { "epoch": 13.964497041420119, "grad_norm": 6.642873287200928, "learning_rate": 0.0001, "loss": 0.0812, "step": 2360 }, { "epoch": 14.023668639053254, "grad_norm": 5.995068073272705, "learning_rate": 0.0001, "loss": 0.0796, "step": 2370 }, { "epoch": 14.08284023668639, "grad_norm": 6.190224647521973, "learning_rate": 0.0001, "loss": 0.078, "step": 2380 }, { "epoch": 14.142011834319527, "grad_norm": 6.190382480621338, "learning_rate": 0.0001, "loss": 0.0773, "step": 2390 }, { "epoch": 14.201183431952662, "grad_norm": 6.278234958648682, "learning_rate": 0.0001, "loss": 0.0788, "step": 2400 }, { "epoch": 14.2603550295858, "grad_norm": 6.13463020324707, "learning_rate": 0.0001, "loss": 0.078, "step": 2410 }, { "epoch": 14.319526627218934, "grad_norm": 5.632469177246094, "learning_rate": 0.0001, "loss": 0.0761, "step": 2420 }, { "epoch": 14.378698224852071, "grad_norm": 5.733285903930664, "learning_rate": 0.0001, "loss": 0.0754, "step": 2430 }, { "epoch": 14.437869822485208, "grad_norm": 5.600717067718506, "learning_rate": 0.0001, "loss": 0.0748, "step": 2440 }, { "epoch": 14.497041420118343, "grad_norm": 5.641643524169922, "learning_rate": 0.0001, "loss": 0.075, "step": 2450 }, { "epoch": 14.55621301775148, "grad_norm": 5.252694129943848, "learning_rate": 0.0001, "loss": 0.0752, "step": 2460 }, { "epoch": 14.615384615384615, "grad_norm": 5.068181037902832, "learning_rate": 0.0001, "loss": 0.0773, "step": 2470 }, { "epoch": 14.674556213017752, "grad_norm": 5.114899158477783, "learning_rate": 0.0001, "loss": 0.0757, "step": 2480 }, { "epoch": 14.733727810650887, "grad_norm": 5.831292629241943, "learning_rate": 0.0001, "loss": 0.0787, "step": 2490 }, { "epoch": 14.792899408284024, "grad_norm": 5.028982162475586, "learning_rate": 0.0001, "loss": 0.0761, "step": 2500 }, { "epoch": 14.85207100591716, "grad_norm": 5.014527797698975, "learning_rate": 0.0001, "loss": 0.0776, "step": 2510 }, { "epoch": 14.911242603550296, "grad_norm": 6.221988201141357, "learning_rate": 0.0001, "loss": 0.0775, "step": 2520 }, { "epoch": 14.970414201183432, "grad_norm": 5.2427568435668945, "learning_rate": 0.0001, "loss": 0.0792, "step": 2530 }, { "epoch": 15.029585798816568, "grad_norm": 6.16827917098999, "learning_rate": 0.0001, "loss": 0.0799, "step": 2540 }, { "epoch": 15.088757396449704, "grad_norm": 6.8596954345703125, "learning_rate": 0.0001, "loss": 0.0771, "step": 2550 }, { "epoch": 15.14792899408284, "grad_norm": 5.609550476074219, "learning_rate": 0.0001, "loss": 0.0745, "step": 2560 }, { "epoch": 15.207100591715976, "grad_norm": 5.917194366455078, "learning_rate": 0.0001, "loss": 0.0753, "step": 2570 }, { "epoch": 15.266272189349113, "grad_norm": 5.112740516662598, "learning_rate": 0.0001, "loss": 0.0767, "step": 2580 }, { "epoch": 15.325443786982248, "grad_norm": 5.134215831756592, "learning_rate": 0.0001, "loss": 0.0741, "step": 2590 }, { "epoch": 15.384615384615385, "grad_norm": 6.14559268951416, "learning_rate": 0.0001, "loss": 0.0727, "step": 2600 }, { "epoch": 15.44378698224852, "grad_norm": 6.110940456390381, "learning_rate": 0.0001, "loss": 0.0717, "step": 2610 }, { "epoch": 15.502958579881657, "grad_norm": 5.124517440795898, "learning_rate": 0.0001, "loss": 0.0723, "step": 2620 }, { "epoch": 15.562130177514792, "grad_norm": 4.919792175292969, "learning_rate": 0.0001, "loss": 0.0708, "step": 2630 }, { "epoch": 15.621301775147929, "grad_norm": 5.667159557342529, "learning_rate": 0.0001, "loss": 0.0705, "step": 2640 }, { "epoch": 15.680473372781066, "grad_norm": 4.846826553344727, "learning_rate": 0.0001, "loss": 0.0692, "step": 2650 }, { "epoch": 15.7396449704142, "grad_norm": 5.483339786529541, "learning_rate": 0.0001, "loss": 0.0737, "step": 2660 }, { "epoch": 15.798816568047338, "grad_norm": 5.044246673583984, "learning_rate": 0.0001, "loss": 0.0719, "step": 2670 }, { "epoch": 15.857988165680473, "grad_norm": 5.147960186004639, "learning_rate": 0.0001, "loss": 0.0705, "step": 2680 }, { "epoch": 15.91715976331361, "grad_norm": 4.518064022064209, "learning_rate": 0.0001, "loss": 0.0695, "step": 2690 }, { "epoch": 15.976331360946746, "grad_norm": 5.760453224182129, "learning_rate": 0.0001, "loss": 0.0694, "step": 2700 }, { "epoch": 16.035502958579883, "grad_norm": 6.740314960479736, "learning_rate": 0.0001, "loss": 0.0731, "step": 2710 }, { "epoch": 16.09467455621302, "grad_norm": 5.94819450378418, "learning_rate": 0.0001, "loss": 0.071, "step": 2720 }, { "epoch": 16.153846153846153, "grad_norm": 4.684739112854004, "learning_rate": 0.0001, "loss": 0.0735, "step": 2730 }, { "epoch": 16.21301775147929, "grad_norm": 5.762418746948242, "learning_rate": 0.0001, "loss": 0.0739, "step": 2740 }, { "epoch": 16.272189349112427, "grad_norm": 5.973446369171143, "learning_rate": 0.0001, "loss": 0.0744, "step": 2750 }, { "epoch": 16.331360946745562, "grad_norm": 6.8376359939575195, "learning_rate": 0.0001, "loss": 0.0723, "step": 2760 }, { "epoch": 16.390532544378697, "grad_norm": 5.875669956207275, "learning_rate": 0.0001, "loss": 0.0754, "step": 2770 }, { "epoch": 16.449704142011836, "grad_norm": 5.758673667907715, "learning_rate": 0.0001, "loss": 0.0699, "step": 2780 }, { "epoch": 16.50887573964497, "grad_norm": 5.225910186767578, "learning_rate": 0.0001, "loss": 0.0701, "step": 2790 }, { "epoch": 16.568047337278106, "grad_norm": 4.90148401260376, "learning_rate": 0.0001, "loss": 0.0667, "step": 2800 }, { "epoch": 16.62721893491124, "grad_norm": 4.839040756225586, "learning_rate": 0.0001, "loss": 0.0674, "step": 2810 }, { "epoch": 16.68639053254438, "grad_norm": 5.368873119354248, "learning_rate": 0.0001, "loss": 0.0651, "step": 2820 }, { "epoch": 16.745562130177515, "grad_norm": 4.516157150268555, "learning_rate": 0.0001, "loss": 0.0687, "step": 2830 }, { "epoch": 16.80473372781065, "grad_norm": 4.945244789123535, "learning_rate": 0.0001, "loss": 0.0681, "step": 2840 }, { "epoch": 16.86390532544379, "grad_norm": 5.3112616539001465, "learning_rate": 0.0001, "loss": 0.0704, "step": 2850 }, { "epoch": 16.923076923076923, "grad_norm": 5.015364646911621, "learning_rate": 0.0001, "loss": 0.0689, "step": 2860 }, { "epoch": 16.98224852071006, "grad_norm": 4.8346428871154785, "learning_rate": 0.0001, "loss": 0.0659, "step": 2870 }, { "epoch": 17.041420118343197, "grad_norm": 5.080626487731934, "learning_rate": 0.0001, "loss": 0.0684, "step": 2880 }, { "epoch": 17.100591715976332, "grad_norm": 5.496791839599609, "learning_rate": 0.0001, "loss": 0.0685, "step": 2890 }, { "epoch": 17.159763313609467, "grad_norm": 5.485905647277832, "learning_rate": 0.0001, "loss": 0.0705, "step": 2900 }, { "epoch": 17.218934911242602, "grad_norm": 4.698921203613281, "learning_rate": 0.0001, "loss": 0.0674, "step": 2910 }, { "epoch": 17.27810650887574, "grad_norm": 4.950345993041992, "learning_rate": 0.0001, "loss": 0.0659, "step": 2920 }, { "epoch": 17.337278106508876, "grad_norm": 5.225870132446289, "learning_rate": 0.0001, "loss": 0.0634, "step": 2930 }, { "epoch": 17.39644970414201, "grad_norm": 5.579608917236328, "learning_rate": 0.0001, "loss": 0.0686, "step": 2940 }, { "epoch": 17.45562130177515, "grad_norm": 4.690073013305664, "learning_rate": 0.0001, "loss": 0.0635, "step": 2950 }, { "epoch": 17.514792899408285, "grad_norm": 5.519926071166992, "learning_rate": 0.0001, "loss": 0.0627, "step": 2960 }, { "epoch": 17.57396449704142, "grad_norm": 4.3772783279418945, "learning_rate": 0.0001, "loss": 0.0607, "step": 2970 }, { "epoch": 17.633136094674555, "grad_norm": 5.161226749420166, "learning_rate": 0.0001, "loss": 0.068, "step": 2980 }, { "epoch": 17.692307692307693, "grad_norm": 4.978453159332275, "learning_rate": 0.0001, "loss": 0.0634, "step": 2990 }, { "epoch": 17.75147928994083, "grad_norm": 4.79890251159668, "learning_rate": 0.0001, "loss": 0.0668, "step": 3000 }, { "epoch": 17.810650887573964, "grad_norm": 4.7277750968933105, "learning_rate": 0.0001, "loss": 0.0638, "step": 3010 }, { "epoch": 17.869822485207102, "grad_norm": 4.501612186431885, "learning_rate": 0.0001, "loss": 0.0646, "step": 3020 }, { "epoch": 17.928994082840237, "grad_norm": 4.2967305183410645, "learning_rate": 0.0001, "loss": 0.0612, "step": 3030 }, { "epoch": 17.988165680473372, "grad_norm": 4.73740291595459, "learning_rate": 0.0001, "loss": 0.0643, "step": 3040 }, { "epoch": 18.047337278106507, "grad_norm": 4.182480812072754, "learning_rate": 0.0001, "loss": 0.0617, "step": 3050 }, { "epoch": 18.106508875739646, "grad_norm": 4.691254615783691, "learning_rate": 0.0001, "loss": 0.0639, "step": 3060 }, { "epoch": 18.16568047337278, "grad_norm": 5.7945098876953125, "learning_rate": 0.0001, "loss": 0.0636, "step": 3070 }, { "epoch": 18.224852071005916, "grad_norm": 5.551990509033203, "learning_rate": 0.0001, "loss": 0.0639, "step": 3080 }, { "epoch": 18.284023668639055, "grad_norm": 4.545684337615967, "learning_rate": 0.0001, "loss": 0.0619, "step": 3090 }, { "epoch": 18.34319526627219, "grad_norm": 4.125000953674316, "learning_rate": 0.0001, "loss": 0.0655, "step": 3100 }, { "epoch": 18.402366863905325, "grad_norm": 4.190248012542725, "learning_rate": 0.0001, "loss": 0.0633, "step": 3110 }, { "epoch": 18.46153846153846, "grad_norm": 5.697264194488525, "learning_rate": 0.0001, "loss": 0.0669, "step": 3120 }, { "epoch": 18.5207100591716, "grad_norm": 5.412989139556885, "learning_rate": 0.0001, "loss": 0.0676, "step": 3130 }, { "epoch": 18.579881656804734, "grad_norm": 5.2546210289001465, "learning_rate": 0.0001, "loss": 0.0623, "step": 3140 }, { "epoch": 18.63905325443787, "grad_norm": 5.374125003814697, "learning_rate": 0.0001, "loss": 0.0615, "step": 3150 }, { "epoch": 18.698224852071007, "grad_norm": 4.924867153167725, "learning_rate": 0.0001, "loss": 0.065, "step": 3160 }, { "epoch": 18.757396449704142, "grad_norm": 4.952323913574219, "learning_rate": 0.0001, "loss": 0.0657, "step": 3170 }, { "epoch": 18.816568047337277, "grad_norm": 4.7922773361206055, "learning_rate": 0.0001, "loss": 0.0627, "step": 3180 }, { "epoch": 18.875739644970416, "grad_norm": 4.317700386047363, "learning_rate": 0.0001, "loss": 0.0598, "step": 3190 }, { "epoch": 18.93491124260355, "grad_norm": 4.585310459136963, "learning_rate": 0.0001, "loss": 0.0627, "step": 3200 }, { "epoch": 18.994082840236686, "grad_norm": 4.464968204498291, "learning_rate": 0.0001, "loss": 0.0611, "step": 3210 }, { "epoch": 19.05325443786982, "grad_norm": 4.571560382843018, "learning_rate": 0.0001, "loss": 0.0643, "step": 3220 }, { "epoch": 19.11242603550296, "grad_norm": 4.702658653259277, "learning_rate": 0.0001, "loss": 0.0633, "step": 3230 }, { "epoch": 19.171597633136095, "grad_norm": 4.890661716461182, "learning_rate": 0.0001, "loss": 0.0652, "step": 3240 }, { "epoch": 19.23076923076923, "grad_norm": 4.615457057952881, "learning_rate": 0.0001, "loss": 0.0635, "step": 3250 }, { "epoch": 19.28994082840237, "grad_norm": 4.495664596557617, "learning_rate": 0.0001, "loss": 0.0627, "step": 3260 }, { "epoch": 19.349112426035504, "grad_norm": 4.19334077835083, "learning_rate": 0.0001, "loss": 0.0592, "step": 3270 }, { "epoch": 19.40828402366864, "grad_norm": 4.484364032745361, "learning_rate": 0.0001, "loss": 0.0598, "step": 3280 }, { "epoch": 19.467455621301774, "grad_norm": 4.6658549308776855, "learning_rate": 0.0001, "loss": 0.0575, "step": 3290 }, { "epoch": 19.526627218934912, "grad_norm": 4.810914993286133, "learning_rate": 0.0001, "loss": 0.0605, "step": 3300 }, { "epoch": 19.585798816568047, "grad_norm": 4.6907830238342285, "learning_rate": 0.0001, "loss": 0.0594, "step": 3310 }, { "epoch": 19.644970414201183, "grad_norm": 4.44071102142334, "learning_rate": 0.0001, "loss": 0.0598, "step": 3320 }, { "epoch": 19.70414201183432, "grad_norm": 4.208802700042725, "learning_rate": 0.0001, "loss": 0.0619, "step": 3330 }, { "epoch": 19.763313609467456, "grad_norm": 4.254820823669434, "learning_rate": 0.0001, "loss": 0.0581, "step": 3340 }, { "epoch": 19.82248520710059, "grad_norm": 3.9939687252044678, "learning_rate": 0.0001, "loss": 0.0578, "step": 3350 }, { "epoch": 19.881656804733726, "grad_norm": 4.191391468048096, "learning_rate": 0.0001, "loss": 0.0595, "step": 3360 }, { "epoch": 19.940828402366865, "grad_norm": 4.916447162628174, "learning_rate": 0.0001, "loss": 0.0592, "step": 3370 }, { "epoch": 20.0, "grad_norm": 3.8993639945983887, "learning_rate": 0.0001, "loss": 0.0576, "step": 3380 }, { "epoch": 20.059171597633135, "grad_norm": 4.171012878417969, "learning_rate": 0.0001, "loss": 0.0585, "step": 3390 }, { "epoch": 20.118343195266274, "grad_norm": 3.9670820236206055, "learning_rate": 0.0001, "loss": 0.0594, "step": 3400 }, { "epoch": 20.17751479289941, "grad_norm": 4.461684703826904, "learning_rate": 0.0001, "loss": 0.061, "step": 3410 }, { "epoch": 20.236686390532544, "grad_norm": 4.963307857513428, "learning_rate": 0.0001, "loss": 0.057, "step": 3420 }, { "epoch": 20.29585798816568, "grad_norm": 3.5942862033843994, "learning_rate": 0.0001, "loss": 0.0571, "step": 3430 }, { "epoch": 20.355029585798817, "grad_norm": 4.112346172332764, "learning_rate": 0.0001, "loss": 0.0596, "step": 3440 }, { "epoch": 20.414201183431953, "grad_norm": 4.4417829513549805, "learning_rate": 0.0001, "loss": 0.0601, "step": 3450 }, { "epoch": 20.473372781065088, "grad_norm": 4.229283332824707, "learning_rate": 0.0001, "loss": 0.0584, "step": 3460 }, { "epoch": 20.532544378698226, "grad_norm": 4.025295257568359, "learning_rate": 0.0001, "loss": 0.0617, "step": 3470 }, { "epoch": 20.59171597633136, "grad_norm": 3.750819683074951, "learning_rate": 0.0001, "loss": 0.059, "step": 3480 }, { "epoch": 20.650887573964496, "grad_norm": 4.331761360168457, "learning_rate": 0.0001, "loss": 0.0585, "step": 3490 }, { "epoch": 20.71005917159763, "grad_norm": 3.9007086753845215, "learning_rate": 0.0001, "loss": 0.0586, "step": 3500 }, { "epoch": 20.76923076923077, "grad_norm": 4.194283485412598, "learning_rate": 0.0001, "loss": 0.0583, "step": 3510 }, { "epoch": 20.828402366863905, "grad_norm": 3.815523386001587, "learning_rate": 0.0001, "loss": 0.0578, "step": 3520 }, { "epoch": 20.88757396449704, "grad_norm": 4.607956886291504, "learning_rate": 0.0001, "loss": 0.0588, "step": 3530 }, { "epoch": 20.94674556213018, "grad_norm": 3.908703565597534, "learning_rate": 0.0001, "loss": 0.0571, "step": 3540 }, { "epoch": 21.005917159763314, "grad_norm": 4.0248212814331055, "learning_rate": 0.0001, "loss": 0.0578, "step": 3550 }, { "epoch": 21.06508875739645, "grad_norm": 4.371763706207275, "learning_rate": 0.0001, "loss": 0.0597, "step": 3560 }, { "epoch": 21.124260355029588, "grad_norm": 3.9722981452941895, "learning_rate": 0.0001, "loss": 0.0604, "step": 3570 }, { "epoch": 21.183431952662723, "grad_norm": 3.8038413524627686, "learning_rate": 0.0001, "loss": 0.0591, "step": 3580 }, { "epoch": 21.242603550295858, "grad_norm": 4.230653285980225, "learning_rate": 0.0001, "loss": 0.0584, "step": 3590 }, { "epoch": 21.301775147928993, "grad_norm": 4.20985221862793, "learning_rate": 0.0001, "loss": 0.0561, "step": 3600 }, { "epoch": 21.36094674556213, "grad_norm": 4.018280029296875, "learning_rate": 0.0001, "loss": 0.0562, "step": 3610 }, { "epoch": 21.420118343195266, "grad_norm": 4.242932319641113, "learning_rate": 0.0001, "loss": 0.0581, "step": 3620 }, { "epoch": 21.4792899408284, "grad_norm": 3.529261350631714, "learning_rate": 0.0001, "loss": 0.0583, "step": 3630 }, { "epoch": 21.53846153846154, "grad_norm": 4.085992813110352, "learning_rate": 0.0001, "loss": 0.0557, "step": 3640 }, { "epoch": 21.597633136094675, "grad_norm": 3.731935501098633, "learning_rate": 0.0001, "loss": 0.0584, "step": 3650 }, { "epoch": 21.65680473372781, "grad_norm": 4.552347660064697, "learning_rate": 0.0001, "loss": 0.0566, "step": 3660 }, { "epoch": 21.715976331360945, "grad_norm": 4.493063449859619, "learning_rate": 0.0001, "loss": 0.0566, "step": 3670 }, { "epoch": 21.775147928994084, "grad_norm": 3.812150239944458, "learning_rate": 0.0001, "loss": 0.0555, "step": 3680 }, { "epoch": 21.83431952662722, "grad_norm": 3.6088874340057373, "learning_rate": 0.0001, "loss": 0.0588, "step": 3690 }, { "epoch": 21.893491124260354, "grad_norm": 5.0085296630859375, "learning_rate": 0.0001, "loss": 0.0601, "step": 3700 }, { "epoch": 21.952662721893493, "grad_norm": 3.730077028274536, "learning_rate": 0.0001, "loss": 0.0589, "step": 3710 }, { "epoch": 22.011834319526628, "grad_norm": 4.414837837219238, "learning_rate": 0.0001, "loss": 0.0566, "step": 3720 }, { "epoch": 22.071005917159763, "grad_norm": 3.701369524002075, "learning_rate": 0.0001, "loss": 0.0555, "step": 3730 }, { "epoch": 22.130177514792898, "grad_norm": 4.643394947052002, "learning_rate": 0.0001, "loss": 0.0563, "step": 3740 }, { "epoch": 22.189349112426036, "grad_norm": 4.171213626861572, "learning_rate": 0.0001, "loss": 0.0566, "step": 3750 }, { "epoch": 22.24852071005917, "grad_norm": 4.128825664520264, "learning_rate": 0.0001, "loss": 0.0554, "step": 3760 }, { "epoch": 22.307692307692307, "grad_norm": 3.825991153717041, "learning_rate": 0.0001, "loss": 0.057, "step": 3770 }, { "epoch": 22.366863905325445, "grad_norm": 3.8491384983062744, "learning_rate": 0.0001, "loss": 0.0557, "step": 3780 }, { "epoch": 22.42603550295858, "grad_norm": 3.8951408863067627, "learning_rate": 0.0001, "loss": 0.0578, "step": 3790 }, { "epoch": 22.485207100591715, "grad_norm": 4.184350490570068, "learning_rate": 0.0001, "loss": 0.0551, "step": 3800 }, { "epoch": 22.54437869822485, "grad_norm": 3.931697368621826, "learning_rate": 0.0001, "loss": 0.0579, "step": 3810 }, { "epoch": 22.60355029585799, "grad_norm": 4.078036308288574, "learning_rate": 0.0001, "loss": 0.0606, "step": 3820 }, { "epoch": 22.662721893491124, "grad_norm": 3.798717975616455, "learning_rate": 0.0001, "loss": 0.0551, "step": 3830 }, { "epoch": 22.72189349112426, "grad_norm": 3.4056642055511475, "learning_rate": 0.0001, "loss": 0.0537, "step": 3840 }, { "epoch": 22.781065088757398, "grad_norm": 3.554065704345703, "learning_rate": 0.0001, "loss": 0.052, "step": 3850 }, { "epoch": 22.840236686390533, "grad_norm": 4.242760181427002, "learning_rate": 0.0001, "loss": 0.0543, "step": 3860 }, { "epoch": 22.899408284023668, "grad_norm": 3.3069067001342773, "learning_rate": 0.0001, "loss": 0.055, "step": 3870 }, { "epoch": 22.958579881656803, "grad_norm": 3.9037585258483887, "learning_rate": 0.0001, "loss": 0.0542, "step": 3880 }, { "epoch": 23.01775147928994, "grad_norm": 4.014892101287842, "learning_rate": 0.0001, "loss": 0.0545, "step": 3890 }, { "epoch": 23.076923076923077, "grad_norm": 3.604482889175415, "learning_rate": 0.0001, "loss": 0.0552, "step": 3900 }, { "epoch": 23.13609467455621, "grad_norm": 3.4665873050689697, "learning_rate": 0.0001, "loss": 0.053, "step": 3910 }, { "epoch": 23.19526627218935, "grad_norm": 3.513502836227417, "learning_rate": 0.0001, "loss": 0.0513, "step": 3920 }, { "epoch": 23.254437869822485, "grad_norm": 2.978473424911499, "learning_rate": 0.0001, "loss": 0.0518, "step": 3930 }, { "epoch": 23.31360946745562, "grad_norm": 3.6420693397521973, "learning_rate": 0.0001, "loss": 0.0512, "step": 3940 }, { "epoch": 23.37278106508876, "grad_norm": 3.730600357055664, "learning_rate": 0.0001, "loss": 0.0519, "step": 3950 }, { "epoch": 23.431952662721894, "grad_norm": 3.7250564098358154, "learning_rate": 0.0001, "loss": 0.0558, "step": 3960 }, { "epoch": 23.49112426035503, "grad_norm": 3.734006643295288, "learning_rate": 0.0001, "loss": 0.0529, "step": 3970 }, { "epoch": 23.550295857988164, "grad_norm": 4.51144552230835, "learning_rate": 0.0001, "loss": 0.0524, "step": 3980 }, { "epoch": 23.609467455621303, "grad_norm": 3.9028196334838867, "learning_rate": 0.0001, "loss": 0.0533, "step": 3990 }, { "epoch": 23.668639053254438, "grad_norm": 3.373577833175659, "learning_rate": 0.0001, "loss": 0.0502, "step": 4000 }, { "epoch": 23.727810650887573, "grad_norm": 3.3119897842407227, "learning_rate": 0.0001, "loss": 0.0509, "step": 4010 }, { "epoch": 23.78698224852071, "grad_norm": 3.2265708446502686, "learning_rate": 0.0001, "loss": 0.0533, "step": 4020 }, { "epoch": 23.846153846153847, "grad_norm": 3.7243387699127197, "learning_rate": 0.0001, "loss": 0.0512, "step": 4030 }, { "epoch": 23.90532544378698, "grad_norm": 3.4187121391296387, "learning_rate": 0.0001, "loss": 0.052, "step": 4040 }, { "epoch": 23.964497041420117, "grad_norm": 3.660780429840088, "learning_rate": 0.0001, "loss": 0.0542, "step": 4050 }, { "epoch": 24.023668639053255, "grad_norm": 3.5427377223968506, "learning_rate": 0.0001, "loss": 0.0535, "step": 4060 }, { "epoch": 24.08284023668639, "grad_norm": 4.219526767730713, "learning_rate": 0.0001, "loss": 0.053, "step": 4070 }, { "epoch": 24.142011834319526, "grad_norm": 3.995042562484741, "learning_rate": 0.0001, "loss": 0.0515, "step": 4080 }, { "epoch": 24.201183431952664, "grad_norm": 3.6195766925811768, "learning_rate": 0.0001, "loss": 0.0526, "step": 4090 }, { "epoch": 24.2603550295858, "grad_norm": 3.3234012126922607, "learning_rate": 0.0001, "loss": 0.0511, "step": 4100 }, { "epoch": 24.319526627218934, "grad_norm": 3.398486375808716, "learning_rate": 0.0001, "loss": 0.0511, "step": 4110 }, { "epoch": 24.37869822485207, "grad_norm": 3.367313861846924, "learning_rate": 0.0001, "loss": 0.0512, "step": 4120 }, { "epoch": 24.437869822485208, "grad_norm": 3.353222608566284, "learning_rate": 0.0001, "loss": 0.0531, "step": 4130 }, { "epoch": 24.497041420118343, "grad_norm": 3.0433566570281982, "learning_rate": 0.0001, "loss": 0.0542, "step": 4140 }, { "epoch": 24.556213017751478, "grad_norm": 3.9570515155792236, "learning_rate": 0.0001, "loss": 0.0538, "step": 4150 }, { "epoch": 24.615384615384617, "grad_norm": 3.636683464050293, "learning_rate": 0.0001, "loss": 0.0521, "step": 4160 }, { "epoch": 24.674556213017752, "grad_norm": 3.8787691593170166, "learning_rate": 0.0001, "loss": 0.0523, "step": 4170 }, { "epoch": 24.733727810650887, "grad_norm": 3.0641114711761475, "learning_rate": 0.0001, "loss": 0.0515, "step": 4180 }, { "epoch": 24.792899408284022, "grad_norm": 2.9384024143218994, "learning_rate": 0.0001, "loss": 0.0523, "step": 4190 }, { "epoch": 24.85207100591716, "grad_norm": 3.4934983253479004, "learning_rate": 0.0001, "loss": 0.0522, "step": 4200 }, { "epoch": 24.911242603550296, "grad_norm": 3.90394926071167, "learning_rate": 0.0001, "loss": 0.0507, "step": 4210 }, { "epoch": 24.97041420118343, "grad_norm": 3.584527015686035, "learning_rate": 0.0001, "loss": 0.0538, "step": 4220 }, { "epoch": 25.02958579881657, "grad_norm": 3.329538345336914, "learning_rate": 0.0001, "loss": 0.0552, "step": 4230 }, { "epoch": 25.088757396449704, "grad_norm": 3.6959550380706787, "learning_rate": 0.0001, "loss": 0.052, "step": 4240 }, { "epoch": 25.14792899408284, "grad_norm": 3.3862521648406982, "learning_rate": 0.0001, "loss": 0.0521, "step": 4250 }, { "epoch": 25.207100591715978, "grad_norm": 3.348890781402588, "learning_rate": 0.0001, "loss": 0.0523, "step": 4260 }, { "epoch": 25.266272189349113, "grad_norm": 3.2840206623077393, "learning_rate": 0.0001, "loss": 0.0526, "step": 4270 }, { "epoch": 25.325443786982248, "grad_norm": 3.6258771419525146, "learning_rate": 0.0001, "loss": 0.0533, "step": 4280 }, { "epoch": 25.384615384615383, "grad_norm": 3.4843764305114746, "learning_rate": 0.0001, "loss": 0.0523, "step": 4290 }, { "epoch": 25.443786982248522, "grad_norm": 3.1796298027038574, "learning_rate": 0.0001, "loss": 0.0499, "step": 4300 }, { "epoch": 25.502958579881657, "grad_norm": 3.5479328632354736, "learning_rate": 0.0001, "loss": 0.0508, "step": 4310 }, { "epoch": 25.562130177514792, "grad_norm": 3.268686056137085, "learning_rate": 0.0001, "loss": 0.0502, "step": 4320 }, { "epoch": 25.62130177514793, "grad_norm": 3.7202608585357666, "learning_rate": 0.0001, "loss": 0.0526, "step": 4330 }, { "epoch": 25.680473372781066, "grad_norm": 3.890648126602173, "learning_rate": 0.0001, "loss": 0.0529, "step": 4340 }, { "epoch": 25.7396449704142, "grad_norm": 2.921858310699463, "learning_rate": 0.0001, "loss": 0.0487, "step": 4350 }, { "epoch": 25.798816568047336, "grad_norm": 2.7523033618927, "learning_rate": 0.0001, "loss": 0.0511, "step": 4360 }, { "epoch": 25.857988165680474, "grad_norm": 3.05062198638916, "learning_rate": 0.0001, "loss": 0.051, "step": 4370 }, { "epoch": 25.91715976331361, "grad_norm": 3.3395676612854004, "learning_rate": 0.0001, "loss": 0.0498, "step": 4380 }, { "epoch": 25.976331360946745, "grad_norm": 3.496117115020752, "learning_rate": 0.0001, "loss": 0.0545, "step": 4390 }, { "epoch": 26.035502958579883, "grad_norm": 3.011644124984741, "learning_rate": 0.0001, "loss": 0.0502, "step": 4400 }, { "epoch": 26.09467455621302, "grad_norm": 3.873725652694702, "learning_rate": 0.0001, "loss": 0.0548, "step": 4410 }, { "epoch": 26.153846153846153, "grad_norm": 4.164061546325684, "learning_rate": 0.0001, "loss": 0.0524, "step": 4420 }, { "epoch": 26.21301775147929, "grad_norm": 4.157295227050781, "learning_rate": 0.0001, "loss": 0.0522, "step": 4430 }, { "epoch": 26.272189349112427, "grad_norm": 3.528493881225586, "learning_rate": 0.0001, "loss": 0.049, "step": 4440 }, { "epoch": 26.331360946745562, "grad_norm": 4.085198879241943, "learning_rate": 0.0001, "loss": 0.0496, "step": 4450 }, { "epoch": 26.390532544378697, "grad_norm": 4.024105072021484, "learning_rate": 0.0001, "loss": 0.0503, "step": 4460 }, { "epoch": 26.449704142011836, "grad_norm": 3.6402428150177, "learning_rate": 0.0001, "loss": 0.0506, "step": 4470 }, { "epoch": 26.50887573964497, "grad_norm": 3.30226731300354, "learning_rate": 0.0001, "loss": 0.0485, "step": 4480 }, { "epoch": 26.568047337278106, "grad_norm": 3.2739925384521484, "learning_rate": 0.0001, "loss": 0.0483, "step": 4490 }, { "epoch": 26.62721893491124, "grad_norm": 2.8552041053771973, "learning_rate": 0.0001, "loss": 0.0488, "step": 4500 }, { "epoch": 26.68639053254438, "grad_norm": 3.918619155883789, "learning_rate": 0.0001, "loss": 0.0476, "step": 4510 }, { "epoch": 26.745562130177515, "grad_norm": 3.399871349334717, "learning_rate": 0.0001, "loss": 0.0464, "step": 4520 }, { "epoch": 26.80473372781065, "grad_norm": 2.7649810314178467, "learning_rate": 0.0001, "loss": 0.0489, "step": 4530 }, { "epoch": 26.86390532544379, "grad_norm": 3.3189356327056885, "learning_rate": 0.0001, "loss": 0.0499, "step": 4540 }, { "epoch": 26.923076923076923, "grad_norm": 3.706620693206787, "learning_rate": 0.0001, "loss": 0.0488, "step": 4550 }, { "epoch": 26.98224852071006, "grad_norm": 2.588131904602051, "learning_rate": 0.0001, "loss": 0.0482, "step": 4560 }, { "epoch": 27.041420118343197, "grad_norm": 3.447209119796753, "learning_rate": 0.0001, "loss": 0.05, "step": 4570 }, { "epoch": 27.100591715976332, "grad_norm": 3.263761043548584, "learning_rate": 0.0001, "loss": 0.0507, "step": 4580 }, { "epoch": 27.159763313609467, "grad_norm": 3.568690299987793, "learning_rate": 0.0001, "loss": 0.0485, "step": 4590 }, { "epoch": 27.218934911242602, "grad_norm": 3.20157527923584, "learning_rate": 0.0001, "loss": 0.0488, "step": 4600 }, { "epoch": 27.27810650887574, "grad_norm": 3.2346811294555664, "learning_rate": 0.0001, "loss": 0.0495, "step": 4610 }, { "epoch": 27.337278106508876, "grad_norm": 3.430814027786255, "learning_rate": 0.0001, "loss": 0.0491, "step": 4620 }, { "epoch": 27.39644970414201, "grad_norm": 3.30088210105896, "learning_rate": 0.0001, "loss": 0.0504, "step": 4630 }, { "epoch": 27.45562130177515, "grad_norm": 3.387882709503174, "learning_rate": 0.0001, "loss": 0.0472, "step": 4640 }, { "epoch": 27.514792899408285, "grad_norm": 3.3953683376312256, "learning_rate": 0.0001, "loss": 0.048, "step": 4650 }, { "epoch": 27.57396449704142, "grad_norm": 3.596088409423828, "learning_rate": 0.0001, "loss": 0.0499, "step": 4660 }, { "epoch": 27.633136094674555, "grad_norm": 2.8967795372009277, "learning_rate": 0.0001, "loss": 0.048, "step": 4670 }, { "epoch": 27.692307692307693, "grad_norm": 3.1011993885040283, "learning_rate": 0.0001, "loss": 0.0487, "step": 4680 }, { "epoch": 27.75147928994083, "grad_norm": 2.729262590408325, "learning_rate": 0.0001, "loss": 0.0464, "step": 4690 }, { "epoch": 27.810650887573964, "grad_norm": 2.7790896892547607, "learning_rate": 0.0001, "loss": 0.0473, "step": 4700 }, { "epoch": 27.869822485207102, "grad_norm": 3.0907204151153564, "learning_rate": 0.0001, "loss": 0.0494, "step": 4710 }, { "epoch": 27.928994082840237, "grad_norm": 2.7839932441711426, "learning_rate": 0.0001, "loss": 0.0488, "step": 4720 }, { "epoch": 27.988165680473372, "grad_norm": 4.726795196533203, "learning_rate": 0.0001, "loss": 0.0504, "step": 4730 }, { "epoch": 28.047337278106507, "grad_norm": 3.1381053924560547, "learning_rate": 0.0001, "loss": 0.0477, "step": 4740 }, { "epoch": 28.106508875739646, "grad_norm": 2.9918739795684814, "learning_rate": 0.0001, "loss": 0.0465, "step": 4750 }, { "epoch": 28.16568047337278, "grad_norm": 2.5469489097595215, "learning_rate": 0.0001, "loss": 0.0477, "step": 4760 }, { "epoch": 28.224852071005916, "grad_norm": 2.6705563068389893, "learning_rate": 0.0001, "loss": 0.0474, "step": 4770 }, { "epoch": 28.284023668639055, "grad_norm": 3.340939521789551, "learning_rate": 0.0001, "loss": 0.0471, "step": 4780 }, { "epoch": 28.34319526627219, "grad_norm": 3.2973732948303223, "learning_rate": 0.0001, "loss": 0.0493, "step": 4790 }, { "epoch": 28.402366863905325, "grad_norm": 2.9594225883483887, "learning_rate": 0.0001, "loss": 0.0499, "step": 4800 }, { "epoch": 28.46153846153846, "grad_norm": 2.7446987628936768, "learning_rate": 0.0001, "loss": 0.0486, "step": 4810 }, { "epoch": 28.5207100591716, "grad_norm": 3.203244209289551, "learning_rate": 0.0001, "loss": 0.0476, "step": 4820 }, { "epoch": 28.579881656804734, "grad_norm": 2.95894193649292, "learning_rate": 0.0001, "loss": 0.049, "step": 4830 }, { "epoch": 28.63905325443787, "grad_norm": 3.9406661987304688, "learning_rate": 0.0001, "loss": 0.0471, "step": 4840 }, { "epoch": 28.698224852071007, "grad_norm": 2.8097972869873047, "learning_rate": 0.0001, "loss": 0.0476, "step": 4850 }, { "epoch": 28.757396449704142, "grad_norm": 4.286942958831787, "learning_rate": 0.0001, "loss": 0.0503, "step": 4860 }, { "epoch": 28.816568047337277, "grad_norm": 2.8839805126190186, "learning_rate": 0.0001, "loss": 0.0473, "step": 4870 }, { "epoch": 28.875739644970416, "grad_norm": 3.6535980701446533, "learning_rate": 0.0001, "loss": 0.0487, "step": 4880 }, { "epoch": 28.93491124260355, "grad_norm": 2.854344606399536, "learning_rate": 0.0001, "loss": 0.0465, "step": 4890 }, { "epoch": 28.994082840236686, "grad_norm": 3.467682123184204, "learning_rate": 0.0001, "loss": 0.0441, "step": 4900 }, { "epoch": 29.05325443786982, "grad_norm": 2.890685796737671, "learning_rate": 0.0001, "loss": 0.046, "step": 4910 }, { "epoch": 29.11242603550296, "grad_norm": 2.8189759254455566, "learning_rate": 0.0001, "loss": 0.0453, "step": 4920 }, { "epoch": 29.171597633136095, "grad_norm": 3.1231541633605957, "learning_rate": 0.0001, "loss": 0.0457, "step": 4930 }, { "epoch": 29.23076923076923, "grad_norm": 2.984151840209961, "learning_rate": 0.0001, "loss": 0.0464, "step": 4940 }, { "epoch": 29.28994082840237, "grad_norm": 2.881073474884033, "learning_rate": 0.0001, "loss": 0.0457, "step": 4950 }, { "epoch": 29.349112426035504, "grad_norm": 2.8085427284240723, "learning_rate": 0.0001, "loss": 0.0503, "step": 4960 }, { "epoch": 29.40828402366864, "grad_norm": 2.727107286453247, "learning_rate": 0.0001, "loss": 0.0477, "step": 4970 }, { "epoch": 29.467455621301774, "grad_norm": 2.7320921421051025, "learning_rate": 0.0001, "loss": 0.0456, "step": 4980 }, { "epoch": 29.526627218934912, "grad_norm": 2.7662909030914307, "learning_rate": 0.0001, "loss": 0.049, "step": 4990 }, { "epoch": 29.585798816568047, "grad_norm": 3.1728057861328125, "learning_rate": 0.0001, "loss": 0.0459, "step": 5000 }, { "epoch": 29.644970414201183, "grad_norm": 3.5613670349121094, "learning_rate": 0.0001, "loss": 0.046, "step": 5010 }, { "epoch": 29.70414201183432, "grad_norm": 2.885639190673828, "learning_rate": 0.0001, "loss": 0.0481, "step": 5020 }, { "epoch": 29.763313609467456, "grad_norm": 3.0232927799224854, "learning_rate": 0.0001, "loss": 0.0446, "step": 5030 }, { "epoch": 29.82248520710059, "grad_norm": 2.6348934173583984, "learning_rate": 0.0001, "loss": 0.0459, "step": 5040 }, { "epoch": 29.881656804733726, "grad_norm": 2.8704535961151123, "learning_rate": 0.0001, "loss": 0.0454, "step": 5050 }, { "epoch": 29.940828402366865, "grad_norm": 2.537104845046997, "learning_rate": 0.0001, "loss": 0.0441, "step": 5060 }, { "epoch": 30.0, "grad_norm": 2.582810401916504, "learning_rate": 0.0001, "loss": 0.0457, "step": 5070 }, { "epoch": 30.059171597633135, "grad_norm": 2.5242960453033447, "learning_rate": 0.0001, "loss": 0.0465, "step": 5080 }, { "epoch": 30.118343195266274, "grad_norm": 3.7850000858306885, "learning_rate": 0.0001, "loss": 0.0489, "step": 5090 }, { "epoch": 30.17751479289941, "grad_norm": 2.6954445838928223, "learning_rate": 0.0001, "loss": 0.048, "step": 5100 }, { "epoch": 30.236686390532544, "grad_norm": 4.050235748291016, "learning_rate": 0.0001, "loss": 0.0474, "step": 5110 }, { "epoch": 30.29585798816568, "grad_norm": 3.3292500972747803, "learning_rate": 0.0001, "loss": 0.0481, "step": 5120 }, { "epoch": 30.355029585798817, "grad_norm": 3.0318655967712402, "learning_rate": 0.0001, "loss": 0.0452, "step": 5130 }, { "epoch": 30.414201183431953, "grad_norm": 2.6199731826782227, "learning_rate": 0.0001, "loss": 0.0459, "step": 5140 }, { "epoch": 30.473372781065088, "grad_norm": 2.7977418899536133, "learning_rate": 0.0001, "loss": 0.045, "step": 5150 }, { "epoch": 30.532544378698226, "grad_norm": 3.4884848594665527, "learning_rate": 0.0001, "loss": 0.0462, "step": 5160 }, { "epoch": 30.59171597633136, "grad_norm": 2.7553083896636963, "learning_rate": 0.0001, "loss": 0.0439, "step": 5170 }, { "epoch": 30.650887573964496, "grad_norm": 3.0813822746276855, "learning_rate": 0.0001, "loss": 0.0457, "step": 5180 }, { "epoch": 30.71005917159763, "grad_norm": 2.7993428707122803, "learning_rate": 0.0001, "loss": 0.0443, "step": 5190 }, { "epoch": 30.76923076923077, "grad_norm": 2.818657636642456, "learning_rate": 0.0001, "loss": 0.0456, "step": 5200 }, { "epoch": 30.828402366863905, "grad_norm": 2.848336696624756, "learning_rate": 0.0001, "loss": 0.0441, "step": 5210 }, { "epoch": 30.88757396449704, "grad_norm": 2.4188754558563232, "learning_rate": 0.0001, "loss": 0.0442, "step": 5220 }, { "epoch": 30.94674556213018, "grad_norm": 2.600083351135254, "learning_rate": 0.0001, "loss": 0.0443, "step": 5230 }, { "epoch": 31.005917159763314, "grad_norm": 2.300328254699707, "learning_rate": 0.0001, "loss": 0.0459, "step": 5240 }, { "epoch": 31.06508875739645, "grad_norm": 2.6209487915039062, "learning_rate": 0.0001, "loss": 0.0452, "step": 5250 }, { "epoch": 31.124260355029588, "grad_norm": 2.987748384475708, "learning_rate": 0.0001, "loss": 0.0461, "step": 5260 }, { "epoch": 31.183431952662723, "grad_norm": 2.613102912902832, "learning_rate": 0.0001, "loss": 0.0472, "step": 5270 }, { "epoch": 31.242603550295858, "grad_norm": 2.5168280601501465, "learning_rate": 0.0001, "loss": 0.044, "step": 5280 }, { "epoch": 31.301775147928993, "grad_norm": 3.5299110412597656, "learning_rate": 0.0001, "loss": 0.0472, "step": 5290 }, { "epoch": 31.36094674556213, "grad_norm": 3.15811824798584, "learning_rate": 0.0001, "loss": 0.0468, "step": 5300 }, { "epoch": 31.420118343195266, "grad_norm": 2.2796008586883545, "learning_rate": 0.0001, "loss": 0.0444, "step": 5310 }, { "epoch": 31.4792899408284, "grad_norm": 2.5788214206695557, "learning_rate": 0.0001, "loss": 0.0442, "step": 5320 }, { "epoch": 31.53846153846154, "grad_norm": 2.801330089569092, "learning_rate": 0.0001, "loss": 0.0469, "step": 5330 }, { "epoch": 31.597633136094675, "grad_norm": 3.317760467529297, "learning_rate": 0.0001, "loss": 0.045, "step": 5340 }, { "epoch": 31.65680473372781, "grad_norm": 2.664973497390747, "learning_rate": 0.0001, "loss": 0.0511, "step": 5350 }, { "epoch": 31.715976331360945, "grad_norm": 3.7047719955444336, "learning_rate": 0.0001, "loss": 0.0491, "step": 5360 }, { "epoch": 31.775147928994084, "grad_norm": 2.385334014892578, "learning_rate": 0.0001, "loss": 0.046, "step": 5370 }, { "epoch": 31.83431952662722, "grad_norm": 2.969276189804077, "learning_rate": 0.0001, "loss": 0.0429, "step": 5380 }, { "epoch": 31.893491124260354, "grad_norm": 3.0268564224243164, "learning_rate": 0.0001, "loss": 0.044, "step": 5390 }, { "epoch": 31.952662721893493, "grad_norm": 2.694345235824585, "learning_rate": 0.0001, "loss": 0.0449, "step": 5400 }, { "epoch": 32.01183431952663, "grad_norm": 3.0807995796203613, "learning_rate": 0.0001, "loss": 0.0437, "step": 5410 }, { "epoch": 32.071005917159766, "grad_norm": 2.9837307929992676, "learning_rate": 0.0001, "loss": 0.0426, "step": 5420 }, { "epoch": 32.1301775147929, "grad_norm": 2.4928839206695557, "learning_rate": 0.0001, "loss": 0.0422, "step": 5430 }, { "epoch": 32.18934911242604, "grad_norm": 3.0148327350616455, "learning_rate": 0.0001, "loss": 0.0445, "step": 5440 }, { "epoch": 32.248520710059175, "grad_norm": 2.72228741645813, "learning_rate": 0.0001, "loss": 0.0421, "step": 5450 }, { "epoch": 32.30769230769231, "grad_norm": 3.302428960800171, "learning_rate": 0.0001, "loss": 0.0442, "step": 5460 }, { "epoch": 32.366863905325445, "grad_norm": 3.163412094116211, "learning_rate": 0.0001, "loss": 0.0434, "step": 5470 }, { "epoch": 32.42603550295858, "grad_norm": 2.7851364612579346, "learning_rate": 0.0001, "loss": 0.0401, "step": 5480 }, { "epoch": 32.485207100591715, "grad_norm": 2.821331739425659, "learning_rate": 0.0001, "loss": 0.0439, "step": 5490 }, { "epoch": 32.544378698224854, "grad_norm": 3.5618996620178223, "learning_rate": 0.0001, "loss": 0.0426, "step": 5500 }, { "epoch": 32.603550295857985, "grad_norm": 2.24983811378479, "learning_rate": 0.0001, "loss": 0.0461, "step": 5510 }, { "epoch": 32.662721893491124, "grad_norm": 2.925349712371826, "learning_rate": 0.0001, "loss": 0.0444, "step": 5520 }, { "epoch": 32.72189349112426, "grad_norm": 3.051266670227051, "learning_rate": 0.0001, "loss": 0.0439, "step": 5530 }, { "epoch": 32.781065088757394, "grad_norm": 2.977635145187378, "learning_rate": 0.0001, "loss": 0.0442, "step": 5540 }, { "epoch": 32.84023668639053, "grad_norm": 2.791907548904419, "learning_rate": 0.0001, "loss": 0.0436, "step": 5550 }, { "epoch": 32.89940828402367, "grad_norm": 2.556138753890991, "learning_rate": 0.0001, "loss": 0.0442, "step": 5560 }, { "epoch": 32.9585798816568, "grad_norm": 2.3039357662200928, "learning_rate": 0.0001, "loss": 0.0424, "step": 5570 }, { "epoch": 33.01775147928994, "grad_norm": 2.389270067214966, "learning_rate": 0.0001, "loss": 0.0441, "step": 5580 }, { "epoch": 33.07692307692308, "grad_norm": 2.700026750564575, "learning_rate": 0.0001, "loss": 0.0442, "step": 5590 }, { "epoch": 33.13609467455621, "grad_norm": 2.4434988498687744, "learning_rate": 0.0001, "loss": 0.045, "step": 5600 }, { "epoch": 33.19526627218935, "grad_norm": 2.418679714202881, "learning_rate": 0.0001, "loss": 0.0431, "step": 5610 }, { "epoch": 33.25443786982248, "grad_norm": 2.664915084838867, "learning_rate": 0.0001, "loss": 0.0473, "step": 5620 }, { "epoch": 33.31360946745562, "grad_norm": 2.341679811477661, "learning_rate": 0.0001, "loss": 0.0446, "step": 5630 }, { "epoch": 33.37278106508876, "grad_norm": 2.9755518436431885, "learning_rate": 0.0001, "loss": 0.0429, "step": 5640 }, { "epoch": 33.43195266272189, "grad_norm": 2.6128480434417725, "learning_rate": 0.0001, "loss": 0.0446, "step": 5650 }, { "epoch": 33.49112426035503, "grad_norm": 2.495741128921509, "learning_rate": 0.0001, "loss": 0.0453, "step": 5660 }, { "epoch": 33.55029585798817, "grad_norm": 2.383582830429077, "learning_rate": 0.0001, "loss": 0.0426, "step": 5670 }, { "epoch": 33.6094674556213, "grad_norm": 2.6396753787994385, "learning_rate": 0.0001, "loss": 0.0411, "step": 5680 }, { "epoch": 33.66863905325444, "grad_norm": 2.630791664123535, "learning_rate": 0.0001, "loss": 0.0418, "step": 5690 }, { "epoch": 33.72781065088758, "grad_norm": 2.8465545177459717, "learning_rate": 0.0001, "loss": 0.043, "step": 5700 }, { "epoch": 33.78698224852071, "grad_norm": 2.8943662643432617, "learning_rate": 0.0001, "loss": 0.0438, "step": 5710 }, { "epoch": 33.84615384615385, "grad_norm": 2.929270029067993, "learning_rate": 0.0001, "loss": 0.0439, "step": 5720 }, { "epoch": 33.905325443786985, "grad_norm": 2.992579460144043, "learning_rate": 0.0001, "loss": 0.0431, "step": 5730 }, { "epoch": 33.96449704142012, "grad_norm": 3.3918285369873047, "learning_rate": 0.0001, "loss": 0.0442, "step": 5740 }, { "epoch": 34.023668639053255, "grad_norm": 3.2850544452667236, "learning_rate": 0.0001, "loss": 0.0426, "step": 5750 }, { "epoch": 34.082840236686394, "grad_norm": 2.928415298461914, "learning_rate": 0.0001, "loss": 0.0441, "step": 5760 }, { "epoch": 34.142011834319526, "grad_norm": 2.258923292160034, "learning_rate": 0.0001, "loss": 0.0425, "step": 5770 }, { "epoch": 34.201183431952664, "grad_norm": 4.597163677215576, "learning_rate": 0.0001, "loss": 0.0415, "step": 5780 }, { "epoch": 34.260355029585796, "grad_norm": 2.29013991355896, "learning_rate": 0.0001, "loss": 0.0434, "step": 5790 }, { "epoch": 34.319526627218934, "grad_norm": 2.7559680938720703, "learning_rate": 0.0001, "loss": 0.0456, "step": 5800 }, { "epoch": 34.37869822485207, "grad_norm": 3.2377777099609375, "learning_rate": 0.0001, "loss": 0.0444, "step": 5810 }, { "epoch": 34.437869822485204, "grad_norm": 2.5860226154327393, "learning_rate": 0.0001, "loss": 0.0428, "step": 5820 }, { "epoch": 34.49704142011834, "grad_norm": 2.291794538497925, "learning_rate": 0.0001, "loss": 0.0429, "step": 5830 }, { "epoch": 34.55621301775148, "grad_norm": 2.7579596042633057, "learning_rate": 0.0001, "loss": 0.0425, "step": 5840 }, { "epoch": 34.61538461538461, "grad_norm": 2.5573623180389404, "learning_rate": 0.0001, "loss": 0.0422, "step": 5850 }, { "epoch": 34.67455621301775, "grad_norm": 2.342090368270874, "learning_rate": 0.0001, "loss": 0.0417, "step": 5860 }, { "epoch": 34.73372781065089, "grad_norm": 2.5002968311309814, "learning_rate": 0.0001, "loss": 0.0415, "step": 5870 }, { "epoch": 34.79289940828402, "grad_norm": 2.326442241668701, "learning_rate": 0.0001, "loss": 0.0405, "step": 5880 }, { "epoch": 34.85207100591716, "grad_norm": 2.730483055114746, "learning_rate": 0.0001, "loss": 0.042, "step": 5890 }, { "epoch": 34.9112426035503, "grad_norm": 3.062039613723755, "learning_rate": 0.0001, "loss": 0.0444, "step": 5900 }, { "epoch": 34.97041420118343, "grad_norm": 2.323446273803711, "learning_rate": 0.0001, "loss": 0.0421, "step": 5910 }, { "epoch": 35.02958579881657, "grad_norm": 2.583503484725952, "learning_rate": 0.0001, "loss": 0.042, "step": 5920 }, { "epoch": 35.0887573964497, "grad_norm": 2.21854829788208, "learning_rate": 0.0001, "loss": 0.0439, "step": 5930 }, { "epoch": 35.14792899408284, "grad_norm": 2.742907762527466, "learning_rate": 0.0001, "loss": 0.0428, "step": 5940 }, { "epoch": 35.20710059171598, "grad_norm": 2.0343527793884277, "learning_rate": 0.0001, "loss": 0.04, "step": 5950 }, { "epoch": 35.26627218934911, "grad_norm": 1.917546272277832, "learning_rate": 0.0001, "loss": 0.0416, "step": 5960 }, { "epoch": 35.32544378698225, "grad_norm": 2.4178991317749023, "learning_rate": 0.0001, "loss": 0.0397, "step": 5970 }, { "epoch": 35.38461538461539, "grad_norm": 2.566964626312256, "learning_rate": 0.0001, "loss": 0.0404, "step": 5980 }, { "epoch": 35.44378698224852, "grad_norm": 3.17779541015625, "learning_rate": 0.0001, "loss": 0.0419, "step": 5990 }, { "epoch": 35.50295857988166, "grad_norm": 2.4985430240631104, "learning_rate": 0.0001, "loss": 0.0419, "step": 6000 }, { "epoch": 35.562130177514796, "grad_norm": 2.741398572921753, "learning_rate": 0.0001, "loss": 0.0425, "step": 6010 }, { "epoch": 35.62130177514793, "grad_norm": 2.0012359619140625, "learning_rate": 0.0001, "loss": 0.0419, "step": 6020 }, { "epoch": 35.680473372781066, "grad_norm": 2.6605968475341797, "learning_rate": 0.0001, "loss": 0.0406, "step": 6030 }, { "epoch": 35.739644970414204, "grad_norm": 2.525573492050171, "learning_rate": 0.0001, "loss": 0.0435, "step": 6040 }, { "epoch": 35.798816568047336, "grad_norm": 2.467519760131836, "learning_rate": 0.0001, "loss": 0.0415, "step": 6050 }, { "epoch": 35.857988165680474, "grad_norm": 2.615041494369507, "learning_rate": 0.0001, "loss": 0.0437, "step": 6060 }, { "epoch": 35.917159763313606, "grad_norm": 2.470360040664673, "learning_rate": 0.0001, "loss": 0.0423, "step": 6070 }, { "epoch": 35.976331360946745, "grad_norm": 3.0041985511779785, "learning_rate": 0.0001, "loss": 0.0406, "step": 6080 }, { "epoch": 36.03550295857988, "grad_norm": 2.110048532485962, "learning_rate": 0.0001, "loss": 0.0408, "step": 6090 }, { "epoch": 36.094674556213015, "grad_norm": 2.157137155532837, "learning_rate": 0.0001, "loss": 0.0417, "step": 6100 }, { "epoch": 36.15384615384615, "grad_norm": 2.3281562328338623, "learning_rate": 0.0001, "loss": 0.0403, "step": 6110 }, { "epoch": 36.21301775147929, "grad_norm": 2.0512025356292725, "learning_rate": 0.0001, "loss": 0.0406, "step": 6120 }, { "epoch": 36.27218934911242, "grad_norm": 2.278346300125122, "learning_rate": 0.0001, "loss": 0.0408, "step": 6130 }, { "epoch": 36.33136094674556, "grad_norm": 2.5638973712921143, "learning_rate": 0.0001, "loss": 0.0394, "step": 6140 }, { "epoch": 36.3905325443787, "grad_norm": 2.694349527359009, "learning_rate": 0.0001, "loss": 0.0407, "step": 6150 }, { "epoch": 36.44970414201183, "grad_norm": 3.001039981842041, "learning_rate": 0.0001, "loss": 0.0398, "step": 6160 }, { "epoch": 36.50887573964497, "grad_norm": 2.834965229034424, "learning_rate": 0.0001, "loss": 0.041, "step": 6170 }, { "epoch": 36.56804733727811, "grad_norm": 2.2614688873291016, "learning_rate": 0.0001, "loss": 0.0403, "step": 6180 }, { "epoch": 36.62721893491124, "grad_norm": 2.6870603561401367, "learning_rate": 0.0001, "loss": 0.0398, "step": 6190 }, { "epoch": 36.68639053254438, "grad_norm": 2.4931278228759766, "learning_rate": 0.0001, "loss": 0.0404, "step": 6200 }, { "epoch": 36.74556213017752, "grad_norm": 2.340984582901001, "learning_rate": 0.0001, "loss": 0.039, "step": 6210 }, { "epoch": 36.80473372781065, "grad_norm": 2.5397956371307373, "learning_rate": 0.0001, "loss": 0.039, "step": 6220 }, { "epoch": 36.86390532544379, "grad_norm": 2.9098761081695557, "learning_rate": 0.0001, "loss": 0.0413, "step": 6230 }, { "epoch": 36.92307692307692, "grad_norm": 2.5594727993011475, "learning_rate": 0.0001, "loss": 0.0407, "step": 6240 }, { "epoch": 36.98224852071006, "grad_norm": 2.944187641143799, "learning_rate": 0.0001, "loss": 0.0403, "step": 6250 }, { "epoch": 37.0414201183432, "grad_norm": 2.7487967014312744, "learning_rate": 0.0001, "loss": 0.0426, "step": 6260 }, { "epoch": 37.10059171597633, "grad_norm": 2.329335927963257, "learning_rate": 0.0001, "loss": 0.0398, "step": 6270 }, { "epoch": 37.15976331360947, "grad_norm": 2.451202392578125, "learning_rate": 0.0001, "loss": 0.0407, "step": 6280 }, { "epoch": 37.218934911242606, "grad_norm": 2.4972989559173584, "learning_rate": 0.0001, "loss": 0.0401, "step": 6290 }, { "epoch": 37.27810650887574, "grad_norm": 2.6726067066192627, "learning_rate": 0.0001, "loss": 0.0406, "step": 6300 }, { "epoch": 37.337278106508876, "grad_norm": 2.4876670837402344, "learning_rate": 0.0001, "loss": 0.0395, "step": 6310 }, { "epoch": 37.396449704142015, "grad_norm": 2.243919610977173, "learning_rate": 0.0001, "loss": 0.0404, "step": 6320 }, { "epoch": 37.455621301775146, "grad_norm": 3.023308753967285, "learning_rate": 0.0001, "loss": 0.042, "step": 6330 }, { "epoch": 37.514792899408285, "grad_norm": 2.4316606521606445, "learning_rate": 0.0001, "loss": 0.0424, "step": 6340 }, { "epoch": 37.57396449704142, "grad_norm": 2.4590184688568115, "learning_rate": 0.0001, "loss": 0.0428, "step": 6350 }, { "epoch": 37.633136094674555, "grad_norm": 2.658515691757202, "learning_rate": 0.0001, "loss": 0.0428, "step": 6360 }, { "epoch": 37.69230769230769, "grad_norm": 2.5840909481048584, "learning_rate": 0.0001, "loss": 0.0417, "step": 6370 }, { "epoch": 37.751479289940825, "grad_norm": 2.451486825942993, "learning_rate": 0.0001, "loss": 0.0407, "step": 6380 }, { "epoch": 37.81065088757396, "grad_norm": 3.480821371078491, "learning_rate": 0.0001, "loss": 0.0407, "step": 6390 }, { "epoch": 37.8698224852071, "grad_norm": 2.7321033477783203, "learning_rate": 0.0001, "loss": 0.0393, "step": 6400 }, { "epoch": 37.928994082840234, "grad_norm": 2.7242612838745117, "learning_rate": 0.0001, "loss": 0.0395, "step": 6410 }, { "epoch": 37.98816568047337, "grad_norm": 2.597043037414551, "learning_rate": 0.0001, "loss": 0.0403, "step": 6420 }, { "epoch": 38.04733727810651, "grad_norm": 2.4965641498565674, "learning_rate": 0.0001, "loss": 0.0382, "step": 6430 }, { "epoch": 38.10650887573964, "grad_norm": 2.785860061645508, "learning_rate": 0.0001, "loss": 0.0383, "step": 6440 }, { "epoch": 38.16568047337278, "grad_norm": 2.8015520572662354, "learning_rate": 0.0001, "loss": 0.0376, "step": 6450 }, { "epoch": 38.22485207100592, "grad_norm": 2.345874547958374, "learning_rate": 0.0001, "loss": 0.0385, "step": 6460 }, { "epoch": 38.28402366863905, "grad_norm": 2.3189609050750732, "learning_rate": 0.0001, "loss": 0.0394, "step": 6470 }, { "epoch": 38.34319526627219, "grad_norm": 2.5552563667297363, "learning_rate": 0.0001, "loss": 0.0386, "step": 6480 }, { "epoch": 38.40236686390533, "grad_norm": 2.3291563987731934, "learning_rate": 0.0001, "loss": 0.0381, "step": 6490 }, { "epoch": 38.46153846153846, "grad_norm": 2.3333356380462646, "learning_rate": 0.0001, "loss": 0.0381, "step": 6500 }, { "epoch": 38.5207100591716, "grad_norm": 2.2243802547454834, "learning_rate": 0.0001, "loss": 0.039, "step": 6510 }, { "epoch": 38.57988165680474, "grad_norm": 2.8184762001037598, "learning_rate": 0.0001, "loss": 0.0397, "step": 6520 }, { "epoch": 38.63905325443787, "grad_norm": 3.096283435821533, "learning_rate": 0.0001, "loss": 0.0415, "step": 6530 }, { "epoch": 38.69822485207101, "grad_norm": 2.144503593444824, "learning_rate": 0.0001, "loss": 0.042, "step": 6540 }, { "epoch": 38.75739644970414, "grad_norm": 2.2234437465667725, "learning_rate": 0.0001, "loss": 0.0411, "step": 6550 }, { "epoch": 38.81656804733728, "grad_norm": 2.2934796810150146, "learning_rate": 0.0001, "loss": 0.04, "step": 6560 }, { "epoch": 38.875739644970416, "grad_norm": 1.9745124578475952, "learning_rate": 0.0001, "loss": 0.0415, "step": 6570 }, { "epoch": 38.93491124260355, "grad_norm": 1.850346565246582, "learning_rate": 0.0001, "loss": 0.0411, "step": 6580 }, { "epoch": 38.994082840236686, "grad_norm": 2.921921730041504, "learning_rate": 0.0001, "loss": 0.0413, "step": 6590 }, { "epoch": 39.053254437869825, "grad_norm": 2.343167781829834, "learning_rate": 0.0001, "loss": 0.0425, "step": 6600 }, { "epoch": 39.112426035502956, "grad_norm": 2.2628204822540283, "learning_rate": 0.0001, "loss": 0.0415, "step": 6610 }, { "epoch": 39.171597633136095, "grad_norm": 2.038224697113037, "learning_rate": 0.0001, "loss": 0.0415, "step": 6620 }, { "epoch": 39.23076923076923, "grad_norm": 2.7431020736694336, "learning_rate": 0.0001, "loss": 0.0394, "step": 6630 }, { "epoch": 39.289940828402365, "grad_norm": 2.323495626449585, "learning_rate": 0.0001, "loss": 0.0385, "step": 6640 }, { "epoch": 39.349112426035504, "grad_norm": 2.543616771697998, "learning_rate": 0.0001, "loss": 0.0392, "step": 6650 }, { "epoch": 39.40828402366864, "grad_norm": 3.057415246963501, "learning_rate": 0.0001, "loss": 0.0402, "step": 6660 }, { "epoch": 39.467455621301774, "grad_norm": 2.823976755142212, "learning_rate": 0.0001, "loss": 0.0384, "step": 6670 }, { "epoch": 39.52662721893491, "grad_norm": 2.3139243125915527, "learning_rate": 0.0001, "loss": 0.0389, "step": 6680 }, { "epoch": 39.585798816568044, "grad_norm": 2.14479923248291, "learning_rate": 0.0001, "loss": 0.0387, "step": 6690 }, { "epoch": 39.64497041420118, "grad_norm": 2.440619468688965, "learning_rate": 0.0001, "loss": 0.0387, "step": 6700 }, { "epoch": 39.70414201183432, "grad_norm": 2.863781213760376, "learning_rate": 0.0001, "loss": 0.0383, "step": 6710 }, { "epoch": 39.76331360946745, "grad_norm": 2.52275013923645, "learning_rate": 0.0001, "loss": 0.0392, "step": 6720 }, { "epoch": 39.82248520710059, "grad_norm": 2.6787893772125244, "learning_rate": 0.0001, "loss": 0.0418, "step": 6730 }, { "epoch": 39.88165680473373, "grad_norm": 2.145162343978882, "learning_rate": 0.0001, "loss": 0.0408, "step": 6740 }, { "epoch": 39.94082840236686, "grad_norm": 2.698439836502075, "learning_rate": 0.0001, "loss": 0.0395, "step": 6750 }, { "epoch": 40.0, "grad_norm": 2.7360434532165527, "learning_rate": 0.0001, "loss": 0.0416, "step": 6760 }, { "epoch": 40.05917159763314, "grad_norm": 1.9875502586364746, "learning_rate": 0.0001, "loss": 0.0379, "step": 6770 }, { "epoch": 40.11834319526627, "grad_norm": 2.3649742603302, "learning_rate": 0.0001, "loss": 0.0394, "step": 6780 }, { "epoch": 40.17751479289941, "grad_norm": 2.4011642932891846, "learning_rate": 0.0001, "loss": 0.0371, "step": 6790 }, { "epoch": 40.23668639053255, "grad_norm": 1.8787661790847778, "learning_rate": 0.0001, "loss": 0.037, "step": 6800 }, { "epoch": 40.29585798816568, "grad_norm": 1.7769991159439087, "learning_rate": 0.0001, "loss": 0.0384, "step": 6810 }, { "epoch": 40.35502958579882, "grad_norm": 2.3936350345611572, "learning_rate": 0.0001, "loss": 0.0393, "step": 6820 }, { "epoch": 40.414201183431956, "grad_norm": 1.8941088914871216, "learning_rate": 0.0001, "loss": 0.038, "step": 6830 }, { "epoch": 40.47337278106509, "grad_norm": 2.3759191036224365, "learning_rate": 0.0001, "loss": 0.04, "step": 6840 }, { "epoch": 40.532544378698226, "grad_norm": 2.524066209793091, "learning_rate": 0.0001, "loss": 0.0381, "step": 6850 }, { "epoch": 40.59171597633136, "grad_norm": 2.533029079437256, "learning_rate": 0.0001, "loss": 0.0391, "step": 6860 }, { "epoch": 40.650887573964496, "grad_norm": 2.9354441165924072, "learning_rate": 0.0001, "loss": 0.0388, "step": 6870 }, { "epoch": 40.710059171597635, "grad_norm": 2.2446532249450684, "learning_rate": 0.0001, "loss": 0.0399, "step": 6880 }, { "epoch": 40.76923076923077, "grad_norm": 2.5661752223968506, "learning_rate": 0.0001, "loss": 0.0407, "step": 6890 }, { "epoch": 40.828402366863905, "grad_norm": 2.0824480056762695, "learning_rate": 0.0001, "loss": 0.0383, "step": 6900 }, { "epoch": 40.887573964497044, "grad_norm": 2.1552536487579346, "learning_rate": 0.0001, "loss": 0.0391, "step": 6910 }, { "epoch": 40.946745562130175, "grad_norm": 2.556182384490967, "learning_rate": 0.0001, "loss": 0.0397, "step": 6920 }, { "epoch": 41.005917159763314, "grad_norm": 2.405585765838623, "learning_rate": 0.0001, "loss": 0.0389, "step": 6930 }, { "epoch": 41.06508875739645, "grad_norm": 2.83366322517395, "learning_rate": 0.0001, "loss": 0.0372, "step": 6940 }, { "epoch": 41.124260355029584, "grad_norm": 2.6361775398254395, "learning_rate": 0.0001, "loss": 0.038, "step": 6950 }, { "epoch": 41.18343195266272, "grad_norm": 2.9444990158081055, "learning_rate": 0.0001, "loss": 0.0371, "step": 6960 }, { "epoch": 41.24260355029586, "grad_norm": 2.317779541015625, "learning_rate": 0.0001, "loss": 0.0395, "step": 6970 }, { "epoch": 41.30177514792899, "grad_norm": 2.462730884552002, "learning_rate": 0.0001, "loss": 0.0366, "step": 6980 }, { "epoch": 41.36094674556213, "grad_norm": 2.24108624458313, "learning_rate": 0.0001, "loss": 0.0368, "step": 6990 }, { "epoch": 41.42011834319526, "grad_norm": 1.98497474193573, "learning_rate": 0.0001, "loss": 0.0376, "step": 7000 }, { "epoch": 41.4792899408284, "grad_norm": 1.9790728092193604, "learning_rate": 0.0001, "loss": 0.0373, "step": 7010 }, { "epoch": 41.53846153846154, "grad_norm": 2.2604503631591797, "learning_rate": 0.0001, "loss": 0.0376, "step": 7020 }, { "epoch": 41.59763313609467, "grad_norm": 2.372873544692993, "learning_rate": 0.0001, "loss": 0.037, "step": 7030 }, { "epoch": 41.65680473372781, "grad_norm": 2.627620220184326, "learning_rate": 0.0001, "loss": 0.0398, "step": 7040 }, { "epoch": 41.71597633136095, "grad_norm": 1.7841624021530151, "learning_rate": 0.0001, "loss": 0.0364, "step": 7050 }, { "epoch": 41.77514792899408, "grad_norm": 2.072183847427368, "learning_rate": 0.0001, "loss": 0.0367, "step": 7060 }, { "epoch": 41.83431952662722, "grad_norm": 2.1780316829681396, "learning_rate": 0.0001, "loss": 0.0381, "step": 7070 }, { "epoch": 41.89349112426036, "grad_norm": 2.1168179512023926, "learning_rate": 0.0001, "loss": 0.0384, "step": 7080 }, { "epoch": 41.95266272189349, "grad_norm": 2.244992971420288, "learning_rate": 0.0001, "loss": 0.038, "step": 7090 }, { "epoch": 42.01183431952663, "grad_norm": 2.3387107849121094, "learning_rate": 0.0001, "loss": 0.039, "step": 7100 }, { "epoch": 42.071005917159766, "grad_norm": 2.9245083332061768, "learning_rate": 0.0001, "loss": 0.0371, "step": 7110 }, { "epoch": 42.1301775147929, "grad_norm": 2.449031352996826, "learning_rate": 0.0001, "loss": 0.0392, "step": 7120 }, { "epoch": 42.18934911242604, "grad_norm": 2.3910062313079834, "learning_rate": 0.0001, "loss": 0.0363, "step": 7130 }, { "epoch": 42.248520710059175, "grad_norm": 2.2225966453552246, "learning_rate": 0.0001, "loss": 0.0366, "step": 7140 }, { "epoch": 42.30769230769231, "grad_norm": 2.571887731552124, "learning_rate": 0.0001, "loss": 0.0364, "step": 7150 }, { "epoch": 42.366863905325445, "grad_norm": 2.100313663482666, "learning_rate": 0.0001, "loss": 0.037, "step": 7160 }, { "epoch": 42.42603550295858, "grad_norm": 2.255918025970459, "learning_rate": 0.0001, "loss": 0.0373, "step": 7170 }, { "epoch": 42.485207100591715, "grad_norm": 2.474910020828247, "learning_rate": 0.0001, "loss": 0.0368, "step": 7180 }, { "epoch": 42.544378698224854, "grad_norm": 2.125667095184326, "learning_rate": 0.0001, "loss": 0.038, "step": 7190 }, { "epoch": 42.603550295857985, "grad_norm": 2.146454334259033, "learning_rate": 0.0001, "loss": 0.0369, "step": 7200 }, { "epoch": 42.662721893491124, "grad_norm": 2.3661751747131348, "learning_rate": 0.0001, "loss": 0.0377, "step": 7210 }, { "epoch": 42.72189349112426, "grad_norm": 2.3355252742767334, "learning_rate": 0.0001, "loss": 0.0383, "step": 7220 }, { "epoch": 42.781065088757394, "grad_norm": 2.025534152984619, "learning_rate": 0.0001, "loss": 0.0392, "step": 7230 }, { "epoch": 42.84023668639053, "grad_norm": 2.061997890472412, "learning_rate": 0.0001, "loss": 0.0378, "step": 7240 }, { "epoch": 42.89940828402367, "grad_norm": 2.460223913192749, "learning_rate": 0.0001, "loss": 0.0393, "step": 7250 }, { "epoch": 42.9585798816568, "grad_norm": 2.4932968616485596, "learning_rate": 0.0001, "loss": 0.0398, "step": 7260 }, { "epoch": 43.01775147928994, "grad_norm": 2.393789529800415, "learning_rate": 0.0001, "loss": 0.0369, "step": 7270 }, { "epoch": 43.07692307692308, "grad_norm": 1.6913036108016968, "learning_rate": 0.0001, "loss": 0.0375, "step": 7280 }, { "epoch": 43.13609467455621, "grad_norm": 1.9966392517089844, "learning_rate": 0.0001, "loss": 0.0362, "step": 7290 }, { "epoch": 43.19526627218935, "grad_norm": 2.2647085189819336, "learning_rate": 0.0001, "loss": 0.0352, "step": 7300 }, { "epoch": 43.25443786982248, "grad_norm": 2.1094040870666504, "learning_rate": 0.0001, "loss": 0.0372, "step": 7310 }, { "epoch": 43.31360946745562, "grad_norm": 2.3744027614593506, "learning_rate": 0.0001, "loss": 0.0374, "step": 7320 }, { "epoch": 43.37278106508876, "grad_norm": 1.905428171157837, "learning_rate": 0.0001, "loss": 0.0355, "step": 7330 }, { "epoch": 43.43195266272189, "grad_norm": 2.3744521141052246, "learning_rate": 0.0001, "loss": 0.0358, "step": 7340 }, { "epoch": 43.49112426035503, "grad_norm": 2.3459866046905518, "learning_rate": 0.0001, "loss": 0.0361, "step": 7350 }, { "epoch": 43.55029585798817, "grad_norm": 2.2260873317718506, "learning_rate": 0.0001, "loss": 0.0364, "step": 7360 }, { "epoch": 43.6094674556213, "grad_norm": 2.2043254375457764, "learning_rate": 0.0001, "loss": 0.0377, "step": 7370 }, { "epoch": 43.66863905325444, "grad_norm": 2.517984628677368, "learning_rate": 0.0001, "loss": 0.0373, "step": 7380 }, { "epoch": 43.72781065088758, "grad_norm": 2.196340799331665, "learning_rate": 0.0001, "loss": 0.0369, "step": 7390 }, { "epoch": 43.78698224852071, "grad_norm": 2.33845591545105, "learning_rate": 0.0001, "loss": 0.0374, "step": 7400 }, { "epoch": 43.84615384615385, "grad_norm": 2.240640640258789, "learning_rate": 0.0001, "loss": 0.0349, "step": 7410 }, { "epoch": 43.905325443786985, "grad_norm": 1.7366080284118652, "learning_rate": 0.0001, "loss": 0.0358, "step": 7420 }, { "epoch": 43.96449704142012, "grad_norm": 2.531285524368286, "learning_rate": 0.0001, "loss": 0.0371, "step": 7430 }, { "epoch": 44.023668639053255, "grad_norm": 2.8422353267669678, "learning_rate": 0.0001, "loss": 0.0368, "step": 7440 }, { "epoch": 44.082840236686394, "grad_norm": 2.0423431396484375, "learning_rate": 0.0001, "loss": 0.0355, "step": 7450 }, { "epoch": 44.142011834319526, "grad_norm": 2.000582456588745, "learning_rate": 0.0001, "loss": 0.0353, "step": 7460 }, { "epoch": 44.201183431952664, "grad_norm": 2.1669435501098633, "learning_rate": 0.0001, "loss": 0.0349, "step": 7470 }, { "epoch": 44.260355029585796, "grad_norm": 2.0764126777648926, "learning_rate": 0.0001, "loss": 0.0358, "step": 7480 }, { "epoch": 44.319526627218934, "grad_norm": 2.3324406147003174, "learning_rate": 0.0001, "loss": 0.0375, "step": 7490 }, { "epoch": 44.37869822485207, "grad_norm": 2.483893394470215, "learning_rate": 0.0001, "loss": 0.0356, "step": 7500 }, { "epoch": 44.437869822485204, "grad_norm": 2.401151418685913, "learning_rate": 0.0001, "loss": 0.0375, "step": 7510 }, { "epoch": 44.49704142011834, "grad_norm": 2.0265400409698486, "learning_rate": 0.0001, "loss": 0.037, "step": 7520 }, { "epoch": 44.55621301775148, "grad_norm": 2.3261024951934814, "learning_rate": 0.0001, "loss": 0.0376, "step": 7530 }, { "epoch": 44.61538461538461, "grad_norm": 1.8654568195343018, "learning_rate": 0.0001, "loss": 0.0366, "step": 7540 }, { "epoch": 44.67455621301775, "grad_norm": 2.409536600112915, "learning_rate": 0.0001, "loss": 0.0389, "step": 7550 }, { "epoch": 44.73372781065089, "grad_norm": 2.2529664039611816, "learning_rate": 0.0001, "loss": 0.0386, "step": 7560 }, { "epoch": 44.79289940828402, "grad_norm": 2.470036506652832, "learning_rate": 0.0001, "loss": 0.0394, "step": 7570 }, { "epoch": 44.85207100591716, "grad_norm": 2.8320062160491943, "learning_rate": 0.0001, "loss": 0.0417, "step": 7580 }, { "epoch": 44.9112426035503, "grad_norm": 2.4825825691223145, "learning_rate": 0.0001, "loss": 0.0392, "step": 7590 }, { "epoch": 44.97041420118343, "grad_norm": 2.9774324893951416, "learning_rate": 0.0001, "loss": 0.0395, "step": 7600 }, { "epoch": 45.02958579881657, "grad_norm": 2.0142269134521484, "learning_rate": 0.0001, "loss": 0.0364, "step": 7610 }, { "epoch": 45.0887573964497, "grad_norm": 1.9716883897781372, "learning_rate": 0.0001, "loss": 0.0357, "step": 7620 }, { "epoch": 45.14792899408284, "grad_norm": 2.119669198989868, "learning_rate": 0.0001, "loss": 0.0349, "step": 7630 }, { "epoch": 45.20710059171598, "grad_norm": 1.601367473602295, "learning_rate": 0.0001, "loss": 0.036, "step": 7640 }, { "epoch": 45.26627218934911, "grad_norm": 2.3662831783294678, "learning_rate": 0.0001, "loss": 0.0393, "step": 7650 }, { "epoch": 45.32544378698225, "grad_norm": 2.0687904357910156, "learning_rate": 0.0001, "loss": 0.0365, "step": 7660 }, { "epoch": 45.38461538461539, "grad_norm": 2.5651841163635254, "learning_rate": 0.0001, "loss": 0.0378, "step": 7670 }, { "epoch": 45.44378698224852, "grad_norm": 1.9394583702087402, "learning_rate": 0.0001, "loss": 0.0377, "step": 7680 }, { "epoch": 45.50295857988166, "grad_norm": 2.532410144805908, "learning_rate": 0.0001, "loss": 0.0368, "step": 7690 }, { "epoch": 45.562130177514796, "grad_norm": 1.6165274381637573, "learning_rate": 0.0001, "loss": 0.0385, "step": 7700 }, { "epoch": 45.62130177514793, "grad_norm": 1.6849991083145142, "learning_rate": 0.0001, "loss": 0.0379, "step": 7710 }, { "epoch": 45.680473372781066, "grad_norm": 2.335282325744629, "learning_rate": 0.0001, "loss": 0.0365, "step": 7720 }, { "epoch": 45.739644970414204, "grad_norm": 1.9209465980529785, "learning_rate": 0.0001, "loss": 0.0358, "step": 7730 }, { "epoch": 45.798816568047336, "grad_norm": 2.09179425239563, "learning_rate": 0.0001, "loss": 0.0372, "step": 7740 }, { "epoch": 45.857988165680474, "grad_norm": 2.4752023220062256, "learning_rate": 0.0001, "loss": 0.0374, "step": 7750 }, { "epoch": 45.917159763313606, "grad_norm": 2.222156047821045, "learning_rate": 0.0001, "loss": 0.0377, "step": 7760 }, { "epoch": 45.976331360946745, "grad_norm": 2.185572624206543, "learning_rate": 0.0001, "loss": 0.0372, "step": 7770 }, { "epoch": 46.03550295857988, "grad_norm": 2.3295931816101074, "learning_rate": 0.0001, "loss": 0.0377, "step": 7780 }, { "epoch": 46.094674556213015, "grad_norm": 2.135991096496582, "learning_rate": 0.0001, "loss": 0.0374, "step": 7790 }, { "epoch": 46.15384615384615, "grad_norm": 2.3086888790130615, "learning_rate": 0.0001, "loss": 0.0358, "step": 7800 }, { "epoch": 46.21301775147929, "grad_norm": 2.5818839073181152, "learning_rate": 0.0001, "loss": 0.0366, "step": 7810 }, { "epoch": 46.27218934911242, "grad_norm": 2.3107752799987793, "learning_rate": 0.0001, "loss": 0.0354, "step": 7820 }, { "epoch": 46.33136094674556, "grad_norm": 2.1417198181152344, "learning_rate": 0.0001, "loss": 0.036, "step": 7830 }, { "epoch": 46.3905325443787, "grad_norm": 2.6400084495544434, "learning_rate": 0.0001, "loss": 0.0358, "step": 7840 }, { "epoch": 46.44970414201183, "grad_norm": 2.5576725006103516, "learning_rate": 0.0001, "loss": 0.0348, "step": 7850 }, { "epoch": 46.50887573964497, "grad_norm": 1.9282197952270508, "learning_rate": 0.0001, "loss": 0.0364, "step": 7860 }, { "epoch": 46.56804733727811, "grad_norm": 3.2247376441955566, "learning_rate": 0.0001, "loss": 0.0373, "step": 7870 }, { "epoch": 46.62721893491124, "grad_norm": 1.7528022527694702, "learning_rate": 0.0001, "loss": 0.0352, "step": 7880 }, { "epoch": 46.68639053254438, "grad_norm": 2.1539173126220703, "learning_rate": 0.0001, "loss": 0.0356, "step": 7890 }, { "epoch": 46.74556213017752, "grad_norm": 2.4724009037017822, "learning_rate": 0.0001, "loss": 0.0346, "step": 7900 }, { "epoch": 46.80473372781065, "grad_norm": 1.8680654764175415, "learning_rate": 0.0001, "loss": 0.0358, "step": 7910 }, { "epoch": 46.86390532544379, "grad_norm": 1.7780545949935913, "learning_rate": 0.0001, "loss": 0.036, "step": 7920 }, { "epoch": 46.92307692307692, "grad_norm": 2.3983092308044434, "learning_rate": 0.0001, "loss": 0.0356, "step": 7930 }, { "epoch": 46.98224852071006, "grad_norm": 2.4637339115142822, "learning_rate": 0.0001, "loss": 0.034, "step": 7940 }, { "epoch": 47.0414201183432, "grad_norm": 2.1025006771087646, "learning_rate": 0.0001, "loss": 0.0345, "step": 7950 }, { "epoch": 47.10059171597633, "grad_norm": 2.3651907444000244, "learning_rate": 0.0001, "loss": 0.0346, "step": 7960 }, { "epoch": 47.15976331360947, "grad_norm": 2.4315719604492188, "learning_rate": 0.0001, "loss": 0.0348, "step": 7970 }, { "epoch": 47.218934911242606, "grad_norm": 2.1637113094329834, "learning_rate": 0.0001, "loss": 0.0329, "step": 7980 }, { "epoch": 47.27810650887574, "grad_norm": 2.0689680576324463, "learning_rate": 0.0001, "loss": 0.0355, "step": 7990 }, { "epoch": 47.337278106508876, "grad_norm": 2.0118582248687744, "learning_rate": 0.0001, "loss": 0.0382, "step": 8000 }, { "epoch": 47.396449704142015, "grad_norm": 1.8387001752853394, "learning_rate": 0.0001, "loss": 0.0362, "step": 8010 }, { "epoch": 47.455621301775146, "grad_norm": 1.924631953239441, "learning_rate": 0.0001, "loss": 0.0371, "step": 8020 }, { "epoch": 47.514792899408285, "grad_norm": 2.161264181137085, "learning_rate": 0.0001, "loss": 0.0369, "step": 8030 }, { "epoch": 47.57396449704142, "grad_norm": 1.9545276165008545, "learning_rate": 0.0001, "loss": 0.0371, "step": 8040 }, { "epoch": 47.633136094674555, "grad_norm": 3.087484359741211, "learning_rate": 0.0001, "loss": 0.038, "step": 8050 }, { "epoch": 47.69230769230769, "grad_norm": 1.9491214752197266, "learning_rate": 0.0001, "loss": 0.0394, "step": 8060 }, { "epoch": 47.751479289940825, "grad_norm": 2.1664047241210938, "learning_rate": 0.0001, "loss": 0.0379, "step": 8070 }, { "epoch": 47.81065088757396, "grad_norm": 1.704701542854309, "learning_rate": 0.0001, "loss": 0.0358, "step": 8080 }, { "epoch": 47.8698224852071, "grad_norm": 2.185610055923462, "learning_rate": 0.0001, "loss": 0.0379, "step": 8090 }, { "epoch": 47.928994082840234, "grad_norm": 2.2190492153167725, "learning_rate": 0.0001, "loss": 0.0354, "step": 8100 }, { "epoch": 47.98816568047337, "grad_norm": 1.807153582572937, "learning_rate": 0.0001, "loss": 0.0358, "step": 8110 }, { "epoch": 48.04733727810651, "grad_norm": 2.288585901260376, "learning_rate": 0.0001, "loss": 0.0356, "step": 8120 }, { "epoch": 48.10650887573964, "grad_norm": 1.9735490083694458, "learning_rate": 0.0001, "loss": 0.0367, "step": 8130 }, { "epoch": 48.16568047337278, "grad_norm": 2.257453203201294, "learning_rate": 0.0001, "loss": 0.0359, "step": 8140 }, { "epoch": 48.22485207100592, "grad_norm": 1.8805813789367676, "learning_rate": 0.0001, "loss": 0.0361, "step": 8150 }, { "epoch": 48.28402366863905, "grad_norm": 1.8604196310043335, "learning_rate": 0.0001, "loss": 0.0353, "step": 8160 }, { "epoch": 48.34319526627219, "grad_norm": 1.9062142372131348, "learning_rate": 0.0001, "loss": 0.0379, "step": 8170 }, { "epoch": 48.40236686390533, "grad_norm": 2.375729560852051, "learning_rate": 0.0001, "loss": 0.0352, "step": 8180 }, { "epoch": 48.46153846153846, "grad_norm": 1.8140190839767456, "learning_rate": 0.0001, "loss": 0.0353, "step": 8190 }, { "epoch": 48.5207100591716, "grad_norm": 2.3980181217193604, "learning_rate": 0.0001, "loss": 0.0363, "step": 8200 }, { "epoch": 48.57988165680474, "grad_norm": 1.788995623588562, "learning_rate": 0.0001, "loss": 0.0381, "step": 8210 }, { "epoch": 48.63905325443787, "grad_norm": 2.039411783218384, "learning_rate": 0.0001, "loss": 0.0366, "step": 8220 }, { "epoch": 48.69822485207101, "grad_norm": 1.8218327760696411, "learning_rate": 0.0001, "loss": 0.0376, "step": 8230 }, { "epoch": 48.75739644970414, "grad_norm": 2.196699619293213, "learning_rate": 0.0001, "loss": 0.0359, "step": 8240 }, { "epoch": 48.81656804733728, "grad_norm": 2.199312925338745, "learning_rate": 0.0001, "loss": 0.0349, "step": 8250 }, { "epoch": 48.875739644970416, "grad_norm": 2.1916627883911133, "learning_rate": 0.0001, "loss": 0.0357, "step": 8260 }, { "epoch": 48.93491124260355, "grad_norm": 2.057276487350464, "learning_rate": 0.0001, "loss": 0.0353, "step": 8270 }, { "epoch": 48.994082840236686, "grad_norm": 2.2466375827789307, "learning_rate": 0.0001, "loss": 0.0348, "step": 8280 }, { "epoch": 49.053254437869825, "grad_norm": 2.3100528717041016, "learning_rate": 0.0001, "loss": 0.0352, "step": 8290 }, { "epoch": 49.112426035502956, "grad_norm": 2.336313486099243, "learning_rate": 0.0001, "loss": 0.0357, "step": 8300 }, { "epoch": 49.171597633136095, "grad_norm": 2.304906129837036, "learning_rate": 0.0001, "loss": 0.0332, "step": 8310 }, { "epoch": 49.23076923076923, "grad_norm": 2.021641254425049, "learning_rate": 0.0001, "loss": 0.0347, "step": 8320 }, { "epoch": 49.289940828402365, "grad_norm": 2.107557773590088, "learning_rate": 0.0001, "loss": 0.034, "step": 8330 }, { "epoch": 49.349112426035504, "grad_norm": 1.8544540405273438, "learning_rate": 0.0001, "loss": 0.0338, "step": 8340 }, { "epoch": 49.40828402366864, "grad_norm": 1.6679387092590332, "learning_rate": 0.0001, "loss": 0.036, "step": 8350 }, { "epoch": 49.467455621301774, "grad_norm": 1.9550069570541382, "learning_rate": 0.0001, "loss": 0.0366, "step": 8360 }, { "epoch": 49.52662721893491, "grad_norm": 1.724850058555603, "learning_rate": 0.0001, "loss": 0.0347, "step": 8370 }, { "epoch": 49.585798816568044, "grad_norm": 2.5529556274414062, "learning_rate": 0.0001, "loss": 0.0345, "step": 8380 }, { "epoch": 49.64497041420118, "grad_norm": 2.073111057281494, "learning_rate": 0.0001, "loss": 0.0357, "step": 8390 }, { "epoch": 49.70414201183432, "grad_norm": 2.484541416168213, "learning_rate": 0.0001, "loss": 0.034, "step": 8400 }, { "epoch": 49.76331360946745, "grad_norm": 2.450557231903076, "learning_rate": 0.0001, "loss": 0.0349, "step": 8410 }, { "epoch": 49.82248520710059, "grad_norm": 2.0338938236236572, "learning_rate": 0.0001, "loss": 0.0324, "step": 8420 }, { "epoch": 49.88165680473373, "grad_norm": 1.981462001800537, "learning_rate": 0.0001, "loss": 0.0346, "step": 8430 }, { "epoch": 49.94082840236686, "grad_norm": 2.7802672386169434, "learning_rate": 0.0001, "loss": 0.0346, "step": 8440 }, { "epoch": 50.0, "grad_norm": 2.114373207092285, "learning_rate": 0.0001, "loss": 0.0365, "step": 8450 }, { "epoch": 50.05917159763314, "grad_norm": 2.6551225185394287, "learning_rate": 0.0001, "loss": 0.0357, "step": 8460 }, { "epoch": 50.11834319526627, "grad_norm": 2.2854244709014893, "learning_rate": 0.0001, "loss": 0.0375, "step": 8470 }, { "epoch": 50.17751479289941, "grad_norm": 1.9731049537658691, "learning_rate": 0.0001, "loss": 0.0343, "step": 8480 }, { "epoch": 50.23668639053255, "grad_norm": 1.8862687349319458, "learning_rate": 0.0001, "loss": 0.0333, "step": 8490 }, { "epoch": 50.29585798816568, "grad_norm": 2.014892101287842, "learning_rate": 0.0001, "loss": 0.0352, "step": 8500 }, { "epoch": 50.35502958579882, "grad_norm": 2.9678854942321777, "learning_rate": 0.0001, "loss": 0.034, "step": 8510 }, { "epoch": 50.414201183431956, "grad_norm": 2.1976568698883057, "learning_rate": 0.0001, "loss": 0.0322, "step": 8520 }, { "epoch": 50.47337278106509, "grad_norm": 2.1121325492858887, "learning_rate": 0.0001, "loss": 0.0313, "step": 8530 }, { "epoch": 50.532544378698226, "grad_norm": 1.8821831941604614, "learning_rate": 0.0001, "loss": 0.0343, "step": 8540 }, { "epoch": 50.59171597633136, "grad_norm": 1.6109462976455688, "learning_rate": 0.0001, "loss": 0.0334, "step": 8550 }, { "epoch": 50.650887573964496, "grad_norm": 2.0709567070007324, "learning_rate": 0.0001, "loss": 0.0333, "step": 8560 }, { "epoch": 50.710059171597635, "grad_norm": 2.294217586517334, "learning_rate": 0.0001, "loss": 0.0336, "step": 8570 }, { "epoch": 50.76923076923077, "grad_norm": 2.676820755004883, "learning_rate": 0.0001, "loss": 0.0349, "step": 8580 }, { "epoch": 50.828402366863905, "grad_norm": 2.2816860675811768, "learning_rate": 0.0001, "loss": 0.0351, "step": 8590 }, { "epoch": 50.887573964497044, "grad_norm": 2.0957345962524414, "learning_rate": 0.0001, "loss": 0.0352, "step": 8600 }, { "epoch": 50.946745562130175, "grad_norm": 2.493990659713745, "learning_rate": 0.0001, "loss": 0.0339, "step": 8610 }, { "epoch": 51.005917159763314, "grad_norm": 2.00517201423645, "learning_rate": 0.0001, "loss": 0.0353, "step": 8620 }, { "epoch": 51.06508875739645, "grad_norm": 2.196530342102051, "learning_rate": 0.0001, "loss": 0.0345, "step": 8630 }, { "epoch": 51.124260355029584, "grad_norm": 2.09260630607605, "learning_rate": 0.0001, "loss": 0.0352, "step": 8640 }, { "epoch": 51.18343195266272, "grad_norm": 2.5671181678771973, "learning_rate": 0.0001, "loss": 0.0352, "step": 8650 }, { "epoch": 51.24260355029586, "grad_norm": 1.8794385194778442, "learning_rate": 0.0001, "loss": 0.0343, "step": 8660 }, { "epoch": 51.30177514792899, "grad_norm": 2.4146440029144287, "learning_rate": 0.0001, "loss": 0.036, "step": 8670 }, { "epoch": 51.36094674556213, "grad_norm": 2.2436561584472656, "learning_rate": 0.0001, "loss": 0.0345, "step": 8680 }, { "epoch": 51.42011834319526, "grad_norm": 1.977684497833252, "learning_rate": 0.0001, "loss": 0.0339, "step": 8690 }, { "epoch": 51.4792899408284, "grad_norm": 2.258413314819336, "learning_rate": 0.0001, "loss": 0.0345, "step": 8700 }, { "epoch": 51.53846153846154, "grad_norm": 1.9720079898834229, "learning_rate": 0.0001, "loss": 0.034, "step": 8710 }, { "epoch": 51.59763313609467, "grad_norm": 2.0522758960723877, "learning_rate": 0.0001, "loss": 0.0338, "step": 8720 }, { "epoch": 51.65680473372781, "grad_norm": 1.741957664489746, "learning_rate": 0.0001, "loss": 0.0339, "step": 8730 }, { "epoch": 51.71597633136095, "grad_norm": 2.025357961654663, "learning_rate": 0.0001, "loss": 0.0333, "step": 8740 }, { "epoch": 51.77514792899408, "grad_norm": 2.1275250911712646, "learning_rate": 0.0001, "loss": 0.034, "step": 8750 }, { "epoch": 51.83431952662722, "grad_norm": 2.0279955863952637, "learning_rate": 0.0001, "loss": 0.0343, "step": 8760 }, { "epoch": 51.89349112426036, "grad_norm": 2.138831615447998, "learning_rate": 0.0001, "loss": 0.0351, "step": 8770 }, { "epoch": 51.95266272189349, "grad_norm": 2.0339295864105225, "learning_rate": 0.0001, "loss": 0.0339, "step": 8780 }, { "epoch": 52.01183431952663, "grad_norm": 2.109532117843628, "learning_rate": 0.0001, "loss": 0.0377, "step": 8790 }, { "epoch": 52.071005917159766, "grad_norm": 2.523235559463501, "learning_rate": 0.0001, "loss": 0.0362, "step": 8800 }, { "epoch": 52.1301775147929, "grad_norm": 1.8649858236312866, "learning_rate": 0.0001, "loss": 0.0356, "step": 8810 }, { "epoch": 52.18934911242604, "grad_norm": 1.890131950378418, "learning_rate": 0.0001, "loss": 0.0335, "step": 8820 }, { "epoch": 52.248520710059175, "grad_norm": 2.37738299369812, "learning_rate": 0.0001, "loss": 0.0345, "step": 8830 }, { "epoch": 52.30769230769231, "grad_norm": 2.4639177322387695, "learning_rate": 0.0001, "loss": 0.0352, "step": 8840 }, { "epoch": 52.366863905325445, "grad_norm": 2.622772455215454, "learning_rate": 0.0001, "loss": 0.0363, "step": 8850 }, { "epoch": 52.42603550295858, "grad_norm": 2.184363603591919, "learning_rate": 0.0001, "loss": 0.0344, "step": 8860 }, { "epoch": 52.485207100591715, "grad_norm": 1.7605830430984497, "learning_rate": 0.0001, "loss": 0.035, "step": 8870 }, { "epoch": 52.544378698224854, "grad_norm": 2.0559444427490234, "learning_rate": 0.0001, "loss": 0.0359, "step": 8880 }, { "epoch": 52.603550295857985, "grad_norm": 2.359647512435913, "learning_rate": 0.0001, "loss": 0.035, "step": 8890 }, { "epoch": 52.662721893491124, "grad_norm": 2.0853419303894043, "learning_rate": 0.0001, "loss": 0.0354, "step": 8900 }, { "epoch": 52.72189349112426, "grad_norm": 1.706286907196045, "learning_rate": 0.0001, "loss": 0.0327, "step": 8910 }, { "epoch": 52.781065088757394, "grad_norm": 1.9099857807159424, "learning_rate": 0.0001, "loss": 0.0338, "step": 8920 }, { "epoch": 52.84023668639053, "grad_norm": 1.9418201446533203, "learning_rate": 0.0001, "loss": 0.0336, "step": 8930 }, { "epoch": 52.89940828402367, "grad_norm": 2.295734167098999, "learning_rate": 0.0001, "loss": 0.0336, "step": 8940 }, { "epoch": 52.9585798816568, "grad_norm": 1.8070427179336548, "learning_rate": 0.0001, "loss": 0.0326, "step": 8950 }, { "epoch": 53.01775147928994, "grad_norm": 2.1482555866241455, "learning_rate": 0.0001, "loss": 0.0327, "step": 8960 }, { "epoch": 53.07692307692308, "grad_norm": 2.322425365447998, "learning_rate": 0.0001, "loss": 0.0334, "step": 8970 }, { "epoch": 53.13609467455621, "grad_norm": 1.8825812339782715, "learning_rate": 0.0001, "loss": 0.0353, "step": 8980 }, { "epoch": 53.19526627218935, "grad_norm": 1.7828145027160645, "learning_rate": 0.0001, "loss": 0.0332, "step": 8990 }, { "epoch": 53.25443786982248, "grad_norm": 1.7910603284835815, "learning_rate": 0.0001, "loss": 0.034, "step": 9000 }, { "epoch": 53.31360946745562, "grad_norm": 1.8862606287002563, "learning_rate": 0.0001, "loss": 0.0347, "step": 9010 }, { "epoch": 53.37278106508876, "grad_norm": 2.2099781036376953, "learning_rate": 0.0001, "loss": 0.0336, "step": 9020 }, { "epoch": 53.43195266272189, "grad_norm": 2.273287296295166, "learning_rate": 0.0001, "loss": 0.0348, "step": 9030 }, { "epoch": 53.49112426035503, "grad_norm": 1.9388978481292725, "learning_rate": 0.0001, "loss": 0.0335, "step": 9040 }, { "epoch": 53.55029585798817, "grad_norm": 1.755401611328125, "learning_rate": 0.0001, "loss": 0.035, "step": 9050 }, { "epoch": 53.6094674556213, "grad_norm": 1.8453328609466553, "learning_rate": 0.0001, "loss": 0.0337, "step": 9060 }, { "epoch": 53.66863905325444, "grad_norm": 2.3412225246429443, "learning_rate": 0.0001, "loss": 0.0327, "step": 9070 }, { "epoch": 53.72781065088758, "grad_norm": 2.343360185623169, "learning_rate": 0.0001, "loss": 0.0343, "step": 9080 }, { "epoch": 53.78698224852071, "grad_norm": 2.270538091659546, "learning_rate": 0.0001, "loss": 0.0348, "step": 9090 }, { "epoch": 53.84615384615385, "grad_norm": 2.4746041297912598, "learning_rate": 0.0001, "loss": 0.0337, "step": 9100 }, { "epoch": 53.905325443786985, "grad_norm": 2.308973550796509, "learning_rate": 0.0001, "loss": 0.0357, "step": 9110 }, { "epoch": 53.96449704142012, "grad_norm": 2.441244125366211, "learning_rate": 0.0001, "loss": 0.0351, "step": 9120 }, { "epoch": 54.023668639053255, "grad_norm": 1.900873064994812, "learning_rate": 0.0001, "loss": 0.0352, "step": 9130 }, { "epoch": 54.082840236686394, "grad_norm": 2.586348056793213, "learning_rate": 0.0001, "loss": 0.0334, "step": 9140 }, { "epoch": 54.142011834319526, "grad_norm": 1.8043115139007568, "learning_rate": 0.0001, "loss": 0.0334, "step": 9150 }, { "epoch": 54.201183431952664, "grad_norm": 2.8228139877319336, "learning_rate": 0.0001, "loss": 0.0346, "step": 9160 }, { "epoch": 54.260355029585796, "grad_norm": 2.2045090198516846, "learning_rate": 0.0001, "loss": 0.0368, "step": 9170 }, { "epoch": 54.319526627218934, "grad_norm": 2.132056951522827, "learning_rate": 0.0001, "loss": 0.0344, "step": 9180 }, { "epoch": 54.37869822485207, "grad_norm": 2.0247793197631836, "learning_rate": 0.0001, "loss": 0.0356, "step": 9190 }, { "epoch": 54.437869822485204, "grad_norm": 1.8188918828964233, "learning_rate": 0.0001, "loss": 0.0336, "step": 9200 }, { "epoch": 54.49704142011834, "grad_norm": 2.3375794887542725, "learning_rate": 0.0001, "loss": 0.0353, "step": 9210 }, { "epoch": 54.55621301775148, "grad_norm": 2.1382980346679688, "learning_rate": 0.0001, "loss": 0.0351, "step": 9220 }, { "epoch": 54.61538461538461, "grad_norm": 1.945704698562622, "learning_rate": 0.0001, "loss": 0.0345, "step": 9230 }, { "epoch": 54.67455621301775, "grad_norm": 1.9352401494979858, "learning_rate": 0.0001, "loss": 0.0331, "step": 9240 }, { "epoch": 54.73372781065089, "grad_norm": 1.845940351486206, "learning_rate": 0.0001, "loss": 0.0327, "step": 9250 }, { "epoch": 54.79289940828402, "grad_norm": 2.6265604496002197, "learning_rate": 0.0001, "loss": 0.0341, "step": 9260 }, { "epoch": 54.85207100591716, "grad_norm": 2.6199448108673096, "learning_rate": 0.0001, "loss": 0.0341, "step": 9270 }, { "epoch": 54.9112426035503, "grad_norm": 2.0406994819641113, "learning_rate": 0.0001, "loss": 0.0342, "step": 9280 }, { "epoch": 54.97041420118343, "grad_norm": 2.3151042461395264, "learning_rate": 0.0001, "loss": 0.0334, "step": 9290 }, { "epoch": 55.02958579881657, "grad_norm": 1.6355373859405518, "learning_rate": 0.0001, "loss": 0.0328, "step": 9300 }, { "epoch": 55.0887573964497, "grad_norm": 2.239894390106201, "learning_rate": 0.0001, "loss": 0.0333, "step": 9310 }, { "epoch": 55.14792899408284, "grad_norm": 1.913952350616455, "learning_rate": 0.0001, "loss": 0.0328, "step": 9320 }, { "epoch": 55.20710059171598, "grad_norm": 1.6797480583190918, "learning_rate": 0.0001, "loss": 0.0314, "step": 9330 }, { "epoch": 55.26627218934911, "grad_norm": 1.8226990699768066, "learning_rate": 0.0001, "loss": 0.0319, "step": 9340 }, { "epoch": 55.32544378698225, "grad_norm": 1.924145221710205, "learning_rate": 0.0001, "loss": 0.031, "step": 9350 }, { "epoch": 55.38461538461539, "grad_norm": 2.2468395233154297, "learning_rate": 0.0001, "loss": 0.0342, "step": 9360 }, { "epoch": 55.44378698224852, "grad_norm": 3.0827219486236572, "learning_rate": 0.0001, "loss": 0.0347, "step": 9370 }, { "epoch": 55.50295857988166, "grad_norm": 2.0826690196990967, "learning_rate": 0.0001, "loss": 0.0347, "step": 9380 }, { "epoch": 55.562130177514796, "grad_norm": 1.735480785369873, "learning_rate": 0.0001, "loss": 0.032, "step": 9390 }, { "epoch": 55.62130177514793, "grad_norm": 2.650045156478882, "learning_rate": 0.0001, "loss": 0.0318, "step": 9400 }, { "epoch": 55.680473372781066, "grad_norm": 1.7872580289840698, "learning_rate": 0.0001, "loss": 0.0336, "step": 9410 }, { "epoch": 55.739644970414204, "grad_norm": 1.812215805053711, "learning_rate": 0.0001, "loss": 0.0323, "step": 9420 }, { "epoch": 55.798816568047336, "grad_norm": 1.9940041303634644, "learning_rate": 0.0001, "loss": 0.0329, "step": 9430 }, { "epoch": 55.857988165680474, "grad_norm": 1.7573813199996948, "learning_rate": 0.0001, "loss": 0.0318, "step": 9440 }, { "epoch": 55.917159763313606, "grad_norm": 1.726826548576355, "learning_rate": 0.0001, "loss": 0.0346, "step": 9450 }, { "epoch": 55.976331360946745, "grad_norm": 2.1589999198913574, "learning_rate": 0.0001, "loss": 0.0345, "step": 9460 }, { "epoch": 56.03550295857988, "grad_norm": 1.9314491748809814, "learning_rate": 0.0001, "loss": 0.0353, "step": 9470 }, { "epoch": 56.094674556213015, "grad_norm": 2.285259962081909, "learning_rate": 0.0001, "loss": 0.0346, "step": 9480 }, { "epoch": 56.15384615384615, "grad_norm": 2.063056468963623, "learning_rate": 0.0001, "loss": 0.0319, "step": 9490 }, { "epoch": 56.21301775147929, "grad_norm": 1.6753276586532593, "learning_rate": 0.0001, "loss": 0.0333, "step": 9500 }, { "epoch": 56.27218934911242, "grad_norm": 1.8650649785995483, "learning_rate": 0.0001, "loss": 0.0348, "step": 9510 }, { "epoch": 56.33136094674556, "grad_norm": 1.899198055267334, "learning_rate": 0.0001, "loss": 0.0351, "step": 9520 }, { "epoch": 56.3905325443787, "grad_norm": 1.7436795234680176, "learning_rate": 0.0001, "loss": 0.0355, "step": 9530 }, { "epoch": 56.44970414201183, "grad_norm": 2.628329277038574, "learning_rate": 0.0001, "loss": 0.0338, "step": 9540 }, { "epoch": 56.50887573964497, "grad_norm": 2.1853408813476562, "learning_rate": 0.0001, "loss": 0.0348, "step": 9550 }, { "epoch": 56.56804733727811, "grad_norm": 1.9088613986968994, "learning_rate": 0.0001, "loss": 0.0333, "step": 9560 }, { "epoch": 56.62721893491124, "grad_norm": 1.7584059238433838, "learning_rate": 0.0001, "loss": 0.0344, "step": 9570 }, { "epoch": 56.68639053254438, "grad_norm": 2.0921554565429688, "learning_rate": 0.0001, "loss": 0.0338, "step": 9580 }, { "epoch": 56.74556213017752, "grad_norm": 1.6412562131881714, "learning_rate": 0.0001, "loss": 0.0338, "step": 9590 }, { "epoch": 56.80473372781065, "grad_norm": 2.7710649967193604, "learning_rate": 0.0001, "loss": 0.0332, "step": 9600 }, { "epoch": 56.86390532544379, "grad_norm": 1.8369601964950562, "learning_rate": 0.0001, "loss": 0.0331, "step": 9610 }, { "epoch": 56.92307692307692, "grad_norm": 1.670758605003357, "learning_rate": 0.0001, "loss": 0.0345, "step": 9620 }, { "epoch": 56.98224852071006, "grad_norm": 2.5321967601776123, "learning_rate": 0.0001, "loss": 0.0345, "step": 9630 }, { "epoch": 57.0414201183432, "grad_norm": 1.8813183307647705, "learning_rate": 0.0001, "loss": 0.032, "step": 9640 }, { "epoch": 57.10059171597633, "grad_norm": 2.062991142272949, "learning_rate": 0.0001, "loss": 0.032, "step": 9650 }, { "epoch": 57.15976331360947, "grad_norm": 1.9301185607910156, "learning_rate": 0.0001, "loss": 0.0318, "step": 9660 }, { "epoch": 57.218934911242606, "grad_norm": 1.9805328845977783, "learning_rate": 0.0001, "loss": 0.0317, "step": 9670 }, { "epoch": 57.27810650887574, "grad_norm": 2.0994582176208496, "learning_rate": 0.0001, "loss": 0.0313, "step": 9680 }, { "epoch": 57.337278106508876, "grad_norm": 1.9402161836624146, "learning_rate": 0.0001, "loss": 0.0336, "step": 9690 }, { "epoch": 57.396449704142015, "grad_norm": 2.5956501960754395, "learning_rate": 0.0001, "loss": 0.0327, "step": 9700 }, { "epoch": 57.455621301775146, "grad_norm": 1.6578630208969116, "learning_rate": 0.0001, "loss": 0.0318, "step": 9710 }, { "epoch": 57.514792899408285, "grad_norm": 2.2813684940338135, "learning_rate": 0.0001, "loss": 0.0327, "step": 9720 }, { "epoch": 57.57396449704142, "grad_norm": 1.4863901138305664, "learning_rate": 0.0001, "loss": 0.0314, "step": 9730 }, { "epoch": 57.633136094674555, "grad_norm": 2.322230339050293, "learning_rate": 0.0001, "loss": 0.0309, "step": 9740 }, { "epoch": 57.69230769230769, "grad_norm": 1.744634747505188, "learning_rate": 0.0001, "loss": 0.0314, "step": 9750 }, { "epoch": 57.751479289940825, "grad_norm": 1.7688119411468506, "learning_rate": 0.0001, "loss": 0.0326, "step": 9760 }, { "epoch": 57.81065088757396, "grad_norm": 1.9524624347686768, "learning_rate": 0.0001, "loss": 0.0317, "step": 9770 }, { "epoch": 57.8698224852071, "grad_norm": 1.7301290035247803, "learning_rate": 0.0001, "loss": 0.0323, "step": 9780 }, { "epoch": 57.928994082840234, "grad_norm": 2.084578514099121, "learning_rate": 0.0001, "loss": 0.0332, "step": 9790 }, { "epoch": 57.98816568047337, "grad_norm": 1.910014033317566, "learning_rate": 0.0001, "loss": 0.0359, "step": 9800 }, { "epoch": 58.04733727810651, "grad_norm": 2.2146341800689697, "learning_rate": 0.0001, "loss": 0.033, "step": 9810 }, { "epoch": 58.10650887573964, "grad_norm": 2.111356019973755, "learning_rate": 0.0001, "loss": 0.0325, "step": 9820 }, { "epoch": 58.16568047337278, "grad_norm": 1.6786781549453735, "learning_rate": 0.0001, "loss": 0.0332, "step": 9830 }, { "epoch": 58.22485207100592, "grad_norm": 2.3371386528015137, "learning_rate": 0.0001, "loss": 0.0339, "step": 9840 }, { "epoch": 58.28402366863905, "grad_norm": 2.2871005535125732, "learning_rate": 0.0001, "loss": 0.033, "step": 9850 }, { "epoch": 58.34319526627219, "grad_norm": 2.565744638442993, "learning_rate": 0.0001, "loss": 0.0317, "step": 9860 }, { "epoch": 58.40236686390533, "grad_norm": 1.8770043849945068, "learning_rate": 0.0001, "loss": 0.0309, "step": 9870 }, { "epoch": 58.46153846153846, "grad_norm": 1.9037714004516602, "learning_rate": 0.0001, "loss": 0.0317, "step": 9880 }, { "epoch": 58.5207100591716, "grad_norm": 1.7115402221679688, "learning_rate": 0.0001, "loss": 0.0319, "step": 9890 }, { "epoch": 58.57988165680474, "grad_norm": 1.89598548412323, "learning_rate": 0.0001, "loss": 0.0319, "step": 9900 }, { "epoch": 58.63905325443787, "grad_norm": 1.9760292768478394, "learning_rate": 0.0001, "loss": 0.0319, "step": 9910 }, { "epoch": 58.69822485207101, "grad_norm": 1.6471452713012695, "learning_rate": 0.0001, "loss": 0.0318, "step": 9920 }, { "epoch": 58.75739644970414, "grad_norm": 1.8194528818130493, "learning_rate": 0.0001, "loss": 0.0324, "step": 9930 }, { "epoch": 58.81656804733728, "grad_norm": 1.846859097480774, "learning_rate": 0.0001, "loss": 0.0308, "step": 9940 }, { "epoch": 58.875739644970416, "grad_norm": 1.988418459892273, "learning_rate": 0.0001, "loss": 0.0313, "step": 9950 }, { "epoch": 58.93491124260355, "grad_norm": 1.9187120199203491, "learning_rate": 0.0001, "loss": 0.0323, "step": 9960 }, { "epoch": 58.994082840236686, "grad_norm": 1.9139988422393799, "learning_rate": 0.0001, "loss": 0.0345, "step": 9970 }, { "epoch": 59.053254437869825, "grad_norm": 1.8829811811447144, "learning_rate": 0.0001, "loss": 0.0327, "step": 9980 }, { "epoch": 59.112426035502956, "grad_norm": 2.203744888305664, "learning_rate": 0.0001, "loss": 0.0343, "step": 9990 }, { "epoch": 59.171597633136095, "grad_norm": 2.0762901306152344, "learning_rate": 0.0001, "loss": 0.0318, "step": 10000 }, { "epoch": 59.23076923076923, "grad_norm": 1.9356908798217773, "learning_rate": 0.0001, "loss": 0.0325, "step": 10010 }, { "epoch": 59.289940828402365, "grad_norm": 2.209742307662964, "learning_rate": 0.0001, "loss": 0.0323, "step": 10020 }, { "epoch": 59.349112426035504, "grad_norm": 1.9134081602096558, "learning_rate": 0.0001, "loss": 0.0315, "step": 10030 }, { "epoch": 59.40828402366864, "grad_norm": 1.762304425239563, "learning_rate": 0.0001, "loss": 0.0338, "step": 10040 }, { "epoch": 59.467455621301774, "grad_norm": 2.5498056411743164, "learning_rate": 0.0001, "loss": 0.0307, "step": 10050 }, { "epoch": 59.52662721893491, "grad_norm": 1.9540791511535645, "learning_rate": 0.0001, "loss": 0.0332, "step": 10060 }, { "epoch": 59.585798816568044, "grad_norm": 1.905519723892212, "learning_rate": 0.0001, "loss": 0.0324, "step": 10070 }, { "epoch": 59.64497041420118, "grad_norm": 1.9742188453674316, "learning_rate": 0.0001, "loss": 0.0314, "step": 10080 }, { "epoch": 59.70414201183432, "grad_norm": 1.8438178300857544, "learning_rate": 0.0001, "loss": 0.0323, "step": 10090 }, { "epoch": 59.76331360946745, "grad_norm": 2.273754358291626, "learning_rate": 0.0001, "loss": 0.0302, "step": 10100 }, { "epoch": 59.82248520710059, "grad_norm": 1.8264018297195435, "learning_rate": 0.0001, "loss": 0.0318, "step": 10110 }, { "epoch": 59.88165680473373, "grad_norm": 1.4410581588745117, "learning_rate": 0.0001, "loss": 0.0323, "step": 10120 }, { "epoch": 59.94082840236686, "grad_norm": 2.144582986831665, "learning_rate": 0.0001, "loss": 0.0324, "step": 10130 }, { "epoch": 60.0, "grad_norm": 2.5620033740997314, "learning_rate": 0.0001, "loss": 0.0324, "step": 10140 }, { "epoch": 60.05917159763314, "grad_norm": 2.0461485385894775, "learning_rate": 0.0001, "loss": 0.0342, "step": 10150 }, { "epoch": 60.11834319526627, "grad_norm": 1.9092527627944946, "learning_rate": 0.0001, "loss": 0.0313, "step": 10160 }, { "epoch": 60.17751479289941, "grad_norm": 1.8468396663665771, "learning_rate": 0.0001, "loss": 0.0313, "step": 10170 }, { "epoch": 60.23668639053255, "grad_norm": 2.1154861450195312, "learning_rate": 0.0001, "loss": 0.0315, "step": 10180 }, { "epoch": 60.29585798816568, "grad_norm": 1.7729041576385498, "learning_rate": 0.0001, "loss": 0.0314, "step": 10190 }, { "epoch": 60.35502958579882, "grad_norm": 1.6574442386627197, "learning_rate": 0.0001, "loss": 0.0318, "step": 10200 }, { "epoch": 60.414201183431956, "grad_norm": 2.367227554321289, "learning_rate": 0.0001, "loss": 0.0309, "step": 10210 }, { "epoch": 60.47337278106509, "grad_norm": 2.2383310794830322, "learning_rate": 0.0001, "loss": 0.0314, "step": 10220 }, { "epoch": 60.532544378698226, "grad_norm": 2.322610855102539, "learning_rate": 0.0001, "loss": 0.0342, "step": 10230 }, { "epoch": 60.59171597633136, "grad_norm": 2.1273884773254395, "learning_rate": 0.0001, "loss": 0.0327, "step": 10240 }, { "epoch": 60.650887573964496, "grad_norm": 1.7425283193588257, "learning_rate": 0.0001, "loss": 0.0322, "step": 10250 }, { "epoch": 60.710059171597635, "grad_norm": 2.4041225910186768, "learning_rate": 0.0001, "loss": 0.0319, "step": 10260 }, { "epoch": 60.76923076923077, "grad_norm": 1.4759029150009155, "learning_rate": 0.0001, "loss": 0.0317, "step": 10270 }, { "epoch": 60.828402366863905, "grad_norm": 1.968384027481079, "learning_rate": 0.0001, "loss": 0.0314, "step": 10280 }, { "epoch": 60.887573964497044, "grad_norm": 2.237220287322998, "learning_rate": 0.0001, "loss": 0.0317, "step": 10290 }, { "epoch": 60.946745562130175, "grad_norm": 1.9968286752700806, "learning_rate": 0.0001, "loss": 0.0322, "step": 10300 }, { "epoch": 61.005917159763314, "grad_norm": 2.2673959732055664, "learning_rate": 0.0001, "loss": 0.0305, "step": 10310 }, { "epoch": 61.06508875739645, "grad_norm": 1.9304046630859375, "learning_rate": 0.0001, "loss": 0.0303, "step": 10320 }, { "epoch": 61.124260355029584, "grad_norm": 2.2834107875823975, "learning_rate": 0.0001, "loss": 0.0311, "step": 10330 }, { "epoch": 61.18343195266272, "grad_norm": 1.6877681016921997, "learning_rate": 0.0001, "loss": 0.0324, "step": 10340 }, { "epoch": 61.24260355029586, "grad_norm": 1.630195140838623, "learning_rate": 0.0001, "loss": 0.0338, "step": 10350 }, { "epoch": 61.30177514792899, "grad_norm": 1.4190911054611206, "learning_rate": 0.0001, "loss": 0.0322, "step": 10360 }, { "epoch": 61.36094674556213, "grad_norm": 1.7607923746109009, "learning_rate": 0.0001, "loss": 0.0334, "step": 10370 }, { "epoch": 61.42011834319526, "grad_norm": 1.9183456897735596, "learning_rate": 0.0001, "loss": 0.032, "step": 10380 }, { "epoch": 61.4792899408284, "grad_norm": 1.8912131786346436, "learning_rate": 0.0001, "loss": 0.0334, "step": 10390 }, { "epoch": 61.53846153846154, "grad_norm": 1.68782639503479, "learning_rate": 0.0001, "loss": 0.0327, "step": 10400 }, { "epoch": 61.59763313609467, "grad_norm": 2.022503614425659, "learning_rate": 0.0001, "loss": 0.0309, "step": 10410 }, { "epoch": 61.65680473372781, "grad_norm": 1.5390222072601318, "learning_rate": 0.0001, "loss": 0.0317, "step": 10420 }, { "epoch": 61.71597633136095, "grad_norm": 1.7554931640625, "learning_rate": 0.0001, "loss": 0.0325, "step": 10430 }, { "epoch": 61.77514792899408, "grad_norm": 1.5205967426300049, "learning_rate": 0.0001, "loss": 0.0328, "step": 10440 }, { "epoch": 61.83431952662722, "grad_norm": 1.5686376094818115, "learning_rate": 0.0001, "loss": 0.032, "step": 10450 }, { "epoch": 61.89349112426036, "grad_norm": 1.7493751049041748, "learning_rate": 0.0001, "loss": 0.0337, "step": 10460 }, { "epoch": 61.95266272189349, "grad_norm": 1.98441481590271, "learning_rate": 0.0001, "loss": 0.0315, "step": 10470 }, { "epoch": 62.01183431952663, "grad_norm": 2.361846685409546, "learning_rate": 0.0001, "loss": 0.0317, "step": 10480 }, { "epoch": 62.071005917159766, "grad_norm": 1.7203779220581055, "learning_rate": 0.0001, "loss": 0.0313, "step": 10490 }, { "epoch": 62.1301775147929, "grad_norm": 1.7348029613494873, "learning_rate": 0.0001, "loss": 0.032, "step": 10500 }, { "epoch": 62.18934911242604, "grad_norm": 1.9366225004196167, "learning_rate": 0.0001, "loss": 0.0324, "step": 10510 }, { "epoch": 62.248520710059175, "grad_norm": 2.231485366821289, "learning_rate": 0.0001, "loss": 0.0337, "step": 10520 }, { "epoch": 62.30769230769231, "grad_norm": 1.759955644607544, "learning_rate": 0.0001, "loss": 0.033, "step": 10530 }, { "epoch": 62.366863905325445, "grad_norm": 1.8047053813934326, "learning_rate": 0.0001, "loss": 0.0318, "step": 10540 }, { "epoch": 62.42603550295858, "grad_norm": 1.8735262155532837, "learning_rate": 0.0001, "loss": 0.0324, "step": 10550 }, { "epoch": 62.485207100591715, "grad_norm": 2.2876079082489014, "learning_rate": 0.0001, "loss": 0.0321, "step": 10560 }, { "epoch": 62.544378698224854, "grad_norm": 1.9605590105056763, "learning_rate": 0.0001, "loss": 0.0311, "step": 10570 }, { "epoch": 62.603550295857985, "grad_norm": 2.257314920425415, "learning_rate": 0.0001, "loss": 0.0319, "step": 10580 }, { "epoch": 62.662721893491124, "grad_norm": 2.1091785430908203, "learning_rate": 0.0001, "loss": 0.0321, "step": 10590 }, { "epoch": 62.72189349112426, "grad_norm": 1.8176383972167969, "learning_rate": 0.0001, "loss": 0.0324, "step": 10600 }, { "epoch": 62.781065088757394, "grad_norm": 1.7573214769363403, "learning_rate": 0.0001, "loss": 0.0318, "step": 10610 }, { "epoch": 62.84023668639053, "grad_norm": 1.709418535232544, "learning_rate": 0.0001, "loss": 0.0307, "step": 10620 }, { "epoch": 62.89940828402367, "grad_norm": 2.1767807006835938, "learning_rate": 0.0001, "loss": 0.0314, "step": 10630 }, { "epoch": 62.9585798816568, "grad_norm": 2.4363853931427, "learning_rate": 0.0001, "loss": 0.0307, "step": 10640 }, { "epoch": 63.01775147928994, "grad_norm": 2.403937578201294, "learning_rate": 0.0001, "loss": 0.0312, "step": 10650 }, { "epoch": 63.07692307692308, "grad_norm": 2.152991533279419, "learning_rate": 0.0001, "loss": 0.0319, "step": 10660 }, { "epoch": 63.13609467455621, "grad_norm": 2.4945948123931885, "learning_rate": 0.0001, "loss": 0.0308, "step": 10670 }, { "epoch": 63.19526627218935, "grad_norm": 1.7772583961486816, "learning_rate": 0.0001, "loss": 0.0322, "step": 10680 }, { "epoch": 63.25443786982248, "grad_norm": 1.9819145202636719, "learning_rate": 0.0001, "loss": 0.0312, "step": 10690 }, { "epoch": 63.31360946745562, "grad_norm": 2.0326600074768066, "learning_rate": 0.0001, "loss": 0.0313, "step": 10700 }, { "epoch": 63.37278106508876, "grad_norm": 1.8770262002944946, "learning_rate": 0.0001, "loss": 0.0318, "step": 10710 }, { "epoch": 63.43195266272189, "grad_norm": 1.4358137845993042, "learning_rate": 0.0001, "loss": 0.0317, "step": 10720 }, { "epoch": 63.49112426035503, "grad_norm": 1.4048131704330444, "learning_rate": 0.0001, "loss": 0.0323, "step": 10730 }, { "epoch": 63.55029585798817, "grad_norm": 1.7929816246032715, "learning_rate": 0.0001, "loss": 0.0301, "step": 10740 }, { "epoch": 63.6094674556213, "grad_norm": 2.1733293533325195, "learning_rate": 0.0001, "loss": 0.0308, "step": 10750 }, { "epoch": 63.66863905325444, "grad_norm": 2.088038206100464, "learning_rate": 0.0001, "loss": 0.0306, "step": 10760 }, { "epoch": 63.72781065088758, "grad_norm": 1.619127869606018, "learning_rate": 0.0001, "loss": 0.0298, "step": 10770 }, { "epoch": 63.78698224852071, "grad_norm": 1.643599271774292, "learning_rate": 0.0001, "loss": 0.0314, "step": 10780 }, { "epoch": 63.84615384615385, "grad_norm": 2.1038687229156494, "learning_rate": 0.0001, "loss": 0.0337, "step": 10790 }, { "epoch": 63.905325443786985, "grad_norm": 1.4514011144638062, "learning_rate": 0.0001, "loss": 0.0318, "step": 10800 }, { "epoch": 63.96449704142012, "grad_norm": 1.8902044296264648, "learning_rate": 0.0001, "loss": 0.0311, "step": 10810 }, { "epoch": 64.02366863905326, "grad_norm": 2.1965935230255127, "learning_rate": 0.0001, "loss": 0.0319, "step": 10820 }, { "epoch": 64.0828402366864, "grad_norm": 1.5566917657852173, "learning_rate": 0.0001, "loss": 0.0311, "step": 10830 }, { "epoch": 64.14201183431953, "grad_norm": 1.55888831615448, "learning_rate": 0.0001, "loss": 0.0306, "step": 10840 }, { "epoch": 64.20118343195266, "grad_norm": 1.7706501483917236, "learning_rate": 0.0001, "loss": 0.0306, "step": 10850 }, { "epoch": 64.2603550295858, "grad_norm": 1.6196694374084473, "learning_rate": 0.0001, "loss": 0.0321, "step": 10860 }, { "epoch": 64.31952662721893, "grad_norm": 2.3305320739746094, "learning_rate": 0.0001, "loss": 0.0321, "step": 10870 }, { "epoch": 64.37869822485207, "grad_norm": 2.0537736415863037, "learning_rate": 0.0001, "loss": 0.0295, "step": 10880 }, { "epoch": 64.43786982248521, "grad_norm": 1.881563425064087, "learning_rate": 0.0001, "loss": 0.0306, "step": 10890 }, { "epoch": 64.49704142011835, "grad_norm": 1.95639967918396, "learning_rate": 0.0001, "loss": 0.0319, "step": 10900 }, { "epoch": 64.55621301775147, "grad_norm": 1.6108462810516357, "learning_rate": 0.0001, "loss": 0.0308, "step": 10910 }, { "epoch": 64.61538461538461, "grad_norm": 2.028263568878174, "learning_rate": 0.0001, "loss": 0.0314, "step": 10920 }, { "epoch": 64.67455621301775, "grad_norm": 2.217963695526123, "learning_rate": 0.0001, "loss": 0.0304, "step": 10930 }, { "epoch": 64.73372781065089, "grad_norm": 1.960577130317688, "learning_rate": 0.0001, "loss": 0.0319, "step": 10940 }, { "epoch": 64.79289940828403, "grad_norm": 1.7715681791305542, "learning_rate": 0.0001, "loss": 0.0318, "step": 10950 }, { "epoch": 64.85207100591715, "grad_norm": 1.741167664527893, "learning_rate": 0.0001, "loss": 0.0322, "step": 10960 }, { "epoch": 64.91124260355029, "grad_norm": 2.068405866622925, "learning_rate": 0.0001, "loss": 0.0326, "step": 10970 }, { "epoch": 64.97041420118343, "grad_norm": 1.8818506002426147, "learning_rate": 0.0001, "loss": 0.0335, "step": 10980 }, { "epoch": 65.02958579881657, "grad_norm": 1.656443476676941, "learning_rate": 0.0001, "loss": 0.0335, "step": 10990 }, { "epoch": 65.08875739644971, "grad_norm": 1.569478154182434, "learning_rate": 0.0001, "loss": 0.0315, "step": 11000 }, { "epoch": 65.14792899408285, "grad_norm": 1.3876622915267944, "learning_rate": 0.0001, "loss": 0.0295, "step": 11010 }, { "epoch": 65.20710059171597, "grad_norm": 2.528808355331421, "learning_rate": 0.0001, "loss": 0.0319, "step": 11020 }, { "epoch": 65.26627218934911, "grad_norm": 1.571533441543579, "learning_rate": 0.0001, "loss": 0.0309, "step": 11030 }, { "epoch": 65.32544378698225, "grad_norm": 1.6685060262680054, "learning_rate": 0.0001, "loss": 0.0323, "step": 11040 }, { "epoch": 65.38461538461539, "grad_norm": 2.014483690261841, "learning_rate": 0.0001, "loss": 0.0318, "step": 11050 }, { "epoch": 65.44378698224853, "grad_norm": 1.5756182670593262, "learning_rate": 0.0001, "loss": 0.0312, "step": 11060 }, { "epoch": 65.50295857988165, "grad_norm": 1.6968916654586792, "learning_rate": 0.0001, "loss": 0.0314, "step": 11070 }, { "epoch": 65.56213017751479, "grad_norm": 2.7476720809936523, "learning_rate": 0.0001, "loss": 0.0318, "step": 11080 }, { "epoch": 65.62130177514793, "grad_norm": 2.2051241397857666, "learning_rate": 0.0001, "loss": 0.0329, "step": 11090 }, { "epoch": 65.68047337278107, "grad_norm": 2.003256320953369, "learning_rate": 0.0001, "loss": 0.0332, "step": 11100 }, { "epoch": 65.7396449704142, "grad_norm": 2.1924686431884766, "learning_rate": 0.0001, "loss": 0.0323, "step": 11110 }, { "epoch": 65.79881656804734, "grad_norm": 1.369592308998108, "learning_rate": 0.0001, "loss": 0.0341, "step": 11120 }, { "epoch": 65.85798816568047, "grad_norm": 1.3310401439666748, "learning_rate": 0.0001, "loss": 0.0303, "step": 11130 }, { "epoch": 65.9171597633136, "grad_norm": 1.7399251461029053, "learning_rate": 0.0001, "loss": 0.0327, "step": 11140 }, { "epoch": 65.97633136094674, "grad_norm": 1.5883480310440063, "learning_rate": 0.0001, "loss": 0.0346, "step": 11150 }, { "epoch": 66.03550295857988, "grad_norm": 2.1667933464050293, "learning_rate": 0.0001, "loss": 0.034, "step": 11160 }, { "epoch": 66.09467455621302, "grad_norm": 2.4345896244049072, "learning_rate": 0.0001, "loss": 0.0334, "step": 11170 }, { "epoch": 66.15384615384616, "grad_norm": 2.653026580810547, "learning_rate": 0.0001, "loss": 0.0339, "step": 11180 }, { "epoch": 66.21301775147928, "grad_norm": 2.647099018096924, "learning_rate": 0.0001, "loss": 0.0337, "step": 11190 }, { "epoch": 66.27218934911242, "grad_norm": 2.142895221710205, "learning_rate": 0.0001, "loss": 0.033, "step": 11200 }, { "epoch": 66.33136094674556, "grad_norm": 2.0091168880462646, "learning_rate": 0.0001, "loss": 0.0315, "step": 11210 }, { "epoch": 66.3905325443787, "grad_norm": 1.6778756380081177, "learning_rate": 0.0001, "loss": 0.0312, "step": 11220 }, { "epoch": 66.44970414201184, "grad_norm": 1.6732847690582275, "learning_rate": 0.0001, "loss": 0.0301, "step": 11230 }, { "epoch": 66.50887573964496, "grad_norm": 1.8221442699432373, "learning_rate": 0.0001, "loss": 0.0312, "step": 11240 }, { "epoch": 66.5680473372781, "grad_norm": 1.5187722444534302, "learning_rate": 0.0001, "loss": 0.0312, "step": 11250 }, { "epoch": 66.62721893491124, "grad_norm": 1.8520886898040771, "learning_rate": 0.0001, "loss": 0.0334, "step": 11260 }, { "epoch": 66.68639053254438, "grad_norm": 1.9234912395477295, "learning_rate": 0.0001, "loss": 0.0314, "step": 11270 }, { "epoch": 66.74556213017752, "grad_norm": 1.4833284616470337, "learning_rate": 0.0001, "loss": 0.0333, "step": 11280 }, { "epoch": 66.80473372781066, "grad_norm": 1.5189228057861328, "learning_rate": 0.0001, "loss": 0.0324, "step": 11290 }, { "epoch": 66.86390532544378, "grad_norm": 1.524887204170227, "learning_rate": 0.0001, "loss": 0.0316, "step": 11300 }, { "epoch": 66.92307692307692, "grad_norm": 1.8200448751449585, "learning_rate": 0.0001, "loss": 0.0314, "step": 11310 }, { "epoch": 66.98224852071006, "grad_norm": 1.5401345491409302, "learning_rate": 0.0001, "loss": 0.0301, "step": 11320 }, { "epoch": 67.0414201183432, "grad_norm": 2.1249403953552246, "learning_rate": 0.0001, "loss": 0.033, "step": 11330 }, { "epoch": 67.10059171597634, "grad_norm": 2.0871784687042236, "learning_rate": 0.0001, "loss": 0.0312, "step": 11340 }, { "epoch": 67.15976331360947, "grad_norm": 2.6198816299438477, "learning_rate": 0.0001, "loss": 0.03, "step": 11350 }, { "epoch": 67.2189349112426, "grad_norm": 1.9927914142608643, "learning_rate": 0.0001, "loss": 0.0291, "step": 11360 }, { "epoch": 67.27810650887574, "grad_norm": 2.0338242053985596, "learning_rate": 0.0001, "loss": 0.0312, "step": 11370 }, { "epoch": 67.33727810650888, "grad_norm": 1.6952403783798218, "learning_rate": 0.0001, "loss": 0.0307, "step": 11380 }, { "epoch": 67.39644970414201, "grad_norm": 2.281118869781494, "learning_rate": 0.0001, "loss": 0.0319, "step": 11390 }, { "epoch": 67.45562130177515, "grad_norm": 2.000225782394409, "learning_rate": 0.0001, "loss": 0.0313, "step": 11400 }, { "epoch": 67.51479289940828, "grad_norm": 2.561842679977417, "learning_rate": 0.0001, "loss": 0.0334, "step": 11410 }, { "epoch": 67.57396449704142, "grad_norm": 1.6910122632980347, "learning_rate": 0.0001, "loss": 0.0306, "step": 11420 }, { "epoch": 67.63313609467455, "grad_norm": 1.4641021490097046, "learning_rate": 0.0001, "loss": 0.0299, "step": 11430 }, { "epoch": 67.6923076923077, "grad_norm": 1.3346073627471924, "learning_rate": 0.0001, "loss": 0.0296, "step": 11440 }, { "epoch": 67.75147928994083, "grad_norm": 1.702642560005188, "learning_rate": 0.0001, "loss": 0.0308, "step": 11450 }, { "epoch": 67.81065088757397, "grad_norm": 2.0324418544769287, "learning_rate": 0.0001, "loss": 0.0312, "step": 11460 }, { "epoch": 67.8698224852071, "grad_norm": 1.9170523881912231, "learning_rate": 0.0001, "loss": 0.0295, "step": 11470 }, { "epoch": 67.92899408284023, "grad_norm": 1.8339183330535889, "learning_rate": 0.0001, "loss": 0.03, "step": 11480 }, { "epoch": 67.98816568047337, "grad_norm": 1.4196767807006836, "learning_rate": 0.0001, "loss": 0.0301, "step": 11490 }, { "epoch": 68.04733727810651, "grad_norm": 2.0290603637695312, "learning_rate": 0.0001, "loss": 0.0298, "step": 11500 }, { "epoch": 68.10650887573965, "grad_norm": 2.2383458614349365, "learning_rate": 0.0001, "loss": 0.0301, "step": 11510 }, { "epoch": 68.16568047337279, "grad_norm": 1.7861320972442627, "learning_rate": 0.0001, "loss": 0.0308, "step": 11520 }, { "epoch": 68.22485207100591, "grad_norm": 1.765856385231018, "learning_rate": 0.0001, "loss": 0.0304, "step": 11530 }, { "epoch": 68.28402366863905, "grad_norm": 1.5731621980667114, "learning_rate": 0.0001, "loss": 0.0305, "step": 11540 }, { "epoch": 68.34319526627219, "grad_norm": 1.9535126686096191, "learning_rate": 0.0001, "loss": 0.0305, "step": 11550 }, { "epoch": 68.40236686390533, "grad_norm": 2.090447187423706, "learning_rate": 0.0001, "loss": 0.0308, "step": 11560 }, { "epoch": 68.46153846153847, "grad_norm": 1.685531735420227, "learning_rate": 0.0001, "loss": 0.0306, "step": 11570 }, { "epoch": 68.52071005917159, "grad_norm": 1.8909152746200562, "learning_rate": 0.0001, "loss": 0.031, "step": 11580 }, { "epoch": 68.57988165680473, "grad_norm": 1.7439522743225098, "learning_rate": 0.0001, "loss": 0.0293, "step": 11590 }, { "epoch": 68.63905325443787, "grad_norm": 2.198719024658203, "learning_rate": 0.0001, "loss": 0.0303, "step": 11600 }, { "epoch": 68.69822485207101, "grad_norm": 1.8812614679336548, "learning_rate": 0.0001, "loss": 0.0292, "step": 11610 }, { "epoch": 68.75739644970415, "grad_norm": 1.7807092666625977, "learning_rate": 0.0001, "loss": 0.0307, "step": 11620 }, { "epoch": 68.81656804733728, "grad_norm": 1.6185169219970703, "learning_rate": 0.0001, "loss": 0.0307, "step": 11630 }, { "epoch": 68.87573964497041, "grad_norm": 1.3825422525405884, "learning_rate": 0.0001, "loss": 0.032, "step": 11640 }, { "epoch": 68.93491124260355, "grad_norm": 2.83486008644104, "learning_rate": 0.0001, "loss": 0.0316, "step": 11650 }, { "epoch": 68.99408284023669, "grad_norm": 1.57773756980896, "learning_rate": 0.0001, "loss": 0.0299, "step": 11660 }, { "epoch": 69.05325443786982, "grad_norm": 1.922195553779602, "learning_rate": 0.0001, "loss": 0.0296, "step": 11670 }, { "epoch": 69.11242603550296, "grad_norm": 1.8412562608718872, "learning_rate": 0.0001, "loss": 0.0281, "step": 11680 }, { "epoch": 69.17159763313609, "grad_norm": 1.8923451900482178, "learning_rate": 0.0001, "loss": 0.0307, "step": 11690 }, { "epoch": 69.23076923076923, "grad_norm": 1.889294981956482, "learning_rate": 0.0001, "loss": 0.0299, "step": 11700 }, { "epoch": 69.28994082840237, "grad_norm": 2.372689962387085, "learning_rate": 0.0001, "loss": 0.0304, "step": 11710 }, { "epoch": 69.3491124260355, "grad_norm": 1.8316903114318848, "learning_rate": 0.0001, "loss": 0.0317, "step": 11720 }, { "epoch": 69.40828402366864, "grad_norm": 2.3629822731018066, "learning_rate": 0.0001, "loss": 0.0297, "step": 11730 }, { "epoch": 69.46745562130178, "grad_norm": 1.7951258420944214, "learning_rate": 0.0001, "loss": 0.0295, "step": 11740 }, { "epoch": 69.5266272189349, "grad_norm": 2.1720645427703857, "learning_rate": 0.0001, "loss": 0.0299, "step": 11750 }, { "epoch": 69.58579881656804, "grad_norm": 1.8459903001785278, "learning_rate": 0.0001, "loss": 0.0298, "step": 11760 }, { "epoch": 69.64497041420118, "grad_norm": 1.4678620100021362, "learning_rate": 0.0001, "loss": 0.0302, "step": 11770 }, { "epoch": 69.70414201183432, "grad_norm": 1.8121380805969238, "learning_rate": 0.0001, "loss": 0.0309, "step": 11780 }, { "epoch": 69.76331360946746, "grad_norm": 1.8132323026657104, "learning_rate": 0.0001, "loss": 0.0304, "step": 11790 }, { "epoch": 69.8224852071006, "grad_norm": 1.985331654548645, "learning_rate": 0.0001, "loss": 0.0316, "step": 11800 }, { "epoch": 69.88165680473372, "grad_norm": 2.293292760848999, "learning_rate": 0.0001, "loss": 0.0314, "step": 11810 }, { "epoch": 69.94082840236686, "grad_norm": 2.1988415718078613, "learning_rate": 0.0001, "loss": 0.0303, "step": 11820 }, { "epoch": 70.0, "grad_norm": 2.400019407272339, "learning_rate": 0.0001, "loss": 0.0302, "step": 11830 }, { "epoch": 70.05917159763314, "grad_norm": 2.2668423652648926, "learning_rate": 0.0001, "loss": 0.0294, "step": 11840 }, { "epoch": 70.11834319526628, "grad_norm": 1.7891758680343628, "learning_rate": 0.0001, "loss": 0.0283, "step": 11850 }, { "epoch": 70.1775147928994, "grad_norm": 1.7806328535079956, "learning_rate": 0.0001, "loss": 0.0293, "step": 11860 }, { "epoch": 70.23668639053254, "grad_norm": 1.5784519910812378, "learning_rate": 0.0001, "loss": 0.029, "step": 11870 }, { "epoch": 70.29585798816568, "grad_norm": 1.9202901124954224, "learning_rate": 0.0001, "loss": 0.0293, "step": 11880 }, { "epoch": 70.35502958579882, "grad_norm": 1.8924599885940552, "learning_rate": 0.0001, "loss": 0.0294, "step": 11890 }, { "epoch": 70.41420118343196, "grad_norm": 1.8414102792739868, "learning_rate": 0.0001, "loss": 0.0296, "step": 11900 }, { "epoch": 70.4733727810651, "grad_norm": 1.3567508459091187, "learning_rate": 0.0001, "loss": 0.0301, "step": 11910 }, { "epoch": 70.53254437869822, "grad_norm": 1.3349461555480957, "learning_rate": 0.0001, "loss": 0.0292, "step": 11920 }, { "epoch": 70.59171597633136, "grad_norm": 1.5805789232254028, "learning_rate": 0.0001, "loss": 0.0308, "step": 11930 }, { "epoch": 70.6508875739645, "grad_norm": 1.268980622291565, "learning_rate": 0.0001, "loss": 0.0296, "step": 11940 }, { "epoch": 70.71005917159763, "grad_norm": 1.6477490663528442, "learning_rate": 0.0001, "loss": 0.0301, "step": 11950 }, { "epoch": 70.76923076923077, "grad_norm": 1.6105555295944214, "learning_rate": 0.0001, "loss": 0.0322, "step": 11960 }, { "epoch": 70.82840236686391, "grad_norm": 2.200436592102051, "learning_rate": 0.0001, "loss": 0.0313, "step": 11970 }, { "epoch": 70.88757396449704, "grad_norm": 1.9743613004684448, "learning_rate": 0.0001, "loss": 0.0303, "step": 11980 }, { "epoch": 70.94674556213018, "grad_norm": 1.9796311855316162, "learning_rate": 0.0001, "loss": 0.0309, "step": 11990 }, { "epoch": 71.00591715976331, "grad_norm": 2.362837076187134, "learning_rate": 0.0001, "loss": 0.0312, "step": 12000 }, { "epoch": 71.06508875739645, "grad_norm": 1.5439103841781616, "learning_rate": 0.0001, "loss": 0.0308, "step": 12010 }, { "epoch": 71.12426035502959, "grad_norm": 1.6639682054519653, "learning_rate": 0.0001, "loss": 0.0308, "step": 12020 }, { "epoch": 71.18343195266272, "grad_norm": 2.2812182903289795, "learning_rate": 0.0001, "loss": 0.0304, "step": 12030 }, { "epoch": 71.24260355029585, "grad_norm": 1.551010251045227, "learning_rate": 0.0001, "loss": 0.0305, "step": 12040 }, { "epoch": 71.30177514792899, "grad_norm": 1.9372248649597168, "learning_rate": 0.0001, "loss": 0.0295, "step": 12050 }, { "epoch": 71.36094674556213, "grad_norm": 1.8353811502456665, "learning_rate": 0.0001, "loss": 0.0312, "step": 12060 }, { "epoch": 71.42011834319527, "grad_norm": 1.7521955966949463, "learning_rate": 0.0001, "loss": 0.031, "step": 12070 }, { "epoch": 71.47928994082841, "grad_norm": 1.9031767845153809, "learning_rate": 0.0001, "loss": 0.0301, "step": 12080 }, { "epoch": 71.53846153846153, "grad_norm": 2.1645452976226807, "learning_rate": 0.0001, "loss": 0.0298, "step": 12090 }, { "epoch": 71.59763313609467, "grad_norm": 1.4442960023880005, "learning_rate": 0.0001, "loss": 0.0298, "step": 12100 }, { "epoch": 71.65680473372781, "grad_norm": 1.6947964429855347, "learning_rate": 0.0001, "loss": 0.0306, "step": 12110 }, { "epoch": 71.71597633136095, "grad_norm": 1.6309479475021362, "learning_rate": 0.0001, "loss": 0.0303, "step": 12120 }, { "epoch": 71.77514792899409, "grad_norm": 1.5988664627075195, "learning_rate": 0.0001, "loss": 0.0315, "step": 12130 }, { "epoch": 71.83431952662721, "grad_norm": 2.105926275253296, "learning_rate": 0.0001, "loss": 0.0306, "step": 12140 }, { "epoch": 71.89349112426035, "grad_norm": 1.663974642753601, "learning_rate": 0.0001, "loss": 0.0317, "step": 12150 }, { "epoch": 71.95266272189349, "grad_norm": 1.6369376182556152, "learning_rate": 0.0001, "loss": 0.0311, "step": 12160 }, { "epoch": 72.01183431952663, "grad_norm": 1.464241623878479, "learning_rate": 0.0001, "loss": 0.0316, "step": 12170 }, { "epoch": 72.07100591715977, "grad_norm": 1.6406829357147217, "learning_rate": 0.0001, "loss": 0.0315, "step": 12180 }, { "epoch": 72.1301775147929, "grad_norm": 1.9081898927688599, "learning_rate": 0.0001, "loss": 0.0299, "step": 12190 }, { "epoch": 72.18934911242603, "grad_norm": 1.7045892477035522, "learning_rate": 0.0001, "loss": 0.0304, "step": 12200 }, { "epoch": 72.24852071005917, "grad_norm": 1.7533644437789917, "learning_rate": 0.0001, "loss": 0.0298, "step": 12210 }, { "epoch": 72.3076923076923, "grad_norm": 1.2213021516799927, "learning_rate": 0.0001, "loss": 0.0301, "step": 12220 }, { "epoch": 72.36686390532545, "grad_norm": 1.6063940525054932, "learning_rate": 0.0001, "loss": 0.03, "step": 12230 }, { "epoch": 72.42603550295858, "grad_norm": 2.637014389038086, "learning_rate": 0.0001, "loss": 0.0315, "step": 12240 }, { "epoch": 72.48520710059172, "grad_norm": 1.8713611364364624, "learning_rate": 0.0001, "loss": 0.0337, "step": 12250 }, { "epoch": 72.54437869822485, "grad_norm": 1.9296704530715942, "learning_rate": 0.0001, "loss": 0.0305, "step": 12260 }, { "epoch": 72.60355029585799, "grad_norm": 1.5642050504684448, "learning_rate": 0.0001, "loss": 0.0293, "step": 12270 }, { "epoch": 72.66272189349112, "grad_norm": 2.1466023921966553, "learning_rate": 0.0001, "loss": 0.0291, "step": 12280 }, { "epoch": 72.72189349112426, "grad_norm": 2.1915152072906494, "learning_rate": 0.0001, "loss": 0.029, "step": 12290 }, { "epoch": 72.7810650887574, "grad_norm": 1.6483856439590454, "learning_rate": 0.0001, "loss": 0.03, "step": 12300 }, { "epoch": 72.84023668639053, "grad_norm": 1.5721471309661865, "learning_rate": 0.0001, "loss": 0.0301, "step": 12310 }, { "epoch": 72.89940828402366, "grad_norm": 1.584708333015442, "learning_rate": 0.0001, "loss": 0.0312, "step": 12320 }, { "epoch": 72.9585798816568, "grad_norm": 1.6179853677749634, "learning_rate": 0.0001, "loss": 0.0278, "step": 12330 }, { "epoch": 73.01775147928994, "grad_norm": 1.6778215169906616, "learning_rate": 0.0001, "loss": 0.0298, "step": 12340 }, { "epoch": 73.07692307692308, "grad_norm": 2.002941131591797, "learning_rate": 0.0001, "loss": 0.0281, "step": 12350 }, { "epoch": 73.13609467455622, "grad_norm": 1.7035313844680786, "learning_rate": 0.0001, "loss": 0.0306, "step": 12360 }, { "epoch": 73.19526627218934, "grad_norm": 2.0504090785980225, "learning_rate": 0.0001, "loss": 0.0303, "step": 12370 }, { "epoch": 73.25443786982248, "grad_norm": 1.632315993309021, "learning_rate": 0.0001, "loss": 0.0292, "step": 12380 }, { "epoch": 73.31360946745562, "grad_norm": 1.794150710105896, "learning_rate": 0.0001, "loss": 0.0297, "step": 12390 }, { "epoch": 73.37278106508876, "grad_norm": 1.9577139616012573, "learning_rate": 0.0001, "loss": 0.0295, "step": 12400 }, { "epoch": 73.4319526627219, "grad_norm": 2.077632427215576, "learning_rate": 0.0001, "loss": 0.03, "step": 12410 }, { "epoch": 73.49112426035504, "grad_norm": 2.118983268737793, "learning_rate": 0.0001, "loss": 0.0303, "step": 12420 }, { "epoch": 73.55029585798816, "grad_norm": 1.9092583656311035, "learning_rate": 0.0001, "loss": 0.0299, "step": 12430 }, { "epoch": 73.6094674556213, "grad_norm": 1.6623069047927856, "learning_rate": 0.0001, "loss": 0.029, "step": 12440 }, { "epoch": 73.66863905325444, "grad_norm": 1.8864250183105469, "learning_rate": 0.0001, "loss": 0.0273, "step": 12450 }, { "epoch": 73.72781065088758, "grad_norm": 1.4074801206588745, "learning_rate": 0.0001, "loss": 0.0287, "step": 12460 }, { "epoch": 73.78698224852072, "grad_norm": 1.4596788883209229, "learning_rate": 0.0001, "loss": 0.0282, "step": 12470 }, { "epoch": 73.84615384615384, "grad_norm": 1.370743989944458, "learning_rate": 0.0001, "loss": 0.0289, "step": 12480 }, { "epoch": 73.90532544378698, "grad_norm": 1.8263006210327148, "learning_rate": 0.0001, "loss": 0.0297, "step": 12490 }, { "epoch": 73.96449704142012, "grad_norm": 1.8158409595489502, "learning_rate": 0.0001, "loss": 0.0285, "step": 12500 }, { "epoch": 74.02366863905326, "grad_norm": 1.7067906856536865, "learning_rate": 0.0001, "loss": 0.0291, "step": 12510 }, { "epoch": 74.0828402366864, "grad_norm": 2.593698263168335, "learning_rate": 0.0001, "loss": 0.0296, "step": 12520 }, { "epoch": 74.14201183431953, "grad_norm": 2.4969241619110107, "learning_rate": 0.0001, "loss": 0.0304, "step": 12530 }, { "epoch": 74.20118343195266, "grad_norm": 2.07260799407959, "learning_rate": 0.0001, "loss": 0.0305, "step": 12540 }, { "epoch": 74.2603550295858, "grad_norm": 2.015178680419922, "learning_rate": 0.0001, "loss": 0.0297, "step": 12550 }, { "epoch": 74.31952662721893, "grad_norm": 2.173203945159912, "learning_rate": 0.0001, "loss": 0.0302, "step": 12560 }, { "epoch": 74.37869822485207, "grad_norm": 1.8017369508743286, "learning_rate": 0.0001, "loss": 0.0304, "step": 12570 }, { "epoch": 74.43786982248521, "grad_norm": 2.094944477081299, "learning_rate": 0.0001, "loss": 0.0312, "step": 12580 }, { "epoch": 74.49704142011835, "grad_norm": 2.2708051204681396, "learning_rate": 0.0001, "loss": 0.0309, "step": 12590 }, { "epoch": 74.55621301775147, "grad_norm": 2.1914515495300293, "learning_rate": 0.0001, "loss": 0.0318, "step": 12600 }, { "epoch": 74.61538461538461, "grad_norm": 2.024599075317383, "learning_rate": 0.0001, "loss": 0.0304, "step": 12610 }, { "epoch": 74.67455621301775, "grad_norm": 1.9934560060501099, "learning_rate": 0.0001, "loss": 0.0301, "step": 12620 }, { "epoch": 74.73372781065089, "grad_norm": 1.5091840028762817, "learning_rate": 0.0001, "loss": 0.0299, "step": 12630 }, { "epoch": 74.79289940828403, "grad_norm": 1.7699127197265625, "learning_rate": 0.0001, "loss": 0.0303, "step": 12640 }, { "epoch": 74.85207100591715, "grad_norm": 1.8536765575408936, "learning_rate": 0.0001, "loss": 0.0294, "step": 12650 }, { "epoch": 74.91124260355029, "grad_norm": 2.575226068496704, "learning_rate": 0.0001, "loss": 0.0297, "step": 12660 }, { "epoch": 74.97041420118343, "grad_norm": 1.2151219844818115, "learning_rate": 0.0001, "loss": 0.0295, "step": 12670 }, { "epoch": 75.02958579881657, "grad_norm": 2.640342950820923, "learning_rate": 0.0001, "loss": 0.0281, "step": 12680 }, { "epoch": 75.08875739644971, "grad_norm": 1.3766578435897827, "learning_rate": 0.0001, "loss": 0.0285, "step": 12690 }, { "epoch": 75.14792899408285, "grad_norm": 1.787416934967041, "learning_rate": 0.0001, "loss": 0.0287, "step": 12700 }, { "epoch": 75.20710059171597, "grad_norm": 1.252223014831543, "learning_rate": 0.0001, "loss": 0.0283, "step": 12710 }, { "epoch": 75.26627218934911, "grad_norm": 2.238722085952759, "learning_rate": 0.0001, "loss": 0.0291, "step": 12720 }, { "epoch": 75.32544378698225, "grad_norm": 1.8951098918914795, "learning_rate": 0.0001, "loss": 0.0293, "step": 12730 }, { "epoch": 75.38461538461539, "grad_norm": 1.652716875076294, "learning_rate": 0.0001, "loss": 0.0286, "step": 12740 }, { "epoch": 75.44378698224853, "grad_norm": 1.8918410539627075, "learning_rate": 0.0001, "loss": 0.0298, "step": 12750 }, { "epoch": 75.50295857988165, "grad_norm": 1.8305774927139282, "learning_rate": 0.0001, "loss": 0.0309, "step": 12760 }, { "epoch": 75.56213017751479, "grad_norm": 1.6599955558776855, "learning_rate": 0.0001, "loss": 0.0298, "step": 12770 }, { "epoch": 75.62130177514793, "grad_norm": 1.7720906734466553, "learning_rate": 0.0001, "loss": 0.0299, "step": 12780 }, { "epoch": 75.68047337278107, "grad_norm": 1.3876029253005981, "learning_rate": 0.0001, "loss": 0.0293, "step": 12790 }, { "epoch": 75.7396449704142, "grad_norm": 1.4683743715286255, "learning_rate": 0.0001, "loss": 0.0287, "step": 12800 }, { "epoch": 75.79881656804734, "grad_norm": 1.504238486289978, "learning_rate": 0.0001, "loss": 0.0284, "step": 12810 }, { "epoch": 75.85798816568047, "grad_norm": 1.5836963653564453, "learning_rate": 0.0001, "loss": 0.0292, "step": 12820 }, { "epoch": 75.9171597633136, "grad_norm": 1.5867384672164917, "learning_rate": 0.0001, "loss": 0.0314, "step": 12830 }, { "epoch": 75.97633136094674, "grad_norm": 1.650313377380371, "learning_rate": 0.0001, "loss": 0.0309, "step": 12840 }, { "epoch": 76.03550295857988, "grad_norm": 2.0597691535949707, "learning_rate": 0.0001, "loss": 0.0315, "step": 12850 }, { "epoch": 76.09467455621302, "grad_norm": 1.542629361152649, "learning_rate": 0.0001, "loss": 0.0309, "step": 12860 }, { "epoch": 76.15384615384616, "grad_norm": 2.1849324703216553, "learning_rate": 0.0001, "loss": 0.0292, "step": 12870 }, { "epoch": 76.21301775147928, "grad_norm": 1.787174105644226, "learning_rate": 0.0001, "loss": 0.0305, "step": 12880 }, { "epoch": 76.27218934911242, "grad_norm": 2.3207204341888428, "learning_rate": 0.0001, "loss": 0.0307, "step": 12890 }, { "epoch": 76.33136094674556, "grad_norm": 1.6957664489746094, "learning_rate": 0.0001, "loss": 0.0304, "step": 12900 }, { "epoch": 76.3905325443787, "grad_norm": 1.5988353490829468, "learning_rate": 0.0001, "loss": 0.0287, "step": 12910 }, { "epoch": 76.44970414201184, "grad_norm": 1.3811252117156982, "learning_rate": 0.0001, "loss": 0.0301, "step": 12920 }, { "epoch": 76.50887573964496, "grad_norm": 1.732063889503479, "learning_rate": 0.0001, "loss": 0.0286, "step": 12930 }, { "epoch": 76.5680473372781, "grad_norm": 1.6476237773895264, "learning_rate": 0.0001, "loss": 0.0287, "step": 12940 }, { "epoch": 76.62721893491124, "grad_norm": 1.6677615642547607, "learning_rate": 0.0001, "loss": 0.0282, "step": 12950 }, { "epoch": 76.68639053254438, "grad_norm": 1.5645583868026733, "learning_rate": 0.0001, "loss": 0.0308, "step": 12960 }, { "epoch": 76.74556213017752, "grad_norm": 1.5712158679962158, "learning_rate": 0.0001, "loss": 0.0303, "step": 12970 }, { "epoch": 76.80473372781066, "grad_norm": 1.5865414142608643, "learning_rate": 0.0001, "loss": 0.0299, "step": 12980 }, { "epoch": 76.86390532544378, "grad_norm": 1.8486440181732178, "learning_rate": 0.0001, "loss": 0.0287, "step": 12990 }, { "epoch": 76.92307692307692, "grad_norm": 1.8056178092956543, "learning_rate": 0.0001, "loss": 0.0296, "step": 13000 }, { "epoch": 76.98224852071006, "grad_norm": 1.5240130424499512, "learning_rate": 0.0001, "loss": 0.0302, "step": 13010 }, { "epoch": 77.0414201183432, "grad_norm": 1.337066650390625, "learning_rate": 0.0001, "loss": 0.0295, "step": 13020 }, { "epoch": 77.10059171597634, "grad_norm": 1.6543998718261719, "learning_rate": 0.0001, "loss": 0.0308, "step": 13030 }, { "epoch": 77.15976331360947, "grad_norm": 1.5613868236541748, "learning_rate": 0.0001, "loss": 0.0293, "step": 13040 }, { "epoch": 77.2189349112426, "grad_norm": 1.8683534860610962, "learning_rate": 0.0001, "loss": 0.0304, "step": 13050 }, { "epoch": 77.27810650887574, "grad_norm": 1.3576754331588745, "learning_rate": 0.0001, "loss": 0.0291, "step": 13060 }, { "epoch": 77.33727810650888, "grad_norm": 1.4709784984588623, "learning_rate": 0.0001, "loss": 0.0294, "step": 13070 }, { "epoch": 77.39644970414201, "grad_norm": 2.007960081100464, "learning_rate": 0.0001, "loss": 0.0294, "step": 13080 }, { "epoch": 77.45562130177515, "grad_norm": 2.1055684089660645, "learning_rate": 0.0001, "loss": 0.0318, "step": 13090 }, { "epoch": 77.51479289940828, "grad_norm": 1.3696526288986206, "learning_rate": 0.0001, "loss": 0.0289, "step": 13100 }, { "epoch": 77.57396449704142, "grad_norm": 1.7728135585784912, "learning_rate": 0.0001, "loss": 0.029, "step": 13110 }, { "epoch": 77.63313609467455, "grad_norm": 1.9723602533340454, "learning_rate": 0.0001, "loss": 0.029, "step": 13120 }, { "epoch": 77.6923076923077, "grad_norm": 2.215200901031494, "learning_rate": 0.0001, "loss": 0.03, "step": 13130 }, { "epoch": 77.75147928994083, "grad_norm": 1.594514012336731, "learning_rate": 0.0001, "loss": 0.029, "step": 13140 }, { "epoch": 77.81065088757397, "grad_norm": 1.7701891660690308, "learning_rate": 0.0001, "loss": 0.0286, "step": 13150 }, { "epoch": 77.8698224852071, "grad_norm": 2.200824022293091, "learning_rate": 0.0001, "loss": 0.0293, "step": 13160 }, { "epoch": 77.92899408284023, "grad_norm": 2.1316826343536377, "learning_rate": 0.0001, "loss": 0.0297, "step": 13170 }, { "epoch": 77.98816568047337, "grad_norm": 1.7729662656784058, "learning_rate": 0.0001, "loss": 0.0277, "step": 13180 }, { "epoch": 78.04733727810651, "grad_norm": 1.574045181274414, "learning_rate": 0.0001, "loss": 0.0282, "step": 13190 }, { "epoch": 78.10650887573965, "grad_norm": 1.8935548067092896, "learning_rate": 0.0001, "loss": 0.0278, "step": 13200 }, { "epoch": 78.16568047337279, "grad_norm": 2.188845634460449, "learning_rate": 0.0001, "loss": 0.028, "step": 13210 }, { "epoch": 78.22485207100591, "grad_norm": 1.5271364450454712, "learning_rate": 0.0001, "loss": 0.0283, "step": 13220 }, { "epoch": 78.28402366863905, "grad_norm": 1.709799885749817, "learning_rate": 0.0001, "loss": 0.0278, "step": 13230 }, { "epoch": 78.34319526627219, "grad_norm": 1.8126262426376343, "learning_rate": 0.0001, "loss": 0.0267, "step": 13240 }, { "epoch": 78.40236686390533, "grad_norm": 1.7274236679077148, "learning_rate": 0.0001, "loss": 0.029, "step": 13250 }, { "epoch": 78.46153846153847, "grad_norm": 1.9450957775115967, "learning_rate": 0.0001, "loss": 0.0286, "step": 13260 }, { "epoch": 78.52071005917159, "grad_norm": 1.5503954887390137, "learning_rate": 0.0001, "loss": 0.0276, "step": 13270 }, { "epoch": 78.57988165680473, "grad_norm": 1.7177788019180298, "learning_rate": 0.0001, "loss": 0.0272, "step": 13280 }, { "epoch": 78.63905325443787, "grad_norm": 2.198521852493286, "learning_rate": 0.0001, "loss": 0.0286, "step": 13290 }, { "epoch": 78.69822485207101, "grad_norm": 1.7270867824554443, "learning_rate": 0.0001, "loss": 0.0291, "step": 13300 }, { "epoch": 78.75739644970415, "grad_norm": 1.4182158708572388, "learning_rate": 0.0001, "loss": 0.0285, "step": 13310 }, { "epoch": 78.81656804733728, "grad_norm": 2.0431549549102783, "learning_rate": 0.0001, "loss": 0.0297, "step": 13320 }, { "epoch": 78.87573964497041, "grad_norm": 1.3427083492279053, "learning_rate": 0.0001, "loss": 0.0298, "step": 13330 }, { "epoch": 78.93491124260355, "grad_norm": 1.7023308277130127, "learning_rate": 0.0001, "loss": 0.0294, "step": 13340 }, { "epoch": 78.99408284023669, "grad_norm": 1.6448678970336914, "learning_rate": 0.0001, "loss": 0.0287, "step": 13350 }, { "epoch": 79.05325443786982, "grad_norm": 1.3475834131240845, "learning_rate": 0.0001, "loss": 0.0286, "step": 13360 }, { "epoch": 79.11242603550296, "grad_norm": 1.5937806367874146, "learning_rate": 0.0001, "loss": 0.0299, "step": 13370 }, { "epoch": 79.17159763313609, "grad_norm": 1.7989280223846436, "learning_rate": 0.0001, "loss": 0.0283, "step": 13380 }, { "epoch": 79.23076923076923, "grad_norm": 1.396798014640808, "learning_rate": 0.0001, "loss": 0.0291, "step": 13390 }, { "epoch": 79.28994082840237, "grad_norm": 1.7228373289108276, "learning_rate": 0.0001, "loss": 0.0307, "step": 13400 }, { "epoch": 79.3491124260355, "grad_norm": 2.07952618598938, "learning_rate": 0.0001, "loss": 0.0313, "step": 13410 }, { "epoch": 79.40828402366864, "grad_norm": 1.828514814376831, "learning_rate": 0.0001, "loss": 0.0301, "step": 13420 }, { "epoch": 79.46745562130178, "grad_norm": 1.8261467218399048, "learning_rate": 0.0001, "loss": 0.0322, "step": 13430 }, { "epoch": 79.5266272189349, "grad_norm": 1.5058075189590454, "learning_rate": 0.0001, "loss": 0.0289, "step": 13440 }, { "epoch": 79.58579881656804, "grad_norm": 1.8194470405578613, "learning_rate": 0.0001, "loss": 0.0307, "step": 13450 }, { "epoch": 79.64497041420118, "grad_norm": 1.9430017471313477, "learning_rate": 0.0001, "loss": 0.0291, "step": 13460 }, { "epoch": 79.70414201183432, "grad_norm": 2.4839885234832764, "learning_rate": 0.0001, "loss": 0.0292, "step": 13470 }, { "epoch": 79.76331360946746, "grad_norm": 2.0954408645629883, "learning_rate": 0.0001, "loss": 0.0292, "step": 13480 }, { "epoch": 79.8224852071006, "grad_norm": 1.4980049133300781, "learning_rate": 0.0001, "loss": 0.0287, "step": 13490 }, { "epoch": 79.88165680473372, "grad_norm": 1.7274476289749146, "learning_rate": 0.0001, "loss": 0.027, "step": 13500 }, { "epoch": 79.94082840236686, "grad_norm": 1.6477634906768799, "learning_rate": 0.0001, "loss": 0.0277, "step": 13510 }, { "epoch": 80.0, "grad_norm": 1.6456685066223145, "learning_rate": 0.0001, "loss": 0.0276, "step": 13520 }, { "epoch": 80.05917159763314, "grad_norm": 1.9690388441085815, "learning_rate": 0.0001, "loss": 0.0285, "step": 13530 }, { "epoch": 80.11834319526628, "grad_norm": 1.6115995645523071, "learning_rate": 0.0001, "loss": 0.0283, "step": 13540 }, { "epoch": 80.1775147928994, "grad_norm": 1.5597548484802246, "learning_rate": 0.0001, "loss": 0.028, "step": 13550 }, { "epoch": 80.23668639053254, "grad_norm": 1.4994943141937256, "learning_rate": 0.0001, "loss": 0.0275, "step": 13560 }, { "epoch": 80.29585798816568, "grad_norm": 1.6371421813964844, "learning_rate": 0.0001, "loss": 0.0283, "step": 13570 }, { "epoch": 80.35502958579882, "grad_norm": 1.5609456300735474, "learning_rate": 0.0001, "loss": 0.0274, "step": 13580 }, { "epoch": 80.41420118343196, "grad_norm": 1.927188754081726, "learning_rate": 0.0001, "loss": 0.0282, "step": 13590 }, { "epoch": 80.4733727810651, "grad_norm": 1.5500059127807617, "learning_rate": 0.0001, "loss": 0.0309, "step": 13600 }, { "epoch": 80.53254437869822, "grad_norm": 1.844635009765625, "learning_rate": 0.0001, "loss": 0.0284, "step": 13610 }, { "epoch": 80.59171597633136, "grad_norm": 2.3151895999908447, "learning_rate": 0.0001, "loss": 0.0295, "step": 13620 }, { "epoch": 80.6508875739645, "grad_norm": 1.9789940118789673, "learning_rate": 0.0001, "loss": 0.0295, "step": 13630 }, { "epoch": 80.71005917159763, "grad_norm": 2.3001701831817627, "learning_rate": 0.0001, "loss": 0.0298, "step": 13640 }, { "epoch": 80.76923076923077, "grad_norm": 1.5195906162261963, "learning_rate": 0.0001, "loss": 0.0292, "step": 13650 }, { "epoch": 80.82840236686391, "grad_norm": 2.4520936012268066, "learning_rate": 0.0001, "loss": 0.0277, "step": 13660 }, { "epoch": 80.88757396449704, "grad_norm": 1.895772099494934, "learning_rate": 0.0001, "loss": 0.0281, "step": 13670 }, { "epoch": 80.94674556213018, "grad_norm": 1.409948468208313, "learning_rate": 0.0001, "loss": 0.0264, "step": 13680 }, { "epoch": 81.00591715976331, "grad_norm": 1.6955418586730957, "learning_rate": 0.0001, "loss": 0.0283, "step": 13690 }, { "epoch": 81.06508875739645, "grad_norm": 1.6297917366027832, "learning_rate": 0.0001, "loss": 0.0277, "step": 13700 }, { "epoch": 81.12426035502959, "grad_norm": 1.6393722295761108, "learning_rate": 0.0001, "loss": 0.0272, "step": 13710 }, { "epoch": 81.18343195266272, "grad_norm": 1.8834614753723145, "learning_rate": 0.0001, "loss": 0.0285, "step": 13720 }, { "epoch": 81.24260355029585, "grad_norm": 1.8427009582519531, "learning_rate": 0.0001, "loss": 0.028, "step": 13730 }, { "epoch": 81.30177514792899, "grad_norm": 1.2885452508926392, "learning_rate": 0.0001, "loss": 0.0278, "step": 13740 }, { "epoch": 81.36094674556213, "grad_norm": 1.3599108457565308, "learning_rate": 0.0001, "loss": 0.0295, "step": 13750 }, { "epoch": 81.42011834319527, "grad_norm": 1.5153740644454956, "learning_rate": 0.0001, "loss": 0.029, "step": 13760 }, { "epoch": 81.47928994082841, "grad_norm": 1.4208409786224365, "learning_rate": 0.0001, "loss": 0.0289, "step": 13770 }, { "epoch": 81.53846153846153, "grad_norm": 1.3228154182434082, "learning_rate": 0.0001, "loss": 0.0289, "step": 13780 }, { "epoch": 81.59763313609467, "grad_norm": 2.2498724460601807, "learning_rate": 0.0001, "loss": 0.03, "step": 13790 }, { "epoch": 81.65680473372781, "grad_norm": 2.2454402446746826, "learning_rate": 0.0001, "loss": 0.029, "step": 13800 }, { "epoch": 81.71597633136095, "grad_norm": 1.952128291130066, "learning_rate": 0.0001, "loss": 0.0281, "step": 13810 }, { "epoch": 81.77514792899409, "grad_norm": 1.5461188554763794, "learning_rate": 0.0001, "loss": 0.0283, "step": 13820 }, { "epoch": 81.83431952662721, "grad_norm": 2.115417957305908, "learning_rate": 0.0001, "loss": 0.0288, "step": 13830 }, { "epoch": 81.89349112426035, "grad_norm": 1.737291932106018, "learning_rate": 0.0001, "loss": 0.0269, "step": 13840 }, { "epoch": 81.95266272189349, "grad_norm": 1.7582436800003052, "learning_rate": 0.0001, "loss": 0.0279, "step": 13850 }, { "epoch": 82.01183431952663, "grad_norm": 1.6190907955169678, "learning_rate": 0.0001, "loss": 0.0285, "step": 13860 }, { "epoch": 82.07100591715977, "grad_norm": 1.7788535356521606, "learning_rate": 0.0001, "loss": 0.0288, "step": 13870 }, { "epoch": 82.1301775147929, "grad_norm": 1.7750798463821411, "learning_rate": 0.0001, "loss": 0.0289, "step": 13880 }, { "epoch": 82.18934911242603, "grad_norm": 1.513469934463501, "learning_rate": 0.0001, "loss": 0.0275, "step": 13890 }, { "epoch": 82.24852071005917, "grad_norm": 1.4091371297836304, "learning_rate": 0.0001, "loss": 0.0273, "step": 13900 }, { "epoch": 82.3076923076923, "grad_norm": 1.6552488803863525, "learning_rate": 0.0001, "loss": 0.0266, "step": 13910 }, { "epoch": 82.36686390532545, "grad_norm": 1.8825668096542358, "learning_rate": 0.0001, "loss": 0.0282, "step": 13920 }, { "epoch": 82.42603550295858, "grad_norm": 1.8455760478973389, "learning_rate": 0.0001, "loss": 0.0284, "step": 13930 }, { "epoch": 82.48520710059172, "grad_norm": 2.381021022796631, "learning_rate": 0.0001, "loss": 0.0308, "step": 13940 }, { "epoch": 82.54437869822485, "grad_norm": 1.2054821252822876, "learning_rate": 0.0001, "loss": 0.0285, "step": 13950 }, { "epoch": 82.60355029585799, "grad_norm": 1.5205910205841064, "learning_rate": 0.0001, "loss": 0.0283, "step": 13960 }, { "epoch": 82.66272189349112, "grad_norm": 1.9282556772232056, "learning_rate": 0.0001, "loss": 0.0291, "step": 13970 }, { "epoch": 82.72189349112426, "grad_norm": 1.7727046012878418, "learning_rate": 0.0001, "loss": 0.0295, "step": 13980 }, { "epoch": 82.7810650887574, "grad_norm": 1.866040825843811, "learning_rate": 0.0001, "loss": 0.0291, "step": 13990 }, { "epoch": 82.84023668639053, "grad_norm": 1.263092041015625, "learning_rate": 0.0001, "loss": 0.0271, "step": 14000 }, { "epoch": 82.89940828402366, "grad_norm": 1.7061164379119873, "learning_rate": 0.0001, "loss": 0.0277, "step": 14010 }, { "epoch": 82.9585798816568, "grad_norm": 2.0393588542938232, "learning_rate": 0.0001, "loss": 0.029, "step": 14020 }, { "epoch": 83.01775147928994, "grad_norm": 1.4109842777252197, "learning_rate": 0.0001, "loss": 0.0288, "step": 14030 }, { "epoch": 83.07692307692308, "grad_norm": 1.232629418373108, "learning_rate": 0.0001, "loss": 0.0291, "step": 14040 }, { "epoch": 83.13609467455622, "grad_norm": 1.2593978643417358, "learning_rate": 0.0001, "loss": 0.03, "step": 14050 }, { "epoch": 83.19526627218934, "grad_norm": 1.3894357681274414, "learning_rate": 0.0001, "loss": 0.0282, "step": 14060 }, { "epoch": 83.25443786982248, "grad_norm": 1.6489630937576294, "learning_rate": 0.0001, "loss": 0.0287, "step": 14070 }, { "epoch": 83.31360946745562, "grad_norm": 1.464470624923706, "learning_rate": 0.0001, "loss": 0.0296, "step": 14080 }, { "epoch": 83.37278106508876, "grad_norm": 1.0723363161087036, "learning_rate": 0.0001, "loss": 0.0306, "step": 14090 }, { "epoch": 83.4319526627219, "grad_norm": 1.8792723417282104, "learning_rate": 0.0001, "loss": 0.0315, "step": 14100 }, { "epoch": 83.49112426035504, "grad_norm": 1.9182002544403076, "learning_rate": 0.0001, "loss": 0.029, "step": 14110 }, { "epoch": 83.55029585798816, "grad_norm": 1.6577177047729492, "learning_rate": 0.0001, "loss": 0.0293, "step": 14120 }, { "epoch": 83.6094674556213, "grad_norm": 2.364114761352539, "learning_rate": 0.0001, "loss": 0.0285, "step": 14130 }, { "epoch": 83.66863905325444, "grad_norm": 1.4208087921142578, "learning_rate": 0.0001, "loss": 0.0286, "step": 14140 }, { "epoch": 83.72781065088758, "grad_norm": 1.6258130073547363, "learning_rate": 0.0001, "loss": 0.0288, "step": 14150 }, { "epoch": 83.78698224852072, "grad_norm": 1.3894916772842407, "learning_rate": 0.0001, "loss": 0.0297, "step": 14160 }, { "epoch": 83.84615384615384, "grad_norm": 1.7538460493087769, "learning_rate": 0.0001, "loss": 0.0291, "step": 14170 }, { "epoch": 83.90532544378698, "grad_norm": 1.6357338428497314, "learning_rate": 0.0001, "loss": 0.0281, "step": 14180 }, { "epoch": 83.96449704142012, "grad_norm": 1.7782138586044312, "learning_rate": 0.0001, "loss": 0.0285, "step": 14190 }, { "epoch": 84.02366863905326, "grad_norm": 1.9577884674072266, "learning_rate": 0.0001, "loss": 0.0282, "step": 14200 }, { "epoch": 84.0828402366864, "grad_norm": 1.1637879610061646, "learning_rate": 0.0001, "loss": 0.0292, "step": 14210 }, { "epoch": 84.14201183431953, "grad_norm": 1.745080828666687, "learning_rate": 0.0001, "loss": 0.0296, "step": 14220 }, { "epoch": 84.20118343195266, "grad_norm": 1.9295228719711304, "learning_rate": 0.0001, "loss": 0.0303, "step": 14230 }, { "epoch": 84.2603550295858, "grad_norm": 1.8617806434631348, "learning_rate": 0.0001, "loss": 0.028, "step": 14240 }, { "epoch": 84.31952662721893, "grad_norm": 1.3378702402114868, "learning_rate": 0.0001, "loss": 0.0289, "step": 14250 }, { "epoch": 84.37869822485207, "grad_norm": 1.6640417575836182, "learning_rate": 0.0001, "loss": 0.0288, "step": 14260 }, { "epoch": 84.43786982248521, "grad_norm": 1.9496287107467651, "learning_rate": 0.0001, "loss": 0.0285, "step": 14270 }, { "epoch": 84.49704142011835, "grad_norm": 1.5225979089736938, "learning_rate": 0.0001, "loss": 0.029, "step": 14280 }, { "epoch": 84.55621301775147, "grad_norm": 1.827197551727295, "learning_rate": 0.0001, "loss": 0.0278, "step": 14290 }, { "epoch": 84.61538461538461, "grad_norm": 1.6674708127975464, "learning_rate": 0.0001, "loss": 0.0272, "step": 14300 }, { "epoch": 84.67455621301775, "grad_norm": 1.6737569570541382, "learning_rate": 0.0001, "loss": 0.0279, "step": 14310 }, { "epoch": 84.73372781065089, "grad_norm": 1.8727366924285889, "learning_rate": 0.0001, "loss": 0.0293, "step": 14320 }, { "epoch": 84.79289940828403, "grad_norm": 1.5261051654815674, "learning_rate": 0.0001, "loss": 0.0287, "step": 14330 }, { "epoch": 84.85207100591715, "grad_norm": 2.490265130996704, "learning_rate": 0.0001, "loss": 0.0285, "step": 14340 }, { "epoch": 84.91124260355029, "grad_norm": 2.421513795852661, "learning_rate": 0.0001, "loss": 0.0311, "step": 14350 }, { "epoch": 84.97041420118343, "grad_norm": 1.7775557041168213, "learning_rate": 0.0001, "loss": 0.028, "step": 14360 }, { "epoch": 85.02958579881657, "grad_norm": 1.5255711078643799, "learning_rate": 0.0001, "loss": 0.0282, "step": 14370 }, { "epoch": 85.08875739644971, "grad_norm": 1.2676481008529663, "learning_rate": 0.0001, "loss": 0.0275, "step": 14380 }, { "epoch": 85.14792899408285, "grad_norm": 2.041949510574341, "learning_rate": 0.0001, "loss": 0.0296, "step": 14390 }, { "epoch": 85.20710059171597, "grad_norm": 2.0712430477142334, "learning_rate": 0.0001, "loss": 0.0292, "step": 14400 }, { "epoch": 85.26627218934911, "grad_norm": 1.5517901182174683, "learning_rate": 0.0001, "loss": 0.0288, "step": 14410 }, { "epoch": 85.32544378698225, "grad_norm": 1.3186544179916382, "learning_rate": 0.0001, "loss": 0.0275, "step": 14420 }, { "epoch": 85.38461538461539, "grad_norm": 1.8532809019088745, "learning_rate": 0.0001, "loss": 0.0266, "step": 14430 }, { "epoch": 85.44378698224853, "grad_norm": 1.1717849969863892, "learning_rate": 0.0001, "loss": 0.0282, "step": 14440 }, { "epoch": 85.50295857988165, "grad_norm": 1.1426750421524048, "learning_rate": 0.0001, "loss": 0.0279, "step": 14450 }, { "epoch": 85.56213017751479, "grad_norm": 1.705265998840332, "learning_rate": 0.0001, "loss": 0.0277, "step": 14460 }, { "epoch": 85.62130177514793, "grad_norm": 1.6591438055038452, "learning_rate": 0.0001, "loss": 0.0287, "step": 14470 }, { "epoch": 85.68047337278107, "grad_norm": 1.8544031381607056, "learning_rate": 0.0001, "loss": 0.0279, "step": 14480 }, { "epoch": 85.7396449704142, "grad_norm": 1.53345787525177, "learning_rate": 0.0001, "loss": 0.0265, "step": 14490 }, { "epoch": 85.79881656804734, "grad_norm": 1.594197154045105, "learning_rate": 0.0001, "loss": 0.0278, "step": 14500 }, { "epoch": 85.85798816568047, "grad_norm": 1.1395113468170166, "learning_rate": 0.0001, "loss": 0.0283, "step": 14510 }, { "epoch": 85.9171597633136, "grad_norm": 1.7510695457458496, "learning_rate": 0.0001, "loss": 0.0289, "step": 14520 }, { "epoch": 85.97633136094674, "grad_norm": 1.4207432270050049, "learning_rate": 0.0001, "loss": 0.0289, "step": 14530 }, { "epoch": 86.03550295857988, "grad_norm": 1.8491264581680298, "learning_rate": 0.0001, "loss": 0.0291, "step": 14540 }, { "epoch": 86.09467455621302, "grad_norm": 1.9184045791625977, "learning_rate": 0.0001, "loss": 0.0304, "step": 14550 }, { "epoch": 86.15384615384616, "grad_norm": 1.3591728210449219, "learning_rate": 0.0001, "loss": 0.0284, "step": 14560 }, { "epoch": 86.21301775147928, "grad_norm": 1.1978782415390015, "learning_rate": 0.0001, "loss": 0.0292, "step": 14570 }, { "epoch": 86.27218934911242, "grad_norm": 1.2397853136062622, "learning_rate": 0.0001, "loss": 0.0281, "step": 14580 }, { "epoch": 86.33136094674556, "grad_norm": 1.8540549278259277, "learning_rate": 0.0001, "loss": 0.0281, "step": 14590 }, { "epoch": 86.3905325443787, "grad_norm": 1.2607190608978271, "learning_rate": 0.0001, "loss": 0.0289, "step": 14600 }, { "epoch": 86.44970414201184, "grad_norm": 1.7354896068572998, "learning_rate": 0.0001, "loss": 0.028, "step": 14610 }, { "epoch": 86.50887573964496, "grad_norm": 1.8879687786102295, "learning_rate": 0.0001, "loss": 0.0285, "step": 14620 }, { "epoch": 86.5680473372781, "grad_norm": 2.0774261951446533, "learning_rate": 0.0001, "loss": 0.0284, "step": 14630 }, { "epoch": 86.62721893491124, "grad_norm": 1.663346290588379, "learning_rate": 0.0001, "loss": 0.0281, "step": 14640 }, { "epoch": 86.68639053254438, "grad_norm": 2.1256046295166016, "learning_rate": 0.0001, "loss": 0.0279, "step": 14650 }, { "epoch": 86.74556213017752, "grad_norm": 2.651665449142456, "learning_rate": 0.0001, "loss": 0.0298, "step": 14660 }, { "epoch": 86.80473372781066, "grad_norm": 1.3334369659423828, "learning_rate": 0.0001, "loss": 0.0281, "step": 14670 }, { "epoch": 86.86390532544378, "grad_norm": 1.8218896389007568, "learning_rate": 0.0001, "loss": 0.0285, "step": 14680 }, { "epoch": 86.92307692307692, "grad_norm": 2.2073845863342285, "learning_rate": 0.0001, "loss": 0.0284, "step": 14690 }, { "epoch": 86.98224852071006, "grad_norm": 2.39620304107666, "learning_rate": 0.0001, "loss": 0.0276, "step": 14700 }, { "epoch": 87.0414201183432, "grad_norm": 1.7653082609176636, "learning_rate": 0.0001, "loss": 0.0285, "step": 14710 }, { "epoch": 87.10059171597634, "grad_norm": 1.667904257774353, "learning_rate": 0.0001, "loss": 0.0276, "step": 14720 }, { "epoch": 87.15976331360947, "grad_norm": 1.8011410236358643, "learning_rate": 0.0001, "loss": 0.0279, "step": 14730 }, { "epoch": 87.2189349112426, "grad_norm": 1.2450700998306274, "learning_rate": 0.0001, "loss": 0.027, "step": 14740 }, { "epoch": 87.27810650887574, "grad_norm": 1.7694300413131714, "learning_rate": 0.0001, "loss": 0.0294, "step": 14750 }, { "epoch": 87.33727810650888, "grad_norm": 2.4683752059936523, "learning_rate": 0.0001, "loss": 0.0275, "step": 14760 }, { "epoch": 87.39644970414201, "grad_norm": 1.6653796434402466, "learning_rate": 0.0001, "loss": 0.0281, "step": 14770 }, { "epoch": 87.45562130177515, "grad_norm": 1.5716357231140137, "learning_rate": 0.0001, "loss": 0.0262, "step": 14780 }, { "epoch": 87.51479289940828, "grad_norm": 2.1472644805908203, "learning_rate": 0.0001, "loss": 0.0265, "step": 14790 }, { "epoch": 87.57396449704142, "grad_norm": 1.8734604120254517, "learning_rate": 0.0001, "loss": 0.0276, "step": 14800 }, { "epoch": 87.63313609467455, "grad_norm": 1.7445017099380493, "learning_rate": 0.0001, "loss": 0.0274, "step": 14810 }, { "epoch": 87.6923076923077, "grad_norm": 1.4185367822647095, "learning_rate": 0.0001, "loss": 0.0285, "step": 14820 }, { "epoch": 87.75147928994083, "grad_norm": 1.3925119638442993, "learning_rate": 0.0001, "loss": 0.0267, "step": 14830 }, { "epoch": 87.81065088757397, "grad_norm": 2.087134838104248, "learning_rate": 0.0001, "loss": 0.0288, "step": 14840 }, { "epoch": 87.8698224852071, "grad_norm": 1.8860834836959839, "learning_rate": 0.0001, "loss": 0.0284, "step": 14850 }, { "epoch": 87.92899408284023, "grad_norm": 1.1717151403427124, "learning_rate": 0.0001, "loss": 0.0267, "step": 14860 }, { "epoch": 87.98816568047337, "grad_norm": 1.9156579971313477, "learning_rate": 0.0001, "loss": 0.0286, "step": 14870 }, { "epoch": 88.04733727810651, "grad_norm": 1.6831936836242676, "learning_rate": 0.0001, "loss": 0.0265, "step": 14880 }, { "epoch": 88.10650887573965, "grad_norm": 1.634172797203064, "learning_rate": 0.0001, "loss": 0.0274, "step": 14890 }, { "epoch": 88.16568047337279, "grad_norm": 1.5360110998153687, "learning_rate": 0.0001, "loss": 0.0266, "step": 14900 }, { "epoch": 88.22485207100591, "grad_norm": 1.4418463706970215, "learning_rate": 0.0001, "loss": 0.0278, "step": 14910 }, { "epoch": 88.28402366863905, "grad_norm": 1.570798635482788, "learning_rate": 0.0001, "loss": 0.0276, "step": 14920 }, { "epoch": 88.34319526627219, "grad_norm": 1.7708101272583008, "learning_rate": 0.0001, "loss": 0.0264, "step": 14930 }, { "epoch": 88.40236686390533, "grad_norm": 1.5047050714492798, "learning_rate": 0.0001, "loss": 0.0277, "step": 14940 }, { "epoch": 88.46153846153847, "grad_norm": 1.7861096858978271, "learning_rate": 0.0001, "loss": 0.0264, "step": 14950 }, { "epoch": 88.52071005917159, "grad_norm": 1.6341906785964966, "learning_rate": 0.0001, "loss": 0.0272, "step": 14960 }, { "epoch": 88.57988165680473, "grad_norm": 1.6625375747680664, "learning_rate": 0.0001, "loss": 0.0271, "step": 14970 }, { "epoch": 88.63905325443787, "grad_norm": 1.6868609189987183, "learning_rate": 0.0001, "loss": 0.0258, "step": 14980 }, { "epoch": 88.69822485207101, "grad_norm": 1.6275410652160645, "learning_rate": 0.0001, "loss": 0.0276, "step": 14990 }, { "epoch": 88.75739644970415, "grad_norm": 2.1598098278045654, "learning_rate": 0.0001, "loss": 0.0295, "step": 15000 }, { "epoch": 88.81656804733728, "grad_norm": 1.5022637844085693, "learning_rate": 0.0001, "loss": 0.028, "step": 15010 }, { "epoch": 88.87573964497041, "grad_norm": 2.021944761276245, "learning_rate": 0.0001, "loss": 0.0269, "step": 15020 }, { "epoch": 88.93491124260355, "grad_norm": 1.6724249124526978, "learning_rate": 0.0001, "loss": 0.0268, "step": 15030 }, { "epoch": 88.99408284023669, "grad_norm": 1.3633445501327515, "learning_rate": 0.0001, "loss": 0.0264, "step": 15040 }, { "epoch": 89.05325443786982, "grad_norm": 1.7915236949920654, "learning_rate": 0.0001, "loss": 0.0269, "step": 15050 }, { "epoch": 89.11242603550296, "grad_norm": 1.6124951839447021, "learning_rate": 0.0001, "loss": 0.028, "step": 15060 }, { "epoch": 89.17159763313609, "grad_norm": 1.5810753107070923, "learning_rate": 0.0001, "loss": 0.0277, "step": 15070 }, { "epoch": 89.23076923076923, "grad_norm": 2.3158791065216064, "learning_rate": 0.0001, "loss": 0.0256, "step": 15080 }, { "epoch": 89.28994082840237, "grad_norm": 1.6969501972198486, "learning_rate": 0.0001, "loss": 0.0257, "step": 15090 }, { "epoch": 89.3491124260355, "grad_norm": 1.7599455118179321, "learning_rate": 0.0001, "loss": 0.0251, "step": 15100 }, { "epoch": 89.40828402366864, "grad_norm": 1.4445723295211792, "learning_rate": 0.0001, "loss": 0.0263, "step": 15110 }, { "epoch": 89.46745562130178, "grad_norm": 1.4228659868240356, "learning_rate": 0.0001, "loss": 0.0252, "step": 15120 }, { "epoch": 89.5266272189349, "grad_norm": 1.3319268226623535, "learning_rate": 0.0001, "loss": 0.0266, "step": 15130 }, { "epoch": 89.58579881656804, "grad_norm": 1.746983289718628, "learning_rate": 0.0001, "loss": 0.0276, "step": 15140 }, { "epoch": 89.64497041420118, "grad_norm": 1.3390052318572998, "learning_rate": 0.0001, "loss": 0.0283, "step": 15150 }, { "epoch": 89.70414201183432, "grad_norm": 1.6228300333023071, "learning_rate": 0.0001, "loss": 0.0268, "step": 15160 }, { "epoch": 89.76331360946746, "grad_norm": 2.2015228271484375, "learning_rate": 0.0001, "loss": 0.0285, "step": 15170 }, { "epoch": 89.8224852071006, "grad_norm": 2.098471164703369, "learning_rate": 0.0001, "loss": 0.0284, "step": 15180 }, { "epoch": 89.88165680473372, "grad_norm": 2.2601654529571533, "learning_rate": 0.0001, "loss": 0.0273, "step": 15190 }, { "epoch": 89.94082840236686, "grad_norm": 2.0032951831817627, "learning_rate": 0.0001, "loss": 0.026, "step": 15200 }, { "epoch": 90.0, "grad_norm": 2.158338785171509, "learning_rate": 0.0001, "loss": 0.0276, "step": 15210 }, { "epoch": 90.05917159763314, "grad_norm": 2.011505126953125, "learning_rate": 0.0001, "loss": 0.0274, "step": 15220 }, { "epoch": 90.11834319526628, "grad_norm": 1.8215605020523071, "learning_rate": 0.0001, "loss": 0.0255, "step": 15230 }, { "epoch": 90.1775147928994, "grad_norm": 1.3463451862335205, "learning_rate": 0.0001, "loss": 0.027, "step": 15240 }, { "epoch": 90.23668639053254, "grad_norm": 1.1675927639007568, "learning_rate": 0.0001, "loss": 0.0266, "step": 15250 }, { "epoch": 90.29585798816568, "grad_norm": 1.8829050064086914, "learning_rate": 0.0001, "loss": 0.029, "step": 15260 }, { "epoch": 90.35502958579882, "grad_norm": 1.6154104471206665, "learning_rate": 0.0001, "loss": 0.0271, "step": 15270 }, { "epoch": 90.41420118343196, "grad_norm": 1.849914789199829, "learning_rate": 0.0001, "loss": 0.0268, "step": 15280 }, { "epoch": 90.4733727810651, "grad_norm": 1.89186429977417, "learning_rate": 0.0001, "loss": 0.0278, "step": 15290 }, { "epoch": 90.53254437869822, "grad_norm": 1.839965581893921, "learning_rate": 0.0001, "loss": 0.0262, "step": 15300 }, { "epoch": 90.59171597633136, "grad_norm": 1.6318309307098389, "learning_rate": 0.0001, "loss": 0.0268, "step": 15310 }, { "epoch": 90.6508875739645, "grad_norm": 2.1691970825195312, "learning_rate": 0.0001, "loss": 0.0286, "step": 15320 }, { "epoch": 90.71005917159763, "grad_norm": 1.272350788116455, "learning_rate": 0.0001, "loss": 0.0283, "step": 15330 }, { "epoch": 90.76923076923077, "grad_norm": 1.512602686882019, "learning_rate": 0.0001, "loss": 0.0257, "step": 15340 }, { "epoch": 90.82840236686391, "grad_norm": 1.6773158311843872, "learning_rate": 0.0001, "loss": 0.0267, "step": 15350 }, { "epoch": 90.88757396449704, "grad_norm": 1.5169200897216797, "learning_rate": 0.0001, "loss": 0.0263, "step": 15360 }, { "epoch": 90.94674556213018, "grad_norm": 1.859876036643982, "learning_rate": 0.0001, "loss": 0.0265, "step": 15370 }, { "epoch": 91.00591715976331, "grad_norm": 1.6127303838729858, "learning_rate": 0.0001, "loss": 0.0249, "step": 15380 }, { "epoch": 91.06508875739645, "grad_norm": 1.4845398664474487, "learning_rate": 0.0001, "loss": 0.0279, "step": 15390 }, { "epoch": 91.12426035502959, "grad_norm": 1.160044550895691, "learning_rate": 0.0001, "loss": 0.0274, "step": 15400 }, { "epoch": 91.18343195266272, "grad_norm": 2.0908052921295166, "learning_rate": 0.0001, "loss": 0.0268, "step": 15410 }, { "epoch": 91.24260355029585, "grad_norm": 1.797471046447754, "learning_rate": 0.0001, "loss": 0.0286, "step": 15420 }, { "epoch": 91.30177514792899, "grad_norm": 1.6939975023269653, "learning_rate": 0.0001, "loss": 0.0268, "step": 15430 }, { "epoch": 91.36094674556213, "grad_norm": 1.4528217315673828, "learning_rate": 0.0001, "loss": 0.0267, "step": 15440 }, { "epoch": 91.42011834319527, "grad_norm": 1.222514271736145, "learning_rate": 0.0001, "loss": 0.0263, "step": 15450 }, { "epoch": 91.47928994082841, "grad_norm": 1.5007128715515137, "learning_rate": 0.0001, "loss": 0.0269, "step": 15460 }, { "epoch": 91.53846153846153, "grad_norm": 1.9214576482772827, "learning_rate": 0.0001, "loss": 0.0266, "step": 15470 }, { "epoch": 91.59763313609467, "grad_norm": 1.710034728050232, "learning_rate": 0.0001, "loss": 0.0284, "step": 15480 }, { "epoch": 91.65680473372781, "grad_norm": 2.1379175186157227, "learning_rate": 0.0001, "loss": 0.027, "step": 15490 }, { "epoch": 91.71597633136095, "grad_norm": 1.4170961380004883, "learning_rate": 0.0001, "loss": 0.0266, "step": 15500 }, { "epoch": 91.77514792899409, "grad_norm": 1.3856899738311768, "learning_rate": 0.0001, "loss": 0.0264, "step": 15510 }, { "epoch": 91.83431952662721, "grad_norm": 1.8492748737335205, "learning_rate": 0.0001, "loss": 0.027, "step": 15520 }, { "epoch": 91.89349112426035, "grad_norm": 1.6489924192428589, "learning_rate": 0.0001, "loss": 0.0268, "step": 15530 }, { "epoch": 91.95266272189349, "grad_norm": 1.6302393674850464, "learning_rate": 0.0001, "loss": 0.0275, "step": 15540 }, { "epoch": 92.01183431952663, "grad_norm": 1.5825130939483643, "learning_rate": 0.0001, "loss": 0.026, "step": 15550 }, { "epoch": 92.07100591715977, "grad_norm": 1.7874207496643066, "learning_rate": 0.0001, "loss": 0.026, "step": 15560 }, { "epoch": 92.1301775147929, "grad_norm": 1.7968064546585083, "learning_rate": 0.0001, "loss": 0.0259, "step": 15570 }, { "epoch": 92.18934911242603, "grad_norm": 1.6556369066238403, "learning_rate": 0.0001, "loss": 0.0256, "step": 15580 }, { "epoch": 92.24852071005917, "grad_norm": 1.5757861137390137, "learning_rate": 0.0001, "loss": 0.0257, "step": 15590 }, { "epoch": 92.3076923076923, "grad_norm": 1.7952519655227661, "learning_rate": 0.0001, "loss": 0.025, "step": 15600 }, { "epoch": 92.36686390532545, "grad_norm": 2.3028202056884766, "learning_rate": 0.0001, "loss": 0.0259, "step": 15610 }, { "epoch": 92.42603550295858, "grad_norm": 2.0497164726257324, "learning_rate": 0.0001, "loss": 0.0271, "step": 15620 }, { "epoch": 92.48520710059172, "grad_norm": 1.571807861328125, "learning_rate": 0.0001, "loss": 0.0263, "step": 15630 }, { "epoch": 92.54437869822485, "grad_norm": 1.4125871658325195, "learning_rate": 0.0001, "loss": 0.0247, "step": 15640 }, { "epoch": 92.60355029585799, "grad_norm": 1.424944519996643, "learning_rate": 0.0001, "loss": 0.0261, "step": 15650 }, { "epoch": 92.66272189349112, "grad_norm": 1.8362457752227783, "learning_rate": 0.0001, "loss": 0.0268, "step": 15660 }, { "epoch": 92.72189349112426, "grad_norm": 1.5764145851135254, "learning_rate": 0.0001, "loss": 0.0257, "step": 15670 }, { "epoch": 92.7810650887574, "grad_norm": 1.2949786186218262, "learning_rate": 0.0001, "loss": 0.0273, "step": 15680 }, { "epoch": 92.84023668639053, "grad_norm": 1.552280068397522, "learning_rate": 0.0001, "loss": 0.0271, "step": 15690 }, { "epoch": 92.89940828402366, "grad_norm": 1.9072961807250977, "learning_rate": 0.0001, "loss": 0.028, "step": 15700 }, { "epoch": 92.9585798816568, "grad_norm": 1.3595942258834839, "learning_rate": 0.0001, "loss": 0.031, "step": 15710 }, { "epoch": 93.01775147928994, "grad_norm": 1.5796197652816772, "learning_rate": 0.0001, "loss": 0.0283, "step": 15720 }, { "epoch": 93.07692307692308, "grad_norm": 1.4288289546966553, "learning_rate": 0.0001, "loss": 0.0283, "step": 15730 }, { "epoch": 93.13609467455622, "grad_norm": 1.7119187116622925, "learning_rate": 0.0001, "loss": 0.0267, "step": 15740 }, { "epoch": 93.19526627218934, "grad_norm": 1.841143012046814, "learning_rate": 0.0001, "loss": 0.0272, "step": 15750 }, { "epoch": 93.25443786982248, "grad_norm": 1.5557856559753418, "learning_rate": 0.0001, "loss": 0.0271, "step": 15760 }, { "epoch": 93.31360946745562, "grad_norm": 2.1959996223449707, "learning_rate": 0.0001, "loss": 0.0271, "step": 15770 }, { "epoch": 93.37278106508876, "grad_norm": 1.4359514713287354, "learning_rate": 0.0001, "loss": 0.0266, "step": 15780 }, { "epoch": 93.4319526627219, "grad_norm": 1.6071994304656982, "learning_rate": 0.0001, "loss": 0.0271, "step": 15790 }, { "epoch": 93.49112426035504, "grad_norm": 1.6520476341247559, "learning_rate": 0.0001, "loss": 0.0271, "step": 15800 }, { "epoch": 93.55029585798816, "grad_norm": 1.891273021697998, "learning_rate": 0.0001, "loss": 0.0272, "step": 15810 }, { "epoch": 93.6094674556213, "grad_norm": 1.7385085821151733, "learning_rate": 0.0001, "loss": 0.027, "step": 15820 }, { "epoch": 93.66863905325444, "grad_norm": 1.561846137046814, "learning_rate": 0.0001, "loss": 0.0269, "step": 15830 }, { "epoch": 93.72781065088758, "grad_norm": 2.0920162200927734, "learning_rate": 0.0001, "loss": 0.027, "step": 15840 }, { "epoch": 93.78698224852072, "grad_norm": 1.5875986814498901, "learning_rate": 0.0001, "loss": 0.0278, "step": 15850 }, { "epoch": 93.84615384615384, "grad_norm": 1.7659938335418701, "learning_rate": 0.0001, "loss": 0.0273, "step": 15860 }, { "epoch": 93.90532544378698, "grad_norm": 2.017791986465454, "learning_rate": 0.0001, "loss": 0.027, "step": 15870 }, { "epoch": 93.96449704142012, "grad_norm": 1.6771608591079712, "learning_rate": 0.0001, "loss": 0.0275, "step": 15880 }, { "epoch": 94.02366863905326, "grad_norm": 1.7506619691848755, "learning_rate": 0.0001, "loss": 0.0272, "step": 15890 }, { "epoch": 94.0828402366864, "grad_norm": 1.5292552709579468, "learning_rate": 0.0001, "loss": 0.0263, "step": 15900 }, { "epoch": 94.14201183431953, "grad_norm": 1.036218523979187, "learning_rate": 0.0001, "loss": 0.0254, "step": 15910 }, { "epoch": 94.20118343195266, "grad_norm": 1.589901328086853, "learning_rate": 0.0001, "loss": 0.0271, "step": 15920 }, { "epoch": 94.2603550295858, "grad_norm": 1.5320048332214355, "learning_rate": 0.0001, "loss": 0.0288, "step": 15930 }, { "epoch": 94.31952662721893, "grad_norm": 1.6022655963897705, "learning_rate": 0.0001, "loss": 0.0275, "step": 15940 }, { "epoch": 94.37869822485207, "grad_norm": 1.7221791744232178, "learning_rate": 0.0001, "loss": 0.0284, "step": 15950 }, { "epoch": 94.43786982248521, "grad_norm": 2.0784683227539062, "learning_rate": 0.0001, "loss": 0.027, "step": 15960 }, { "epoch": 94.49704142011835, "grad_norm": 1.9355556964874268, "learning_rate": 0.0001, "loss": 0.0283, "step": 15970 }, { "epoch": 94.55621301775147, "grad_norm": 1.7169827222824097, "learning_rate": 0.0001, "loss": 0.0291, "step": 15980 }, { "epoch": 94.61538461538461, "grad_norm": 1.5959008932113647, "learning_rate": 0.0001, "loss": 0.027, "step": 15990 }, { "epoch": 94.67455621301775, "grad_norm": 1.7118438482284546, "learning_rate": 0.0001, "loss": 0.0282, "step": 16000 }, { "epoch": 94.73372781065089, "grad_norm": 1.4468134641647339, "learning_rate": 0.0001, "loss": 0.0265, "step": 16010 }, { "epoch": 94.79289940828403, "grad_norm": 1.4634939432144165, "learning_rate": 0.0001, "loss": 0.0276, "step": 16020 }, { "epoch": 94.85207100591715, "grad_norm": 1.7240033149719238, "learning_rate": 0.0001, "loss": 0.027, "step": 16030 }, { "epoch": 94.91124260355029, "grad_norm": 1.738993763923645, "learning_rate": 0.0001, "loss": 0.0267, "step": 16040 }, { "epoch": 94.97041420118343, "grad_norm": 1.5479215383529663, "learning_rate": 0.0001, "loss": 0.0264, "step": 16050 }, { "epoch": 95.02958579881657, "grad_norm": 1.3362421989440918, "learning_rate": 0.0001, "loss": 0.0265, "step": 16060 }, { "epoch": 95.08875739644971, "grad_norm": 1.5550212860107422, "learning_rate": 0.0001, "loss": 0.0272, "step": 16070 }, { "epoch": 95.14792899408285, "grad_norm": 1.61492919921875, "learning_rate": 0.0001, "loss": 0.0262, "step": 16080 }, { "epoch": 95.20710059171597, "grad_norm": 1.8070425987243652, "learning_rate": 0.0001, "loss": 0.0264, "step": 16090 }, { "epoch": 95.26627218934911, "grad_norm": 1.3418387174606323, "learning_rate": 0.0001, "loss": 0.0273, "step": 16100 }, { "epoch": 95.32544378698225, "grad_norm": 1.653890609741211, "learning_rate": 0.0001, "loss": 0.0259, "step": 16110 }, { "epoch": 95.38461538461539, "grad_norm": 1.6647813320159912, "learning_rate": 0.0001, "loss": 0.0258, "step": 16120 }, { "epoch": 95.44378698224853, "grad_norm": 1.5109667778015137, "learning_rate": 0.0001, "loss": 0.0264, "step": 16130 }, { "epoch": 95.50295857988165, "grad_norm": 1.4142593145370483, "learning_rate": 0.0001, "loss": 0.0264, "step": 16140 }, { "epoch": 95.56213017751479, "grad_norm": 1.3701032400131226, "learning_rate": 0.0001, "loss": 0.0271, "step": 16150 }, { "epoch": 95.62130177514793, "grad_norm": 1.7060127258300781, "learning_rate": 0.0001, "loss": 0.0267, "step": 16160 }, { "epoch": 95.68047337278107, "grad_norm": 2.1572930812835693, "learning_rate": 0.0001, "loss": 0.0261, "step": 16170 }, { "epoch": 95.7396449704142, "grad_norm": 1.5954173803329468, "learning_rate": 0.0001, "loss": 0.0283, "step": 16180 }, { "epoch": 95.79881656804734, "grad_norm": 1.547423243522644, "learning_rate": 0.0001, "loss": 0.0255, "step": 16190 }, { "epoch": 95.85798816568047, "grad_norm": 1.3531512022018433, "learning_rate": 0.0001, "loss": 0.0259, "step": 16200 }, { "epoch": 95.9171597633136, "grad_norm": 1.9303975105285645, "learning_rate": 0.0001, "loss": 0.0262, "step": 16210 }, { "epoch": 95.97633136094674, "grad_norm": 2.4396984577178955, "learning_rate": 0.0001, "loss": 0.0277, "step": 16220 }, { "epoch": 96.03550295857988, "grad_norm": 1.445061445236206, "learning_rate": 0.0001, "loss": 0.0267, "step": 16230 }, { "epoch": 96.09467455621302, "grad_norm": 1.9003947973251343, "learning_rate": 0.0001, "loss": 0.0277, "step": 16240 }, { "epoch": 96.15384615384616, "grad_norm": 1.8887784481048584, "learning_rate": 0.0001, "loss": 0.0267, "step": 16250 }, { "epoch": 96.21301775147928, "grad_norm": 1.4374672174453735, "learning_rate": 0.0001, "loss": 0.0265, "step": 16260 }, { "epoch": 96.27218934911242, "grad_norm": 2.373680830001831, "learning_rate": 0.0001, "loss": 0.0252, "step": 16270 }, { "epoch": 96.33136094674556, "grad_norm": 1.4005873203277588, "learning_rate": 0.0001, "loss": 0.0262, "step": 16280 }, { "epoch": 96.3905325443787, "grad_norm": 1.7103840112686157, "learning_rate": 0.0001, "loss": 0.0261, "step": 16290 }, { "epoch": 96.44970414201184, "grad_norm": 1.2638014554977417, "learning_rate": 0.0001, "loss": 0.0271, "step": 16300 }, { "epoch": 96.50887573964496, "grad_norm": 1.4438227415084839, "learning_rate": 0.0001, "loss": 0.0261, "step": 16310 }, { "epoch": 96.5680473372781, "grad_norm": 1.4212442636489868, "learning_rate": 0.0001, "loss": 0.0276, "step": 16320 }, { "epoch": 96.62721893491124, "grad_norm": 1.588212251663208, "learning_rate": 0.0001, "loss": 0.0279, "step": 16330 }, { "epoch": 96.68639053254438, "grad_norm": 1.7317638397216797, "learning_rate": 0.0001, "loss": 0.0271, "step": 16340 }, { "epoch": 96.74556213017752, "grad_norm": 1.6484242677688599, "learning_rate": 0.0001, "loss": 0.0271, "step": 16350 }, { "epoch": 96.80473372781066, "grad_norm": 2.218794822692871, "learning_rate": 0.0001, "loss": 0.0273, "step": 16360 }, { "epoch": 96.86390532544378, "grad_norm": 1.8884763717651367, "learning_rate": 0.0001, "loss": 0.0285, "step": 16370 }, { "epoch": 96.92307692307692, "grad_norm": 1.9927177429199219, "learning_rate": 0.0001, "loss": 0.0276, "step": 16380 }, { "epoch": 96.98224852071006, "grad_norm": 1.8023303747177124, "learning_rate": 0.0001, "loss": 0.0268, "step": 16390 }, { "epoch": 97.0414201183432, "grad_norm": 1.5173333883285522, "learning_rate": 0.0001, "loss": 0.0269, "step": 16400 }, { "epoch": 97.10059171597634, "grad_norm": 1.9109643697738647, "learning_rate": 0.0001, "loss": 0.0258, "step": 16410 }, { "epoch": 97.15976331360947, "grad_norm": 1.485317587852478, "learning_rate": 0.0001, "loss": 0.0269, "step": 16420 }, { "epoch": 97.2189349112426, "grad_norm": 1.8494558334350586, "learning_rate": 0.0001, "loss": 0.0267, "step": 16430 }, { "epoch": 97.27810650887574, "grad_norm": 1.5114191770553589, "learning_rate": 0.0001, "loss": 0.0276, "step": 16440 }, { "epoch": 97.33727810650888, "grad_norm": 1.9201053380966187, "learning_rate": 0.0001, "loss": 0.0255, "step": 16450 }, { "epoch": 97.39644970414201, "grad_norm": 1.520798683166504, "learning_rate": 0.0001, "loss": 0.0272, "step": 16460 }, { "epoch": 97.45562130177515, "grad_norm": 1.8467390537261963, "learning_rate": 0.0001, "loss": 0.0279, "step": 16470 }, { "epoch": 97.51479289940828, "grad_norm": 1.6369913816452026, "learning_rate": 0.0001, "loss": 0.0268, "step": 16480 }, { "epoch": 97.57396449704142, "grad_norm": 1.6534287929534912, "learning_rate": 0.0001, "loss": 0.0258, "step": 16490 }, { "epoch": 97.63313609467455, "grad_norm": 1.4711045026779175, "learning_rate": 0.0001, "loss": 0.0262, "step": 16500 }, { "epoch": 97.6923076923077, "grad_norm": 1.76560640335083, "learning_rate": 0.0001, "loss": 0.0258, "step": 16510 }, { "epoch": 97.75147928994083, "grad_norm": 1.7870081663131714, "learning_rate": 0.0001, "loss": 0.0266, "step": 16520 }, { "epoch": 97.81065088757397, "grad_norm": 1.494162917137146, "learning_rate": 0.0001, "loss": 0.0277, "step": 16530 }, { "epoch": 97.8698224852071, "grad_norm": 1.7298845052719116, "learning_rate": 0.0001, "loss": 0.0258, "step": 16540 }, { "epoch": 97.92899408284023, "grad_norm": 1.4313361644744873, "learning_rate": 0.0001, "loss": 0.0256, "step": 16550 }, { "epoch": 97.98816568047337, "grad_norm": 1.3597660064697266, "learning_rate": 0.0001, "loss": 0.0265, "step": 16560 }, { "epoch": 98.04733727810651, "grad_norm": 1.6125025749206543, "learning_rate": 0.0001, "loss": 0.0252, "step": 16570 }, { "epoch": 98.10650887573965, "grad_norm": 1.6149601936340332, "learning_rate": 0.0001, "loss": 0.026, "step": 16580 }, { "epoch": 98.16568047337279, "grad_norm": 2.0037436485290527, "learning_rate": 0.0001, "loss": 0.0266, "step": 16590 }, { "epoch": 98.22485207100591, "grad_norm": 1.7898390293121338, "learning_rate": 0.0001, "loss": 0.0258, "step": 16600 }, { "epoch": 98.28402366863905, "grad_norm": 1.5894354581832886, "learning_rate": 0.0001, "loss": 0.0259, "step": 16610 }, { "epoch": 98.34319526627219, "grad_norm": 1.3405784368515015, "learning_rate": 0.0001, "loss": 0.0253, "step": 16620 }, { "epoch": 98.40236686390533, "grad_norm": 1.6774709224700928, "learning_rate": 0.0001, "loss": 0.0259, "step": 16630 }, { "epoch": 98.46153846153847, "grad_norm": 1.432090401649475, "learning_rate": 0.0001, "loss": 0.0256, "step": 16640 }, { "epoch": 98.52071005917159, "grad_norm": 1.3350290060043335, "learning_rate": 0.0001, "loss": 0.0258, "step": 16650 }, { "epoch": 98.57988165680473, "grad_norm": 1.835278034210205, "learning_rate": 0.0001, "loss": 0.0266, "step": 16660 }, { "epoch": 98.63905325443787, "grad_norm": 1.606459140777588, "learning_rate": 0.0001, "loss": 0.0261, "step": 16670 }, { "epoch": 98.69822485207101, "grad_norm": 1.7699569463729858, "learning_rate": 0.0001, "loss": 0.0267, "step": 16680 }, { "epoch": 98.75739644970415, "grad_norm": 1.5623904466629028, "learning_rate": 0.0001, "loss": 0.0263, "step": 16690 }, { "epoch": 98.81656804733728, "grad_norm": 1.4613949060440063, "learning_rate": 0.0001, "loss": 0.0267, "step": 16700 }, { "epoch": 98.87573964497041, "grad_norm": 1.277431607246399, "learning_rate": 0.0001, "loss": 0.0255, "step": 16710 }, { "epoch": 98.93491124260355, "grad_norm": 1.694647192955017, "learning_rate": 0.0001, "loss": 0.0245, "step": 16720 }, { "epoch": 98.99408284023669, "grad_norm": 2.0295796394348145, "learning_rate": 0.0001, "loss": 0.0265, "step": 16730 }, { "epoch": 99.05325443786982, "grad_norm": 2.0689375400543213, "learning_rate": 0.0001, "loss": 0.0265, "step": 16740 }, { "epoch": 99.11242603550296, "grad_norm": 1.3141193389892578, "learning_rate": 0.0001, "loss": 0.025, "step": 16750 }, { "epoch": 99.17159763313609, "grad_norm": 1.3493739366531372, "learning_rate": 0.0001, "loss": 0.0264, "step": 16760 }, { "epoch": 99.23076923076923, "grad_norm": 1.6883349418640137, "learning_rate": 0.0001, "loss": 0.0255, "step": 16770 }, { "epoch": 99.28994082840237, "grad_norm": 1.5592509508132935, "learning_rate": 0.0001, "loss": 0.0245, "step": 16780 }, { "epoch": 99.3491124260355, "grad_norm": 1.5131289958953857, "learning_rate": 0.0001, "loss": 0.0252, "step": 16790 }, { "epoch": 99.40828402366864, "grad_norm": 1.2498494386672974, "learning_rate": 0.0001, "loss": 0.0255, "step": 16800 }, { "epoch": 99.46745562130178, "grad_norm": 1.3289576768875122, "learning_rate": 0.0001, "loss": 0.0244, "step": 16810 }, { "epoch": 99.5266272189349, "grad_norm": 1.6332381963729858, "learning_rate": 0.0001, "loss": 0.0267, "step": 16820 }, { "epoch": 99.58579881656804, "grad_norm": 2.0991885662078857, "learning_rate": 0.0001, "loss": 0.027, "step": 16830 }, { "epoch": 99.64497041420118, "grad_norm": 1.8775763511657715, "learning_rate": 0.0001, "loss": 0.0258, "step": 16840 }, { "epoch": 99.70414201183432, "grad_norm": 1.8826587200164795, "learning_rate": 0.0001, "loss": 0.0258, "step": 16850 }, { "epoch": 99.76331360946746, "grad_norm": 1.5205402374267578, "learning_rate": 0.0001, "loss": 0.0248, "step": 16860 }, { "epoch": 99.8224852071006, "grad_norm": 1.6357060670852661, "learning_rate": 0.0001, "loss": 0.026, "step": 16870 }, { "epoch": 99.88165680473372, "grad_norm": 1.3361115455627441, "learning_rate": 0.0001, "loss": 0.0266, "step": 16880 }, { "epoch": 99.94082840236686, "grad_norm": 1.682531476020813, "learning_rate": 0.0001, "loss": 0.0268, "step": 16890 }, { "epoch": 100.0, "grad_norm": 1.3777778148651123, "learning_rate": 0.0001, "loss": 0.0265, "step": 16900 }, { "epoch": 100.05917159763314, "grad_norm": 2.043806791305542, "learning_rate": 0.0001, "loss": 0.027, "step": 16910 }, { "epoch": 100.11834319526628, "grad_norm": 1.5332293510437012, "learning_rate": 0.0001, "loss": 0.0279, "step": 16920 }, { "epoch": 100.1775147928994, "grad_norm": 1.5093722343444824, "learning_rate": 0.0001, "loss": 0.0278, "step": 16930 }, { "epoch": 100.23668639053254, "grad_norm": 1.1633856296539307, "learning_rate": 0.0001, "loss": 0.0257, "step": 16940 }, { "epoch": 100.29585798816568, "grad_norm": 1.4330850839614868, "learning_rate": 0.0001, "loss": 0.0262, "step": 16950 }, { "epoch": 100.35502958579882, "grad_norm": 1.708530068397522, "learning_rate": 0.0001, "loss": 0.0262, "step": 16960 }, { "epoch": 100.41420118343196, "grad_norm": 1.524470567703247, "learning_rate": 0.0001, "loss": 0.0259, "step": 16970 }, { "epoch": 100.4733727810651, "grad_norm": 1.4476228952407837, "learning_rate": 0.0001, "loss": 0.0264, "step": 16980 }, { "epoch": 100.53254437869822, "grad_norm": 1.4369665384292603, "learning_rate": 0.0001, "loss": 0.0269, "step": 16990 }, { "epoch": 100.59171597633136, "grad_norm": 1.3812737464904785, "learning_rate": 0.0001, "loss": 0.0261, "step": 17000 }, { "epoch": 100.6508875739645, "grad_norm": 2.1099750995635986, "learning_rate": 0.0001, "loss": 0.0275, "step": 17010 }, { "epoch": 100.71005917159763, "grad_norm": 1.6349544525146484, "learning_rate": 0.0001, "loss": 0.0272, "step": 17020 }, { "epoch": 100.76923076923077, "grad_norm": 1.673108696937561, "learning_rate": 0.0001, "loss": 0.0276, "step": 17030 }, { "epoch": 100.82840236686391, "grad_norm": 1.5189217329025269, "learning_rate": 0.0001, "loss": 0.0268, "step": 17040 }, { "epoch": 100.88757396449704, "grad_norm": 1.6288717985153198, "learning_rate": 0.0001, "loss": 0.0266, "step": 17050 }, { "epoch": 100.94674556213018, "grad_norm": 1.5159913301467896, "learning_rate": 0.0001, "loss": 0.0248, "step": 17060 }, { "epoch": 101.00591715976331, "grad_norm": 1.472872257232666, "learning_rate": 0.0001, "loss": 0.024, "step": 17070 }, { "epoch": 101.06508875739645, "grad_norm": 1.9575871229171753, "learning_rate": 0.0001, "loss": 0.0254, "step": 17080 }, { "epoch": 101.12426035502959, "grad_norm": 2.385197162628174, "learning_rate": 0.0001, "loss": 0.0271, "step": 17090 }, { "epoch": 101.18343195266272, "grad_norm": 2.215818166732788, "learning_rate": 0.0001, "loss": 0.025, "step": 17100 }, { "epoch": 101.24260355029585, "grad_norm": 1.977542757987976, "learning_rate": 0.0001, "loss": 0.0255, "step": 17110 }, { "epoch": 101.30177514792899, "grad_norm": 2.0294933319091797, "learning_rate": 0.0001, "loss": 0.0258, "step": 17120 }, { "epoch": 101.36094674556213, "grad_norm": 1.7324378490447998, "learning_rate": 0.0001, "loss": 0.0252, "step": 17130 }, { "epoch": 101.42011834319527, "grad_norm": 1.630754828453064, "learning_rate": 0.0001, "loss": 0.026, "step": 17140 }, { "epoch": 101.47928994082841, "grad_norm": 2.1423933506011963, "learning_rate": 0.0001, "loss": 0.0254, "step": 17150 }, { "epoch": 101.53846153846153, "grad_norm": 1.910994052886963, "learning_rate": 0.0001, "loss": 0.0252, "step": 17160 }, { "epoch": 101.59763313609467, "grad_norm": 1.8440018892288208, "learning_rate": 0.0001, "loss": 0.026, "step": 17170 }, { "epoch": 101.65680473372781, "grad_norm": 1.5927495956420898, "learning_rate": 0.0001, "loss": 0.0252, "step": 17180 }, { "epoch": 101.71597633136095, "grad_norm": 1.640611171722412, "learning_rate": 0.0001, "loss": 0.0248, "step": 17190 }, { "epoch": 101.77514792899409, "grad_norm": 1.640289068222046, "learning_rate": 0.0001, "loss": 0.025, "step": 17200 }, { "epoch": 101.83431952662721, "grad_norm": 1.825135350227356, "learning_rate": 0.0001, "loss": 0.0261, "step": 17210 }, { "epoch": 101.89349112426035, "grad_norm": 2.1115877628326416, "learning_rate": 0.0001, "loss": 0.0254, "step": 17220 }, { "epoch": 101.95266272189349, "grad_norm": 1.9629405736923218, "learning_rate": 0.0001, "loss": 0.0259, "step": 17230 }, { "epoch": 102.01183431952663, "grad_norm": 1.6097747087478638, "learning_rate": 0.0001, "loss": 0.0251, "step": 17240 }, { "epoch": 102.07100591715977, "grad_norm": 2.1289632320404053, "learning_rate": 0.0001, "loss": 0.0258, "step": 17250 }, { "epoch": 102.1301775147929, "grad_norm": 1.3834342956542969, "learning_rate": 0.0001, "loss": 0.0247, "step": 17260 }, { "epoch": 102.18934911242603, "grad_norm": 1.6639059782028198, "learning_rate": 0.0001, "loss": 0.0264, "step": 17270 }, { "epoch": 102.24852071005917, "grad_norm": 1.5481454133987427, "learning_rate": 0.0001, "loss": 0.024, "step": 17280 }, { "epoch": 102.3076923076923, "grad_norm": 1.5337421894073486, "learning_rate": 0.0001, "loss": 0.0258, "step": 17290 }, { "epoch": 102.36686390532545, "grad_norm": 1.332580804824829, "learning_rate": 0.0001, "loss": 0.025, "step": 17300 }, { "epoch": 102.42603550295858, "grad_norm": 1.4046190977096558, "learning_rate": 0.0001, "loss": 0.0264, "step": 17310 }, { "epoch": 102.48520710059172, "grad_norm": 1.9278885126113892, "learning_rate": 0.0001, "loss": 0.0262, "step": 17320 }, { "epoch": 102.54437869822485, "grad_norm": 1.3687036037445068, "learning_rate": 0.0001, "loss": 0.0269, "step": 17330 }, { "epoch": 102.60355029585799, "grad_norm": 1.3971467018127441, "learning_rate": 0.0001, "loss": 0.0265, "step": 17340 }, { "epoch": 102.66272189349112, "grad_norm": 1.7415915727615356, "learning_rate": 0.0001, "loss": 0.0264, "step": 17350 }, { "epoch": 102.72189349112426, "grad_norm": 1.5043632984161377, "learning_rate": 0.0001, "loss": 0.0269, "step": 17360 }, { "epoch": 102.7810650887574, "grad_norm": 1.066420316696167, "learning_rate": 0.0001, "loss": 0.0268, "step": 17370 }, { "epoch": 102.84023668639053, "grad_norm": 1.7254488468170166, "learning_rate": 0.0001, "loss": 0.0246, "step": 17380 }, { "epoch": 102.89940828402366, "grad_norm": 1.7191846370697021, "learning_rate": 0.0001, "loss": 0.0255, "step": 17390 }, { "epoch": 102.9585798816568, "grad_norm": 1.7242511510849, "learning_rate": 0.0001, "loss": 0.0275, "step": 17400 }, { "epoch": 103.01775147928994, "grad_norm": 1.5717693567276, "learning_rate": 0.0001, "loss": 0.0259, "step": 17410 }, { "epoch": 103.07692307692308, "grad_norm": 1.9666684865951538, "learning_rate": 0.0001, "loss": 0.0244, "step": 17420 }, { "epoch": 103.13609467455622, "grad_norm": 1.8861017227172852, "learning_rate": 0.0001, "loss": 0.0258, "step": 17430 }, { "epoch": 103.19526627218934, "grad_norm": 1.2661702632904053, "learning_rate": 0.0001, "loss": 0.026, "step": 17440 }, { "epoch": 103.25443786982248, "grad_norm": 1.5506561994552612, "learning_rate": 0.0001, "loss": 0.0258, "step": 17450 }, { "epoch": 103.31360946745562, "grad_norm": 1.3063440322875977, "learning_rate": 0.0001, "loss": 0.0242, "step": 17460 }, { "epoch": 103.37278106508876, "grad_norm": 1.601036548614502, "learning_rate": 0.0001, "loss": 0.0263, "step": 17470 }, { "epoch": 103.4319526627219, "grad_norm": 1.5009862184524536, "learning_rate": 0.0001, "loss": 0.0259, "step": 17480 }, { "epoch": 103.49112426035504, "grad_norm": 1.2007697820663452, "learning_rate": 0.0001, "loss": 0.0261, "step": 17490 }, { "epoch": 103.55029585798816, "grad_norm": 1.7272799015045166, "learning_rate": 0.0001, "loss": 0.0273, "step": 17500 }, { "epoch": 103.6094674556213, "grad_norm": 1.3488199710845947, "learning_rate": 0.0001, "loss": 0.0284, "step": 17510 }, { "epoch": 103.66863905325444, "grad_norm": 1.095036268234253, "learning_rate": 0.0001, "loss": 0.0269, "step": 17520 }, { "epoch": 103.72781065088758, "grad_norm": 1.6861222982406616, "learning_rate": 0.0001, "loss": 0.0266, "step": 17530 }, { "epoch": 103.78698224852072, "grad_norm": 1.9763951301574707, "learning_rate": 0.0001, "loss": 0.0266, "step": 17540 }, { "epoch": 103.84615384615384, "grad_norm": 1.536069631576538, "learning_rate": 0.0001, "loss": 0.0268, "step": 17550 }, { "epoch": 103.90532544378698, "grad_norm": 1.362142562866211, "learning_rate": 0.0001, "loss": 0.0269, "step": 17560 }, { "epoch": 103.96449704142012, "grad_norm": 1.4171984195709229, "learning_rate": 0.0001, "loss": 0.0267, "step": 17570 }, { "epoch": 104.02366863905326, "grad_norm": 1.3235714435577393, "learning_rate": 0.0001, "loss": 0.0277, "step": 17580 }, { "epoch": 104.0828402366864, "grad_norm": 2.038267135620117, "learning_rate": 0.0001, "loss": 0.0275, "step": 17590 }, { "epoch": 104.14201183431953, "grad_norm": 1.3383609056472778, "learning_rate": 0.0001, "loss": 0.0256, "step": 17600 }, { "epoch": 104.20118343195266, "grad_norm": 1.6321433782577515, "learning_rate": 0.0001, "loss": 0.0273, "step": 17610 }, { "epoch": 104.2603550295858, "grad_norm": 1.647024393081665, "learning_rate": 0.0001, "loss": 0.0264, "step": 17620 }, { "epoch": 104.31952662721893, "grad_norm": 1.809311866760254, "learning_rate": 0.0001, "loss": 0.027, "step": 17630 }, { "epoch": 104.37869822485207, "grad_norm": 1.6921018362045288, "learning_rate": 0.0001, "loss": 0.0262, "step": 17640 }, { "epoch": 104.43786982248521, "grad_norm": 1.6428320407867432, "learning_rate": 0.0001, "loss": 0.0265, "step": 17650 }, { "epoch": 104.49704142011835, "grad_norm": 1.3917149305343628, "learning_rate": 0.0001, "loss": 0.025, "step": 17660 }, { "epoch": 104.55621301775147, "grad_norm": 1.3470455408096313, "learning_rate": 0.0001, "loss": 0.0242, "step": 17670 }, { "epoch": 104.61538461538461, "grad_norm": 1.5173072814941406, "learning_rate": 0.0001, "loss": 0.0259, "step": 17680 }, { "epoch": 104.67455621301775, "grad_norm": 1.0590153932571411, "learning_rate": 0.0001, "loss": 0.0262, "step": 17690 }, { "epoch": 104.73372781065089, "grad_norm": 1.273803472518921, "learning_rate": 0.0001, "loss": 0.0269, "step": 17700 }, { "epoch": 104.79289940828403, "grad_norm": 1.394396185874939, "learning_rate": 0.0001, "loss": 0.0272, "step": 17710 }, { "epoch": 104.85207100591715, "grad_norm": 1.2598323822021484, "learning_rate": 0.0001, "loss": 0.0289, "step": 17720 }, { "epoch": 104.91124260355029, "grad_norm": 2.0856642723083496, "learning_rate": 0.0001, "loss": 0.0307, "step": 17730 }, { "epoch": 104.97041420118343, "grad_norm": 2.8122613430023193, "learning_rate": 0.0001, "loss": 0.0284, "step": 17740 }, { "epoch": 105.02958579881657, "grad_norm": 1.9192427396774292, "learning_rate": 0.0001, "loss": 0.0259, "step": 17750 }, { "epoch": 105.08875739644971, "grad_norm": 1.6123884916305542, "learning_rate": 0.0001, "loss": 0.0261, "step": 17760 }, { "epoch": 105.14792899408285, "grad_norm": 1.7245540618896484, "learning_rate": 0.0001, "loss": 0.0263, "step": 17770 }, { "epoch": 105.20710059171597, "grad_norm": 2.3444323539733887, "learning_rate": 0.0001, "loss": 0.0257, "step": 17780 }, { "epoch": 105.26627218934911, "grad_norm": 1.5307093858718872, "learning_rate": 0.0001, "loss": 0.0262, "step": 17790 }, { "epoch": 105.32544378698225, "grad_norm": 1.663991093635559, "learning_rate": 0.0001, "loss": 0.0264, "step": 17800 }, { "epoch": 105.38461538461539, "grad_norm": 1.242200255393982, "learning_rate": 0.0001, "loss": 0.0247, "step": 17810 }, { "epoch": 105.44378698224853, "grad_norm": 1.1324563026428223, "learning_rate": 0.0001, "loss": 0.0261, "step": 17820 }, { "epoch": 105.50295857988165, "grad_norm": 1.4513096809387207, "learning_rate": 0.0001, "loss": 0.0268, "step": 17830 }, { "epoch": 105.56213017751479, "grad_norm": 1.2292907238006592, "learning_rate": 0.0001, "loss": 0.0277, "step": 17840 }, { "epoch": 105.62130177514793, "grad_norm": 1.8985364437103271, "learning_rate": 0.0001, "loss": 0.0286, "step": 17850 }, { "epoch": 105.68047337278107, "grad_norm": 1.344464898109436, "learning_rate": 0.0001, "loss": 0.0275, "step": 17860 }, { "epoch": 105.7396449704142, "grad_norm": 1.3315728902816772, "learning_rate": 0.0001, "loss": 0.0259, "step": 17870 }, { "epoch": 105.79881656804734, "grad_norm": 1.3947885036468506, "learning_rate": 0.0001, "loss": 0.0258, "step": 17880 }, { "epoch": 105.85798816568047, "grad_norm": 1.4972177743911743, "learning_rate": 0.0001, "loss": 0.0258, "step": 17890 }, { "epoch": 105.9171597633136, "grad_norm": 1.479939341545105, "learning_rate": 0.0001, "loss": 0.025, "step": 17900 }, { "epoch": 105.97633136094674, "grad_norm": 1.5584843158721924, "learning_rate": 0.0001, "loss": 0.0257, "step": 17910 }, { "epoch": 106.03550295857988, "grad_norm": 1.155634880065918, "learning_rate": 0.0001, "loss": 0.0262, "step": 17920 }, { "epoch": 106.09467455621302, "grad_norm": 1.7053613662719727, "learning_rate": 0.0001, "loss": 0.0257, "step": 17930 }, { "epoch": 106.15384615384616, "grad_norm": 1.8626021146774292, "learning_rate": 0.0001, "loss": 0.0262, "step": 17940 }, { "epoch": 106.21301775147928, "grad_norm": 1.7530268430709839, "learning_rate": 0.0001, "loss": 0.0261, "step": 17950 }, { "epoch": 106.27218934911242, "grad_norm": 2.008701801300049, "learning_rate": 0.0001, "loss": 0.0259, "step": 17960 }, { "epoch": 106.33136094674556, "grad_norm": 1.3362863063812256, "learning_rate": 0.0001, "loss": 0.0254, "step": 17970 }, { "epoch": 106.3905325443787, "grad_norm": 1.2728350162506104, "learning_rate": 0.0001, "loss": 0.0245, "step": 17980 }, { "epoch": 106.44970414201184, "grad_norm": 1.0574220418930054, "learning_rate": 0.0001, "loss": 0.0243, "step": 17990 }, { "epoch": 106.50887573964496, "grad_norm": 1.77041494846344, "learning_rate": 0.0001, "loss": 0.0255, "step": 18000 }, { "epoch": 106.5680473372781, "grad_norm": 1.334742546081543, "learning_rate": 0.0001, "loss": 0.0272, "step": 18010 }, { "epoch": 106.62721893491124, "grad_norm": 1.1167922019958496, "learning_rate": 0.0001, "loss": 0.0254, "step": 18020 }, { "epoch": 106.68639053254438, "grad_norm": 1.591597318649292, "learning_rate": 0.0001, "loss": 0.0258, "step": 18030 }, { "epoch": 106.74556213017752, "grad_norm": 1.681300163269043, "learning_rate": 0.0001, "loss": 0.0267, "step": 18040 }, { "epoch": 106.80473372781066, "grad_norm": 1.9211896657943726, "learning_rate": 0.0001, "loss": 0.0274, "step": 18050 }, { "epoch": 106.86390532544378, "grad_norm": 1.4798047542572021, "learning_rate": 0.0001, "loss": 0.0279, "step": 18060 }, { "epoch": 106.92307692307692, "grad_norm": 1.3561455011367798, "learning_rate": 0.0001, "loss": 0.027, "step": 18070 }, { "epoch": 106.98224852071006, "grad_norm": 1.3977030515670776, "learning_rate": 0.0001, "loss": 0.026, "step": 18080 }, { "epoch": 107.0414201183432, "grad_norm": 1.5403028726577759, "learning_rate": 0.0001, "loss": 0.0259, "step": 18090 }, { "epoch": 107.10059171597634, "grad_norm": 1.9070069789886475, "learning_rate": 0.0001, "loss": 0.0263, "step": 18100 }, { "epoch": 107.15976331360947, "grad_norm": 1.4401390552520752, "learning_rate": 0.0001, "loss": 0.0266, "step": 18110 }, { "epoch": 107.2189349112426, "grad_norm": 1.4564379453659058, "learning_rate": 0.0001, "loss": 0.0269, "step": 18120 }, { "epoch": 107.27810650887574, "grad_norm": 1.508639931678772, "learning_rate": 0.0001, "loss": 0.0254, "step": 18130 }, { "epoch": 107.33727810650888, "grad_norm": 1.5949206352233887, "learning_rate": 0.0001, "loss": 0.0247, "step": 18140 }, { "epoch": 107.39644970414201, "grad_norm": 1.2537730932235718, "learning_rate": 0.0001, "loss": 0.0259, "step": 18150 }, { "epoch": 107.45562130177515, "grad_norm": 1.4299670457839966, "learning_rate": 0.0001, "loss": 0.0246, "step": 18160 }, { "epoch": 107.51479289940828, "grad_norm": 1.2867765426635742, "learning_rate": 0.0001, "loss": 0.0253, "step": 18170 }, { "epoch": 107.57396449704142, "grad_norm": 1.6854437589645386, "learning_rate": 0.0001, "loss": 0.0249, "step": 18180 }, { "epoch": 107.63313609467455, "grad_norm": 1.501104712486267, "learning_rate": 0.0001, "loss": 0.0249, "step": 18190 }, { "epoch": 107.6923076923077, "grad_norm": 1.2526355981826782, "learning_rate": 0.0001, "loss": 0.0254, "step": 18200 }, { "epoch": 107.75147928994083, "grad_norm": 1.354607343673706, "learning_rate": 0.0001, "loss": 0.0254, "step": 18210 }, { "epoch": 107.81065088757397, "grad_norm": 1.4535508155822754, "learning_rate": 0.0001, "loss": 0.0248, "step": 18220 }, { "epoch": 107.8698224852071, "grad_norm": 1.483656406402588, "learning_rate": 0.0001, "loss": 0.0255, "step": 18230 }, { "epoch": 107.92899408284023, "grad_norm": 1.1930395364761353, "learning_rate": 0.0001, "loss": 0.0256, "step": 18240 }, { "epoch": 107.98816568047337, "grad_norm": 1.8803009986877441, "learning_rate": 0.0001, "loss": 0.0268, "step": 18250 }, { "epoch": 108.04733727810651, "grad_norm": 1.4863253831863403, "learning_rate": 0.0001, "loss": 0.0263, "step": 18260 }, { "epoch": 108.10650887573965, "grad_norm": 1.7983930110931396, "learning_rate": 0.0001, "loss": 0.0257, "step": 18270 }, { "epoch": 108.16568047337279, "grad_norm": 1.7867889404296875, "learning_rate": 0.0001, "loss": 0.0251, "step": 18280 }, { "epoch": 108.22485207100591, "grad_norm": 1.4212646484375, "learning_rate": 0.0001, "loss": 0.0257, "step": 18290 }, { "epoch": 108.28402366863905, "grad_norm": 1.6550354957580566, "learning_rate": 0.0001, "loss": 0.0265, "step": 18300 }, { "epoch": 108.34319526627219, "grad_norm": 1.667275071144104, "learning_rate": 0.0001, "loss": 0.0248, "step": 18310 }, { "epoch": 108.40236686390533, "grad_norm": 1.6337451934814453, "learning_rate": 0.0001, "loss": 0.0279, "step": 18320 }, { "epoch": 108.46153846153847, "grad_norm": 1.832651972770691, "learning_rate": 0.0001, "loss": 0.0265, "step": 18330 }, { "epoch": 108.52071005917159, "grad_norm": 1.3788131475448608, "learning_rate": 0.0001, "loss": 0.0258, "step": 18340 }, { "epoch": 108.57988165680473, "grad_norm": 1.3384666442871094, "learning_rate": 0.0001, "loss": 0.0266, "step": 18350 }, { "epoch": 108.63905325443787, "grad_norm": 1.795141339302063, "learning_rate": 0.0001, "loss": 0.025, "step": 18360 }, { "epoch": 108.69822485207101, "grad_norm": 1.7532826662063599, "learning_rate": 0.0001, "loss": 0.0262, "step": 18370 }, { "epoch": 108.75739644970415, "grad_norm": 2.0311408042907715, "learning_rate": 0.0001, "loss": 0.0265, "step": 18380 }, { "epoch": 108.81656804733728, "grad_norm": 1.901776671409607, "learning_rate": 0.0001, "loss": 0.0257, "step": 18390 }, { "epoch": 108.87573964497041, "grad_norm": 1.2265090942382812, "learning_rate": 0.0001, "loss": 0.0258, "step": 18400 }, { "epoch": 108.93491124260355, "grad_norm": 1.4487773180007935, "learning_rate": 0.0001, "loss": 0.0247, "step": 18410 }, { "epoch": 108.99408284023669, "grad_norm": 2.1685791015625, "learning_rate": 0.0001, "loss": 0.0264, "step": 18420 }, { "epoch": 109.05325443786982, "grad_norm": 1.5457946062088013, "learning_rate": 0.0001, "loss": 0.0259, "step": 18430 }, { "epoch": 109.11242603550296, "grad_norm": 1.2115870714187622, "learning_rate": 0.0001, "loss": 0.025, "step": 18440 }, { "epoch": 109.17159763313609, "grad_norm": 1.1764168739318848, "learning_rate": 0.0001, "loss": 0.0252, "step": 18450 }, { "epoch": 109.23076923076923, "grad_norm": 1.382271647453308, "learning_rate": 0.0001, "loss": 0.0251, "step": 18460 }, { "epoch": 109.28994082840237, "grad_norm": 2.392472267150879, "learning_rate": 0.0001, "loss": 0.026, "step": 18470 }, { "epoch": 109.3491124260355, "grad_norm": 1.9167773723602295, "learning_rate": 0.0001, "loss": 0.0259, "step": 18480 }, { "epoch": 109.40828402366864, "grad_norm": 1.8893789052963257, "learning_rate": 0.0001, "loss": 0.0253, "step": 18490 }, { "epoch": 109.46745562130178, "grad_norm": 1.807668685913086, "learning_rate": 0.0001, "loss": 0.0266, "step": 18500 }, { "epoch": 109.5266272189349, "grad_norm": 1.9658818244934082, "learning_rate": 0.0001, "loss": 0.0242, "step": 18510 }, { "epoch": 109.58579881656804, "grad_norm": 1.7297940254211426, "learning_rate": 0.0001, "loss": 0.0239, "step": 18520 }, { "epoch": 109.64497041420118, "grad_norm": 1.1871163845062256, "learning_rate": 0.0001, "loss": 0.0252, "step": 18530 }, { "epoch": 109.70414201183432, "grad_norm": 1.6238046884536743, "learning_rate": 0.0001, "loss": 0.0253, "step": 18540 }, { "epoch": 109.76331360946746, "grad_norm": 1.5665812492370605, "learning_rate": 0.0001, "loss": 0.0241, "step": 18550 }, { "epoch": 109.8224852071006, "grad_norm": 2.194464683532715, "learning_rate": 0.0001, "loss": 0.0238, "step": 18560 }, { "epoch": 109.88165680473372, "grad_norm": 1.3683253526687622, "learning_rate": 0.0001, "loss": 0.0255, "step": 18570 }, { "epoch": 109.94082840236686, "grad_norm": 1.3032909631729126, "learning_rate": 0.0001, "loss": 0.0271, "step": 18580 }, { "epoch": 110.0, "grad_norm": 1.2106890678405762, "learning_rate": 0.0001, "loss": 0.0257, "step": 18590 }, { "epoch": 110.05917159763314, "grad_norm": 1.0278964042663574, "learning_rate": 0.0001, "loss": 0.0265, "step": 18600 }, { "epoch": 110.11834319526628, "grad_norm": 1.281351923942566, "learning_rate": 0.0001, "loss": 0.0252, "step": 18610 }, { "epoch": 110.1775147928994, "grad_norm": 1.3651962280273438, "learning_rate": 0.0001, "loss": 0.0255, "step": 18620 }, { "epoch": 110.23668639053254, "grad_norm": 1.5737565755844116, "learning_rate": 0.0001, "loss": 0.027, "step": 18630 }, { "epoch": 110.29585798816568, "grad_norm": 1.7021540403366089, "learning_rate": 0.0001, "loss": 0.0248, "step": 18640 }, { "epoch": 110.35502958579882, "grad_norm": 1.7169647216796875, "learning_rate": 0.0001, "loss": 0.0239, "step": 18650 }, { "epoch": 110.41420118343196, "grad_norm": 1.957826852798462, "learning_rate": 0.0001, "loss": 0.0256, "step": 18660 }, { "epoch": 110.4733727810651, "grad_norm": 1.7240309715270996, "learning_rate": 0.0001, "loss": 0.0239, "step": 18670 }, { "epoch": 110.53254437869822, "grad_norm": 1.6830214262008667, "learning_rate": 0.0001, "loss": 0.0245, "step": 18680 }, { "epoch": 110.59171597633136, "grad_norm": 2.5226330757141113, "learning_rate": 0.0001, "loss": 0.025, "step": 18690 }, { "epoch": 110.6508875739645, "grad_norm": 1.4356735944747925, "learning_rate": 0.0001, "loss": 0.0248, "step": 18700 }, { "epoch": 110.71005917159763, "grad_norm": 1.8798269033432007, "learning_rate": 0.0001, "loss": 0.0239, "step": 18710 }, { "epoch": 110.76923076923077, "grad_norm": 1.3705815076828003, "learning_rate": 0.0001, "loss": 0.0231, "step": 18720 }, { "epoch": 110.82840236686391, "grad_norm": 1.1489572525024414, "learning_rate": 0.0001, "loss": 0.0231, "step": 18730 }, { "epoch": 110.88757396449704, "grad_norm": 1.482344388961792, "learning_rate": 0.0001, "loss": 0.0243, "step": 18740 }, { "epoch": 110.94674556213018, "grad_norm": 1.514339566230774, "learning_rate": 0.0001, "loss": 0.0235, "step": 18750 }, { "epoch": 111.00591715976331, "grad_norm": 1.4481362104415894, "learning_rate": 0.0001, "loss": 0.0235, "step": 18760 }, { "epoch": 111.06508875739645, "grad_norm": 1.7595734596252441, "learning_rate": 0.0001, "loss": 0.024, "step": 18770 }, { "epoch": 111.12426035502959, "grad_norm": 1.4881128072738647, "learning_rate": 0.0001, "loss": 0.025, "step": 18780 }, { "epoch": 111.18343195266272, "grad_norm": 1.862440824508667, "learning_rate": 0.0001, "loss": 0.0253, "step": 18790 }, { "epoch": 111.24260355029585, "grad_norm": 1.4463809728622437, "learning_rate": 0.0001, "loss": 0.0248, "step": 18800 }, { "epoch": 111.30177514792899, "grad_norm": 1.3877615928649902, "learning_rate": 0.0001, "loss": 0.0254, "step": 18810 }, { "epoch": 111.36094674556213, "grad_norm": 1.6177059412002563, "learning_rate": 0.0001, "loss": 0.0241, "step": 18820 }, { "epoch": 111.42011834319527, "grad_norm": 1.8177719116210938, "learning_rate": 0.0001, "loss": 0.0264, "step": 18830 }, { "epoch": 111.47928994082841, "grad_norm": 2.059906005859375, "learning_rate": 0.0001, "loss": 0.0259, "step": 18840 }, { "epoch": 111.53846153846153, "grad_norm": 2.123096227645874, "learning_rate": 0.0001, "loss": 0.0253, "step": 18850 }, { "epoch": 111.59763313609467, "grad_norm": 1.5619239807128906, "learning_rate": 0.0001, "loss": 0.0258, "step": 18860 }, { "epoch": 111.65680473372781, "grad_norm": 1.464411973953247, "learning_rate": 0.0001, "loss": 0.0253, "step": 18870 }, { "epoch": 111.71597633136095, "grad_norm": 1.3420404195785522, "learning_rate": 0.0001, "loss": 0.025, "step": 18880 }, { "epoch": 111.77514792899409, "grad_norm": 1.271925926208496, "learning_rate": 0.0001, "loss": 0.0237, "step": 18890 }, { "epoch": 111.83431952662721, "grad_norm": 1.2641996145248413, "learning_rate": 0.0001, "loss": 0.0249, "step": 18900 }, { "epoch": 111.89349112426035, "grad_norm": 1.4037072658538818, "learning_rate": 0.0001, "loss": 0.0254, "step": 18910 }, { "epoch": 111.95266272189349, "grad_norm": 0.9968383312225342, "learning_rate": 0.0001, "loss": 0.0262, "step": 18920 }, { "epoch": 112.01183431952663, "grad_norm": 1.7601560354232788, "learning_rate": 0.0001, "loss": 0.0264, "step": 18930 }, { "epoch": 112.07100591715977, "grad_norm": 1.8560253381729126, "learning_rate": 0.0001, "loss": 0.0255, "step": 18940 }, { "epoch": 112.1301775147929, "grad_norm": 1.562642216682434, "learning_rate": 0.0001, "loss": 0.025, "step": 18950 }, { "epoch": 112.18934911242603, "grad_norm": 2.0420007705688477, "learning_rate": 0.0001, "loss": 0.025, "step": 18960 }, { "epoch": 112.24852071005917, "grad_norm": 1.6989458799362183, "learning_rate": 0.0001, "loss": 0.0237, "step": 18970 }, { "epoch": 112.3076923076923, "grad_norm": 1.3139822483062744, "learning_rate": 0.0001, "loss": 0.0241, "step": 18980 }, { "epoch": 112.36686390532545, "grad_norm": 2.159208297729492, "learning_rate": 0.0001, "loss": 0.0248, "step": 18990 }, { "epoch": 112.42603550295858, "grad_norm": 1.9208314418792725, "learning_rate": 0.0001, "loss": 0.0238, "step": 19000 }, { "epoch": 112.48520710059172, "grad_norm": 1.5137537717819214, "learning_rate": 0.0001, "loss": 0.025, "step": 19010 }, { "epoch": 112.54437869822485, "grad_norm": 1.5662624835968018, "learning_rate": 0.0001, "loss": 0.0241, "step": 19020 }, { "epoch": 112.60355029585799, "grad_norm": 1.5151339769363403, "learning_rate": 0.0001, "loss": 0.0229, "step": 19030 }, { "epoch": 112.66272189349112, "grad_norm": 1.9542598724365234, "learning_rate": 0.0001, "loss": 0.0241, "step": 19040 }, { "epoch": 112.72189349112426, "grad_norm": 1.6627751588821411, "learning_rate": 0.0001, "loss": 0.0244, "step": 19050 }, { "epoch": 112.7810650887574, "grad_norm": 1.7156360149383545, "learning_rate": 0.0001, "loss": 0.0242, "step": 19060 }, { "epoch": 112.84023668639053, "grad_norm": 1.4786920547485352, "learning_rate": 0.0001, "loss": 0.0242, "step": 19070 }, { "epoch": 112.89940828402366, "grad_norm": 1.3446840047836304, "learning_rate": 0.0001, "loss": 0.0232, "step": 19080 }, { "epoch": 112.9585798816568, "grad_norm": 1.6260424852371216, "learning_rate": 0.0001, "loss": 0.0246, "step": 19090 }, { "epoch": 113.01775147928994, "grad_norm": 1.7538962364196777, "learning_rate": 0.0001, "loss": 0.0249, "step": 19100 }, { "epoch": 113.07692307692308, "grad_norm": 1.6095421314239502, "learning_rate": 0.0001, "loss": 0.0248, "step": 19110 }, { "epoch": 113.13609467455622, "grad_norm": 1.9238210916519165, "learning_rate": 0.0001, "loss": 0.0245, "step": 19120 }, { "epoch": 113.19526627218934, "grad_norm": 1.4737799167633057, "learning_rate": 0.0001, "loss": 0.025, "step": 19130 }, { "epoch": 113.25443786982248, "grad_norm": 1.5817943811416626, "learning_rate": 0.0001, "loss": 0.0248, "step": 19140 }, { "epoch": 113.31360946745562, "grad_norm": 1.3485914468765259, "learning_rate": 0.0001, "loss": 0.0244, "step": 19150 }, { "epoch": 113.37278106508876, "grad_norm": 1.383191466331482, "learning_rate": 0.0001, "loss": 0.0246, "step": 19160 }, { "epoch": 113.4319526627219, "grad_norm": 1.4589033126831055, "learning_rate": 0.0001, "loss": 0.0244, "step": 19170 }, { "epoch": 113.49112426035504, "grad_norm": 1.9018789529800415, "learning_rate": 0.0001, "loss": 0.0258, "step": 19180 }, { "epoch": 113.55029585798816, "grad_norm": 1.5085923671722412, "learning_rate": 0.0001, "loss": 0.0261, "step": 19190 }, { "epoch": 113.6094674556213, "grad_norm": 1.0397355556488037, "learning_rate": 0.0001, "loss": 0.0271, "step": 19200 }, { "epoch": 113.66863905325444, "grad_norm": 1.6652554273605347, "learning_rate": 0.0001, "loss": 0.0265, "step": 19210 }, { "epoch": 113.72781065088758, "grad_norm": 1.445265293121338, "learning_rate": 0.0001, "loss": 0.0241, "step": 19220 }, { "epoch": 113.78698224852072, "grad_norm": 2.000033140182495, "learning_rate": 0.0001, "loss": 0.0252, "step": 19230 }, { "epoch": 113.84615384615384, "grad_norm": 1.548280119895935, "learning_rate": 0.0001, "loss": 0.0249, "step": 19240 }, { "epoch": 113.90532544378698, "grad_norm": 1.6990638971328735, "learning_rate": 0.0001, "loss": 0.024, "step": 19250 }, { "epoch": 113.96449704142012, "grad_norm": 1.7526310682296753, "learning_rate": 0.0001, "loss": 0.0249, "step": 19260 }, { "epoch": 114.02366863905326, "grad_norm": 1.5447546243667603, "learning_rate": 0.0001, "loss": 0.024, "step": 19270 }, { "epoch": 114.0828402366864, "grad_norm": 1.7399991750717163, "learning_rate": 0.0001, "loss": 0.0231, "step": 19280 }, { "epoch": 114.14201183431953, "grad_norm": 1.492555856704712, "learning_rate": 0.0001, "loss": 0.0239, "step": 19290 }, { "epoch": 114.20118343195266, "grad_norm": 1.8574066162109375, "learning_rate": 0.0001, "loss": 0.0254, "step": 19300 }, { "epoch": 114.2603550295858, "grad_norm": 1.7632348537445068, "learning_rate": 0.0001, "loss": 0.0229, "step": 19310 }, { "epoch": 114.31952662721893, "grad_norm": 1.8171968460083008, "learning_rate": 0.0001, "loss": 0.0247, "step": 19320 }, { "epoch": 114.37869822485207, "grad_norm": 1.855109691619873, "learning_rate": 0.0001, "loss": 0.0242, "step": 19330 }, { "epoch": 114.43786982248521, "grad_norm": 1.9745033979415894, "learning_rate": 0.0001, "loss": 0.0257, "step": 19340 }, { "epoch": 114.49704142011835, "grad_norm": 1.5975289344787598, "learning_rate": 0.0001, "loss": 0.0246, "step": 19350 }, { "epoch": 114.55621301775147, "grad_norm": 1.6421700716018677, "learning_rate": 0.0001, "loss": 0.0249, "step": 19360 }, { "epoch": 114.61538461538461, "grad_norm": 1.4898629188537598, "learning_rate": 0.0001, "loss": 0.0241, "step": 19370 }, { "epoch": 114.67455621301775, "grad_norm": 1.4664137363433838, "learning_rate": 0.0001, "loss": 0.0241, "step": 19380 }, { "epoch": 114.73372781065089, "grad_norm": 1.464479684829712, "learning_rate": 0.0001, "loss": 0.025, "step": 19390 }, { "epoch": 114.79289940828403, "grad_norm": 2.442481517791748, "learning_rate": 0.0001, "loss": 0.0257, "step": 19400 }, { "epoch": 114.85207100591715, "grad_norm": 1.60806143283844, "learning_rate": 0.0001, "loss": 0.0263, "step": 19410 }, { "epoch": 114.91124260355029, "grad_norm": 1.592771053314209, "learning_rate": 0.0001, "loss": 0.0269, "step": 19420 }, { "epoch": 114.97041420118343, "grad_norm": 1.271399974822998, "learning_rate": 0.0001, "loss": 0.0254, "step": 19430 }, { "epoch": 115.02958579881657, "grad_norm": 1.5474998950958252, "learning_rate": 0.0001, "loss": 0.0253, "step": 19440 }, { "epoch": 115.08875739644971, "grad_norm": 1.320244550704956, "learning_rate": 0.0001, "loss": 0.0233, "step": 19450 }, { "epoch": 115.14792899408285, "grad_norm": 1.6675114631652832, "learning_rate": 0.0001, "loss": 0.0253, "step": 19460 }, { "epoch": 115.20710059171597, "grad_norm": 1.760861873626709, "learning_rate": 0.0001, "loss": 0.0237, "step": 19470 }, { "epoch": 115.26627218934911, "grad_norm": 1.8309366703033447, "learning_rate": 0.0001, "loss": 0.0238, "step": 19480 }, { "epoch": 115.32544378698225, "grad_norm": 1.250541090965271, "learning_rate": 0.0001, "loss": 0.0258, "step": 19490 }, { "epoch": 115.38461538461539, "grad_norm": 1.5982918739318848, "learning_rate": 0.0001, "loss": 0.0258, "step": 19500 }, { "epoch": 115.44378698224853, "grad_norm": 1.6114037036895752, "learning_rate": 0.0001, "loss": 0.0253, "step": 19510 }, { "epoch": 115.50295857988165, "grad_norm": 1.600756287574768, "learning_rate": 0.0001, "loss": 0.0264, "step": 19520 }, { "epoch": 115.56213017751479, "grad_norm": 1.3714040517807007, "learning_rate": 0.0001, "loss": 0.0251, "step": 19530 }, { "epoch": 115.62130177514793, "grad_norm": 1.8954532146453857, "learning_rate": 0.0001, "loss": 0.0257, "step": 19540 }, { "epoch": 115.68047337278107, "grad_norm": 2.0173275470733643, "learning_rate": 0.0001, "loss": 0.0259, "step": 19550 }, { "epoch": 115.7396449704142, "grad_norm": 1.7344890832901, "learning_rate": 0.0001, "loss": 0.0242, "step": 19560 }, { "epoch": 115.79881656804734, "grad_norm": 1.4744304418563843, "learning_rate": 0.0001, "loss": 0.0251, "step": 19570 }, { "epoch": 115.85798816568047, "grad_norm": 1.5546965599060059, "learning_rate": 0.0001, "loss": 0.0237, "step": 19580 }, { "epoch": 115.9171597633136, "grad_norm": 1.4469677209854126, "learning_rate": 0.0001, "loss": 0.0246, "step": 19590 }, { "epoch": 115.97633136094674, "grad_norm": 1.701549768447876, "learning_rate": 0.0001, "loss": 0.0258, "step": 19600 }, { "epoch": 116.03550295857988, "grad_norm": 1.9932337999343872, "learning_rate": 0.0001, "loss": 0.0259, "step": 19610 }, { "epoch": 116.09467455621302, "grad_norm": 1.5876764059066772, "learning_rate": 0.0001, "loss": 0.0247, "step": 19620 }, { "epoch": 116.15384615384616, "grad_norm": 1.2691922187805176, "learning_rate": 0.0001, "loss": 0.0244, "step": 19630 }, { "epoch": 116.21301775147928, "grad_norm": 1.2456001043319702, "learning_rate": 0.0001, "loss": 0.0236, "step": 19640 }, { "epoch": 116.27218934911242, "grad_norm": 1.515762209892273, "learning_rate": 0.0001, "loss": 0.0238, "step": 19650 }, { "epoch": 116.33136094674556, "grad_norm": 2.003148078918457, "learning_rate": 0.0001, "loss": 0.024, "step": 19660 }, { "epoch": 116.3905325443787, "grad_norm": 1.7398144006729126, "learning_rate": 0.0001, "loss": 0.0245, "step": 19670 }, { "epoch": 116.44970414201184, "grad_norm": 1.5896118879318237, "learning_rate": 0.0001, "loss": 0.0261, "step": 19680 }, { "epoch": 116.50887573964496, "grad_norm": 1.3802510499954224, "learning_rate": 0.0001, "loss": 0.025, "step": 19690 }, { "epoch": 116.5680473372781, "grad_norm": 1.5616990327835083, "learning_rate": 0.0001, "loss": 0.0258, "step": 19700 }, { "epoch": 116.62721893491124, "grad_norm": 1.3298535346984863, "learning_rate": 0.0001, "loss": 0.0251, "step": 19710 }, { "epoch": 116.68639053254438, "grad_norm": 1.450252890586853, "learning_rate": 0.0001, "loss": 0.024, "step": 19720 }, { "epoch": 116.74556213017752, "grad_norm": 1.211578607559204, "learning_rate": 0.0001, "loss": 0.0246, "step": 19730 }, { "epoch": 116.80473372781066, "grad_norm": 1.435779094696045, "learning_rate": 0.0001, "loss": 0.0249, "step": 19740 }, { "epoch": 116.86390532544378, "grad_norm": 1.1626797914505005, "learning_rate": 0.0001, "loss": 0.0249, "step": 19750 }, { "epoch": 116.92307692307692, "grad_norm": 1.4214140176773071, "learning_rate": 0.0001, "loss": 0.0257, "step": 19760 }, { "epoch": 116.98224852071006, "grad_norm": 1.1732285022735596, "learning_rate": 0.0001, "loss": 0.0252, "step": 19770 }, { "epoch": 117.0414201183432, "grad_norm": 1.3342543840408325, "learning_rate": 0.0001, "loss": 0.0264, "step": 19780 }, { "epoch": 117.10059171597634, "grad_norm": 1.5154694318771362, "learning_rate": 0.0001, "loss": 0.0255, "step": 19790 }, { "epoch": 117.15976331360947, "grad_norm": 2.0330324172973633, "learning_rate": 0.0001, "loss": 0.0253, "step": 19800 }, { "epoch": 117.2189349112426, "grad_norm": 1.3093416690826416, "learning_rate": 0.0001, "loss": 0.0249, "step": 19810 }, { "epoch": 117.27810650887574, "grad_norm": 1.491545557975769, "learning_rate": 0.0001, "loss": 0.0247, "step": 19820 }, { "epoch": 117.33727810650888, "grad_norm": 1.8200998306274414, "learning_rate": 0.0001, "loss": 0.0255, "step": 19830 }, { "epoch": 117.39644970414201, "grad_norm": 1.3772296905517578, "learning_rate": 0.0001, "loss": 0.0246, "step": 19840 }, { "epoch": 117.45562130177515, "grad_norm": 1.596791386604309, "learning_rate": 0.0001, "loss": 0.0241, "step": 19850 }, { "epoch": 117.51479289940828, "grad_norm": 1.3623021841049194, "learning_rate": 0.0001, "loss": 0.024, "step": 19860 }, { "epoch": 117.57396449704142, "grad_norm": 1.4141747951507568, "learning_rate": 0.0001, "loss": 0.0242, "step": 19870 }, { "epoch": 117.63313609467455, "grad_norm": 1.6432946920394897, "learning_rate": 0.0001, "loss": 0.0249, "step": 19880 }, { "epoch": 117.6923076923077, "grad_norm": 1.4928805828094482, "learning_rate": 0.0001, "loss": 0.0238, "step": 19890 }, { "epoch": 117.75147928994083, "grad_norm": 1.3720744848251343, "learning_rate": 0.0001, "loss": 0.0234, "step": 19900 }, { "epoch": 117.81065088757397, "grad_norm": 1.3393833637237549, "learning_rate": 0.0001, "loss": 0.0236, "step": 19910 }, { "epoch": 117.8698224852071, "grad_norm": 1.6519807577133179, "learning_rate": 0.0001, "loss": 0.0242, "step": 19920 }, { "epoch": 117.92899408284023, "grad_norm": 1.5151872634887695, "learning_rate": 0.0001, "loss": 0.0236, "step": 19930 }, { "epoch": 117.98816568047337, "grad_norm": 1.5610730648040771, "learning_rate": 0.0001, "loss": 0.0246, "step": 19940 }, { "epoch": 118.04733727810651, "grad_norm": 1.584161400794983, "learning_rate": 0.0001, "loss": 0.0232, "step": 19950 }, { "epoch": 118.10650887573965, "grad_norm": 1.8299890756607056, "learning_rate": 0.0001, "loss": 0.0235, "step": 19960 }, { "epoch": 118.16568047337279, "grad_norm": 1.6389787197113037, "learning_rate": 0.0001, "loss": 0.0257, "step": 19970 }, { "epoch": 118.22485207100591, "grad_norm": 1.6875072717666626, "learning_rate": 0.0001, "loss": 0.0245, "step": 19980 }, { "epoch": 118.28402366863905, "grad_norm": 1.0670231580734253, "learning_rate": 0.0001, "loss": 0.0243, "step": 19990 }, { "epoch": 118.34319526627219, "grad_norm": 1.5046017169952393, "learning_rate": 0.0001, "loss": 0.0245, "step": 20000 }, { "epoch": 118.40236686390533, "grad_norm": 1.3336238861083984, "learning_rate": 0.0001, "loss": 0.0241, "step": 20010 }, { "epoch": 118.46153846153847, "grad_norm": 1.8793874979019165, "learning_rate": 0.0001, "loss": 0.0245, "step": 20020 }, { "epoch": 118.52071005917159, "grad_norm": 1.840494990348816, "learning_rate": 0.0001, "loss": 0.0243, "step": 20030 }, { "epoch": 118.57988165680473, "grad_norm": 1.8920868635177612, "learning_rate": 0.0001, "loss": 0.0247, "step": 20040 }, { "epoch": 118.63905325443787, "grad_norm": 1.702409267425537, "learning_rate": 0.0001, "loss": 0.0239, "step": 20050 }, { "epoch": 118.69822485207101, "grad_norm": 2.423318862915039, "learning_rate": 0.0001, "loss": 0.0239, "step": 20060 }, { "epoch": 118.75739644970415, "grad_norm": 1.636331558227539, "learning_rate": 0.0001, "loss": 0.0241, "step": 20070 }, { "epoch": 118.81656804733728, "grad_norm": 1.5501534938812256, "learning_rate": 0.0001, "loss": 0.0243, "step": 20080 }, { "epoch": 118.87573964497041, "grad_norm": 1.5468095541000366, "learning_rate": 0.0001, "loss": 0.024, "step": 20090 }, { "epoch": 118.93491124260355, "grad_norm": 1.9341951608657837, "learning_rate": 0.0001, "loss": 0.0242, "step": 20100 }, { "epoch": 118.99408284023669, "grad_norm": 1.5052398443222046, "learning_rate": 0.0001, "loss": 0.0244, "step": 20110 }, { "epoch": 119.05325443786982, "grad_norm": 1.9163942337036133, "learning_rate": 0.0001, "loss": 0.0256, "step": 20120 }, { "epoch": 119.11242603550296, "grad_norm": 2.711479902267456, "learning_rate": 0.0001, "loss": 0.0253, "step": 20130 }, { "epoch": 119.17159763313609, "grad_norm": 2.302992582321167, "learning_rate": 0.0001, "loss": 0.0243, "step": 20140 }, { "epoch": 119.23076923076923, "grad_norm": 2.043517827987671, "learning_rate": 0.0001, "loss": 0.0248, "step": 20150 }, { "epoch": 119.28994082840237, "grad_norm": 1.8621257543563843, "learning_rate": 0.0001, "loss": 0.0253, "step": 20160 }, { "epoch": 119.3491124260355, "grad_norm": 2.019651412963867, "learning_rate": 0.0001, "loss": 0.0241, "step": 20170 }, { "epoch": 119.40828402366864, "grad_norm": 1.751746416091919, "learning_rate": 0.0001, "loss": 0.0251, "step": 20180 }, { "epoch": 119.46745562130178, "grad_norm": 1.6983256340026855, "learning_rate": 0.0001, "loss": 0.0224, "step": 20190 }, { "epoch": 119.5266272189349, "grad_norm": 1.620203971862793, "learning_rate": 0.0001, "loss": 0.0221, "step": 20200 }, { "epoch": 119.58579881656804, "grad_norm": 1.2470026016235352, "learning_rate": 0.0001, "loss": 0.0227, "step": 20210 }, { "epoch": 119.64497041420118, "grad_norm": 1.777154564857483, "learning_rate": 0.0001, "loss": 0.0239, "step": 20220 }, { "epoch": 119.70414201183432, "grad_norm": 1.4579102993011475, "learning_rate": 0.0001, "loss": 0.023, "step": 20230 }, { "epoch": 119.76331360946746, "grad_norm": 1.4480316638946533, "learning_rate": 0.0001, "loss": 0.0245, "step": 20240 }, { "epoch": 119.8224852071006, "grad_norm": 1.5789780616760254, "learning_rate": 0.0001, "loss": 0.0244, "step": 20250 }, { "epoch": 119.88165680473372, "grad_norm": 1.66701078414917, "learning_rate": 0.0001, "loss": 0.0227, "step": 20260 }, { "epoch": 119.94082840236686, "grad_norm": 1.494914174079895, "learning_rate": 0.0001, "loss": 0.0233, "step": 20270 }, { "epoch": 120.0, "grad_norm": 1.1893357038497925, "learning_rate": 0.0001, "loss": 0.0238, "step": 20280 }, { "epoch": 120.05917159763314, "grad_norm": 1.888107180595398, "learning_rate": 0.0001, "loss": 0.0243, "step": 20290 }, { "epoch": 120.11834319526628, "grad_norm": 1.7388328313827515, "learning_rate": 0.0001, "loss": 0.0234, "step": 20300 }, { "epoch": 120.1775147928994, "grad_norm": 1.1984046697616577, "learning_rate": 0.0001, "loss": 0.0239, "step": 20310 }, { "epoch": 120.23668639053254, "grad_norm": 1.9203356504440308, "learning_rate": 0.0001, "loss": 0.025, "step": 20320 }, { "epoch": 120.29585798816568, "grad_norm": 1.3447134494781494, "learning_rate": 0.0001, "loss": 0.0232, "step": 20330 }, { "epoch": 120.35502958579882, "grad_norm": 1.7264684438705444, "learning_rate": 0.0001, "loss": 0.0233, "step": 20340 }, { "epoch": 120.41420118343196, "grad_norm": 1.4838932752609253, "learning_rate": 0.0001, "loss": 0.0243, "step": 20350 }, { "epoch": 120.4733727810651, "grad_norm": 1.2301445007324219, "learning_rate": 0.0001, "loss": 0.0235, "step": 20360 }, { "epoch": 120.53254437869822, "grad_norm": 1.72585129737854, "learning_rate": 0.0001, "loss": 0.0235, "step": 20370 }, { "epoch": 120.59171597633136, "grad_norm": 1.480351209640503, "learning_rate": 0.0001, "loss": 0.024, "step": 20380 }, { "epoch": 120.6508875739645, "grad_norm": 1.3412723541259766, "learning_rate": 0.0001, "loss": 0.0241, "step": 20390 }, { "epoch": 120.71005917159763, "grad_norm": 2.0568559169769287, "learning_rate": 0.0001, "loss": 0.0242, "step": 20400 }, { "epoch": 120.76923076923077, "grad_norm": 2.2854654788970947, "learning_rate": 0.0001, "loss": 0.0244, "step": 20410 }, { "epoch": 120.82840236686391, "grad_norm": 1.3693511486053467, "learning_rate": 0.0001, "loss": 0.0244, "step": 20420 }, { "epoch": 120.88757396449704, "grad_norm": 1.4728013277053833, "learning_rate": 0.0001, "loss": 0.0243, "step": 20430 }, { "epoch": 120.94674556213018, "grad_norm": 1.8962215185165405, "learning_rate": 0.0001, "loss": 0.0258, "step": 20440 }, { "epoch": 121.00591715976331, "grad_norm": 1.650237798690796, "learning_rate": 0.0001, "loss": 0.0236, "step": 20450 }, { "epoch": 121.06508875739645, "grad_norm": 1.7355082035064697, "learning_rate": 0.0001, "loss": 0.0241, "step": 20460 }, { "epoch": 121.12426035502959, "grad_norm": 1.8244450092315674, "learning_rate": 0.0001, "loss": 0.0243, "step": 20470 }, { "epoch": 121.18343195266272, "grad_norm": 1.3406902551651, "learning_rate": 0.0001, "loss": 0.0251, "step": 20480 }, { "epoch": 121.24260355029585, "grad_norm": 1.3993268013000488, "learning_rate": 0.0001, "loss": 0.0242, "step": 20490 }, { "epoch": 121.30177514792899, "grad_norm": 1.1069599390029907, "learning_rate": 0.0001, "loss": 0.0229, "step": 20500 }, { "epoch": 121.36094674556213, "grad_norm": 1.293418288230896, "learning_rate": 0.0001, "loss": 0.0243, "step": 20510 }, { "epoch": 121.42011834319527, "grad_norm": 1.6708757877349854, "learning_rate": 0.0001, "loss": 0.0245, "step": 20520 }, { "epoch": 121.47928994082841, "grad_norm": 1.4860033988952637, "learning_rate": 0.0001, "loss": 0.0246, "step": 20530 }, { "epoch": 121.53846153846153, "grad_norm": 1.4900178909301758, "learning_rate": 0.0001, "loss": 0.024, "step": 20540 }, { "epoch": 121.59763313609467, "grad_norm": 1.4090445041656494, "learning_rate": 0.0001, "loss": 0.0245, "step": 20550 }, { "epoch": 121.65680473372781, "grad_norm": 1.229798674583435, "learning_rate": 0.0001, "loss": 0.0251, "step": 20560 }, { "epoch": 121.71597633136095, "grad_norm": 1.7193299531936646, "learning_rate": 0.0001, "loss": 0.0244, "step": 20570 }, { "epoch": 121.77514792899409, "grad_norm": 1.8368279933929443, "learning_rate": 0.0001, "loss": 0.0235, "step": 20580 }, { "epoch": 121.83431952662721, "grad_norm": 2.168823719024658, "learning_rate": 0.0001, "loss": 0.025, "step": 20590 }, { "epoch": 121.89349112426035, "grad_norm": 1.355351209640503, "learning_rate": 0.0001, "loss": 0.0249, "step": 20600 }, { "epoch": 121.95266272189349, "grad_norm": 1.4996570348739624, "learning_rate": 0.0001, "loss": 0.0251, "step": 20610 }, { "epoch": 122.01183431952663, "grad_norm": 1.2516422271728516, "learning_rate": 0.0001, "loss": 0.0237, "step": 20620 }, { "epoch": 122.07100591715977, "grad_norm": 1.4136583805084229, "learning_rate": 0.0001, "loss": 0.024, "step": 20630 }, { "epoch": 122.1301775147929, "grad_norm": 1.6435188055038452, "learning_rate": 0.0001, "loss": 0.0251, "step": 20640 }, { "epoch": 122.18934911242603, "grad_norm": 1.33233642578125, "learning_rate": 0.0001, "loss": 0.0246, "step": 20650 }, { "epoch": 122.24852071005917, "grad_norm": 1.2726895809173584, "learning_rate": 0.0001, "loss": 0.025, "step": 20660 }, { "epoch": 122.3076923076923, "grad_norm": 1.6699985265731812, "learning_rate": 0.0001, "loss": 0.0245, "step": 20670 }, { "epoch": 122.36686390532545, "grad_norm": 1.9436041116714478, "learning_rate": 0.0001, "loss": 0.0249, "step": 20680 }, { "epoch": 122.42603550295858, "grad_norm": 1.5204527378082275, "learning_rate": 0.0001, "loss": 0.0251, "step": 20690 }, { "epoch": 122.48520710059172, "grad_norm": 1.2812682390213013, "learning_rate": 0.0001, "loss": 0.0248, "step": 20700 }, { "epoch": 122.54437869822485, "grad_norm": 1.337295651435852, "learning_rate": 0.0001, "loss": 0.0238, "step": 20710 }, { "epoch": 122.60355029585799, "grad_norm": 1.46759831905365, "learning_rate": 0.0001, "loss": 0.0255, "step": 20720 }, { "epoch": 122.66272189349112, "grad_norm": 1.7598927021026611, "learning_rate": 0.0001, "loss": 0.0251, "step": 20730 }, { "epoch": 122.72189349112426, "grad_norm": 1.9579120874404907, "learning_rate": 0.0001, "loss": 0.0259, "step": 20740 }, { "epoch": 122.7810650887574, "grad_norm": 1.47304105758667, "learning_rate": 0.0001, "loss": 0.0236, "step": 20750 }, { "epoch": 122.84023668639053, "grad_norm": 2.0689642429351807, "learning_rate": 0.0001, "loss": 0.0246, "step": 20760 }, { "epoch": 122.89940828402366, "grad_norm": 1.569043517112732, "learning_rate": 0.0001, "loss": 0.0246, "step": 20770 }, { "epoch": 122.9585798816568, "grad_norm": 1.383862018585205, "learning_rate": 0.0001, "loss": 0.0238, "step": 20780 }, { "epoch": 123.01775147928994, "grad_norm": 1.2182120084762573, "learning_rate": 0.0001, "loss": 0.025, "step": 20790 }, { "epoch": 123.07692307692308, "grad_norm": 1.7168728113174438, "learning_rate": 0.0001, "loss": 0.0244, "step": 20800 }, { "epoch": 123.13609467455622, "grad_norm": 1.7645381689071655, "learning_rate": 0.0001, "loss": 0.0245, "step": 20810 }, { "epoch": 123.19526627218934, "grad_norm": 1.8119664192199707, "learning_rate": 0.0001, "loss": 0.0253, "step": 20820 }, { "epoch": 123.25443786982248, "grad_norm": 1.5536601543426514, "learning_rate": 0.0001, "loss": 0.0245, "step": 20830 }, { "epoch": 123.31360946745562, "grad_norm": 1.643082618713379, "learning_rate": 0.0001, "loss": 0.0257, "step": 20840 }, { "epoch": 123.37278106508876, "grad_norm": 1.1931909322738647, "learning_rate": 0.0001, "loss": 0.0239, "step": 20850 }, { "epoch": 123.4319526627219, "grad_norm": 1.4043633937835693, "learning_rate": 0.0001, "loss": 0.023, "step": 20860 }, { "epoch": 123.49112426035504, "grad_norm": 1.7250332832336426, "learning_rate": 0.0001, "loss": 0.0238, "step": 20870 }, { "epoch": 123.55029585798816, "grad_norm": 1.6145695447921753, "learning_rate": 0.0001, "loss": 0.0248, "step": 20880 }, { "epoch": 123.6094674556213, "grad_norm": 2.091442584991455, "learning_rate": 0.0001, "loss": 0.0232, "step": 20890 }, { "epoch": 123.66863905325444, "grad_norm": 1.4025990962982178, "learning_rate": 0.0001, "loss": 0.0244, "step": 20900 }, { "epoch": 123.72781065088758, "grad_norm": 2.0694684982299805, "learning_rate": 0.0001, "loss": 0.0247, "step": 20910 }, { "epoch": 123.78698224852072, "grad_norm": 1.3531553745269775, "learning_rate": 0.0001, "loss": 0.024, "step": 20920 }, { "epoch": 123.84615384615384, "grad_norm": 1.7315845489501953, "learning_rate": 0.0001, "loss": 0.0238, "step": 20930 }, { "epoch": 123.90532544378698, "grad_norm": 1.3910056352615356, "learning_rate": 0.0001, "loss": 0.0241, "step": 20940 }, { "epoch": 123.96449704142012, "grad_norm": 1.7779054641723633, "learning_rate": 0.0001, "loss": 0.0237, "step": 20950 }, { "epoch": 124.02366863905326, "grad_norm": 2.0063276290893555, "learning_rate": 0.0001, "loss": 0.0242, "step": 20960 }, { "epoch": 124.0828402366864, "grad_norm": 1.5343997478485107, "learning_rate": 0.0001, "loss": 0.024, "step": 20970 }, { "epoch": 124.14201183431953, "grad_norm": 2.0881474018096924, "learning_rate": 0.0001, "loss": 0.0241, "step": 20980 }, { "epoch": 124.20118343195266, "grad_norm": 1.090989112854004, "learning_rate": 0.0001, "loss": 0.0238, "step": 20990 }, { "epoch": 124.2603550295858, "grad_norm": 1.8592829704284668, "learning_rate": 0.0001, "loss": 0.0246, "step": 21000 }, { "epoch": 124.31952662721893, "grad_norm": 1.451992154121399, "learning_rate": 0.0001, "loss": 0.0235, "step": 21010 }, { "epoch": 124.37869822485207, "grad_norm": 1.3477333784103394, "learning_rate": 0.0001, "loss": 0.0252, "step": 21020 }, { "epoch": 124.43786982248521, "grad_norm": 1.4557844400405884, "learning_rate": 0.0001, "loss": 0.025, "step": 21030 }, { "epoch": 124.49704142011835, "grad_norm": 2.204045295715332, "learning_rate": 0.0001, "loss": 0.0244, "step": 21040 }, { "epoch": 124.55621301775147, "grad_norm": 1.6557952165603638, "learning_rate": 0.0001, "loss": 0.024, "step": 21050 }, { "epoch": 124.61538461538461, "grad_norm": 1.356551170349121, "learning_rate": 0.0001, "loss": 0.0237, "step": 21060 }, { "epoch": 124.67455621301775, "grad_norm": 1.4055249691009521, "learning_rate": 0.0001, "loss": 0.0235, "step": 21070 }, { "epoch": 124.73372781065089, "grad_norm": 1.1654703617095947, "learning_rate": 0.0001, "loss": 0.024, "step": 21080 }, { "epoch": 124.79289940828403, "grad_norm": 1.2038614749908447, "learning_rate": 0.0001, "loss": 0.0227, "step": 21090 }, { "epoch": 124.85207100591715, "grad_norm": 1.262635588645935, "learning_rate": 0.0001, "loss": 0.0234, "step": 21100 }, { "epoch": 124.91124260355029, "grad_norm": 1.4357316493988037, "learning_rate": 0.0001, "loss": 0.0251, "step": 21110 }, { "epoch": 124.97041420118343, "grad_norm": 1.2594584226608276, "learning_rate": 0.0001, "loss": 0.0239, "step": 21120 }, { "epoch": 125.02958579881657, "grad_norm": 1.2901742458343506, "learning_rate": 0.0001, "loss": 0.0255, "step": 21130 }, { "epoch": 125.08875739644971, "grad_norm": 1.1420270204544067, "learning_rate": 0.0001, "loss": 0.024, "step": 21140 }, { "epoch": 125.14792899408285, "grad_norm": 1.5096663236618042, "learning_rate": 0.0001, "loss": 0.025, "step": 21150 }, { "epoch": 125.20710059171597, "grad_norm": 1.262708067893982, "learning_rate": 0.0001, "loss": 0.0224, "step": 21160 }, { "epoch": 125.26627218934911, "grad_norm": 1.5808833837509155, "learning_rate": 0.0001, "loss": 0.0235, "step": 21170 }, { "epoch": 125.32544378698225, "grad_norm": 1.7921563386917114, "learning_rate": 0.0001, "loss": 0.0254, "step": 21180 }, { "epoch": 125.38461538461539, "grad_norm": 1.5138548612594604, "learning_rate": 0.0001, "loss": 0.0244, "step": 21190 }, { "epoch": 125.44378698224853, "grad_norm": 1.5701154470443726, "learning_rate": 0.0001, "loss": 0.0232, "step": 21200 }, { "epoch": 125.50295857988165, "grad_norm": 1.4337513446807861, "learning_rate": 0.0001, "loss": 0.0243, "step": 21210 }, { "epoch": 125.56213017751479, "grad_norm": 1.5205641984939575, "learning_rate": 0.0001, "loss": 0.0239, "step": 21220 }, { "epoch": 125.62130177514793, "grad_norm": 1.2746695280075073, "learning_rate": 0.0001, "loss": 0.0236, "step": 21230 }, { "epoch": 125.68047337278107, "grad_norm": 1.3662751913070679, "learning_rate": 0.0001, "loss": 0.0236, "step": 21240 }, { "epoch": 125.7396449704142, "grad_norm": 1.374159812927246, "learning_rate": 0.0001, "loss": 0.025, "step": 21250 }, { "epoch": 125.79881656804734, "grad_norm": 1.3573168516159058, "learning_rate": 0.0001, "loss": 0.0246, "step": 21260 }, { "epoch": 125.85798816568047, "grad_norm": 1.5369594097137451, "learning_rate": 0.0001, "loss": 0.0252, "step": 21270 }, { "epoch": 125.9171597633136, "grad_norm": 1.767720103263855, "learning_rate": 0.0001, "loss": 0.0242, "step": 21280 }, { "epoch": 125.97633136094674, "grad_norm": 1.8196474313735962, "learning_rate": 0.0001, "loss": 0.0241, "step": 21290 }, { "epoch": 126.03550295857988, "grad_norm": 1.566155195236206, "learning_rate": 0.0001, "loss": 0.0236, "step": 21300 }, { "epoch": 126.09467455621302, "grad_norm": 1.7110573053359985, "learning_rate": 0.0001, "loss": 0.0245, "step": 21310 }, { "epoch": 126.15384615384616, "grad_norm": 1.0103895664215088, "learning_rate": 0.0001, "loss": 0.0227, "step": 21320 }, { "epoch": 126.21301775147928, "grad_norm": 1.5894798040390015, "learning_rate": 0.0001, "loss": 0.0233, "step": 21330 }, { "epoch": 126.27218934911242, "grad_norm": 2.3615386486053467, "learning_rate": 0.0001, "loss": 0.0237, "step": 21340 }, { "epoch": 126.33136094674556, "grad_norm": 2.0063915252685547, "learning_rate": 0.0001, "loss": 0.0239, "step": 21350 }, { "epoch": 126.3905325443787, "grad_norm": 1.7906142473220825, "learning_rate": 0.0001, "loss": 0.0231, "step": 21360 }, { "epoch": 126.44970414201184, "grad_norm": 1.3752309083938599, "learning_rate": 0.0001, "loss": 0.0238, "step": 21370 }, { "epoch": 126.50887573964496, "grad_norm": 1.958687424659729, "learning_rate": 0.0001, "loss": 0.0239, "step": 21380 }, { "epoch": 126.5680473372781, "grad_norm": 1.340692400932312, "learning_rate": 0.0001, "loss": 0.0241, "step": 21390 }, { "epoch": 126.62721893491124, "grad_norm": 2.1110787391662598, "learning_rate": 0.0001, "loss": 0.0244, "step": 21400 }, { "epoch": 126.68639053254438, "grad_norm": 1.5755714178085327, "learning_rate": 0.0001, "loss": 0.024, "step": 21410 }, { "epoch": 126.74556213017752, "grad_norm": 1.8172893524169922, "learning_rate": 0.0001, "loss": 0.0242, "step": 21420 }, { "epoch": 126.80473372781066, "grad_norm": 1.704957127571106, "learning_rate": 0.0001, "loss": 0.0236, "step": 21430 }, { "epoch": 126.86390532544378, "grad_norm": 1.7099320888519287, "learning_rate": 0.0001, "loss": 0.0227, "step": 21440 }, { "epoch": 126.92307692307692, "grad_norm": 1.8021202087402344, "learning_rate": 0.0001, "loss": 0.0229, "step": 21450 }, { "epoch": 126.98224852071006, "grad_norm": 1.4481946229934692, "learning_rate": 0.0001, "loss": 0.025, "step": 21460 }, { "epoch": 127.0414201183432, "grad_norm": 1.306019902229309, "learning_rate": 0.0001, "loss": 0.0227, "step": 21470 }, { "epoch": 127.10059171597634, "grad_norm": 1.3825839757919312, "learning_rate": 0.0001, "loss": 0.0235, "step": 21480 }, { "epoch": 127.15976331360947, "grad_norm": 1.1438040733337402, "learning_rate": 0.0001, "loss": 0.0235, "step": 21490 }, { "epoch": 127.2189349112426, "grad_norm": 1.3650189638137817, "learning_rate": 0.0001, "loss": 0.0234, "step": 21500 }, { "epoch": 127.27810650887574, "grad_norm": 1.407728910446167, "learning_rate": 0.0001, "loss": 0.0235, "step": 21510 }, { "epoch": 127.33727810650888, "grad_norm": 1.390440821647644, "learning_rate": 0.0001, "loss": 0.0251, "step": 21520 }, { "epoch": 127.39644970414201, "grad_norm": 1.4849590063095093, "learning_rate": 0.0001, "loss": 0.0243, "step": 21530 }, { "epoch": 127.45562130177515, "grad_norm": 1.3170475959777832, "learning_rate": 0.0001, "loss": 0.0239, "step": 21540 }, { "epoch": 127.51479289940828, "grad_norm": 1.319886326789856, "learning_rate": 0.0001, "loss": 0.0241, "step": 21550 }, { "epoch": 127.57396449704142, "grad_norm": 2.4190988540649414, "learning_rate": 0.0001, "loss": 0.0231, "step": 21560 }, { "epoch": 127.63313609467455, "grad_norm": 1.246935486793518, "learning_rate": 0.0001, "loss": 0.0246, "step": 21570 }, { "epoch": 127.6923076923077, "grad_norm": 1.0962272882461548, "learning_rate": 0.0001, "loss": 0.0228, "step": 21580 }, { "epoch": 127.75147928994083, "grad_norm": 1.5150156021118164, "learning_rate": 0.0001, "loss": 0.0238, "step": 21590 }, { "epoch": 127.81065088757397, "grad_norm": 1.0444529056549072, "learning_rate": 0.0001, "loss": 0.0227, "step": 21600 }, { "epoch": 127.8698224852071, "grad_norm": 1.2725692987442017, "learning_rate": 0.0001, "loss": 0.0241, "step": 21610 }, { "epoch": 127.92899408284023, "grad_norm": 1.4964576959609985, "learning_rate": 0.0001, "loss": 0.024, "step": 21620 }, { "epoch": 127.98816568047337, "grad_norm": 1.3288843631744385, "learning_rate": 0.0001, "loss": 0.0229, "step": 21630 }, { "epoch": 128.0473372781065, "grad_norm": 1.3586474657058716, "learning_rate": 0.0001, "loss": 0.0241, "step": 21640 }, { "epoch": 128.10650887573965, "grad_norm": 2.1283700466156006, "learning_rate": 0.0001, "loss": 0.0237, "step": 21650 }, { "epoch": 128.1656804733728, "grad_norm": 2.1375021934509277, "learning_rate": 0.0001, "loss": 0.0239, "step": 21660 }, { "epoch": 128.22485207100593, "grad_norm": 1.7728155851364136, "learning_rate": 0.0001, "loss": 0.0242, "step": 21670 }, { "epoch": 128.28402366863907, "grad_norm": 1.4814608097076416, "learning_rate": 0.0001, "loss": 0.025, "step": 21680 }, { "epoch": 128.34319526627218, "grad_norm": 1.8674674034118652, "learning_rate": 0.0001, "loss": 0.0237, "step": 21690 }, { "epoch": 128.4023668639053, "grad_norm": 1.34758460521698, "learning_rate": 0.0001, "loss": 0.0232, "step": 21700 }, { "epoch": 128.46153846153845, "grad_norm": 1.4363784790039062, "learning_rate": 0.0001, "loss": 0.0243, "step": 21710 }, { "epoch": 128.5207100591716, "grad_norm": 1.5251965522766113, "learning_rate": 0.0001, "loss": 0.023, "step": 21720 }, { "epoch": 128.57988165680473, "grad_norm": 1.375557780265808, "learning_rate": 0.0001, "loss": 0.0233, "step": 21730 }, { "epoch": 128.63905325443787, "grad_norm": 1.6310837268829346, "learning_rate": 0.0001, "loss": 0.0235, "step": 21740 }, { "epoch": 128.698224852071, "grad_norm": 1.4078294038772583, "learning_rate": 0.0001, "loss": 0.0236, "step": 21750 }, { "epoch": 128.75739644970415, "grad_norm": 1.5379143953323364, "learning_rate": 0.0001, "loss": 0.0233, "step": 21760 }, { "epoch": 128.81656804733728, "grad_norm": 1.5616847276687622, "learning_rate": 0.0001, "loss": 0.0227, "step": 21770 }, { "epoch": 128.87573964497042, "grad_norm": 1.3290719985961914, "learning_rate": 0.0001, "loss": 0.0245, "step": 21780 }, { "epoch": 128.93491124260356, "grad_norm": 1.836195468902588, "learning_rate": 0.0001, "loss": 0.0234, "step": 21790 }, { "epoch": 128.9940828402367, "grad_norm": 2.794248342514038, "learning_rate": 0.0001, "loss": 0.0245, "step": 21800 }, { "epoch": 129.0532544378698, "grad_norm": 1.3407576084136963, "learning_rate": 0.0001, "loss": 0.0252, "step": 21810 }, { "epoch": 129.11242603550295, "grad_norm": 1.259130597114563, "learning_rate": 0.0001, "loss": 0.0238, "step": 21820 }, { "epoch": 129.1715976331361, "grad_norm": 1.6667972803115845, "learning_rate": 0.0001, "loss": 0.0237, "step": 21830 }, { "epoch": 129.23076923076923, "grad_norm": 1.5170016288757324, "learning_rate": 0.0001, "loss": 0.0235, "step": 21840 }, { "epoch": 129.28994082840237, "grad_norm": 1.391515851020813, "learning_rate": 0.0001, "loss": 0.0246, "step": 21850 }, { "epoch": 129.3491124260355, "grad_norm": 1.2307288646697998, "learning_rate": 0.0001, "loss": 0.0239, "step": 21860 }, { "epoch": 129.40828402366864, "grad_norm": 1.196203351020813, "learning_rate": 0.0001, "loss": 0.0244, "step": 21870 }, { "epoch": 129.46745562130178, "grad_norm": 1.090560793876648, "learning_rate": 0.0001, "loss": 0.0229, "step": 21880 }, { "epoch": 129.52662721893492, "grad_norm": 1.3238881826400757, "learning_rate": 0.0001, "loss": 0.022, "step": 21890 }, { "epoch": 129.58579881656806, "grad_norm": 1.794948697090149, "learning_rate": 0.0001, "loss": 0.0233, "step": 21900 }, { "epoch": 129.6449704142012, "grad_norm": 1.1953095197677612, "learning_rate": 0.0001, "loss": 0.0223, "step": 21910 }, { "epoch": 129.7041420118343, "grad_norm": 1.3214365243911743, "learning_rate": 0.0001, "loss": 0.0225, "step": 21920 }, { "epoch": 129.76331360946745, "grad_norm": 1.6785128116607666, "learning_rate": 0.0001, "loss": 0.0228, "step": 21930 }, { "epoch": 129.82248520710058, "grad_norm": 1.9589238166809082, "learning_rate": 0.0001, "loss": 0.023, "step": 21940 }, { "epoch": 129.88165680473372, "grad_norm": 1.3584564924240112, "learning_rate": 0.0001, "loss": 0.0231, "step": 21950 }, { "epoch": 129.94082840236686, "grad_norm": 1.8160459995269775, "learning_rate": 0.0001, "loss": 0.0238, "step": 21960 }, { "epoch": 130.0, "grad_norm": 1.6693834066390991, "learning_rate": 0.0001, "loss": 0.0236, "step": 21970 }, { "epoch": 130.05917159763314, "grad_norm": 1.3691976070404053, "learning_rate": 0.0001, "loss": 0.0228, "step": 21980 }, { "epoch": 130.11834319526628, "grad_norm": 1.6098825931549072, "learning_rate": 0.0001, "loss": 0.0234, "step": 21990 }, { "epoch": 130.17751479289942, "grad_norm": 1.588507056236267, "learning_rate": 0.0001, "loss": 0.0241, "step": 22000 }, { "epoch": 130.23668639053255, "grad_norm": 1.537980318069458, "learning_rate": 0.0001, "loss": 0.0242, "step": 22010 }, { "epoch": 130.2958579881657, "grad_norm": 1.6927415132522583, "learning_rate": 0.0001, "loss": 0.0236, "step": 22020 }, { "epoch": 130.3550295857988, "grad_norm": 1.6545333862304688, "learning_rate": 0.0001, "loss": 0.0235, "step": 22030 }, { "epoch": 130.41420118343194, "grad_norm": 1.5116757154464722, "learning_rate": 0.0001, "loss": 0.0227, "step": 22040 }, { "epoch": 130.47337278106508, "grad_norm": 1.3643717765808105, "learning_rate": 0.0001, "loss": 0.0241, "step": 22050 }, { "epoch": 130.53254437869822, "grad_norm": 1.5945848226547241, "learning_rate": 0.0001, "loss": 0.0234, "step": 22060 }, { "epoch": 130.59171597633136, "grad_norm": 1.0960053205490112, "learning_rate": 0.0001, "loss": 0.0233, "step": 22070 }, { "epoch": 130.6508875739645, "grad_norm": 1.3388361930847168, "learning_rate": 0.0001, "loss": 0.0233, "step": 22080 }, { "epoch": 130.71005917159763, "grad_norm": 1.3517833948135376, "learning_rate": 0.0001, "loss": 0.0241, "step": 22090 }, { "epoch": 130.76923076923077, "grad_norm": 1.1483101844787598, "learning_rate": 0.0001, "loss": 0.0235, "step": 22100 }, { "epoch": 130.8284023668639, "grad_norm": 1.757896065711975, "learning_rate": 0.0001, "loss": 0.0245, "step": 22110 }, { "epoch": 130.88757396449705, "grad_norm": 1.8782812356948853, "learning_rate": 0.0001, "loss": 0.025, "step": 22120 }, { "epoch": 130.9467455621302, "grad_norm": 2.0116934776306152, "learning_rate": 0.0001, "loss": 0.0235, "step": 22130 }, { "epoch": 131.0059171597633, "grad_norm": 1.3743361234664917, "learning_rate": 0.0001, "loss": 0.0239, "step": 22140 }, { "epoch": 131.06508875739644, "grad_norm": 1.1905078887939453, "learning_rate": 0.0001, "loss": 0.0239, "step": 22150 }, { "epoch": 131.12426035502958, "grad_norm": 1.214572548866272, "learning_rate": 0.0001, "loss": 0.0239, "step": 22160 }, { "epoch": 131.18343195266272, "grad_norm": 1.0491257905960083, "learning_rate": 0.0001, "loss": 0.0229, "step": 22170 }, { "epoch": 131.24260355029585, "grad_norm": 1.4397287368774414, "learning_rate": 0.0001, "loss": 0.0243, "step": 22180 }, { "epoch": 131.301775147929, "grad_norm": 1.3486584424972534, "learning_rate": 0.0001, "loss": 0.0232, "step": 22190 }, { "epoch": 131.36094674556213, "grad_norm": 1.691930890083313, "learning_rate": 0.0001, "loss": 0.0246, "step": 22200 }, { "epoch": 131.42011834319527, "grad_norm": 1.3248236179351807, "learning_rate": 0.0001, "loss": 0.0251, "step": 22210 }, { "epoch": 131.4792899408284, "grad_norm": 0.9425277709960938, "learning_rate": 0.0001, "loss": 0.023, "step": 22220 }, { "epoch": 131.53846153846155, "grad_norm": 1.3046852350234985, "learning_rate": 0.0001, "loss": 0.0242, "step": 22230 }, { "epoch": 131.5976331360947, "grad_norm": 1.652566909790039, "learning_rate": 0.0001, "loss": 0.0245, "step": 22240 }, { "epoch": 131.65680473372782, "grad_norm": 1.423359751701355, "learning_rate": 0.0001, "loss": 0.0236, "step": 22250 }, { "epoch": 131.71597633136093, "grad_norm": 1.549715518951416, "learning_rate": 0.0001, "loss": 0.0243, "step": 22260 }, { "epoch": 131.77514792899407, "grad_norm": 1.877761960029602, "learning_rate": 0.0001, "loss": 0.0249, "step": 22270 }, { "epoch": 131.8343195266272, "grad_norm": 1.6790283918380737, "learning_rate": 0.0001, "loss": 0.0237, "step": 22280 }, { "epoch": 131.89349112426035, "grad_norm": 1.672221302986145, "learning_rate": 0.0001, "loss": 0.0232, "step": 22290 }, { "epoch": 131.9526627218935, "grad_norm": 1.9710195064544678, "learning_rate": 0.0001, "loss": 0.0247, "step": 22300 }, { "epoch": 132.01183431952663, "grad_norm": 2.183898687362671, "learning_rate": 0.0001, "loss": 0.0247, "step": 22310 }, { "epoch": 132.07100591715977, "grad_norm": 2.1521010398864746, "learning_rate": 0.0001, "loss": 0.0238, "step": 22320 }, { "epoch": 132.1301775147929, "grad_norm": 1.191962718963623, "learning_rate": 0.0001, "loss": 0.0229, "step": 22330 }, { "epoch": 132.18934911242604, "grad_norm": 1.4443455934524536, "learning_rate": 0.0001, "loss": 0.0232, "step": 22340 }, { "epoch": 132.24852071005918, "grad_norm": 1.250766634941101, "learning_rate": 0.0001, "loss": 0.0219, "step": 22350 }, { "epoch": 132.30769230769232, "grad_norm": 1.9157243967056274, "learning_rate": 0.0001, "loss": 0.0234, "step": 22360 }, { "epoch": 132.36686390532543, "grad_norm": 1.6203584671020508, "learning_rate": 0.0001, "loss": 0.0225, "step": 22370 }, { "epoch": 132.42603550295857, "grad_norm": 1.9990142583847046, "learning_rate": 0.0001, "loss": 0.0225, "step": 22380 }, { "epoch": 132.4852071005917, "grad_norm": 1.636446237564087, "learning_rate": 0.0001, "loss": 0.0233, "step": 22390 }, { "epoch": 132.54437869822485, "grad_norm": 1.6315312385559082, "learning_rate": 0.0001, "loss": 0.0234, "step": 22400 }, { "epoch": 132.60355029585799, "grad_norm": 1.6616979837417603, "learning_rate": 0.0001, "loss": 0.0238, "step": 22410 }, { "epoch": 132.66272189349112, "grad_norm": 1.5046131610870361, "learning_rate": 0.0001, "loss": 0.0232, "step": 22420 }, { "epoch": 132.72189349112426, "grad_norm": 1.9462499618530273, "learning_rate": 0.0001, "loss": 0.0227, "step": 22430 }, { "epoch": 132.7810650887574, "grad_norm": 1.6640390157699585, "learning_rate": 0.0001, "loss": 0.0232, "step": 22440 }, { "epoch": 132.84023668639054, "grad_norm": 1.5582449436187744, "learning_rate": 0.0001, "loss": 0.0247, "step": 22450 }, { "epoch": 132.89940828402368, "grad_norm": 2.039883613586426, "learning_rate": 0.0001, "loss": 0.0227, "step": 22460 }, { "epoch": 132.95857988165682, "grad_norm": 1.1407829523086548, "learning_rate": 0.0001, "loss": 0.0225, "step": 22470 }, { "epoch": 133.01775147928993, "grad_norm": 1.611054539680481, "learning_rate": 0.0001, "loss": 0.0232, "step": 22480 }, { "epoch": 133.07692307692307, "grad_norm": 1.5888770818710327, "learning_rate": 0.0001, "loss": 0.0227, "step": 22490 }, { "epoch": 133.1360946745562, "grad_norm": 1.1413401365280151, "learning_rate": 0.0001, "loss": 0.0238, "step": 22500 }, { "epoch": 133.19526627218934, "grad_norm": 1.2442901134490967, "learning_rate": 0.0001, "loss": 0.0239, "step": 22510 }, { "epoch": 133.25443786982248, "grad_norm": 1.790948510169983, "learning_rate": 0.0001, "loss": 0.0248, "step": 22520 }, { "epoch": 133.31360946745562, "grad_norm": 1.4156562089920044, "learning_rate": 0.0001, "loss": 0.024, "step": 22530 }, { "epoch": 133.37278106508876, "grad_norm": 1.2033716440200806, "learning_rate": 0.0001, "loss": 0.0221, "step": 22540 }, { "epoch": 133.4319526627219, "grad_norm": 1.1005910634994507, "learning_rate": 0.0001, "loss": 0.0236, "step": 22550 }, { "epoch": 133.49112426035504, "grad_norm": 1.6103978157043457, "learning_rate": 0.0001, "loss": 0.0246, "step": 22560 }, { "epoch": 133.55029585798817, "grad_norm": 1.3207180500030518, "learning_rate": 0.0001, "loss": 0.0233, "step": 22570 }, { "epoch": 133.6094674556213, "grad_norm": 1.6736252307891846, "learning_rate": 0.0001, "loss": 0.0234, "step": 22580 }, { "epoch": 133.66863905325442, "grad_norm": 1.9210213422775269, "learning_rate": 0.0001, "loss": 0.0248, "step": 22590 }, { "epoch": 133.72781065088756, "grad_norm": 2.080117702484131, "learning_rate": 0.0001, "loss": 0.0255, "step": 22600 }, { "epoch": 133.7869822485207, "grad_norm": 1.3049381971359253, "learning_rate": 0.0001, "loss": 0.0241, "step": 22610 }, { "epoch": 133.84615384615384, "grad_norm": 2.0115435123443604, "learning_rate": 0.0001, "loss": 0.0242, "step": 22620 }, { "epoch": 133.90532544378698, "grad_norm": 1.686875581741333, "learning_rate": 0.0001, "loss": 0.0238, "step": 22630 }, { "epoch": 133.96449704142012, "grad_norm": 1.0315933227539062, "learning_rate": 0.0001, "loss": 0.0234, "step": 22640 }, { "epoch": 134.02366863905326, "grad_norm": 1.2340424060821533, "learning_rate": 0.0001, "loss": 0.0223, "step": 22650 }, { "epoch": 134.0828402366864, "grad_norm": 1.4103429317474365, "learning_rate": 0.0001, "loss": 0.0231, "step": 22660 }, { "epoch": 134.14201183431953, "grad_norm": 1.297670841217041, "learning_rate": 0.0001, "loss": 0.0227, "step": 22670 }, { "epoch": 134.20118343195267, "grad_norm": 1.2303435802459717, "learning_rate": 0.0001, "loss": 0.0228, "step": 22680 }, { "epoch": 134.2603550295858, "grad_norm": 1.421676516532898, "learning_rate": 0.0001, "loss": 0.0231, "step": 22690 }, { "epoch": 134.31952662721895, "grad_norm": 1.498890995979309, "learning_rate": 0.0001, "loss": 0.0234, "step": 22700 }, { "epoch": 134.37869822485206, "grad_norm": 1.9806216955184937, "learning_rate": 0.0001, "loss": 0.0233, "step": 22710 }, { "epoch": 134.4378698224852, "grad_norm": 1.2969321012496948, "learning_rate": 0.0001, "loss": 0.0236, "step": 22720 }, { "epoch": 134.49704142011834, "grad_norm": 1.4738191366195679, "learning_rate": 0.0001, "loss": 0.023, "step": 22730 }, { "epoch": 134.55621301775147, "grad_norm": 1.2421598434448242, "learning_rate": 0.0001, "loss": 0.022, "step": 22740 }, { "epoch": 134.6153846153846, "grad_norm": 1.070454478263855, "learning_rate": 0.0001, "loss": 0.0213, "step": 22750 }, { "epoch": 134.67455621301775, "grad_norm": 2.3466315269470215, "learning_rate": 0.0001, "loss": 0.0221, "step": 22760 }, { "epoch": 134.7337278106509, "grad_norm": 1.8768153190612793, "learning_rate": 0.0001, "loss": 0.0231, "step": 22770 }, { "epoch": 134.79289940828403, "grad_norm": 1.7109942436218262, "learning_rate": 0.0001, "loss": 0.0221, "step": 22780 }, { "epoch": 134.85207100591717, "grad_norm": 1.6232916116714478, "learning_rate": 0.0001, "loss": 0.0226, "step": 22790 }, { "epoch": 134.9112426035503, "grad_norm": 1.238905906677246, "learning_rate": 0.0001, "loss": 0.0219, "step": 22800 }, { "epoch": 134.97041420118344, "grad_norm": 1.4628806114196777, "learning_rate": 0.0001, "loss": 0.0216, "step": 22810 }, { "epoch": 135.02958579881656, "grad_norm": 1.3500521183013916, "learning_rate": 0.0001, "loss": 0.0226, "step": 22820 }, { "epoch": 135.0887573964497, "grad_norm": 1.1362117528915405, "learning_rate": 0.0001, "loss": 0.0223, "step": 22830 }, { "epoch": 135.14792899408283, "grad_norm": 1.6222641468048096, "learning_rate": 0.0001, "loss": 0.0233, "step": 22840 }, { "epoch": 135.20710059171597, "grad_norm": 1.4515498876571655, "learning_rate": 0.0001, "loss": 0.0222, "step": 22850 }, { "epoch": 135.2662721893491, "grad_norm": 1.3743890523910522, "learning_rate": 0.0001, "loss": 0.0228, "step": 22860 }, { "epoch": 135.32544378698225, "grad_norm": 1.2989169359207153, "learning_rate": 0.0001, "loss": 0.0225, "step": 22870 }, { "epoch": 135.3846153846154, "grad_norm": 1.4397659301757812, "learning_rate": 0.0001, "loss": 0.0225, "step": 22880 }, { "epoch": 135.44378698224853, "grad_norm": 1.3164410591125488, "learning_rate": 0.0001, "loss": 0.0221, "step": 22890 }, { "epoch": 135.50295857988166, "grad_norm": 1.5367070436477661, "learning_rate": 0.0001, "loss": 0.022, "step": 22900 }, { "epoch": 135.5621301775148, "grad_norm": 1.5941411256790161, "learning_rate": 0.0001, "loss": 0.0239, "step": 22910 }, { "epoch": 135.62130177514794, "grad_norm": 1.742397665977478, "learning_rate": 0.0001, "loss": 0.0236, "step": 22920 }, { "epoch": 135.68047337278105, "grad_norm": 1.9385242462158203, "learning_rate": 0.0001, "loss": 0.0228, "step": 22930 }, { "epoch": 135.7396449704142, "grad_norm": 1.390820860862732, "learning_rate": 0.0001, "loss": 0.0229, "step": 22940 }, { "epoch": 135.79881656804733, "grad_norm": 1.5467450618743896, "learning_rate": 0.0001, "loss": 0.0226, "step": 22950 }, { "epoch": 135.85798816568047, "grad_norm": 1.380983829498291, "learning_rate": 0.0001, "loss": 0.023, "step": 22960 }, { "epoch": 135.9171597633136, "grad_norm": 1.4439271688461304, "learning_rate": 0.0001, "loss": 0.0231, "step": 22970 }, { "epoch": 135.97633136094674, "grad_norm": 1.4563738107681274, "learning_rate": 0.0001, "loss": 0.0232, "step": 22980 }, { "epoch": 136.03550295857988, "grad_norm": 1.008076786994934, "learning_rate": 0.0001, "loss": 0.0227, "step": 22990 }, { "epoch": 136.09467455621302, "grad_norm": 1.5027406215667725, "learning_rate": 0.0001, "loss": 0.0231, "step": 23000 }, { "epoch": 136.15384615384616, "grad_norm": 1.7240090370178223, "learning_rate": 0.0001, "loss": 0.0228, "step": 23010 }, { "epoch": 136.2130177514793, "grad_norm": 1.8829364776611328, "learning_rate": 0.0001, "loss": 0.0222, "step": 23020 }, { "epoch": 136.27218934911244, "grad_norm": 1.7445006370544434, "learning_rate": 0.0001, "loss": 0.0239, "step": 23030 }, { "epoch": 136.33136094674558, "grad_norm": 1.8162493705749512, "learning_rate": 0.0001, "loss": 0.0238, "step": 23040 }, { "epoch": 136.3905325443787, "grad_norm": 1.407775640487671, "learning_rate": 0.0001, "loss": 0.0234, "step": 23050 }, { "epoch": 136.44970414201183, "grad_norm": 1.226261019706726, "learning_rate": 0.0001, "loss": 0.0218, "step": 23060 }, { "epoch": 136.50887573964496, "grad_norm": 1.166310429573059, "learning_rate": 0.0001, "loss": 0.0225, "step": 23070 }, { "epoch": 136.5680473372781, "grad_norm": 1.3146368265151978, "learning_rate": 0.0001, "loss": 0.0224, "step": 23080 }, { "epoch": 136.62721893491124, "grad_norm": 1.058558464050293, "learning_rate": 0.0001, "loss": 0.0234, "step": 23090 }, { "epoch": 136.68639053254438, "grad_norm": 1.312732458114624, "learning_rate": 0.0001, "loss": 0.0224, "step": 23100 }, { "epoch": 136.74556213017752, "grad_norm": 1.6454449892044067, "learning_rate": 0.0001, "loss": 0.0234, "step": 23110 }, { "epoch": 136.80473372781066, "grad_norm": 1.4270521402359009, "learning_rate": 0.0001, "loss": 0.0225, "step": 23120 }, { "epoch": 136.8639053254438, "grad_norm": 1.2697011232376099, "learning_rate": 0.0001, "loss": 0.0232, "step": 23130 }, { "epoch": 136.92307692307693, "grad_norm": 1.3287811279296875, "learning_rate": 0.0001, "loss": 0.0229, "step": 23140 }, { "epoch": 136.98224852071007, "grad_norm": 1.1909235715866089, "learning_rate": 0.0001, "loss": 0.0237, "step": 23150 }, { "epoch": 137.04142011834318, "grad_norm": 1.4343959093093872, "learning_rate": 0.0001, "loss": 0.0229, "step": 23160 }, { "epoch": 137.10059171597632, "grad_norm": 1.331602692604065, "learning_rate": 0.0001, "loss": 0.0233, "step": 23170 }, { "epoch": 137.15976331360946, "grad_norm": 1.5790141820907593, "learning_rate": 0.0001, "loss": 0.0233, "step": 23180 }, { "epoch": 137.2189349112426, "grad_norm": 1.0270471572875977, "learning_rate": 0.0001, "loss": 0.0235, "step": 23190 }, { "epoch": 137.27810650887574, "grad_norm": 2.0987930297851562, "learning_rate": 0.0001, "loss": 0.0239, "step": 23200 }, { "epoch": 137.33727810650888, "grad_norm": 1.552823543548584, "learning_rate": 0.0001, "loss": 0.0233, "step": 23210 }, { "epoch": 137.39644970414201, "grad_norm": 1.3183848857879639, "learning_rate": 0.0001, "loss": 0.024, "step": 23220 }, { "epoch": 137.45562130177515, "grad_norm": 1.5275496244430542, "learning_rate": 0.0001, "loss": 0.0238, "step": 23230 }, { "epoch": 137.5147928994083, "grad_norm": 1.5433861017227173, "learning_rate": 0.0001, "loss": 0.0235, "step": 23240 }, { "epoch": 137.57396449704143, "grad_norm": 1.3984718322753906, "learning_rate": 0.0001, "loss": 0.0235, "step": 23250 }, { "epoch": 137.63313609467457, "grad_norm": 1.606903076171875, "learning_rate": 0.0001, "loss": 0.022, "step": 23260 }, { "epoch": 137.69230769230768, "grad_norm": 1.1285206079483032, "learning_rate": 0.0001, "loss": 0.0235, "step": 23270 }, { "epoch": 137.75147928994082, "grad_norm": 1.87546706199646, "learning_rate": 0.0001, "loss": 0.0234, "step": 23280 }, { "epoch": 137.81065088757396, "grad_norm": 2.1279125213623047, "learning_rate": 0.0001, "loss": 0.0225, "step": 23290 }, { "epoch": 137.8698224852071, "grad_norm": 1.0663539171218872, "learning_rate": 0.0001, "loss": 0.0214, "step": 23300 }, { "epoch": 137.92899408284023, "grad_norm": 1.1506255865097046, "learning_rate": 0.0001, "loss": 0.0222, "step": 23310 }, { "epoch": 137.98816568047337, "grad_norm": 1.4599038362503052, "learning_rate": 0.0001, "loss": 0.0243, "step": 23320 }, { "epoch": 138.0473372781065, "grad_norm": 1.5501033067703247, "learning_rate": 0.0001, "loss": 0.0235, "step": 23330 }, { "epoch": 138.10650887573965, "grad_norm": 1.4991542100906372, "learning_rate": 0.0001, "loss": 0.0242, "step": 23340 }, { "epoch": 138.1656804733728, "grad_norm": 1.196536898612976, "learning_rate": 0.0001, "loss": 0.0242, "step": 23350 }, { "epoch": 138.22485207100593, "grad_norm": 1.582877516746521, "learning_rate": 0.0001, "loss": 0.0247, "step": 23360 }, { "epoch": 138.28402366863907, "grad_norm": 1.261749029159546, "learning_rate": 0.0001, "loss": 0.0241, "step": 23370 }, { "epoch": 138.34319526627218, "grad_norm": 1.1079758405685425, "learning_rate": 0.0001, "loss": 0.0236, "step": 23380 }, { "epoch": 138.4023668639053, "grad_norm": 1.0441226959228516, "learning_rate": 0.0001, "loss": 0.0234, "step": 23390 }, { "epoch": 138.46153846153845, "grad_norm": 1.4128137826919556, "learning_rate": 0.0001, "loss": 0.0233, "step": 23400 }, { "epoch": 138.5207100591716, "grad_norm": 1.5187100172042847, "learning_rate": 0.0001, "loss": 0.0229, "step": 23410 }, { "epoch": 138.57988165680473, "grad_norm": 1.2997010946273804, "learning_rate": 0.0001, "loss": 0.0246, "step": 23420 }, { "epoch": 138.63905325443787, "grad_norm": 1.3031842708587646, "learning_rate": 0.0001, "loss": 0.0234, "step": 23430 }, { "epoch": 138.698224852071, "grad_norm": 1.5750266313552856, "learning_rate": 0.0001, "loss": 0.0246, "step": 23440 }, { "epoch": 138.75739644970415, "grad_norm": 2.6712992191314697, "learning_rate": 0.0001, "loss": 0.0238, "step": 23450 }, { "epoch": 138.81656804733728, "grad_norm": 1.392214059829712, "learning_rate": 0.0001, "loss": 0.0236, "step": 23460 }, { "epoch": 138.87573964497042, "grad_norm": 1.5283150672912598, "learning_rate": 0.0001, "loss": 0.0234, "step": 23470 }, { "epoch": 138.93491124260356, "grad_norm": 1.3901969194412231, "learning_rate": 0.0001, "loss": 0.0225, "step": 23480 }, { "epoch": 138.9940828402367, "grad_norm": 1.9193007946014404, "learning_rate": 0.0001, "loss": 0.0226, "step": 23490 }, { "epoch": 139.0532544378698, "grad_norm": 1.8145160675048828, "learning_rate": 0.0001, "loss": 0.0241, "step": 23500 }, { "epoch": 139.11242603550295, "grad_norm": 1.2250152826309204, "learning_rate": 0.0001, "loss": 0.0217, "step": 23510 }, { "epoch": 139.1715976331361, "grad_norm": 1.696818470954895, "learning_rate": 0.0001, "loss": 0.0229, "step": 23520 }, { "epoch": 139.23076923076923, "grad_norm": 1.3276128768920898, "learning_rate": 0.0001, "loss": 0.0226, "step": 23530 }, { "epoch": 139.28994082840237, "grad_norm": 1.547763466835022, "learning_rate": 0.0001, "loss": 0.0221, "step": 23540 }, { "epoch": 139.3491124260355, "grad_norm": 1.4323389530181885, "learning_rate": 0.0001, "loss": 0.0217, "step": 23550 }, { "epoch": 139.40828402366864, "grad_norm": 1.3492522239685059, "learning_rate": 0.0001, "loss": 0.0225, "step": 23560 }, { "epoch": 139.46745562130178, "grad_norm": 1.037255048751831, "learning_rate": 0.0001, "loss": 0.0225, "step": 23570 }, { "epoch": 139.52662721893492, "grad_norm": 1.5966969728469849, "learning_rate": 0.0001, "loss": 0.0235, "step": 23580 }, { "epoch": 139.58579881656806, "grad_norm": 1.6263082027435303, "learning_rate": 0.0001, "loss": 0.0236, "step": 23590 }, { "epoch": 139.6449704142012, "grad_norm": 1.4830694198608398, "learning_rate": 0.0001, "loss": 0.023, "step": 23600 }, { "epoch": 139.7041420118343, "grad_norm": 1.5873267650604248, "learning_rate": 0.0001, "loss": 0.0217, "step": 23610 }, { "epoch": 139.76331360946745, "grad_norm": 1.2400026321411133, "learning_rate": 0.0001, "loss": 0.0226, "step": 23620 }, { "epoch": 139.82248520710058, "grad_norm": 1.5581709146499634, "learning_rate": 0.0001, "loss": 0.0222, "step": 23630 }, { "epoch": 139.88165680473372, "grad_norm": 1.217904806137085, "learning_rate": 0.0001, "loss": 0.0225, "step": 23640 }, { "epoch": 139.94082840236686, "grad_norm": 1.657213807106018, "learning_rate": 0.0001, "loss": 0.0217, "step": 23650 }, { "epoch": 140.0, "grad_norm": 1.3090523481369019, "learning_rate": 0.0001, "loss": 0.0213, "step": 23660 }, { "epoch": 140.05917159763314, "grad_norm": 1.7182044982910156, "learning_rate": 0.0001, "loss": 0.0223, "step": 23670 }, { "epoch": 140.11834319526628, "grad_norm": 1.5938249826431274, "learning_rate": 0.0001, "loss": 0.0227, "step": 23680 }, { "epoch": 140.17751479289942, "grad_norm": 1.423177719116211, "learning_rate": 0.0001, "loss": 0.022, "step": 23690 }, { "epoch": 140.23668639053255, "grad_norm": 1.4663641452789307, "learning_rate": 0.0001, "loss": 0.0232, "step": 23700 }, { "epoch": 140.2958579881657, "grad_norm": 1.6691519021987915, "learning_rate": 0.0001, "loss": 0.0234, "step": 23710 }, { "epoch": 140.3550295857988, "grad_norm": 1.2974470853805542, "learning_rate": 0.0001, "loss": 0.023, "step": 23720 }, { "epoch": 140.41420118343194, "grad_norm": 1.558093786239624, "learning_rate": 0.0001, "loss": 0.0228, "step": 23730 }, { "epoch": 140.47337278106508, "grad_norm": 2.1833391189575195, "learning_rate": 0.0001, "loss": 0.024, "step": 23740 }, { "epoch": 140.53254437869822, "grad_norm": 1.322705626487732, "learning_rate": 0.0001, "loss": 0.0219, "step": 23750 }, { "epoch": 140.59171597633136, "grad_norm": 2.5005176067352295, "learning_rate": 0.0001, "loss": 0.0228, "step": 23760 }, { "epoch": 140.6508875739645, "grad_norm": 2.457139253616333, "learning_rate": 0.0001, "loss": 0.0232, "step": 23770 }, { "epoch": 140.71005917159763, "grad_norm": 1.6916102170944214, "learning_rate": 0.0001, "loss": 0.0232, "step": 23780 }, { "epoch": 140.76923076923077, "grad_norm": 1.3022249937057495, "learning_rate": 0.0001, "loss": 0.0223, "step": 23790 }, { "epoch": 140.8284023668639, "grad_norm": 1.1824474334716797, "learning_rate": 0.0001, "loss": 0.0222, "step": 23800 }, { "epoch": 140.88757396449705, "grad_norm": 1.287595272064209, "learning_rate": 0.0001, "loss": 0.022, "step": 23810 }, { "epoch": 140.9467455621302, "grad_norm": 1.2182550430297852, "learning_rate": 0.0001, "loss": 0.023, "step": 23820 }, { "epoch": 141.0059171597633, "grad_norm": 1.4441462755203247, "learning_rate": 0.0001, "loss": 0.0229, "step": 23830 }, { "epoch": 141.06508875739644, "grad_norm": 1.278725028038025, "learning_rate": 0.0001, "loss": 0.0224, "step": 23840 }, { "epoch": 141.12426035502958, "grad_norm": 1.117855429649353, "learning_rate": 0.0001, "loss": 0.0237, "step": 23850 }, { "epoch": 141.18343195266272, "grad_norm": 1.4191057682037354, "learning_rate": 0.0001, "loss": 0.0249, "step": 23860 }, { "epoch": 141.24260355029585, "grad_norm": 1.540771484375, "learning_rate": 0.0001, "loss": 0.0249, "step": 23870 }, { "epoch": 141.301775147929, "grad_norm": 1.0990530252456665, "learning_rate": 0.0001, "loss": 0.0244, "step": 23880 }, { "epoch": 141.36094674556213, "grad_norm": 1.415777325630188, "learning_rate": 0.0001, "loss": 0.0228, "step": 23890 }, { "epoch": 141.42011834319527, "grad_norm": 1.3027247190475464, "learning_rate": 0.0001, "loss": 0.0226, "step": 23900 }, { "epoch": 141.4792899408284, "grad_norm": 1.2007877826690674, "learning_rate": 0.0001, "loss": 0.0237, "step": 23910 }, { "epoch": 141.53846153846155, "grad_norm": 1.4063174724578857, "learning_rate": 0.0001, "loss": 0.0238, "step": 23920 }, { "epoch": 141.5976331360947, "grad_norm": 1.3678361177444458, "learning_rate": 0.0001, "loss": 0.024, "step": 23930 }, { "epoch": 141.65680473372782, "grad_norm": 1.5221940279006958, "learning_rate": 0.0001, "loss": 0.0231, "step": 23940 }, { "epoch": 141.71597633136093, "grad_norm": 1.4238009452819824, "learning_rate": 0.0001, "loss": 0.0227, "step": 23950 }, { "epoch": 141.77514792899407, "grad_norm": 1.5237504243850708, "learning_rate": 0.0001, "loss": 0.0238, "step": 23960 }, { "epoch": 141.8343195266272, "grad_norm": 2.0769155025482178, "learning_rate": 0.0001, "loss": 0.0253, "step": 23970 }, { "epoch": 141.89349112426035, "grad_norm": 1.419219732284546, "learning_rate": 0.0001, "loss": 0.0226, "step": 23980 }, { "epoch": 141.9526627218935, "grad_norm": 1.108001947402954, "learning_rate": 0.0001, "loss": 0.0232, "step": 23990 }, { "epoch": 142.01183431952663, "grad_norm": 1.2419253587722778, "learning_rate": 0.0001, "loss": 0.022, "step": 24000 }, { "epoch": 142.07100591715977, "grad_norm": 1.2445439100265503, "learning_rate": 0.0001, "loss": 0.0219, "step": 24010 }, { "epoch": 142.1301775147929, "grad_norm": 1.7760157585144043, "learning_rate": 0.0001, "loss": 0.0229, "step": 24020 }, { "epoch": 142.18934911242604, "grad_norm": 0.9730281233787537, "learning_rate": 0.0001, "loss": 0.0234, "step": 24030 }, { "epoch": 142.24852071005918, "grad_norm": 1.1950058937072754, "learning_rate": 0.0001, "loss": 0.0229, "step": 24040 }, { "epoch": 142.30769230769232, "grad_norm": 1.6115761995315552, "learning_rate": 0.0001, "loss": 0.0231, "step": 24050 }, { "epoch": 142.36686390532543, "grad_norm": 1.3578972816467285, "learning_rate": 0.0001, "loss": 0.0229, "step": 24060 }, { "epoch": 142.42603550295857, "grad_norm": 1.7409601211547852, "learning_rate": 0.0001, "loss": 0.0236, "step": 24070 }, { "epoch": 142.4852071005917, "grad_norm": 1.8936048746109009, "learning_rate": 0.0001, "loss": 0.0237, "step": 24080 }, { "epoch": 142.54437869822485, "grad_norm": 1.7024095058441162, "learning_rate": 0.0001, "loss": 0.0218, "step": 24090 }, { "epoch": 142.60355029585799, "grad_norm": 1.8986092805862427, "learning_rate": 0.0001, "loss": 0.023, "step": 24100 }, { "epoch": 142.66272189349112, "grad_norm": 1.5966641902923584, "learning_rate": 0.0001, "loss": 0.0221, "step": 24110 }, { "epoch": 142.72189349112426, "grad_norm": 1.3769018650054932, "learning_rate": 0.0001, "loss": 0.0237, "step": 24120 }, { "epoch": 142.7810650887574, "grad_norm": 1.4885668754577637, "learning_rate": 0.0001, "loss": 0.0229, "step": 24130 }, { "epoch": 142.84023668639054, "grad_norm": 1.3013654947280884, "learning_rate": 0.0001, "loss": 0.0224, "step": 24140 }, { "epoch": 142.89940828402368, "grad_norm": 2.171419858932495, "learning_rate": 0.0001, "loss": 0.0219, "step": 24150 }, { "epoch": 142.95857988165682, "grad_norm": 1.5883712768554688, "learning_rate": 0.0001, "loss": 0.0232, "step": 24160 }, { "epoch": 143.01775147928993, "grad_norm": 1.6839144229888916, "learning_rate": 0.0001, "loss": 0.0223, "step": 24170 }, { "epoch": 143.07692307692307, "grad_norm": 1.884848952293396, "learning_rate": 0.0001, "loss": 0.0232, "step": 24180 }, { "epoch": 143.1360946745562, "grad_norm": 1.3516279458999634, "learning_rate": 0.0001, "loss": 0.0236, "step": 24190 }, { "epoch": 143.19526627218934, "grad_norm": 1.2143150568008423, "learning_rate": 0.0001, "loss": 0.0238, "step": 24200 }, { "epoch": 143.25443786982248, "grad_norm": 1.1210254430770874, "learning_rate": 0.0001, "loss": 0.0216, "step": 24210 }, { "epoch": 143.31360946745562, "grad_norm": 1.3645530939102173, "learning_rate": 0.0001, "loss": 0.0212, "step": 24220 }, { "epoch": 143.37278106508876, "grad_norm": 1.4452154636383057, "learning_rate": 0.0001, "loss": 0.0224, "step": 24230 }, { "epoch": 143.4319526627219, "grad_norm": 1.3181666135787964, "learning_rate": 0.0001, "loss": 0.0212, "step": 24240 }, { "epoch": 143.49112426035504, "grad_norm": 1.0862276554107666, "learning_rate": 0.0001, "loss": 0.0217, "step": 24250 }, { "epoch": 143.55029585798817, "grad_norm": 1.0731688737869263, "learning_rate": 0.0001, "loss": 0.0213, "step": 24260 }, { "epoch": 143.6094674556213, "grad_norm": 1.2234258651733398, "learning_rate": 0.0001, "loss": 0.0223, "step": 24270 }, { "epoch": 143.66863905325442, "grad_norm": 1.6040650606155396, "learning_rate": 0.0001, "loss": 0.0213, "step": 24280 }, { "epoch": 143.72781065088756, "grad_norm": 1.111960768699646, "learning_rate": 0.0001, "loss": 0.0232, "step": 24290 }, { "epoch": 143.7869822485207, "grad_norm": 1.448182463645935, "learning_rate": 0.0001, "loss": 0.0223, "step": 24300 }, { "epoch": 143.84615384615384, "grad_norm": 1.320658564567566, "learning_rate": 0.0001, "loss": 0.0224, "step": 24310 }, { "epoch": 143.90532544378698, "grad_norm": 1.1499472856521606, "learning_rate": 0.0001, "loss": 0.023, "step": 24320 }, { "epoch": 143.96449704142012, "grad_norm": 1.4925100803375244, "learning_rate": 0.0001, "loss": 0.0224, "step": 24330 }, { "epoch": 144.02366863905326, "grad_norm": 1.2242377996444702, "learning_rate": 0.0001, "loss": 0.0224, "step": 24340 }, { "epoch": 144.0828402366864, "grad_norm": 1.143158197402954, "learning_rate": 0.0001, "loss": 0.0224, "step": 24350 }, { "epoch": 144.14201183431953, "grad_norm": 1.391683578491211, "learning_rate": 0.0001, "loss": 0.0238, "step": 24360 }, { "epoch": 144.20118343195267, "grad_norm": 1.6641006469726562, "learning_rate": 0.0001, "loss": 0.0231, "step": 24370 }, { "epoch": 144.2603550295858, "grad_norm": 1.3952569961547852, "learning_rate": 0.0001, "loss": 0.0236, "step": 24380 }, { "epoch": 144.31952662721895, "grad_norm": 1.3652406930923462, "learning_rate": 0.0001, "loss": 0.0234, "step": 24390 }, { "epoch": 144.37869822485206, "grad_norm": 1.9934130907058716, "learning_rate": 0.0001, "loss": 0.0237, "step": 24400 }, { "epoch": 144.4378698224852, "grad_norm": 1.7354295253753662, "learning_rate": 0.0001, "loss": 0.0224, "step": 24410 }, { "epoch": 144.49704142011834, "grad_norm": 1.6311091184616089, "learning_rate": 0.0001, "loss": 0.0231, "step": 24420 }, { "epoch": 144.55621301775147, "grad_norm": 1.5923337936401367, "learning_rate": 0.0001, "loss": 0.0233, "step": 24430 }, { "epoch": 144.6153846153846, "grad_norm": 1.4344260692596436, "learning_rate": 0.0001, "loss": 0.0221, "step": 24440 }, { "epoch": 144.67455621301775, "grad_norm": 1.227130651473999, "learning_rate": 0.0001, "loss": 0.023, "step": 24450 }, { "epoch": 144.7337278106509, "grad_norm": 0.9901591539382935, "learning_rate": 0.0001, "loss": 0.0231, "step": 24460 }, { "epoch": 144.79289940828403, "grad_norm": 1.2369555234909058, "learning_rate": 0.0001, "loss": 0.0225, "step": 24470 }, { "epoch": 144.85207100591717, "grad_norm": 1.1737309694290161, "learning_rate": 0.0001, "loss": 0.0228, "step": 24480 }, { "epoch": 144.9112426035503, "grad_norm": 1.3004835844039917, "learning_rate": 0.0001, "loss": 0.0231, "step": 24490 }, { "epoch": 144.97041420118344, "grad_norm": 1.6025450229644775, "learning_rate": 0.0001, "loss": 0.024, "step": 24500 }, { "epoch": 145.02958579881656, "grad_norm": 1.2681784629821777, "learning_rate": 0.0001, "loss": 0.0235, "step": 24510 }, { "epoch": 145.0887573964497, "grad_norm": 1.2675150632858276, "learning_rate": 0.0001, "loss": 0.0233, "step": 24520 }, { "epoch": 145.14792899408283, "grad_norm": 2.4846136569976807, "learning_rate": 0.0001, "loss": 0.0242, "step": 24530 }, { "epoch": 145.20710059171597, "grad_norm": 1.7916557788848877, "learning_rate": 0.0001, "loss": 0.0236, "step": 24540 }, { "epoch": 145.2662721893491, "grad_norm": 1.6869090795516968, "learning_rate": 0.0001, "loss": 0.0233, "step": 24550 }, { "epoch": 145.32544378698225, "grad_norm": 1.9680649042129517, "learning_rate": 0.0001, "loss": 0.0225, "step": 24560 }, { "epoch": 145.3846153846154, "grad_norm": 1.8574966192245483, "learning_rate": 0.0001, "loss": 0.0224, "step": 24570 }, { "epoch": 145.44378698224853, "grad_norm": 1.4449577331542969, "learning_rate": 0.0001, "loss": 0.0213, "step": 24580 }, { "epoch": 145.50295857988166, "grad_norm": 1.3018991947174072, "learning_rate": 0.0001, "loss": 0.0216, "step": 24590 }, { "epoch": 145.5621301775148, "grad_norm": 1.3078190088272095, "learning_rate": 0.0001, "loss": 0.0212, "step": 24600 }, { "epoch": 145.62130177514794, "grad_norm": 1.5029728412628174, "learning_rate": 0.0001, "loss": 0.0214, "step": 24610 }, { "epoch": 145.68047337278105, "grad_norm": 1.6518278121948242, "learning_rate": 0.0001, "loss": 0.0227, "step": 24620 }, { "epoch": 145.7396449704142, "grad_norm": 1.3797104358673096, "learning_rate": 0.0001, "loss": 0.0224, "step": 24630 }, { "epoch": 145.79881656804733, "grad_norm": 1.337508201599121, "learning_rate": 0.0001, "loss": 0.0222, "step": 24640 }, { "epoch": 145.85798816568047, "grad_norm": 1.3512612581253052, "learning_rate": 0.0001, "loss": 0.0226, "step": 24650 }, { "epoch": 145.9171597633136, "grad_norm": 1.4625164270401, "learning_rate": 0.0001, "loss": 0.0239, "step": 24660 }, { "epoch": 145.97633136094674, "grad_norm": 0.9829410910606384, "learning_rate": 0.0001, "loss": 0.0221, "step": 24670 }, { "epoch": 146.03550295857988, "grad_norm": 1.2950361967086792, "learning_rate": 0.0001, "loss": 0.0232, "step": 24680 }, { "epoch": 146.09467455621302, "grad_norm": 1.5028809309005737, "learning_rate": 0.0001, "loss": 0.0219, "step": 24690 }, { "epoch": 146.15384615384616, "grad_norm": 1.550463080406189, "learning_rate": 0.0001, "loss": 0.0231, "step": 24700 }, { "epoch": 146.2130177514793, "grad_norm": 1.546109914779663, "learning_rate": 0.0001, "loss": 0.0233, "step": 24710 }, { "epoch": 146.27218934911244, "grad_norm": 1.5201597213745117, "learning_rate": 0.0001, "loss": 0.0226, "step": 24720 }, { "epoch": 146.33136094674558, "grad_norm": 1.410349726676941, "learning_rate": 0.0001, "loss": 0.0222, "step": 24730 }, { "epoch": 146.3905325443787, "grad_norm": 1.0742242336273193, "learning_rate": 0.0001, "loss": 0.0223, "step": 24740 }, { "epoch": 146.44970414201183, "grad_norm": 1.6277532577514648, "learning_rate": 0.0001, "loss": 0.0229, "step": 24750 }, { "epoch": 146.50887573964496, "grad_norm": 1.9191012382507324, "learning_rate": 0.0001, "loss": 0.023, "step": 24760 }, { "epoch": 146.5680473372781, "grad_norm": 1.3771097660064697, "learning_rate": 0.0001, "loss": 0.023, "step": 24770 }, { "epoch": 146.62721893491124, "grad_norm": 1.3655176162719727, "learning_rate": 0.0001, "loss": 0.0225, "step": 24780 }, { "epoch": 146.68639053254438, "grad_norm": 1.851922869682312, "learning_rate": 0.0001, "loss": 0.0227, "step": 24790 }, { "epoch": 146.74556213017752, "grad_norm": 1.6313717365264893, "learning_rate": 0.0001, "loss": 0.0228, "step": 24800 }, { "epoch": 146.80473372781066, "grad_norm": 1.718437671661377, "learning_rate": 0.0001, "loss": 0.023, "step": 24810 }, { "epoch": 146.8639053254438, "grad_norm": 1.4559136629104614, "learning_rate": 0.0001, "loss": 0.0216, "step": 24820 }, { "epoch": 146.92307692307693, "grad_norm": 1.497010588645935, "learning_rate": 0.0001, "loss": 0.0213, "step": 24830 }, { "epoch": 146.98224852071007, "grad_norm": 1.5541388988494873, "learning_rate": 0.0001, "loss": 0.0221, "step": 24840 }, { "epoch": 147.04142011834318, "grad_norm": 1.5442196130752563, "learning_rate": 0.0001, "loss": 0.0214, "step": 24850 }, { "epoch": 147.10059171597632, "grad_norm": 1.7613492012023926, "learning_rate": 0.0001, "loss": 0.0218, "step": 24860 }, { "epoch": 147.15976331360946, "grad_norm": 1.2829681634902954, "learning_rate": 0.0001, "loss": 0.021, "step": 24870 }, { "epoch": 147.2189349112426, "grad_norm": 1.1418352127075195, "learning_rate": 0.0001, "loss": 0.0224, "step": 24880 }, { "epoch": 147.27810650887574, "grad_norm": 1.2026394605636597, "learning_rate": 0.0001, "loss": 0.0211, "step": 24890 }, { "epoch": 147.33727810650888, "grad_norm": 1.1818193197250366, "learning_rate": 0.0001, "loss": 0.0216, "step": 24900 }, { "epoch": 147.39644970414201, "grad_norm": 1.0214804410934448, "learning_rate": 0.0001, "loss": 0.0226, "step": 24910 }, { "epoch": 147.45562130177515, "grad_norm": 1.3492696285247803, "learning_rate": 0.0001, "loss": 0.0224, "step": 24920 }, { "epoch": 147.5147928994083, "grad_norm": 1.5501071214675903, "learning_rate": 0.0001, "loss": 0.0233, "step": 24930 }, { "epoch": 147.57396449704143, "grad_norm": 1.0075403451919556, "learning_rate": 0.0001, "loss": 0.0228, "step": 24940 }, { "epoch": 147.63313609467457, "grad_norm": 2.275965929031372, "learning_rate": 0.0001, "loss": 0.0233, "step": 24950 }, { "epoch": 147.69230769230768, "grad_norm": 1.9871184825897217, "learning_rate": 0.0001, "loss": 0.0229, "step": 24960 }, { "epoch": 147.75147928994082, "grad_norm": 1.1846410036087036, "learning_rate": 0.0001, "loss": 0.023, "step": 24970 }, { "epoch": 147.81065088757396, "grad_norm": 1.3013665676116943, "learning_rate": 0.0001, "loss": 0.0213, "step": 24980 }, { "epoch": 147.8698224852071, "grad_norm": 1.4388583898544312, "learning_rate": 0.0001, "loss": 0.0221, "step": 24990 }, { "epoch": 147.92899408284023, "grad_norm": 1.7013663053512573, "learning_rate": 0.0001, "loss": 0.0222, "step": 25000 }, { "epoch": 147.98816568047337, "grad_norm": 1.9687913656234741, "learning_rate": 0.0001, "loss": 0.0229, "step": 25010 }, { "epoch": 148.0473372781065, "grad_norm": 2.1248157024383545, "learning_rate": 0.0001, "loss": 0.0232, "step": 25020 }, { "epoch": 148.10650887573965, "grad_norm": 1.3683364391326904, "learning_rate": 0.0001, "loss": 0.0221, "step": 25030 }, { "epoch": 148.1656804733728, "grad_norm": 1.4235748052597046, "learning_rate": 0.0001, "loss": 0.0223, "step": 25040 }, { "epoch": 148.22485207100593, "grad_norm": 1.3020665645599365, "learning_rate": 0.0001, "loss": 0.0214, "step": 25050 }, { "epoch": 148.28402366863907, "grad_norm": 1.2908762693405151, "learning_rate": 0.0001, "loss": 0.0216, "step": 25060 }, { "epoch": 148.34319526627218, "grad_norm": 1.7880098819732666, "learning_rate": 0.0001, "loss": 0.0224, "step": 25070 }, { "epoch": 148.4023668639053, "grad_norm": 1.6447688341140747, "learning_rate": 0.0001, "loss": 0.0232, "step": 25080 }, { "epoch": 148.46153846153845, "grad_norm": 1.8929938077926636, "learning_rate": 0.0001, "loss": 0.0221, "step": 25090 }, { "epoch": 148.5207100591716, "grad_norm": 1.5567349195480347, "learning_rate": 0.0001, "loss": 0.0221, "step": 25100 }, { "epoch": 148.57988165680473, "grad_norm": 1.8570616245269775, "learning_rate": 0.0001, "loss": 0.0214, "step": 25110 }, { "epoch": 148.63905325443787, "grad_norm": 1.4416215419769287, "learning_rate": 0.0001, "loss": 0.0227, "step": 25120 }, { "epoch": 148.698224852071, "grad_norm": 1.380882740020752, "learning_rate": 0.0001, "loss": 0.0213, "step": 25130 }, { "epoch": 148.75739644970415, "grad_norm": 1.4313751459121704, "learning_rate": 0.0001, "loss": 0.0216, "step": 25140 }, { "epoch": 148.81656804733728, "grad_norm": 1.2441447973251343, "learning_rate": 0.0001, "loss": 0.022, "step": 25150 }, { "epoch": 148.87573964497042, "grad_norm": 1.6935595273971558, "learning_rate": 0.0001, "loss": 0.0204, "step": 25160 }, { "epoch": 148.93491124260356, "grad_norm": 1.8206140995025635, "learning_rate": 0.0001, "loss": 0.0227, "step": 25170 }, { "epoch": 148.9940828402367, "grad_norm": 1.460291862487793, "learning_rate": 0.0001, "loss": 0.0228, "step": 25180 }, { "epoch": 149.0532544378698, "grad_norm": 1.7869505882263184, "learning_rate": 0.0001, "loss": 0.0219, "step": 25190 }, { "epoch": 149.11242603550295, "grad_norm": 1.3300869464874268, "learning_rate": 0.0001, "loss": 0.022, "step": 25200 }, { "epoch": 149.1715976331361, "grad_norm": 1.5892833471298218, "learning_rate": 0.0001, "loss": 0.0219, "step": 25210 }, { "epoch": 149.23076923076923, "grad_norm": 1.846030831336975, "learning_rate": 0.0001, "loss": 0.0224, "step": 25220 }, { "epoch": 149.28994082840237, "grad_norm": 1.3531348705291748, "learning_rate": 0.0001, "loss": 0.0232, "step": 25230 }, { "epoch": 149.3491124260355, "grad_norm": 1.0698927640914917, "learning_rate": 0.0001, "loss": 0.024, "step": 25240 }, { "epoch": 149.40828402366864, "grad_norm": 1.4911789894104004, "learning_rate": 0.0001, "loss": 0.023, "step": 25250 }, { "epoch": 149.46745562130178, "grad_norm": 1.2789133787155151, "learning_rate": 0.0001, "loss": 0.0227, "step": 25260 }, { "epoch": 149.52662721893492, "grad_norm": 1.2878471612930298, "learning_rate": 0.0001, "loss": 0.0227, "step": 25270 }, { "epoch": 149.58579881656806, "grad_norm": 1.2039148807525635, "learning_rate": 0.0001, "loss": 0.0219, "step": 25280 }, { "epoch": 149.6449704142012, "grad_norm": 1.7013520002365112, "learning_rate": 0.0001, "loss": 0.0227, "step": 25290 }, { "epoch": 149.7041420118343, "grad_norm": 1.6520761251449585, "learning_rate": 0.0001, "loss": 0.0223, "step": 25300 }, { "epoch": 149.76331360946745, "grad_norm": 1.3770825862884521, "learning_rate": 0.0001, "loss": 0.0221, "step": 25310 }, { "epoch": 149.82248520710058, "grad_norm": 1.2222503423690796, "learning_rate": 0.0001, "loss": 0.0226, "step": 25320 }, { "epoch": 149.88165680473372, "grad_norm": 1.3172694444656372, "learning_rate": 0.0001, "loss": 0.0226, "step": 25330 }, { "epoch": 149.94082840236686, "grad_norm": 1.5450624227523804, "learning_rate": 0.0001, "loss": 0.0227, "step": 25340 }, { "epoch": 150.0, "grad_norm": 1.2983654737472534, "learning_rate": 0.0001, "loss": 0.0229, "step": 25350 }, { "epoch": 150.05917159763314, "grad_norm": 1.435376763343811, "learning_rate": 0.0001, "loss": 0.022, "step": 25360 }, { "epoch": 150.11834319526628, "grad_norm": 1.5601383447647095, "learning_rate": 0.0001, "loss": 0.0238, "step": 25370 }, { "epoch": 150.17751479289942, "grad_norm": 1.3944593667984009, "learning_rate": 0.0001, "loss": 0.0236, "step": 25380 }, { "epoch": 150.23668639053255, "grad_norm": 1.286478877067566, "learning_rate": 0.0001, "loss": 0.023, "step": 25390 }, { "epoch": 150.2958579881657, "grad_norm": 1.3614124059677124, "learning_rate": 0.0001, "loss": 0.0229, "step": 25400 }, { "epoch": 150.3550295857988, "grad_norm": 1.3912999629974365, "learning_rate": 0.0001, "loss": 0.0222, "step": 25410 }, { "epoch": 150.41420118343194, "grad_norm": 1.6693559885025024, "learning_rate": 0.0001, "loss": 0.0227, "step": 25420 }, { "epoch": 150.47337278106508, "grad_norm": 1.266960859298706, "learning_rate": 0.0001, "loss": 0.0216, "step": 25430 }, { "epoch": 150.53254437869822, "grad_norm": 1.3553344011306763, "learning_rate": 0.0001, "loss": 0.0221, "step": 25440 }, { "epoch": 150.59171597633136, "grad_norm": 1.275038242340088, "learning_rate": 0.0001, "loss": 0.0226, "step": 25450 }, { "epoch": 150.6508875739645, "grad_norm": 1.1660391092300415, "learning_rate": 0.0001, "loss": 0.022, "step": 25460 }, { "epoch": 150.71005917159763, "grad_norm": 1.0408915281295776, "learning_rate": 0.0001, "loss": 0.0215, "step": 25470 }, { "epoch": 150.76923076923077, "grad_norm": 1.391427993774414, "learning_rate": 0.0001, "loss": 0.0221, "step": 25480 }, { "epoch": 150.8284023668639, "grad_norm": 1.6635313034057617, "learning_rate": 0.0001, "loss": 0.022, "step": 25490 }, { "epoch": 150.88757396449705, "grad_norm": 2.160356044769287, "learning_rate": 0.0001, "loss": 0.0237, "step": 25500 }, { "epoch": 150.9467455621302, "grad_norm": 2.2749130725860596, "learning_rate": 0.0001, "loss": 0.0235, "step": 25510 }, { "epoch": 151.0059171597633, "grad_norm": 1.6757078170776367, "learning_rate": 0.0001, "loss": 0.0245, "step": 25520 }, { "epoch": 151.06508875739644, "grad_norm": 1.7509537935256958, "learning_rate": 0.0001, "loss": 0.0233, "step": 25530 }, { "epoch": 151.12426035502958, "grad_norm": 1.517418384552002, "learning_rate": 0.0001, "loss": 0.0223, "step": 25540 }, { "epoch": 151.18343195266272, "grad_norm": 1.6271003484725952, "learning_rate": 0.0001, "loss": 0.0219, "step": 25550 }, { "epoch": 151.24260355029585, "grad_norm": 1.7227970361709595, "learning_rate": 0.0001, "loss": 0.0217, "step": 25560 }, { "epoch": 151.301775147929, "grad_norm": 2.092853546142578, "learning_rate": 0.0001, "loss": 0.0225, "step": 25570 }, { "epoch": 151.36094674556213, "grad_norm": 1.5334817171096802, "learning_rate": 0.0001, "loss": 0.0219, "step": 25580 }, { "epoch": 151.42011834319527, "grad_norm": 1.4114316701889038, "learning_rate": 0.0001, "loss": 0.023, "step": 25590 }, { "epoch": 151.4792899408284, "grad_norm": 1.811838150024414, "learning_rate": 0.0001, "loss": 0.0231, "step": 25600 }, { "epoch": 151.53846153846155, "grad_norm": 1.4995858669281006, "learning_rate": 0.0001, "loss": 0.0233, "step": 25610 }, { "epoch": 151.5976331360947, "grad_norm": 1.9088345766067505, "learning_rate": 0.0001, "loss": 0.0236, "step": 25620 }, { "epoch": 151.65680473372782, "grad_norm": 1.8351342678070068, "learning_rate": 0.0001, "loss": 0.0213, "step": 25630 }, { "epoch": 151.71597633136093, "grad_norm": 1.5224294662475586, "learning_rate": 0.0001, "loss": 0.0227, "step": 25640 }, { "epoch": 151.77514792899407, "grad_norm": 1.6776913404464722, "learning_rate": 0.0001, "loss": 0.0212, "step": 25650 }, { "epoch": 151.8343195266272, "grad_norm": 1.5520858764648438, "learning_rate": 0.0001, "loss": 0.0224, "step": 25660 }, { "epoch": 151.89349112426035, "grad_norm": 1.3121304512023926, "learning_rate": 0.0001, "loss": 0.0218, "step": 25670 }, { "epoch": 151.9526627218935, "grad_norm": 1.5871235132217407, "learning_rate": 0.0001, "loss": 0.0219, "step": 25680 }, { "epoch": 152.01183431952663, "grad_norm": 1.3929308652877808, "learning_rate": 0.0001, "loss": 0.0223, "step": 25690 }, { "epoch": 152.07100591715977, "grad_norm": 1.794174313545227, "learning_rate": 0.0001, "loss": 0.0227, "step": 25700 }, { "epoch": 152.1301775147929, "grad_norm": 1.4306180477142334, "learning_rate": 0.0001, "loss": 0.0234, "step": 25710 }, { "epoch": 152.18934911242604, "grad_norm": 1.5918587446212769, "learning_rate": 0.0001, "loss": 0.0231, "step": 25720 }, { "epoch": 152.24852071005918, "grad_norm": 1.3437448740005493, "learning_rate": 0.0001, "loss": 0.0218, "step": 25730 }, { "epoch": 152.30769230769232, "grad_norm": 1.422795295715332, "learning_rate": 0.0001, "loss": 0.0218, "step": 25740 }, { "epoch": 152.36686390532543, "grad_norm": 1.0995882749557495, "learning_rate": 0.0001, "loss": 0.022, "step": 25750 }, { "epoch": 152.42603550295857, "grad_norm": 2.1809849739074707, "learning_rate": 0.0001, "loss": 0.0221, "step": 25760 }, { "epoch": 152.4852071005917, "grad_norm": 1.6117023229599, "learning_rate": 0.0001, "loss": 0.0221, "step": 25770 }, { "epoch": 152.54437869822485, "grad_norm": 1.487176775932312, "learning_rate": 0.0001, "loss": 0.0217, "step": 25780 }, { "epoch": 152.60355029585799, "grad_norm": 1.6727238893508911, "learning_rate": 0.0001, "loss": 0.0231, "step": 25790 }, { "epoch": 152.66272189349112, "grad_norm": 1.2094824314117432, "learning_rate": 0.0001, "loss": 0.0231, "step": 25800 }, { "epoch": 152.72189349112426, "grad_norm": 1.553799033164978, "learning_rate": 0.0001, "loss": 0.0217, "step": 25810 }, { "epoch": 152.7810650887574, "grad_norm": 1.7227586507797241, "learning_rate": 0.0001, "loss": 0.0215, "step": 25820 }, { "epoch": 152.84023668639054, "grad_norm": 1.757019281387329, "learning_rate": 0.0001, "loss": 0.0219, "step": 25830 }, { "epoch": 152.89940828402368, "grad_norm": 1.2971783876419067, "learning_rate": 0.0001, "loss": 0.0207, "step": 25840 }, { "epoch": 152.95857988165682, "grad_norm": 1.3304834365844727, "learning_rate": 0.0001, "loss": 0.0213, "step": 25850 }, { "epoch": 153.01775147928993, "grad_norm": 1.495637059211731, "learning_rate": 0.0001, "loss": 0.0218, "step": 25860 }, { "epoch": 153.07692307692307, "grad_norm": 1.7224271297454834, "learning_rate": 0.0001, "loss": 0.021, "step": 25870 }, { "epoch": 153.1360946745562, "grad_norm": 1.189470887184143, "learning_rate": 0.0001, "loss": 0.0223, "step": 25880 }, { "epoch": 153.19526627218934, "grad_norm": 1.4131863117218018, "learning_rate": 0.0001, "loss": 0.0221, "step": 25890 }, { "epoch": 153.25443786982248, "grad_norm": 1.4546622037887573, "learning_rate": 0.0001, "loss": 0.0219, "step": 25900 }, { "epoch": 153.31360946745562, "grad_norm": 1.9694441556930542, "learning_rate": 0.0001, "loss": 0.0226, "step": 25910 }, { "epoch": 153.37278106508876, "grad_norm": 1.6388379335403442, "learning_rate": 0.0001, "loss": 0.0219, "step": 25920 }, { "epoch": 153.4319526627219, "grad_norm": 1.3594251871109009, "learning_rate": 0.0001, "loss": 0.0214, "step": 25930 }, { "epoch": 153.49112426035504, "grad_norm": 1.5911802053451538, "learning_rate": 0.0001, "loss": 0.0219, "step": 25940 }, { "epoch": 153.55029585798817, "grad_norm": 1.729836106300354, "learning_rate": 0.0001, "loss": 0.0223, "step": 25950 }, { "epoch": 153.6094674556213, "grad_norm": 1.2635977268218994, "learning_rate": 0.0001, "loss": 0.0209, "step": 25960 }, { "epoch": 153.66863905325442, "grad_norm": 1.202386736869812, "learning_rate": 0.0001, "loss": 0.0218, "step": 25970 }, { "epoch": 153.72781065088756, "grad_norm": 1.6901218891143799, "learning_rate": 0.0001, "loss": 0.0211, "step": 25980 }, { "epoch": 153.7869822485207, "grad_norm": 1.1808362007141113, "learning_rate": 0.0001, "loss": 0.0211, "step": 25990 }, { "epoch": 153.84615384615384, "grad_norm": 1.4825023412704468, "learning_rate": 0.0001, "loss": 0.0209, "step": 26000 }, { "epoch": 153.90532544378698, "grad_norm": 1.2620928287506104, "learning_rate": 0.0001, "loss": 0.0221, "step": 26010 }, { "epoch": 153.96449704142012, "grad_norm": 1.4508581161499023, "learning_rate": 0.0001, "loss": 0.0217, "step": 26020 }, { "epoch": 154.02366863905326, "grad_norm": 1.293116569519043, "learning_rate": 0.0001, "loss": 0.0226, "step": 26030 }, { "epoch": 154.0828402366864, "grad_norm": 1.558098554611206, "learning_rate": 0.0001, "loss": 0.0235, "step": 26040 }, { "epoch": 154.14201183431953, "grad_norm": 1.4404038190841675, "learning_rate": 0.0001, "loss": 0.0237, "step": 26050 }, { "epoch": 154.20118343195267, "grad_norm": 1.5918000936508179, "learning_rate": 0.0001, "loss": 0.0228, "step": 26060 }, { "epoch": 154.2603550295858, "grad_norm": 1.4861390590667725, "learning_rate": 0.0001, "loss": 0.0226, "step": 26070 }, { "epoch": 154.31952662721895, "grad_norm": 1.310438871383667, "learning_rate": 0.0001, "loss": 0.0222, "step": 26080 }, { "epoch": 154.37869822485206, "grad_norm": 1.3104629516601562, "learning_rate": 0.0001, "loss": 0.0221, "step": 26090 }, { "epoch": 154.4378698224852, "grad_norm": 1.839911937713623, "learning_rate": 0.0001, "loss": 0.0225, "step": 26100 }, { "epoch": 154.49704142011834, "grad_norm": 1.215262770652771, "learning_rate": 0.0001, "loss": 0.0223, "step": 26110 }, { "epoch": 154.55621301775147, "grad_norm": 1.2230534553527832, "learning_rate": 0.0001, "loss": 0.0218, "step": 26120 }, { "epoch": 154.6153846153846, "grad_norm": 1.0349701642990112, "learning_rate": 0.0001, "loss": 0.0225, "step": 26130 }, { "epoch": 154.67455621301775, "grad_norm": 1.4535049200057983, "learning_rate": 0.0001, "loss": 0.0226, "step": 26140 }, { "epoch": 154.7337278106509, "grad_norm": 1.533899188041687, "learning_rate": 0.0001, "loss": 0.0231, "step": 26150 }, { "epoch": 154.79289940828403, "grad_norm": 1.1345545053482056, "learning_rate": 0.0001, "loss": 0.0235, "step": 26160 }, { "epoch": 154.85207100591717, "grad_norm": 1.0522841215133667, "learning_rate": 0.0001, "loss": 0.0234, "step": 26170 }, { "epoch": 154.9112426035503, "grad_norm": 1.2687146663665771, "learning_rate": 0.0001, "loss": 0.0224, "step": 26180 }, { "epoch": 154.97041420118344, "grad_norm": 1.0288112163543701, "learning_rate": 0.0001, "loss": 0.0229, "step": 26190 }, { "epoch": 155.02958579881656, "grad_norm": 1.6130789518356323, "learning_rate": 0.0001, "loss": 0.0228, "step": 26200 }, { "epoch": 155.0887573964497, "grad_norm": 1.1799490451812744, "learning_rate": 0.0001, "loss": 0.0233, "step": 26210 }, { "epoch": 155.14792899408283, "grad_norm": 1.1963014602661133, "learning_rate": 0.0001, "loss": 0.024, "step": 26220 }, { "epoch": 155.20710059171597, "grad_norm": 1.3003185987472534, "learning_rate": 0.0001, "loss": 0.0244, "step": 26230 }, { "epoch": 155.2662721893491, "grad_norm": 1.4847569465637207, "learning_rate": 0.0001, "loss": 0.0225, "step": 26240 }, { "epoch": 155.32544378698225, "grad_norm": 1.1167497634887695, "learning_rate": 0.0001, "loss": 0.0223, "step": 26250 }, { "epoch": 155.3846153846154, "grad_norm": 1.8370153903961182, "learning_rate": 0.0001, "loss": 0.0237, "step": 26260 }, { "epoch": 155.44378698224853, "grad_norm": 1.478394627571106, "learning_rate": 0.0001, "loss": 0.0228, "step": 26270 }, { "epoch": 155.50295857988166, "grad_norm": 1.0731257200241089, "learning_rate": 0.0001, "loss": 0.0221, "step": 26280 }, { "epoch": 155.5621301775148, "grad_norm": 1.0769842863082886, "learning_rate": 0.0001, "loss": 0.0222, "step": 26290 }, { "epoch": 155.62130177514794, "grad_norm": 1.1063053607940674, "learning_rate": 0.0001, "loss": 0.0224, "step": 26300 }, { "epoch": 155.68047337278105, "grad_norm": 1.5925976037979126, "learning_rate": 0.0001, "loss": 0.0219, "step": 26310 }, { "epoch": 155.7396449704142, "grad_norm": 1.2397032976150513, "learning_rate": 0.0001, "loss": 0.0217, "step": 26320 }, { "epoch": 155.79881656804733, "grad_norm": 1.2539854049682617, "learning_rate": 0.0001, "loss": 0.0217, "step": 26330 }, { "epoch": 155.85798816568047, "grad_norm": 1.462418556213379, "learning_rate": 0.0001, "loss": 0.0219, "step": 26340 }, { "epoch": 155.9171597633136, "grad_norm": 0.8448836803436279, "learning_rate": 0.0001, "loss": 0.0222, "step": 26350 }, { "epoch": 155.97633136094674, "grad_norm": 1.1645684242248535, "learning_rate": 0.0001, "loss": 0.0225, "step": 26360 }, { "epoch": 156.03550295857988, "grad_norm": 1.5057579278945923, "learning_rate": 0.0001, "loss": 0.0234, "step": 26370 }, { "epoch": 156.09467455621302, "grad_norm": 1.678221583366394, "learning_rate": 0.0001, "loss": 0.023, "step": 26380 }, { "epoch": 156.15384615384616, "grad_norm": 1.9284495115280151, "learning_rate": 0.0001, "loss": 0.022, "step": 26390 }, { "epoch": 156.2130177514793, "grad_norm": 1.4299999475479126, "learning_rate": 0.0001, "loss": 0.022, "step": 26400 }, { "epoch": 156.27218934911244, "grad_norm": 1.8384315967559814, "learning_rate": 0.0001, "loss": 0.0228, "step": 26410 }, { "epoch": 156.33136094674558, "grad_norm": 1.5994479656219482, "learning_rate": 0.0001, "loss": 0.0213, "step": 26420 }, { "epoch": 156.3905325443787, "grad_norm": 1.5976701974868774, "learning_rate": 0.0001, "loss": 0.0219, "step": 26430 }, { "epoch": 156.44970414201183, "grad_norm": 1.4358421564102173, "learning_rate": 0.0001, "loss": 0.0215, "step": 26440 }, { "epoch": 156.50887573964496, "grad_norm": 2.100712776184082, "learning_rate": 0.0001, "loss": 0.0211, "step": 26450 }, { "epoch": 156.5680473372781, "grad_norm": 2.29032039642334, "learning_rate": 0.0001, "loss": 0.0213, "step": 26460 }, { "epoch": 156.62721893491124, "grad_norm": 1.8885473012924194, "learning_rate": 0.0001, "loss": 0.0198, "step": 26470 }, { "epoch": 156.68639053254438, "grad_norm": 1.6040687561035156, "learning_rate": 0.0001, "loss": 0.0203, "step": 26480 }, { "epoch": 156.74556213017752, "grad_norm": 1.368825912475586, "learning_rate": 0.0001, "loss": 0.0211, "step": 26490 }, { "epoch": 156.80473372781066, "grad_norm": 1.5657836198806763, "learning_rate": 0.0001, "loss": 0.0202, "step": 26500 }, { "epoch": 156.8639053254438, "grad_norm": 1.537675380706787, "learning_rate": 0.0001, "loss": 0.0211, "step": 26510 }, { "epoch": 156.92307692307693, "grad_norm": 1.3927247524261475, "learning_rate": 0.0001, "loss": 0.0209, "step": 26520 }, { "epoch": 156.98224852071007, "grad_norm": 2.044133186340332, "learning_rate": 0.0001, "loss": 0.0218, "step": 26530 }, { "epoch": 157.04142011834318, "grad_norm": 1.6439173221588135, "learning_rate": 0.0001, "loss": 0.0212, "step": 26540 }, { "epoch": 157.10059171597632, "grad_norm": 1.103346824645996, "learning_rate": 0.0001, "loss": 0.0209, "step": 26550 }, { "epoch": 157.15976331360946, "grad_norm": 1.562464952468872, "learning_rate": 0.0001, "loss": 0.021, "step": 26560 }, { "epoch": 157.2189349112426, "grad_norm": 1.5315884351730347, "learning_rate": 0.0001, "loss": 0.022, "step": 26570 }, { "epoch": 157.27810650887574, "grad_norm": 1.500468134880066, "learning_rate": 0.0001, "loss": 0.021, "step": 26580 }, { "epoch": 157.33727810650888, "grad_norm": 1.8212531805038452, "learning_rate": 0.0001, "loss": 0.0211, "step": 26590 }, { "epoch": 157.39644970414201, "grad_norm": 1.5647459030151367, "learning_rate": 0.0001, "loss": 0.0214, "step": 26600 }, { "epoch": 157.45562130177515, "grad_norm": 1.7350488901138306, "learning_rate": 0.0001, "loss": 0.0224, "step": 26610 }, { "epoch": 157.5147928994083, "grad_norm": 1.4916423559188843, "learning_rate": 0.0001, "loss": 0.0222, "step": 26620 }, { "epoch": 157.57396449704143, "grad_norm": 1.5763801336288452, "learning_rate": 0.0001, "loss": 0.022, "step": 26630 }, { "epoch": 157.63313609467457, "grad_norm": 1.5139557123184204, "learning_rate": 0.0001, "loss": 0.023, "step": 26640 }, { "epoch": 157.69230769230768, "grad_norm": 1.1644935607910156, "learning_rate": 0.0001, "loss": 0.0223, "step": 26650 }, { "epoch": 157.75147928994082, "grad_norm": 1.5471253395080566, "learning_rate": 0.0001, "loss": 0.0212, "step": 26660 }, { "epoch": 157.81065088757396, "grad_norm": 1.215193748474121, "learning_rate": 0.0001, "loss": 0.0216, "step": 26670 }, { "epoch": 157.8698224852071, "grad_norm": 1.4326555728912354, "learning_rate": 0.0001, "loss": 0.022, "step": 26680 }, { "epoch": 157.92899408284023, "grad_norm": 1.3947888612747192, "learning_rate": 0.0001, "loss": 0.0223, "step": 26690 }, { "epoch": 157.98816568047337, "grad_norm": 1.4785107374191284, "learning_rate": 0.0001, "loss": 0.0218, "step": 26700 }, { "epoch": 158.0473372781065, "grad_norm": 2.206202983856201, "learning_rate": 0.0001, "loss": 0.0226, "step": 26710 }, { "epoch": 158.10650887573965, "grad_norm": 1.010291337966919, "learning_rate": 0.0001, "loss": 0.0217, "step": 26720 }, { "epoch": 158.1656804733728, "grad_norm": 1.6590285301208496, "learning_rate": 0.0001, "loss": 0.022, "step": 26730 }, { "epoch": 158.22485207100593, "grad_norm": 1.3147261142730713, "learning_rate": 0.0001, "loss": 0.0213, "step": 26740 }, { "epoch": 158.28402366863907, "grad_norm": 0.8524631261825562, "learning_rate": 0.0001, "loss": 0.0216, "step": 26750 }, { "epoch": 158.34319526627218, "grad_norm": 1.9306163787841797, "learning_rate": 0.0001, "loss": 0.0205, "step": 26760 }, { "epoch": 158.4023668639053, "grad_norm": 1.8532675504684448, "learning_rate": 0.0001, "loss": 0.0218, "step": 26770 }, { "epoch": 158.46153846153845, "grad_norm": 1.7881739139556885, "learning_rate": 0.0001, "loss": 0.0226, "step": 26780 }, { "epoch": 158.5207100591716, "grad_norm": 1.0406032800674438, "learning_rate": 0.0001, "loss": 0.021, "step": 26790 }, { "epoch": 158.57988165680473, "grad_norm": 1.8241533041000366, "learning_rate": 0.0001, "loss": 0.0217, "step": 26800 }, { "epoch": 158.63905325443787, "grad_norm": 1.3107279539108276, "learning_rate": 0.0001, "loss": 0.0214, "step": 26810 }, { "epoch": 158.698224852071, "grad_norm": 1.6401652097702026, "learning_rate": 0.0001, "loss": 0.0216, "step": 26820 }, { "epoch": 158.75739644970415, "grad_norm": 1.7146565914154053, "learning_rate": 0.0001, "loss": 0.0207, "step": 26830 }, { "epoch": 158.81656804733728, "grad_norm": 1.332451581954956, "learning_rate": 0.0001, "loss": 0.0216, "step": 26840 }, { "epoch": 158.87573964497042, "grad_norm": 0.921194314956665, "learning_rate": 0.0001, "loss": 0.022, "step": 26850 }, { "epoch": 158.93491124260356, "grad_norm": 1.5866293907165527, "learning_rate": 0.0001, "loss": 0.0209, "step": 26860 }, { "epoch": 158.9940828402367, "grad_norm": 1.4004827737808228, "learning_rate": 0.0001, "loss": 0.0222, "step": 26870 }, { "epoch": 159.0532544378698, "grad_norm": 1.4005285501480103, "learning_rate": 0.0001, "loss": 0.021, "step": 26880 }, { "epoch": 159.11242603550295, "grad_norm": 1.3820929527282715, "learning_rate": 0.0001, "loss": 0.0216, "step": 26890 }, { "epoch": 159.1715976331361, "grad_norm": 1.3166232109069824, "learning_rate": 0.0001, "loss": 0.0216, "step": 26900 }, { "epoch": 159.23076923076923, "grad_norm": 1.4804691076278687, "learning_rate": 0.0001, "loss": 0.0219, "step": 26910 }, { "epoch": 159.28994082840237, "grad_norm": 1.5785678625106812, "learning_rate": 0.0001, "loss": 0.0218, "step": 26920 }, { "epoch": 159.3491124260355, "grad_norm": 1.4850966930389404, "learning_rate": 0.0001, "loss": 0.0228, "step": 26930 }, { "epoch": 159.40828402366864, "grad_norm": 1.1138055324554443, "learning_rate": 0.0001, "loss": 0.0213, "step": 26940 }, { "epoch": 159.46745562130178, "grad_norm": 1.4456262588500977, "learning_rate": 0.0001, "loss": 0.0217, "step": 26950 }, { "epoch": 159.52662721893492, "grad_norm": 1.313808560371399, "learning_rate": 0.0001, "loss": 0.0225, "step": 26960 }, { "epoch": 159.58579881656806, "grad_norm": 1.4746092557907104, "learning_rate": 0.0001, "loss": 0.0213, "step": 26970 }, { "epoch": 159.6449704142012, "grad_norm": 1.2902840375900269, "learning_rate": 0.0001, "loss": 0.0219, "step": 26980 }, { "epoch": 159.7041420118343, "grad_norm": 1.4305999279022217, "learning_rate": 0.0001, "loss": 0.023, "step": 26990 }, { "epoch": 159.76331360946745, "grad_norm": 1.426767110824585, "learning_rate": 0.0001, "loss": 0.0226, "step": 27000 }, { "epoch": 159.82248520710058, "grad_norm": 1.5294355154037476, "learning_rate": 0.0001, "loss": 0.0215, "step": 27010 }, { "epoch": 159.88165680473372, "grad_norm": 1.376447081565857, "learning_rate": 0.0001, "loss": 0.0219, "step": 27020 }, { "epoch": 159.94082840236686, "grad_norm": 1.552646279335022, "learning_rate": 0.0001, "loss": 0.0206, "step": 27030 }, { "epoch": 160.0, "grad_norm": 1.5042287111282349, "learning_rate": 0.0001, "loss": 0.0221, "step": 27040 }, { "epoch": 160.05917159763314, "grad_norm": 1.638505458831787, "learning_rate": 0.0001, "loss": 0.0216, "step": 27050 }, { "epoch": 160.11834319526628, "grad_norm": 1.0262218713760376, "learning_rate": 0.0001, "loss": 0.0223, "step": 27060 }, { "epoch": 160.17751479289942, "grad_norm": 1.69545316696167, "learning_rate": 0.0001, "loss": 0.0215, "step": 27070 }, { "epoch": 160.23668639053255, "grad_norm": 1.486509919166565, "learning_rate": 0.0001, "loss": 0.0216, "step": 27080 }, { "epoch": 160.2958579881657, "grad_norm": 1.5458452701568604, "learning_rate": 0.0001, "loss": 0.0219, "step": 27090 }, { "epoch": 160.3550295857988, "grad_norm": 1.0300132036209106, "learning_rate": 0.0001, "loss": 0.0218, "step": 27100 }, { "epoch": 160.41420118343194, "grad_norm": 1.2342203855514526, "learning_rate": 0.0001, "loss": 0.0209, "step": 27110 }, { "epoch": 160.47337278106508, "grad_norm": 0.8687087893486023, "learning_rate": 0.0001, "loss": 0.0207, "step": 27120 }, { "epoch": 160.53254437869822, "grad_norm": 1.1241142749786377, "learning_rate": 0.0001, "loss": 0.022, "step": 27130 }, { "epoch": 160.59171597633136, "grad_norm": 1.2471928596496582, "learning_rate": 0.0001, "loss": 0.0225, "step": 27140 }, { "epoch": 160.6508875739645, "grad_norm": 1.2565630674362183, "learning_rate": 0.0001, "loss": 0.0216, "step": 27150 }, { "epoch": 160.71005917159763, "grad_norm": 1.4829155206680298, "learning_rate": 0.0001, "loss": 0.0226, "step": 27160 }, { "epoch": 160.76923076923077, "grad_norm": 1.2747150659561157, "learning_rate": 0.0001, "loss": 0.0217, "step": 27170 }, { "epoch": 160.8284023668639, "grad_norm": 1.122566819190979, "learning_rate": 0.0001, "loss": 0.0221, "step": 27180 }, { "epoch": 160.88757396449705, "grad_norm": 1.437368392944336, "learning_rate": 0.0001, "loss": 0.0225, "step": 27190 }, { "epoch": 160.9467455621302, "grad_norm": 1.7764850854873657, "learning_rate": 0.0001, "loss": 0.0224, "step": 27200 }, { "epoch": 161.0059171597633, "grad_norm": 1.1024115085601807, "learning_rate": 0.0001, "loss": 0.0229, "step": 27210 }, { "epoch": 161.06508875739644, "grad_norm": 1.6187024116516113, "learning_rate": 0.0001, "loss": 0.0214, "step": 27220 }, { "epoch": 161.12426035502958, "grad_norm": 1.137972354888916, "learning_rate": 0.0001, "loss": 0.0216, "step": 27230 }, { "epoch": 161.18343195266272, "grad_norm": 0.9170749187469482, "learning_rate": 0.0001, "loss": 0.021, "step": 27240 }, { "epoch": 161.24260355029585, "grad_norm": 1.309045433998108, "learning_rate": 0.0001, "loss": 0.0221, "step": 27250 }, { "epoch": 161.301775147929, "grad_norm": 1.3391776084899902, "learning_rate": 0.0001, "loss": 0.0214, "step": 27260 }, { "epoch": 161.36094674556213, "grad_norm": 1.734952688217163, "learning_rate": 0.0001, "loss": 0.0216, "step": 27270 }, { "epoch": 161.42011834319527, "grad_norm": 2.128032922744751, "learning_rate": 0.0001, "loss": 0.0221, "step": 27280 }, { "epoch": 161.4792899408284, "grad_norm": 1.7427895069122314, "learning_rate": 0.0001, "loss": 0.0212, "step": 27290 }, { "epoch": 161.53846153846155, "grad_norm": 2.028836488723755, "learning_rate": 0.0001, "loss": 0.0206, "step": 27300 }, { "epoch": 161.5976331360947, "grad_norm": 1.3314130306243896, "learning_rate": 0.0001, "loss": 0.0209, "step": 27310 }, { "epoch": 161.65680473372782, "grad_norm": 1.874639868736267, "learning_rate": 0.0001, "loss": 0.021, "step": 27320 }, { "epoch": 161.71597633136093, "grad_norm": 1.389588713645935, "learning_rate": 0.0001, "loss": 0.0201, "step": 27330 }, { "epoch": 161.77514792899407, "grad_norm": 1.1544586420059204, "learning_rate": 0.0001, "loss": 0.0189, "step": 27340 }, { "epoch": 161.8343195266272, "grad_norm": 1.355604887008667, "learning_rate": 0.0001, "loss": 0.0212, "step": 27350 }, { "epoch": 161.89349112426035, "grad_norm": 1.3631455898284912, "learning_rate": 0.0001, "loss": 0.0208, "step": 27360 }, { "epoch": 161.9526627218935, "grad_norm": 1.2096742391586304, "learning_rate": 0.0001, "loss": 0.0206, "step": 27370 }, { "epoch": 162.01183431952663, "grad_norm": 1.8768879175186157, "learning_rate": 0.0001, "loss": 0.0209, "step": 27380 }, { "epoch": 162.07100591715977, "grad_norm": 1.7900421619415283, "learning_rate": 0.0001, "loss": 0.0205, "step": 27390 }, { "epoch": 162.1301775147929, "grad_norm": 2.3071255683898926, "learning_rate": 0.0001, "loss": 0.0222, "step": 27400 }, { "epoch": 162.18934911242604, "grad_norm": 2.415456771850586, "learning_rate": 0.0001, "loss": 0.0221, "step": 27410 }, { "epoch": 162.24852071005918, "grad_norm": 2.58857798576355, "learning_rate": 0.0001, "loss": 0.0219, "step": 27420 }, { "epoch": 162.30769230769232, "grad_norm": 2.234882116317749, "learning_rate": 0.0001, "loss": 0.0218, "step": 27430 }, { "epoch": 162.36686390532543, "grad_norm": 1.587390661239624, "learning_rate": 0.0001, "loss": 0.0211, "step": 27440 }, { "epoch": 162.42603550295857, "grad_norm": 1.649149775505066, "learning_rate": 0.0001, "loss": 0.0201, "step": 27450 }, { "epoch": 162.4852071005917, "grad_norm": 1.8875675201416016, "learning_rate": 0.0001, "loss": 0.0201, "step": 27460 }, { "epoch": 162.54437869822485, "grad_norm": 2.5878124237060547, "learning_rate": 0.0001, "loss": 0.0207, "step": 27470 }, { "epoch": 162.60355029585799, "grad_norm": 2.2977333068847656, "learning_rate": 0.0001, "loss": 0.0208, "step": 27480 }, { "epoch": 162.66272189349112, "grad_norm": 2.288339138031006, "learning_rate": 0.0001, "loss": 0.0213, "step": 27490 }, { "epoch": 162.72189349112426, "grad_norm": 1.84151029586792, "learning_rate": 0.0001, "loss": 0.0206, "step": 27500 }, { "epoch": 162.7810650887574, "grad_norm": 1.5507512092590332, "learning_rate": 0.0001, "loss": 0.0212, "step": 27510 }, { "epoch": 162.84023668639054, "grad_norm": 1.1983089447021484, "learning_rate": 0.0001, "loss": 0.0209, "step": 27520 }, { "epoch": 162.89940828402368, "grad_norm": 1.3389161825180054, "learning_rate": 0.0001, "loss": 0.0199, "step": 27530 }, { "epoch": 162.95857988165682, "grad_norm": 1.4682458639144897, "learning_rate": 0.0001, "loss": 0.0216, "step": 27540 }, { "epoch": 163.01775147928993, "grad_norm": 1.7260527610778809, "learning_rate": 0.0001, "loss": 0.0212, "step": 27550 }, { "epoch": 163.07692307692307, "grad_norm": 1.373134732246399, "learning_rate": 0.0001, "loss": 0.0215, "step": 27560 }, { "epoch": 163.1360946745562, "grad_norm": 1.6699734926223755, "learning_rate": 0.0001, "loss": 0.0213, "step": 27570 }, { "epoch": 163.19526627218934, "grad_norm": 1.4435162544250488, "learning_rate": 0.0001, "loss": 0.0216, "step": 27580 }, { "epoch": 163.25443786982248, "grad_norm": 1.1775407791137695, "learning_rate": 0.0001, "loss": 0.0218, "step": 27590 }, { "epoch": 163.31360946745562, "grad_norm": 1.5373494625091553, "learning_rate": 0.0001, "loss": 0.0202, "step": 27600 }, { "epoch": 163.37278106508876, "grad_norm": 1.3716343641281128, "learning_rate": 0.0001, "loss": 0.0224, "step": 27610 }, { "epoch": 163.4319526627219, "grad_norm": 1.4562708139419556, "learning_rate": 0.0001, "loss": 0.0226, "step": 27620 }, { "epoch": 163.49112426035504, "grad_norm": 2.694645643234253, "learning_rate": 0.0001, "loss": 0.0219, "step": 27630 }, { "epoch": 163.55029585798817, "grad_norm": 2.3930859565734863, "learning_rate": 0.0001, "loss": 0.0227, "step": 27640 }, { "epoch": 163.6094674556213, "grad_norm": 1.627415418624878, "learning_rate": 0.0001, "loss": 0.0219, "step": 27650 }, { "epoch": 163.66863905325442, "grad_norm": 1.5821336507797241, "learning_rate": 0.0001, "loss": 0.0213, "step": 27660 }, { "epoch": 163.72781065088756, "grad_norm": 1.251909852027893, "learning_rate": 0.0001, "loss": 0.0213, "step": 27670 }, { "epoch": 163.7869822485207, "grad_norm": 1.12010657787323, "learning_rate": 0.0001, "loss": 0.0207, "step": 27680 }, { "epoch": 163.84615384615384, "grad_norm": 1.5870567560195923, "learning_rate": 0.0001, "loss": 0.0225, "step": 27690 }, { "epoch": 163.90532544378698, "grad_norm": 1.215476393699646, "learning_rate": 0.0001, "loss": 0.021, "step": 27700 }, { "epoch": 163.96449704142012, "grad_norm": 1.4277054071426392, "learning_rate": 0.0001, "loss": 0.0219, "step": 27710 }, { "epoch": 164.02366863905326, "grad_norm": 1.6708316802978516, "learning_rate": 0.0001, "loss": 0.0214, "step": 27720 }, { "epoch": 164.0828402366864, "grad_norm": 2.018012285232544, "learning_rate": 0.0001, "loss": 0.0216, "step": 27730 }, { "epoch": 164.14201183431953, "grad_norm": 1.8078501224517822, "learning_rate": 0.0001, "loss": 0.0219, "step": 27740 }, { "epoch": 164.20118343195267, "grad_norm": 1.5459134578704834, "learning_rate": 0.0001, "loss": 0.0211, "step": 27750 }, { "epoch": 164.2603550295858, "grad_norm": 1.6514554023742676, "learning_rate": 0.0001, "loss": 0.0204, "step": 27760 }, { "epoch": 164.31952662721895, "grad_norm": 1.8051819801330566, "learning_rate": 0.0001, "loss": 0.0211, "step": 27770 }, { "epoch": 164.37869822485206, "grad_norm": 1.9423887729644775, "learning_rate": 0.0001, "loss": 0.0217, "step": 27780 }, { "epoch": 164.4378698224852, "grad_norm": 1.2297255992889404, "learning_rate": 0.0001, "loss": 0.021, "step": 27790 }, { "epoch": 164.49704142011834, "grad_norm": 1.6890573501586914, "learning_rate": 0.0001, "loss": 0.0209, "step": 27800 }, { "epoch": 164.55621301775147, "grad_norm": 1.343502163887024, "learning_rate": 0.0001, "loss": 0.0209, "step": 27810 }, { "epoch": 164.6153846153846, "grad_norm": 1.7594904899597168, "learning_rate": 0.0001, "loss": 0.0211, "step": 27820 }, { "epoch": 164.67455621301775, "grad_norm": 1.500885248184204, "learning_rate": 0.0001, "loss": 0.0213, "step": 27830 }, { "epoch": 164.7337278106509, "grad_norm": 1.3914772272109985, "learning_rate": 0.0001, "loss": 0.0215, "step": 27840 }, { "epoch": 164.79289940828403, "grad_norm": 1.9226372241973877, "learning_rate": 0.0001, "loss": 0.0207, "step": 27850 }, { "epoch": 164.85207100591717, "grad_norm": 1.6534043550491333, "learning_rate": 0.0001, "loss": 0.0212, "step": 27860 }, { "epoch": 164.9112426035503, "grad_norm": 1.76524019241333, "learning_rate": 0.0001, "loss": 0.021, "step": 27870 }, { "epoch": 164.97041420118344, "grad_norm": 1.2286968231201172, "learning_rate": 0.0001, "loss": 0.0212, "step": 27880 }, { "epoch": 165.02958579881656, "grad_norm": 1.443534016609192, "learning_rate": 0.0001, "loss": 0.0216, "step": 27890 }, { "epoch": 165.0887573964497, "grad_norm": 1.0847570896148682, "learning_rate": 0.0001, "loss": 0.0206, "step": 27900 }, { "epoch": 165.14792899408283, "grad_norm": 1.1272469758987427, "learning_rate": 0.0001, "loss": 0.0205, "step": 27910 }, { "epoch": 165.20710059171597, "grad_norm": 1.3934510946273804, "learning_rate": 0.0001, "loss": 0.0204, "step": 27920 }, { "epoch": 165.2662721893491, "grad_norm": 1.929265022277832, "learning_rate": 0.0001, "loss": 0.0211, "step": 27930 }, { "epoch": 165.32544378698225, "grad_norm": 1.7919000387191772, "learning_rate": 0.0001, "loss": 0.0221, "step": 27940 }, { "epoch": 165.3846153846154, "grad_norm": 1.4646437168121338, "learning_rate": 0.0001, "loss": 0.0218, "step": 27950 }, { "epoch": 165.44378698224853, "grad_norm": 1.2067160606384277, "learning_rate": 0.0001, "loss": 0.0213, "step": 27960 }, { "epoch": 165.50295857988166, "grad_norm": 1.6266251802444458, "learning_rate": 0.0001, "loss": 0.02, "step": 27970 }, { "epoch": 165.5621301775148, "grad_norm": 1.1977802515029907, "learning_rate": 0.0001, "loss": 0.0206, "step": 27980 }, { "epoch": 165.62130177514794, "grad_norm": 1.0527373552322388, "learning_rate": 0.0001, "loss": 0.0206, "step": 27990 }, { "epoch": 165.68047337278105, "grad_norm": 1.781083583831787, "learning_rate": 0.0001, "loss": 0.0211, "step": 28000 }, { "epoch": 165.7396449704142, "grad_norm": 1.4081273078918457, "learning_rate": 0.0001, "loss": 0.022, "step": 28010 }, { "epoch": 165.79881656804733, "grad_norm": 1.271606683731079, "learning_rate": 0.0001, "loss": 0.0215, "step": 28020 }, { "epoch": 165.85798816568047, "grad_norm": 1.7599005699157715, "learning_rate": 0.0001, "loss": 0.0219, "step": 28030 }, { "epoch": 165.9171597633136, "grad_norm": 1.1158000230789185, "learning_rate": 0.0001, "loss": 0.0224, "step": 28040 }, { "epoch": 165.97633136094674, "grad_norm": 1.8900973796844482, "learning_rate": 0.0001, "loss": 0.0215, "step": 28050 }, { "epoch": 166.03550295857988, "grad_norm": 1.992554783821106, "learning_rate": 0.0001, "loss": 0.0229, "step": 28060 }, { "epoch": 166.09467455621302, "grad_norm": 1.3816494941711426, "learning_rate": 0.0001, "loss": 0.0221, "step": 28070 }, { "epoch": 166.15384615384616, "grad_norm": 1.087001085281372, "learning_rate": 0.0001, "loss": 0.022, "step": 28080 }, { "epoch": 166.2130177514793, "grad_norm": 1.5316556692123413, "learning_rate": 0.0001, "loss": 0.0218, "step": 28090 }, { "epoch": 166.27218934911244, "grad_norm": 1.687706708908081, "learning_rate": 0.0001, "loss": 0.0217, "step": 28100 }, { "epoch": 166.33136094674558, "grad_norm": 1.1238642930984497, "learning_rate": 0.0001, "loss": 0.0216, "step": 28110 }, { "epoch": 166.3905325443787, "grad_norm": 1.627586007118225, "learning_rate": 0.0001, "loss": 0.0212, "step": 28120 }, { "epoch": 166.44970414201183, "grad_norm": 1.140436053276062, "learning_rate": 0.0001, "loss": 0.0223, "step": 28130 }, { "epoch": 166.50887573964496, "grad_norm": 1.424795389175415, "learning_rate": 0.0001, "loss": 0.0228, "step": 28140 }, { "epoch": 166.5680473372781, "grad_norm": 1.4458030462265015, "learning_rate": 0.0001, "loss": 0.0215, "step": 28150 }, { "epoch": 166.62721893491124, "grad_norm": 1.498218059539795, "learning_rate": 0.0001, "loss": 0.0232, "step": 28160 }, { "epoch": 166.68639053254438, "grad_norm": 1.484844446182251, "learning_rate": 0.0001, "loss": 0.0214, "step": 28170 }, { "epoch": 166.74556213017752, "grad_norm": 1.5950144529342651, "learning_rate": 0.0001, "loss": 0.0222, "step": 28180 }, { "epoch": 166.80473372781066, "grad_norm": 1.2728520631790161, "learning_rate": 0.0001, "loss": 0.022, "step": 28190 }, { "epoch": 166.8639053254438, "grad_norm": 1.305268406867981, "learning_rate": 0.0001, "loss": 0.0219, "step": 28200 }, { "epoch": 166.92307692307693, "grad_norm": 1.6219807863235474, "learning_rate": 0.0001, "loss": 0.0213, "step": 28210 }, { "epoch": 166.98224852071007, "grad_norm": 1.6583833694458008, "learning_rate": 0.0001, "loss": 0.021, "step": 28220 }, { "epoch": 167.04142011834318, "grad_norm": 1.6135121583938599, "learning_rate": 0.0001, "loss": 0.0219, "step": 28230 }, { "epoch": 167.10059171597632, "grad_norm": 1.662473201751709, "learning_rate": 0.0001, "loss": 0.0217, "step": 28240 }, { "epoch": 167.15976331360946, "grad_norm": 1.846085786819458, "learning_rate": 0.0001, "loss": 0.0225, "step": 28250 }, { "epoch": 167.2189349112426, "grad_norm": 1.4071134328842163, "learning_rate": 0.0001, "loss": 0.021, "step": 28260 }, { "epoch": 167.27810650887574, "grad_norm": 1.4139795303344727, "learning_rate": 0.0001, "loss": 0.0209, "step": 28270 }, { "epoch": 167.33727810650888, "grad_norm": 1.661391258239746, "learning_rate": 0.0001, "loss": 0.021, "step": 28280 }, { "epoch": 167.39644970414201, "grad_norm": 1.4083620309829712, "learning_rate": 0.0001, "loss": 0.0215, "step": 28290 }, { "epoch": 167.45562130177515, "grad_norm": 1.3975025415420532, "learning_rate": 0.0001, "loss": 0.0211, "step": 28300 }, { "epoch": 167.5147928994083, "grad_norm": 1.1331710815429688, "learning_rate": 0.0001, "loss": 0.0225, "step": 28310 }, { "epoch": 167.57396449704143, "grad_norm": 1.4921400547027588, "learning_rate": 0.0001, "loss": 0.0207, "step": 28320 }, { "epoch": 167.63313609467457, "grad_norm": 0.9900938272476196, "learning_rate": 0.0001, "loss": 0.022, "step": 28330 }, { "epoch": 167.69230769230768, "grad_norm": 1.5407072305679321, "learning_rate": 0.0001, "loss": 0.0213, "step": 28340 }, { "epoch": 167.75147928994082, "grad_norm": 1.450128197669983, "learning_rate": 0.0001, "loss": 0.0227, "step": 28350 }, { "epoch": 167.81065088757396, "grad_norm": 1.6829075813293457, "learning_rate": 0.0001, "loss": 0.0219, "step": 28360 }, { "epoch": 167.8698224852071, "grad_norm": 1.7010612487792969, "learning_rate": 0.0001, "loss": 0.0216, "step": 28370 }, { "epoch": 167.92899408284023, "grad_norm": 1.1187907457351685, "learning_rate": 0.0001, "loss": 0.0219, "step": 28380 }, { "epoch": 167.98816568047337, "grad_norm": 1.0838987827301025, "learning_rate": 0.0001, "loss": 0.0207, "step": 28390 }, { "epoch": 168.0473372781065, "grad_norm": 1.1639500856399536, "learning_rate": 0.0001, "loss": 0.022, "step": 28400 }, { "epoch": 168.10650887573965, "grad_norm": 1.407523274421692, "learning_rate": 0.0001, "loss": 0.0208, "step": 28410 }, { "epoch": 168.1656804733728, "grad_norm": 1.9700653553009033, "learning_rate": 0.0001, "loss": 0.0214, "step": 28420 }, { "epoch": 168.22485207100593, "grad_norm": 1.4049782752990723, "learning_rate": 0.0001, "loss": 0.0223, "step": 28430 }, { "epoch": 168.28402366863907, "grad_norm": 1.4907617568969727, "learning_rate": 0.0001, "loss": 0.0206, "step": 28440 }, { "epoch": 168.34319526627218, "grad_norm": 1.5994184017181396, "learning_rate": 0.0001, "loss": 0.0216, "step": 28450 }, { "epoch": 168.4023668639053, "grad_norm": 1.6920254230499268, "learning_rate": 0.0001, "loss": 0.0212, "step": 28460 }, { "epoch": 168.46153846153845, "grad_norm": 1.4232008457183838, "learning_rate": 0.0001, "loss": 0.0205, "step": 28470 }, { "epoch": 168.5207100591716, "grad_norm": 1.1945669651031494, "learning_rate": 0.0001, "loss": 0.0208, "step": 28480 }, { "epoch": 168.57988165680473, "grad_norm": 1.3460824489593506, "learning_rate": 0.0001, "loss": 0.0218, "step": 28490 }, { "epoch": 168.63905325443787, "grad_norm": 1.665669322013855, "learning_rate": 0.0001, "loss": 0.021, "step": 28500 }, { "epoch": 168.698224852071, "grad_norm": 1.051431655883789, "learning_rate": 0.0001, "loss": 0.0208, "step": 28510 }, { "epoch": 168.75739644970415, "grad_norm": 2.1920337677001953, "learning_rate": 0.0001, "loss": 0.0215, "step": 28520 }, { "epoch": 168.81656804733728, "grad_norm": 0.9811384677886963, "learning_rate": 0.0001, "loss": 0.022, "step": 28530 }, { "epoch": 168.87573964497042, "grad_norm": 1.5311201810836792, "learning_rate": 0.0001, "loss": 0.0215, "step": 28540 }, { "epoch": 168.93491124260356, "grad_norm": 1.7538957595825195, "learning_rate": 0.0001, "loss": 0.0218, "step": 28550 }, { "epoch": 168.9940828402367, "grad_norm": 1.6627498865127563, "learning_rate": 0.0001, "loss": 0.0212, "step": 28560 }, { "epoch": 169.0532544378698, "grad_norm": 1.848078727722168, "learning_rate": 0.0001, "loss": 0.0211, "step": 28570 }, { "epoch": 169.11242603550295, "grad_norm": 1.2957017421722412, "learning_rate": 0.0001, "loss": 0.0201, "step": 28580 }, { "epoch": 169.1715976331361, "grad_norm": 1.4336774349212646, "learning_rate": 0.0001, "loss": 0.0212, "step": 28590 }, { "epoch": 169.23076923076923, "grad_norm": 1.802724003791809, "learning_rate": 0.0001, "loss": 0.0214, "step": 28600 }, { "epoch": 169.28994082840237, "grad_norm": 0.9775751233100891, "learning_rate": 0.0001, "loss": 0.0204, "step": 28610 }, { "epoch": 169.3491124260355, "grad_norm": 1.2969083786010742, "learning_rate": 0.0001, "loss": 0.0204, "step": 28620 }, { "epoch": 169.40828402366864, "grad_norm": 1.6688518524169922, "learning_rate": 0.0001, "loss": 0.0201, "step": 28630 }, { "epoch": 169.46745562130178, "grad_norm": 1.2649345397949219, "learning_rate": 0.0001, "loss": 0.0209, "step": 28640 }, { "epoch": 169.52662721893492, "grad_norm": 1.6280725002288818, "learning_rate": 0.0001, "loss": 0.0192, "step": 28650 }, { "epoch": 169.58579881656806, "grad_norm": 1.4836041927337646, "learning_rate": 0.0001, "loss": 0.0201, "step": 28660 }, { "epoch": 169.6449704142012, "grad_norm": 1.532708764076233, "learning_rate": 0.0001, "loss": 0.0206, "step": 28670 }, { "epoch": 169.7041420118343, "grad_norm": 1.527893304824829, "learning_rate": 0.0001, "loss": 0.0206, "step": 28680 }, { "epoch": 169.76331360946745, "grad_norm": 1.6263209581375122, "learning_rate": 0.0001, "loss": 0.0203, "step": 28690 }, { "epoch": 169.82248520710058, "grad_norm": 1.4840314388275146, "learning_rate": 0.0001, "loss": 0.0205, "step": 28700 }, { "epoch": 169.88165680473372, "grad_norm": 1.2375150918960571, "learning_rate": 0.0001, "loss": 0.0207, "step": 28710 }, { "epoch": 169.94082840236686, "grad_norm": 1.8745750188827515, "learning_rate": 0.0001, "loss": 0.0213, "step": 28720 }, { "epoch": 170.0, "grad_norm": 1.5996590852737427, "learning_rate": 0.0001, "loss": 0.0208, "step": 28730 }, { "epoch": 170.05917159763314, "grad_norm": 2.414541721343994, "learning_rate": 0.0001, "loss": 0.0206, "step": 28740 }, { "epoch": 170.11834319526628, "grad_norm": 2.0853919982910156, "learning_rate": 0.0001, "loss": 0.0209, "step": 28750 }, { "epoch": 170.17751479289942, "grad_norm": 1.5541712045669556, "learning_rate": 0.0001, "loss": 0.021, "step": 28760 }, { "epoch": 170.23668639053255, "grad_norm": 1.475528359413147, "learning_rate": 0.0001, "loss": 0.0206, "step": 28770 }, { "epoch": 170.2958579881657, "grad_norm": 1.5770002603530884, "learning_rate": 0.0001, "loss": 0.02, "step": 28780 }, { "epoch": 170.3550295857988, "grad_norm": 1.3127965927124023, "learning_rate": 0.0001, "loss": 0.0206, "step": 28790 }, { "epoch": 170.41420118343194, "grad_norm": 1.4835312366485596, "learning_rate": 0.0001, "loss": 0.0212, "step": 28800 }, { "epoch": 170.47337278106508, "grad_norm": 1.170835018157959, "learning_rate": 0.0001, "loss": 0.0207, "step": 28810 }, { "epoch": 170.53254437869822, "grad_norm": 1.169121503829956, "learning_rate": 0.0001, "loss": 0.0209, "step": 28820 }, { "epoch": 170.59171597633136, "grad_norm": 1.374738097190857, "learning_rate": 0.0001, "loss": 0.0217, "step": 28830 }, { "epoch": 170.6508875739645, "grad_norm": 1.6377079486846924, "learning_rate": 0.0001, "loss": 0.0203, "step": 28840 }, { "epoch": 170.71005917159763, "grad_norm": 1.885909080505371, "learning_rate": 0.0001, "loss": 0.0207, "step": 28850 }, { "epoch": 170.76923076923077, "grad_norm": 1.9170241355895996, "learning_rate": 0.0001, "loss": 0.0207, "step": 28860 }, { "epoch": 170.8284023668639, "grad_norm": 1.4142237901687622, "learning_rate": 0.0001, "loss": 0.0208, "step": 28870 }, { "epoch": 170.88757396449705, "grad_norm": 1.774638295173645, "learning_rate": 0.0001, "loss": 0.0206, "step": 28880 }, { "epoch": 170.9467455621302, "grad_norm": 1.4028193950653076, "learning_rate": 0.0001, "loss": 0.0213, "step": 28890 }, { "epoch": 171.0059171597633, "grad_norm": 1.2838197946548462, "learning_rate": 0.0001, "loss": 0.021, "step": 28900 }, { "epoch": 171.06508875739644, "grad_norm": 1.2840327024459839, "learning_rate": 0.0001, "loss": 0.0211, "step": 28910 }, { "epoch": 171.12426035502958, "grad_norm": 1.5011248588562012, "learning_rate": 0.0001, "loss": 0.021, "step": 28920 }, { "epoch": 171.18343195266272, "grad_norm": 1.3768430948257446, "learning_rate": 0.0001, "loss": 0.0208, "step": 28930 }, { "epoch": 171.24260355029585, "grad_norm": 1.4080450534820557, "learning_rate": 0.0001, "loss": 0.022, "step": 28940 }, { "epoch": 171.301775147929, "grad_norm": 1.7597965002059937, "learning_rate": 0.0001, "loss": 0.021, "step": 28950 }, { "epoch": 171.36094674556213, "grad_norm": 1.3593099117279053, "learning_rate": 0.0001, "loss": 0.0226, "step": 28960 }, { "epoch": 171.42011834319527, "grad_norm": 1.318086862564087, "learning_rate": 0.0001, "loss": 0.0213, "step": 28970 }, { "epoch": 171.4792899408284, "grad_norm": 1.6329866647720337, "learning_rate": 0.0001, "loss": 0.0214, "step": 28980 }, { "epoch": 171.53846153846155, "grad_norm": 1.7899540662765503, "learning_rate": 0.0001, "loss": 0.0238, "step": 28990 }, { "epoch": 171.5976331360947, "grad_norm": 1.6459089517593384, "learning_rate": 0.0001, "loss": 0.0226, "step": 29000 }, { "epoch": 171.65680473372782, "grad_norm": 1.502548336982727, "learning_rate": 0.0001, "loss": 0.0233, "step": 29010 }, { "epoch": 171.71597633136093, "grad_norm": 1.1113921403884888, "learning_rate": 0.0001, "loss": 0.0217, "step": 29020 }, { "epoch": 171.77514792899407, "grad_norm": 1.235133409500122, "learning_rate": 0.0001, "loss": 0.0223, "step": 29030 }, { "epoch": 171.8343195266272, "grad_norm": 1.4972188472747803, "learning_rate": 0.0001, "loss": 0.0207, "step": 29040 }, { "epoch": 171.89349112426035, "grad_norm": 1.2817412614822388, "learning_rate": 0.0001, "loss": 0.0214, "step": 29050 }, { "epoch": 171.9526627218935, "grad_norm": 1.1549391746520996, "learning_rate": 0.0001, "loss": 0.0209, "step": 29060 }, { "epoch": 172.01183431952663, "grad_norm": 1.6784253120422363, "learning_rate": 0.0001, "loss": 0.0224, "step": 29070 }, { "epoch": 172.07100591715977, "grad_norm": 1.6289777755737305, "learning_rate": 0.0001, "loss": 0.0221, "step": 29080 }, { "epoch": 172.1301775147929, "grad_norm": 1.7584434747695923, "learning_rate": 0.0001, "loss": 0.0215, "step": 29090 }, { "epoch": 172.18934911242604, "grad_norm": 1.5722213983535767, "learning_rate": 0.0001, "loss": 0.0209, "step": 29100 }, { "epoch": 172.24852071005918, "grad_norm": 1.3313138484954834, "learning_rate": 0.0001, "loss": 0.022, "step": 29110 }, { "epoch": 172.30769230769232, "grad_norm": 1.650519847869873, "learning_rate": 0.0001, "loss": 0.0219, "step": 29120 }, { "epoch": 172.36686390532543, "grad_norm": 1.6545190811157227, "learning_rate": 0.0001, "loss": 0.0232, "step": 29130 }, { "epoch": 172.42603550295857, "grad_norm": 1.4269858598709106, "learning_rate": 0.0001, "loss": 0.0234, "step": 29140 }, { "epoch": 172.4852071005917, "grad_norm": 2.047924757003784, "learning_rate": 0.0001, "loss": 0.0219, "step": 29150 }, { "epoch": 172.54437869822485, "grad_norm": 2.1008055210113525, "learning_rate": 0.0001, "loss": 0.0228, "step": 29160 }, { "epoch": 172.60355029585799, "grad_norm": 2.20617413520813, "learning_rate": 0.0001, "loss": 0.0221, "step": 29170 }, { "epoch": 172.66272189349112, "grad_norm": 1.7124793529510498, "learning_rate": 0.0001, "loss": 0.0219, "step": 29180 }, { "epoch": 172.72189349112426, "grad_norm": 1.304251790046692, "learning_rate": 0.0001, "loss": 0.0208, "step": 29190 }, { "epoch": 172.7810650887574, "grad_norm": 1.3082438707351685, "learning_rate": 0.0001, "loss": 0.0216, "step": 29200 }, { "epoch": 172.84023668639054, "grad_norm": 1.687892198562622, "learning_rate": 0.0001, "loss": 0.0219, "step": 29210 }, { "epoch": 172.89940828402368, "grad_norm": 1.4199796915054321, "learning_rate": 0.0001, "loss": 0.0217, "step": 29220 }, { "epoch": 172.95857988165682, "grad_norm": 1.2600358724594116, "learning_rate": 0.0001, "loss": 0.0207, "step": 29230 }, { "epoch": 173.01775147928993, "grad_norm": 1.6842923164367676, "learning_rate": 0.0001, "loss": 0.0211, "step": 29240 }, { "epoch": 173.07692307692307, "grad_norm": 1.1859503984451294, "learning_rate": 0.0001, "loss": 0.0219, "step": 29250 }, { "epoch": 173.1360946745562, "grad_norm": 1.3835941553115845, "learning_rate": 0.0001, "loss": 0.0214, "step": 29260 }, { "epoch": 173.19526627218934, "grad_norm": 1.494677186012268, "learning_rate": 0.0001, "loss": 0.0208, "step": 29270 }, { "epoch": 173.25443786982248, "grad_norm": 1.4768859148025513, "learning_rate": 0.0001, "loss": 0.022, "step": 29280 }, { "epoch": 173.31360946745562, "grad_norm": 1.3132171630859375, "learning_rate": 0.0001, "loss": 0.0212, "step": 29290 }, { "epoch": 173.37278106508876, "grad_norm": 1.5731886625289917, "learning_rate": 0.0001, "loss": 0.0213, "step": 29300 }, { "epoch": 173.4319526627219, "grad_norm": 1.412389874458313, "learning_rate": 0.0001, "loss": 0.0212, "step": 29310 }, { "epoch": 173.49112426035504, "grad_norm": 1.1720194816589355, "learning_rate": 0.0001, "loss": 0.0204, "step": 29320 }, { "epoch": 173.55029585798817, "grad_norm": 1.525128722190857, "learning_rate": 0.0001, "loss": 0.02, "step": 29330 }, { "epoch": 173.6094674556213, "grad_norm": 1.5158592462539673, "learning_rate": 0.0001, "loss": 0.0201, "step": 29340 }, { "epoch": 173.66863905325442, "grad_norm": 1.4475696086883545, "learning_rate": 0.0001, "loss": 0.0218, "step": 29350 }, { "epoch": 173.72781065088756, "grad_norm": 1.4352666139602661, "learning_rate": 0.0001, "loss": 0.0213, "step": 29360 }, { "epoch": 173.7869822485207, "grad_norm": 1.7554782629013062, "learning_rate": 0.0001, "loss": 0.0212, "step": 29370 }, { "epoch": 173.84615384615384, "grad_norm": 1.4401869773864746, "learning_rate": 0.0001, "loss": 0.0216, "step": 29380 }, { "epoch": 173.90532544378698, "grad_norm": 1.3011507987976074, "learning_rate": 0.0001, "loss": 0.0201, "step": 29390 }, { "epoch": 173.96449704142012, "grad_norm": 1.663869857788086, "learning_rate": 0.0001, "loss": 0.0201, "step": 29400 }, { "epoch": 174.02366863905326, "grad_norm": 1.4407769441604614, "learning_rate": 0.0001, "loss": 0.0207, "step": 29410 }, { "epoch": 174.0828402366864, "grad_norm": 1.5006027221679688, "learning_rate": 0.0001, "loss": 0.0208, "step": 29420 }, { "epoch": 174.14201183431953, "grad_norm": 1.8231219053268433, "learning_rate": 0.0001, "loss": 0.0203, "step": 29430 }, { "epoch": 174.20118343195267, "grad_norm": 1.6331323385238647, "learning_rate": 0.0001, "loss": 0.0207, "step": 29440 }, { "epoch": 174.2603550295858, "grad_norm": 1.5572057962417603, "learning_rate": 0.0001, "loss": 0.0205, "step": 29450 }, { "epoch": 174.31952662721895, "grad_norm": 1.7315846681594849, "learning_rate": 0.0001, "loss": 0.0211, "step": 29460 }, { "epoch": 174.37869822485206, "grad_norm": 1.883667230606079, "learning_rate": 0.0001, "loss": 0.022, "step": 29470 }, { "epoch": 174.4378698224852, "grad_norm": 0.9886133670806885, "learning_rate": 0.0001, "loss": 0.0219, "step": 29480 }, { "epoch": 174.49704142011834, "grad_norm": 0.9564580321311951, "learning_rate": 0.0001, "loss": 0.0202, "step": 29490 }, { "epoch": 174.55621301775147, "grad_norm": 1.3308346271514893, "learning_rate": 0.0001, "loss": 0.021, "step": 29500 }, { "epoch": 174.6153846153846, "grad_norm": 2.0455198287963867, "learning_rate": 0.0001, "loss": 0.0214, "step": 29510 }, { "epoch": 174.67455621301775, "grad_norm": 1.4653739929199219, "learning_rate": 0.0001, "loss": 0.022, "step": 29520 }, { "epoch": 174.7337278106509, "grad_norm": 1.5578733682632446, "learning_rate": 0.0001, "loss": 0.0207, "step": 29530 }, { "epoch": 174.79289940828403, "grad_norm": 1.4361023902893066, "learning_rate": 0.0001, "loss": 0.0222, "step": 29540 }, { "epoch": 174.85207100591717, "grad_norm": 0.9807714819908142, "learning_rate": 0.0001, "loss": 0.0206, "step": 29550 }, { "epoch": 174.9112426035503, "grad_norm": 1.484761357307434, "learning_rate": 0.0001, "loss": 0.021, "step": 29560 }, { "epoch": 174.97041420118344, "grad_norm": 1.1971189975738525, "learning_rate": 0.0001, "loss": 0.0216, "step": 29570 }, { "epoch": 175.02958579881656, "grad_norm": 1.5236179828643799, "learning_rate": 0.0001, "loss": 0.0222, "step": 29580 }, { "epoch": 175.0887573964497, "grad_norm": 1.258272409439087, "learning_rate": 0.0001, "loss": 0.0216, "step": 29590 }, { "epoch": 175.14792899408283, "grad_norm": 1.234546184539795, "learning_rate": 0.0001, "loss": 0.0207, "step": 29600 }, { "epoch": 175.20710059171597, "grad_norm": 1.5035619735717773, "learning_rate": 0.0001, "loss": 0.021, "step": 29610 }, { "epoch": 175.2662721893491, "grad_norm": 1.504460096359253, "learning_rate": 0.0001, "loss": 0.0219, "step": 29620 }, { "epoch": 175.32544378698225, "grad_norm": 1.9628832340240479, "learning_rate": 0.0001, "loss": 0.0215, "step": 29630 }, { "epoch": 175.3846153846154, "grad_norm": 1.906305193901062, "learning_rate": 0.0001, "loss": 0.0232, "step": 29640 }, { "epoch": 175.44378698224853, "grad_norm": 1.5419796705245972, "learning_rate": 0.0001, "loss": 0.0222, "step": 29650 }, { "epoch": 175.50295857988166, "grad_norm": 1.2505013942718506, "learning_rate": 0.0001, "loss": 0.0222, "step": 29660 }, { "epoch": 175.5621301775148, "grad_norm": 1.4763752222061157, "learning_rate": 0.0001, "loss": 0.0219, "step": 29670 }, { "epoch": 175.62130177514794, "grad_norm": 1.1079888343811035, "learning_rate": 0.0001, "loss": 0.0213, "step": 29680 }, { "epoch": 175.68047337278105, "grad_norm": 1.435534119606018, "learning_rate": 0.0001, "loss": 0.0205, "step": 29690 }, { "epoch": 175.7396449704142, "grad_norm": 1.4825403690338135, "learning_rate": 0.0001, "loss": 0.0209, "step": 29700 }, { "epoch": 175.79881656804733, "grad_norm": 1.743375301361084, "learning_rate": 0.0001, "loss": 0.0203, "step": 29710 }, { "epoch": 175.85798816568047, "grad_norm": 1.3275151252746582, "learning_rate": 0.0001, "loss": 0.0208, "step": 29720 }, { "epoch": 175.9171597633136, "grad_norm": 1.6027990579605103, "learning_rate": 0.0001, "loss": 0.0211, "step": 29730 }, { "epoch": 175.97633136094674, "grad_norm": 1.72512686252594, "learning_rate": 0.0001, "loss": 0.0214, "step": 29740 }, { "epoch": 176.03550295857988, "grad_norm": 1.6897835731506348, "learning_rate": 0.0001, "loss": 0.0217, "step": 29750 }, { "epoch": 176.09467455621302, "grad_norm": 1.202134132385254, "learning_rate": 0.0001, "loss": 0.0198, "step": 29760 }, { "epoch": 176.15384615384616, "grad_norm": 1.6858726739883423, "learning_rate": 0.0001, "loss": 0.0209, "step": 29770 }, { "epoch": 176.2130177514793, "grad_norm": 1.775199294090271, "learning_rate": 0.0001, "loss": 0.021, "step": 29780 }, { "epoch": 176.27218934911244, "grad_norm": 1.1621134281158447, "learning_rate": 0.0001, "loss": 0.0214, "step": 29790 }, { "epoch": 176.33136094674558, "grad_norm": 1.376298189163208, "learning_rate": 0.0001, "loss": 0.0215, "step": 29800 }, { "epoch": 176.3905325443787, "grad_norm": 1.4587548971176147, "learning_rate": 0.0001, "loss": 0.0216, "step": 29810 }, { "epoch": 176.44970414201183, "grad_norm": 1.388917326927185, "learning_rate": 0.0001, "loss": 0.0209, "step": 29820 }, { "epoch": 176.50887573964496, "grad_norm": 1.019277572631836, "learning_rate": 0.0001, "loss": 0.0216, "step": 29830 }, { "epoch": 176.5680473372781, "grad_norm": 1.7484475374221802, "learning_rate": 0.0001, "loss": 0.0218, "step": 29840 }, { "epoch": 176.62721893491124, "grad_norm": 1.3332982063293457, "learning_rate": 0.0001, "loss": 0.0203, "step": 29850 }, { "epoch": 176.68639053254438, "grad_norm": 1.4240779876708984, "learning_rate": 0.0001, "loss": 0.0203, "step": 29860 }, { "epoch": 176.74556213017752, "grad_norm": 1.4058678150177002, "learning_rate": 0.0001, "loss": 0.0209, "step": 29870 }, { "epoch": 176.80473372781066, "grad_norm": 1.5967134237289429, "learning_rate": 0.0001, "loss": 0.0201, "step": 29880 }, { "epoch": 176.8639053254438, "grad_norm": 2.2750649452209473, "learning_rate": 0.0001, "loss": 0.0204, "step": 29890 }, { "epoch": 176.92307692307693, "grad_norm": 1.3144826889038086, "learning_rate": 0.0001, "loss": 0.0212, "step": 29900 }, { "epoch": 176.98224852071007, "grad_norm": 2.062574863433838, "learning_rate": 0.0001, "loss": 0.0205, "step": 29910 }, { "epoch": 177.04142011834318, "grad_norm": 2.260805368423462, "learning_rate": 0.0001, "loss": 0.02, "step": 29920 }, { "epoch": 177.10059171597632, "grad_norm": 1.5518171787261963, "learning_rate": 0.0001, "loss": 0.0203, "step": 29930 }, { "epoch": 177.15976331360946, "grad_norm": 1.4742319583892822, "learning_rate": 0.0001, "loss": 0.0198, "step": 29940 }, { "epoch": 177.2189349112426, "grad_norm": 1.6139997243881226, "learning_rate": 0.0001, "loss": 0.0209, "step": 29950 }, { "epoch": 177.27810650887574, "grad_norm": 1.5775812864303589, "learning_rate": 0.0001, "loss": 0.0213, "step": 29960 }, { "epoch": 177.33727810650888, "grad_norm": 0.9223977327346802, "learning_rate": 0.0001, "loss": 0.0199, "step": 29970 }, { "epoch": 177.39644970414201, "grad_norm": 1.3155505657196045, "learning_rate": 0.0001, "loss": 0.0197, "step": 29980 }, { "epoch": 177.45562130177515, "grad_norm": 1.1749366521835327, "learning_rate": 0.0001, "loss": 0.0207, "step": 29990 }, { "epoch": 177.5147928994083, "grad_norm": 1.178136944770813, "learning_rate": 0.0001, "loss": 0.0207, "step": 30000 }, { "epoch": 177.5147928994083, "step": 30000, "total_flos": 0.0, "train_loss": 0.06381190015673638, "train_runtime": 8120.8163, "train_samples_per_second": 236.429, "train_steps_per_second": 3.694 } ], "logging_steps": 10, "max_steps": 30000, "num_input_tokens_seen": 0, "num_train_epochs": 178, "save_steps": 5000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 64, "trial_name": null, "trial_params": null }