{ "best_metric": null, "best_model_checkpoint": null, "epoch": 318.47133757961785, "eval_steps": 500, "global_step": 50000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.06369426751592357, "grad_norm": 156.41769409179688, "learning_rate": 0.0001, "loss": 19.9817, "step": 10 }, { "epoch": 0.12738853503184713, "grad_norm": 93.97779846191406, "learning_rate": 0.0001, "loss": 4.6049, "step": 20 }, { "epoch": 0.1910828025477707, "grad_norm": 79.27691650390625, "learning_rate": 0.0001, "loss": 3.5609, "step": 30 }, { "epoch": 0.25477707006369427, "grad_norm": 70.33567810058594, "learning_rate": 0.0001, "loss": 3.2326, "step": 40 }, { "epoch": 0.3184713375796178, "grad_norm": 55.35213088989258, "learning_rate": 0.0001, "loss": 3.0514, "step": 50 }, { "epoch": 0.3821656050955414, "grad_norm": 63.50765609741211, "learning_rate": 0.0001, "loss": 2.7915, "step": 60 }, { "epoch": 0.445859872611465, "grad_norm": 68.20964813232422, "learning_rate": 0.0001, "loss": 2.6574, "step": 70 }, { "epoch": 0.5095541401273885, "grad_norm": 59.128395080566406, "learning_rate": 0.0001, "loss": 2.4783, "step": 80 }, { "epoch": 0.5732484076433121, "grad_norm": 67.79703521728516, "learning_rate": 0.0001, "loss": 2.4025, "step": 90 }, { "epoch": 0.6369426751592356, "grad_norm": 54.69905090332031, "learning_rate": 0.0001, "loss": 2.2517, "step": 100 }, { "epoch": 0.7006369426751592, "grad_norm": 69.73320770263672, "learning_rate": 0.0001, "loss": 2.2159, "step": 110 }, { "epoch": 0.7643312101910829, "grad_norm": 67.55083465576172, "learning_rate": 0.0001, "loss": 2.154, "step": 120 }, { "epoch": 0.8280254777070064, "grad_norm": 57.41201400756836, "learning_rate": 0.0001, "loss": 2.0245, "step": 130 }, { "epoch": 0.89171974522293, "grad_norm": 60.017120361328125, "learning_rate": 0.0001, "loss": 1.9063, "step": 140 }, { "epoch": 0.9554140127388535, "grad_norm": 56.31606674194336, "learning_rate": 0.0001, "loss": 1.8264, "step": 150 }, { "epoch": 1.019108280254777, "grad_norm": 47.47957992553711, "learning_rate": 0.0001, "loss": 1.6984, "step": 160 }, { "epoch": 1.0828025477707006, "grad_norm": 37.967716217041016, "learning_rate": 0.0001, "loss": 1.6477, "step": 170 }, { "epoch": 1.1464968152866242, "grad_norm": 48.32626724243164, "learning_rate": 0.0001, "loss": 1.5929, "step": 180 }, { "epoch": 1.2101910828025477, "grad_norm": 50.79423522949219, "learning_rate": 0.0001, "loss": 1.5247, "step": 190 }, { "epoch": 1.2738853503184713, "grad_norm": 42.59910583496094, "learning_rate": 0.0001, "loss": 1.4726, "step": 200 }, { "epoch": 1.3375796178343948, "grad_norm": 46.11872863769531, "learning_rate": 0.0001, "loss": 1.3977, "step": 210 }, { "epoch": 1.4012738853503186, "grad_norm": 43.90074157714844, "learning_rate": 0.0001, "loss": 1.3533, "step": 220 }, { "epoch": 1.4649681528662422, "grad_norm": 42.401405334472656, "learning_rate": 0.0001, "loss": 1.239, "step": 230 }, { "epoch": 1.5286624203821657, "grad_norm": 43.989017486572266, "learning_rate": 0.0001, "loss": 1.1924, "step": 240 }, { "epoch": 1.5923566878980893, "grad_norm": 41.35331344604492, "learning_rate": 0.0001, "loss": 1.157, "step": 250 }, { "epoch": 1.6560509554140128, "grad_norm": 44.31100845336914, "learning_rate": 0.0001, "loss": 1.0794, "step": 260 }, { "epoch": 1.7197452229299364, "grad_norm": 39.254188537597656, "learning_rate": 0.0001, "loss": 1.0066, "step": 270 }, { "epoch": 1.78343949044586, "grad_norm": 40.580623626708984, "learning_rate": 0.0001, "loss": 0.9903, "step": 280 }, { "epoch": 1.8471337579617835, "grad_norm": 37.415767669677734, "learning_rate": 0.0001, "loss": 0.9532, "step": 290 }, { "epoch": 1.910828025477707, "grad_norm": 25.66832733154297, "learning_rate": 0.0001, "loss": 0.8962, "step": 300 }, { "epoch": 1.9745222929936306, "grad_norm": 32.47686767578125, "learning_rate": 0.0001, "loss": 0.8505, "step": 310 }, { "epoch": 2.038216560509554, "grad_norm": 41.00144577026367, "learning_rate": 0.0001, "loss": 0.7955, "step": 320 }, { "epoch": 2.1019108280254777, "grad_norm": 27.022380828857422, "learning_rate": 0.0001, "loss": 0.7689, "step": 330 }, { "epoch": 2.1656050955414012, "grad_norm": 26.71839714050293, "learning_rate": 0.0001, "loss": 0.7415, "step": 340 }, { "epoch": 2.229299363057325, "grad_norm": 26.954849243164062, "learning_rate": 0.0001, "loss": 0.7141, "step": 350 }, { "epoch": 2.2929936305732483, "grad_norm": 30.090211868286133, "learning_rate": 0.0001, "loss": 0.6841, "step": 360 }, { "epoch": 2.356687898089172, "grad_norm": 28.22332763671875, "learning_rate": 0.0001, "loss": 0.6446, "step": 370 }, { "epoch": 2.4203821656050954, "grad_norm": 26.81166648864746, "learning_rate": 0.0001, "loss": 0.6122, "step": 380 }, { "epoch": 2.484076433121019, "grad_norm": 25.748214721679688, "learning_rate": 0.0001, "loss": 0.59, "step": 390 }, { "epoch": 2.5477707006369426, "grad_norm": 25.778013229370117, "learning_rate": 0.0001, "loss": 0.5773, "step": 400 }, { "epoch": 2.611464968152866, "grad_norm": 27.965782165527344, "learning_rate": 0.0001, "loss": 0.5544, "step": 410 }, { "epoch": 2.6751592356687897, "grad_norm": 24.55584144592285, "learning_rate": 0.0001, "loss": 0.5244, "step": 420 }, { "epoch": 2.738853503184713, "grad_norm": 29.259105682373047, "learning_rate": 0.0001, "loss": 0.5109, "step": 430 }, { "epoch": 2.802547770700637, "grad_norm": 28.739282608032227, "learning_rate": 0.0001, "loss": 0.5164, "step": 440 }, { "epoch": 2.8662420382165603, "grad_norm": 31.376846313476562, "learning_rate": 0.0001, "loss": 0.4883, "step": 450 }, { "epoch": 2.9299363057324843, "grad_norm": 25.93101692199707, "learning_rate": 0.0001, "loss": 0.4605, "step": 460 }, { "epoch": 2.9936305732484074, "grad_norm": 24.53693199157715, "learning_rate": 0.0001, "loss": 0.4511, "step": 470 }, { "epoch": 3.0573248407643314, "grad_norm": 23.08462905883789, "learning_rate": 0.0001, "loss": 0.4432, "step": 480 }, { "epoch": 3.121019108280255, "grad_norm": 21.347076416015625, "learning_rate": 0.0001, "loss": 0.4441, "step": 490 }, { "epoch": 3.1847133757961785, "grad_norm": 21.252954483032227, "learning_rate": 0.0001, "loss": 0.4312, "step": 500 }, { "epoch": 3.248407643312102, "grad_norm": 23.383583068847656, "learning_rate": 0.0001, "loss": 0.4211, "step": 510 }, { "epoch": 3.3121019108280256, "grad_norm": 22.73468589782715, "learning_rate": 0.0001, "loss": 0.419, "step": 520 }, { "epoch": 3.375796178343949, "grad_norm": 19.49928092956543, "learning_rate": 0.0001, "loss": 0.4182, "step": 530 }, { "epoch": 3.4394904458598727, "grad_norm": 21.024965286254883, "learning_rate": 0.0001, "loss": 0.408, "step": 540 }, { "epoch": 3.5031847133757963, "grad_norm": 19.52722930908203, "learning_rate": 0.0001, "loss": 0.3896, "step": 550 }, { "epoch": 3.56687898089172, "grad_norm": 20.48493766784668, "learning_rate": 0.0001, "loss": 0.379, "step": 560 }, { "epoch": 3.6305732484076434, "grad_norm": 18.10401153564453, "learning_rate": 0.0001, "loss": 0.3744, "step": 570 }, { "epoch": 3.694267515923567, "grad_norm": 21.007116317749023, "learning_rate": 0.0001, "loss": 0.3695, "step": 580 }, { "epoch": 3.7579617834394905, "grad_norm": 19.996116638183594, "learning_rate": 0.0001, "loss": 0.3664, "step": 590 }, { "epoch": 3.821656050955414, "grad_norm": 17.6173038482666, "learning_rate": 0.0001, "loss": 0.3609, "step": 600 }, { "epoch": 3.8853503184713376, "grad_norm": 19.086990356445312, "learning_rate": 0.0001, "loss": 0.3559, "step": 610 }, { "epoch": 3.949044585987261, "grad_norm": 21.667236328125, "learning_rate": 0.0001, "loss": 0.3592, "step": 620 }, { "epoch": 4.012738853503185, "grad_norm": 18.8443660736084, "learning_rate": 0.0001, "loss": 0.3446, "step": 630 }, { "epoch": 4.076433121019108, "grad_norm": 19.98897361755371, "learning_rate": 0.0001, "loss": 0.3628, "step": 640 }, { "epoch": 4.140127388535032, "grad_norm": 16.95684814453125, "learning_rate": 0.0001, "loss": 0.3394, "step": 650 }, { "epoch": 4.203821656050955, "grad_norm": 17.993148803710938, "learning_rate": 0.0001, "loss": 0.333, "step": 660 }, { "epoch": 4.267515923566879, "grad_norm": 19.32396697998047, "learning_rate": 0.0001, "loss": 0.3301, "step": 670 }, { "epoch": 4.3312101910828025, "grad_norm": 16.6735782623291, "learning_rate": 0.0001, "loss": 0.323, "step": 680 }, { "epoch": 4.3949044585987265, "grad_norm": 18.729185104370117, "learning_rate": 0.0001, "loss": 0.3253, "step": 690 }, { "epoch": 4.45859872611465, "grad_norm": 16.13702964782715, "learning_rate": 0.0001, "loss": 0.323, "step": 700 }, { "epoch": 4.522292993630574, "grad_norm": 18.172847747802734, "learning_rate": 0.0001, "loss": 0.3146, "step": 710 }, { "epoch": 4.585987261146497, "grad_norm": 19.354507446289062, "learning_rate": 0.0001, "loss": 0.3032, "step": 720 }, { "epoch": 4.649681528662421, "grad_norm": 14.950016021728516, "learning_rate": 0.0001, "loss": 0.3082, "step": 730 }, { "epoch": 4.713375796178344, "grad_norm": 17.214807510375977, "learning_rate": 0.0001, "loss": 0.3001, "step": 740 }, { "epoch": 4.777070063694268, "grad_norm": 16.805862426757812, "learning_rate": 0.0001, "loss": 0.3012, "step": 750 }, { "epoch": 4.840764331210191, "grad_norm": 16.260150909423828, "learning_rate": 0.0001, "loss": 0.3063, "step": 760 }, { "epoch": 4.904458598726115, "grad_norm": 17.688682556152344, "learning_rate": 0.0001, "loss": 0.293, "step": 770 }, { "epoch": 4.968152866242038, "grad_norm": 15.148552894592285, "learning_rate": 0.0001, "loss": 0.2893, "step": 780 }, { "epoch": 5.031847133757962, "grad_norm": 16.526531219482422, "learning_rate": 0.0001, "loss": 0.2902, "step": 790 }, { "epoch": 5.095541401273885, "grad_norm": 19.193788528442383, "learning_rate": 0.0001, "loss": 0.2872, "step": 800 }, { "epoch": 5.159235668789809, "grad_norm": 17.030136108398438, "learning_rate": 0.0001, "loss": 0.28, "step": 810 }, { "epoch": 5.222929936305732, "grad_norm": 14.553903579711914, "learning_rate": 0.0001, "loss": 0.2848, "step": 820 }, { "epoch": 5.286624203821656, "grad_norm": 17.029258728027344, "learning_rate": 0.0001, "loss": 0.2918, "step": 830 }, { "epoch": 5.350318471337579, "grad_norm": 16.388532638549805, "learning_rate": 0.0001, "loss": 0.2823, "step": 840 }, { "epoch": 5.414012738853503, "grad_norm": 14.884830474853516, "learning_rate": 0.0001, "loss": 0.2696, "step": 850 }, { "epoch": 5.477707006369426, "grad_norm": 16.14713478088379, "learning_rate": 0.0001, "loss": 0.2636, "step": 860 }, { "epoch": 5.54140127388535, "grad_norm": 13.79494571685791, "learning_rate": 0.0001, "loss": 0.276, "step": 870 }, { "epoch": 5.6050955414012735, "grad_norm": 16.057527542114258, "learning_rate": 0.0001, "loss": 0.2744, "step": 880 }, { "epoch": 5.6687898089171975, "grad_norm": 13.554064750671387, "learning_rate": 0.0001, "loss": 0.281, "step": 890 }, { "epoch": 5.732484076433121, "grad_norm": 17.668272018432617, "learning_rate": 0.0001, "loss": 0.2664, "step": 900 }, { "epoch": 5.796178343949045, "grad_norm": 16.327777862548828, "learning_rate": 0.0001, "loss": 0.2633, "step": 910 }, { "epoch": 5.859872611464969, "grad_norm": 15.24561882019043, "learning_rate": 0.0001, "loss": 0.2612, "step": 920 }, { "epoch": 5.923566878980892, "grad_norm": 14.129444122314453, "learning_rate": 0.0001, "loss": 0.2612, "step": 930 }, { "epoch": 5.987261146496815, "grad_norm": 15.318094253540039, "learning_rate": 0.0001, "loss": 0.2543, "step": 940 }, { "epoch": 6.050955414012739, "grad_norm": 15.340836524963379, "learning_rate": 0.0001, "loss": 0.2536, "step": 950 }, { "epoch": 6.114649681528663, "grad_norm": 12.637797355651855, "learning_rate": 0.0001, "loss": 0.2549, "step": 960 }, { "epoch": 6.178343949044586, "grad_norm": 11.704322814941406, "learning_rate": 0.0001, "loss": 0.2529, "step": 970 }, { "epoch": 6.24203821656051, "grad_norm": 13.802431106567383, "learning_rate": 0.0001, "loss": 0.2544, "step": 980 }, { "epoch": 6.305732484076433, "grad_norm": 13.279501914978027, "learning_rate": 0.0001, "loss": 0.2535, "step": 990 }, { "epoch": 6.369426751592357, "grad_norm": 11.766366958618164, "learning_rate": 0.0001, "loss": 0.2461, "step": 1000 }, { "epoch": 6.43312101910828, "grad_norm": 13.344511985778809, "learning_rate": 0.0001, "loss": 0.2417, "step": 1010 }, { "epoch": 6.496815286624204, "grad_norm": 14.456836700439453, "learning_rate": 0.0001, "loss": 0.2423, "step": 1020 }, { "epoch": 6.560509554140127, "grad_norm": 13.779865264892578, "learning_rate": 0.0001, "loss": 0.2445, "step": 1030 }, { "epoch": 6.624203821656051, "grad_norm": 15.393293380737305, "learning_rate": 0.0001, "loss": 0.2476, "step": 1040 }, { "epoch": 6.687898089171974, "grad_norm": 11.475264549255371, "learning_rate": 0.0001, "loss": 0.2447, "step": 1050 }, { "epoch": 6.751592356687898, "grad_norm": 14.325343132019043, "learning_rate": 0.0001, "loss": 0.2493, "step": 1060 }, { "epoch": 6.8152866242038215, "grad_norm": 13.485072135925293, "learning_rate": 0.0001, "loss": 0.2323, "step": 1070 }, { "epoch": 6.8789808917197455, "grad_norm": 13.206942558288574, "learning_rate": 0.0001, "loss": 0.242, "step": 1080 }, { "epoch": 6.942675159235669, "grad_norm": 12.959823608398438, "learning_rate": 0.0001, "loss": 0.2386, "step": 1090 }, { "epoch": 7.006369426751593, "grad_norm": 14.822141647338867, "learning_rate": 0.0001, "loss": 0.2421, "step": 1100 }, { "epoch": 7.070063694267516, "grad_norm": 12.270977973937988, "learning_rate": 0.0001, "loss": 0.2286, "step": 1110 }, { "epoch": 7.13375796178344, "grad_norm": 13.021611213684082, "learning_rate": 0.0001, "loss": 0.23, "step": 1120 }, { "epoch": 7.197452229299363, "grad_norm": 13.643697738647461, "learning_rate": 0.0001, "loss": 0.2289, "step": 1130 }, { "epoch": 7.261146496815287, "grad_norm": 14.550004005432129, "learning_rate": 0.0001, "loss": 0.2308, "step": 1140 }, { "epoch": 7.32484076433121, "grad_norm": 12.778716087341309, "learning_rate": 0.0001, "loss": 0.233, "step": 1150 }, { "epoch": 7.388535031847134, "grad_norm": 11.426482200622559, "learning_rate": 0.0001, "loss": 0.2303, "step": 1160 }, { "epoch": 7.452229299363057, "grad_norm": 12.455096244812012, "learning_rate": 0.0001, "loss": 0.2355, "step": 1170 }, { "epoch": 7.515923566878981, "grad_norm": 9.93753433227539, "learning_rate": 0.0001, "loss": 0.2278, "step": 1180 }, { "epoch": 7.579617834394904, "grad_norm": 12.400490760803223, "learning_rate": 0.0001, "loss": 0.2239, "step": 1190 }, { "epoch": 7.643312101910828, "grad_norm": 13.721969604492188, "learning_rate": 0.0001, "loss": 0.2205, "step": 1200 }, { "epoch": 7.707006369426751, "grad_norm": 12.098180770874023, "learning_rate": 0.0001, "loss": 0.217, "step": 1210 }, { "epoch": 7.770700636942675, "grad_norm": 11.372154235839844, "learning_rate": 0.0001, "loss": 0.2185, "step": 1220 }, { "epoch": 7.834394904458598, "grad_norm": 10.8790864944458, "learning_rate": 0.0001, "loss": 0.209, "step": 1230 }, { "epoch": 7.898089171974522, "grad_norm": 13.411617279052734, "learning_rate": 0.0001, "loss": 0.2192, "step": 1240 }, { "epoch": 7.961783439490446, "grad_norm": 10.100236892700195, "learning_rate": 0.0001, "loss": 0.2099, "step": 1250 }, { "epoch": 8.02547770700637, "grad_norm": 11.320174217224121, "learning_rate": 0.0001, "loss": 0.2094, "step": 1260 }, { "epoch": 8.089171974522293, "grad_norm": 11.76405143737793, "learning_rate": 0.0001, "loss": 0.21, "step": 1270 }, { "epoch": 8.152866242038217, "grad_norm": 8.77917766571045, "learning_rate": 0.0001, "loss": 0.2134, "step": 1280 }, { "epoch": 8.21656050955414, "grad_norm": 11.561644554138184, "learning_rate": 0.0001, "loss": 0.2099, "step": 1290 }, { "epoch": 8.280254777070065, "grad_norm": 11.617578506469727, "learning_rate": 0.0001, "loss": 0.2162, "step": 1300 }, { "epoch": 8.343949044585987, "grad_norm": 10.836274147033691, "learning_rate": 0.0001, "loss": 0.2159, "step": 1310 }, { "epoch": 8.40764331210191, "grad_norm": 9.55105209350586, "learning_rate": 0.0001, "loss": 0.2072, "step": 1320 }, { "epoch": 8.471337579617835, "grad_norm": 11.458130836486816, "learning_rate": 0.0001, "loss": 0.207, "step": 1330 }, { "epoch": 8.535031847133759, "grad_norm": 10.765179634094238, "learning_rate": 0.0001, "loss": 0.2078, "step": 1340 }, { "epoch": 8.598726114649681, "grad_norm": 11.6493558883667, "learning_rate": 0.0001, "loss": 0.2042, "step": 1350 }, { "epoch": 8.662420382165605, "grad_norm": 11.217811584472656, "learning_rate": 0.0001, "loss": 0.2112, "step": 1360 }, { "epoch": 8.726114649681529, "grad_norm": 11.198540687561035, "learning_rate": 0.0001, "loss": 0.2032, "step": 1370 }, { "epoch": 8.789808917197453, "grad_norm": 11.205888748168945, "learning_rate": 0.0001, "loss": 0.2051, "step": 1380 }, { "epoch": 8.853503184713375, "grad_norm": 10.559950828552246, "learning_rate": 0.0001, "loss": 0.2035, "step": 1390 }, { "epoch": 8.9171974522293, "grad_norm": 12.299173355102539, "learning_rate": 0.0001, "loss": 0.1982, "step": 1400 }, { "epoch": 8.980891719745223, "grad_norm": 10.343477249145508, "learning_rate": 0.0001, "loss": 0.2019, "step": 1410 }, { "epoch": 9.044585987261147, "grad_norm": 10.841248512268066, "learning_rate": 0.0001, "loss": 0.2072, "step": 1420 }, { "epoch": 9.10828025477707, "grad_norm": 11.12691593170166, "learning_rate": 0.0001, "loss": 0.1911, "step": 1430 }, { "epoch": 9.171974522292993, "grad_norm": 11.083256721496582, "learning_rate": 0.0001, "loss": 0.1957, "step": 1440 }, { "epoch": 9.235668789808917, "grad_norm": 11.296979904174805, "learning_rate": 0.0001, "loss": 0.1921, "step": 1450 }, { "epoch": 9.299363057324841, "grad_norm": 11.228266716003418, "learning_rate": 0.0001, "loss": 0.1964, "step": 1460 }, { "epoch": 9.363057324840764, "grad_norm": 11.303832054138184, "learning_rate": 0.0001, "loss": 0.1999, "step": 1470 }, { "epoch": 9.426751592356688, "grad_norm": 9.267276763916016, "learning_rate": 0.0001, "loss": 0.196, "step": 1480 }, { "epoch": 9.490445859872612, "grad_norm": 9.71078109741211, "learning_rate": 0.0001, "loss": 0.1957, "step": 1490 }, { "epoch": 9.554140127388536, "grad_norm": 10.620139122009277, "learning_rate": 0.0001, "loss": 0.1968, "step": 1500 }, { "epoch": 9.617834394904458, "grad_norm": 9.609233856201172, "learning_rate": 0.0001, "loss": 0.201, "step": 1510 }, { "epoch": 9.681528662420382, "grad_norm": 9.643800735473633, "learning_rate": 0.0001, "loss": 0.1936, "step": 1520 }, { "epoch": 9.745222929936306, "grad_norm": 11.300803184509277, "learning_rate": 0.0001, "loss": 0.1846, "step": 1530 }, { "epoch": 9.80891719745223, "grad_norm": 10.362200736999512, "learning_rate": 0.0001, "loss": 0.1918, "step": 1540 }, { "epoch": 9.872611464968152, "grad_norm": 10.284977912902832, "learning_rate": 0.0001, "loss": 0.192, "step": 1550 }, { "epoch": 9.936305732484076, "grad_norm": 10.4548978805542, "learning_rate": 0.0001, "loss": 0.2004, "step": 1560 }, { "epoch": 10.0, "grad_norm": 10.07293701171875, "learning_rate": 0.0001, "loss": 0.1887, "step": 1570 }, { "epoch": 10.063694267515924, "grad_norm": 11.317354202270508, "learning_rate": 0.0001, "loss": 0.1834, "step": 1580 }, { "epoch": 10.127388535031848, "grad_norm": 9.264471054077148, "learning_rate": 0.0001, "loss": 0.1784, "step": 1590 }, { "epoch": 10.19108280254777, "grad_norm": 9.09054183959961, "learning_rate": 0.0001, "loss": 0.1797, "step": 1600 }, { "epoch": 10.254777070063694, "grad_norm": 11.144927024841309, "learning_rate": 0.0001, "loss": 0.1837, "step": 1610 }, { "epoch": 10.318471337579618, "grad_norm": 10.504740715026855, "learning_rate": 0.0001, "loss": 0.1795, "step": 1620 }, { "epoch": 10.382165605095542, "grad_norm": 11.077340126037598, "learning_rate": 0.0001, "loss": 0.1802, "step": 1630 }, { "epoch": 10.445859872611464, "grad_norm": 9.78762435913086, "learning_rate": 0.0001, "loss": 0.1791, "step": 1640 }, { "epoch": 10.509554140127388, "grad_norm": 11.205950736999512, "learning_rate": 0.0001, "loss": 0.1806, "step": 1650 }, { "epoch": 10.573248407643312, "grad_norm": 8.84304428100586, "learning_rate": 0.0001, "loss": 0.1744, "step": 1660 }, { "epoch": 10.636942675159236, "grad_norm": 8.417778015136719, "learning_rate": 0.0001, "loss": 0.1842, "step": 1670 }, { "epoch": 10.700636942675159, "grad_norm": 10.047447204589844, "learning_rate": 0.0001, "loss": 0.1769, "step": 1680 }, { "epoch": 10.764331210191083, "grad_norm": 10.44043254852295, "learning_rate": 0.0001, "loss": 0.1846, "step": 1690 }, { "epoch": 10.828025477707007, "grad_norm": 11.355624198913574, "learning_rate": 0.0001, "loss": 0.1866, "step": 1700 }, { "epoch": 10.89171974522293, "grad_norm": 10.384223937988281, "learning_rate": 0.0001, "loss": 0.1823, "step": 1710 }, { "epoch": 10.955414012738853, "grad_norm": 9.754363059997559, "learning_rate": 0.0001, "loss": 0.1768, "step": 1720 }, { "epoch": 11.019108280254777, "grad_norm": 9.550606727600098, "learning_rate": 0.0001, "loss": 0.1837, "step": 1730 }, { "epoch": 11.0828025477707, "grad_norm": 9.771001815795898, "learning_rate": 0.0001, "loss": 0.1743, "step": 1740 }, { "epoch": 11.146496815286625, "grad_norm": 10.218421936035156, "learning_rate": 0.0001, "loss": 0.1696, "step": 1750 }, { "epoch": 11.210191082802547, "grad_norm": 9.99843692779541, "learning_rate": 0.0001, "loss": 0.1705, "step": 1760 }, { "epoch": 11.273885350318471, "grad_norm": 8.858160018920898, "learning_rate": 0.0001, "loss": 0.1745, "step": 1770 }, { "epoch": 11.337579617834395, "grad_norm": 10.064541816711426, "learning_rate": 0.0001, "loss": 0.1698, "step": 1780 }, { "epoch": 11.401273885350319, "grad_norm": 8.834334373474121, "learning_rate": 0.0001, "loss": 0.1806, "step": 1790 }, { "epoch": 11.464968152866241, "grad_norm": 9.378928184509277, "learning_rate": 0.0001, "loss": 0.1703, "step": 1800 }, { "epoch": 11.528662420382165, "grad_norm": 8.881120681762695, "learning_rate": 0.0001, "loss": 0.1702, "step": 1810 }, { "epoch": 11.59235668789809, "grad_norm": 9.405679702758789, "learning_rate": 0.0001, "loss": 0.1701, "step": 1820 }, { "epoch": 11.656050955414013, "grad_norm": 9.466071128845215, "learning_rate": 0.0001, "loss": 0.1674, "step": 1830 }, { "epoch": 11.719745222929935, "grad_norm": 8.758752822875977, "learning_rate": 0.0001, "loss": 0.1676, "step": 1840 }, { "epoch": 11.78343949044586, "grad_norm": 7.843833923339844, "learning_rate": 0.0001, "loss": 0.1665, "step": 1850 }, { "epoch": 11.847133757961783, "grad_norm": 10.438551902770996, "learning_rate": 0.0001, "loss": 0.1624, "step": 1860 }, { "epoch": 11.910828025477707, "grad_norm": 8.850358009338379, "learning_rate": 0.0001, "loss": 0.1675, "step": 1870 }, { "epoch": 11.97452229299363, "grad_norm": 7.863918304443359, "learning_rate": 0.0001, "loss": 0.1636, "step": 1880 }, { "epoch": 12.038216560509554, "grad_norm": 10.414758682250977, "learning_rate": 0.0001, "loss": 0.1662, "step": 1890 }, { "epoch": 12.101910828025478, "grad_norm": 8.102476119995117, "learning_rate": 0.0001, "loss": 0.1609, "step": 1900 }, { "epoch": 12.165605095541402, "grad_norm": 9.28372859954834, "learning_rate": 0.0001, "loss": 0.166, "step": 1910 }, { "epoch": 12.229299363057326, "grad_norm": 8.731134414672852, "learning_rate": 0.0001, "loss": 0.1614, "step": 1920 }, { "epoch": 12.292993630573248, "grad_norm": 8.715885162353516, "learning_rate": 0.0001, "loss": 0.1578, "step": 1930 }, { "epoch": 12.356687898089172, "grad_norm": 8.768304824829102, "learning_rate": 0.0001, "loss": 0.1583, "step": 1940 }, { "epoch": 12.420382165605096, "grad_norm": 9.938737869262695, "learning_rate": 0.0001, "loss": 0.159, "step": 1950 }, { "epoch": 12.48407643312102, "grad_norm": 9.116240501403809, "learning_rate": 0.0001, "loss": 0.1583, "step": 1960 }, { "epoch": 12.547770700636942, "grad_norm": 9.079319953918457, "learning_rate": 0.0001, "loss": 0.1583, "step": 1970 }, { "epoch": 12.611464968152866, "grad_norm": 7.534732818603516, "learning_rate": 0.0001, "loss": 0.1559, "step": 1980 }, { "epoch": 12.67515923566879, "grad_norm": 9.383201599121094, "learning_rate": 0.0001, "loss": 0.1663, "step": 1990 }, { "epoch": 12.738853503184714, "grad_norm": 8.013839721679688, "learning_rate": 0.0001, "loss": 0.158, "step": 2000 }, { "epoch": 12.802547770700636, "grad_norm": 7.988734245300293, "learning_rate": 0.0001, "loss": 0.1563, "step": 2010 }, { "epoch": 12.86624203821656, "grad_norm": 8.618616104125977, "learning_rate": 0.0001, "loss": 0.1557, "step": 2020 }, { "epoch": 12.929936305732484, "grad_norm": 8.028732299804688, "learning_rate": 0.0001, "loss": 0.1549, "step": 2030 }, { "epoch": 12.993630573248408, "grad_norm": 8.4745512008667, "learning_rate": 0.0001, "loss": 0.1545, "step": 2040 }, { "epoch": 13.05732484076433, "grad_norm": 8.45059871673584, "learning_rate": 0.0001, "loss": 0.1558, "step": 2050 }, { "epoch": 13.121019108280255, "grad_norm": 7.333597183227539, "learning_rate": 0.0001, "loss": 0.1534, "step": 2060 }, { "epoch": 13.184713375796179, "grad_norm": 7.4386444091796875, "learning_rate": 0.0001, "loss": 0.1529, "step": 2070 }, { "epoch": 13.248407643312103, "grad_norm": 9.791462898254395, "learning_rate": 0.0001, "loss": 0.1568, "step": 2080 }, { "epoch": 13.312101910828025, "grad_norm": 7.422467231750488, "learning_rate": 0.0001, "loss": 0.1482, "step": 2090 }, { "epoch": 13.375796178343949, "grad_norm": 10.29655933380127, "learning_rate": 0.0001, "loss": 0.1559, "step": 2100 }, { "epoch": 13.439490445859873, "grad_norm": 8.220909118652344, "learning_rate": 0.0001, "loss": 0.1532, "step": 2110 }, { "epoch": 13.503184713375797, "grad_norm": 8.924830436706543, "learning_rate": 0.0001, "loss": 0.1531, "step": 2120 }, { "epoch": 13.566878980891719, "grad_norm": 6.793705940246582, "learning_rate": 0.0001, "loss": 0.1488, "step": 2130 }, { "epoch": 13.630573248407643, "grad_norm": 7.66600227355957, "learning_rate": 0.0001, "loss": 0.1456, "step": 2140 }, { "epoch": 13.694267515923567, "grad_norm": 7.503973484039307, "learning_rate": 0.0001, "loss": 0.1549, "step": 2150 }, { "epoch": 13.757961783439491, "grad_norm": 8.10653305053711, "learning_rate": 0.0001, "loss": 0.1529, "step": 2160 }, { "epoch": 13.821656050955415, "grad_norm": 8.253316879272461, "learning_rate": 0.0001, "loss": 0.1461, "step": 2170 }, { "epoch": 13.885350318471337, "grad_norm": 7.344550132751465, "learning_rate": 0.0001, "loss": 0.1428, "step": 2180 }, { "epoch": 13.949044585987261, "grad_norm": 9.408671379089355, "learning_rate": 0.0001, "loss": 0.1498, "step": 2190 }, { "epoch": 14.012738853503185, "grad_norm": 8.770389556884766, "learning_rate": 0.0001, "loss": 0.1499, "step": 2200 }, { "epoch": 14.07643312101911, "grad_norm": 8.123120307922363, "learning_rate": 0.0001, "loss": 0.143, "step": 2210 }, { "epoch": 14.140127388535031, "grad_norm": 8.865860939025879, "learning_rate": 0.0001, "loss": 0.1488, "step": 2220 }, { "epoch": 14.203821656050955, "grad_norm": 6.812440872192383, "learning_rate": 0.0001, "loss": 0.1458, "step": 2230 }, { "epoch": 14.26751592356688, "grad_norm": 7.896829128265381, "learning_rate": 0.0001, "loss": 0.1417, "step": 2240 }, { "epoch": 14.331210191082803, "grad_norm": 9.157877922058105, "learning_rate": 0.0001, "loss": 0.1492, "step": 2250 }, { "epoch": 14.394904458598726, "grad_norm": 8.309847831726074, "learning_rate": 0.0001, "loss": 0.1517, "step": 2260 }, { "epoch": 14.45859872611465, "grad_norm": 7.3151021003723145, "learning_rate": 0.0001, "loss": 0.1455, "step": 2270 }, { "epoch": 14.522292993630574, "grad_norm": 8.25631046295166, "learning_rate": 0.0001, "loss": 0.1407, "step": 2280 }, { "epoch": 14.585987261146498, "grad_norm": 8.295973777770996, "learning_rate": 0.0001, "loss": 0.1423, "step": 2290 }, { "epoch": 14.64968152866242, "grad_norm": 9.054570198059082, "learning_rate": 0.0001, "loss": 0.1391, "step": 2300 }, { "epoch": 14.713375796178344, "grad_norm": 6.856845378875732, "learning_rate": 0.0001, "loss": 0.1414, "step": 2310 }, { "epoch": 14.777070063694268, "grad_norm": 7.702650547027588, "learning_rate": 0.0001, "loss": 0.1444, "step": 2320 }, { "epoch": 14.840764331210192, "grad_norm": 9.196455955505371, "learning_rate": 0.0001, "loss": 0.1385, "step": 2330 }, { "epoch": 14.904458598726114, "grad_norm": 8.173233985900879, "learning_rate": 0.0001, "loss": 0.1381, "step": 2340 }, { "epoch": 14.968152866242038, "grad_norm": 8.01815128326416, "learning_rate": 0.0001, "loss": 0.1414, "step": 2350 }, { "epoch": 15.031847133757962, "grad_norm": 8.095742225646973, "learning_rate": 0.0001, "loss": 0.145, "step": 2360 }, { "epoch": 15.095541401273886, "grad_norm": 7.782421588897705, "learning_rate": 0.0001, "loss": 0.1395, "step": 2370 }, { "epoch": 15.159235668789808, "grad_norm": 7.080327987670898, "learning_rate": 0.0001, "loss": 0.1362, "step": 2380 }, { "epoch": 15.222929936305732, "grad_norm": 7.946308135986328, "learning_rate": 0.0001, "loss": 0.1386, "step": 2390 }, { "epoch": 15.286624203821656, "grad_norm": 6.563339710235596, "learning_rate": 0.0001, "loss": 0.1414, "step": 2400 }, { "epoch": 15.35031847133758, "grad_norm": 7.805520534515381, "learning_rate": 0.0001, "loss": 0.1362, "step": 2410 }, { "epoch": 15.414012738853502, "grad_norm": 8.321504592895508, "learning_rate": 0.0001, "loss": 0.1371, "step": 2420 }, { "epoch": 15.477707006369426, "grad_norm": 7.60122537612915, "learning_rate": 0.0001, "loss": 0.1454, "step": 2430 }, { "epoch": 15.54140127388535, "grad_norm": 6.910974979400635, "learning_rate": 0.0001, "loss": 0.1406, "step": 2440 }, { "epoch": 15.605095541401274, "grad_norm": 6.244896411895752, "learning_rate": 0.0001, "loss": 0.1325, "step": 2450 }, { "epoch": 15.668789808917197, "grad_norm": 7.1541948318481445, "learning_rate": 0.0001, "loss": 0.1339, "step": 2460 }, { "epoch": 15.73248407643312, "grad_norm": 6.748579502105713, "learning_rate": 0.0001, "loss": 0.1377, "step": 2470 }, { "epoch": 15.796178343949045, "grad_norm": 6.987049579620361, "learning_rate": 0.0001, "loss": 0.1274, "step": 2480 }, { "epoch": 15.859872611464969, "grad_norm": 5.92326545715332, "learning_rate": 0.0001, "loss": 0.1343, "step": 2490 }, { "epoch": 15.923566878980893, "grad_norm": 6.045137405395508, "learning_rate": 0.0001, "loss": 0.1343, "step": 2500 }, { "epoch": 15.987261146496815, "grad_norm": 5.825116157531738, "learning_rate": 0.0001, "loss": 0.1287, "step": 2510 }, { "epoch": 16.05095541401274, "grad_norm": 6.215178966522217, "learning_rate": 0.0001, "loss": 0.1272, "step": 2520 }, { "epoch": 16.11464968152866, "grad_norm": 7.099797248840332, "learning_rate": 0.0001, "loss": 0.1364, "step": 2530 }, { "epoch": 16.178343949044585, "grad_norm": 8.595040321350098, "learning_rate": 0.0001, "loss": 0.1336, "step": 2540 }, { "epoch": 16.24203821656051, "grad_norm": 7.993870735168457, "learning_rate": 0.0001, "loss": 0.1352, "step": 2550 }, { "epoch": 16.305732484076433, "grad_norm": 7.297451019287109, "learning_rate": 0.0001, "loss": 0.1335, "step": 2560 }, { "epoch": 16.369426751592357, "grad_norm": 6.814788341522217, "learning_rate": 0.0001, "loss": 0.1291, "step": 2570 }, { "epoch": 16.43312101910828, "grad_norm": 6.196445465087891, "learning_rate": 0.0001, "loss": 0.1283, "step": 2580 }, { "epoch": 16.496815286624205, "grad_norm": 6.871626853942871, "learning_rate": 0.0001, "loss": 0.1295, "step": 2590 }, { "epoch": 16.56050955414013, "grad_norm": 8.1677885055542, "learning_rate": 0.0001, "loss": 0.135, "step": 2600 }, { "epoch": 16.62420382165605, "grad_norm": 7.744696617126465, "learning_rate": 0.0001, "loss": 0.1351, "step": 2610 }, { "epoch": 16.687898089171973, "grad_norm": 6.90109920501709, "learning_rate": 0.0001, "loss": 0.122, "step": 2620 }, { "epoch": 16.751592356687897, "grad_norm": 6.0098466873168945, "learning_rate": 0.0001, "loss": 0.1257, "step": 2630 }, { "epoch": 16.81528662420382, "grad_norm": 7.601247787475586, "learning_rate": 0.0001, "loss": 0.127, "step": 2640 }, { "epoch": 16.878980891719745, "grad_norm": 7.116043567657471, "learning_rate": 0.0001, "loss": 0.13, "step": 2650 }, { "epoch": 16.94267515923567, "grad_norm": 6.464692115783691, "learning_rate": 0.0001, "loss": 0.1226, "step": 2660 }, { "epoch": 17.006369426751593, "grad_norm": 5.750668525695801, "learning_rate": 0.0001, "loss": 0.1307, "step": 2670 }, { "epoch": 17.070063694267517, "grad_norm": 6.166470527648926, "learning_rate": 0.0001, "loss": 0.1245, "step": 2680 }, { "epoch": 17.133757961783438, "grad_norm": 7.889608860015869, "learning_rate": 0.0001, "loss": 0.1263, "step": 2690 }, { "epoch": 17.197452229299362, "grad_norm": 6.692622661590576, "learning_rate": 0.0001, "loss": 0.1176, "step": 2700 }, { "epoch": 17.261146496815286, "grad_norm": 6.167306423187256, "learning_rate": 0.0001, "loss": 0.1216, "step": 2710 }, { "epoch": 17.32484076433121, "grad_norm": 6.063767910003662, "learning_rate": 0.0001, "loss": 0.1225, "step": 2720 }, { "epoch": 17.388535031847134, "grad_norm": 6.923361778259277, "learning_rate": 0.0001, "loss": 0.1302, "step": 2730 }, { "epoch": 17.452229299363058, "grad_norm": 6.453667163848877, "learning_rate": 0.0001, "loss": 0.1225, "step": 2740 }, { "epoch": 17.515923566878982, "grad_norm": 6.606473922729492, "learning_rate": 0.0001, "loss": 0.1258, "step": 2750 }, { "epoch": 17.579617834394906, "grad_norm": 7.704937934875488, "learning_rate": 0.0001, "loss": 0.1223, "step": 2760 }, { "epoch": 17.643312101910826, "grad_norm": 6.063183307647705, "learning_rate": 0.0001, "loss": 0.1205, "step": 2770 }, { "epoch": 17.70700636942675, "grad_norm": 6.60798978805542, "learning_rate": 0.0001, "loss": 0.1194, "step": 2780 }, { "epoch": 17.770700636942674, "grad_norm": 7.070493698120117, "learning_rate": 0.0001, "loss": 0.1242, "step": 2790 }, { "epoch": 17.8343949044586, "grad_norm": 6.28232479095459, "learning_rate": 0.0001, "loss": 0.1181, "step": 2800 }, { "epoch": 17.898089171974522, "grad_norm": 7.309407711029053, "learning_rate": 0.0001, "loss": 0.1268, "step": 2810 }, { "epoch": 17.961783439490446, "grad_norm": 6.165432929992676, "learning_rate": 0.0001, "loss": 0.1221, "step": 2820 }, { "epoch": 18.02547770700637, "grad_norm": 5.240340709686279, "learning_rate": 0.0001, "loss": 0.1219, "step": 2830 }, { "epoch": 18.089171974522294, "grad_norm": 6.407044887542725, "learning_rate": 0.0001, "loss": 0.1225, "step": 2840 }, { "epoch": 18.15286624203822, "grad_norm": 6.789233207702637, "learning_rate": 0.0001, "loss": 0.1278, "step": 2850 }, { "epoch": 18.21656050955414, "grad_norm": 7.687410354614258, "learning_rate": 0.0001, "loss": 0.1284, "step": 2860 }, { "epoch": 18.280254777070063, "grad_norm": 5.883219242095947, "learning_rate": 0.0001, "loss": 0.1204, "step": 2870 }, { "epoch": 18.343949044585987, "grad_norm": 5.403461456298828, "learning_rate": 0.0001, "loss": 0.1194, "step": 2880 }, { "epoch": 18.40764331210191, "grad_norm": 6.297480583190918, "learning_rate": 0.0001, "loss": 0.1183, "step": 2890 }, { "epoch": 18.471337579617835, "grad_norm": 6.643802165985107, "learning_rate": 0.0001, "loss": 0.1142, "step": 2900 }, { "epoch": 18.53503184713376, "grad_norm": 7.298845291137695, "learning_rate": 0.0001, "loss": 0.1159, "step": 2910 }, { "epoch": 18.598726114649683, "grad_norm": 6.438895225524902, "learning_rate": 0.0001, "loss": 0.1146, "step": 2920 }, { "epoch": 18.662420382165607, "grad_norm": 6.8430585861206055, "learning_rate": 0.0001, "loss": 0.12, "step": 2930 }, { "epoch": 18.726114649681527, "grad_norm": 6.297487735748291, "learning_rate": 0.0001, "loss": 0.1197, "step": 2940 }, { "epoch": 18.78980891719745, "grad_norm": 5.837064743041992, "learning_rate": 0.0001, "loss": 0.1235, "step": 2950 }, { "epoch": 18.853503184713375, "grad_norm": 5.90976619720459, "learning_rate": 0.0001, "loss": 0.1204, "step": 2960 }, { "epoch": 18.9171974522293, "grad_norm": 5.699254035949707, "learning_rate": 0.0001, "loss": 0.1119, "step": 2970 }, { "epoch": 18.980891719745223, "grad_norm": 6.868646144866943, "learning_rate": 0.0001, "loss": 0.1189, "step": 2980 }, { "epoch": 19.044585987261147, "grad_norm": 6.049735069274902, "learning_rate": 0.0001, "loss": 0.1178, "step": 2990 }, { "epoch": 19.10828025477707, "grad_norm": 5.602506637573242, "learning_rate": 0.0001, "loss": 0.1202, "step": 3000 }, { "epoch": 19.171974522292995, "grad_norm": 6.0019402503967285, "learning_rate": 0.0001, "loss": 0.1158, "step": 3010 }, { "epoch": 19.235668789808916, "grad_norm": 6.5507988929748535, "learning_rate": 0.0001, "loss": 0.118, "step": 3020 }, { "epoch": 19.29936305732484, "grad_norm": 7.107872009277344, "learning_rate": 0.0001, "loss": 0.116, "step": 3030 }, { "epoch": 19.363057324840764, "grad_norm": 6.069121837615967, "learning_rate": 0.0001, "loss": 0.1141, "step": 3040 }, { "epoch": 19.426751592356688, "grad_norm": 5.808029651641846, "learning_rate": 0.0001, "loss": 0.1138, "step": 3050 }, { "epoch": 19.49044585987261, "grad_norm": 5.21278190612793, "learning_rate": 0.0001, "loss": 0.1161, "step": 3060 }, { "epoch": 19.554140127388536, "grad_norm": 5.619228363037109, "learning_rate": 0.0001, "loss": 0.1138, "step": 3070 }, { "epoch": 19.61783439490446, "grad_norm": 5.233692169189453, "learning_rate": 0.0001, "loss": 0.1099, "step": 3080 }, { "epoch": 19.681528662420384, "grad_norm": 6.3453593254089355, "learning_rate": 0.0001, "loss": 0.1103, "step": 3090 }, { "epoch": 19.745222929936304, "grad_norm": 5.9204182624816895, "learning_rate": 0.0001, "loss": 0.1107, "step": 3100 }, { "epoch": 19.808917197452228, "grad_norm": 5.884125232696533, "learning_rate": 0.0001, "loss": 0.1145, "step": 3110 }, { "epoch": 19.872611464968152, "grad_norm": 5.938080310821533, "learning_rate": 0.0001, "loss": 0.1106, "step": 3120 }, { "epoch": 19.936305732484076, "grad_norm": 6.09297513961792, "learning_rate": 0.0001, "loss": 0.113, "step": 3130 }, { "epoch": 20.0, "grad_norm": 5.56687593460083, "learning_rate": 0.0001, "loss": 0.1122, "step": 3140 }, { "epoch": 20.063694267515924, "grad_norm": 6.022899150848389, "learning_rate": 0.0001, "loss": 0.1051, "step": 3150 }, { "epoch": 20.127388535031848, "grad_norm": 5.765618324279785, "learning_rate": 0.0001, "loss": 0.1101, "step": 3160 }, { "epoch": 20.191082802547772, "grad_norm": 5.6903228759765625, "learning_rate": 0.0001, "loss": 0.1129, "step": 3170 }, { "epoch": 20.254777070063696, "grad_norm": 5.77426290512085, "learning_rate": 0.0001, "loss": 0.1091, "step": 3180 }, { "epoch": 20.318471337579616, "grad_norm": 5.717085361480713, "learning_rate": 0.0001, "loss": 0.1071, "step": 3190 }, { "epoch": 20.38216560509554, "grad_norm": 6.123028755187988, "learning_rate": 0.0001, "loss": 0.1098, "step": 3200 }, { "epoch": 20.445859872611464, "grad_norm": 5.226891994476318, "learning_rate": 0.0001, "loss": 0.1097, "step": 3210 }, { "epoch": 20.50955414012739, "grad_norm": 5.224104404449463, "learning_rate": 0.0001, "loss": 0.1111, "step": 3220 }, { "epoch": 20.573248407643312, "grad_norm": 5.149974346160889, "learning_rate": 0.0001, "loss": 0.1058, "step": 3230 }, { "epoch": 20.636942675159236, "grad_norm": 6.077393054962158, "learning_rate": 0.0001, "loss": 0.1108, "step": 3240 }, { "epoch": 20.70063694267516, "grad_norm": 5.508329391479492, "learning_rate": 0.0001, "loss": 0.1095, "step": 3250 }, { "epoch": 20.764331210191084, "grad_norm": 4.65595006942749, "learning_rate": 0.0001, "loss": 0.1091, "step": 3260 }, { "epoch": 20.828025477707005, "grad_norm": 5.213708877563477, "learning_rate": 0.0001, "loss": 0.1066, "step": 3270 }, { "epoch": 20.89171974522293, "grad_norm": 5.390801906585693, "learning_rate": 0.0001, "loss": 0.1062, "step": 3280 }, { "epoch": 20.955414012738853, "grad_norm": 5.097716331481934, "learning_rate": 0.0001, "loss": 0.1087, "step": 3290 }, { "epoch": 21.019108280254777, "grad_norm": 5.796699523925781, "learning_rate": 0.0001, "loss": 0.1093, "step": 3300 }, { "epoch": 21.0828025477707, "grad_norm": 5.684518337249756, "learning_rate": 0.0001, "loss": 0.1093, "step": 3310 }, { "epoch": 21.146496815286625, "grad_norm": 5.469202041625977, "learning_rate": 0.0001, "loss": 0.107, "step": 3320 }, { "epoch": 21.21019108280255, "grad_norm": 5.4239726066589355, "learning_rate": 0.0001, "loss": 0.107, "step": 3330 }, { "epoch": 21.273885350318473, "grad_norm": 5.164736747741699, "learning_rate": 0.0001, "loss": 0.1073, "step": 3340 }, { "epoch": 21.337579617834393, "grad_norm": 5.382824897766113, "learning_rate": 0.0001, "loss": 0.1066, "step": 3350 }, { "epoch": 21.401273885350317, "grad_norm": 6.500673294067383, "learning_rate": 0.0001, "loss": 0.1052, "step": 3360 }, { "epoch": 21.46496815286624, "grad_norm": 6.12367057800293, "learning_rate": 0.0001, "loss": 0.1096, "step": 3370 }, { "epoch": 21.528662420382165, "grad_norm": 5.152108669281006, "learning_rate": 0.0001, "loss": 0.104, "step": 3380 }, { "epoch": 21.59235668789809, "grad_norm": 5.412603378295898, "learning_rate": 0.0001, "loss": 0.1022, "step": 3390 }, { "epoch": 21.656050955414013, "grad_norm": 5.057712078094482, "learning_rate": 0.0001, "loss": 0.1, "step": 3400 }, { "epoch": 21.719745222929937, "grad_norm": 5.858835697174072, "learning_rate": 0.0001, "loss": 0.1061, "step": 3410 }, { "epoch": 21.78343949044586, "grad_norm": 5.921517848968506, "learning_rate": 0.0001, "loss": 0.107, "step": 3420 }, { "epoch": 21.847133757961785, "grad_norm": 5.253808498382568, "learning_rate": 0.0001, "loss": 0.104, "step": 3430 }, { "epoch": 21.910828025477706, "grad_norm": 5.5571675300598145, "learning_rate": 0.0001, "loss": 0.1048, "step": 3440 }, { "epoch": 21.97452229299363, "grad_norm": 5.432802200317383, "learning_rate": 0.0001, "loss": 0.1037, "step": 3450 }, { "epoch": 22.038216560509554, "grad_norm": 4.58548641204834, "learning_rate": 0.0001, "loss": 0.1063, "step": 3460 }, { "epoch": 22.101910828025478, "grad_norm": 4.701617240905762, "learning_rate": 0.0001, "loss": 0.1039, "step": 3470 }, { "epoch": 22.1656050955414, "grad_norm": 5.901213645935059, "learning_rate": 0.0001, "loss": 0.1038, "step": 3480 }, { "epoch": 22.229299363057326, "grad_norm": 5.419118881225586, "learning_rate": 0.0001, "loss": 0.1038, "step": 3490 }, { "epoch": 22.29299363057325, "grad_norm": 5.72907018661499, "learning_rate": 0.0001, "loss": 0.1009, "step": 3500 }, { "epoch": 22.356687898089174, "grad_norm": 5.664834022521973, "learning_rate": 0.0001, "loss": 0.1093, "step": 3510 }, { "epoch": 22.420382165605094, "grad_norm": 5.241158962249756, "learning_rate": 0.0001, "loss": 0.1067, "step": 3520 }, { "epoch": 22.484076433121018, "grad_norm": 5.4284749031066895, "learning_rate": 0.0001, "loss": 0.1032, "step": 3530 }, { "epoch": 22.547770700636942, "grad_norm": 5.034504413604736, "learning_rate": 0.0001, "loss": 0.1031, "step": 3540 }, { "epoch": 22.611464968152866, "grad_norm": 4.7359209060668945, "learning_rate": 0.0001, "loss": 0.1035, "step": 3550 }, { "epoch": 22.67515923566879, "grad_norm": 5.288575649261475, "learning_rate": 0.0001, "loss": 0.0993, "step": 3560 }, { "epoch": 22.738853503184714, "grad_norm": 3.775801658630371, "learning_rate": 0.0001, "loss": 0.1004, "step": 3570 }, { "epoch": 22.802547770700638, "grad_norm": 5.596522331237793, "learning_rate": 0.0001, "loss": 0.0962, "step": 3580 }, { "epoch": 22.866242038216562, "grad_norm": 4.869710445404053, "learning_rate": 0.0001, "loss": 0.1022, "step": 3590 }, { "epoch": 22.929936305732483, "grad_norm": 5.0813751220703125, "learning_rate": 0.0001, "loss": 0.1003, "step": 3600 }, { "epoch": 22.993630573248407, "grad_norm": 5.06503963470459, "learning_rate": 0.0001, "loss": 0.1008, "step": 3610 }, { "epoch": 23.05732484076433, "grad_norm": 5.3498406410217285, "learning_rate": 0.0001, "loss": 0.0986, "step": 3620 }, { "epoch": 23.121019108280255, "grad_norm": 4.930723667144775, "learning_rate": 0.0001, "loss": 0.1065, "step": 3630 }, { "epoch": 23.18471337579618, "grad_norm": 4.646623611450195, "learning_rate": 0.0001, "loss": 0.0972, "step": 3640 }, { "epoch": 23.248407643312103, "grad_norm": 5.322900295257568, "learning_rate": 0.0001, "loss": 0.0978, "step": 3650 }, { "epoch": 23.312101910828027, "grad_norm": 4.55164098739624, "learning_rate": 0.0001, "loss": 0.0959, "step": 3660 }, { "epoch": 23.37579617834395, "grad_norm": 4.5190815925598145, "learning_rate": 0.0001, "loss": 0.099, "step": 3670 }, { "epoch": 23.43949044585987, "grad_norm": 5.236412048339844, "learning_rate": 0.0001, "loss": 0.1, "step": 3680 }, { "epoch": 23.503184713375795, "grad_norm": 5.238063335418701, "learning_rate": 0.0001, "loss": 0.1015, "step": 3690 }, { "epoch": 23.56687898089172, "grad_norm": 6.107975959777832, "learning_rate": 0.0001, "loss": 0.0997, "step": 3700 }, { "epoch": 23.630573248407643, "grad_norm": 4.7621002197265625, "learning_rate": 0.0001, "loss": 0.0963, "step": 3710 }, { "epoch": 23.694267515923567, "grad_norm": 5.488010883331299, "learning_rate": 0.0001, "loss": 0.0988, "step": 3720 }, { "epoch": 23.75796178343949, "grad_norm": 4.810127258300781, "learning_rate": 0.0001, "loss": 0.0942, "step": 3730 }, { "epoch": 23.821656050955415, "grad_norm": 4.569536209106445, "learning_rate": 0.0001, "loss": 0.0966, "step": 3740 }, { "epoch": 23.88535031847134, "grad_norm": 4.511500835418701, "learning_rate": 0.0001, "loss": 0.0976, "step": 3750 }, { "epoch": 23.94904458598726, "grad_norm": 5.017733097076416, "learning_rate": 0.0001, "loss": 0.0978, "step": 3760 }, { "epoch": 24.012738853503183, "grad_norm": 4.793808460235596, "learning_rate": 0.0001, "loss": 0.0968, "step": 3770 }, { "epoch": 24.076433121019107, "grad_norm": 5.298835277557373, "learning_rate": 0.0001, "loss": 0.0934, "step": 3780 }, { "epoch": 24.14012738853503, "grad_norm": 4.2714385986328125, "learning_rate": 0.0001, "loss": 0.0982, "step": 3790 }, { "epoch": 24.203821656050955, "grad_norm": 4.507734298706055, "learning_rate": 0.0001, "loss": 0.0957, "step": 3800 }, { "epoch": 24.26751592356688, "grad_norm": 5.709644794464111, "learning_rate": 0.0001, "loss": 0.0981, "step": 3810 }, { "epoch": 24.331210191082803, "grad_norm": 4.628353595733643, "learning_rate": 0.0001, "loss": 0.0976, "step": 3820 }, { "epoch": 24.394904458598727, "grad_norm": 4.926641941070557, "learning_rate": 0.0001, "loss": 0.0957, "step": 3830 }, { "epoch": 24.45859872611465, "grad_norm": 4.965304851531982, "learning_rate": 0.0001, "loss": 0.0949, "step": 3840 }, { "epoch": 24.522292993630572, "grad_norm": 4.885852336883545, "learning_rate": 0.0001, "loss": 0.0989, "step": 3850 }, { "epoch": 24.585987261146496, "grad_norm": 4.273346900939941, "learning_rate": 0.0001, "loss": 0.0932, "step": 3860 }, { "epoch": 24.64968152866242, "grad_norm": 5.059970378875732, "learning_rate": 0.0001, "loss": 0.1034, "step": 3870 }, { "epoch": 24.713375796178344, "grad_norm": 4.052926540374756, "learning_rate": 0.0001, "loss": 0.0987, "step": 3880 }, { "epoch": 24.777070063694268, "grad_norm": 4.6308417320251465, "learning_rate": 0.0001, "loss": 0.0975, "step": 3890 }, { "epoch": 24.840764331210192, "grad_norm": 3.9879705905914307, "learning_rate": 0.0001, "loss": 0.1003, "step": 3900 }, { "epoch": 24.904458598726116, "grad_norm": 4.621618270874023, "learning_rate": 0.0001, "loss": 0.093, "step": 3910 }, { "epoch": 24.96815286624204, "grad_norm": 4.531450271606445, "learning_rate": 0.0001, "loss": 0.0938, "step": 3920 }, { "epoch": 25.03184713375796, "grad_norm": 3.4791336059570312, "learning_rate": 0.0001, "loss": 0.1002, "step": 3930 }, { "epoch": 25.095541401273884, "grad_norm": 5.072641372680664, "learning_rate": 0.0001, "loss": 0.0958, "step": 3940 }, { "epoch": 25.159235668789808, "grad_norm": 4.610130310058594, "learning_rate": 0.0001, "loss": 0.0951, "step": 3950 }, { "epoch": 25.222929936305732, "grad_norm": 5.349038600921631, "learning_rate": 0.0001, "loss": 0.0945, "step": 3960 }, { "epoch": 25.286624203821656, "grad_norm": 4.983922958374023, "learning_rate": 0.0001, "loss": 0.0986, "step": 3970 }, { "epoch": 25.35031847133758, "grad_norm": 5.032786846160889, "learning_rate": 0.0001, "loss": 0.0958, "step": 3980 }, { "epoch": 25.414012738853504, "grad_norm": 5.449321746826172, "learning_rate": 0.0001, "loss": 0.1016, "step": 3990 }, { "epoch": 25.477707006369428, "grad_norm": 4.11007022857666, "learning_rate": 0.0001, "loss": 0.0932, "step": 4000 }, { "epoch": 25.54140127388535, "grad_norm": 4.6925048828125, "learning_rate": 0.0001, "loss": 0.0938, "step": 4010 }, { "epoch": 25.605095541401273, "grad_norm": 4.059266090393066, "learning_rate": 0.0001, "loss": 0.0895, "step": 4020 }, { "epoch": 25.668789808917197, "grad_norm": 4.113833427429199, "learning_rate": 0.0001, "loss": 0.0864, "step": 4030 }, { "epoch": 25.73248407643312, "grad_norm": 4.368225574493408, "learning_rate": 0.0001, "loss": 0.0915, "step": 4040 }, { "epoch": 25.796178343949045, "grad_norm": 4.320700168609619, "learning_rate": 0.0001, "loss": 0.0914, "step": 4050 }, { "epoch": 25.85987261146497, "grad_norm": 4.060349464416504, "learning_rate": 0.0001, "loss": 0.0922, "step": 4060 }, { "epoch": 25.923566878980893, "grad_norm": 4.847778797149658, "learning_rate": 0.0001, "loss": 0.091, "step": 4070 }, { "epoch": 25.987261146496817, "grad_norm": 4.089770793914795, "learning_rate": 0.0001, "loss": 0.0933, "step": 4080 }, { "epoch": 26.05095541401274, "grad_norm": 3.8009731769561768, "learning_rate": 0.0001, "loss": 0.0895, "step": 4090 }, { "epoch": 26.11464968152866, "grad_norm": 4.191619396209717, "learning_rate": 0.0001, "loss": 0.0901, "step": 4100 }, { "epoch": 26.178343949044585, "grad_norm": 4.088994026184082, "learning_rate": 0.0001, "loss": 0.0883, "step": 4110 }, { "epoch": 26.24203821656051, "grad_norm": 5.114682674407959, "learning_rate": 0.0001, "loss": 0.0866, "step": 4120 }, { "epoch": 26.305732484076433, "grad_norm": 4.130494594573975, "learning_rate": 0.0001, "loss": 0.0896, "step": 4130 }, { "epoch": 26.369426751592357, "grad_norm": 4.842663764953613, "learning_rate": 0.0001, "loss": 0.0909, "step": 4140 }, { "epoch": 26.43312101910828, "grad_norm": 4.133167743682861, "learning_rate": 0.0001, "loss": 0.0891, "step": 4150 }, { "epoch": 26.496815286624205, "grad_norm": 4.38192892074585, "learning_rate": 0.0001, "loss": 0.09, "step": 4160 }, { "epoch": 26.56050955414013, "grad_norm": 4.354903697967529, "learning_rate": 0.0001, "loss": 0.0928, "step": 4170 }, { "epoch": 26.62420382165605, "grad_norm": 4.129147529602051, "learning_rate": 0.0001, "loss": 0.0878, "step": 4180 }, { "epoch": 26.687898089171973, "grad_norm": 4.406203269958496, "learning_rate": 0.0001, "loss": 0.0947, "step": 4190 }, { "epoch": 26.751592356687897, "grad_norm": 4.069082260131836, "learning_rate": 0.0001, "loss": 0.0893, "step": 4200 }, { "epoch": 26.81528662420382, "grad_norm": 3.8925418853759766, "learning_rate": 0.0001, "loss": 0.0884, "step": 4210 }, { "epoch": 26.878980891719745, "grad_norm": 4.458364009857178, "learning_rate": 0.0001, "loss": 0.0915, "step": 4220 }, { "epoch": 26.94267515923567, "grad_norm": 4.492340087890625, "learning_rate": 0.0001, "loss": 0.0906, "step": 4230 }, { "epoch": 27.006369426751593, "grad_norm": 4.659182071685791, "learning_rate": 0.0001, "loss": 0.0866, "step": 4240 }, { "epoch": 27.070063694267517, "grad_norm": 3.7178611755371094, "learning_rate": 0.0001, "loss": 0.0935, "step": 4250 }, { "epoch": 27.133757961783438, "grad_norm": 4.201383113861084, "learning_rate": 0.0001, "loss": 0.0887, "step": 4260 }, { "epoch": 27.197452229299362, "grad_norm": 4.269254207611084, "learning_rate": 0.0001, "loss": 0.0908, "step": 4270 }, { "epoch": 27.261146496815286, "grad_norm": 3.330920934677124, "learning_rate": 0.0001, "loss": 0.0918, "step": 4280 }, { "epoch": 27.32484076433121, "grad_norm": 4.162090301513672, "learning_rate": 0.0001, "loss": 0.095, "step": 4290 }, { "epoch": 27.388535031847134, "grad_norm": 3.851008653640747, "learning_rate": 0.0001, "loss": 0.0841, "step": 4300 }, { "epoch": 27.452229299363058, "grad_norm": 3.4643893241882324, "learning_rate": 0.0001, "loss": 0.088, "step": 4310 }, { "epoch": 27.515923566878982, "grad_norm": 3.7494192123413086, "learning_rate": 0.0001, "loss": 0.086, "step": 4320 }, { "epoch": 27.579617834394906, "grad_norm": 3.286935806274414, "learning_rate": 0.0001, "loss": 0.0851, "step": 4330 }, { "epoch": 27.643312101910826, "grad_norm": 3.8144218921661377, "learning_rate": 0.0001, "loss": 0.0881, "step": 4340 }, { "epoch": 27.70700636942675, "grad_norm": 3.9793155193328857, "learning_rate": 0.0001, "loss": 0.0881, "step": 4350 }, { "epoch": 27.770700636942674, "grad_norm": 3.8037986755371094, "learning_rate": 0.0001, "loss": 0.093, "step": 4360 }, { "epoch": 27.8343949044586, "grad_norm": 4.432806491851807, "learning_rate": 0.0001, "loss": 0.0855, "step": 4370 }, { "epoch": 27.898089171974522, "grad_norm": 4.082523822784424, "learning_rate": 0.0001, "loss": 0.0907, "step": 4380 }, { "epoch": 27.961783439490446, "grad_norm": 3.8679046630859375, "learning_rate": 0.0001, "loss": 0.0846, "step": 4390 }, { "epoch": 28.02547770700637, "grad_norm": 4.145493984222412, "learning_rate": 0.0001, "loss": 0.0875, "step": 4400 }, { "epoch": 28.089171974522294, "grad_norm": 4.879427909851074, "learning_rate": 0.0001, "loss": 0.0867, "step": 4410 }, { "epoch": 28.15286624203822, "grad_norm": 4.397915363311768, "learning_rate": 0.0001, "loss": 0.0841, "step": 4420 }, { "epoch": 28.21656050955414, "grad_norm": 3.7157669067382812, "learning_rate": 0.0001, "loss": 0.0879, "step": 4430 }, { "epoch": 28.280254777070063, "grad_norm": 3.545579671859741, "learning_rate": 0.0001, "loss": 0.087, "step": 4440 }, { "epoch": 28.343949044585987, "grad_norm": 3.612354040145874, "learning_rate": 0.0001, "loss": 0.0903, "step": 4450 }, { "epoch": 28.40764331210191, "grad_norm": 3.6179628372192383, "learning_rate": 0.0001, "loss": 0.0838, "step": 4460 }, { "epoch": 28.471337579617835, "grad_norm": 3.685458183288574, "learning_rate": 0.0001, "loss": 0.0861, "step": 4470 }, { "epoch": 28.53503184713376, "grad_norm": 4.3621087074279785, "learning_rate": 0.0001, "loss": 0.0885, "step": 4480 }, { "epoch": 28.598726114649683, "grad_norm": 4.860956192016602, "learning_rate": 0.0001, "loss": 0.0853, "step": 4490 }, { "epoch": 28.662420382165607, "grad_norm": 4.349576473236084, "learning_rate": 0.0001, "loss": 0.0847, "step": 4500 }, { "epoch": 28.726114649681527, "grad_norm": 3.570477247238159, "learning_rate": 0.0001, "loss": 0.0857, "step": 4510 }, { "epoch": 28.78980891719745, "grad_norm": 3.5203237533569336, "learning_rate": 0.0001, "loss": 0.0828, "step": 4520 }, { "epoch": 28.853503184713375, "grad_norm": 4.0334858894348145, "learning_rate": 0.0001, "loss": 0.0836, "step": 4530 }, { "epoch": 28.9171974522293, "grad_norm": 3.5357394218444824, "learning_rate": 0.0001, "loss": 0.0843, "step": 4540 }, { "epoch": 28.980891719745223, "grad_norm": 3.633251190185547, "learning_rate": 0.0001, "loss": 0.085, "step": 4550 }, { "epoch": 29.044585987261147, "grad_norm": 3.3884119987487793, "learning_rate": 0.0001, "loss": 0.085, "step": 4560 }, { "epoch": 29.10828025477707, "grad_norm": 3.70686411857605, "learning_rate": 0.0001, "loss": 0.084, "step": 4570 }, { "epoch": 29.171974522292995, "grad_norm": 3.5423591136932373, "learning_rate": 0.0001, "loss": 0.0867, "step": 4580 }, { "epoch": 29.235668789808916, "grad_norm": 3.8151609897613525, "learning_rate": 0.0001, "loss": 0.0793, "step": 4590 }, { "epoch": 29.29936305732484, "grad_norm": 3.586094617843628, "learning_rate": 0.0001, "loss": 0.0851, "step": 4600 }, { "epoch": 29.363057324840764, "grad_norm": 4.5711188316345215, "learning_rate": 0.0001, "loss": 0.0871, "step": 4610 }, { "epoch": 29.426751592356688, "grad_norm": 3.9298312664031982, "learning_rate": 0.0001, "loss": 0.0813, "step": 4620 }, { "epoch": 29.49044585987261, "grad_norm": 3.6020238399505615, "learning_rate": 0.0001, "loss": 0.0855, "step": 4630 }, { "epoch": 29.554140127388536, "grad_norm": 3.6229474544525146, "learning_rate": 0.0001, "loss": 0.0832, "step": 4640 }, { "epoch": 29.61783439490446, "grad_norm": 3.5059056282043457, "learning_rate": 0.0001, "loss": 0.086, "step": 4650 }, { "epoch": 29.681528662420384, "grad_norm": 3.4405934810638428, "learning_rate": 0.0001, "loss": 0.091, "step": 4660 }, { "epoch": 29.745222929936304, "grad_norm": 3.3424925804138184, "learning_rate": 0.0001, "loss": 0.0868, "step": 4670 }, { "epoch": 29.808917197452228, "grad_norm": 3.2606148719787598, "learning_rate": 0.0001, "loss": 0.0884, "step": 4680 }, { "epoch": 29.872611464968152, "grad_norm": 3.559434175491333, "learning_rate": 0.0001, "loss": 0.0882, "step": 4690 }, { "epoch": 29.936305732484076, "grad_norm": 3.5333163738250732, "learning_rate": 0.0001, "loss": 0.0846, "step": 4700 }, { "epoch": 30.0, "grad_norm": 3.953310251235962, "learning_rate": 0.0001, "loss": 0.0855, "step": 4710 }, { "epoch": 30.063694267515924, "grad_norm": 3.706470251083374, "learning_rate": 0.0001, "loss": 0.0825, "step": 4720 }, { "epoch": 30.127388535031848, "grad_norm": 4.40505838394165, "learning_rate": 0.0001, "loss": 0.0847, "step": 4730 }, { "epoch": 30.191082802547772, "grad_norm": 4.208911895751953, "learning_rate": 0.0001, "loss": 0.0856, "step": 4740 }, { "epoch": 30.254777070063696, "grad_norm": 3.361865520477295, "learning_rate": 0.0001, "loss": 0.083, "step": 4750 }, { "epoch": 30.318471337579616, "grad_norm": 3.417301654815674, "learning_rate": 0.0001, "loss": 0.0851, "step": 4760 }, { "epoch": 30.38216560509554, "grad_norm": 2.9796133041381836, "learning_rate": 0.0001, "loss": 0.0821, "step": 4770 }, { "epoch": 30.445859872611464, "grad_norm": 3.4666595458984375, "learning_rate": 0.0001, "loss": 0.0799, "step": 4780 }, { "epoch": 30.50955414012739, "grad_norm": 3.5004804134368896, "learning_rate": 0.0001, "loss": 0.0824, "step": 4790 }, { "epoch": 30.573248407643312, "grad_norm": 4.036218166351318, "learning_rate": 0.0001, "loss": 0.084, "step": 4800 }, { "epoch": 30.636942675159236, "grad_norm": 3.642204999923706, "learning_rate": 0.0001, "loss": 0.0858, "step": 4810 }, { "epoch": 30.70063694267516, "grad_norm": 3.8037166595458984, "learning_rate": 0.0001, "loss": 0.085, "step": 4820 }, { "epoch": 30.764331210191084, "grad_norm": 2.996929883956909, "learning_rate": 0.0001, "loss": 0.0854, "step": 4830 }, { "epoch": 30.828025477707005, "grad_norm": 4.026613235473633, "learning_rate": 0.0001, "loss": 0.0824, "step": 4840 }, { "epoch": 30.89171974522293, "grad_norm": 4.073195457458496, "learning_rate": 0.0001, "loss": 0.082, "step": 4850 }, { "epoch": 30.955414012738853, "grad_norm": 3.9058356285095215, "learning_rate": 0.0001, "loss": 0.0783, "step": 4860 }, { "epoch": 31.019108280254777, "grad_norm": 3.9580917358398438, "learning_rate": 0.0001, "loss": 0.0812, "step": 4870 }, { "epoch": 31.0828025477707, "grad_norm": 3.4045586585998535, "learning_rate": 0.0001, "loss": 0.0767, "step": 4880 }, { "epoch": 31.146496815286625, "grad_norm": 3.567366123199463, "learning_rate": 0.0001, "loss": 0.0799, "step": 4890 }, { "epoch": 31.21019108280255, "grad_norm": 3.685333013534546, "learning_rate": 0.0001, "loss": 0.0789, "step": 4900 }, { "epoch": 31.273885350318473, "grad_norm": 4.082586765289307, "learning_rate": 0.0001, "loss": 0.0778, "step": 4910 }, { "epoch": 31.337579617834393, "grad_norm": 4.009594440460205, "learning_rate": 0.0001, "loss": 0.0797, "step": 4920 }, { "epoch": 31.401273885350317, "grad_norm": 3.7715065479278564, "learning_rate": 0.0001, "loss": 0.0813, "step": 4930 }, { "epoch": 31.46496815286624, "grad_norm": 3.737119197845459, "learning_rate": 0.0001, "loss": 0.0781, "step": 4940 }, { "epoch": 31.528662420382165, "grad_norm": 3.606182336807251, "learning_rate": 0.0001, "loss": 0.0786, "step": 4950 }, { "epoch": 31.59235668789809, "grad_norm": 3.1105425357818604, "learning_rate": 0.0001, "loss": 0.0816, "step": 4960 }, { "epoch": 31.656050955414013, "grad_norm": 3.5713303089141846, "learning_rate": 0.0001, "loss": 0.0783, "step": 4970 }, { "epoch": 31.719745222929937, "grad_norm": 3.5171139240264893, "learning_rate": 0.0001, "loss": 0.0778, "step": 4980 }, { "epoch": 31.78343949044586, "grad_norm": 3.7038919925689697, "learning_rate": 0.0001, "loss": 0.0799, "step": 4990 }, { "epoch": 31.847133757961785, "grad_norm": 4.336887359619141, "learning_rate": 0.0001, "loss": 0.082, "step": 5000 }, { "epoch": 31.910828025477706, "grad_norm": 3.610889196395874, "learning_rate": 0.0001, "loss": 0.0812, "step": 5010 }, { "epoch": 31.97452229299363, "grad_norm": 3.4893581867218018, "learning_rate": 0.0001, "loss": 0.0781, "step": 5020 }, { "epoch": 32.038216560509554, "grad_norm": 3.284306526184082, "learning_rate": 0.0001, "loss": 0.0754, "step": 5030 }, { "epoch": 32.10191082802548, "grad_norm": 3.205719470977783, "learning_rate": 0.0001, "loss": 0.0779, "step": 5040 }, { "epoch": 32.1656050955414, "grad_norm": 3.2632720470428467, "learning_rate": 0.0001, "loss": 0.0818, "step": 5050 }, { "epoch": 32.22929936305732, "grad_norm": 3.739823341369629, "learning_rate": 0.0001, "loss": 0.0756, "step": 5060 }, { "epoch": 32.29299363057325, "grad_norm": 3.3696680068969727, "learning_rate": 0.0001, "loss": 0.0819, "step": 5070 }, { "epoch": 32.35668789808917, "grad_norm": 2.646751642227173, "learning_rate": 0.0001, "loss": 0.0801, "step": 5080 }, { "epoch": 32.4203821656051, "grad_norm": 3.4773032665252686, "learning_rate": 0.0001, "loss": 0.0834, "step": 5090 }, { "epoch": 32.48407643312102, "grad_norm": 3.502204418182373, "learning_rate": 0.0001, "loss": 0.0843, "step": 5100 }, { "epoch": 32.547770700636946, "grad_norm": 3.4109928607940674, "learning_rate": 0.0001, "loss": 0.0824, "step": 5110 }, { "epoch": 32.611464968152866, "grad_norm": 3.613319158554077, "learning_rate": 0.0001, "loss": 0.0824, "step": 5120 }, { "epoch": 32.67515923566879, "grad_norm": 3.348965644836426, "learning_rate": 0.0001, "loss": 0.0793, "step": 5130 }, { "epoch": 32.738853503184714, "grad_norm": 3.403973340988159, "learning_rate": 0.0001, "loss": 0.0789, "step": 5140 }, { "epoch": 32.802547770700635, "grad_norm": 3.8807437419891357, "learning_rate": 0.0001, "loss": 0.0812, "step": 5150 }, { "epoch": 32.86624203821656, "grad_norm": 3.311023712158203, "learning_rate": 0.0001, "loss": 0.0828, "step": 5160 }, { "epoch": 32.92993630573248, "grad_norm": 2.768989086151123, "learning_rate": 0.0001, "loss": 0.0779, "step": 5170 }, { "epoch": 32.99363057324841, "grad_norm": 3.506288528442383, "learning_rate": 0.0001, "loss": 0.0773, "step": 5180 }, { "epoch": 33.05732484076433, "grad_norm": 2.9270668029785156, "learning_rate": 0.0001, "loss": 0.0739, "step": 5190 }, { "epoch": 33.12101910828026, "grad_norm": 3.3132762908935547, "learning_rate": 0.0001, "loss": 0.079, "step": 5200 }, { "epoch": 33.18471337579618, "grad_norm": 3.998896598815918, "learning_rate": 0.0001, "loss": 0.0802, "step": 5210 }, { "epoch": 33.2484076433121, "grad_norm": 4.167168617248535, "learning_rate": 0.0001, "loss": 0.076, "step": 5220 }, { "epoch": 33.31210191082803, "grad_norm": 3.723484992980957, "learning_rate": 0.0001, "loss": 0.074, "step": 5230 }, { "epoch": 33.37579617834395, "grad_norm": 3.183976650238037, "learning_rate": 0.0001, "loss": 0.0782, "step": 5240 }, { "epoch": 33.439490445859875, "grad_norm": 2.9413907527923584, "learning_rate": 0.0001, "loss": 0.0761, "step": 5250 }, { "epoch": 33.503184713375795, "grad_norm": 3.593597173690796, "learning_rate": 0.0001, "loss": 0.0729, "step": 5260 }, { "epoch": 33.56687898089172, "grad_norm": 3.4241836071014404, "learning_rate": 0.0001, "loss": 0.0772, "step": 5270 }, { "epoch": 33.63057324840764, "grad_norm": 3.6275641918182373, "learning_rate": 0.0001, "loss": 0.0722, "step": 5280 }, { "epoch": 33.69426751592356, "grad_norm": 3.564967632293701, "learning_rate": 0.0001, "loss": 0.0782, "step": 5290 }, { "epoch": 33.75796178343949, "grad_norm": 3.1977665424346924, "learning_rate": 0.0001, "loss": 0.0771, "step": 5300 }, { "epoch": 33.82165605095541, "grad_norm": 3.3054115772247314, "learning_rate": 0.0001, "loss": 0.0737, "step": 5310 }, { "epoch": 33.88535031847134, "grad_norm": 3.0096094608306885, "learning_rate": 0.0001, "loss": 0.0763, "step": 5320 }, { "epoch": 33.94904458598726, "grad_norm": 3.115355968475342, "learning_rate": 0.0001, "loss": 0.0735, "step": 5330 }, { "epoch": 34.01273885350319, "grad_norm": 3.0741329193115234, "learning_rate": 0.0001, "loss": 0.074, "step": 5340 }, { "epoch": 34.07643312101911, "grad_norm": 3.145265579223633, "learning_rate": 0.0001, "loss": 0.0727, "step": 5350 }, { "epoch": 34.140127388535035, "grad_norm": 3.229097843170166, "learning_rate": 0.0001, "loss": 0.0777, "step": 5360 }, { "epoch": 34.203821656050955, "grad_norm": 3.2248287200927734, "learning_rate": 0.0001, "loss": 0.0769, "step": 5370 }, { "epoch": 34.267515923566876, "grad_norm": 3.3750481605529785, "learning_rate": 0.0001, "loss": 0.0791, "step": 5380 }, { "epoch": 34.3312101910828, "grad_norm": 3.409576416015625, "learning_rate": 0.0001, "loss": 0.0747, "step": 5390 }, { "epoch": 34.394904458598724, "grad_norm": 3.6772301197052, "learning_rate": 0.0001, "loss": 0.0762, "step": 5400 }, { "epoch": 34.45859872611465, "grad_norm": 3.6506707668304443, "learning_rate": 0.0001, "loss": 0.077, "step": 5410 }, { "epoch": 34.52229299363057, "grad_norm": 3.5129778385162354, "learning_rate": 0.0001, "loss": 0.0774, "step": 5420 }, { "epoch": 34.5859872611465, "grad_norm": 3.360593318939209, "learning_rate": 0.0001, "loss": 0.0803, "step": 5430 }, { "epoch": 34.64968152866242, "grad_norm": 3.662980794906616, "learning_rate": 0.0001, "loss": 0.0742, "step": 5440 }, { "epoch": 34.71337579617835, "grad_norm": 3.4425623416900635, "learning_rate": 0.0001, "loss": 0.0716, "step": 5450 }, { "epoch": 34.77707006369427, "grad_norm": 3.1326253414154053, "learning_rate": 0.0001, "loss": 0.0748, "step": 5460 }, { "epoch": 34.84076433121019, "grad_norm": 3.493950843811035, "learning_rate": 0.0001, "loss": 0.0717, "step": 5470 }, { "epoch": 34.904458598726116, "grad_norm": 3.327911376953125, "learning_rate": 0.0001, "loss": 0.0763, "step": 5480 }, { "epoch": 34.968152866242036, "grad_norm": 3.264493703842163, "learning_rate": 0.0001, "loss": 0.0707, "step": 5490 }, { "epoch": 35.031847133757964, "grad_norm": 2.9397904872894287, "learning_rate": 0.0001, "loss": 0.0714, "step": 5500 }, { "epoch": 35.095541401273884, "grad_norm": 3.2123944759368896, "learning_rate": 0.0001, "loss": 0.0738, "step": 5510 }, { "epoch": 35.15923566878981, "grad_norm": 3.376662492752075, "learning_rate": 0.0001, "loss": 0.0759, "step": 5520 }, { "epoch": 35.22292993630573, "grad_norm": 3.7555837631225586, "learning_rate": 0.0001, "loss": 0.0715, "step": 5530 }, { "epoch": 35.28662420382165, "grad_norm": 3.0577166080474854, "learning_rate": 0.0001, "loss": 0.0708, "step": 5540 }, { "epoch": 35.35031847133758, "grad_norm": 3.9232118129730225, "learning_rate": 0.0001, "loss": 0.0776, "step": 5550 }, { "epoch": 35.4140127388535, "grad_norm": 2.956372022628784, "learning_rate": 0.0001, "loss": 0.0777, "step": 5560 }, { "epoch": 35.47770700636943, "grad_norm": 3.59306263923645, "learning_rate": 0.0001, "loss": 0.0743, "step": 5570 }, { "epoch": 35.54140127388535, "grad_norm": 3.162421703338623, "learning_rate": 0.0001, "loss": 0.0751, "step": 5580 }, { "epoch": 35.605095541401276, "grad_norm": 3.34190034866333, "learning_rate": 0.0001, "loss": 0.0769, "step": 5590 }, { "epoch": 35.6687898089172, "grad_norm": 3.203977108001709, "learning_rate": 0.0001, "loss": 0.0711, "step": 5600 }, { "epoch": 35.732484076433124, "grad_norm": 3.1731083393096924, "learning_rate": 0.0001, "loss": 0.0787, "step": 5610 }, { "epoch": 35.796178343949045, "grad_norm": 3.63537859916687, "learning_rate": 0.0001, "loss": 0.0733, "step": 5620 }, { "epoch": 35.859872611464965, "grad_norm": 3.125520944595337, "learning_rate": 0.0001, "loss": 0.0752, "step": 5630 }, { "epoch": 35.92356687898089, "grad_norm": 2.987882614135742, "learning_rate": 0.0001, "loss": 0.0723, "step": 5640 }, { "epoch": 35.98726114649681, "grad_norm": 3.0677311420440674, "learning_rate": 0.0001, "loss": 0.0721, "step": 5650 }, { "epoch": 36.05095541401274, "grad_norm": 3.22471022605896, "learning_rate": 0.0001, "loss": 0.0711, "step": 5660 }, { "epoch": 36.11464968152866, "grad_norm": 3.541494846343994, "learning_rate": 0.0001, "loss": 0.0782, "step": 5670 }, { "epoch": 36.17834394904459, "grad_norm": 2.916961908340454, "learning_rate": 0.0001, "loss": 0.071, "step": 5680 }, { "epoch": 36.24203821656051, "grad_norm": 3.2141411304473877, "learning_rate": 0.0001, "loss": 0.0741, "step": 5690 }, { "epoch": 36.30573248407644, "grad_norm": 2.792628526687622, "learning_rate": 0.0001, "loss": 0.0741, "step": 5700 }, { "epoch": 36.36942675159236, "grad_norm": 2.848093032836914, "learning_rate": 0.0001, "loss": 0.0776, "step": 5710 }, { "epoch": 36.43312101910828, "grad_norm": 3.1893539428710938, "learning_rate": 0.0001, "loss": 0.0779, "step": 5720 }, { "epoch": 36.496815286624205, "grad_norm": 2.8400566577911377, "learning_rate": 0.0001, "loss": 0.0758, "step": 5730 }, { "epoch": 36.560509554140125, "grad_norm": 3.383619785308838, "learning_rate": 0.0001, "loss": 0.0799, "step": 5740 }, { "epoch": 36.62420382165605, "grad_norm": 3.049561023712158, "learning_rate": 0.0001, "loss": 0.0807, "step": 5750 }, { "epoch": 36.68789808917197, "grad_norm": 2.8253884315490723, "learning_rate": 0.0001, "loss": 0.077, "step": 5760 }, { "epoch": 36.7515923566879, "grad_norm": 3.033073663711548, "learning_rate": 0.0001, "loss": 0.0728, "step": 5770 }, { "epoch": 36.81528662420382, "grad_norm": 2.827857494354248, "learning_rate": 0.0001, "loss": 0.0778, "step": 5780 }, { "epoch": 36.87898089171974, "grad_norm": 3.066443681716919, "learning_rate": 0.0001, "loss": 0.0757, "step": 5790 }, { "epoch": 36.94267515923567, "grad_norm": 2.5931429862976074, "learning_rate": 0.0001, "loss": 0.0738, "step": 5800 }, { "epoch": 37.00636942675159, "grad_norm": 3.0347537994384766, "learning_rate": 0.0001, "loss": 0.0739, "step": 5810 }, { "epoch": 37.07006369426752, "grad_norm": 2.777911424636841, "learning_rate": 0.0001, "loss": 0.0748, "step": 5820 }, { "epoch": 37.13375796178344, "grad_norm": 3.1631205081939697, "learning_rate": 0.0001, "loss": 0.0777, "step": 5830 }, { "epoch": 37.197452229299365, "grad_norm": 3.3032479286193848, "learning_rate": 0.0001, "loss": 0.0734, "step": 5840 }, { "epoch": 37.261146496815286, "grad_norm": 3.1174168586730957, "learning_rate": 0.0001, "loss": 0.0726, "step": 5850 }, { "epoch": 37.32484076433121, "grad_norm": 3.033862590789795, "learning_rate": 0.0001, "loss": 0.077, "step": 5860 }, { "epoch": 37.388535031847134, "grad_norm": 2.4495224952697754, "learning_rate": 0.0001, "loss": 0.0751, "step": 5870 }, { "epoch": 37.452229299363054, "grad_norm": 2.6152760982513428, "learning_rate": 0.0001, "loss": 0.0723, "step": 5880 }, { "epoch": 37.51592356687898, "grad_norm": 2.8707997798919678, "learning_rate": 0.0001, "loss": 0.0743, "step": 5890 }, { "epoch": 37.5796178343949, "grad_norm": 2.7761616706848145, "learning_rate": 0.0001, "loss": 0.0733, "step": 5900 }, { "epoch": 37.64331210191083, "grad_norm": 3.5413320064544678, "learning_rate": 0.0001, "loss": 0.073, "step": 5910 }, { "epoch": 37.70700636942675, "grad_norm": 3.2843964099884033, "learning_rate": 0.0001, "loss": 0.0681, "step": 5920 }, { "epoch": 37.77070063694268, "grad_norm": 3.010165214538574, "learning_rate": 0.0001, "loss": 0.0712, "step": 5930 }, { "epoch": 37.8343949044586, "grad_norm": 3.393099784851074, "learning_rate": 0.0001, "loss": 0.07, "step": 5940 }, { "epoch": 37.89808917197452, "grad_norm": 3.2045674324035645, "learning_rate": 0.0001, "loss": 0.067, "step": 5950 }, { "epoch": 37.961783439490446, "grad_norm": 3.12638258934021, "learning_rate": 0.0001, "loss": 0.0751, "step": 5960 }, { "epoch": 38.02547770700637, "grad_norm": 2.962380886077881, "learning_rate": 0.0001, "loss": 0.0691, "step": 5970 }, { "epoch": 38.089171974522294, "grad_norm": 2.8237619400024414, "learning_rate": 0.0001, "loss": 0.0689, "step": 5980 }, { "epoch": 38.152866242038215, "grad_norm": 2.81015944480896, "learning_rate": 0.0001, "loss": 0.0697, "step": 5990 }, { "epoch": 38.21656050955414, "grad_norm": 2.590351104736328, "learning_rate": 0.0001, "loss": 0.0712, "step": 6000 }, { "epoch": 38.28025477707006, "grad_norm": 2.849849224090576, "learning_rate": 0.0001, "loss": 0.0729, "step": 6010 }, { "epoch": 38.34394904458599, "grad_norm": 2.8964099884033203, "learning_rate": 0.0001, "loss": 0.0705, "step": 6020 }, { "epoch": 38.40764331210191, "grad_norm": 2.4432287216186523, "learning_rate": 0.0001, "loss": 0.0681, "step": 6030 }, { "epoch": 38.47133757961783, "grad_norm": 3.0105953216552734, "learning_rate": 0.0001, "loss": 0.0678, "step": 6040 }, { "epoch": 38.53503184713376, "grad_norm": 2.4857983589172363, "learning_rate": 0.0001, "loss": 0.0696, "step": 6050 }, { "epoch": 38.59872611464968, "grad_norm": 2.8537991046905518, "learning_rate": 0.0001, "loss": 0.0694, "step": 6060 }, { "epoch": 38.66242038216561, "grad_norm": 2.471968650817871, "learning_rate": 0.0001, "loss": 0.0685, "step": 6070 }, { "epoch": 38.72611464968153, "grad_norm": 3.065091848373413, "learning_rate": 0.0001, "loss": 0.0685, "step": 6080 }, { "epoch": 38.789808917197455, "grad_norm": 2.8296918869018555, "learning_rate": 0.0001, "loss": 0.0707, "step": 6090 }, { "epoch": 38.853503184713375, "grad_norm": 2.9093809127807617, "learning_rate": 0.0001, "loss": 0.0693, "step": 6100 }, { "epoch": 38.9171974522293, "grad_norm": 2.8551552295684814, "learning_rate": 0.0001, "loss": 0.069, "step": 6110 }, { "epoch": 38.98089171974522, "grad_norm": 2.778142213821411, "learning_rate": 0.0001, "loss": 0.0703, "step": 6120 }, { "epoch": 39.044585987261144, "grad_norm": 2.9948389530181885, "learning_rate": 0.0001, "loss": 0.0698, "step": 6130 }, { "epoch": 39.10828025477707, "grad_norm": 2.7430057525634766, "learning_rate": 0.0001, "loss": 0.0695, "step": 6140 }, { "epoch": 39.17197452229299, "grad_norm": 3.07726788520813, "learning_rate": 0.0001, "loss": 0.0699, "step": 6150 }, { "epoch": 39.23566878980892, "grad_norm": 2.6856625080108643, "learning_rate": 0.0001, "loss": 0.0678, "step": 6160 }, { "epoch": 39.29936305732484, "grad_norm": 3.3479177951812744, "learning_rate": 0.0001, "loss": 0.0733, "step": 6170 }, { "epoch": 39.36305732484077, "grad_norm": 3.028226375579834, "learning_rate": 0.0001, "loss": 0.0699, "step": 6180 }, { "epoch": 39.42675159235669, "grad_norm": 3.397296905517578, "learning_rate": 0.0001, "loss": 0.0684, "step": 6190 }, { "epoch": 39.49044585987261, "grad_norm": 2.846815824508667, "learning_rate": 0.0001, "loss": 0.07, "step": 6200 }, { "epoch": 39.554140127388536, "grad_norm": 2.7761709690093994, "learning_rate": 0.0001, "loss": 0.072, "step": 6210 }, { "epoch": 39.617834394904456, "grad_norm": 3.260953664779663, "learning_rate": 0.0001, "loss": 0.0718, "step": 6220 }, { "epoch": 39.681528662420384, "grad_norm": 2.407089948654175, "learning_rate": 0.0001, "loss": 0.0705, "step": 6230 }, { "epoch": 39.745222929936304, "grad_norm": 2.920893907546997, "learning_rate": 0.0001, "loss": 0.069, "step": 6240 }, { "epoch": 39.80891719745223, "grad_norm": 2.628615617752075, "learning_rate": 0.0001, "loss": 0.0673, "step": 6250 }, { "epoch": 39.87261146496815, "grad_norm": 3.0904760360717773, "learning_rate": 0.0001, "loss": 0.0698, "step": 6260 }, { "epoch": 39.93630573248408, "grad_norm": 2.731755018234253, "learning_rate": 0.0001, "loss": 0.0709, "step": 6270 }, { "epoch": 40.0, "grad_norm": 2.5712578296661377, "learning_rate": 0.0001, "loss": 0.0673, "step": 6280 }, { "epoch": 40.06369426751592, "grad_norm": 2.927391767501831, "learning_rate": 0.0001, "loss": 0.0657, "step": 6290 }, { "epoch": 40.12738853503185, "grad_norm": 2.723423957824707, "learning_rate": 0.0001, "loss": 0.0674, "step": 6300 }, { "epoch": 40.19108280254777, "grad_norm": 2.5323219299316406, "learning_rate": 0.0001, "loss": 0.0727, "step": 6310 }, { "epoch": 40.254777070063696, "grad_norm": 2.708711862564087, "learning_rate": 0.0001, "loss": 0.0755, "step": 6320 }, { "epoch": 40.318471337579616, "grad_norm": 2.786921262741089, "learning_rate": 0.0001, "loss": 0.0715, "step": 6330 }, { "epoch": 40.382165605095544, "grad_norm": 2.708756446838379, "learning_rate": 0.0001, "loss": 0.0709, "step": 6340 }, { "epoch": 40.445859872611464, "grad_norm": 3.1912639141082764, "learning_rate": 0.0001, "loss": 0.068, "step": 6350 }, { "epoch": 40.50955414012739, "grad_norm": 2.962975263595581, "learning_rate": 0.0001, "loss": 0.0683, "step": 6360 }, { "epoch": 40.57324840764331, "grad_norm": 2.6959781646728516, "learning_rate": 0.0001, "loss": 0.0713, "step": 6370 }, { "epoch": 40.63694267515923, "grad_norm": 2.9620680809020996, "learning_rate": 0.0001, "loss": 0.0728, "step": 6380 }, { "epoch": 40.70063694267516, "grad_norm": 2.9797191619873047, "learning_rate": 0.0001, "loss": 0.0708, "step": 6390 }, { "epoch": 40.76433121019108, "grad_norm": 2.4183712005615234, "learning_rate": 0.0001, "loss": 0.0711, "step": 6400 }, { "epoch": 40.82802547770701, "grad_norm": 2.409252882003784, "learning_rate": 0.0001, "loss": 0.0689, "step": 6410 }, { "epoch": 40.89171974522293, "grad_norm": 2.261404514312744, "learning_rate": 0.0001, "loss": 0.0721, "step": 6420 }, { "epoch": 40.955414012738856, "grad_norm": 2.704946279525757, "learning_rate": 0.0001, "loss": 0.073, "step": 6430 }, { "epoch": 41.01910828025478, "grad_norm": 2.4236435890197754, "learning_rate": 0.0001, "loss": 0.0713, "step": 6440 }, { "epoch": 41.0828025477707, "grad_norm": 2.574367046356201, "learning_rate": 0.0001, "loss": 0.0681, "step": 6450 }, { "epoch": 41.146496815286625, "grad_norm": 2.8261237144470215, "learning_rate": 0.0001, "loss": 0.0733, "step": 6460 }, { "epoch": 41.210191082802545, "grad_norm": 2.4770219326019287, "learning_rate": 0.0001, "loss": 0.0735, "step": 6470 }, { "epoch": 41.27388535031847, "grad_norm": 3.4225709438323975, "learning_rate": 0.0001, "loss": 0.0727, "step": 6480 }, { "epoch": 41.33757961783439, "grad_norm": 3.0220494270324707, "learning_rate": 0.0001, "loss": 0.0676, "step": 6490 }, { "epoch": 41.40127388535032, "grad_norm": 2.862534523010254, "learning_rate": 0.0001, "loss": 0.0648, "step": 6500 }, { "epoch": 41.46496815286624, "grad_norm": 2.6460349559783936, "learning_rate": 0.0001, "loss": 0.0684, "step": 6510 }, { "epoch": 41.52866242038217, "grad_norm": 2.84767484664917, "learning_rate": 0.0001, "loss": 0.067, "step": 6520 }, { "epoch": 41.59235668789809, "grad_norm": 2.793433427810669, "learning_rate": 0.0001, "loss": 0.0675, "step": 6530 }, { "epoch": 41.65605095541401, "grad_norm": 2.5421557426452637, "learning_rate": 0.0001, "loss": 0.0685, "step": 6540 }, { "epoch": 41.71974522292994, "grad_norm": 2.600630760192871, "learning_rate": 0.0001, "loss": 0.0691, "step": 6550 }, { "epoch": 41.78343949044586, "grad_norm": 2.878746271133423, "learning_rate": 0.0001, "loss": 0.0674, "step": 6560 }, { "epoch": 41.847133757961785, "grad_norm": 2.781005859375, "learning_rate": 0.0001, "loss": 0.0637, "step": 6570 }, { "epoch": 41.910828025477706, "grad_norm": 2.8469104766845703, "learning_rate": 0.0001, "loss": 0.0641, "step": 6580 }, { "epoch": 41.97452229299363, "grad_norm": 2.8917415142059326, "learning_rate": 0.0001, "loss": 0.0696, "step": 6590 }, { "epoch": 42.038216560509554, "grad_norm": 2.511179208755493, "learning_rate": 0.0001, "loss": 0.0694, "step": 6600 }, { "epoch": 42.10191082802548, "grad_norm": 2.707634449005127, "learning_rate": 0.0001, "loss": 0.0701, "step": 6610 }, { "epoch": 42.1656050955414, "grad_norm": 2.2659246921539307, "learning_rate": 0.0001, "loss": 0.0702, "step": 6620 }, { "epoch": 42.22929936305732, "grad_norm": 3.1509242057800293, "learning_rate": 0.0001, "loss": 0.0673, "step": 6630 }, { "epoch": 42.29299363057325, "grad_norm": 2.6814181804656982, "learning_rate": 0.0001, "loss": 0.0665, "step": 6640 }, { "epoch": 42.35668789808917, "grad_norm": 3.9488563537597656, "learning_rate": 0.0001, "loss": 0.0699, "step": 6650 }, { "epoch": 42.4203821656051, "grad_norm": 2.8163771629333496, "learning_rate": 0.0001, "loss": 0.067, "step": 6660 }, { "epoch": 42.48407643312102, "grad_norm": 3.175305128097534, "learning_rate": 0.0001, "loss": 0.0633, "step": 6670 }, { "epoch": 42.547770700636946, "grad_norm": 3.2769086360931396, "learning_rate": 0.0001, "loss": 0.0651, "step": 6680 }, { "epoch": 42.611464968152866, "grad_norm": 2.6479954719543457, "learning_rate": 0.0001, "loss": 0.0597, "step": 6690 }, { "epoch": 42.67515923566879, "grad_norm": 2.692216634750366, "learning_rate": 0.0001, "loss": 0.0656, "step": 6700 }, { "epoch": 42.738853503184714, "grad_norm": 2.6727182865142822, "learning_rate": 0.0001, "loss": 0.0665, "step": 6710 }, { "epoch": 42.802547770700635, "grad_norm": 2.8885445594787598, "learning_rate": 0.0001, "loss": 0.0632, "step": 6720 }, { "epoch": 42.86624203821656, "grad_norm": 2.6378605365753174, "learning_rate": 0.0001, "loss": 0.0681, "step": 6730 }, { "epoch": 42.92993630573248, "grad_norm": 2.5806725025177, "learning_rate": 0.0001, "loss": 0.0632, "step": 6740 }, { "epoch": 42.99363057324841, "grad_norm": 2.655748128890991, "learning_rate": 0.0001, "loss": 0.064, "step": 6750 }, { "epoch": 43.05732484076433, "grad_norm": 2.6751420497894287, "learning_rate": 0.0001, "loss": 0.0668, "step": 6760 }, { "epoch": 43.12101910828026, "grad_norm": 3.0048224925994873, "learning_rate": 0.0001, "loss": 0.0618, "step": 6770 }, { "epoch": 43.18471337579618, "grad_norm": 2.1828463077545166, "learning_rate": 0.0001, "loss": 0.0607, "step": 6780 }, { "epoch": 43.2484076433121, "grad_norm": 2.4600672721862793, "learning_rate": 0.0001, "loss": 0.0668, "step": 6790 }, { "epoch": 43.31210191082803, "grad_norm": 3.159379482269287, "learning_rate": 0.0001, "loss": 0.0677, "step": 6800 }, { "epoch": 43.37579617834395, "grad_norm": 2.7637484073638916, "learning_rate": 0.0001, "loss": 0.0676, "step": 6810 }, { "epoch": 43.439490445859875, "grad_norm": 2.6934993267059326, "learning_rate": 0.0001, "loss": 0.0635, "step": 6820 }, { "epoch": 43.503184713375795, "grad_norm": 2.607598066329956, "learning_rate": 0.0001, "loss": 0.0689, "step": 6830 }, { "epoch": 43.56687898089172, "grad_norm": 2.3252906799316406, "learning_rate": 0.0001, "loss": 0.0669, "step": 6840 }, { "epoch": 43.63057324840764, "grad_norm": 2.419931411743164, "learning_rate": 0.0001, "loss": 0.0645, "step": 6850 }, { "epoch": 43.69426751592356, "grad_norm": 2.66821551322937, "learning_rate": 0.0001, "loss": 0.0669, "step": 6860 }, { "epoch": 43.75796178343949, "grad_norm": 2.7597317695617676, "learning_rate": 0.0001, "loss": 0.0652, "step": 6870 }, { "epoch": 43.82165605095541, "grad_norm": 2.510902166366577, "learning_rate": 0.0001, "loss": 0.0674, "step": 6880 }, { "epoch": 43.88535031847134, "grad_norm": 2.8532636165618896, "learning_rate": 0.0001, "loss": 0.0651, "step": 6890 }, { "epoch": 43.94904458598726, "grad_norm": 2.616150379180908, "learning_rate": 0.0001, "loss": 0.0679, "step": 6900 }, { "epoch": 44.01273885350319, "grad_norm": 2.6965529918670654, "learning_rate": 0.0001, "loss": 0.0688, "step": 6910 }, { "epoch": 44.07643312101911, "grad_norm": 2.286072254180908, "learning_rate": 0.0001, "loss": 0.0653, "step": 6920 }, { "epoch": 44.140127388535035, "grad_norm": 2.586181640625, "learning_rate": 0.0001, "loss": 0.0672, "step": 6930 }, { "epoch": 44.203821656050955, "grad_norm": 2.784291982650757, "learning_rate": 0.0001, "loss": 0.0668, "step": 6940 }, { "epoch": 44.267515923566876, "grad_norm": 2.824005603790283, "learning_rate": 0.0001, "loss": 0.0681, "step": 6950 }, { "epoch": 44.3312101910828, "grad_norm": 2.655280590057373, "learning_rate": 0.0001, "loss": 0.0621, "step": 6960 }, { "epoch": 44.394904458598724, "grad_norm": 2.437988519668579, "learning_rate": 0.0001, "loss": 0.0663, "step": 6970 }, { "epoch": 44.45859872611465, "grad_norm": 2.366197109222412, "learning_rate": 0.0001, "loss": 0.0652, "step": 6980 }, { "epoch": 44.52229299363057, "grad_norm": 2.4231009483337402, "learning_rate": 0.0001, "loss": 0.0679, "step": 6990 }, { "epoch": 44.5859872611465, "grad_norm": 2.379183053970337, "learning_rate": 0.0001, "loss": 0.066, "step": 7000 }, { "epoch": 44.64968152866242, "grad_norm": 2.523529529571533, "learning_rate": 0.0001, "loss": 0.0724, "step": 7010 }, { "epoch": 44.71337579617835, "grad_norm": 2.739490270614624, "learning_rate": 0.0001, "loss": 0.0734, "step": 7020 }, { "epoch": 44.77707006369427, "grad_norm": 2.201383352279663, "learning_rate": 0.0001, "loss": 0.0647, "step": 7030 }, { "epoch": 44.84076433121019, "grad_norm": 2.705928325653076, "learning_rate": 0.0001, "loss": 0.0666, "step": 7040 }, { "epoch": 44.904458598726116, "grad_norm": 2.949538469314575, "learning_rate": 0.0001, "loss": 0.0699, "step": 7050 }, { "epoch": 44.968152866242036, "grad_norm": 2.7554826736450195, "learning_rate": 0.0001, "loss": 0.0663, "step": 7060 }, { "epoch": 45.031847133757964, "grad_norm": 2.619295120239258, "learning_rate": 0.0001, "loss": 0.0698, "step": 7070 }, { "epoch": 45.095541401273884, "grad_norm": 2.776186466217041, "learning_rate": 0.0001, "loss": 0.0663, "step": 7080 }, { "epoch": 45.15923566878981, "grad_norm": 2.6860883235931396, "learning_rate": 0.0001, "loss": 0.064, "step": 7090 }, { "epoch": 45.22292993630573, "grad_norm": 2.43002986907959, "learning_rate": 0.0001, "loss": 0.0678, "step": 7100 }, { "epoch": 45.28662420382165, "grad_norm": 2.398918628692627, "learning_rate": 0.0001, "loss": 0.0682, "step": 7110 }, { "epoch": 45.35031847133758, "grad_norm": 2.3382205963134766, "learning_rate": 0.0001, "loss": 0.0693, "step": 7120 }, { "epoch": 45.4140127388535, "grad_norm": 2.0985167026519775, "learning_rate": 0.0001, "loss": 0.0625, "step": 7130 }, { "epoch": 45.47770700636943, "grad_norm": 2.657869577407837, "learning_rate": 0.0001, "loss": 0.0644, "step": 7140 }, { "epoch": 45.54140127388535, "grad_norm": 2.4839930534362793, "learning_rate": 0.0001, "loss": 0.0703, "step": 7150 }, { "epoch": 45.605095541401276, "grad_norm": 2.5947165489196777, "learning_rate": 0.0001, "loss": 0.0665, "step": 7160 }, { "epoch": 45.6687898089172, "grad_norm": 2.189042568206787, "learning_rate": 0.0001, "loss": 0.0681, "step": 7170 }, { "epoch": 45.732484076433124, "grad_norm": 2.15224027633667, "learning_rate": 0.0001, "loss": 0.0642, "step": 7180 }, { "epoch": 45.796178343949045, "grad_norm": 2.220731258392334, "learning_rate": 0.0001, "loss": 0.0646, "step": 7190 }, { "epoch": 45.859872611464965, "grad_norm": 2.711808204650879, "learning_rate": 0.0001, "loss": 0.0654, "step": 7200 }, { "epoch": 45.92356687898089, "grad_norm": 2.2829248905181885, "learning_rate": 0.0001, "loss": 0.0645, "step": 7210 }, { "epoch": 45.98726114649681, "grad_norm": 2.6629161834716797, "learning_rate": 0.0001, "loss": 0.065, "step": 7220 }, { "epoch": 46.05095541401274, "grad_norm": 2.72788143157959, "learning_rate": 0.0001, "loss": 0.0647, "step": 7230 }, { "epoch": 46.11464968152866, "grad_norm": 2.622267961502075, "learning_rate": 0.0001, "loss": 0.0664, "step": 7240 }, { "epoch": 46.17834394904459, "grad_norm": 2.842200517654419, "learning_rate": 0.0001, "loss": 0.0697, "step": 7250 }, { "epoch": 46.24203821656051, "grad_norm": 2.6117236614227295, "learning_rate": 0.0001, "loss": 0.0635, "step": 7260 }, { "epoch": 46.30573248407644, "grad_norm": 2.7575292587280273, "learning_rate": 0.0001, "loss": 0.0662, "step": 7270 }, { "epoch": 46.36942675159236, "grad_norm": 2.3936774730682373, "learning_rate": 0.0001, "loss": 0.07, "step": 7280 }, { "epoch": 46.43312101910828, "grad_norm": 2.370028018951416, "learning_rate": 0.0001, "loss": 0.0625, "step": 7290 }, { "epoch": 46.496815286624205, "grad_norm": 2.40263295173645, "learning_rate": 0.0001, "loss": 0.0686, "step": 7300 }, { "epoch": 46.560509554140125, "grad_norm": 1.9591044187545776, "learning_rate": 0.0001, "loss": 0.0635, "step": 7310 }, { "epoch": 46.62420382165605, "grad_norm": 2.6083009243011475, "learning_rate": 0.0001, "loss": 0.0647, "step": 7320 }, { "epoch": 46.68789808917197, "grad_norm": 2.3341236114501953, "learning_rate": 0.0001, "loss": 0.0648, "step": 7330 }, { "epoch": 46.7515923566879, "grad_norm": 2.599531412124634, "learning_rate": 0.0001, "loss": 0.0661, "step": 7340 }, { "epoch": 46.81528662420382, "grad_norm": 2.233501672744751, "learning_rate": 0.0001, "loss": 0.0675, "step": 7350 }, { "epoch": 46.87898089171974, "grad_norm": 2.2574212551116943, "learning_rate": 0.0001, "loss": 0.0611, "step": 7360 }, { "epoch": 46.94267515923567, "grad_norm": 2.464907646179199, "learning_rate": 0.0001, "loss": 0.0674, "step": 7370 }, { "epoch": 47.00636942675159, "grad_norm": 2.5567779541015625, "learning_rate": 0.0001, "loss": 0.0615, "step": 7380 }, { "epoch": 47.07006369426752, "grad_norm": 2.307438850402832, "learning_rate": 0.0001, "loss": 0.0674, "step": 7390 }, { "epoch": 47.13375796178344, "grad_norm": 2.2819576263427734, "learning_rate": 0.0001, "loss": 0.0633, "step": 7400 }, { "epoch": 47.197452229299365, "grad_norm": 2.3263370990753174, "learning_rate": 0.0001, "loss": 0.065, "step": 7410 }, { "epoch": 47.261146496815286, "grad_norm": 2.364497661590576, "learning_rate": 0.0001, "loss": 0.0629, "step": 7420 }, { "epoch": 47.32484076433121, "grad_norm": 2.4211342334747314, "learning_rate": 0.0001, "loss": 0.0621, "step": 7430 }, { "epoch": 47.388535031847134, "grad_norm": 2.3898682594299316, "learning_rate": 0.0001, "loss": 0.0627, "step": 7440 }, { "epoch": 47.452229299363054, "grad_norm": 2.625920057296753, "learning_rate": 0.0001, "loss": 0.0666, "step": 7450 }, { "epoch": 47.51592356687898, "grad_norm": 2.524195671081543, "learning_rate": 0.0001, "loss": 0.0687, "step": 7460 }, { "epoch": 47.5796178343949, "grad_norm": 2.5412731170654297, "learning_rate": 0.0001, "loss": 0.0664, "step": 7470 }, { "epoch": 47.64331210191083, "grad_norm": 2.2503561973571777, "learning_rate": 0.0001, "loss": 0.0645, "step": 7480 }, { "epoch": 47.70700636942675, "grad_norm": 2.4177491664886475, "learning_rate": 0.0001, "loss": 0.0667, "step": 7490 }, { "epoch": 47.77070063694268, "grad_norm": 2.260131359100342, "learning_rate": 0.0001, "loss": 0.0635, "step": 7500 }, { "epoch": 47.8343949044586, "grad_norm": 2.161890983581543, "learning_rate": 0.0001, "loss": 0.064, "step": 7510 }, { "epoch": 47.89808917197452, "grad_norm": 2.224578857421875, "learning_rate": 0.0001, "loss": 0.066, "step": 7520 }, { "epoch": 47.961783439490446, "grad_norm": 4.481538772583008, "learning_rate": 0.0001, "loss": 0.0694, "step": 7530 }, { "epoch": 48.02547770700637, "grad_norm": 5.4635467529296875, "learning_rate": 0.0001, "loss": 0.076, "step": 7540 }, { "epoch": 48.089171974522294, "grad_norm": 3.3924317359924316, "learning_rate": 0.0001, "loss": 0.0679, "step": 7550 }, { "epoch": 48.152866242038215, "grad_norm": 3.663449764251709, "learning_rate": 0.0001, "loss": 0.0663, "step": 7560 }, { "epoch": 48.21656050955414, "grad_norm": 3.0875015258789062, "learning_rate": 0.0001, "loss": 0.062, "step": 7570 }, { "epoch": 48.28025477707006, "grad_norm": 2.8413894176483154, "learning_rate": 0.0001, "loss": 0.0625, "step": 7580 }, { "epoch": 48.34394904458599, "grad_norm": 2.7076492309570312, "learning_rate": 0.0001, "loss": 0.0641, "step": 7590 }, { "epoch": 48.40764331210191, "grad_norm": 2.5717813968658447, "learning_rate": 0.0001, "loss": 0.0618, "step": 7600 }, { "epoch": 48.47133757961783, "grad_norm": 2.254018545150757, "learning_rate": 0.0001, "loss": 0.0631, "step": 7610 }, { "epoch": 48.53503184713376, "grad_norm": 2.6733853816986084, "learning_rate": 0.0001, "loss": 0.0624, "step": 7620 }, { "epoch": 48.59872611464968, "grad_norm": 2.2812564373016357, "learning_rate": 0.0001, "loss": 0.0618, "step": 7630 }, { "epoch": 48.66242038216561, "grad_norm": 3.067448616027832, "learning_rate": 0.0001, "loss": 0.0612, "step": 7640 }, { "epoch": 48.72611464968153, "grad_norm": 2.1545510292053223, "learning_rate": 0.0001, "loss": 0.0647, "step": 7650 }, { "epoch": 48.789808917197455, "grad_norm": 2.150322914123535, "learning_rate": 0.0001, "loss": 0.0584, "step": 7660 }, { "epoch": 48.853503184713375, "grad_norm": 2.4993960857391357, "learning_rate": 0.0001, "loss": 0.0637, "step": 7670 }, { "epoch": 48.9171974522293, "grad_norm": 2.3427574634552, "learning_rate": 0.0001, "loss": 0.0639, "step": 7680 }, { "epoch": 48.98089171974522, "grad_norm": 2.3255908489227295, "learning_rate": 0.0001, "loss": 0.0636, "step": 7690 }, { "epoch": 49.044585987261144, "grad_norm": 2.461595296859741, "learning_rate": 0.0001, "loss": 0.0609, "step": 7700 }, { "epoch": 49.10828025477707, "grad_norm": 2.2738327980041504, "learning_rate": 0.0001, "loss": 0.062, "step": 7710 }, { "epoch": 49.17197452229299, "grad_norm": 2.4575259685516357, "learning_rate": 0.0001, "loss": 0.0626, "step": 7720 }, { "epoch": 49.23566878980892, "grad_norm": 2.508897066116333, "learning_rate": 0.0001, "loss": 0.06, "step": 7730 }, { "epoch": 49.29936305732484, "grad_norm": 2.395838737487793, "learning_rate": 0.0001, "loss": 0.0596, "step": 7740 }, { "epoch": 49.36305732484077, "grad_norm": 2.755337953567505, "learning_rate": 0.0001, "loss": 0.0638, "step": 7750 }, { "epoch": 49.42675159235669, "grad_norm": 2.521503210067749, "learning_rate": 0.0001, "loss": 0.062, "step": 7760 }, { "epoch": 49.49044585987261, "grad_norm": 3.2699108123779297, "learning_rate": 0.0001, "loss": 0.0667, "step": 7770 }, { "epoch": 49.554140127388536, "grad_norm": 2.358887195587158, "learning_rate": 0.0001, "loss": 0.059, "step": 7780 }, { "epoch": 49.617834394904456, "grad_norm": 2.3485734462738037, "learning_rate": 0.0001, "loss": 0.0604, "step": 7790 }, { "epoch": 49.681528662420384, "grad_norm": 2.2272520065307617, "learning_rate": 0.0001, "loss": 0.0591, "step": 7800 }, { "epoch": 49.745222929936304, "grad_norm": 2.0136051177978516, "learning_rate": 0.0001, "loss": 0.0605, "step": 7810 }, { "epoch": 49.80891719745223, "grad_norm": 2.445220708847046, "learning_rate": 0.0001, "loss": 0.0636, "step": 7820 }, { "epoch": 49.87261146496815, "grad_norm": 2.5135135650634766, "learning_rate": 0.0001, "loss": 0.0618, "step": 7830 }, { "epoch": 49.93630573248408, "grad_norm": 2.035425901412964, "learning_rate": 0.0001, "loss": 0.0654, "step": 7840 }, { "epoch": 50.0, "grad_norm": 2.293149709701538, "learning_rate": 0.0001, "loss": 0.0606, "step": 7850 }, { "epoch": 50.06369426751592, "grad_norm": 2.2461605072021484, "learning_rate": 0.0001, "loss": 0.0585, "step": 7860 }, { "epoch": 50.12738853503185, "grad_norm": 2.334841012954712, "learning_rate": 0.0001, "loss": 0.0622, "step": 7870 }, { "epoch": 50.19108280254777, "grad_norm": 2.193125009536743, "learning_rate": 0.0001, "loss": 0.0596, "step": 7880 }, { "epoch": 50.254777070063696, "grad_norm": 2.6340036392211914, "learning_rate": 0.0001, "loss": 0.061, "step": 7890 }, { "epoch": 50.318471337579616, "grad_norm": 2.3753435611724854, "learning_rate": 0.0001, "loss": 0.063, "step": 7900 }, { "epoch": 50.382165605095544, "grad_norm": 2.1447181701660156, "learning_rate": 0.0001, "loss": 0.0606, "step": 7910 }, { "epoch": 50.445859872611464, "grad_norm": 2.322434425354004, "learning_rate": 0.0001, "loss": 0.0563, "step": 7920 }, { "epoch": 50.50955414012739, "grad_norm": 2.416712522506714, "learning_rate": 0.0001, "loss": 0.059, "step": 7930 }, { "epoch": 50.57324840764331, "grad_norm": 2.265883207321167, "learning_rate": 0.0001, "loss": 0.0598, "step": 7940 }, { "epoch": 50.63694267515923, "grad_norm": 2.0436618328094482, "learning_rate": 0.0001, "loss": 0.0605, "step": 7950 }, { "epoch": 50.70063694267516, "grad_norm": 2.6686179637908936, "learning_rate": 0.0001, "loss": 0.0619, "step": 7960 }, { "epoch": 50.76433121019108, "grad_norm": 2.571150302886963, "learning_rate": 0.0001, "loss": 0.0619, "step": 7970 }, { "epoch": 50.82802547770701, "grad_norm": 2.1832380294799805, "learning_rate": 0.0001, "loss": 0.0589, "step": 7980 }, { "epoch": 50.89171974522293, "grad_norm": 2.595506429672241, "learning_rate": 0.0001, "loss": 0.068, "step": 7990 }, { "epoch": 50.955414012738856, "grad_norm": 2.1001131534576416, "learning_rate": 0.0001, "loss": 0.0638, "step": 8000 }, { "epoch": 51.01910828025478, "grad_norm": 2.1582157611846924, "learning_rate": 0.0001, "loss": 0.0593, "step": 8010 }, { "epoch": 51.0828025477707, "grad_norm": 1.9467597007751465, "learning_rate": 0.0001, "loss": 0.0649, "step": 8020 }, { "epoch": 51.146496815286625, "grad_norm": 2.2735583782196045, "learning_rate": 0.0001, "loss": 0.062, "step": 8030 }, { "epoch": 51.210191082802545, "grad_norm": 2.18703556060791, "learning_rate": 0.0001, "loss": 0.0638, "step": 8040 }, { "epoch": 51.27388535031847, "grad_norm": 2.0311028957366943, "learning_rate": 0.0001, "loss": 0.0606, "step": 8050 }, { "epoch": 51.33757961783439, "grad_norm": 2.4906489849090576, "learning_rate": 0.0001, "loss": 0.0621, "step": 8060 }, { "epoch": 51.40127388535032, "grad_norm": 2.0712673664093018, "learning_rate": 0.0001, "loss": 0.0579, "step": 8070 }, { "epoch": 51.46496815286624, "grad_norm": 2.213200569152832, "learning_rate": 0.0001, "loss": 0.0621, "step": 8080 }, { "epoch": 51.52866242038217, "grad_norm": 2.294825315475464, "learning_rate": 0.0001, "loss": 0.0617, "step": 8090 }, { "epoch": 51.59235668789809, "grad_norm": 2.043128490447998, "learning_rate": 0.0001, "loss": 0.0626, "step": 8100 }, { "epoch": 51.65605095541401, "grad_norm": 2.1958224773406982, "learning_rate": 0.0001, "loss": 0.0593, "step": 8110 }, { "epoch": 51.71974522292994, "grad_norm": 2.391763925552368, "learning_rate": 0.0001, "loss": 0.0613, "step": 8120 }, { "epoch": 51.78343949044586, "grad_norm": 2.261824607849121, "learning_rate": 0.0001, "loss": 0.0601, "step": 8130 }, { "epoch": 51.847133757961785, "grad_norm": 2.181659698486328, "learning_rate": 0.0001, "loss": 0.0643, "step": 8140 }, { "epoch": 51.910828025477706, "grad_norm": 2.1882381439208984, "learning_rate": 0.0001, "loss": 0.0621, "step": 8150 }, { "epoch": 51.97452229299363, "grad_norm": 2.181298017501831, "learning_rate": 0.0001, "loss": 0.0611, "step": 8160 }, { "epoch": 52.038216560509554, "grad_norm": 2.0960328578948975, "learning_rate": 0.0001, "loss": 0.0646, "step": 8170 }, { "epoch": 52.10191082802548, "grad_norm": 1.8877384662628174, "learning_rate": 0.0001, "loss": 0.0593, "step": 8180 }, { "epoch": 52.1656050955414, "grad_norm": 2.3050284385681152, "learning_rate": 0.0001, "loss": 0.0584, "step": 8190 }, { "epoch": 52.22929936305732, "grad_norm": 2.2422308921813965, "learning_rate": 0.0001, "loss": 0.0602, "step": 8200 }, { "epoch": 52.29299363057325, "grad_norm": 1.9882867336273193, "learning_rate": 0.0001, "loss": 0.0601, "step": 8210 }, { "epoch": 52.35668789808917, "grad_norm": 1.9608228206634521, "learning_rate": 0.0001, "loss": 0.0562, "step": 8220 }, { "epoch": 52.4203821656051, "grad_norm": 2.1209702491760254, "learning_rate": 0.0001, "loss": 0.0614, "step": 8230 }, { "epoch": 52.48407643312102, "grad_norm": 2.1777074337005615, "learning_rate": 0.0001, "loss": 0.0623, "step": 8240 }, { "epoch": 52.547770700636946, "grad_norm": 2.035221815109253, "learning_rate": 0.0001, "loss": 0.0628, "step": 8250 }, { "epoch": 52.611464968152866, "grad_norm": 2.0458855628967285, "learning_rate": 0.0001, "loss": 0.0636, "step": 8260 }, { "epoch": 52.67515923566879, "grad_norm": 2.0692312717437744, "learning_rate": 0.0001, "loss": 0.068, "step": 8270 }, { "epoch": 52.738853503184714, "grad_norm": 1.991896152496338, "learning_rate": 0.0001, "loss": 0.0648, "step": 8280 }, { "epoch": 52.802547770700635, "grad_norm": 1.878970742225647, "learning_rate": 0.0001, "loss": 0.068, "step": 8290 }, { "epoch": 52.86624203821656, "grad_norm": 2.1788079738616943, "learning_rate": 0.0001, "loss": 0.0634, "step": 8300 }, { "epoch": 52.92993630573248, "grad_norm": 2.3087518215179443, "learning_rate": 0.0001, "loss": 0.0634, "step": 8310 }, { "epoch": 52.99363057324841, "grad_norm": 2.3174006938934326, "learning_rate": 0.0001, "loss": 0.0579, "step": 8320 }, { "epoch": 53.05732484076433, "grad_norm": 1.7299202680587769, "learning_rate": 0.0001, "loss": 0.0628, "step": 8330 }, { "epoch": 53.12101910828026, "grad_norm": 2.3164398670196533, "learning_rate": 0.0001, "loss": 0.0592, "step": 8340 }, { "epoch": 53.18471337579618, "grad_norm": 2.3017163276672363, "learning_rate": 0.0001, "loss": 0.0613, "step": 8350 }, { "epoch": 53.2484076433121, "grad_norm": 2.1909844875335693, "learning_rate": 0.0001, "loss": 0.0593, "step": 8360 }, { "epoch": 53.31210191082803, "grad_norm": 2.5799436569213867, "learning_rate": 0.0001, "loss": 0.0583, "step": 8370 }, { "epoch": 53.37579617834395, "grad_norm": 2.385820150375366, "learning_rate": 0.0001, "loss": 0.0615, "step": 8380 }, { "epoch": 53.439490445859875, "grad_norm": 2.1424314975738525, "learning_rate": 0.0001, "loss": 0.0608, "step": 8390 }, { "epoch": 53.503184713375795, "grad_norm": 2.0919950008392334, "learning_rate": 0.0001, "loss": 0.059, "step": 8400 }, { "epoch": 53.56687898089172, "grad_norm": 2.381542444229126, "learning_rate": 0.0001, "loss": 0.0567, "step": 8410 }, { "epoch": 53.63057324840764, "grad_norm": 2.1572699546813965, "learning_rate": 0.0001, "loss": 0.0596, "step": 8420 }, { "epoch": 53.69426751592356, "grad_norm": 1.7683459520339966, "learning_rate": 0.0001, "loss": 0.0632, "step": 8430 }, { "epoch": 53.75796178343949, "grad_norm": 2.300126791000366, "learning_rate": 0.0001, "loss": 0.0651, "step": 8440 }, { "epoch": 53.82165605095541, "grad_norm": 2.3747289180755615, "learning_rate": 0.0001, "loss": 0.0612, "step": 8450 }, { "epoch": 53.88535031847134, "grad_norm": 2.2909796237945557, "learning_rate": 0.0001, "loss": 0.0594, "step": 8460 }, { "epoch": 53.94904458598726, "grad_norm": 1.957448124885559, "learning_rate": 0.0001, "loss": 0.0596, "step": 8470 }, { "epoch": 54.01273885350319, "grad_norm": 2.117011547088623, "learning_rate": 0.0001, "loss": 0.0605, "step": 8480 }, { "epoch": 54.07643312101911, "grad_norm": 2.1813671588897705, "learning_rate": 0.0001, "loss": 0.0597, "step": 8490 }, { "epoch": 54.140127388535035, "grad_norm": 2.2538459300994873, "learning_rate": 0.0001, "loss": 0.0603, "step": 8500 }, { "epoch": 54.203821656050955, "grad_norm": 1.9029637575149536, "learning_rate": 0.0001, "loss": 0.063, "step": 8510 }, { "epoch": 54.267515923566876, "grad_norm": 2.283144950866699, "learning_rate": 0.0001, "loss": 0.0545, "step": 8520 }, { "epoch": 54.3312101910828, "grad_norm": 2.8843955993652344, "learning_rate": 0.0001, "loss": 0.0599, "step": 8530 }, { "epoch": 54.394904458598724, "grad_norm": 3.529932975769043, "learning_rate": 0.0001, "loss": 0.0615, "step": 8540 }, { "epoch": 54.45859872611465, "grad_norm": 2.800692558288574, "learning_rate": 0.0001, "loss": 0.0596, "step": 8550 }, { "epoch": 54.52229299363057, "grad_norm": 2.419069766998291, "learning_rate": 0.0001, "loss": 0.0603, "step": 8560 }, { "epoch": 54.5859872611465, "grad_norm": 2.203068971633911, "learning_rate": 0.0001, "loss": 0.0573, "step": 8570 }, { "epoch": 54.64968152866242, "grad_norm": 1.8778634071350098, "learning_rate": 0.0001, "loss": 0.0543, "step": 8580 }, { "epoch": 54.71337579617835, "grad_norm": 2.2702441215515137, "learning_rate": 0.0001, "loss": 0.0563, "step": 8590 }, { "epoch": 54.77707006369427, "grad_norm": 2.3016202449798584, "learning_rate": 0.0001, "loss": 0.0582, "step": 8600 }, { "epoch": 54.84076433121019, "grad_norm": 2.517343521118164, "learning_rate": 0.0001, "loss": 0.0591, "step": 8610 }, { "epoch": 54.904458598726116, "grad_norm": 2.0755958557128906, "learning_rate": 0.0001, "loss": 0.062, "step": 8620 }, { "epoch": 54.968152866242036, "grad_norm": 2.379730463027954, "learning_rate": 0.0001, "loss": 0.0572, "step": 8630 }, { "epoch": 55.031847133757964, "grad_norm": 1.9931505918502808, "learning_rate": 0.0001, "loss": 0.0546, "step": 8640 }, { "epoch": 55.095541401273884, "grad_norm": 2.3119890689849854, "learning_rate": 0.0001, "loss": 0.0561, "step": 8650 }, { "epoch": 55.15923566878981, "grad_norm": 2.262129545211792, "learning_rate": 0.0001, "loss": 0.0584, "step": 8660 }, { "epoch": 55.22292993630573, "grad_norm": 2.1885366439819336, "learning_rate": 0.0001, "loss": 0.0561, "step": 8670 }, { "epoch": 55.28662420382165, "grad_norm": 1.9871503114700317, "learning_rate": 0.0001, "loss": 0.0603, "step": 8680 }, { "epoch": 55.35031847133758, "grad_norm": 2.380505084991455, "learning_rate": 0.0001, "loss": 0.0568, "step": 8690 }, { "epoch": 55.4140127388535, "grad_norm": 2.408733606338501, "learning_rate": 0.0001, "loss": 0.0612, "step": 8700 }, { "epoch": 55.47770700636943, "grad_norm": 2.3647005558013916, "learning_rate": 0.0001, "loss": 0.0602, "step": 8710 }, { "epoch": 55.54140127388535, "grad_norm": 1.9911019802093506, "learning_rate": 0.0001, "loss": 0.0606, "step": 8720 }, { "epoch": 55.605095541401276, "grad_norm": 1.9794394969940186, "learning_rate": 0.0001, "loss": 0.0622, "step": 8730 }, { "epoch": 55.6687898089172, "grad_norm": 2.225670337677002, "learning_rate": 0.0001, "loss": 0.0615, "step": 8740 }, { "epoch": 55.732484076433124, "grad_norm": 2.204319953918457, "learning_rate": 0.0001, "loss": 0.0593, "step": 8750 }, { "epoch": 55.796178343949045, "grad_norm": 2.327968120574951, "learning_rate": 0.0001, "loss": 0.0618, "step": 8760 }, { "epoch": 55.859872611464965, "grad_norm": 1.9408026933670044, "learning_rate": 0.0001, "loss": 0.0584, "step": 8770 }, { "epoch": 55.92356687898089, "grad_norm": 2.2269721031188965, "learning_rate": 0.0001, "loss": 0.0599, "step": 8780 }, { "epoch": 55.98726114649681, "grad_norm": 2.101301431655884, "learning_rate": 0.0001, "loss": 0.0566, "step": 8790 }, { "epoch": 56.05095541401274, "grad_norm": 2.1420910358428955, "learning_rate": 0.0001, "loss": 0.0612, "step": 8800 }, { "epoch": 56.11464968152866, "grad_norm": 2.2949187755584717, "learning_rate": 0.0001, "loss": 0.0556, "step": 8810 }, { "epoch": 56.17834394904459, "grad_norm": 2.357858896255493, "learning_rate": 0.0001, "loss": 0.0536, "step": 8820 }, { "epoch": 56.24203821656051, "grad_norm": 2.2129430770874023, "learning_rate": 0.0001, "loss": 0.0625, "step": 8830 }, { "epoch": 56.30573248407644, "grad_norm": 1.9557701349258423, "learning_rate": 0.0001, "loss": 0.0601, "step": 8840 }, { "epoch": 56.36942675159236, "grad_norm": 2.0575153827667236, "learning_rate": 0.0001, "loss": 0.0561, "step": 8850 }, { "epoch": 56.43312101910828, "grad_norm": 2.193183183670044, "learning_rate": 0.0001, "loss": 0.0581, "step": 8860 }, { "epoch": 56.496815286624205, "grad_norm": 1.8481546640396118, "learning_rate": 0.0001, "loss": 0.0595, "step": 8870 }, { "epoch": 56.560509554140125, "grad_norm": 2.394832134246826, "learning_rate": 0.0001, "loss": 0.0606, "step": 8880 }, { "epoch": 56.62420382165605, "grad_norm": 2.3085615634918213, "learning_rate": 0.0001, "loss": 0.0613, "step": 8890 }, { "epoch": 56.68789808917197, "grad_norm": 2.4557697772979736, "learning_rate": 0.0001, "loss": 0.0571, "step": 8900 }, { "epoch": 56.7515923566879, "grad_norm": 1.9624885320663452, "learning_rate": 0.0001, "loss": 0.0608, "step": 8910 }, { "epoch": 56.81528662420382, "grad_norm": 1.7230887413024902, "learning_rate": 0.0001, "loss": 0.0551, "step": 8920 }, { "epoch": 56.87898089171974, "grad_norm": 2.078625202178955, "learning_rate": 0.0001, "loss": 0.0556, "step": 8930 }, { "epoch": 56.94267515923567, "grad_norm": 2.202491283416748, "learning_rate": 0.0001, "loss": 0.0583, "step": 8940 }, { "epoch": 57.00636942675159, "grad_norm": 1.8416975736618042, "learning_rate": 0.0001, "loss": 0.0555, "step": 8950 }, { "epoch": 57.07006369426752, "grad_norm": 2.0693910121917725, "learning_rate": 0.0001, "loss": 0.0611, "step": 8960 }, { "epoch": 57.13375796178344, "grad_norm": 1.9069418907165527, "learning_rate": 0.0001, "loss": 0.0534, "step": 8970 }, { "epoch": 57.197452229299365, "grad_norm": 1.973203182220459, "learning_rate": 0.0001, "loss": 0.0555, "step": 8980 }, { "epoch": 57.261146496815286, "grad_norm": 2.43104887008667, "learning_rate": 0.0001, "loss": 0.0604, "step": 8990 }, { "epoch": 57.32484076433121, "grad_norm": 2.038259506225586, "learning_rate": 0.0001, "loss": 0.056, "step": 9000 }, { "epoch": 57.388535031847134, "grad_norm": 2.2269093990325928, "learning_rate": 0.0001, "loss": 0.0574, "step": 9010 }, { "epoch": 57.452229299363054, "grad_norm": 1.7998794317245483, "learning_rate": 0.0001, "loss": 0.0566, "step": 9020 }, { "epoch": 57.51592356687898, "grad_norm": 2.072557210922241, "learning_rate": 0.0001, "loss": 0.0617, "step": 9030 }, { "epoch": 57.5796178343949, "grad_norm": 1.7840062379837036, "learning_rate": 0.0001, "loss": 0.0603, "step": 9040 }, { "epoch": 57.64331210191083, "grad_norm": 1.8380111455917358, "learning_rate": 0.0001, "loss": 0.0562, "step": 9050 }, { "epoch": 57.70700636942675, "grad_norm": 1.8964613676071167, "learning_rate": 0.0001, "loss": 0.0621, "step": 9060 }, { "epoch": 57.77070063694268, "grad_norm": 1.9906922578811646, "learning_rate": 0.0001, "loss": 0.06, "step": 9070 }, { "epoch": 57.8343949044586, "grad_norm": 2.1781437397003174, "learning_rate": 0.0001, "loss": 0.063, "step": 9080 }, { "epoch": 57.89808917197452, "grad_norm": 2.452876567840576, "learning_rate": 0.0001, "loss": 0.0579, "step": 9090 }, { "epoch": 57.961783439490446, "grad_norm": 2.2490134239196777, "learning_rate": 0.0001, "loss": 0.059, "step": 9100 }, { "epoch": 58.02547770700637, "grad_norm": 2.214207649230957, "learning_rate": 0.0001, "loss": 0.06, "step": 9110 }, { "epoch": 58.089171974522294, "grad_norm": 2.2149953842163086, "learning_rate": 0.0001, "loss": 0.0592, "step": 9120 }, { "epoch": 58.152866242038215, "grad_norm": 2.9927570819854736, "learning_rate": 0.0001, "loss": 0.0581, "step": 9130 }, { "epoch": 58.21656050955414, "grad_norm": 2.428122043609619, "learning_rate": 0.0001, "loss": 0.0567, "step": 9140 }, { "epoch": 58.28025477707006, "grad_norm": 2.457392692565918, "learning_rate": 0.0001, "loss": 0.0555, "step": 9150 }, { "epoch": 58.34394904458599, "grad_norm": 2.481559991836548, "learning_rate": 0.0001, "loss": 0.0564, "step": 9160 }, { "epoch": 58.40764331210191, "grad_norm": 2.1260337829589844, "learning_rate": 0.0001, "loss": 0.0539, "step": 9170 }, { "epoch": 58.47133757961783, "grad_norm": 2.4016799926757812, "learning_rate": 0.0001, "loss": 0.0581, "step": 9180 }, { "epoch": 58.53503184713376, "grad_norm": 2.2390074729919434, "learning_rate": 0.0001, "loss": 0.0542, "step": 9190 }, { "epoch": 58.59872611464968, "grad_norm": 2.006378412246704, "learning_rate": 0.0001, "loss": 0.058, "step": 9200 }, { "epoch": 58.66242038216561, "grad_norm": 2.3712291717529297, "learning_rate": 0.0001, "loss": 0.0585, "step": 9210 }, { "epoch": 58.72611464968153, "grad_norm": 2.0532424449920654, "learning_rate": 0.0001, "loss": 0.0561, "step": 9220 }, { "epoch": 58.789808917197455, "grad_norm": 2.0927493572235107, "learning_rate": 0.0001, "loss": 0.0577, "step": 9230 }, { "epoch": 58.853503184713375, "grad_norm": 2.329864740371704, "learning_rate": 0.0001, "loss": 0.0589, "step": 9240 }, { "epoch": 58.9171974522293, "grad_norm": 2.2195420265197754, "learning_rate": 0.0001, "loss": 0.0629, "step": 9250 }, { "epoch": 58.98089171974522, "grad_norm": 2.6388068199157715, "learning_rate": 0.0001, "loss": 0.0572, "step": 9260 }, { "epoch": 59.044585987261144, "grad_norm": 2.0936646461486816, "learning_rate": 0.0001, "loss": 0.0584, "step": 9270 }, { "epoch": 59.10828025477707, "grad_norm": 2.2923836708068848, "learning_rate": 0.0001, "loss": 0.0559, "step": 9280 }, { "epoch": 59.17197452229299, "grad_norm": 2.293351411819458, "learning_rate": 0.0001, "loss": 0.0584, "step": 9290 }, { "epoch": 59.23566878980892, "grad_norm": 2.1421825885772705, "learning_rate": 0.0001, "loss": 0.058, "step": 9300 }, { "epoch": 59.29936305732484, "grad_norm": 2.0888516902923584, "learning_rate": 0.0001, "loss": 0.0544, "step": 9310 }, { "epoch": 59.36305732484077, "grad_norm": 2.109684467315674, "learning_rate": 0.0001, "loss": 0.0576, "step": 9320 }, { "epoch": 59.42675159235669, "grad_norm": 2.25069260597229, "learning_rate": 0.0001, "loss": 0.054, "step": 9330 }, { "epoch": 59.49044585987261, "grad_norm": 2.0621745586395264, "learning_rate": 0.0001, "loss": 0.0548, "step": 9340 }, { "epoch": 59.554140127388536, "grad_norm": 1.9061620235443115, "learning_rate": 0.0001, "loss": 0.0604, "step": 9350 }, { "epoch": 59.617834394904456, "grad_norm": 2.080768585205078, "learning_rate": 0.0001, "loss": 0.0561, "step": 9360 }, { "epoch": 59.681528662420384, "grad_norm": 1.7822362184524536, "learning_rate": 0.0001, "loss": 0.0561, "step": 9370 }, { "epoch": 59.745222929936304, "grad_norm": 1.9460077285766602, "learning_rate": 0.0001, "loss": 0.0607, "step": 9380 }, { "epoch": 59.80891719745223, "grad_norm": 2.1300525665283203, "learning_rate": 0.0001, "loss": 0.0579, "step": 9390 }, { "epoch": 59.87261146496815, "grad_norm": 2.4152767658233643, "learning_rate": 0.0001, "loss": 0.0583, "step": 9400 }, { "epoch": 59.93630573248408, "grad_norm": 1.9109939336776733, "learning_rate": 0.0001, "loss": 0.0587, "step": 9410 }, { "epoch": 60.0, "grad_norm": 1.7983086109161377, "learning_rate": 0.0001, "loss": 0.0572, "step": 9420 }, { "epoch": 60.06369426751592, "grad_norm": 1.673050880432129, "learning_rate": 0.0001, "loss": 0.0587, "step": 9430 }, { "epoch": 60.12738853503185, "grad_norm": 1.597550392150879, "learning_rate": 0.0001, "loss": 0.0564, "step": 9440 }, { "epoch": 60.19108280254777, "grad_norm": 1.7368433475494385, "learning_rate": 0.0001, "loss": 0.0596, "step": 9450 }, { "epoch": 60.254777070063696, "grad_norm": 2.1020681858062744, "learning_rate": 0.0001, "loss": 0.0562, "step": 9460 }, { "epoch": 60.318471337579616, "grad_norm": 2.0337889194488525, "learning_rate": 0.0001, "loss": 0.0597, "step": 9470 }, { "epoch": 60.382165605095544, "grad_norm": 1.8028478622436523, "learning_rate": 0.0001, "loss": 0.0584, "step": 9480 }, { "epoch": 60.445859872611464, "grad_norm": 2.115612268447876, "learning_rate": 0.0001, "loss": 0.0572, "step": 9490 }, { "epoch": 60.50955414012739, "grad_norm": 2.2593164443969727, "learning_rate": 0.0001, "loss": 0.0551, "step": 9500 }, { "epoch": 60.57324840764331, "grad_norm": 1.9045567512512207, "learning_rate": 0.0001, "loss": 0.0601, "step": 9510 }, { "epoch": 60.63694267515923, "grad_norm": 2.0510497093200684, "learning_rate": 0.0001, "loss": 0.0584, "step": 9520 }, { "epoch": 60.70063694267516, "grad_norm": 2.3130340576171875, "learning_rate": 0.0001, "loss": 0.058, "step": 9530 }, { "epoch": 60.76433121019108, "grad_norm": 1.8961944580078125, "learning_rate": 0.0001, "loss": 0.0573, "step": 9540 }, { "epoch": 60.82802547770701, "grad_norm": 1.9158601760864258, "learning_rate": 0.0001, "loss": 0.0558, "step": 9550 }, { "epoch": 60.89171974522293, "grad_norm": 1.656625747680664, "learning_rate": 0.0001, "loss": 0.0596, "step": 9560 }, { "epoch": 60.955414012738856, "grad_norm": 1.9255701303482056, "learning_rate": 0.0001, "loss": 0.062, "step": 9570 }, { "epoch": 61.01910828025478, "grad_norm": 1.5261919498443604, "learning_rate": 0.0001, "loss": 0.0576, "step": 9580 }, { "epoch": 61.0828025477707, "grad_norm": 1.8539847135543823, "learning_rate": 0.0001, "loss": 0.0571, "step": 9590 }, { "epoch": 61.146496815286625, "grad_norm": 2.016038417816162, "learning_rate": 0.0001, "loss": 0.0582, "step": 9600 }, { "epoch": 61.210191082802545, "grad_norm": 1.8645352125167847, "learning_rate": 0.0001, "loss": 0.0582, "step": 9610 }, { "epoch": 61.27388535031847, "grad_norm": 1.950600266456604, "learning_rate": 0.0001, "loss": 0.0528, "step": 9620 }, { "epoch": 61.33757961783439, "grad_norm": 2.0023508071899414, "learning_rate": 0.0001, "loss": 0.0607, "step": 9630 }, { "epoch": 61.40127388535032, "grad_norm": 2.1755897998809814, "learning_rate": 0.0001, "loss": 0.0605, "step": 9640 }, { "epoch": 61.46496815286624, "grad_norm": 2.2576301097869873, "learning_rate": 0.0001, "loss": 0.0594, "step": 9650 }, { "epoch": 61.52866242038217, "grad_norm": 2.063849449157715, "learning_rate": 0.0001, "loss": 0.0586, "step": 9660 }, { "epoch": 61.59235668789809, "grad_norm": 1.6141220331192017, "learning_rate": 0.0001, "loss": 0.0533, "step": 9670 }, { "epoch": 61.65605095541401, "grad_norm": 1.9523097276687622, "learning_rate": 0.0001, "loss": 0.0577, "step": 9680 }, { "epoch": 61.71974522292994, "grad_norm": 1.8736836910247803, "learning_rate": 0.0001, "loss": 0.0552, "step": 9690 }, { "epoch": 61.78343949044586, "grad_norm": 1.5462660789489746, "learning_rate": 0.0001, "loss": 0.0569, "step": 9700 }, { "epoch": 61.847133757961785, "grad_norm": 1.870153784751892, "learning_rate": 0.0001, "loss": 0.0581, "step": 9710 }, { "epoch": 61.910828025477706, "grad_norm": 1.819495439529419, "learning_rate": 0.0001, "loss": 0.0589, "step": 9720 }, { "epoch": 61.97452229299363, "grad_norm": 1.764788269996643, "learning_rate": 0.0001, "loss": 0.0546, "step": 9730 }, { "epoch": 62.038216560509554, "grad_norm": 1.748396396636963, "learning_rate": 0.0001, "loss": 0.0589, "step": 9740 }, { "epoch": 62.10191082802548, "grad_norm": 2.630002737045288, "learning_rate": 0.0001, "loss": 0.0565, "step": 9750 }, { "epoch": 62.1656050955414, "grad_norm": 2.240643262863159, "learning_rate": 0.0001, "loss": 0.0592, "step": 9760 }, { "epoch": 62.22929936305732, "grad_norm": 2.583648204803467, "learning_rate": 0.0001, "loss": 0.0572, "step": 9770 }, { "epoch": 62.29299363057325, "grad_norm": 2.202775239944458, "learning_rate": 0.0001, "loss": 0.0589, "step": 9780 }, { "epoch": 62.35668789808917, "grad_norm": 2.0077316761016846, "learning_rate": 0.0001, "loss": 0.0576, "step": 9790 }, { "epoch": 62.4203821656051, "grad_norm": 2.2732698917388916, "learning_rate": 0.0001, "loss": 0.0583, "step": 9800 }, { "epoch": 62.48407643312102, "grad_norm": 1.7405937910079956, "learning_rate": 0.0001, "loss": 0.0544, "step": 9810 }, { "epoch": 62.547770700636946, "grad_norm": 1.9091839790344238, "learning_rate": 0.0001, "loss": 0.0547, "step": 9820 }, { "epoch": 62.611464968152866, "grad_norm": 1.9828511476516724, "learning_rate": 0.0001, "loss": 0.0568, "step": 9830 }, { "epoch": 62.67515923566879, "grad_norm": 1.9150711297988892, "learning_rate": 0.0001, "loss": 0.0605, "step": 9840 }, { "epoch": 62.738853503184714, "grad_norm": 1.7578699588775635, "learning_rate": 0.0001, "loss": 0.0583, "step": 9850 }, { "epoch": 62.802547770700635, "grad_norm": 1.9258337020874023, "learning_rate": 0.0001, "loss": 0.0549, "step": 9860 }, { "epoch": 62.86624203821656, "grad_norm": 1.8123828172683716, "learning_rate": 0.0001, "loss": 0.0513, "step": 9870 }, { "epoch": 62.92993630573248, "grad_norm": 2.0562644004821777, "learning_rate": 0.0001, "loss": 0.0566, "step": 9880 }, { "epoch": 62.99363057324841, "grad_norm": 1.8881092071533203, "learning_rate": 0.0001, "loss": 0.0562, "step": 9890 }, { "epoch": 63.05732484076433, "grad_norm": 1.926159381866455, "learning_rate": 0.0001, "loss": 0.0562, "step": 9900 }, { "epoch": 63.12101910828026, "grad_norm": 1.9985334873199463, "learning_rate": 0.0001, "loss": 0.0589, "step": 9910 }, { "epoch": 63.18471337579618, "grad_norm": 2.3501367568969727, "learning_rate": 0.0001, "loss": 0.0578, "step": 9920 }, { "epoch": 63.2484076433121, "grad_norm": 2.0403571128845215, "learning_rate": 0.0001, "loss": 0.0583, "step": 9930 }, { "epoch": 63.31210191082803, "grad_norm": 1.9714267253875732, "learning_rate": 0.0001, "loss": 0.0569, "step": 9940 }, { "epoch": 63.37579617834395, "grad_norm": 2.1726670265197754, "learning_rate": 0.0001, "loss": 0.0607, "step": 9950 }, { "epoch": 63.439490445859875, "grad_norm": 1.929787516593933, "learning_rate": 0.0001, "loss": 0.0598, "step": 9960 }, { "epoch": 63.503184713375795, "grad_norm": 1.7903579473495483, "learning_rate": 0.0001, "loss": 0.0552, "step": 9970 }, { "epoch": 63.56687898089172, "grad_norm": 1.8409336805343628, "learning_rate": 0.0001, "loss": 0.0527, "step": 9980 }, { "epoch": 63.63057324840764, "grad_norm": 1.6869924068450928, "learning_rate": 0.0001, "loss": 0.0551, "step": 9990 }, { "epoch": 63.69426751592356, "grad_norm": 2.709916353225708, "learning_rate": 0.0001, "loss": 0.0589, "step": 10000 }, { "epoch": 63.75796178343949, "grad_norm": 2.3819057941436768, "learning_rate": 0.0001, "loss": 0.0632, "step": 10010 }, { "epoch": 63.82165605095541, "grad_norm": 1.6012017726898193, "learning_rate": 0.0001, "loss": 0.0566, "step": 10020 }, { "epoch": 63.88535031847134, "grad_norm": 1.662710189819336, "learning_rate": 0.0001, "loss": 0.0613, "step": 10030 }, { "epoch": 63.94904458598726, "grad_norm": 1.9464925527572632, "learning_rate": 0.0001, "loss": 0.0598, "step": 10040 }, { "epoch": 64.01273885350318, "grad_norm": 1.818298101425171, "learning_rate": 0.0001, "loss": 0.0511, "step": 10050 }, { "epoch": 64.07643312101911, "grad_norm": 1.7114503383636475, "learning_rate": 0.0001, "loss": 0.0548, "step": 10060 }, { "epoch": 64.14012738853503, "grad_norm": 1.6352883577346802, "learning_rate": 0.0001, "loss": 0.057, "step": 10070 }, { "epoch": 64.20382165605096, "grad_norm": 1.7248188257217407, "learning_rate": 0.0001, "loss": 0.0546, "step": 10080 }, { "epoch": 64.26751592356688, "grad_norm": 2.372760534286499, "learning_rate": 0.0001, "loss": 0.0598, "step": 10090 }, { "epoch": 64.3312101910828, "grad_norm": 1.7333749532699585, "learning_rate": 0.0001, "loss": 0.0551, "step": 10100 }, { "epoch": 64.39490445859873, "grad_norm": 2.2532756328582764, "learning_rate": 0.0001, "loss": 0.0552, "step": 10110 }, { "epoch": 64.45859872611464, "grad_norm": 1.950847864151001, "learning_rate": 0.0001, "loss": 0.0566, "step": 10120 }, { "epoch": 64.52229299363057, "grad_norm": 2.150973081588745, "learning_rate": 0.0001, "loss": 0.0576, "step": 10130 }, { "epoch": 64.5859872611465, "grad_norm": 2.175238609313965, "learning_rate": 0.0001, "loss": 0.0563, "step": 10140 }, { "epoch": 64.64968152866243, "grad_norm": 2.498021125793457, "learning_rate": 0.0001, "loss": 0.055, "step": 10150 }, { "epoch": 64.71337579617834, "grad_norm": 2.1340434551239014, "learning_rate": 0.0001, "loss": 0.0581, "step": 10160 }, { "epoch": 64.77707006369427, "grad_norm": 1.9452824592590332, "learning_rate": 0.0001, "loss": 0.0539, "step": 10170 }, { "epoch": 64.8407643312102, "grad_norm": 1.9772199392318726, "learning_rate": 0.0001, "loss": 0.0526, "step": 10180 }, { "epoch": 64.90445859872611, "grad_norm": 1.9267945289611816, "learning_rate": 0.0001, "loss": 0.0522, "step": 10190 }, { "epoch": 64.96815286624204, "grad_norm": 1.8902651071548462, "learning_rate": 0.0001, "loss": 0.0563, "step": 10200 }, { "epoch": 65.03184713375796, "grad_norm": 2.097879409790039, "learning_rate": 0.0001, "loss": 0.0573, "step": 10210 }, { "epoch": 65.09554140127389, "grad_norm": 2.126025915145874, "learning_rate": 0.0001, "loss": 0.0501, "step": 10220 }, { "epoch": 65.1592356687898, "grad_norm": 2.125105619430542, "learning_rate": 0.0001, "loss": 0.0547, "step": 10230 }, { "epoch": 65.22292993630573, "grad_norm": 2.3525772094726562, "learning_rate": 0.0001, "loss": 0.056, "step": 10240 }, { "epoch": 65.28662420382166, "grad_norm": 2.193100690841675, "learning_rate": 0.0001, "loss": 0.0523, "step": 10250 }, { "epoch": 65.35031847133757, "grad_norm": 2.2143428325653076, "learning_rate": 0.0001, "loss": 0.0556, "step": 10260 }, { "epoch": 65.4140127388535, "grad_norm": 1.742448091506958, "learning_rate": 0.0001, "loss": 0.0511, "step": 10270 }, { "epoch": 65.47770700636943, "grad_norm": 2.139549970626831, "learning_rate": 0.0001, "loss": 0.0555, "step": 10280 }, { "epoch": 65.54140127388536, "grad_norm": 1.8370410203933716, "learning_rate": 0.0001, "loss": 0.0515, "step": 10290 }, { "epoch": 65.60509554140127, "grad_norm": 2.25881028175354, "learning_rate": 0.0001, "loss": 0.0518, "step": 10300 }, { "epoch": 65.6687898089172, "grad_norm": 2.3993685245513916, "learning_rate": 0.0001, "loss": 0.0556, "step": 10310 }, { "epoch": 65.73248407643312, "grad_norm": 2.015014410018921, "learning_rate": 0.0001, "loss": 0.0483, "step": 10320 }, { "epoch": 65.79617834394904, "grad_norm": 1.829719066619873, "learning_rate": 0.0001, "loss": 0.0583, "step": 10330 }, { "epoch": 65.85987261146497, "grad_norm": 1.76730215549469, "learning_rate": 0.0001, "loss": 0.056, "step": 10340 }, { "epoch": 65.92356687898089, "grad_norm": 2.12646222114563, "learning_rate": 0.0001, "loss": 0.055, "step": 10350 }, { "epoch": 65.98726114649682, "grad_norm": 1.835974931716919, "learning_rate": 0.0001, "loss": 0.0579, "step": 10360 }, { "epoch": 66.05095541401273, "grad_norm": 1.7636247873306274, "learning_rate": 0.0001, "loss": 0.0543, "step": 10370 }, { "epoch": 66.11464968152866, "grad_norm": 1.8226345777511597, "learning_rate": 0.0001, "loss": 0.0518, "step": 10380 }, { "epoch": 66.17834394904459, "grad_norm": 1.9204694032669067, "learning_rate": 0.0001, "loss": 0.0519, "step": 10390 }, { "epoch": 66.24203821656052, "grad_norm": 1.9366486072540283, "learning_rate": 0.0001, "loss": 0.0538, "step": 10400 }, { "epoch": 66.30573248407643, "grad_norm": 1.8173679113388062, "learning_rate": 0.0001, "loss": 0.0574, "step": 10410 }, { "epoch": 66.36942675159236, "grad_norm": 2.136437177658081, "learning_rate": 0.0001, "loss": 0.0584, "step": 10420 }, { "epoch": 66.43312101910828, "grad_norm": 1.821769118309021, "learning_rate": 0.0001, "loss": 0.056, "step": 10430 }, { "epoch": 66.4968152866242, "grad_norm": 1.7254918813705444, "learning_rate": 0.0001, "loss": 0.0579, "step": 10440 }, { "epoch": 66.56050955414013, "grad_norm": 2.0237646102905273, "learning_rate": 0.0001, "loss": 0.0572, "step": 10450 }, { "epoch": 66.62420382165605, "grad_norm": 1.91119384765625, "learning_rate": 0.0001, "loss": 0.0659, "step": 10460 }, { "epoch": 66.68789808917198, "grad_norm": 1.848382592201233, "learning_rate": 0.0001, "loss": 0.0571, "step": 10470 }, { "epoch": 66.7515923566879, "grad_norm": 2.2833211421966553, "learning_rate": 0.0001, "loss": 0.0651, "step": 10480 }, { "epoch": 66.81528662420382, "grad_norm": 2.1656978130340576, "learning_rate": 0.0001, "loss": 0.0599, "step": 10490 }, { "epoch": 66.87898089171975, "grad_norm": 1.662873387336731, "learning_rate": 0.0001, "loss": 0.058, "step": 10500 }, { "epoch": 66.94267515923566, "grad_norm": 1.888155221939087, "learning_rate": 0.0001, "loss": 0.0562, "step": 10510 }, { "epoch": 67.00636942675159, "grad_norm": 2.2426154613494873, "learning_rate": 0.0001, "loss": 0.0593, "step": 10520 }, { "epoch": 67.07006369426752, "grad_norm": 2.199026107788086, "learning_rate": 0.0001, "loss": 0.0553, "step": 10530 }, { "epoch": 67.13375796178345, "grad_norm": 1.9341243505477905, "learning_rate": 0.0001, "loss": 0.0552, "step": 10540 }, { "epoch": 67.19745222929936, "grad_norm": 2.0347983837127686, "learning_rate": 0.0001, "loss": 0.0546, "step": 10550 }, { "epoch": 67.26114649681529, "grad_norm": 1.8350305557250977, "learning_rate": 0.0001, "loss": 0.0555, "step": 10560 }, { "epoch": 67.32484076433121, "grad_norm": 1.7520166635513306, "learning_rate": 0.0001, "loss": 0.0581, "step": 10570 }, { "epoch": 67.38853503184713, "grad_norm": 1.7529175281524658, "learning_rate": 0.0001, "loss": 0.0562, "step": 10580 }, { "epoch": 67.45222929936305, "grad_norm": 1.8133114576339722, "learning_rate": 0.0001, "loss": 0.0575, "step": 10590 }, { "epoch": 67.51592356687898, "grad_norm": 1.6690889596939087, "learning_rate": 0.0001, "loss": 0.0573, "step": 10600 }, { "epoch": 67.57961783439491, "grad_norm": 1.5533227920532227, "learning_rate": 0.0001, "loss": 0.0555, "step": 10610 }, { "epoch": 67.64331210191082, "grad_norm": 1.9376810789108276, "learning_rate": 0.0001, "loss": 0.0564, "step": 10620 }, { "epoch": 67.70700636942675, "grad_norm": 1.9370397329330444, "learning_rate": 0.0001, "loss": 0.0541, "step": 10630 }, { "epoch": 67.77070063694268, "grad_norm": 2.0784623622894287, "learning_rate": 0.0001, "loss": 0.0548, "step": 10640 }, { "epoch": 67.8343949044586, "grad_norm": 2.1036009788513184, "learning_rate": 0.0001, "loss": 0.0539, "step": 10650 }, { "epoch": 67.89808917197452, "grad_norm": 1.6862165927886963, "learning_rate": 0.0001, "loss": 0.0533, "step": 10660 }, { "epoch": 67.96178343949045, "grad_norm": 1.9237582683563232, "learning_rate": 0.0001, "loss": 0.0569, "step": 10670 }, { "epoch": 68.02547770700637, "grad_norm": 1.7960420846939087, "learning_rate": 0.0001, "loss": 0.0484, "step": 10680 }, { "epoch": 68.08917197452229, "grad_norm": 1.6653668880462646, "learning_rate": 0.0001, "loss": 0.0508, "step": 10690 }, { "epoch": 68.15286624203821, "grad_norm": 1.9786585569381714, "learning_rate": 0.0001, "loss": 0.0516, "step": 10700 }, { "epoch": 68.21656050955414, "grad_norm": 2.0557777881622314, "learning_rate": 0.0001, "loss": 0.0511, "step": 10710 }, { "epoch": 68.28025477707007, "grad_norm": 2.0634148120880127, "learning_rate": 0.0001, "loss": 0.0528, "step": 10720 }, { "epoch": 68.34394904458598, "grad_norm": 1.8734732866287231, "learning_rate": 0.0001, "loss": 0.0546, "step": 10730 }, { "epoch": 68.40764331210191, "grad_norm": 1.9182634353637695, "learning_rate": 0.0001, "loss": 0.0534, "step": 10740 }, { "epoch": 68.47133757961784, "grad_norm": 1.7177423238754272, "learning_rate": 0.0001, "loss": 0.0513, "step": 10750 }, { "epoch": 68.53503184713375, "grad_norm": 1.9399218559265137, "learning_rate": 0.0001, "loss": 0.0515, "step": 10760 }, { "epoch": 68.59872611464968, "grad_norm": 1.790362000465393, "learning_rate": 0.0001, "loss": 0.0554, "step": 10770 }, { "epoch": 68.6624203821656, "grad_norm": 2.174116611480713, "learning_rate": 0.0001, "loss": 0.0623, "step": 10780 }, { "epoch": 68.72611464968153, "grad_norm": 1.83573317527771, "learning_rate": 0.0001, "loss": 0.0556, "step": 10790 }, { "epoch": 68.78980891719745, "grad_norm": 2.03324031829834, "learning_rate": 0.0001, "loss": 0.0559, "step": 10800 }, { "epoch": 68.85350318471338, "grad_norm": 1.6938214302062988, "learning_rate": 0.0001, "loss": 0.0525, "step": 10810 }, { "epoch": 68.9171974522293, "grad_norm": 1.4340629577636719, "learning_rate": 0.0001, "loss": 0.0546, "step": 10820 }, { "epoch": 68.98089171974522, "grad_norm": 1.5692317485809326, "learning_rate": 0.0001, "loss": 0.0502, "step": 10830 }, { "epoch": 69.04458598726114, "grad_norm": 1.7593803405761719, "learning_rate": 0.0001, "loss": 0.0498, "step": 10840 }, { "epoch": 69.10828025477707, "grad_norm": 1.7776451110839844, "learning_rate": 0.0001, "loss": 0.05, "step": 10850 }, { "epoch": 69.171974522293, "grad_norm": 1.7624714374542236, "learning_rate": 0.0001, "loss": 0.0534, "step": 10860 }, { "epoch": 69.23566878980891, "grad_norm": 2.422492504119873, "learning_rate": 0.0001, "loss": 0.0519, "step": 10870 }, { "epoch": 69.29936305732484, "grad_norm": 1.920953631401062, "learning_rate": 0.0001, "loss": 0.0546, "step": 10880 }, { "epoch": 69.36305732484077, "grad_norm": 1.9107297658920288, "learning_rate": 0.0001, "loss": 0.0523, "step": 10890 }, { "epoch": 69.4267515923567, "grad_norm": 2.225682258605957, "learning_rate": 0.0001, "loss": 0.0518, "step": 10900 }, { "epoch": 69.49044585987261, "grad_norm": 2.223067045211792, "learning_rate": 0.0001, "loss": 0.0547, "step": 10910 }, { "epoch": 69.55414012738854, "grad_norm": 2.0467538833618164, "learning_rate": 0.0001, "loss": 0.0498, "step": 10920 }, { "epoch": 69.61783439490446, "grad_norm": 1.7044579982757568, "learning_rate": 0.0001, "loss": 0.0528, "step": 10930 }, { "epoch": 69.68152866242038, "grad_norm": 2.01537823677063, "learning_rate": 0.0001, "loss": 0.0522, "step": 10940 }, { "epoch": 69.7452229299363, "grad_norm": 1.985940933227539, "learning_rate": 0.0001, "loss": 0.0535, "step": 10950 }, { "epoch": 69.80891719745223, "grad_norm": 2.088958978652954, "learning_rate": 0.0001, "loss": 0.0539, "step": 10960 }, { "epoch": 69.87261146496816, "grad_norm": 1.6607328653335571, "learning_rate": 0.0001, "loss": 0.0488, "step": 10970 }, { "epoch": 69.93630573248407, "grad_norm": 1.853441834449768, "learning_rate": 0.0001, "loss": 0.052, "step": 10980 }, { "epoch": 70.0, "grad_norm": 1.6448442935943604, "learning_rate": 0.0001, "loss": 0.0501, "step": 10990 }, { "epoch": 70.06369426751593, "grad_norm": 1.677259087562561, "learning_rate": 0.0001, "loss": 0.0489, "step": 11000 }, { "epoch": 70.12738853503184, "grad_norm": 1.6033686399459839, "learning_rate": 0.0001, "loss": 0.0491, "step": 11010 }, { "epoch": 70.19108280254777, "grad_norm": 1.494301199913025, "learning_rate": 0.0001, "loss": 0.0523, "step": 11020 }, { "epoch": 70.2547770700637, "grad_norm": 2.003443717956543, "learning_rate": 0.0001, "loss": 0.0526, "step": 11030 }, { "epoch": 70.31847133757962, "grad_norm": 1.5664259195327759, "learning_rate": 0.0001, "loss": 0.0541, "step": 11040 }, { "epoch": 70.38216560509554, "grad_norm": 1.6851779222488403, "learning_rate": 0.0001, "loss": 0.0547, "step": 11050 }, { "epoch": 70.44585987261146, "grad_norm": 1.528714656829834, "learning_rate": 0.0001, "loss": 0.0498, "step": 11060 }, { "epoch": 70.50955414012739, "grad_norm": 1.6736868619918823, "learning_rate": 0.0001, "loss": 0.0507, "step": 11070 }, { "epoch": 70.5732484076433, "grad_norm": 2.3379549980163574, "learning_rate": 0.0001, "loss": 0.0544, "step": 11080 }, { "epoch": 70.63694267515923, "grad_norm": 1.9576376676559448, "learning_rate": 0.0001, "loss": 0.0528, "step": 11090 }, { "epoch": 70.70063694267516, "grad_norm": 1.5554176568984985, "learning_rate": 0.0001, "loss": 0.0608, "step": 11100 }, { "epoch": 70.76433121019109, "grad_norm": 1.7643511295318604, "learning_rate": 0.0001, "loss": 0.0553, "step": 11110 }, { "epoch": 70.828025477707, "grad_norm": 1.9938466548919678, "learning_rate": 0.0001, "loss": 0.058, "step": 11120 }, { "epoch": 70.89171974522293, "grad_norm": 2.1049139499664307, "learning_rate": 0.0001, "loss": 0.0545, "step": 11130 }, { "epoch": 70.95541401273886, "grad_norm": 1.731493353843689, "learning_rate": 0.0001, "loss": 0.0499, "step": 11140 }, { "epoch": 71.01910828025478, "grad_norm": 2.2636356353759766, "learning_rate": 0.0001, "loss": 0.054, "step": 11150 }, { "epoch": 71.0828025477707, "grad_norm": 2.233635663986206, "learning_rate": 0.0001, "loss": 0.0583, "step": 11160 }, { "epoch": 71.14649681528662, "grad_norm": 2.0341622829437256, "learning_rate": 0.0001, "loss": 0.0517, "step": 11170 }, { "epoch": 71.21019108280255, "grad_norm": 2.0676679611206055, "learning_rate": 0.0001, "loss": 0.0501, "step": 11180 }, { "epoch": 71.27388535031847, "grad_norm": 1.5837899446487427, "learning_rate": 0.0001, "loss": 0.0508, "step": 11190 }, { "epoch": 71.3375796178344, "grad_norm": 2.148590564727783, "learning_rate": 0.0001, "loss": 0.0555, "step": 11200 }, { "epoch": 71.40127388535032, "grad_norm": 1.9741541147232056, "learning_rate": 0.0001, "loss": 0.0542, "step": 11210 }, { "epoch": 71.46496815286625, "grad_norm": 2.1217901706695557, "learning_rate": 0.0001, "loss": 0.0539, "step": 11220 }, { "epoch": 71.52866242038216, "grad_norm": 2.1403188705444336, "learning_rate": 0.0001, "loss": 0.0525, "step": 11230 }, { "epoch": 71.59235668789809, "grad_norm": 1.971739411354065, "learning_rate": 0.0001, "loss": 0.0539, "step": 11240 }, { "epoch": 71.65605095541402, "grad_norm": 1.8207457065582275, "learning_rate": 0.0001, "loss": 0.0526, "step": 11250 }, { "epoch": 71.71974522292993, "grad_norm": 1.7685319185256958, "learning_rate": 0.0001, "loss": 0.0507, "step": 11260 }, { "epoch": 71.78343949044586, "grad_norm": 1.8234423398971558, "learning_rate": 0.0001, "loss": 0.0557, "step": 11270 }, { "epoch": 71.84713375796179, "grad_norm": 1.6248924732208252, "learning_rate": 0.0001, "loss": 0.0525, "step": 11280 }, { "epoch": 71.91082802547771, "grad_norm": 1.759057641029358, "learning_rate": 0.0001, "loss": 0.0539, "step": 11290 }, { "epoch": 71.97452229299363, "grad_norm": 1.5748484134674072, "learning_rate": 0.0001, "loss": 0.0543, "step": 11300 }, { "epoch": 72.03821656050955, "grad_norm": 1.4182177782058716, "learning_rate": 0.0001, "loss": 0.0534, "step": 11310 }, { "epoch": 72.10191082802548, "grad_norm": 1.7235502004623413, "learning_rate": 0.0001, "loss": 0.053, "step": 11320 }, { "epoch": 72.1656050955414, "grad_norm": 1.6157959699630737, "learning_rate": 0.0001, "loss": 0.0588, "step": 11330 }, { "epoch": 72.22929936305732, "grad_norm": 1.837053656578064, "learning_rate": 0.0001, "loss": 0.0568, "step": 11340 }, { "epoch": 72.29299363057325, "grad_norm": 1.4150809049606323, "learning_rate": 0.0001, "loss": 0.0536, "step": 11350 }, { "epoch": 72.35668789808918, "grad_norm": 1.574402093887329, "learning_rate": 0.0001, "loss": 0.051, "step": 11360 }, { "epoch": 72.42038216560509, "grad_norm": 1.6626657247543335, "learning_rate": 0.0001, "loss": 0.0531, "step": 11370 }, { "epoch": 72.48407643312102, "grad_norm": 1.6227893829345703, "learning_rate": 0.0001, "loss": 0.0523, "step": 11380 }, { "epoch": 72.54777070063695, "grad_norm": 1.3444489240646362, "learning_rate": 0.0001, "loss": 0.053, "step": 11390 }, { "epoch": 72.61146496815287, "grad_norm": 1.6321101188659668, "learning_rate": 0.0001, "loss": 0.0554, "step": 11400 }, { "epoch": 72.67515923566879, "grad_norm": 1.6227716207504272, "learning_rate": 0.0001, "loss": 0.0539, "step": 11410 }, { "epoch": 72.73885350318471, "grad_norm": 1.5774481296539307, "learning_rate": 0.0001, "loss": 0.054, "step": 11420 }, { "epoch": 72.80254777070064, "grad_norm": 1.8820441961288452, "learning_rate": 0.0001, "loss": 0.0552, "step": 11430 }, { "epoch": 72.86624203821655, "grad_norm": 1.6420966386795044, "learning_rate": 0.0001, "loss": 0.0539, "step": 11440 }, { "epoch": 72.92993630573248, "grad_norm": 1.8904719352722168, "learning_rate": 0.0001, "loss": 0.0562, "step": 11450 }, { "epoch": 72.99363057324841, "grad_norm": 1.770211100578308, "learning_rate": 0.0001, "loss": 0.0533, "step": 11460 }, { "epoch": 73.05732484076434, "grad_norm": 1.6162742376327515, "learning_rate": 0.0001, "loss": 0.0538, "step": 11470 }, { "epoch": 73.12101910828025, "grad_norm": 1.8611490726470947, "learning_rate": 0.0001, "loss": 0.0501, "step": 11480 }, { "epoch": 73.18471337579618, "grad_norm": 1.686840295791626, "learning_rate": 0.0001, "loss": 0.0504, "step": 11490 }, { "epoch": 73.2484076433121, "grad_norm": 1.814237117767334, "learning_rate": 0.0001, "loss": 0.0509, "step": 11500 }, { "epoch": 73.31210191082802, "grad_norm": 1.9183100461959839, "learning_rate": 0.0001, "loss": 0.0537, "step": 11510 }, { "epoch": 73.37579617834395, "grad_norm": 1.5221400260925293, "learning_rate": 0.0001, "loss": 0.0534, "step": 11520 }, { "epoch": 73.43949044585987, "grad_norm": 1.4998258352279663, "learning_rate": 0.0001, "loss": 0.0513, "step": 11530 }, { "epoch": 73.5031847133758, "grad_norm": 1.7876991033554077, "learning_rate": 0.0001, "loss": 0.0566, "step": 11540 }, { "epoch": 73.56687898089172, "grad_norm": 1.6018035411834717, "learning_rate": 0.0001, "loss": 0.0559, "step": 11550 }, { "epoch": 73.63057324840764, "grad_norm": 1.8035871982574463, "learning_rate": 0.0001, "loss": 0.0566, "step": 11560 }, { "epoch": 73.69426751592357, "grad_norm": 1.6869609355926514, "learning_rate": 0.0001, "loss": 0.0535, "step": 11570 }, { "epoch": 73.75796178343948, "grad_norm": 1.661850929260254, "learning_rate": 0.0001, "loss": 0.0536, "step": 11580 }, { "epoch": 73.82165605095541, "grad_norm": 1.670170545578003, "learning_rate": 0.0001, "loss": 0.0554, "step": 11590 }, { "epoch": 73.88535031847134, "grad_norm": 2.0039496421813965, "learning_rate": 0.0001, "loss": 0.0555, "step": 11600 }, { "epoch": 73.94904458598727, "grad_norm": 1.8448761701583862, "learning_rate": 0.0001, "loss": 0.0487, "step": 11610 }, { "epoch": 74.01273885350318, "grad_norm": 1.552302360534668, "learning_rate": 0.0001, "loss": 0.0531, "step": 11620 }, { "epoch": 74.07643312101911, "grad_norm": 1.865772008895874, "learning_rate": 0.0001, "loss": 0.0514, "step": 11630 }, { "epoch": 74.14012738853503, "grad_norm": 1.871740460395813, "learning_rate": 0.0001, "loss": 0.048, "step": 11640 }, { "epoch": 74.20382165605096, "grad_norm": 1.6884204149246216, "learning_rate": 0.0001, "loss": 0.0523, "step": 11650 }, { "epoch": 74.26751592356688, "grad_norm": 1.8703057765960693, "learning_rate": 0.0001, "loss": 0.0538, "step": 11660 }, { "epoch": 74.3312101910828, "grad_norm": 1.479590654373169, "learning_rate": 0.0001, "loss": 0.0537, "step": 11670 }, { "epoch": 74.39490445859873, "grad_norm": 1.6267019510269165, "learning_rate": 0.0001, "loss": 0.0521, "step": 11680 }, { "epoch": 74.45859872611464, "grad_norm": 1.6558144092559814, "learning_rate": 0.0001, "loss": 0.0524, "step": 11690 }, { "epoch": 74.52229299363057, "grad_norm": 1.8338431119918823, "learning_rate": 0.0001, "loss": 0.0506, "step": 11700 }, { "epoch": 74.5859872611465, "grad_norm": 1.569459319114685, "learning_rate": 0.0001, "loss": 0.0586, "step": 11710 }, { "epoch": 74.64968152866243, "grad_norm": 1.697645664215088, "learning_rate": 0.0001, "loss": 0.0561, "step": 11720 }, { "epoch": 74.71337579617834, "grad_norm": 1.8653714656829834, "learning_rate": 0.0001, "loss": 0.0565, "step": 11730 }, { "epoch": 74.77707006369427, "grad_norm": 1.9133257865905762, "learning_rate": 0.0001, "loss": 0.0549, "step": 11740 }, { "epoch": 74.8407643312102, "grad_norm": 1.8773844242095947, "learning_rate": 0.0001, "loss": 0.056, "step": 11750 }, { "epoch": 74.90445859872611, "grad_norm": 1.7885699272155762, "learning_rate": 0.0001, "loss": 0.0481, "step": 11760 }, { "epoch": 74.96815286624204, "grad_norm": 1.5858768224716187, "learning_rate": 0.0001, "loss": 0.0528, "step": 11770 }, { "epoch": 75.03184713375796, "grad_norm": 1.9953739643096924, "learning_rate": 0.0001, "loss": 0.0556, "step": 11780 }, { "epoch": 75.09554140127389, "grad_norm": 1.894490361213684, "learning_rate": 0.0001, "loss": 0.0534, "step": 11790 }, { "epoch": 75.1592356687898, "grad_norm": 1.656960368156433, "learning_rate": 0.0001, "loss": 0.0515, "step": 11800 }, { "epoch": 75.22292993630573, "grad_norm": 1.703127384185791, "learning_rate": 0.0001, "loss": 0.0526, "step": 11810 }, { "epoch": 75.28662420382166, "grad_norm": 1.873714566230774, "learning_rate": 0.0001, "loss": 0.0507, "step": 11820 }, { "epoch": 75.35031847133757, "grad_norm": 2.142831325531006, "learning_rate": 0.0001, "loss": 0.0525, "step": 11830 }, { "epoch": 75.4140127388535, "grad_norm": 1.9689327478408813, "learning_rate": 0.0001, "loss": 0.0527, "step": 11840 }, { "epoch": 75.47770700636943, "grad_norm": 1.6483488082885742, "learning_rate": 0.0001, "loss": 0.0519, "step": 11850 }, { "epoch": 75.54140127388536, "grad_norm": 2.241898536682129, "learning_rate": 0.0001, "loss": 0.0552, "step": 11860 }, { "epoch": 75.60509554140127, "grad_norm": 1.9479236602783203, "learning_rate": 0.0001, "loss": 0.055, "step": 11870 }, { "epoch": 75.6687898089172, "grad_norm": 2.2032084465026855, "learning_rate": 0.0001, "loss": 0.0508, "step": 11880 }, { "epoch": 75.73248407643312, "grad_norm": 2.004916191101074, "learning_rate": 0.0001, "loss": 0.0558, "step": 11890 }, { "epoch": 75.79617834394904, "grad_norm": 1.787864327430725, "learning_rate": 0.0001, "loss": 0.0514, "step": 11900 }, { "epoch": 75.85987261146497, "grad_norm": 2.1651008129119873, "learning_rate": 0.0001, "loss": 0.0518, "step": 11910 }, { "epoch": 75.92356687898089, "grad_norm": 1.8039911985397339, "learning_rate": 0.0001, "loss": 0.0511, "step": 11920 }, { "epoch": 75.98726114649682, "grad_norm": 1.6886134147644043, "learning_rate": 0.0001, "loss": 0.051, "step": 11930 }, { "epoch": 76.05095541401273, "grad_norm": 1.8819880485534668, "learning_rate": 0.0001, "loss": 0.0508, "step": 11940 }, { "epoch": 76.11464968152866, "grad_norm": 1.8052765130996704, "learning_rate": 0.0001, "loss": 0.0527, "step": 11950 }, { "epoch": 76.17834394904459, "grad_norm": 1.7464878559112549, "learning_rate": 0.0001, "loss": 0.0479, "step": 11960 }, { "epoch": 76.24203821656052, "grad_norm": 1.8114702701568604, "learning_rate": 0.0001, "loss": 0.0524, "step": 11970 }, { "epoch": 76.30573248407643, "grad_norm": 1.8462086915969849, "learning_rate": 0.0001, "loss": 0.0517, "step": 11980 }, { "epoch": 76.36942675159236, "grad_norm": 1.780972957611084, "learning_rate": 0.0001, "loss": 0.0505, "step": 11990 }, { "epoch": 76.43312101910828, "grad_norm": 1.7920222282409668, "learning_rate": 0.0001, "loss": 0.0488, "step": 12000 }, { "epoch": 76.4968152866242, "grad_norm": 1.4111169576644897, "learning_rate": 0.0001, "loss": 0.0523, "step": 12010 }, { "epoch": 76.56050955414013, "grad_norm": 1.5461814403533936, "learning_rate": 0.0001, "loss": 0.0521, "step": 12020 }, { "epoch": 76.62420382165605, "grad_norm": 1.6446293592453003, "learning_rate": 0.0001, "loss": 0.0513, "step": 12030 }, { "epoch": 76.68789808917198, "grad_norm": 1.6546131372451782, "learning_rate": 0.0001, "loss": 0.0501, "step": 12040 }, { "epoch": 76.7515923566879, "grad_norm": 1.5294133424758911, "learning_rate": 0.0001, "loss": 0.0521, "step": 12050 }, { "epoch": 76.81528662420382, "grad_norm": 1.5983355045318604, "learning_rate": 0.0001, "loss": 0.0541, "step": 12060 }, { "epoch": 76.87898089171975, "grad_norm": 1.633764386177063, "learning_rate": 0.0001, "loss": 0.0493, "step": 12070 }, { "epoch": 76.94267515923566, "grad_norm": 1.609096884727478, "learning_rate": 0.0001, "loss": 0.0523, "step": 12080 }, { "epoch": 77.00636942675159, "grad_norm": 1.759411334991455, "learning_rate": 0.0001, "loss": 0.0516, "step": 12090 }, { "epoch": 77.07006369426752, "grad_norm": 1.8851250410079956, "learning_rate": 0.0001, "loss": 0.0497, "step": 12100 }, { "epoch": 77.13375796178345, "grad_norm": 1.4698072671890259, "learning_rate": 0.0001, "loss": 0.0497, "step": 12110 }, { "epoch": 77.19745222929936, "grad_norm": 1.7010939121246338, "learning_rate": 0.0001, "loss": 0.0511, "step": 12120 }, { "epoch": 77.26114649681529, "grad_norm": 1.6025100946426392, "learning_rate": 0.0001, "loss": 0.0515, "step": 12130 }, { "epoch": 77.32484076433121, "grad_norm": 1.4332853555679321, "learning_rate": 0.0001, "loss": 0.0481, "step": 12140 }, { "epoch": 77.38853503184713, "grad_norm": 1.8133318424224854, "learning_rate": 0.0001, "loss": 0.0488, "step": 12150 }, { "epoch": 77.45222929936305, "grad_norm": 1.624714732170105, "learning_rate": 0.0001, "loss": 0.0515, "step": 12160 }, { "epoch": 77.51592356687898, "grad_norm": 1.6665362119674683, "learning_rate": 0.0001, "loss": 0.0501, "step": 12170 }, { "epoch": 77.57961783439491, "grad_norm": 1.7437957525253296, "learning_rate": 0.0001, "loss": 0.0503, "step": 12180 }, { "epoch": 77.64331210191082, "grad_norm": 1.7609339952468872, "learning_rate": 0.0001, "loss": 0.048, "step": 12190 }, { "epoch": 77.70700636942675, "grad_norm": 1.730905294418335, "learning_rate": 0.0001, "loss": 0.0509, "step": 12200 }, { "epoch": 77.77070063694268, "grad_norm": 1.6418629884719849, "learning_rate": 0.0001, "loss": 0.0486, "step": 12210 }, { "epoch": 77.8343949044586, "grad_norm": 1.5005050897598267, "learning_rate": 0.0001, "loss": 0.0508, "step": 12220 }, { "epoch": 77.89808917197452, "grad_norm": 1.4765671491622925, "learning_rate": 0.0001, "loss": 0.0513, "step": 12230 }, { "epoch": 77.96178343949045, "grad_norm": 1.6950738430023193, "learning_rate": 0.0001, "loss": 0.0529, "step": 12240 }, { "epoch": 78.02547770700637, "grad_norm": 1.6467220783233643, "learning_rate": 0.0001, "loss": 0.0528, "step": 12250 }, { "epoch": 78.08917197452229, "grad_norm": 1.7577016353607178, "learning_rate": 0.0001, "loss": 0.0499, "step": 12260 }, { "epoch": 78.15286624203821, "grad_norm": 1.7914732694625854, "learning_rate": 0.0001, "loss": 0.0502, "step": 12270 }, { "epoch": 78.21656050955414, "grad_norm": 1.915697455406189, "learning_rate": 0.0001, "loss": 0.0506, "step": 12280 }, { "epoch": 78.28025477707007, "grad_norm": 1.9716625213623047, "learning_rate": 0.0001, "loss": 0.0508, "step": 12290 }, { "epoch": 78.34394904458598, "grad_norm": 2.070148468017578, "learning_rate": 0.0001, "loss": 0.0488, "step": 12300 }, { "epoch": 78.40764331210191, "grad_norm": 1.7839282751083374, "learning_rate": 0.0001, "loss": 0.0511, "step": 12310 }, { "epoch": 78.47133757961784, "grad_norm": 1.8590002059936523, "learning_rate": 0.0001, "loss": 0.0518, "step": 12320 }, { "epoch": 78.53503184713375, "grad_norm": 1.7945940494537354, "learning_rate": 0.0001, "loss": 0.0502, "step": 12330 }, { "epoch": 78.59872611464968, "grad_norm": 1.6794060468673706, "learning_rate": 0.0001, "loss": 0.0505, "step": 12340 }, { "epoch": 78.6624203821656, "grad_norm": 1.8985562324523926, "learning_rate": 0.0001, "loss": 0.053, "step": 12350 }, { "epoch": 78.72611464968153, "grad_norm": 1.7300053834915161, "learning_rate": 0.0001, "loss": 0.0535, "step": 12360 }, { "epoch": 78.78980891719745, "grad_norm": 1.7149120569229126, "learning_rate": 0.0001, "loss": 0.0512, "step": 12370 }, { "epoch": 78.85350318471338, "grad_norm": 1.780017375946045, "learning_rate": 0.0001, "loss": 0.052, "step": 12380 }, { "epoch": 78.9171974522293, "grad_norm": 1.8347100019454956, "learning_rate": 0.0001, "loss": 0.0507, "step": 12390 }, { "epoch": 78.98089171974522, "grad_norm": 1.690537691116333, "learning_rate": 0.0001, "loss": 0.0511, "step": 12400 }, { "epoch": 79.04458598726114, "grad_norm": 2.162991523742676, "learning_rate": 0.0001, "loss": 0.0487, "step": 12410 }, { "epoch": 79.10828025477707, "grad_norm": 1.4668869972229004, "learning_rate": 0.0001, "loss": 0.0488, "step": 12420 }, { "epoch": 79.171974522293, "grad_norm": 1.6337465047836304, "learning_rate": 0.0001, "loss": 0.0502, "step": 12430 }, { "epoch": 79.23566878980891, "grad_norm": 1.887885332107544, "learning_rate": 0.0001, "loss": 0.0481, "step": 12440 }, { "epoch": 79.29936305732484, "grad_norm": 1.725399136543274, "learning_rate": 0.0001, "loss": 0.0489, "step": 12450 }, { "epoch": 79.36305732484077, "grad_norm": 1.383437156677246, "learning_rate": 0.0001, "loss": 0.0468, "step": 12460 }, { "epoch": 79.4267515923567, "grad_norm": 1.9962074756622314, "learning_rate": 0.0001, "loss": 0.0543, "step": 12470 }, { "epoch": 79.49044585987261, "grad_norm": 1.7573449611663818, "learning_rate": 0.0001, "loss": 0.0498, "step": 12480 }, { "epoch": 79.55414012738854, "grad_norm": 1.6104313135147095, "learning_rate": 0.0001, "loss": 0.0468, "step": 12490 }, { "epoch": 79.61783439490446, "grad_norm": 1.5359727144241333, "learning_rate": 0.0001, "loss": 0.0553, "step": 12500 }, { "epoch": 79.68152866242038, "grad_norm": 1.4639062881469727, "learning_rate": 0.0001, "loss": 0.0536, "step": 12510 }, { "epoch": 79.7452229299363, "grad_norm": 1.6044377088546753, "learning_rate": 0.0001, "loss": 0.0534, "step": 12520 }, { "epoch": 79.80891719745223, "grad_norm": 1.7305707931518555, "learning_rate": 0.0001, "loss": 0.0488, "step": 12530 }, { "epoch": 79.87261146496816, "grad_norm": 1.486091136932373, "learning_rate": 0.0001, "loss": 0.0538, "step": 12540 }, { "epoch": 79.93630573248407, "grad_norm": 1.4696218967437744, "learning_rate": 0.0001, "loss": 0.0492, "step": 12550 }, { "epoch": 80.0, "grad_norm": 1.8167083263397217, "learning_rate": 0.0001, "loss": 0.0526, "step": 12560 }, { "epoch": 80.06369426751593, "grad_norm": 2.1027657985687256, "learning_rate": 0.0001, "loss": 0.0522, "step": 12570 }, { "epoch": 80.12738853503184, "grad_norm": 1.8279118537902832, "learning_rate": 0.0001, "loss": 0.052, "step": 12580 }, { "epoch": 80.19108280254777, "grad_norm": 1.622800350189209, "learning_rate": 0.0001, "loss": 0.0502, "step": 12590 }, { "epoch": 80.2547770700637, "grad_norm": 2.0850188732147217, "learning_rate": 0.0001, "loss": 0.05, "step": 12600 }, { "epoch": 80.31847133757962, "grad_norm": 2.0340628623962402, "learning_rate": 0.0001, "loss": 0.0487, "step": 12610 }, { "epoch": 80.38216560509554, "grad_norm": 1.5827347040176392, "learning_rate": 0.0001, "loss": 0.0513, "step": 12620 }, { "epoch": 80.44585987261146, "grad_norm": 1.6292204856872559, "learning_rate": 0.0001, "loss": 0.0499, "step": 12630 }, { "epoch": 80.50955414012739, "grad_norm": 2.0578110218048096, "learning_rate": 0.0001, "loss": 0.0552, "step": 12640 }, { "epoch": 80.5732484076433, "grad_norm": 2.185716152191162, "learning_rate": 0.0001, "loss": 0.0503, "step": 12650 }, { "epoch": 80.63694267515923, "grad_norm": 2.2086429595947266, "learning_rate": 0.0001, "loss": 0.0515, "step": 12660 }, { "epoch": 80.70063694267516, "grad_norm": 2.1864967346191406, "learning_rate": 0.0001, "loss": 0.0529, "step": 12670 }, { "epoch": 80.76433121019109, "grad_norm": 2.0121521949768066, "learning_rate": 0.0001, "loss": 0.0453, "step": 12680 }, { "epoch": 80.828025477707, "grad_norm": 1.8989063501358032, "learning_rate": 0.0001, "loss": 0.0483, "step": 12690 }, { "epoch": 80.89171974522293, "grad_norm": 1.7268705368041992, "learning_rate": 0.0001, "loss": 0.0488, "step": 12700 }, { "epoch": 80.95541401273886, "grad_norm": 1.9325916767120361, "learning_rate": 0.0001, "loss": 0.0473, "step": 12710 }, { "epoch": 81.01910828025478, "grad_norm": 2.090557813644409, "learning_rate": 0.0001, "loss": 0.0485, "step": 12720 }, { "epoch": 81.0828025477707, "grad_norm": 3.502279281616211, "learning_rate": 0.0001, "loss": 0.0498, "step": 12730 }, { "epoch": 81.14649681528662, "grad_norm": 2.7308287620544434, "learning_rate": 0.0001, "loss": 0.0499, "step": 12740 }, { "epoch": 81.21019108280255, "grad_norm": 2.540210485458374, "learning_rate": 0.0001, "loss": 0.0495, "step": 12750 }, { "epoch": 81.27388535031847, "grad_norm": 2.3726377487182617, "learning_rate": 0.0001, "loss": 0.0477, "step": 12760 }, { "epoch": 81.3375796178344, "grad_norm": 2.3219895362854004, "learning_rate": 0.0001, "loss": 0.0488, "step": 12770 }, { "epoch": 81.40127388535032, "grad_norm": 2.1245436668395996, "learning_rate": 0.0001, "loss": 0.0478, "step": 12780 }, { "epoch": 81.46496815286625, "grad_norm": 1.891027569770813, "learning_rate": 0.0001, "loss": 0.0488, "step": 12790 }, { "epoch": 81.52866242038216, "grad_norm": 1.808934211730957, "learning_rate": 0.0001, "loss": 0.0479, "step": 12800 }, { "epoch": 81.59235668789809, "grad_norm": 2.1144394874572754, "learning_rate": 0.0001, "loss": 0.0473, "step": 12810 }, { "epoch": 81.65605095541402, "grad_norm": 1.7516521215438843, "learning_rate": 0.0001, "loss": 0.0444, "step": 12820 }, { "epoch": 81.71974522292993, "grad_norm": 1.7118208408355713, "learning_rate": 0.0001, "loss": 0.0477, "step": 12830 }, { "epoch": 81.78343949044586, "grad_norm": 1.6813533306121826, "learning_rate": 0.0001, "loss": 0.0454, "step": 12840 }, { "epoch": 81.84713375796179, "grad_norm": 1.7281651496887207, "learning_rate": 0.0001, "loss": 0.0496, "step": 12850 }, { "epoch": 81.91082802547771, "grad_norm": 1.9444782733917236, "learning_rate": 0.0001, "loss": 0.0511, "step": 12860 }, { "epoch": 81.97452229299363, "grad_norm": 1.74335777759552, "learning_rate": 0.0001, "loss": 0.0467, "step": 12870 }, { "epoch": 82.03821656050955, "grad_norm": 1.83609938621521, "learning_rate": 0.0001, "loss": 0.0477, "step": 12880 }, { "epoch": 82.10191082802548, "grad_norm": 1.853606104850769, "learning_rate": 0.0001, "loss": 0.0466, "step": 12890 }, { "epoch": 82.1656050955414, "grad_norm": 1.6843804121017456, "learning_rate": 0.0001, "loss": 0.0504, "step": 12900 }, { "epoch": 82.22929936305732, "grad_norm": 1.5620334148406982, "learning_rate": 0.0001, "loss": 0.0496, "step": 12910 }, { "epoch": 82.29299363057325, "grad_norm": 1.542197346687317, "learning_rate": 0.0001, "loss": 0.0513, "step": 12920 }, { "epoch": 82.35668789808918, "grad_norm": 2.2766075134277344, "learning_rate": 0.0001, "loss": 0.0565, "step": 12930 }, { "epoch": 82.42038216560509, "grad_norm": 1.902388095855713, "learning_rate": 0.0001, "loss": 0.0492, "step": 12940 }, { "epoch": 82.48407643312102, "grad_norm": 1.8320927619934082, "learning_rate": 0.0001, "loss": 0.0519, "step": 12950 }, { "epoch": 82.54777070063695, "grad_norm": 1.8929121494293213, "learning_rate": 0.0001, "loss": 0.047, "step": 12960 }, { "epoch": 82.61146496815287, "grad_norm": 1.9219220876693726, "learning_rate": 0.0001, "loss": 0.0461, "step": 12970 }, { "epoch": 82.67515923566879, "grad_norm": 2.0982742309570312, "learning_rate": 0.0001, "loss": 0.0489, "step": 12980 }, { "epoch": 82.73885350318471, "grad_norm": 2.4361064434051514, "learning_rate": 0.0001, "loss": 0.0521, "step": 12990 }, { "epoch": 82.80254777070064, "grad_norm": 2.175934076309204, "learning_rate": 0.0001, "loss": 0.0501, "step": 13000 }, { "epoch": 82.86624203821655, "grad_norm": 2.1100049018859863, "learning_rate": 0.0001, "loss": 0.0468, "step": 13010 }, { "epoch": 82.92993630573248, "grad_norm": 2.0762252807617188, "learning_rate": 0.0001, "loss": 0.0509, "step": 13020 }, { "epoch": 82.99363057324841, "grad_norm": 2.0821499824523926, "learning_rate": 0.0001, "loss": 0.0501, "step": 13030 }, { "epoch": 83.05732484076434, "grad_norm": 1.6105787754058838, "learning_rate": 0.0001, "loss": 0.048, "step": 13040 }, { "epoch": 83.12101910828025, "grad_norm": 2.022373676300049, "learning_rate": 0.0001, "loss": 0.0467, "step": 13050 }, { "epoch": 83.18471337579618, "grad_norm": 2.0437633991241455, "learning_rate": 0.0001, "loss": 0.0492, "step": 13060 }, { "epoch": 83.2484076433121, "grad_norm": 1.8118585348129272, "learning_rate": 0.0001, "loss": 0.0479, "step": 13070 }, { "epoch": 83.31210191082802, "grad_norm": 1.8799163103103638, "learning_rate": 0.0001, "loss": 0.0466, "step": 13080 }, { "epoch": 83.37579617834395, "grad_norm": 1.6529191732406616, "learning_rate": 0.0001, "loss": 0.0473, "step": 13090 }, { "epoch": 83.43949044585987, "grad_norm": 1.6329364776611328, "learning_rate": 0.0001, "loss": 0.0471, "step": 13100 }, { "epoch": 83.5031847133758, "grad_norm": 1.9707340002059937, "learning_rate": 0.0001, "loss": 0.0486, "step": 13110 }, { "epoch": 83.56687898089172, "grad_norm": 1.6363554000854492, "learning_rate": 0.0001, "loss": 0.0478, "step": 13120 }, { "epoch": 83.63057324840764, "grad_norm": 2.09277081489563, "learning_rate": 0.0001, "loss": 0.0512, "step": 13130 }, { "epoch": 83.69426751592357, "grad_norm": 1.82542085647583, "learning_rate": 0.0001, "loss": 0.0501, "step": 13140 }, { "epoch": 83.75796178343948, "grad_norm": 1.757753610610962, "learning_rate": 0.0001, "loss": 0.0494, "step": 13150 }, { "epoch": 83.82165605095541, "grad_norm": 1.6028114557266235, "learning_rate": 0.0001, "loss": 0.0462, "step": 13160 }, { "epoch": 83.88535031847134, "grad_norm": 1.7034083604812622, "learning_rate": 0.0001, "loss": 0.0483, "step": 13170 }, { "epoch": 83.94904458598727, "grad_norm": 1.6255407333374023, "learning_rate": 0.0001, "loss": 0.0507, "step": 13180 }, { "epoch": 84.01273885350318, "grad_norm": 1.4854459762573242, "learning_rate": 0.0001, "loss": 0.0465, "step": 13190 }, { "epoch": 84.07643312101911, "grad_norm": 1.370410442352295, "learning_rate": 0.0001, "loss": 0.0468, "step": 13200 }, { "epoch": 84.14012738853503, "grad_norm": 1.633548378944397, "learning_rate": 0.0001, "loss": 0.0475, "step": 13210 }, { "epoch": 84.20382165605096, "grad_norm": 1.5780941247940063, "learning_rate": 0.0001, "loss": 0.0496, "step": 13220 }, { "epoch": 84.26751592356688, "grad_norm": 1.7333341836929321, "learning_rate": 0.0001, "loss": 0.0496, "step": 13230 }, { "epoch": 84.3312101910828, "grad_norm": 1.5129436254501343, "learning_rate": 0.0001, "loss": 0.0489, "step": 13240 }, { "epoch": 84.39490445859873, "grad_norm": 1.635573148727417, "learning_rate": 0.0001, "loss": 0.0503, "step": 13250 }, { "epoch": 84.45859872611464, "grad_norm": 1.9976927042007446, "learning_rate": 0.0001, "loss": 0.0463, "step": 13260 }, { "epoch": 84.52229299363057, "grad_norm": 1.7515299320220947, "learning_rate": 0.0001, "loss": 0.0504, "step": 13270 }, { "epoch": 84.5859872611465, "grad_norm": 1.6865262985229492, "learning_rate": 0.0001, "loss": 0.0518, "step": 13280 }, { "epoch": 84.64968152866243, "grad_norm": 1.4847381114959717, "learning_rate": 0.0001, "loss": 0.0528, "step": 13290 }, { "epoch": 84.71337579617834, "grad_norm": 1.497086763381958, "learning_rate": 0.0001, "loss": 0.0523, "step": 13300 }, { "epoch": 84.77707006369427, "grad_norm": 1.6409685611724854, "learning_rate": 0.0001, "loss": 0.0559, "step": 13310 }, { "epoch": 84.8407643312102, "grad_norm": 1.5017833709716797, "learning_rate": 0.0001, "loss": 0.0542, "step": 13320 }, { "epoch": 84.90445859872611, "grad_norm": 1.6369761228561401, "learning_rate": 0.0001, "loss": 0.0532, "step": 13330 }, { "epoch": 84.96815286624204, "grad_norm": 1.563537359237671, "learning_rate": 0.0001, "loss": 0.0555, "step": 13340 }, { "epoch": 85.03184713375796, "grad_norm": 1.4730894565582275, "learning_rate": 0.0001, "loss": 0.0508, "step": 13350 }, { "epoch": 85.09554140127389, "grad_norm": 1.9422591924667358, "learning_rate": 0.0001, "loss": 0.0514, "step": 13360 }, { "epoch": 85.1592356687898, "grad_norm": 1.9511085748672485, "learning_rate": 0.0001, "loss": 0.0508, "step": 13370 }, { "epoch": 85.22292993630573, "grad_norm": 2.008615732192993, "learning_rate": 0.0001, "loss": 0.0576, "step": 13380 }, { "epoch": 85.28662420382166, "grad_norm": 1.7340503931045532, "learning_rate": 0.0001, "loss": 0.0554, "step": 13390 }, { "epoch": 85.35031847133757, "grad_norm": 1.8740181922912598, "learning_rate": 0.0001, "loss": 0.0506, "step": 13400 }, { "epoch": 85.4140127388535, "grad_norm": 1.7708492279052734, "learning_rate": 0.0001, "loss": 0.0484, "step": 13410 }, { "epoch": 85.47770700636943, "grad_norm": 1.9248510599136353, "learning_rate": 0.0001, "loss": 0.0479, "step": 13420 }, { "epoch": 85.54140127388536, "grad_norm": 1.6247023344039917, "learning_rate": 0.0001, "loss": 0.0524, "step": 13430 }, { "epoch": 85.60509554140127, "grad_norm": 1.5788811445236206, "learning_rate": 0.0001, "loss": 0.0558, "step": 13440 }, { "epoch": 85.6687898089172, "grad_norm": 1.6494181156158447, "learning_rate": 0.0001, "loss": 0.0476, "step": 13450 }, { "epoch": 85.73248407643312, "grad_norm": 1.3825238943099976, "learning_rate": 0.0001, "loss": 0.0476, "step": 13460 }, { "epoch": 85.79617834394904, "grad_norm": 1.7376831769943237, "learning_rate": 0.0001, "loss": 0.0491, "step": 13470 }, { "epoch": 85.85987261146497, "grad_norm": 1.8842839002609253, "learning_rate": 0.0001, "loss": 0.0497, "step": 13480 }, { "epoch": 85.92356687898089, "grad_norm": 1.9629732370376587, "learning_rate": 0.0001, "loss": 0.0498, "step": 13490 }, { "epoch": 85.98726114649682, "grad_norm": 1.8728734254837036, "learning_rate": 0.0001, "loss": 0.0495, "step": 13500 }, { "epoch": 86.05095541401273, "grad_norm": 1.8000260591506958, "learning_rate": 0.0001, "loss": 0.0477, "step": 13510 }, { "epoch": 86.11464968152866, "grad_norm": 1.7386105060577393, "learning_rate": 0.0001, "loss": 0.0451, "step": 13520 }, { "epoch": 86.17834394904459, "grad_norm": 1.563156008720398, "learning_rate": 0.0001, "loss": 0.0473, "step": 13530 }, { "epoch": 86.24203821656052, "grad_norm": 1.7159897089004517, "learning_rate": 0.0001, "loss": 0.0469, "step": 13540 }, { "epoch": 86.30573248407643, "grad_norm": 1.9237146377563477, "learning_rate": 0.0001, "loss": 0.0483, "step": 13550 }, { "epoch": 86.36942675159236, "grad_norm": 1.4293407201766968, "learning_rate": 0.0001, "loss": 0.0521, "step": 13560 }, { "epoch": 86.43312101910828, "grad_norm": 1.963935375213623, "learning_rate": 0.0001, "loss": 0.0496, "step": 13570 }, { "epoch": 86.4968152866242, "grad_norm": 1.8289490938186646, "learning_rate": 0.0001, "loss": 0.0482, "step": 13580 }, { "epoch": 86.56050955414013, "grad_norm": 1.8148990869522095, "learning_rate": 0.0001, "loss": 0.0486, "step": 13590 }, { "epoch": 86.62420382165605, "grad_norm": 1.6656323671340942, "learning_rate": 0.0001, "loss": 0.0503, "step": 13600 }, { "epoch": 86.68789808917198, "grad_norm": 1.6304277181625366, "learning_rate": 0.0001, "loss": 0.0513, "step": 13610 }, { "epoch": 86.7515923566879, "grad_norm": 1.5648895502090454, "learning_rate": 0.0001, "loss": 0.055, "step": 13620 }, { "epoch": 86.81528662420382, "grad_norm": 1.5770680904388428, "learning_rate": 0.0001, "loss": 0.0544, "step": 13630 }, { "epoch": 86.87898089171975, "grad_norm": 1.6606395244598389, "learning_rate": 0.0001, "loss": 0.0508, "step": 13640 }, { "epoch": 86.94267515923566, "grad_norm": 1.5189493894577026, "learning_rate": 0.0001, "loss": 0.0507, "step": 13650 }, { "epoch": 87.00636942675159, "grad_norm": 1.4936827421188354, "learning_rate": 0.0001, "loss": 0.0498, "step": 13660 }, { "epoch": 87.07006369426752, "grad_norm": 1.7071796655654907, "learning_rate": 0.0001, "loss": 0.0517, "step": 13670 }, { "epoch": 87.13375796178345, "grad_norm": 1.1999627351760864, "learning_rate": 0.0001, "loss": 0.0487, "step": 13680 }, { "epoch": 87.19745222929936, "grad_norm": 1.5409961938858032, "learning_rate": 0.0001, "loss": 0.0495, "step": 13690 }, { "epoch": 87.26114649681529, "grad_norm": 1.8969858884811401, "learning_rate": 0.0001, "loss": 0.0558, "step": 13700 }, { "epoch": 87.32484076433121, "grad_norm": 1.4021774530410767, "learning_rate": 0.0001, "loss": 0.0524, "step": 13710 }, { "epoch": 87.38853503184713, "grad_norm": 1.5925968885421753, "learning_rate": 0.0001, "loss": 0.054, "step": 13720 }, { "epoch": 87.45222929936305, "grad_norm": 1.5209338665008545, "learning_rate": 0.0001, "loss": 0.0546, "step": 13730 }, { "epoch": 87.51592356687898, "grad_norm": 1.3787864446640015, "learning_rate": 0.0001, "loss": 0.0531, "step": 13740 }, { "epoch": 87.57961783439491, "grad_norm": 1.6214630603790283, "learning_rate": 0.0001, "loss": 0.0528, "step": 13750 }, { "epoch": 87.64331210191082, "grad_norm": 1.4972646236419678, "learning_rate": 0.0001, "loss": 0.054, "step": 13760 }, { "epoch": 87.70700636942675, "grad_norm": 1.6363133192062378, "learning_rate": 0.0001, "loss": 0.0544, "step": 13770 }, { "epoch": 87.77070063694268, "grad_norm": 1.6986254453659058, "learning_rate": 0.0001, "loss": 0.0519, "step": 13780 }, { "epoch": 87.8343949044586, "grad_norm": 1.5493896007537842, "learning_rate": 0.0001, "loss": 0.0495, "step": 13790 }, { "epoch": 87.89808917197452, "grad_norm": 1.469539999961853, "learning_rate": 0.0001, "loss": 0.0528, "step": 13800 }, { "epoch": 87.96178343949045, "grad_norm": 1.922215461730957, "learning_rate": 0.0001, "loss": 0.0508, "step": 13810 }, { "epoch": 88.02547770700637, "grad_norm": 1.703686237335205, "learning_rate": 0.0001, "loss": 0.0514, "step": 13820 }, { "epoch": 88.08917197452229, "grad_norm": 1.6973074674606323, "learning_rate": 0.0001, "loss": 0.0507, "step": 13830 }, { "epoch": 88.15286624203821, "grad_norm": 1.72831392288208, "learning_rate": 0.0001, "loss": 0.0503, "step": 13840 }, { "epoch": 88.21656050955414, "grad_norm": 1.491443395614624, "learning_rate": 0.0001, "loss": 0.0493, "step": 13850 }, { "epoch": 88.28025477707007, "grad_norm": 1.5107890367507935, "learning_rate": 0.0001, "loss": 0.0512, "step": 13860 }, { "epoch": 88.34394904458598, "grad_norm": 1.470184564590454, "learning_rate": 0.0001, "loss": 0.0479, "step": 13870 }, { "epoch": 88.40764331210191, "grad_norm": 1.613457202911377, "learning_rate": 0.0001, "loss": 0.0481, "step": 13880 }, { "epoch": 88.47133757961784, "grad_norm": 1.9858452081680298, "learning_rate": 0.0001, "loss": 0.0536, "step": 13890 }, { "epoch": 88.53503184713375, "grad_norm": 1.9290413856506348, "learning_rate": 0.0001, "loss": 0.0539, "step": 13900 }, { "epoch": 88.59872611464968, "grad_norm": 1.6686451435089111, "learning_rate": 0.0001, "loss": 0.0481, "step": 13910 }, { "epoch": 88.6624203821656, "grad_norm": 1.4673595428466797, "learning_rate": 0.0001, "loss": 0.0501, "step": 13920 }, { "epoch": 88.72611464968153, "grad_norm": 1.3007479906082153, "learning_rate": 0.0001, "loss": 0.052, "step": 13930 }, { "epoch": 88.78980891719745, "grad_norm": 1.5992950201034546, "learning_rate": 0.0001, "loss": 0.0498, "step": 13940 }, { "epoch": 88.85350318471338, "grad_norm": 1.7177757024765015, "learning_rate": 0.0001, "loss": 0.0487, "step": 13950 }, { "epoch": 88.9171974522293, "grad_norm": 1.573062539100647, "learning_rate": 0.0001, "loss": 0.0544, "step": 13960 }, { "epoch": 88.98089171974522, "grad_norm": 1.6905701160430908, "learning_rate": 0.0001, "loss": 0.0491, "step": 13970 }, { "epoch": 89.04458598726114, "grad_norm": 1.434324026107788, "learning_rate": 0.0001, "loss": 0.0512, "step": 13980 }, { "epoch": 89.10828025477707, "grad_norm": 1.4276012182235718, "learning_rate": 0.0001, "loss": 0.0516, "step": 13990 }, { "epoch": 89.171974522293, "grad_norm": 1.5272549390792847, "learning_rate": 0.0001, "loss": 0.0502, "step": 14000 }, { "epoch": 89.23566878980891, "grad_norm": 1.666749358177185, "learning_rate": 0.0001, "loss": 0.0513, "step": 14010 }, { "epoch": 89.29936305732484, "grad_norm": 1.249731421470642, "learning_rate": 0.0001, "loss": 0.0492, "step": 14020 }, { "epoch": 89.36305732484077, "grad_norm": 1.335440754890442, "learning_rate": 0.0001, "loss": 0.0493, "step": 14030 }, { "epoch": 89.4267515923567, "grad_norm": 1.5451223850250244, "learning_rate": 0.0001, "loss": 0.0549, "step": 14040 }, { "epoch": 89.49044585987261, "grad_norm": 1.7401175498962402, "learning_rate": 0.0001, "loss": 0.0494, "step": 14050 }, { "epoch": 89.55414012738854, "grad_norm": 1.5365450382232666, "learning_rate": 0.0001, "loss": 0.0545, "step": 14060 }, { "epoch": 89.61783439490446, "grad_norm": 1.5835082530975342, "learning_rate": 0.0001, "loss": 0.0492, "step": 14070 }, { "epoch": 89.68152866242038, "grad_norm": 1.2934550046920776, "learning_rate": 0.0001, "loss": 0.0521, "step": 14080 }, { "epoch": 89.7452229299363, "grad_norm": 1.9727649688720703, "learning_rate": 0.0001, "loss": 0.0519, "step": 14090 }, { "epoch": 89.80891719745223, "grad_norm": 1.6824711561203003, "learning_rate": 0.0001, "loss": 0.0491, "step": 14100 }, { "epoch": 89.87261146496816, "grad_norm": 1.5465885400772095, "learning_rate": 0.0001, "loss": 0.049, "step": 14110 }, { "epoch": 89.93630573248407, "grad_norm": 1.6003862619400024, "learning_rate": 0.0001, "loss": 0.0499, "step": 14120 }, { "epoch": 90.0, "grad_norm": 1.2944692373275757, "learning_rate": 0.0001, "loss": 0.0507, "step": 14130 }, { "epoch": 90.06369426751593, "grad_norm": 1.4662762880325317, "learning_rate": 0.0001, "loss": 0.0445, "step": 14140 }, { "epoch": 90.12738853503184, "grad_norm": 1.2515543699264526, "learning_rate": 0.0001, "loss": 0.0501, "step": 14150 }, { "epoch": 90.19108280254777, "grad_norm": 1.6725555658340454, "learning_rate": 0.0001, "loss": 0.0523, "step": 14160 }, { "epoch": 90.2547770700637, "grad_norm": 1.8474984169006348, "learning_rate": 0.0001, "loss": 0.0521, "step": 14170 }, { "epoch": 90.31847133757962, "grad_norm": 1.7244415283203125, "learning_rate": 0.0001, "loss": 0.0465, "step": 14180 }, { "epoch": 90.38216560509554, "grad_norm": 1.3762623071670532, "learning_rate": 0.0001, "loss": 0.0486, "step": 14190 }, { "epoch": 90.44585987261146, "grad_norm": 1.7953150272369385, "learning_rate": 0.0001, "loss": 0.0488, "step": 14200 }, { "epoch": 90.50955414012739, "grad_norm": 1.5140800476074219, "learning_rate": 0.0001, "loss": 0.0452, "step": 14210 }, { "epoch": 90.5732484076433, "grad_norm": 1.9735854864120483, "learning_rate": 0.0001, "loss": 0.0515, "step": 14220 }, { "epoch": 90.63694267515923, "grad_norm": 1.716007947921753, "learning_rate": 0.0001, "loss": 0.051, "step": 14230 }, { "epoch": 90.70063694267516, "grad_norm": 2.124378204345703, "learning_rate": 0.0001, "loss": 0.0546, "step": 14240 }, { "epoch": 90.76433121019109, "grad_norm": 1.4421651363372803, "learning_rate": 0.0001, "loss": 0.0479, "step": 14250 }, { "epoch": 90.828025477707, "grad_norm": 1.5857638120651245, "learning_rate": 0.0001, "loss": 0.0488, "step": 14260 }, { "epoch": 90.89171974522293, "grad_norm": 1.69535231590271, "learning_rate": 0.0001, "loss": 0.0478, "step": 14270 }, { "epoch": 90.95541401273886, "grad_norm": 1.8273509740829468, "learning_rate": 0.0001, "loss": 0.0451, "step": 14280 }, { "epoch": 91.01910828025478, "grad_norm": 1.7491302490234375, "learning_rate": 0.0001, "loss": 0.0493, "step": 14290 }, { "epoch": 91.0828025477707, "grad_norm": 1.8842182159423828, "learning_rate": 0.0001, "loss": 0.0468, "step": 14300 }, { "epoch": 91.14649681528662, "grad_norm": 1.648573398590088, "learning_rate": 0.0001, "loss": 0.048, "step": 14310 }, { "epoch": 91.21019108280255, "grad_norm": 1.324217438697815, "learning_rate": 0.0001, "loss": 0.0505, "step": 14320 }, { "epoch": 91.27388535031847, "grad_norm": 2.5345499515533447, "learning_rate": 0.0001, "loss": 0.0516, "step": 14330 }, { "epoch": 91.3375796178344, "grad_norm": 2.0903708934783936, "learning_rate": 0.0001, "loss": 0.0493, "step": 14340 }, { "epoch": 91.40127388535032, "grad_norm": 1.7892881631851196, "learning_rate": 0.0001, "loss": 0.045, "step": 14350 }, { "epoch": 91.46496815286625, "grad_norm": 1.7294824123382568, "learning_rate": 0.0001, "loss": 0.0479, "step": 14360 }, { "epoch": 91.52866242038216, "grad_norm": 1.7749409675598145, "learning_rate": 0.0001, "loss": 0.0517, "step": 14370 }, { "epoch": 91.59235668789809, "grad_norm": 1.750657558441162, "learning_rate": 0.0001, "loss": 0.0475, "step": 14380 }, { "epoch": 91.65605095541402, "grad_norm": 2.005683183670044, "learning_rate": 0.0001, "loss": 0.047, "step": 14390 }, { "epoch": 91.71974522292993, "grad_norm": 1.816331386566162, "learning_rate": 0.0001, "loss": 0.047, "step": 14400 }, { "epoch": 91.78343949044586, "grad_norm": 1.7130392789840698, "learning_rate": 0.0001, "loss": 0.0484, "step": 14410 }, { "epoch": 91.84713375796179, "grad_norm": 1.5592385530471802, "learning_rate": 0.0001, "loss": 0.0463, "step": 14420 }, { "epoch": 91.91082802547771, "grad_norm": 1.782791256904602, "learning_rate": 0.0001, "loss": 0.0494, "step": 14430 }, { "epoch": 91.97452229299363, "grad_norm": 1.8127764463424683, "learning_rate": 0.0001, "loss": 0.0448, "step": 14440 }, { "epoch": 92.03821656050955, "grad_norm": 1.6421717405319214, "learning_rate": 0.0001, "loss": 0.0456, "step": 14450 }, { "epoch": 92.10191082802548, "grad_norm": 1.7124961614608765, "learning_rate": 0.0001, "loss": 0.0461, "step": 14460 }, { "epoch": 92.1656050955414, "grad_norm": 1.6722925901412964, "learning_rate": 0.0001, "loss": 0.048, "step": 14470 }, { "epoch": 92.22929936305732, "grad_norm": 1.6453193426132202, "learning_rate": 0.0001, "loss": 0.0461, "step": 14480 }, { "epoch": 92.29299363057325, "grad_norm": 1.402477502822876, "learning_rate": 0.0001, "loss": 0.0467, "step": 14490 }, { "epoch": 92.35668789808918, "grad_norm": 1.4851194620132446, "learning_rate": 0.0001, "loss": 0.0487, "step": 14500 }, { "epoch": 92.42038216560509, "grad_norm": 1.3546311855316162, "learning_rate": 0.0001, "loss": 0.0463, "step": 14510 }, { "epoch": 92.48407643312102, "grad_norm": 1.5455354452133179, "learning_rate": 0.0001, "loss": 0.0494, "step": 14520 }, { "epoch": 92.54777070063695, "grad_norm": 1.4869821071624756, "learning_rate": 0.0001, "loss": 0.0492, "step": 14530 }, { "epoch": 92.61146496815287, "grad_norm": 1.7440129518508911, "learning_rate": 0.0001, "loss": 0.0458, "step": 14540 }, { "epoch": 92.67515923566879, "grad_norm": 1.4171563386917114, "learning_rate": 0.0001, "loss": 0.0488, "step": 14550 }, { "epoch": 92.73885350318471, "grad_norm": 1.8667035102844238, "learning_rate": 0.0001, "loss": 0.0474, "step": 14560 }, { "epoch": 92.80254777070064, "grad_norm": 3.2918927669525146, "learning_rate": 0.0001, "loss": 0.0477, "step": 14570 }, { "epoch": 92.86624203821655, "grad_norm": 2.5170235633850098, "learning_rate": 0.0001, "loss": 0.0514, "step": 14580 }, { "epoch": 92.92993630573248, "grad_norm": 2.043153762817383, "learning_rate": 0.0001, "loss": 0.0532, "step": 14590 }, { "epoch": 92.99363057324841, "grad_norm": 1.914408802986145, "learning_rate": 0.0001, "loss": 0.0476, "step": 14600 }, { "epoch": 93.05732484076434, "grad_norm": 1.8067020177841187, "learning_rate": 0.0001, "loss": 0.0424, "step": 14610 }, { "epoch": 93.12101910828025, "grad_norm": 1.779273271560669, "learning_rate": 0.0001, "loss": 0.0457, "step": 14620 }, { "epoch": 93.18471337579618, "grad_norm": 1.812270164489746, "learning_rate": 0.0001, "loss": 0.0476, "step": 14630 }, { "epoch": 93.2484076433121, "grad_norm": 1.8711782693862915, "learning_rate": 0.0001, "loss": 0.0447, "step": 14640 }, { "epoch": 93.31210191082802, "grad_norm": 1.4985113143920898, "learning_rate": 0.0001, "loss": 0.0458, "step": 14650 }, { "epoch": 93.37579617834395, "grad_norm": 1.7114458084106445, "learning_rate": 0.0001, "loss": 0.0456, "step": 14660 }, { "epoch": 93.43949044585987, "grad_norm": 1.9598371982574463, "learning_rate": 0.0001, "loss": 0.0485, "step": 14670 }, { "epoch": 93.5031847133758, "grad_norm": 1.4980871677398682, "learning_rate": 0.0001, "loss": 0.0513, "step": 14680 }, { "epoch": 93.56687898089172, "grad_norm": 1.7075837850570679, "learning_rate": 0.0001, "loss": 0.049, "step": 14690 }, { "epoch": 93.63057324840764, "grad_norm": 1.6518592834472656, "learning_rate": 0.0001, "loss": 0.0484, "step": 14700 }, { "epoch": 93.69426751592357, "grad_norm": 1.5318683385849, "learning_rate": 0.0001, "loss": 0.0483, "step": 14710 }, { "epoch": 93.75796178343948, "grad_norm": 1.598813533782959, "learning_rate": 0.0001, "loss": 0.0494, "step": 14720 }, { "epoch": 93.82165605095541, "grad_norm": 1.5293395519256592, "learning_rate": 0.0001, "loss": 0.0478, "step": 14730 }, { "epoch": 93.88535031847134, "grad_norm": 1.7172064781188965, "learning_rate": 0.0001, "loss": 0.0465, "step": 14740 }, { "epoch": 93.94904458598727, "grad_norm": 1.6014286279678345, "learning_rate": 0.0001, "loss": 0.0452, "step": 14750 }, { "epoch": 94.01273885350318, "grad_norm": 1.4103577136993408, "learning_rate": 0.0001, "loss": 0.0419, "step": 14760 }, { "epoch": 94.07643312101911, "grad_norm": 1.5981829166412354, "learning_rate": 0.0001, "loss": 0.0466, "step": 14770 }, { "epoch": 94.14012738853503, "grad_norm": 2.0717453956604004, "learning_rate": 0.0001, "loss": 0.0461, "step": 14780 }, { "epoch": 94.20382165605096, "grad_norm": 1.9527339935302734, "learning_rate": 0.0001, "loss": 0.048, "step": 14790 }, { "epoch": 94.26751592356688, "grad_norm": 1.5469748973846436, "learning_rate": 0.0001, "loss": 0.0488, "step": 14800 }, { "epoch": 94.3312101910828, "grad_norm": 1.4868391752243042, "learning_rate": 0.0001, "loss": 0.0454, "step": 14810 }, { "epoch": 94.39490445859873, "grad_norm": 1.2876818180084229, "learning_rate": 0.0001, "loss": 0.045, "step": 14820 }, { "epoch": 94.45859872611464, "grad_norm": 1.5819729566574097, "learning_rate": 0.0001, "loss": 0.0474, "step": 14830 }, { "epoch": 94.52229299363057, "grad_norm": 1.817672848701477, "learning_rate": 0.0001, "loss": 0.0464, "step": 14840 }, { "epoch": 94.5859872611465, "grad_norm": 2.051816701889038, "learning_rate": 0.0001, "loss": 0.0476, "step": 14850 }, { "epoch": 94.64968152866243, "grad_norm": 1.5522571802139282, "learning_rate": 0.0001, "loss": 0.0503, "step": 14860 }, { "epoch": 94.71337579617834, "grad_norm": 1.4614670276641846, "learning_rate": 0.0001, "loss": 0.0467, "step": 14870 }, { "epoch": 94.77707006369427, "grad_norm": 1.61818265914917, "learning_rate": 0.0001, "loss": 0.0468, "step": 14880 }, { "epoch": 94.8407643312102, "grad_norm": 1.5494686365127563, "learning_rate": 0.0001, "loss": 0.046, "step": 14890 }, { "epoch": 94.90445859872611, "grad_norm": 1.7852954864501953, "learning_rate": 0.0001, "loss": 0.0485, "step": 14900 }, { "epoch": 94.96815286624204, "grad_norm": 1.5002213716506958, "learning_rate": 0.0001, "loss": 0.0468, "step": 14910 }, { "epoch": 95.03184713375796, "grad_norm": 1.4567830562591553, "learning_rate": 0.0001, "loss": 0.0481, "step": 14920 }, { "epoch": 95.09554140127389, "grad_norm": 1.4434280395507812, "learning_rate": 0.0001, "loss": 0.0453, "step": 14930 }, { "epoch": 95.1592356687898, "grad_norm": 1.6624534130096436, "learning_rate": 0.0001, "loss": 0.0468, "step": 14940 }, { "epoch": 95.22292993630573, "grad_norm": 1.5593342781066895, "learning_rate": 0.0001, "loss": 0.0484, "step": 14950 }, { "epoch": 95.28662420382166, "grad_norm": 1.444202184677124, "learning_rate": 0.0001, "loss": 0.0467, "step": 14960 }, { "epoch": 95.35031847133757, "grad_norm": 1.6704227924346924, "learning_rate": 0.0001, "loss": 0.0497, "step": 14970 }, { "epoch": 95.4140127388535, "grad_norm": 1.713295340538025, "learning_rate": 0.0001, "loss": 0.0501, "step": 14980 }, { "epoch": 95.47770700636943, "grad_norm": 1.6443310976028442, "learning_rate": 0.0001, "loss": 0.0492, "step": 14990 }, { "epoch": 95.54140127388536, "grad_norm": 1.6062891483306885, "learning_rate": 0.0001, "loss": 0.046, "step": 15000 }, { "epoch": 95.60509554140127, "grad_norm": 2.3452529907226562, "learning_rate": 0.0001, "loss": 0.0479, "step": 15010 }, { "epoch": 95.6687898089172, "grad_norm": 1.8115895986557007, "learning_rate": 0.0001, "loss": 0.0465, "step": 15020 }, { "epoch": 95.73248407643312, "grad_norm": 1.895035982131958, "learning_rate": 0.0001, "loss": 0.0448, "step": 15030 }, { "epoch": 95.79617834394904, "grad_norm": 1.719176173210144, "learning_rate": 0.0001, "loss": 0.0467, "step": 15040 }, { "epoch": 95.85987261146497, "grad_norm": 1.577064037322998, "learning_rate": 0.0001, "loss": 0.0472, "step": 15050 }, { "epoch": 95.92356687898089, "grad_norm": 1.4504649639129639, "learning_rate": 0.0001, "loss": 0.0446, "step": 15060 }, { "epoch": 95.98726114649682, "grad_norm": 1.6175682544708252, "learning_rate": 0.0001, "loss": 0.0488, "step": 15070 }, { "epoch": 96.05095541401273, "grad_norm": 1.5720083713531494, "learning_rate": 0.0001, "loss": 0.0429, "step": 15080 }, { "epoch": 96.11464968152866, "grad_norm": 1.9701557159423828, "learning_rate": 0.0001, "loss": 0.0476, "step": 15090 }, { "epoch": 96.17834394904459, "grad_norm": 1.5687333345413208, "learning_rate": 0.0001, "loss": 0.048, "step": 15100 }, { "epoch": 96.24203821656052, "grad_norm": 1.7455253601074219, "learning_rate": 0.0001, "loss": 0.0477, "step": 15110 }, { "epoch": 96.30573248407643, "grad_norm": 1.6968508958816528, "learning_rate": 0.0001, "loss": 0.046, "step": 15120 }, { "epoch": 96.36942675159236, "grad_norm": 1.586462140083313, "learning_rate": 0.0001, "loss": 0.0458, "step": 15130 }, { "epoch": 96.43312101910828, "grad_norm": 1.7619937658309937, "learning_rate": 0.0001, "loss": 0.0467, "step": 15140 }, { "epoch": 96.4968152866242, "grad_norm": 1.6902202367782593, "learning_rate": 0.0001, "loss": 0.05, "step": 15150 }, { "epoch": 96.56050955414013, "grad_norm": 1.8143349885940552, "learning_rate": 0.0001, "loss": 0.0486, "step": 15160 }, { "epoch": 96.62420382165605, "grad_norm": 1.7370126247406006, "learning_rate": 0.0001, "loss": 0.0504, "step": 15170 }, { "epoch": 96.68789808917198, "grad_norm": 1.6371523141860962, "learning_rate": 0.0001, "loss": 0.0447, "step": 15180 }, { "epoch": 96.7515923566879, "grad_norm": 2.2727279663085938, "learning_rate": 0.0001, "loss": 0.0503, "step": 15190 }, { "epoch": 96.81528662420382, "grad_norm": 1.8998866081237793, "learning_rate": 0.0001, "loss": 0.0443, "step": 15200 }, { "epoch": 96.87898089171975, "grad_norm": 1.846635341644287, "learning_rate": 0.0001, "loss": 0.0493, "step": 15210 }, { "epoch": 96.94267515923566, "grad_norm": 2.127211570739746, "learning_rate": 0.0001, "loss": 0.0456, "step": 15220 }, { "epoch": 97.00636942675159, "grad_norm": 1.7659647464752197, "learning_rate": 0.0001, "loss": 0.0469, "step": 15230 }, { "epoch": 97.07006369426752, "grad_norm": 1.6807130575180054, "learning_rate": 0.0001, "loss": 0.0481, "step": 15240 }, { "epoch": 97.13375796178345, "grad_norm": 1.4129340648651123, "learning_rate": 0.0001, "loss": 0.0469, "step": 15250 }, { "epoch": 97.19745222929936, "grad_norm": 1.4282890558242798, "learning_rate": 0.0001, "loss": 0.05, "step": 15260 }, { "epoch": 97.26114649681529, "grad_norm": 1.9211212396621704, "learning_rate": 0.0001, "loss": 0.0526, "step": 15270 }, { "epoch": 97.32484076433121, "grad_norm": 1.9991379976272583, "learning_rate": 0.0001, "loss": 0.0501, "step": 15280 }, { "epoch": 97.38853503184713, "grad_norm": 1.4237842559814453, "learning_rate": 0.0001, "loss": 0.0484, "step": 15290 }, { "epoch": 97.45222929936305, "grad_norm": 1.4732425212860107, "learning_rate": 0.0001, "loss": 0.0505, "step": 15300 }, { "epoch": 97.51592356687898, "grad_norm": 1.6646931171417236, "learning_rate": 0.0001, "loss": 0.0461, "step": 15310 }, { "epoch": 97.57961783439491, "grad_norm": 1.4995113611221313, "learning_rate": 0.0001, "loss": 0.047, "step": 15320 }, { "epoch": 97.64331210191082, "grad_norm": 1.4406594038009644, "learning_rate": 0.0001, "loss": 0.0482, "step": 15330 }, { "epoch": 97.70700636942675, "grad_norm": 1.506616473197937, "learning_rate": 0.0001, "loss": 0.0473, "step": 15340 }, { "epoch": 97.77070063694268, "grad_norm": 1.6232056617736816, "learning_rate": 0.0001, "loss": 0.0523, "step": 15350 }, { "epoch": 97.8343949044586, "grad_norm": 1.5562934875488281, "learning_rate": 0.0001, "loss": 0.05, "step": 15360 }, { "epoch": 97.89808917197452, "grad_norm": 1.3116145133972168, "learning_rate": 0.0001, "loss": 0.0449, "step": 15370 }, { "epoch": 97.96178343949045, "grad_norm": 2.4029529094696045, "learning_rate": 0.0001, "loss": 0.054, "step": 15380 }, { "epoch": 98.02547770700637, "grad_norm": 2.2698001861572266, "learning_rate": 0.0001, "loss": 0.0484, "step": 15390 }, { "epoch": 98.08917197452229, "grad_norm": 2.2189595699310303, "learning_rate": 0.0001, "loss": 0.0469, "step": 15400 }, { "epoch": 98.15286624203821, "grad_norm": 2.3321213722229004, "learning_rate": 0.0001, "loss": 0.0518, "step": 15410 }, { "epoch": 98.21656050955414, "grad_norm": 2.2914512157440186, "learning_rate": 0.0001, "loss": 0.0457, "step": 15420 }, { "epoch": 98.28025477707007, "grad_norm": 1.9939966201782227, "learning_rate": 0.0001, "loss": 0.0441, "step": 15430 }, { "epoch": 98.34394904458598, "grad_norm": 2.1371543407440186, "learning_rate": 0.0001, "loss": 0.0455, "step": 15440 }, { "epoch": 98.40764331210191, "grad_norm": 1.8496835231781006, "learning_rate": 0.0001, "loss": 0.0443, "step": 15450 }, { "epoch": 98.47133757961784, "grad_norm": 1.543355941772461, "learning_rate": 0.0001, "loss": 0.0469, "step": 15460 }, { "epoch": 98.53503184713375, "grad_norm": 1.7601346969604492, "learning_rate": 0.0001, "loss": 0.0457, "step": 15470 }, { "epoch": 98.59872611464968, "grad_norm": 1.7548848390579224, "learning_rate": 0.0001, "loss": 0.0456, "step": 15480 }, { "epoch": 98.6624203821656, "grad_norm": 1.6936813592910767, "learning_rate": 0.0001, "loss": 0.046, "step": 15490 }, { "epoch": 98.72611464968153, "grad_norm": 1.440000057220459, "learning_rate": 0.0001, "loss": 0.0447, "step": 15500 }, { "epoch": 98.78980891719745, "grad_norm": 1.7370730638504028, "learning_rate": 0.0001, "loss": 0.0482, "step": 15510 }, { "epoch": 98.85350318471338, "grad_norm": 1.5726451873779297, "learning_rate": 0.0001, "loss": 0.0501, "step": 15520 }, { "epoch": 98.9171974522293, "grad_norm": 1.5720367431640625, "learning_rate": 0.0001, "loss": 0.045, "step": 15530 }, { "epoch": 98.98089171974522, "grad_norm": 1.6249558925628662, "learning_rate": 0.0001, "loss": 0.0454, "step": 15540 }, { "epoch": 99.04458598726114, "grad_norm": 1.496238112449646, "learning_rate": 0.0001, "loss": 0.0461, "step": 15550 }, { "epoch": 99.10828025477707, "grad_norm": 1.549331545829773, "learning_rate": 0.0001, "loss": 0.0421, "step": 15560 }, { "epoch": 99.171974522293, "grad_norm": 1.604657769203186, "learning_rate": 0.0001, "loss": 0.0476, "step": 15570 }, { "epoch": 99.23566878980891, "grad_norm": 1.5029650926589966, "learning_rate": 0.0001, "loss": 0.0481, "step": 15580 }, { "epoch": 99.29936305732484, "grad_norm": 1.8527872562408447, "learning_rate": 0.0001, "loss": 0.0489, "step": 15590 }, { "epoch": 99.36305732484077, "grad_norm": 1.7510136365890503, "learning_rate": 0.0001, "loss": 0.0433, "step": 15600 }, { "epoch": 99.4267515923567, "grad_norm": 1.8524667024612427, "learning_rate": 0.0001, "loss": 0.0486, "step": 15610 }, { "epoch": 99.49044585987261, "grad_norm": 1.998898983001709, "learning_rate": 0.0001, "loss": 0.0454, "step": 15620 }, { "epoch": 99.55414012738854, "grad_norm": 2.3312461376190186, "learning_rate": 0.0001, "loss": 0.0469, "step": 15630 }, { "epoch": 99.61783439490446, "grad_norm": 1.7812986373901367, "learning_rate": 0.0001, "loss": 0.0491, "step": 15640 }, { "epoch": 99.68152866242038, "grad_norm": 1.474155306816101, "learning_rate": 0.0001, "loss": 0.0478, "step": 15650 }, { "epoch": 99.7452229299363, "grad_norm": 1.6731098890304565, "learning_rate": 0.0001, "loss": 0.0461, "step": 15660 }, { "epoch": 99.80891719745223, "grad_norm": 1.540347695350647, "learning_rate": 0.0001, "loss": 0.0445, "step": 15670 }, { "epoch": 99.87261146496816, "grad_norm": 1.2254998683929443, "learning_rate": 0.0001, "loss": 0.0452, "step": 15680 }, { "epoch": 99.93630573248407, "grad_norm": 1.6239287853240967, "learning_rate": 0.0001, "loss": 0.0459, "step": 15690 }, { "epoch": 100.0, "grad_norm": 1.6548913717269897, "learning_rate": 0.0001, "loss": 0.0474, "step": 15700 }, { "epoch": 100.06369426751593, "grad_norm": 1.3748579025268555, "learning_rate": 0.0001, "loss": 0.0462, "step": 15710 }, { "epoch": 100.12738853503184, "grad_norm": 1.6290733814239502, "learning_rate": 0.0001, "loss": 0.0473, "step": 15720 }, { "epoch": 100.19108280254777, "grad_norm": 1.5603028535842896, "learning_rate": 0.0001, "loss": 0.0451, "step": 15730 }, { "epoch": 100.2547770700637, "grad_norm": 1.6656951904296875, "learning_rate": 0.0001, "loss": 0.0488, "step": 15740 }, { "epoch": 100.31847133757962, "grad_norm": 1.533202052116394, "learning_rate": 0.0001, "loss": 0.0467, "step": 15750 }, { "epoch": 100.38216560509554, "grad_norm": 1.9591790437698364, "learning_rate": 0.0001, "loss": 0.047, "step": 15760 }, { "epoch": 100.44585987261146, "grad_norm": 1.7327525615692139, "learning_rate": 0.0001, "loss": 0.0531, "step": 15770 }, { "epoch": 100.50955414012739, "grad_norm": 1.8676841259002686, "learning_rate": 0.0001, "loss": 0.0516, "step": 15780 }, { "epoch": 100.5732484076433, "grad_norm": 1.3096623420715332, "learning_rate": 0.0001, "loss": 0.0501, "step": 15790 }, { "epoch": 100.63694267515923, "grad_norm": 1.6795815229415894, "learning_rate": 0.0001, "loss": 0.048, "step": 15800 }, { "epoch": 100.70063694267516, "grad_norm": 1.485032320022583, "learning_rate": 0.0001, "loss": 0.0523, "step": 15810 }, { "epoch": 100.76433121019109, "grad_norm": 1.5108792781829834, "learning_rate": 0.0001, "loss": 0.0459, "step": 15820 }, { "epoch": 100.828025477707, "grad_norm": 1.4687798023223877, "learning_rate": 0.0001, "loss": 0.0496, "step": 15830 }, { "epoch": 100.89171974522293, "grad_norm": 1.3876405954360962, "learning_rate": 0.0001, "loss": 0.0446, "step": 15840 }, { "epoch": 100.95541401273886, "grad_norm": 1.1996430158615112, "learning_rate": 0.0001, "loss": 0.0476, "step": 15850 }, { "epoch": 101.01910828025478, "grad_norm": 1.4326902627944946, "learning_rate": 0.0001, "loss": 0.0469, "step": 15860 }, { "epoch": 101.0828025477707, "grad_norm": 1.4660817384719849, "learning_rate": 0.0001, "loss": 0.0453, "step": 15870 }, { "epoch": 101.14649681528662, "grad_norm": 1.3746232986450195, "learning_rate": 0.0001, "loss": 0.0483, "step": 15880 }, { "epoch": 101.21019108280255, "grad_norm": 1.3414417505264282, "learning_rate": 0.0001, "loss": 0.0488, "step": 15890 }, { "epoch": 101.27388535031847, "grad_norm": 1.3042035102844238, "learning_rate": 0.0001, "loss": 0.0519, "step": 15900 }, { "epoch": 101.3375796178344, "grad_norm": 1.3823853731155396, "learning_rate": 0.0001, "loss": 0.0475, "step": 15910 }, { "epoch": 101.40127388535032, "grad_norm": 1.5211228132247925, "learning_rate": 0.0001, "loss": 0.0507, "step": 15920 }, { "epoch": 101.46496815286625, "grad_norm": 1.9752744436264038, "learning_rate": 0.0001, "loss": 0.0506, "step": 15930 }, { "epoch": 101.52866242038216, "grad_norm": 1.2799954414367676, "learning_rate": 0.0001, "loss": 0.0473, "step": 15940 }, { "epoch": 101.59235668789809, "grad_norm": 1.6192827224731445, "learning_rate": 0.0001, "loss": 0.0481, "step": 15950 }, { "epoch": 101.65605095541402, "grad_norm": 1.4787020683288574, "learning_rate": 0.0001, "loss": 0.0526, "step": 15960 }, { "epoch": 101.71974522292993, "grad_norm": 1.2974525690078735, "learning_rate": 0.0001, "loss": 0.0526, "step": 15970 }, { "epoch": 101.78343949044586, "grad_norm": 1.4257175922393799, "learning_rate": 0.0001, "loss": 0.0457, "step": 15980 }, { "epoch": 101.84713375796179, "grad_norm": 1.2425320148468018, "learning_rate": 0.0001, "loss": 0.0464, "step": 15990 }, { "epoch": 101.91082802547771, "grad_norm": 1.4462467432022095, "learning_rate": 0.0001, "loss": 0.0483, "step": 16000 }, { "epoch": 101.97452229299363, "grad_norm": 1.423478603363037, "learning_rate": 0.0001, "loss": 0.0512, "step": 16010 }, { "epoch": 102.03821656050955, "grad_norm": 1.453244686126709, "learning_rate": 0.0001, "loss": 0.0464, "step": 16020 }, { "epoch": 102.10191082802548, "grad_norm": 1.4474107027053833, "learning_rate": 0.0001, "loss": 0.0445, "step": 16030 }, { "epoch": 102.1656050955414, "grad_norm": 1.587878704071045, "learning_rate": 0.0001, "loss": 0.0483, "step": 16040 }, { "epoch": 102.22929936305732, "grad_norm": 1.6052488088607788, "learning_rate": 0.0001, "loss": 0.0447, "step": 16050 }, { "epoch": 102.29299363057325, "grad_norm": 1.356554627418518, "learning_rate": 0.0001, "loss": 0.0476, "step": 16060 }, { "epoch": 102.35668789808918, "grad_norm": 1.244002103805542, "learning_rate": 0.0001, "loss": 0.0467, "step": 16070 }, { "epoch": 102.42038216560509, "grad_norm": 1.4938507080078125, "learning_rate": 0.0001, "loss": 0.0443, "step": 16080 }, { "epoch": 102.48407643312102, "grad_norm": 1.844598650932312, "learning_rate": 0.0001, "loss": 0.0446, "step": 16090 }, { "epoch": 102.54777070063695, "grad_norm": 1.5406337976455688, "learning_rate": 0.0001, "loss": 0.0481, "step": 16100 }, { "epoch": 102.61146496815287, "grad_norm": 1.491247534751892, "learning_rate": 0.0001, "loss": 0.0477, "step": 16110 }, { "epoch": 102.67515923566879, "grad_norm": 1.4950711727142334, "learning_rate": 0.0001, "loss": 0.0444, "step": 16120 }, { "epoch": 102.73885350318471, "grad_norm": 1.532497525215149, "learning_rate": 0.0001, "loss": 0.0438, "step": 16130 }, { "epoch": 102.80254777070064, "grad_norm": 1.4710065126419067, "learning_rate": 0.0001, "loss": 0.0485, "step": 16140 }, { "epoch": 102.86624203821655, "grad_norm": 1.5599687099456787, "learning_rate": 0.0001, "loss": 0.0448, "step": 16150 }, { "epoch": 102.92993630573248, "grad_norm": 1.254043698310852, "learning_rate": 0.0001, "loss": 0.0484, "step": 16160 }, { "epoch": 102.99363057324841, "grad_norm": 1.4444880485534668, "learning_rate": 0.0001, "loss": 0.0504, "step": 16170 }, { "epoch": 103.05732484076434, "grad_norm": 1.4160369634628296, "learning_rate": 0.0001, "loss": 0.0455, "step": 16180 }, { "epoch": 103.12101910828025, "grad_norm": 1.5365864038467407, "learning_rate": 0.0001, "loss": 0.0477, "step": 16190 }, { "epoch": 103.18471337579618, "grad_norm": 1.8037139177322388, "learning_rate": 0.0001, "loss": 0.0479, "step": 16200 }, { "epoch": 103.2484076433121, "grad_norm": 1.5874030590057373, "learning_rate": 0.0001, "loss": 0.0462, "step": 16210 }, { "epoch": 103.31210191082802, "grad_norm": 1.608213186264038, "learning_rate": 0.0001, "loss": 0.048, "step": 16220 }, { "epoch": 103.37579617834395, "grad_norm": 1.7731430530548096, "learning_rate": 0.0001, "loss": 0.0483, "step": 16230 }, { "epoch": 103.43949044585987, "grad_norm": 2.072140693664551, "learning_rate": 0.0001, "loss": 0.0453, "step": 16240 }, { "epoch": 103.5031847133758, "grad_norm": 1.9810233116149902, "learning_rate": 0.0001, "loss": 0.0466, "step": 16250 }, { "epoch": 103.56687898089172, "grad_norm": 1.7277864217758179, "learning_rate": 0.0001, "loss": 0.0489, "step": 16260 }, { "epoch": 103.63057324840764, "grad_norm": 1.5338349342346191, "learning_rate": 0.0001, "loss": 0.0443, "step": 16270 }, { "epoch": 103.69426751592357, "grad_norm": 1.8362113237380981, "learning_rate": 0.0001, "loss": 0.0451, "step": 16280 }, { "epoch": 103.75796178343948, "grad_norm": 1.5015811920166016, "learning_rate": 0.0001, "loss": 0.0465, "step": 16290 }, { "epoch": 103.82165605095541, "grad_norm": 1.484157919883728, "learning_rate": 0.0001, "loss": 0.0487, "step": 16300 }, { "epoch": 103.88535031847134, "grad_norm": 1.6300702095031738, "learning_rate": 0.0001, "loss": 0.0476, "step": 16310 }, { "epoch": 103.94904458598727, "grad_norm": 1.6016857624053955, "learning_rate": 0.0001, "loss": 0.0486, "step": 16320 }, { "epoch": 104.01273885350318, "grad_norm": 1.780245065689087, "learning_rate": 0.0001, "loss": 0.0446, "step": 16330 }, { "epoch": 104.07643312101911, "grad_norm": 1.728497862815857, "learning_rate": 0.0001, "loss": 0.0478, "step": 16340 }, { "epoch": 104.14012738853503, "grad_norm": 1.6085543632507324, "learning_rate": 0.0001, "loss": 0.0479, "step": 16350 }, { "epoch": 104.20382165605096, "grad_norm": 1.4808381795883179, "learning_rate": 0.0001, "loss": 0.0448, "step": 16360 }, { "epoch": 104.26751592356688, "grad_norm": 1.5553019046783447, "learning_rate": 0.0001, "loss": 0.0454, "step": 16370 }, { "epoch": 104.3312101910828, "grad_norm": 1.6286396980285645, "learning_rate": 0.0001, "loss": 0.0449, "step": 16380 }, { "epoch": 104.39490445859873, "grad_norm": 1.3496997356414795, "learning_rate": 0.0001, "loss": 0.0473, "step": 16390 }, { "epoch": 104.45859872611464, "grad_norm": 1.3056080341339111, "learning_rate": 0.0001, "loss": 0.0424, "step": 16400 }, { "epoch": 104.52229299363057, "grad_norm": 0.9992454648017883, "learning_rate": 0.0001, "loss": 0.0471, "step": 16410 }, { "epoch": 104.5859872611465, "grad_norm": 1.4512847661972046, "learning_rate": 0.0001, "loss": 0.0481, "step": 16420 }, { "epoch": 104.64968152866243, "grad_norm": 1.3293135166168213, "learning_rate": 0.0001, "loss": 0.0522, "step": 16430 }, { "epoch": 104.71337579617834, "grad_norm": 1.363348126411438, "learning_rate": 0.0001, "loss": 0.0482, "step": 16440 }, { "epoch": 104.77707006369427, "grad_norm": 1.5697860717773438, "learning_rate": 0.0001, "loss": 0.0508, "step": 16450 }, { "epoch": 104.8407643312102, "grad_norm": 1.5477944612503052, "learning_rate": 0.0001, "loss": 0.0492, "step": 16460 }, { "epoch": 104.90445859872611, "grad_norm": 1.5148725509643555, "learning_rate": 0.0001, "loss": 0.0506, "step": 16470 }, { "epoch": 104.96815286624204, "grad_norm": 1.163575291633606, "learning_rate": 0.0001, "loss": 0.0498, "step": 16480 }, { "epoch": 105.03184713375796, "grad_norm": 1.1789816617965698, "learning_rate": 0.0001, "loss": 0.0476, "step": 16490 }, { "epoch": 105.09554140127389, "grad_norm": 1.2433836460113525, "learning_rate": 0.0001, "loss": 0.0456, "step": 16500 }, { "epoch": 105.1592356687898, "grad_norm": 1.1905654668807983, "learning_rate": 0.0001, "loss": 0.0501, "step": 16510 }, { "epoch": 105.22292993630573, "grad_norm": 1.3986539840698242, "learning_rate": 0.0001, "loss": 0.0506, "step": 16520 }, { "epoch": 105.28662420382166, "grad_norm": 1.423036813735962, "learning_rate": 0.0001, "loss": 0.0477, "step": 16530 }, { "epoch": 105.35031847133757, "grad_norm": 1.1840684413909912, "learning_rate": 0.0001, "loss": 0.0464, "step": 16540 }, { "epoch": 105.4140127388535, "grad_norm": 1.2519854307174683, "learning_rate": 0.0001, "loss": 0.0466, "step": 16550 }, { "epoch": 105.47770700636943, "grad_norm": 1.1667248010635376, "learning_rate": 0.0001, "loss": 0.0479, "step": 16560 }, { "epoch": 105.54140127388536, "grad_norm": 1.6643168926239014, "learning_rate": 0.0001, "loss": 0.0481, "step": 16570 }, { "epoch": 105.60509554140127, "grad_norm": 1.7580647468566895, "learning_rate": 0.0001, "loss": 0.0505, "step": 16580 }, { "epoch": 105.6687898089172, "grad_norm": 1.919137716293335, "learning_rate": 0.0001, "loss": 0.0472, "step": 16590 }, { "epoch": 105.73248407643312, "grad_norm": 1.802807331085205, "learning_rate": 0.0001, "loss": 0.0465, "step": 16600 }, { "epoch": 105.79617834394904, "grad_norm": 1.8588114976882935, "learning_rate": 0.0001, "loss": 0.0439, "step": 16610 }, { "epoch": 105.85987261146497, "grad_norm": 1.7717069387435913, "learning_rate": 0.0001, "loss": 0.0454, "step": 16620 }, { "epoch": 105.92356687898089, "grad_norm": 1.589817762374878, "learning_rate": 0.0001, "loss": 0.045, "step": 16630 }, { "epoch": 105.98726114649682, "grad_norm": 1.2747911214828491, "learning_rate": 0.0001, "loss": 0.0456, "step": 16640 }, { "epoch": 106.05095541401273, "grad_norm": 1.6424005031585693, "learning_rate": 0.0001, "loss": 0.0406, "step": 16650 }, { "epoch": 106.11464968152866, "grad_norm": 1.7529692649841309, "learning_rate": 0.0001, "loss": 0.0462, "step": 16660 }, { "epoch": 106.17834394904459, "grad_norm": 1.5701767206192017, "learning_rate": 0.0001, "loss": 0.0438, "step": 16670 }, { "epoch": 106.24203821656052, "grad_norm": 1.6088290214538574, "learning_rate": 0.0001, "loss": 0.0438, "step": 16680 }, { "epoch": 106.30573248407643, "grad_norm": 1.842284917831421, "learning_rate": 0.0001, "loss": 0.0445, "step": 16690 }, { "epoch": 106.36942675159236, "grad_norm": 1.5366122722625732, "learning_rate": 0.0001, "loss": 0.0475, "step": 16700 }, { "epoch": 106.43312101910828, "grad_norm": 1.5518441200256348, "learning_rate": 0.0001, "loss": 0.047, "step": 16710 }, { "epoch": 106.4968152866242, "grad_norm": 1.4746352434158325, "learning_rate": 0.0001, "loss": 0.0489, "step": 16720 }, { "epoch": 106.56050955414013, "grad_norm": 1.345232605934143, "learning_rate": 0.0001, "loss": 0.0437, "step": 16730 }, { "epoch": 106.62420382165605, "grad_norm": 1.7930665016174316, "learning_rate": 0.0001, "loss": 0.0488, "step": 16740 }, { "epoch": 106.68789808917198, "grad_norm": 1.734494924545288, "learning_rate": 0.0001, "loss": 0.0474, "step": 16750 }, { "epoch": 106.7515923566879, "grad_norm": 2.3241984844207764, "learning_rate": 0.0001, "loss": 0.0433, "step": 16760 }, { "epoch": 106.81528662420382, "grad_norm": 1.6837457418441772, "learning_rate": 0.0001, "loss": 0.0475, "step": 16770 }, { "epoch": 106.87898089171975, "grad_norm": 1.709365725517273, "learning_rate": 0.0001, "loss": 0.0458, "step": 16780 }, { "epoch": 106.94267515923566, "grad_norm": 1.6238847970962524, "learning_rate": 0.0001, "loss": 0.0454, "step": 16790 }, { "epoch": 107.00636942675159, "grad_norm": 1.5170314311981201, "learning_rate": 0.0001, "loss": 0.0451, "step": 16800 }, { "epoch": 107.07006369426752, "grad_norm": 1.663462519645691, "learning_rate": 0.0001, "loss": 0.0449, "step": 16810 }, { "epoch": 107.13375796178345, "grad_norm": 1.4811636209487915, "learning_rate": 0.0001, "loss": 0.0428, "step": 16820 }, { "epoch": 107.19745222929936, "grad_norm": 1.779457926750183, "learning_rate": 0.0001, "loss": 0.0417, "step": 16830 }, { "epoch": 107.26114649681529, "grad_norm": 1.3897655010223389, "learning_rate": 0.0001, "loss": 0.0434, "step": 16840 }, { "epoch": 107.32484076433121, "grad_norm": 1.5177292823791504, "learning_rate": 0.0001, "loss": 0.0449, "step": 16850 }, { "epoch": 107.38853503184713, "grad_norm": 1.8491867780685425, "learning_rate": 0.0001, "loss": 0.0446, "step": 16860 }, { "epoch": 107.45222929936305, "grad_norm": 1.6017485857009888, "learning_rate": 0.0001, "loss": 0.0429, "step": 16870 }, { "epoch": 107.51592356687898, "grad_norm": 1.398202657699585, "learning_rate": 0.0001, "loss": 0.0451, "step": 16880 }, { "epoch": 107.57961783439491, "grad_norm": 1.5800204277038574, "learning_rate": 0.0001, "loss": 0.0445, "step": 16890 }, { "epoch": 107.64331210191082, "grad_norm": 1.6079550981521606, "learning_rate": 0.0001, "loss": 0.0485, "step": 16900 }, { "epoch": 107.70700636942675, "grad_norm": 1.3223581314086914, "learning_rate": 0.0001, "loss": 0.0443, "step": 16910 }, { "epoch": 107.77070063694268, "grad_norm": 1.575873613357544, "learning_rate": 0.0001, "loss": 0.047, "step": 16920 }, { "epoch": 107.8343949044586, "grad_norm": 1.531509280204773, "learning_rate": 0.0001, "loss": 0.0421, "step": 16930 }, { "epoch": 107.89808917197452, "grad_norm": 1.3847171068191528, "learning_rate": 0.0001, "loss": 0.0462, "step": 16940 }, { "epoch": 107.96178343949045, "grad_norm": 1.4084432125091553, "learning_rate": 0.0001, "loss": 0.0493, "step": 16950 }, { "epoch": 108.02547770700637, "grad_norm": 1.32808518409729, "learning_rate": 0.0001, "loss": 0.0454, "step": 16960 }, { "epoch": 108.08917197452229, "grad_norm": 1.4288136959075928, "learning_rate": 0.0001, "loss": 0.0446, "step": 16970 }, { "epoch": 108.15286624203821, "grad_norm": 1.4444465637207031, "learning_rate": 0.0001, "loss": 0.0488, "step": 16980 }, { "epoch": 108.21656050955414, "grad_norm": 1.467020869255066, "learning_rate": 0.0001, "loss": 0.0474, "step": 16990 }, { "epoch": 108.28025477707007, "grad_norm": 1.3442180156707764, "learning_rate": 0.0001, "loss": 0.0466, "step": 17000 }, { "epoch": 108.34394904458598, "grad_norm": 1.065587043762207, "learning_rate": 0.0001, "loss": 0.0466, "step": 17010 }, { "epoch": 108.40764331210191, "grad_norm": 1.4549305438995361, "learning_rate": 0.0001, "loss": 0.05, "step": 17020 }, { "epoch": 108.47133757961784, "grad_norm": 1.3704864978790283, "learning_rate": 0.0001, "loss": 0.0462, "step": 17030 }, { "epoch": 108.53503184713375, "grad_norm": 1.2309269905090332, "learning_rate": 0.0001, "loss": 0.0456, "step": 17040 }, { "epoch": 108.59872611464968, "grad_norm": 1.421269178390503, "learning_rate": 0.0001, "loss": 0.0439, "step": 17050 }, { "epoch": 108.6624203821656, "grad_norm": 1.5236519575119019, "learning_rate": 0.0001, "loss": 0.0471, "step": 17060 }, { "epoch": 108.72611464968153, "grad_norm": 1.350536584854126, "learning_rate": 0.0001, "loss": 0.0487, "step": 17070 }, { "epoch": 108.78980891719745, "grad_norm": 1.5624173879623413, "learning_rate": 0.0001, "loss": 0.0476, "step": 17080 }, { "epoch": 108.85350318471338, "grad_norm": 1.5134835243225098, "learning_rate": 0.0001, "loss": 0.0431, "step": 17090 }, { "epoch": 108.9171974522293, "grad_norm": 1.3221451044082642, "learning_rate": 0.0001, "loss": 0.0464, "step": 17100 }, { "epoch": 108.98089171974522, "grad_norm": 1.512399435043335, "learning_rate": 0.0001, "loss": 0.0496, "step": 17110 }, { "epoch": 109.04458598726114, "grad_norm": 1.5800961256027222, "learning_rate": 0.0001, "loss": 0.047, "step": 17120 }, { "epoch": 109.10828025477707, "grad_norm": 1.272968053817749, "learning_rate": 0.0001, "loss": 0.046, "step": 17130 }, { "epoch": 109.171974522293, "grad_norm": 1.428163766860962, "learning_rate": 0.0001, "loss": 0.0437, "step": 17140 }, { "epoch": 109.23566878980891, "grad_norm": 1.434929370880127, "learning_rate": 0.0001, "loss": 0.0462, "step": 17150 }, { "epoch": 109.29936305732484, "grad_norm": 1.4337564706802368, "learning_rate": 0.0001, "loss": 0.0483, "step": 17160 }, { "epoch": 109.36305732484077, "grad_norm": 1.2362124919891357, "learning_rate": 0.0001, "loss": 0.0449, "step": 17170 }, { "epoch": 109.4267515923567, "grad_norm": 1.225395917892456, "learning_rate": 0.0001, "loss": 0.0441, "step": 17180 }, { "epoch": 109.49044585987261, "grad_norm": 1.324584722518921, "learning_rate": 0.0001, "loss": 0.0449, "step": 17190 }, { "epoch": 109.55414012738854, "grad_norm": 1.4724845886230469, "learning_rate": 0.0001, "loss": 0.049, "step": 17200 }, { "epoch": 109.61783439490446, "grad_norm": 1.5670366287231445, "learning_rate": 0.0001, "loss": 0.0425, "step": 17210 }, { "epoch": 109.68152866242038, "grad_norm": 1.4772498607635498, "learning_rate": 0.0001, "loss": 0.045, "step": 17220 }, { "epoch": 109.7452229299363, "grad_norm": 1.1654154062271118, "learning_rate": 0.0001, "loss": 0.0431, "step": 17230 }, { "epoch": 109.80891719745223, "grad_norm": 1.3449667692184448, "learning_rate": 0.0001, "loss": 0.0445, "step": 17240 }, { "epoch": 109.87261146496816, "grad_norm": 1.7888953685760498, "learning_rate": 0.0001, "loss": 0.0468, "step": 17250 }, { "epoch": 109.93630573248407, "grad_norm": 1.549422025680542, "learning_rate": 0.0001, "loss": 0.0435, "step": 17260 }, { "epoch": 110.0, "grad_norm": 1.6525212526321411, "learning_rate": 0.0001, "loss": 0.0466, "step": 17270 }, { "epoch": 110.06369426751593, "grad_norm": 1.6411725282669067, "learning_rate": 0.0001, "loss": 0.0449, "step": 17280 }, { "epoch": 110.12738853503184, "grad_norm": 1.1626701354980469, "learning_rate": 0.0001, "loss": 0.0439, "step": 17290 }, { "epoch": 110.19108280254777, "grad_norm": 1.1558561325073242, "learning_rate": 0.0001, "loss": 0.0471, "step": 17300 }, { "epoch": 110.2547770700637, "grad_norm": 1.4721052646636963, "learning_rate": 0.0001, "loss": 0.0433, "step": 17310 }, { "epoch": 110.31847133757962, "grad_norm": 1.2657519578933716, "learning_rate": 0.0001, "loss": 0.0471, "step": 17320 }, { "epoch": 110.38216560509554, "grad_norm": 1.305860996246338, "learning_rate": 0.0001, "loss": 0.0429, "step": 17330 }, { "epoch": 110.44585987261146, "grad_norm": 1.339339256286621, "learning_rate": 0.0001, "loss": 0.0468, "step": 17340 }, { "epoch": 110.50955414012739, "grad_norm": 1.472556471824646, "learning_rate": 0.0001, "loss": 0.048, "step": 17350 }, { "epoch": 110.5732484076433, "grad_norm": 1.644423007965088, "learning_rate": 0.0001, "loss": 0.0484, "step": 17360 }, { "epoch": 110.63694267515923, "grad_norm": 1.737448811531067, "learning_rate": 0.0001, "loss": 0.048, "step": 17370 }, { "epoch": 110.70063694267516, "grad_norm": 1.3361574411392212, "learning_rate": 0.0001, "loss": 0.0487, "step": 17380 }, { "epoch": 110.76433121019109, "grad_norm": 1.6410844326019287, "learning_rate": 0.0001, "loss": 0.0458, "step": 17390 }, { "epoch": 110.828025477707, "grad_norm": 1.5595039129257202, "learning_rate": 0.0001, "loss": 0.0458, "step": 17400 }, { "epoch": 110.89171974522293, "grad_norm": 1.5095250606536865, "learning_rate": 0.0001, "loss": 0.05, "step": 17410 }, { "epoch": 110.95541401273886, "grad_norm": 1.1429041624069214, "learning_rate": 0.0001, "loss": 0.0484, "step": 17420 }, { "epoch": 111.01910828025478, "grad_norm": 1.4661078453063965, "learning_rate": 0.0001, "loss": 0.0453, "step": 17430 }, { "epoch": 111.0828025477707, "grad_norm": 1.4822161197662354, "learning_rate": 0.0001, "loss": 0.0463, "step": 17440 }, { "epoch": 111.14649681528662, "grad_norm": 1.1274099349975586, "learning_rate": 0.0001, "loss": 0.0439, "step": 17450 }, { "epoch": 111.21019108280255, "grad_norm": 1.3333534002304077, "learning_rate": 0.0001, "loss": 0.0446, "step": 17460 }, { "epoch": 111.27388535031847, "grad_norm": 1.21860671043396, "learning_rate": 0.0001, "loss": 0.0453, "step": 17470 }, { "epoch": 111.3375796178344, "grad_norm": 1.1547517776489258, "learning_rate": 0.0001, "loss": 0.0461, "step": 17480 }, { "epoch": 111.40127388535032, "grad_norm": 1.7086591720581055, "learning_rate": 0.0001, "loss": 0.0445, "step": 17490 }, { "epoch": 111.46496815286625, "grad_norm": 1.3174501657485962, "learning_rate": 0.0001, "loss": 0.0445, "step": 17500 }, { "epoch": 111.52866242038216, "grad_norm": 1.3817133903503418, "learning_rate": 0.0001, "loss": 0.0427, "step": 17510 }, { "epoch": 111.59235668789809, "grad_norm": 1.4661567211151123, "learning_rate": 0.0001, "loss": 0.0438, "step": 17520 }, { "epoch": 111.65605095541402, "grad_norm": 1.6230792999267578, "learning_rate": 0.0001, "loss": 0.0471, "step": 17530 }, { "epoch": 111.71974522292993, "grad_norm": 1.6550027132034302, "learning_rate": 0.0001, "loss": 0.048, "step": 17540 }, { "epoch": 111.78343949044586, "grad_norm": 1.4756301641464233, "learning_rate": 0.0001, "loss": 0.0491, "step": 17550 }, { "epoch": 111.84713375796179, "grad_norm": 1.4865578413009644, "learning_rate": 0.0001, "loss": 0.0443, "step": 17560 }, { "epoch": 111.91082802547771, "grad_norm": 1.5582572221755981, "learning_rate": 0.0001, "loss": 0.0448, "step": 17570 }, { "epoch": 111.97452229299363, "grad_norm": 1.5731662511825562, "learning_rate": 0.0001, "loss": 0.0447, "step": 17580 }, { "epoch": 112.03821656050955, "grad_norm": 1.4155879020690918, "learning_rate": 0.0001, "loss": 0.0412, "step": 17590 }, { "epoch": 112.10191082802548, "grad_norm": 1.4289518594741821, "learning_rate": 0.0001, "loss": 0.0406, "step": 17600 }, { "epoch": 112.1656050955414, "grad_norm": 1.495981216430664, "learning_rate": 0.0001, "loss": 0.0449, "step": 17610 }, { "epoch": 112.22929936305732, "grad_norm": 1.8035030364990234, "learning_rate": 0.0001, "loss": 0.0452, "step": 17620 }, { "epoch": 112.29299363057325, "grad_norm": 1.5493617057800293, "learning_rate": 0.0001, "loss": 0.0414, "step": 17630 }, { "epoch": 112.35668789808918, "grad_norm": 1.547324299812317, "learning_rate": 0.0001, "loss": 0.0452, "step": 17640 }, { "epoch": 112.42038216560509, "grad_norm": 1.506306767463684, "learning_rate": 0.0001, "loss": 0.0464, "step": 17650 }, { "epoch": 112.48407643312102, "grad_norm": 1.4001119136810303, "learning_rate": 0.0001, "loss": 0.0422, "step": 17660 }, { "epoch": 112.54777070063695, "grad_norm": 1.5176385641098022, "learning_rate": 0.0001, "loss": 0.0496, "step": 17670 }, { "epoch": 112.61146496815287, "grad_norm": 1.3710986375808716, "learning_rate": 0.0001, "loss": 0.0505, "step": 17680 }, { "epoch": 112.67515923566879, "grad_norm": 1.1120235919952393, "learning_rate": 0.0001, "loss": 0.0498, "step": 17690 }, { "epoch": 112.73885350318471, "grad_norm": 1.2851943969726562, "learning_rate": 0.0001, "loss": 0.0451, "step": 17700 }, { "epoch": 112.80254777070064, "grad_norm": 1.4374011754989624, "learning_rate": 0.0001, "loss": 0.0505, "step": 17710 }, { "epoch": 112.86624203821655, "grad_norm": 1.7765569686889648, "learning_rate": 0.0001, "loss": 0.0445, "step": 17720 }, { "epoch": 112.92993630573248, "grad_norm": 2.147254467010498, "learning_rate": 0.0001, "loss": 0.0463, "step": 17730 }, { "epoch": 112.99363057324841, "grad_norm": 1.902028203010559, "learning_rate": 0.0001, "loss": 0.0451, "step": 17740 }, { "epoch": 113.05732484076434, "grad_norm": 1.5182631015777588, "learning_rate": 0.0001, "loss": 0.045, "step": 17750 }, { "epoch": 113.12101910828025, "grad_norm": 1.4090394973754883, "learning_rate": 0.0001, "loss": 0.0471, "step": 17760 }, { "epoch": 113.18471337579618, "grad_norm": 1.3896172046661377, "learning_rate": 0.0001, "loss": 0.0464, "step": 17770 }, { "epoch": 113.2484076433121, "grad_norm": 1.196099877357483, "learning_rate": 0.0001, "loss": 0.0443, "step": 17780 }, { "epoch": 113.31210191082802, "grad_norm": 1.5249587297439575, "learning_rate": 0.0001, "loss": 0.0424, "step": 17790 }, { "epoch": 113.37579617834395, "grad_norm": 1.231996774673462, "learning_rate": 0.0001, "loss": 0.0466, "step": 17800 }, { "epoch": 113.43949044585987, "grad_norm": 1.1271660327911377, "learning_rate": 0.0001, "loss": 0.046, "step": 17810 }, { "epoch": 113.5031847133758, "grad_norm": 2.016371726989746, "learning_rate": 0.0001, "loss": 0.0446, "step": 17820 }, { "epoch": 113.56687898089172, "grad_norm": 1.6626328229904175, "learning_rate": 0.0001, "loss": 0.0462, "step": 17830 }, { "epoch": 113.63057324840764, "grad_norm": 1.5458223819732666, "learning_rate": 0.0001, "loss": 0.0451, "step": 17840 }, { "epoch": 113.69426751592357, "grad_norm": 1.5014735460281372, "learning_rate": 0.0001, "loss": 0.0406, "step": 17850 }, { "epoch": 113.75796178343948, "grad_norm": 1.4281985759735107, "learning_rate": 0.0001, "loss": 0.0442, "step": 17860 }, { "epoch": 113.82165605095541, "grad_norm": 1.2632782459259033, "learning_rate": 0.0001, "loss": 0.0472, "step": 17870 }, { "epoch": 113.88535031847134, "grad_norm": 1.4833216667175293, "learning_rate": 0.0001, "loss": 0.0468, "step": 17880 }, { "epoch": 113.94904458598727, "grad_norm": 1.2658504247665405, "learning_rate": 0.0001, "loss": 0.0479, "step": 17890 }, { "epoch": 114.01273885350318, "grad_norm": 1.194663405418396, "learning_rate": 0.0001, "loss": 0.0443, "step": 17900 }, { "epoch": 114.07643312101911, "grad_norm": 1.4869022369384766, "learning_rate": 0.0001, "loss": 0.0454, "step": 17910 }, { "epoch": 114.14012738853503, "grad_norm": 1.3944414854049683, "learning_rate": 0.0001, "loss": 0.0448, "step": 17920 }, { "epoch": 114.20382165605096, "grad_norm": 1.2497892379760742, "learning_rate": 0.0001, "loss": 0.0438, "step": 17930 }, { "epoch": 114.26751592356688, "grad_norm": 1.4510273933410645, "learning_rate": 0.0001, "loss": 0.0474, "step": 17940 }, { "epoch": 114.3312101910828, "grad_norm": 1.3743699789047241, "learning_rate": 0.0001, "loss": 0.0479, "step": 17950 }, { "epoch": 114.39490445859873, "grad_norm": 1.1981059312820435, "learning_rate": 0.0001, "loss": 0.0434, "step": 17960 }, { "epoch": 114.45859872611464, "grad_norm": 1.5566157102584839, "learning_rate": 0.0001, "loss": 0.0488, "step": 17970 }, { "epoch": 114.52229299363057, "grad_norm": 1.4602313041687012, "learning_rate": 0.0001, "loss": 0.0441, "step": 17980 }, { "epoch": 114.5859872611465, "grad_norm": 1.5802501440048218, "learning_rate": 0.0001, "loss": 0.0452, "step": 17990 }, { "epoch": 114.64968152866243, "grad_norm": 1.506340503692627, "learning_rate": 0.0001, "loss": 0.0438, "step": 18000 }, { "epoch": 114.71337579617834, "grad_norm": 1.5471497774124146, "learning_rate": 0.0001, "loss": 0.0406, "step": 18010 }, { "epoch": 114.77707006369427, "grad_norm": 1.5858607292175293, "learning_rate": 0.0001, "loss": 0.0409, "step": 18020 }, { "epoch": 114.8407643312102, "grad_norm": 1.1127945184707642, "learning_rate": 0.0001, "loss": 0.0427, "step": 18030 }, { "epoch": 114.90445859872611, "grad_norm": 1.3597806692123413, "learning_rate": 0.0001, "loss": 0.0454, "step": 18040 }, { "epoch": 114.96815286624204, "grad_norm": 1.2921662330627441, "learning_rate": 0.0001, "loss": 0.05, "step": 18050 }, { "epoch": 115.03184713375796, "grad_norm": 1.736646056175232, "learning_rate": 0.0001, "loss": 0.0436, "step": 18060 }, { "epoch": 115.09554140127389, "grad_norm": 1.357574462890625, "learning_rate": 0.0001, "loss": 0.0478, "step": 18070 }, { "epoch": 115.1592356687898, "grad_norm": 1.4753291606903076, "learning_rate": 0.0001, "loss": 0.0448, "step": 18080 }, { "epoch": 115.22292993630573, "grad_norm": 1.271520733833313, "learning_rate": 0.0001, "loss": 0.0401, "step": 18090 }, { "epoch": 115.28662420382166, "grad_norm": 1.4813355207443237, "learning_rate": 0.0001, "loss": 0.0442, "step": 18100 }, { "epoch": 115.35031847133757, "grad_norm": 1.4702059030532837, "learning_rate": 0.0001, "loss": 0.0451, "step": 18110 }, { "epoch": 115.4140127388535, "grad_norm": 1.5228216648101807, "learning_rate": 0.0001, "loss": 0.0433, "step": 18120 }, { "epoch": 115.47770700636943, "grad_norm": 1.3177393674850464, "learning_rate": 0.0001, "loss": 0.0459, "step": 18130 }, { "epoch": 115.54140127388536, "grad_norm": 1.2442667484283447, "learning_rate": 0.0001, "loss": 0.0466, "step": 18140 }, { "epoch": 115.60509554140127, "grad_norm": 1.2591516971588135, "learning_rate": 0.0001, "loss": 0.0452, "step": 18150 }, { "epoch": 115.6687898089172, "grad_norm": 1.5011301040649414, "learning_rate": 0.0001, "loss": 0.0481, "step": 18160 }, { "epoch": 115.73248407643312, "grad_norm": 1.4619784355163574, "learning_rate": 0.0001, "loss": 0.0439, "step": 18170 }, { "epoch": 115.79617834394904, "grad_norm": 1.5224008560180664, "learning_rate": 0.0001, "loss": 0.0449, "step": 18180 }, { "epoch": 115.85987261146497, "grad_norm": 1.4787812232971191, "learning_rate": 0.0001, "loss": 0.0432, "step": 18190 }, { "epoch": 115.92356687898089, "grad_norm": 1.4433623552322388, "learning_rate": 0.0001, "loss": 0.0438, "step": 18200 }, { "epoch": 115.98726114649682, "grad_norm": 1.3799906969070435, "learning_rate": 0.0001, "loss": 0.0489, "step": 18210 }, { "epoch": 116.05095541401273, "grad_norm": 1.4449808597564697, "learning_rate": 0.0001, "loss": 0.0446, "step": 18220 }, { "epoch": 116.11464968152866, "grad_norm": 1.1604746580123901, "learning_rate": 0.0001, "loss": 0.0448, "step": 18230 }, { "epoch": 116.17834394904459, "grad_norm": 1.2282549142837524, "learning_rate": 0.0001, "loss": 0.0475, "step": 18240 }, { "epoch": 116.24203821656052, "grad_norm": 1.4023886919021606, "learning_rate": 0.0001, "loss": 0.0467, "step": 18250 }, { "epoch": 116.30573248407643, "grad_norm": 1.3423497676849365, "learning_rate": 0.0001, "loss": 0.0447, "step": 18260 }, { "epoch": 116.36942675159236, "grad_norm": 1.3066171407699585, "learning_rate": 0.0001, "loss": 0.048, "step": 18270 }, { "epoch": 116.43312101910828, "grad_norm": 1.4349915981292725, "learning_rate": 0.0001, "loss": 0.0433, "step": 18280 }, { "epoch": 116.4968152866242, "grad_norm": 1.1297028064727783, "learning_rate": 0.0001, "loss": 0.0459, "step": 18290 }, { "epoch": 116.56050955414013, "grad_norm": 1.202077031135559, "learning_rate": 0.0001, "loss": 0.0486, "step": 18300 }, { "epoch": 116.62420382165605, "grad_norm": 1.2346173524856567, "learning_rate": 0.0001, "loss": 0.0461, "step": 18310 }, { "epoch": 116.68789808917198, "grad_norm": 1.3296160697937012, "learning_rate": 0.0001, "loss": 0.0446, "step": 18320 }, { "epoch": 116.7515923566879, "grad_norm": 1.541972279548645, "learning_rate": 0.0001, "loss": 0.0476, "step": 18330 }, { "epoch": 116.81528662420382, "grad_norm": 1.2356714010238647, "learning_rate": 0.0001, "loss": 0.0445, "step": 18340 }, { "epoch": 116.87898089171975, "grad_norm": 1.5611233711242676, "learning_rate": 0.0001, "loss": 0.0468, "step": 18350 }, { "epoch": 116.94267515923566, "grad_norm": 1.321467638015747, "learning_rate": 0.0001, "loss": 0.0447, "step": 18360 }, { "epoch": 117.00636942675159, "grad_norm": 1.2449429035186768, "learning_rate": 0.0001, "loss": 0.0452, "step": 18370 }, { "epoch": 117.07006369426752, "grad_norm": 1.418324589729309, "learning_rate": 0.0001, "loss": 0.0448, "step": 18380 }, { "epoch": 117.13375796178345, "grad_norm": 1.4154881238937378, "learning_rate": 0.0001, "loss": 0.044, "step": 18390 }, { "epoch": 117.19745222929936, "grad_norm": 1.7022072076797485, "learning_rate": 0.0001, "loss": 0.0458, "step": 18400 }, { "epoch": 117.26114649681529, "grad_norm": 1.4858900308609009, "learning_rate": 0.0001, "loss": 0.042, "step": 18410 }, { "epoch": 117.32484076433121, "grad_norm": 1.3760398626327515, "learning_rate": 0.0001, "loss": 0.0435, "step": 18420 }, { "epoch": 117.38853503184713, "grad_norm": 1.2909014225006104, "learning_rate": 0.0001, "loss": 0.0447, "step": 18430 }, { "epoch": 117.45222929936305, "grad_norm": 1.1435484886169434, "learning_rate": 0.0001, "loss": 0.0466, "step": 18440 }, { "epoch": 117.51592356687898, "grad_norm": 1.1507290601730347, "learning_rate": 0.0001, "loss": 0.0445, "step": 18450 }, { "epoch": 117.57961783439491, "grad_norm": 1.5613031387329102, "learning_rate": 0.0001, "loss": 0.0439, "step": 18460 }, { "epoch": 117.64331210191082, "grad_norm": 1.3719571828842163, "learning_rate": 0.0001, "loss": 0.0481, "step": 18470 }, { "epoch": 117.70700636942675, "grad_norm": 1.2335853576660156, "learning_rate": 0.0001, "loss": 0.0439, "step": 18480 }, { "epoch": 117.77070063694268, "grad_norm": 1.2843579053878784, "learning_rate": 0.0001, "loss": 0.043, "step": 18490 }, { "epoch": 117.8343949044586, "grad_norm": 1.2104636430740356, "learning_rate": 0.0001, "loss": 0.0434, "step": 18500 }, { "epoch": 117.89808917197452, "grad_norm": 1.2496910095214844, "learning_rate": 0.0001, "loss": 0.0495, "step": 18510 }, { "epoch": 117.96178343949045, "grad_norm": 1.6113117933273315, "learning_rate": 0.0001, "loss": 0.0478, "step": 18520 }, { "epoch": 118.02547770700637, "grad_norm": 1.4338411092758179, "learning_rate": 0.0001, "loss": 0.0437, "step": 18530 }, { "epoch": 118.08917197452229, "grad_norm": 1.784504771232605, "learning_rate": 0.0001, "loss": 0.0433, "step": 18540 }, { "epoch": 118.15286624203821, "grad_norm": 1.4998630285263062, "learning_rate": 0.0001, "loss": 0.0415, "step": 18550 }, { "epoch": 118.21656050955414, "grad_norm": 1.4916315078735352, "learning_rate": 0.0001, "loss": 0.0441, "step": 18560 }, { "epoch": 118.28025477707007, "grad_norm": 0.9879129528999329, "learning_rate": 0.0001, "loss": 0.0433, "step": 18570 }, { "epoch": 118.34394904458598, "grad_norm": 1.3011046648025513, "learning_rate": 0.0001, "loss": 0.042, "step": 18580 }, { "epoch": 118.40764331210191, "grad_norm": 1.515490174293518, "learning_rate": 0.0001, "loss": 0.0444, "step": 18590 }, { "epoch": 118.47133757961784, "grad_norm": 1.3856123685836792, "learning_rate": 0.0001, "loss": 0.0448, "step": 18600 }, { "epoch": 118.53503184713375, "grad_norm": 1.0483989715576172, "learning_rate": 0.0001, "loss": 0.0453, "step": 18610 }, { "epoch": 118.59872611464968, "grad_norm": 1.365037441253662, "learning_rate": 0.0001, "loss": 0.0452, "step": 18620 }, { "epoch": 118.6624203821656, "grad_norm": 1.0043832063674927, "learning_rate": 0.0001, "loss": 0.0446, "step": 18630 }, { "epoch": 118.72611464968153, "grad_norm": 1.3094029426574707, "learning_rate": 0.0001, "loss": 0.0425, "step": 18640 }, { "epoch": 118.78980891719745, "grad_norm": 1.7615092992782593, "learning_rate": 0.0001, "loss": 0.0436, "step": 18650 }, { "epoch": 118.85350318471338, "grad_norm": 1.313910961151123, "learning_rate": 0.0001, "loss": 0.0414, "step": 18660 }, { "epoch": 118.9171974522293, "grad_norm": 1.1371597051620483, "learning_rate": 0.0001, "loss": 0.0439, "step": 18670 }, { "epoch": 118.98089171974522, "grad_norm": 1.4076037406921387, "learning_rate": 0.0001, "loss": 0.0438, "step": 18680 }, { "epoch": 119.04458598726114, "grad_norm": 1.443073034286499, "learning_rate": 0.0001, "loss": 0.0441, "step": 18690 }, { "epoch": 119.10828025477707, "grad_norm": 1.2747516632080078, "learning_rate": 0.0001, "loss": 0.0438, "step": 18700 }, { "epoch": 119.171974522293, "grad_norm": 1.3523818254470825, "learning_rate": 0.0001, "loss": 0.0398, "step": 18710 }, { "epoch": 119.23566878980891, "grad_norm": 1.2801094055175781, "learning_rate": 0.0001, "loss": 0.0459, "step": 18720 }, { "epoch": 119.29936305732484, "grad_norm": 1.7804388999938965, "learning_rate": 0.0001, "loss": 0.0427, "step": 18730 }, { "epoch": 119.36305732484077, "grad_norm": 1.4481210708618164, "learning_rate": 0.0001, "loss": 0.0455, "step": 18740 }, { "epoch": 119.4267515923567, "grad_norm": 1.6692603826522827, "learning_rate": 0.0001, "loss": 0.0423, "step": 18750 }, { "epoch": 119.49044585987261, "grad_norm": 1.3990603685379028, "learning_rate": 0.0001, "loss": 0.0429, "step": 18760 }, { "epoch": 119.55414012738854, "grad_norm": 1.4733755588531494, "learning_rate": 0.0001, "loss": 0.041, "step": 18770 }, { "epoch": 119.61783439490446, "grad_norm": 1.3171707391738892, "learning_rate": 0.0001, "loss": 0.0412, "step": 18780 }, { "epoch": 119.68152866242038, "grad_norm": 2.5641067028045654, "learning_rate": 0.0001, "loss": 0.0424, "step": 18790 }, { "epoch": 119.7452229299363, "grad_norm": 1.6793849468231201, "learning_rate": 0.0001, "loss": 0.0465, "step": 18800 }, { "epoch": 119.80891719745223, "grad_norm": 1.8186689615249634, "learning_rate": 0.0001, "loss": 0.0413, "step": 18810 }, { "epoch": 119.87261146496816, "grad_norm": 2.143751382827759, "learning_rate": 0.0001, "loss": 0.0409, "step": 18820 }, { "epoch": 119.93630573248407, "grad_norm": 1.7928868532180786, "learning_rate": 0.0001, "loss": 0.0393, "step": 18830 }, { "epoch": 120.0, "grad_norm": 1.8675436973571777, "learning_rate": 0.0001, "loss": 0.0449, "step": 18840 }, { "epoch": 120.06369426751593, "grad_norm": 1.6595637798309326, "learning_rate": 0.0001, "loss": 0.0439, "step": 18850 }, { "epoch": 120.12738853503184, "grad_norm": 1.6543625593185425, "learning_rate": 0.0001, "loss": 0.0398, "step": 18860 }, { "epoch": 120.19108280254777, "grad_norm": 1.6734318733215332, "learning_rate": 0.0001, "loss": 0.0449, "step": 18870 }, { "epoch": 120.2547770700637, "grad_norm": 1.3839908838272095, "learning_rate": 0.0001, "loss": 0.0432, "step": 18880 }, { "epoch": 120.31847133757962, "grad_norm": 1.6011502742767334, "learning_rate": 0.0001, "loss": 0.0423, "step": 18890 }, { "epoch": 120.38216560509554, "grad_norm": 1.5156205892562866, "learning_rate": 0.0001, "loss": 0.042, "step": 18900 }, { "epoch": 120.44585987261146, "grad_norm": 1.3252609968185425, "learning_rate": 0.0001, "loss": 0.0386, "step": 18910 }, { "epoch": 120.50955414012739, "grad_norm": 1.465922474861145, "learning_rate": 0.0001, "loss": 0.0434, "step": 18920 }, { "epoch": 120.5732484076433, "grad_norm": 1.3932639360427856, "learning_rate": 0.0001, "loss": 0.0408, "step": 18930 }, { "epoch": 120.63694267515923, "grad_norm": 1.5442124605178833, "learning_rate": 0.0001, "loss": 0.0413, "step": 18940 }, { "epoch": 120.70063694267516, "grad_norm": 1.546379566192627, "learning_rate": 0.0001, "loss": 0.0419, "step": 18950 }, { "epoch": 120.76433121019109, "grad_norm": 1.2519079446792603, "learning_rate": 0.0001, "loss": 0.0401, "step": 18960 }, { "epoch": 120.828025477707, "grad_norm": 1.424184799194336, "learning_rate": 0.0001, "loss": 0.0447, "step": 18970 }, { "epoch": 120.89171974522293, "grad_norm": 1.9328597784042358, "learning_rate": 0.0001, "loss": 0.0435, "step": 18980 }, { "epoch": 120.95541401273886, "grad_norm": 2.30653977394104, "learning_rate": 0.0001, "loss": 0.0441, "step": 18990 }, { "epoch": 121.01910828025478, "grad_norm": 1.4912662506103516, "learning_rate": 0.0001, "loss": 0.0442, "step": 19000 }, { "epoch": 121.0828025477707, "grad_norm": 1.4835511445999146, "learning_rate": 0.0001, "loss": 0.0432, "step": 19010 }, { "epoch": 121.14649681528662, "grad_norm": 1.6730241775512695, "learning_rate": 0.0001, "loss": 0.0412, "step": 19020 }, { "epoch": 121.21019108280255, "grad_norm": 1.6641911268234253, "learning_rate": 0.0001, "loss": 0.0413, "step": 19030 }, { "epoch": 121.27388535031847, "grad_norm": 1.2113441228866577, "learning_rate": 0.0001, "loss": 0.0415, "step": 19040 }, { "epoch": 121.3375796178344, "grad_norm": 1.4150481224060059, "learning_rate": 0.0001, "loss": 0.0429, "step": 19050 }, { "epoch": 121.40127388535032, "grad_norm": 1.6074702739715576, "learning_rate": 0.0001, "loss": 0.0422, "step": 19060 }, { "epoch": 121.46496815286625, "grad_norm": 1.249145269393921, "learning_rate": 0.0001, "loss": 0.04, "step": 19070 }, { "epoch": 121.52866242038216, "grad_norm": 1.4686903953552246, "learning_rate": 0.0001, "loss": 0.0421, "step": 19080 }, { "epoch": 121.59235668789809, "grad_norm": 1.5468615293502808, "learning_rate": 0.0001, "loss": 0.0464, "step": 19090 }, { "epoch": 121.65605095541402, "grad_norm": 1.435880422592163, "learning_rate": 0.0001, "loss": 0.0515, "step": 19100 }, { "epoch": 121.71974522292993, "grad_norm": 1.3568828105926514, "learning_rate": 0.0001, "loss": 0.0438, "step": 19110 }, { "epoch": 121.78343949044586, "grad_norm": 1.5052253007888794, "learning_rate": 0.0001, "loss": 0.0446, "step": 19120 }, { "epoch": 121.84713375796179, "grad_norm": 1.5749281644821167, "learning_rate": 0.0001, "loss": 0.0433, "step": 19130 }, { "epoch": 121.91082802547771, "grad_norm": 1.4376550912857056, "learning_rate": 0.0001, "loss": 0.0451, "step": 19140 }, { "epoch": 121.97452229299363, "grad_norm": 1.5301350355148315, "learning_rate": 0.0001, "loss": 0.0432, "step": 19150 }, { "epoch": 122.03821656050955, "grad_norm": 1.6270490884780884, "learning_rate": 0.0001, "loss": 0.0476, "step": 19160 }, { "epoch": 122.10191082802548, "grad_norm": 1.5058867931365967, "learning_rate": 0.0001, "loss": 0.0489, "step": 19170 }, { "epoch": 122.1656050955414, "grad_norm": 1.3487776517868042, "learning_rate": 0.0001, "loss": 0.0469, "step": 19180 }, { "epoch": 122.22929936305732, "grad_norm": 1.3066048622131348, "learning_rate": 0.0001, "loss": 0.0444, "step": 19190 }, { "epoch": 122.29299363057325, "grad_norm": 1.6006709337234497, "learning_rate": 0.0001, "loss": 0.0453, "step": 19200 }, { "epoch": 122.35668789808918, "grad_norm": 1.396336317062378, "learning_rate": 0.0001, "loss": 0.0434, "step": 19210 }, { "epoch": 122.42038216560509, "grad_norm": 1.467441439628601, "learning_rate": 0.0001, "loss": 0.0489, "step": 19220 }, { "epoch": 122.48407643312102, "grad_norm": 1.469349980354309, "learning_rate": 0.0001, "loss": 0.0424, "step": 19230 }, { "epoch": 122.54777070063695, "grad_norm": 1.3311097621917725, "learning_rate": 0.0001, "loss": 0.0463, "step": 19240 }, { "epoch": 122.61146496815287, "grad_norm": 1.1785115003585815, "learning_rate": 0.0001, "loss": 0.0458, "step": 19250 }, { "epoch": 122.67515923566879, "grad_norm": 1.473096251487732, "learning_rate": 0.0001, "loss": 0.0445, "step": 19260 }, { "epoch": 122.73885350318471, "grad_norm": 1.2282589673995972, "learning_rate": 0.0001, "loss": 0.0417, "step": 19270 }, { "epoch": 122.80254777070064, "grad_norm": 1.4522132873535156, "learning_rate": 0.0001, "loss": 0.0431, "step": 19280 }, { "epoch": 122.86624203821655, "grad_norm": 1.1194076538085938, "learning_rate": 0.0001, "loss": 0.0449, "step": 19290 }, { "epoch": 122.92993630573248, "grad_norm": 1.5055737495422363, "learning_rate": 0.0001, "loss": 0.0439, "step": 19300 }, { "epoch": 122.99363057324841, "grad_norm": 1.190428376197815, "learning_rate": 0.0001, "loss": 0.042, "step": 19310 }, { "epoch": 123.05732484076434, "grad_norm": 1.3019704818725586, "learning_rate": 0.0001, "loss": 0.0426, "step": 19320 }, { "epoch": 123.12101910828025, "grad_norm": 1.1272410154342651, "learning_rate": 0.0001, "loss": 0.0434, "step": 19330 }, { "epoch": 123.18471337579618, "grad_norm": 1.2334132194519043, "learning_rate": 0.0001, "loss": 0.0412, "step": 19340 }, { "epoch": 123.2484076433121, "grad_norm": 1.3781323432922363, "learning_rate": 0.0001, "loss": 0.0433, "step": 19350 }, { "epoch": 123.31210191082802, "grad_norm": 1.4394340515136719, "learning_rate": 0.0001, "loss": 0.048, "step": 19360 }, { "epoch": 123.37579617834395, "grad_norm": 1.3123518228530884, "learning_rate": 0.0001, "loss": 0.0494, "step": 19370 }, { "epoch": 123.43949044585987, "grad_norm": 1.4096390008926392, "learning_rate": 0.0001, "loss": 0.048, "step": 19380 }, { "epoch": 123.5031847133758, "grad_norm": 1.398040771484375, "learning_rate": 0.0001, "loss": 0.054, "step": 19390 }, { "epoch": 123.56687898089172, "grad_norm": 1.359965443611145, "learning_rate": 0.0001, "loss": 0.0483, "step": 19400 }, { "epoch": 123.63057324840764, "grad_norm": 0.8967390060424805, "learning_rate": 0.0001, "loss": 0.047, "step": 19410 }, { "epoch": 123.69426751592357, "grad_norm": 1.277967095375061, "learning_rate": 0.0001, "loss": 0.0502, "step": 19420 }, { "epoch": 123.75796178343948, "grad_norm": 1.290128469467163, "learning_rate": 0.0001, "loss": 0.049, "step": 19430 }, { "epoch": 123.82165605095541, "grad_norm": 1.3999556303024292, "learning_rate": 0.0001, "loss": 0.0479, "step": 19440 }, { "epoch": 123.88535031847134, "grad_norm": 1.2712316513061523, "learning_rate": 0.0001, "loss": 0.0498, "step": 19450 }, { "epoch": 123.94904458598727, "grad_norm": 1.2715013027191162, "learning_rate": 0.0001, "loss": 0.0453, "step": 19460 }, { "epoch": 124.01273885350318, "grad_norm": 1.4762217998504639, "learning_rate": 0.0001, "loss": 0.0438, "step": 19470 }, { "epoch": 124.07643312101911, "grad_norm": 1.3909152746200562, "learning_rate": 0.0001, "loss": 0.0437, "step": 19480 }, { "epoch": 124.14012738853503, "grad_norm": 1.1570583581924438, "learning_rate": 0.0001, "loss": 0.0459, "step": 19490 }, { "epoch": 124.20382165605096, "grad_norm": 1.7598679065704346, "learning_rate": 0.0001, "loss": 0.0456, "step": 19500 }, { "epoch": 124.26751592356688, "grad_norm": 1.3363761901855469, "learning_rate": 0.0001, "loss": 0.0472, "step": 19510 }, { "epoch": 124.3312101910828, "grad_norm": 1.2754638195037842, "learning_rate": 0.0001, "loss": 0.0445, "step": 19520 }, { "epoch": 124.39490445859873, "grad_norm": 1.277768850326538, "learning_rate": 0.0001, "loss": 0.0453, "step": 19530 }, { "epoch": 124.45859872611464, "grad_norm": 1.234714388847351, "learning_rate": 0.0001, "loss": 0.0431, "step": 19540 }, { "epoch": 124.52229299363057, "grad_norm": 1.0694342851638794, "learning_rate": 0.0001, "loss": 0.045, "step": 19550 }, { "epoch": 124.5859872611465, "grad_norm": 1.4698904752731323, "learning_rate": 0.0001, "loss": 0.0462, "step": 19560 }, { "epoch": 124.64968152866243, "grad_norm": 1.5040946006774902, "learning_rate": 0.0001, "loss": 0.046, "step": 19570 }, { "epoch": 124.71337579617834, "grad_norm": 1.4086192846298218, "learning_rate": 0.0001, "loss": 0.0419, "step": 19580 }, { "epoch": 124.77707006369427, "grad_norm": 1.4577008485794067, "learning_rate": 0.0001, "loss": 0.0436, "step": 19590 }, { "epoch": 124.8407643312102, "grad_norm": 1.3134231567382812, "learning_rate": 0.0001, "loss": 0.0453, "step": 19600 }, { "epoch": 124.90445859872611, "grad_norm": 1.2780232429504395, "learning_rate": 0.0001, "loss": 0.0436, "step": 19610 }, { "epoch": 124.96815286624204, "grad_norm": 1.817807912826538, "learning_rate": 0.0001, "loss": 0.0458, "step": 19620 }, { "epoch": 125.03184713375796, "grad_norm": 1.2105482816696167, "learning_rate": 0.0001, "loss": 0.042, "step": 19630 }, { "epoch": 125.09554140127389, "grad_norm": 1.4640356302261353, "learning_rate": 0.0001, "loss": 0.0399, "step": 19640 }, { "epoch": 125.1592356687898, "grad_norm": 1.5069547891616821, "learning_rate": 0.0001, "loss": 0.0412, "step": 19650 }, { "epoch": 125.22292993630573, "grad_norm": 1.4890124797821045, "learning_rate": 0.0001, "loss": 0.0426, "step": 19660 }, { "epoch": 125.28662420382166, "grad_norm": 1.4260509014129639, "learning_rate": 0.0001, "loss": 0.0456, "step": 19670 }, { "epoch": 125.35031847133757, "grad_norm": 1.3099347352981567, "learning_rate": 0.0001, "loss": 0.0457, "step": 19680 }, { "epoch": 125.4140127388535, "grad_norm": 1.2474498748779297, "learning_rate": 0.0001, "loss": 0.0437, "step": 19690 }, { "epoch": 125.47770700636943, "grad_norm": 1.1782113313674927, "learning_rate": 0.0001, "loss": 0.0467, "step": 19700 }, { "epoch": 125.54140127388536, "grad_norm": 1.132495641708374, "learning_rate": 0.0001, "loss": 0.0452, "step": 19710 }, { "epoch": 125.60509554140127, "grad_norm": 1.215091347694397, "learning_rate": 0.0001, "loss": 0.0461, "step": 19720 }, { "epoch": 125.6687898089172, "grad_norm": 1.09493088722229, "learning_rate": 0.0001, "loss": 0.0456, "step": 19730 }, { "epoch": 125.73248407643312, "grad_norm": 1.317816972732544, "learning_rate": 0.0001, "loss": 0.0455, "step": 19740 }, { "epoch": 125.79617834394904, "grad_norm": 1.0608031749725342, "learning_rate": 0.0001, "loss": 0.0488, "step": 19750 }, { "epoch": 125.85987261146497, "grad_norm": 1.1748862266540527, "learning_rate": 0.0001, "loss": 0.0427, "step": 19760 }, { "epoch": 125.92356687898089, "grad_norm": 1.8526133298873901, "learning_rate": 0.0001, "loss": 0.0496, "step": 19770 }, { "epoch": 125.98726114649682, "grad_norm": 2.5082483291625977, "learning_rate": 0.0001, "loss": 0.0436, "step": 19780 }, { "epoch": 126.05095541401273, "grad_norm": 2.264767646789551, "learning_rate": 0.0001, "loss": 0.0469, "step": 19790 }, { "epoch": 126.11464968152866, "grad_norm": 2.2874715328216553, "learning_rate": 0.0001, "loss": 0.0431, "step": 19800 }, { "epoch": 126.17834394904459, "grad_norm": 1.9024156332015991, "learning_rate": 0.0001, "loss": 0.0421, "step": 19810 }, { "epoch": 126.24203821656052, "grad_norm": 1.5734213590621948, "learning_rate": 0.0001, "loss": 0.0406, "step": 19820 }, { "epoch": 126.30573248407643, "grad_norm": 1.380625605583191, "learning_rate": 0.0001, "loss": 0.043, "step": 19830 }, { "epoch": 126.36942675159236, "grad_norm": 2.007199287414551, "learning_rate": 0.0001, "loss": 0.0427, "step": 19840 }, { "epoch": 126.43312101910828, "grad_norm": 1.5006024837493896, "learning_rate": 0.0001, "loss": 0.041, "step": 19850 }, { "epoch": 126.4968152866242, "grad_norm": 1.6618167161941528, "learning_rate": 0.0001, "loss": 0.0437, "step": 19860 }, { "epoch": 126.56050955414013, "grad_norm": 1.6220378875732422, "learning_rate": 0.0001, "loss": 0.0379, "step": 19870 }, { "epoch": 126.62420382165605, "grad_norm": 1.7638481855392456, "learning_rate": 0.0001, "loss": 0.0382, "step": 19880 }, { "epoch": 126.68789808917198, "grad_norm": 1.3409003019332886, "learning_rate": 0.0001, "loss": 0.0428, "step": 19890 }, { "epoch": 126.7515923566879, "grad_norm": 1.7467799186706543, "learning_rate": 0.0001, "loss": 0.0436, "step": 19900 }, { "epoch": 126.81528662420382, "grad_norm": 1.5625790357589722, "learning_rate": 0.0001, "loss": 0.0447, "step": 19910 }, { "epoch": 126.87898089171975, "grad_norm": 1.368428349494934, "learning_rate": 0.0001, "loss": 0.0394, "step": 19920 }, { "epoch": 126.94267515923566, "grad_norm": 1.5386468172073364, "learning_rate": 0.0001, "loss": 0.0427, "step": 19930 }, { "epoch": 127.00636942675159, "grad_norm": 1.363431692123413, "learning_rate": 0.0001, "loss": 0.0414, "step": 19940 }, { "epoch": 127.07006369426752, "grad_norm": 1.490545392036438, "learning_rate": 0.0001, "loss": 0.042, "step": 19950 }, { "epoch": 127.13375796178345, "grad_norm": 1.210688829421997, "learning_rate": 0.0001, "loss": 0.042, "step": 19960 }, { "epoch": 127.19745222929936, "grad_norm": 2.0595524311065674, "learning_rate": 0.0001, "loss": 0.0463, "step": 19970 }, { "epoch": 127.26114649681529, "grad_norm": 1.5464951992034912, "learning_rate": 0.0001, "loss": 0.0469, "step": 19980 }, { "epoch": 127.32484076433121, "grad_norm": 1.640088677406311, "learning_rate": 0.0001, "loss": 0.0388, "step": 19990 }, { "epoch": 127.38853503184713, "grad_norm": 1.7923165559768677, "learning_rate": 0.0001, "loss": 0.0409, "step": 20000 }, { "epoch": 127.45222929936305, "grad_norm": 1.4944814443588257, "learning_rate": 0.0001, "loss": 0.0411, "step": 20010 }, { "epoch": 127.51592356687898, "grad_norm": 1.831305980682373, "learning_rate": 0.0001, "loss": 0.0423, "step": 20020 }, { "epoch": 127.57961783439491, "grad_norm": 1.6783373355865479, "learning_rate": 0.0001, "loss": 0.0429, "step": 20030 }, { "epoch": 127.64331210191082, "grad_norm": 1.4243460893630981, "learning_rate": 0.0001, "loss": 0.0417, "step": 20040 }, { "epoch": 127.70700636942675, "grad_norm": 1.5917372703552246, "learning_rate": 0.0001, "loss": 0.0399, "step": 20050 }, { "epoch": 127.77070063694268, "grad_norm": 1.660276174545288, "learning_rate": 0.0001, "loss": 0.0426, "step": 20060 }, { "epoch": 127.8343949044586, "grad_norm": 1.619864583015442, "learning_rate": 0.0001, "loss": 0.0438, "step": 20070 }, { "epoch": 127.89808917197452, "grad_norm": 1.6513537168502808, "learning_rate": 0.0001, "loss": 0.0434, "step": 20080 }, { "epoch": 127.96178343949045, "grad_norm": 1.7600609064102173, "learning_rate": 0.0001, "loss": 0.0467, "step": 20090 }, { "epoch": 128.02547770700636, "grad_norm": 1.5989487171173096, "learning_rate": 0.0001, "loss": 0.0407, "step": 20100 }, { "epoch": 128.0891719745223, "grad_norm": 1.3013964891433716, "learning_rate": 0.0001, "loss": 0.0447, "step": 20110 }, { "epoch": 128.15286624203821, "grad_norm": 1.0979801416397095, "learning_rate": 0.0001, "loss": 0.0466, "step": 20120 }, { "epoch": 128.21656050955414, "grad_norm": 1.3395881652832031, "learning_rate": 0.0001, "loss": 0.0442, "step": 20130 }, { "epoch": 128.28025477707007, "grad_norm": 1.2808581590652466, "learning_rate": 0.0001, "loss": 0.0419, "step": 20140 }, { "epoch": 128.343949044586, "grad_norm": 1.3671023845672607, "learning_rate": 0.0001, "loss": 0.0431, "step": 20150 }, { "epoch": 128.40764331210192, "grad_norm": 1.1923695802688599, "learning_rate": 0.0001, "loss": 0.0428, "step": 20160 }, { "epoch": 128.47133757961782, "grad_norm": 1.158187747001648, "learning_rate": 0.0001, "loss": 0.0411, "step": 20170 }, { "epoch": 128.53503184713375, "grad_norm": 1.5194103717803955, "learning_rate": 0.0001, "loss": 0.0471, "step": 20180 }, { "epoch": 128.59872611464968, "grad_norm": 1.3241220712661743, "learning_rate": 0.0001, "loss": 0.0434, "step": 20190 }, { "epoch": 128.6624203821656, "grad_norm": 1.4660711288452148, "learning_rate": 0.0001, "loss": 0.0471, "step": 20200 }, { "epoch": 128.72611464968153, "grad_norm": 1.264190673828125, "learning_rate": 0.0001, "loss": 0.044, "step": 20210 }, { "epoch": 128.78980891719746, "grad_norm": 1.5643670558929443, "learning_rate": 0.0001, "loss": 0.0425, "step": 20220 }, { "epoch": 128.8535031847134, "grad_norm": 1.4720720052719116, "learning_rate": 0.0001, "loss": 0.0422, "step": 20230 }, { "epoch": 128.9171974522293, "grad_norm": 1.689847707748413, "learning_rate": 0.0001, "loss": 0.0464, "step": 20240 }, { "epoch": 128.98089171974522, "grad_norm": 1.1831263303756714, "learning_rate": 0.0001, "loss": 0.0452, "step": 20250 }, { "epoch": 129.04458598726114, "grad_norm": 1.6082450151443481, "learning_rate": 0.0001, "loss": 0.045, "step": 20260 }, { "epoch": 129.10828025477707, "grad_norm": 1.1463748216629028, "learning_rate": 0.0001, "loss": 0.0434, "step": 20270 }, { "epoch": 129.171974522293, "grad_norm": 1.5154356956481934, "learning_rate": 0.0001, "loss": 0.0426, "step": 20280 }, { "epoch": 129.23566878980893, "grad_norm": 1.1728535890579224, "learning_rate": 0.0001, "loss": 0.0456, "step": 20290 }, { "epoch": 129.29936305732485, "grad_norm": 1.1014456748962402, "learning_rate": 0.0001, "loss": 0.0423, "step": 20300 }, { "epoch": 129.36305732484075, "grad_norm": 1.1029802560806274, "learning_rate": 0.0001, "loss": 0.0435, "step": 20310 }, { "epoch": 129.42675159235668, "grad_norm": 1.2448194026947021, "learning_rate": 0.0001, "loss": 0.0457, "step": 20320 }, { "epoch": 129.4904458598726, "grad_norm": 1.0045857429504395, "learning_rate": 0.0001, "loss": 0.0435, "step": 20330 }, { "epoch": 129.55414012738854, "grad_norm": 1.2315653562545776, "learning_rate": 0.0001, "loss": 0.0466, "step": 20340 }, { "epoch": 129.61783439490446, "grad_norm": 1.4588046073913574, "learning_rate": 0.0001, "loss": 0.047, "step": 20350 }, { "epoch": 129.6815286624204, "grad_norm": 1.4733781814575195, "learning_rate": 0.0001, "loss": 0.0456, "step": 20360 }, { "epoch": 129.74522292993632, "grad_norm": 1.3275460004806519, "learning_rate": 0.0001, "loss": 0.0404, "step": 20370 }, { "epoch": 129.80891719745222, "grad_norm": 1.401556134223938, "learning_rate": 0.0001, "loss": 0.0451, "step": 20380 }, { "epoch": 129.87261146496814, "grad_norm": 1.2451789379119873, "learning_rate": 0.0001, "loss": 0.0421, "step": 20390 }, { "epoch": 129.93630573248407, "grad_norm": 2.081394910812378, "learning_rate": 0.0001, "loss": 0.0487, "step": 20400 }, { "epoch": 130.0, "grad_norm": 3.193598508834839, "learning_rate": 0.0001, "loss": 0.0429, "step": 20410 }, { "epoch": 130.06369426751593, "grad_norm": 1.9878277778625488, "learning_rate": 0.0001, "loss": 0.0447, "step": 20420 }, { "epoch": 130.12738853503186, "grad_norm": 1.9912248849868774, "learning_rate": 0.0001, "loss": 0.0428, "step": 20430 }, { "epoch": 130.19108280254778, "grad_norm": 1.9500141143798828, "learning_rate": 0.0001, "loss": 0.0417, "step": 20440 }, { "epoch": 130.25477707006368, "grad_norm": 1.5158494710922241, "learning_rate": 0.0001, "loss": 0.0437, "step": 20450 }, { "epoch": 130.3184713375796, "grad_norm": 1.3071136474609375, "learning_rate": 0.0001, "loss": 0.0394, "step": 20460 }, { "epoch": 130.38216560509554, "grad_norm": 1.3677716255187988, "learning_rate": 0.0001, "loss": 0.0407, "step": 20470 }, { "epoch": 130.44585987261146, "grad_norm": 1.7654454708099365, "learning_rate": 0.0001, "loss": 0.0409, "step": 20480 }, { "epoch": 130.5095541401274, "grad_norm": 1.4449845552444458, "learning_rate": 0.0001, "loss": 0.0411, "step": 20490 }, { "epoch": 130.57324840764332, "grad_norm": 1.4250000715255737, "learning_rate": 0.0001, "loss": 0.0394, "step": 20500 }, { "epoch": 130.63694267515925, "grad_norm": 1.2981232404708862, "learning_rate": 0.0001, "loss": 0.0409, "step": 20510 }, { "epoch": 130.70063694267515, "grad_norm": 1.3467835187911987, "learning_rate": 0.0001, "loss": 0.0388, "step": 20520 }, { "epoch": 130.76433121019107, "grad_norm": 1.4715960025787354, "learning_rate": 0.0001, "loss": 0.0467, "step": 20530 }, { "epoch": 130.828025477707, "grad_norm": 1.1987805366516113, "learning_rate": 0.0001, "loss": 0.0419, "step": 20540 }, { "epoch": 130.89171974522293, "grad_norm": 2.3097681999206543, "learning_rate": 0.0001, "loss": 0.0444, "step": 20550 }, { "epoch": 130.95541401273886, "grad_norm": 2.3402061462402344, "learning_rate": 0.0001, "loss": 0.0432, "step": 20560 }, { "epoch": 131.01910828025478, "grad_norm": 2.029860019683838, "learning_rate": 0.0001, "loss": 0.0429, "step": 20570 }, { "epoch": 131.0828025477707, "grad_norm": 1.6264928579330444, "learning_rate": 0.0001, "loss": 0.0416, "step": 20580 }, { "epoch": 131.1464968152866, "grad_norm": 1.7959222793579102, "learning_rate": 0.0001, "loss": 0.0428, "step": 20590 }, { "epoch": 131.21019108280254, "grad_norm": 1.6300286054611206, "learning_rate": 0.0001, "loss": 0.0375, "step": 20600 }, { "epoch": 131.27388535031847, "grad_norm": 1.9607809782028198, "learning_rate": 0.0001, "loss": 0.0357, "step": 20610 }, { "epoch": 131.3375796178344, "grad_norm": 1.456030249595642, "learning_rate": 0.0001, "loss": 0.043, "step": 20620 }, { "epoch": 131.40127388535032, "grad_norm": 1.639656662940979, "learning_rate": 0.0001, "loss": 0.0428, "step": 20630 }, { "epoch": 131.46496815286625, "grad_norm": 1.8076751232147217, "learning_rate": 0.0001, "loss": 0.0417, "step": 20640 }, { "epoch": 131.52866242038218, "grad_norm": 1.6063716411590576, "learning_rate": 0.0001, "loss": 0.0413, "step": 20650 }, { "epoch": 131.59235668789808, "grad_norm": 1.6668081283569336, "learning_rate": 0.0001, "loss": 0.0409, "step": 20660 }, { "epoch": 131.656050955414, "grad_norm": 1.5097194910049438, "learning_rate": 0.0001, "loss": 0.0444, "step": 20670 }, { "epoch": 131.71974522292993, "grad_norm": 1.6661319732666016, "learning_rate": 0.0001, "loss": 0.0455, "step": 20680 }, { "epoch": 131.78343949044586, "grad_norm": 1.3880820274353027, "learning_rate": 0.0001, "loss": 0.0434, "step": 20690 }, { "epoch": 131.84713375796179, "grad_norm": 1.2181768417358398, "learning_rate": 0.0001, "loss": 0.0459, "step": 20700 }, { "epoch": 131.9108280254777, "grad_norm": 1.0328413248062134, "learning_rate": 0.0001, "loss": 0.0457, "step": 20710 }, { "epoch": 131.97452229299364, "grad_norm": 1.4004002809524536, "learning_rate": 0.0001, "loss": 0.0453, "step": 20720 }, { "epoch": 132.03821656050957, "grad_norm": 3.2774226665496826, "learning_rate": 0.0001, "loss": 0.0438, "step": 20730 }, { "epoch": 132.10191082802547, "grad_norm": 1.5900540351867676, "learning_rate": 0.0001, "loss": 0.0399, "step": 20740 }, { "epoch": 132.1656050955414, "grad_norm": 1.391090989112854, "learning_rate": 0.0001, "loss": 0.0448, "step": 20750 }, { "epoch": 132.22929936305732, "grad_norm": 1.201167345046997, "learning_rate": 0.0001, "loss": 0.0437, "step": 20760 }, { "epoch": 132.29299363057325, "grad_norm": 1.5337549448013306, "learning_rate": 0.0001, "loss": 0.0441, "step": 20770 }, { "epoch": 132.35668789808918, "grad_norm": 1.7994818687438965, "learning_rate": 0.0001, "loss": 0.0405, "step": 20780 }, { "epoch": 132.4203821656051, "grad_norm": 1.4965437650680542, "learning_rate": 0.0001, "loss": 0.0421, "step": 20790 }, { "epoch": 132.48407643312103, "grad_norm": 1.3255534172058105, "learning_rate": 0.0001, "loss": 0.044, "step": 20800 }, { "epoch": 132.54777070063693, "grad_norm": 1.3365637063980103, "learning_rate": 0.0001, "loss": 0.0419, "step": 20810 }, { "epoch": 132.61146496815286, "grad_norm": 1.1867365837097168, "learning_rate": 0.0001, "loss": 0.0394, "step": 20820 }, { "epoch": 132.6751592356688, "grad_norm": 1.4334465265274048, "learning_rate": 0.0001, "loss": 0.0402, "step": 20830 }, { "epoch": 132.7388535031847, "grad_norm": 1.4340722560882568, "learning_rate": 0.0001, "loss": 0.0409, "step": 20840 }, { "epoch": 132.80254777070064, "grad_norm": 1.4372605085372925, "learning_rate": 0.0001, "loss": 0.0437, "step": 20850 }, { "epoch": 132.86624203821657, "grad_norm": 1.43801748752594, "learning_rate": 0.0001, "loss": 0.0458, "step": 20860 }, { "epoch": 132.9299363057325, "grad_norm": 1.3370589017868042, "learning_rate": 0.0001, "loss": 0.0421, "step": 20870 }, { "epoch": 132.9936305732484, "grad_norm": 1.4504448175430298, "learning_rate": 0.0001, "loss": 0.0504, "step": 20880 }, { "epoch": 133.05732484076432, "grad_norm": 1.355460286140442, "learning_rate": 0.0001, "loss": 0.0447, "step": 20890 }, { "epoch": 133.12101910828025, "grad_norm": 1.2252970933914185, "learning_rate": 0.0001, "loss": 0.0443, "step": 20900 }, { "epoch": 133.18471337579618, "grad_norm": 1.1552661657333374, "learning_rate": 0.0001, "loss": 0.0469, "step": 20910 }, { "epoch": 133.2484076433121, "grad_norm": 1.589751958847046, "learning_rate": 0.0001, "loss": 0.045, "step": 20920 }, { "epoch": 133.31210191082803, "grad_norm": 1.2793352603912354, "learning_rate": 0.0001, "loss": 0.0466, "step": 20930 }, { "epoch": 133.37579617834396, "grad_norm": 1.3847604990005493, "learning_rate": 0.0001, "loss": 0.0485, "step": 20940 }, { "epoch": 133.43949044585986, "grad_norm": 1.2773122787475586, "learning_rate": 0.0001, "loss": 0.0481, "step": 20950 }, { "epoch": 133.5031847133758, "grad_norm": 0.9975866675376892, "learning_rate": 0.0001, "loss": 0.0437, "step": 20960 }, { "epoch": 133.56687898089172, "grad_norm": 1.4888458251953125, "learning_rate": 0.0001, "loss": 0.0416, "step": 20970 }, { "epoch": 133.63057324840764, "grad_norm": 1.0889434814453125, "learning_rate": 0.0001, "loss": 0.0442, "step": 20980 }, { "epoch": 133.69426751592357, "grad_norm": 1.7690134048461914, "learning_rate": 0.0001, "loss": 0.0458, "step": 20990 }, { "epoch": 133.7579617834395, "grad_norm": 2.252171516418457, "learning_rate": 0.0001, "loss": 0.0439, "step": 21000 }, { "epoch": 133.82165605095543, "grad_norm": 1.8575372695922852, "learning_rate": 0.0001, "loss": 0.0426, "step": 21010 }, { "epoch": 133.88535031847132, "grad_norm": 1.806162714958191, "learning_rate": 0.0001, "loss": 0.0454, "step": 21020 }, { "epoch": 133.94904458598725, "grad_norm": 1.4881763458251953, "learning_rate": 0.0001, "loss": 0.0433, "step": 21030 }, { "epoch": 134.01273885350318, "grad_norm": 1.3085821866989136, "learning_rate": 0.0001, "loss": 0.0421, "step": 21040 }, { "epoch": 134.0764331210191, "grad_norm": 1.8608615398406982, "learning_rate": 0.0001, "loss": 0.0399, "step": 21050 }, { "epoch": 134.14012738853503, "grad_norm": 1.3379052877426147, "learning_rate": 0.0001, "loss": 0.0453, "step": 21060 }, { "epoch": 134.20382165605096, "grad_norm": 1.5407570600509644, "learning_rate": 0.0001, "loss": 0.0408, "step": 21070 }, { "epoch": 134.2675159235669, "grad_norm": 1.3423950672149658, "learning_rate": 0.0001, "loss": 0.0409, "step": 21080 }, { "epoch": 134.3312101910828, "grad_norm": 1.4315638542175293, "learning_rate": 0.0001, "loss": 0.0492, "step": 21090 }, { "epoch": 134.39490445859872, "grad_norm": 1.5275095701217651, "learning_rate": 0.0001, "loss": 0.0454, "step": 21100 }, { "epoch": 134.45859872611464, "grad_norm": 1.2167316675186157, "learning_rate": 0.0001, "loss": 0.0434, "step": 21110 }, { "epoch": 134.52229299363057, "grad_norm": 1.5492379665374756, "learning_rate": 0.0001, "loss": 0.0498, "step": 21120 }, { "epoch": 134.5859872611465, "grad_norm": 1.2599387168884277, "learning_rate": 0.0001, "loss": 0.0439, "step": 21130 }, { "epoch": 134.64968152866243, "grad_norm": 1.2309396266937256, "learning_rate": 0.0001, "loss": 0.0422, "step": 21140 }, { "epoch": 134.71337579617835, "grad_norm": 1.343724250793457, "learning_rate": 0.0001, "loss": 0.0476, "step": 21150 }, { "epoch": 134.77707006369425, "grad_norm": 1.452094554901123, "learning_rate": 0.0001, "loss": 0.0456, "step": 21160 }, { "epoch": 134.84076433121018, "grad_norm": 1.4226797819137573, "learning_rate": 0.0001, "loss": 0.0438, "step": 21170 }, { "epoch": 134.9044585987261, "grad_norm": 1.314632534980774, "learning_rate": 0.0001, "loss": 0.0471, "step": 21180 }, { "epoch": 134.96815286624204, "grad_norm": 1.3474767208099365, "learning_rate": 0.0001, "loss": 0.0443, "step": 21190 }, { "epoch": 135.03184713375796, "grad_norm": 1.5690118074417114, "learning_rate": 0.0001, "loss": 0.0428, "step": 21200 }, { "epoch": 135.0955414012739, "grad_norm": 1.4618099927902222, "learning_rate": 0.0001, "loss": 0.0446, "step": 21210 }, { "epoch": 135.15923566878982, "grad_norm": 1.351894736289978, "learning_rate": 0.0001, "loss": 0.042, "step": 21220 }, { "epoch": 135.22292993630575, "grad_norm": 1.4632763862609863, "learning_rate": 0.0001, "loss": 0.0411, "step": 21230 }, { "epoch": 135.28662420382165, "grad_norm": 1.6056733131408691, "learning_rate": 0.0001, "loss": 0.0414, "step": 21240 }, { "epoch": 135.35031847133757, "grad_norm": 1.745894432067871, "learning_rate": 0.0001, "loss": 0.0457, "step": 21250 }, { "epoch": 135.4140127388535, "grad_norm": 1.2234669923782349, "learning_rate": 0.0001, "loss": 0.0444, "step": 21260 }, { "epoch": 135.47770700636943, "grad_norm": 1.5141963958740234, "learning_rate": 0.0001, "loss": 0.0452, "step": 21270 }, { "epoch": 135.54140127388536, "grad_norm": 1.572986125946045, "learning_rate": 0.0001, "loss": 0.0427, "step": 21280 }, { "epoch": 135.60509554140128, "grad_norm": 1.5517609119415283, "learning_rate": 0.0001, "loss": 0.0482, "step": 21290 }, { "epoch": 135.6687898089172, "grad_norm": 1.5174453258514404, "learning_rate": 0.0001, "loss": 0.0435, "step": 21300 }, { "epoch": 135.7324840764331, "grad_norm": 1.3666679859161377, "learning_rate": 0.0001, "loss": 0.0465, "step": 21310 }, { "epoch": 135.79617834394904, "grad_norm": 1.1580995321273804, "learning_rate": 0.0001, "loss": 0.0441, "step": 21320 }, { "epoch": 135.85987261146497, "grad_norm": 1.208633542060852, "learning_rate": 0.0001, "loss": 0.0441, "step": 21330 }, { "epoch": 135.9235668789809, "grad_norm": 1.670583724975586, "learning_rate": 0.0001, "loss": 0.0432, "step": 21340 }, { "epoch": 135.98726114649682, "grad_norm": 1.5721774101257324, "learning_rate": 0.0001, "loss": 0.0432, "step": 21350 }, { "epoch": 136.05095541401275, "grad_norm": 1.604060411453247, "learning_rate": 0.0001, "loss": 0.0457, "step": 21360 }, { "epoch": 136.11464968152868, "grad_norm": 1.645163655281067, "learning_rate": 0.0001, "loss": 0.0428, "step": 21370 }, { "epoch": 136.17834394904457, "grad_norm": 1.372017502784729, "learning_rate": 0.0001, "loss": 0.0433, "step": 21380 }, { "epoch": 136.2420382165605, "grad_norm": 1.4773629903793335, "learning_rate": 0.0001, "loss": 0.0458, "step": 21390 }, { "epoch": 136.30573248407643, "grad_norm": 1.5948882102966309, "learning_rate": 0.0001, "loss": 0.0457, "step": 21400 }, { "epoch": 136.36942675159236, "grad_norm": 1.507340431213379, "learning_rate": 0.0001, "loss": 0.0418, "step": 21410 }, { "epoch": 136.43312101910828, "grad_norm": 1.1641616821289062, "learning_rate": 0.0001, "loss": 0.0453, "step": 21420 }, { "epoch": 136.4968152866242, "grad_norm": 1.552794098854065, "learning_rate": 0.0001, "loss": 0.0446, "step": 21430 }, { "epoch": 136.56050955414014, "grad_norm": 1.624616265296936, "learning_rate": 0.0001, "loss": 0.0478, "step": 21440 }, { "epoch": 136.62420382165604, "grad_norm": 1.65762460231781, "learning_rate": 0.0001, "loss": 0.0435, "step": 21450 }, { "epoch": 136.68789808917197, "grad_norm": 1.5024904012680054, "learning_rate": 0.0001, "loss": 0.0418, "step": 21460 }, { "epoch": 136.7515923566879, "grad_norm": 1.6629481315612793, "learning_rate": 0.0001, "loss": 0.0448, "step": 21470 }, { "epoch": 136.81528662420382, "grad_norm": 1.5806620121002197, "learning_rate": 0.0001, "loss": 0.0441, "step": 21480 }, { "epoch": 136.87898089171975, "grad_norm": 1.374289870262146, "learning_rate": 0.0001, "loss": 0.0427, "step": 21490 }, { "epoch": 136.94267515923568, "grad_norm": 1.6676571369171143, "learning_rate": 0.0001, "loss": 0.0452, "step": 21500 }, { "epoch": 137.0063694267516, "grad_norm": 1.4934358596801758, "learning_rate": 0.0001, "loss": 0.0399, "step": 21510 }, { "epoch": 137.0700636942675, "grad_norm": 1.2476247549057007, "learning_rate": 0.0001, "loss": 0.0416, "step": 21520 }, { "epoch": 137.13375796178343, "grad_norm": 1.5656541585922241, "learning_rate": 0.0001, "loss": 0.046, "step": 21530 }, { "epoch": 137.19745222929936, "grad_norm": 1.6239479780197144, "learning_rate": 0.0001, "loss": 0.0389, "step": 21540 }, { "epoch": 137.2611464968153, "grad_norm": 2.0390408039093018, "learning_rate": 0.0001, "loss": 0.0426, "step": 21550 }, { "epoch": 137.3248407643312, "grad_norm": 1.9610189199447632, "learning_rate": 0.0001, "loss": 0.0408, "step": 21560 }, { "epoch": 137.38853503184714, "grad_norm": 1.8346928358078003, "learning_rate": 0.0001, "loss": 0.0365, "step": 21570 }, { "epoch": 137.45222929936307, "grad_norm": 1.2199339866638184, "learning_rate": 0.0001, "loss": 0.041, "step": 21580 }, { "epoch": 137.51592356687897, "grad_norm": 1.3178715705871582, "learning_rate": 0.0001, "loss": 0.0441, "step": 21590 }, { "epoch": 137.5796178343949, "grad_norm": 1.197360634803772, "learning_rate": 0.0001, "loss": 0.0401, "step": 21600 }, { "epoch": 137.64331210191082, "grad_norm": 1.4411557912826538, "learning_rate": 0.0001, "loss": 0.0415, "step": 21610 }, { "epoch": 137.70700636942675, "grad_norm": 1.3894309997558594, "learning_rate": 0.0001, "loss": 0.046, "step": 21620 }, { "epoch": 137.77070063694268, "grad_norm": 1.5614526271820068, "learning_rate": 0.0001, "loss": 0.0436, "step": 21630 }, { "epoch": 137.8343949044586, "grad_norm": 1.1123480796813965, "learning_rate": 0.0001, "loss": 0.0432, "step": 21640 }, { "epoch": 137.89808917197453, "grad_norm": 1.7449383735656738, "learning_rate": 0.0001, "loss": 0.0454, "step": 21650 }, { "epoch": 137.96178343949043, "grad_norm": 1.49588942527771, "learning_rate": 0.0001, "loss": 0.0448, "step": 21660 }, { "epoch": 138.02547770700636, "grad_norm": 1.1962698698043823, "learning_rate": 0.0001, "loss": 0.0427, "step": 21670 }, { "epoch": 138.0891719745223, "grad_norm": 1.4299660921096802, "learning_rate": 0.0001, "loss": 0.045, "step": 21680 }, { "epoch": 138.15286624203821, "grad_norm": 1.024121642112732, "learning_rate": 0.0001, "loss": 0.0437, "step": 21690 }, { "epoch": 138.21656050955414, "grad_norm": 1.2534643411636353, "learning_rate": 0.0001, "loss": 0.041, "step": 21700 }, { "epoch": 138.28025477707007, "grad_norm": 1.144388198852539, "learning_rate": 0.0001, "loss": 0.0386, "step": 21710 }, { "epoch": 138.343949044586, "grad_norm": 1.044029712677002, "learning_rate": 0.0001, "loss": 0.0411, "step": 21720 }, { "epoch": 138.40764331210192, "grad_norm": 1.8442903757095337, "learning_rate": 0.0001, "loss": 0.0422, "step": 21730 }, { "epoch": 138.47133757961782, "grad_norm": 1.4041471481323242, "learning_rate": 0.0001, "loss": 0.0446, "step": 21740 }, { "epoch": 138.53503184713375, "grad_norm": 1.314470887184143, "learning_rate": 0.0001, "loss": 0.0426, "step": 21750 }, { "epoch": 138.59872611464968, "grad_norm": 1.4445257186889648, "learning_rate": 0.0001, "loss": 0.0441, "step": 21760 }, { "epoch": 138.6624203821656, "grad_norm": 1.4139341115951538, "learning_rate": 0.0001, "loss": 0.0422, "step": 21770 }, { "epoch": 138.72611464968153, "grad_norm": 1.1768882274627686, "learning_rate": 0.0001, "loss": 0.0422, "step": 21780 }, { "epoch": 138.78980891719746, "grad_norm": 1.4598358869552612, "learning_rate": 0.0001, "loss": 0.0404, "step": 21790 }, { "epoch": 138.8535031847134, "grad_norm": 1.3230352401733398, "learning_rate": 0.0001, "loss": 0.044, "step": 21800 }, { "epoch": 138.9171974522293, "grad_norm": 1.2361992597579956, "learning_rate": 0.0001, "loss": 0.0464, "step": 21810 }, { "epoch": 138.98089171974522, "grad_norm": 1.2741343975067139, "learning_rate": 0.0001, "loss": 0.0413, "step": 21820 }, { "epoch": 139.04458598726114, "grad_norm": 1.220578670501709, "learning_rate": 0.0001, "loss": 0.043, "step": 21830 }, { "epoch": 139.10828025477707, "grad_norm": 1.510947823524475, "learning_rate": 0.0001, "loss": 0.0442, "step": 21840 }, { "epoch": 139.171974522293, "grad_norm": 1.0053277015686035, "learning_rate": 0.0001, "loss": 0.0409, "step": 21850 }, { "epoch": 139.23566878980893, "grad_norm": 1.334843635559082, "learning_rate": 0.0001, "loss": 0.046, "step": 21860 }, { "epoch": 139.29936305732485, "grad_norm": 1.3517807722091675, "learning_rate": 0.0001, "loss": 0.0427, "step": 21870 }, { "epoch": 139.36305732484075, "grad_norm": 1.5772984027862549, "learning_rate": 0.0001, "loss": 0.0431, "step": 21880 }, { "epoch": 139.42675159235668, "grad_norm": 0.9610340595245361, "learning_rate": 0.0001, "loss": 0.0463, "step": 21890 }, { "epoch": 139.4904458598726, "grad_norm": 1.1861521005630493, "learning_rate": 0.0001, "loss": 0.0409, "step": 21900 }, { "epoch": 139.55414012738854, "grad_norm": 1.1799538135528564, "learning_rate": 0.0001, "loss": 0.0427, "step": 21910 }, { "epoch": 139.61783439490446, "grad_norm": 1.2363333702087402, "learning_rate": 0.0001, "loss": 0.0415, "step": 21920 }, { "epoch": 139.6815286624204, "grad_norm": 1.389320731163025, "learning_rate": 0.0001, "loss": 0.0424, "step": 21930 }, { "epoch": 139.74522292993632, "grad_norm": 1.2420525550842285, "learning_rate": 0.0001, "loss": 0.0466, "step": 21940 }, { "epoch": 139.80891719745222, "grad_norm": 1.527825117111206, "learning_rate": 0.0001, "loss": 0.0451, "step": 21950 }, { "epoch": 139.87261146496814, "grad_norm": 1.4327555894851685, "learning_rate": 0.0001, "loss": 0.0441, "step": 21960 }, { "epoch": 139.93630573248407, "grad_norm": 1.1813627481460571, "learning_rate": 0.0001, "loss": 0.0396, "step": 21970 }, { "epoch": 140.0, "grad_norm": 1.3851382732391357, "learning_rate": 0.0001, "loss": 0.0429, "step": 21980 }, { "epoch": 140.06369426751593, "grad_norm": 1.2626839876174927, "learning_rate": 0.0001, "loss": 0.0458, "step": 21990 }, { "epoch": 140.12738853503186, "grad_norm": 1.245194435119629, "learning_rate": 0.0001, "loss": 0.0398, "step": 22000 }, { "epoch": 140.19108280254778, "grad_norm": 1.448860764503479, "learning_rate": 0.0001, "loss": 0.0421, "step": 22010 }, { "epoch": 140.25477707006368, "grad_norm": 1.4270838499069214, "learning_rate": 0.0001, "loss": 0.0459, "step": 22020 }, { "epoch": 140.3184713375796, "grad_norm": 1.2744206190109253, "learning_rate": 0.0001, "loss": 0.0395, "step": 22030 }, { "epoch": 140.38216560509554, "grad_norm": 1.2130823135375977, "learning_rate": 0.0001, "loss": 0.0418, "step": 22040 }, { "epoch": 140.44585987261146, "grad_norm": 1.2961852550506592, "learning_rate": 0.0001, "loss": 0.0419, "step": 22050 }, { "epoch": 140.5095541401274, "grad_norm": 1.3218719959259033, "learning_rate": 0.0001, "loss": 0.0422, "step": 22060 }, { "epoch": 140.57324840764332, "grad_norm": 1.3571218252182007, "learning_rate": 0.0001, "loss": 0.0418, "step": 22070 }, { "epoch": 140.63694267515925, "grad_norm": 1.533380150794983, "learning_rate": 0.0001, "loss": 0.0451, "step": 22080 }, { "epoch": 140.70063694267515, "grad_norm": 1.2963438034057617, "learning_rate": 0.0001, "loss": 0.0394, "step": 22090 }, { "epoch": 140.76433121019107, "grad_norm": 1.0851202011108398, "learning_rate": 0.0001, "loss": 0.0413, "step": 22100 }, { "epoch": 140.828025477707, "grad_norm": 1.1562855243682861, "learning_rate": 0.0001, "loss": 0.0451, "step": 22110 }, { "epoch": 140.89171974522293, "grad_norm": 1.0355968475341797, "learning_rate": 0.0001, "loss": 0.0416, "step": 22120 }, { "epoch": 140.95541401273886, "grad_norm": 0.9164203405380249, "learning_rate": 0.0001, "loss": 0.0444, "step": 22130 }, { "epoch": 141.01910828025478, "grad_norm": 1.3679225444793701, "learning_rate": 0.0001, "loss": 0.0442, "step": 22140 }, { "epoch": 141.0828025477707, "grad_norm": 1.1617250442504883, "learning_rate": 0.0001, "loss": 0.0399, "step": 22150 }, { "epoch": 141.1464968152866, "grad_norm": 1.2494256496429443, "learning_rate": 0.0001, "loss": 0.0444, "step": 22160 }, { "epoch": 141.21019108280254, "grad_norm": 1.3566529750823975, "learning_rate": 0.0001, "loss": 0.0444, "step": 22170 }, { "epoch": 141.27388535031847, "grad_norm": 1.0102306604385376, "learning_rate": 0.0001, "loss": 0.0424, "step": 22180 }, { "epoch": 141.3375796178344, "grad_norm": 1.3702481985092163, "learning_rate": 0.0001, "loss": 0.0444, "step": 22190 }, { "epoch": 141.40127388535032, "grad_norm": 1.2529886960983276, "learning_rate": 0.0001, "loss": 0.0421, "step": 22200 }, { "epoch": 141.46496815286625, "grad_norm": 1.483176827430725, "learning_rate": 0.0001, "loss": 0.0409, "step": 22210 }, { "epoch": 141.52866242038218, "grad_norm": 1.2816500663757324, "learning_rate": 0.0001, "loss": 0.0453, "step": 22220 }, { "epoch": 141.59235668789808, "grad_norm": 1.1576615571975708, "learning_rate": 0.0001, "loss": 0.0439, "step": 22230 }, { "epoch": 141.656050955414, "grad_norm": 1.573695421218872, "learning_rate": 0.0001, "loss": 0.044, "step": 22240 }, { "epoch": 141.71974522292993, "grad_norm": 1.16523277759552, "learning_rate": 0.0001, "loss": 0.0453, "step": 22250 }, { "epoch": 141.78343949044586, "grad_norm": 1.0451310873031616, "learning_rate": 0.0001, "loss": 0.045, "step": 22260 }, { "epoch": 141.84713375796179, "grad_norm": 1.2070637941360474, "learning_rate": 0.0001, "loss": 0.0454, "step": 22270 }, { "epoch": 141.9108280254777, "grad_norm": 1.1656235456466675, "learning_rate": 0.0001, "loss": 0.0447, "step": 22280 }, { "epoch": 141.97452229299364, "grad_norm": 0.9757199883460999, "learning_rate": 0.0001, "loss": 0.0471, "step": 22290 }, { "epoch": 142.03821656050957, "grad_norm": 1.2717387676239014, "learning_rate": 0.0001, "loss": 0.0423, "step": 22300 }, { "epoch": 142.10191082802547, "grad_norm": 1.2716256380081177, "learning_rate": 0.0001, "loss": 0.0466, "step": 22310 }, { "epoch": 142.1656050955414, "grad_norm": 1.0074771642684937, "learning_rate": 0.0001, "loss": 0.0494, "step": 22320 }, { "epoch": 142.22929936305732, "grad_norm": 1.2366642951965332, "learning_rate": 0.0001, "loss": 0.0446, "step": 22330 }, { "epoch": 142.29299363057325, "grad_norm": 1.2824221849441528, "learning_rate": 0.0001, "loss": 0.0412, "step": 22340 }, { "epoch": 142.35668789808918, "grad_norm": 1.2692006826400757, "learning_rate": 0.0001, "loss": 0.0475, "step": 22350 }, { "epoch": 142.4203821656051, "grad_norm": 1.3925373554229736, "learning_rate": 0.0001, "loss": 0.0451, "step": 22360 }, { "epoch": 142.48407643312103, "grad_norm": 1.3039758205413818, "learning_rate": 0.0001, "loss": 0.0433, "step": 22370 }, { "epoch": 142.54777070063693, "grad_norm": 1.2802201509475708, "learning_rate": 0.0001, "loss": 0.0449, "step": 22380 }, { "epoch": 142.61146496815286, "grad_norm": 1.0021967887878418, "learning_rate": 0.0001, "loss": 0.0435, "step": 22390 }, { "epoch": 142.6751592356688, "grad_norm": 1.1789592504501343, "learning_rate": 0.0001, "loss": 0.0415, "step": 22400 }, { "epoch": 142.7388535031847, "grad_norm": 1.3234766721725464, "learning_rate": 0.0001, "loss": 0.0479, "step": 22410 }, { "epoch": 142.80254777070064, "grad_norm": 0.8375675678253174, "learning_rate": 0.0001, "loss": 0.0421, "step": 22420 }, { "epoch": 142.86624203821657, "grad_norm": 1.2763986587524414, "learning_rate": 0.0001, "loss": 0.0432, "step": 22430 }, { "epoch": 142.9299363057325, "grad_norm": 1.2676764726638794, "learning_rate": 0.0001, "loss": 0.0463, "step": 22440 }, { "epoch": 142.9936305732484, "grad_norm": 1.1814168691635132, "learning_rate": 0.0001, "loss": 0.0506, "step": 22450 }, { "epoch": 143.05732484076432, "grad_norm": 1.1591368913650513, "learning_rate": 0.0001, "loss": 0.0422, "step": 22460 }, { "epoch": 143.12101910828025, "grad_norm": 1.4538908004760742, "learning_rate": 0.0001, "loss": 0.0494, "step": 22470 }, { "epoch": 143.18471337579618, "grad_norm": 1.3721246719360352, "learning_rate": 0.0001, "loss": 0.0408, "step": 22480 }, { "epoch": 143.2484076433121, "grad_norm": 1.2310339212417603, "learning_rate": 0.0001, "loss": 0.0418, "step": 22490 }, { "epoch": 143.31210191082803, "grad_norm": 1.444654941558838, "learning_rate": 0.0001, "loss": 0.0435, "step": 22500 }, { "epoch": 143.37579617834396, "grad_norm": 1.2865713834762573, "learning_rate": 0.0001, "loss": 0.0386, "step": 22510 }, { "epoch": 143.43949044585986, "grad_norm": 1.2526521682739258, "learning_rate": 0.0001, "loss": 0.0409, "step": 22520 }, { "epoch": 143.5031847133758, "grad_norm": 1.3417690992355347, "learning_rate": 0.0001, "loss": 0.0449, "step": 22530 }, { "epoch": 143.56687898089172, "grad_norm": 1.2658878564834595, "learning_rate": 0.0001, "loss": 0.0437, "step": 22540 }, { "epoch": 143.63057324840764, "grad_norm": 1.4027117490768433, "learning_rate": 0.0001, "loss": 0.0392, "step": 22550 }, { "epoch": 143.69426751592357, "grad_norm": 0.9965071678161621, "learning_rate": 0.0001, "loss": 0.042, "step": 22560 }, { "epoch": 143.7579617834395, "grad_norm": 1.35568368434906, "learning_rate": 0.0001, "loss": 0.0394, "step": 22570 }, { "epoch": 143.82165605095543, "grad_norm": 1.5253947973251343, "learning_rate": 0.0001, "loss": 0.0449, "step": 22580 }, { "epoch": 143.88535031847132, "grad_norm": 1.0227978229522705, "learning_rate": 0.0001, "loss": 0.0428, "step": 22590 }, { "epoch": 143.94904458598725, "grad_norm": 1.2814702987670898, "learning_rate": 0.0001, "loss": 0.043, "step": 22600 }, { "epoch": 144.01273885350318, "grad_norm": 1.3480545282363892, "learning_rate": 0.0001, "loss": 0.0429, "step": 22610 }, { "epoch": 144.0764331210191, "grad_norm": 1.1268815994262695, "learning_rate": 0.0001, "loss": 0.0426, "step": 22620 }, { "epoch": 144.14012738853503, "grad_norm": 1.1760576963424683, "learning_rate": 0.0001, "loss": 0.0414, "step": 22630 }, { "epoch": 144.20382165605096, "grad_norm": 1.2630411386489868, "learning_rate": 0.0001, "loss": 0.0431, "step": 22640 }, { "epoch": 144.2675159235669, "grad_norm": 1.2833788394927979, "learning_rate": 0.0001, "loss": 0.0372, "step": 22650 }, { "epoch": 144.3312101910828, "grad_norm": 1.1789342164993286, "learning_rate": 0.0001, "loss": 0.0419, "step": 22660 }, { "epoch": 144.39490445859872, "grad_norm": 1.058035969734192, "learning_rate": 0.0001, "loss": 0.044, "step": 22670 }, { "epoch": 144.45859872611464, "grad_norm": 1.2139815092086792, "learning_rate": 0.0001, "loss": 0.0388, "step": 22680 }, { "epoch": 144.52229299363057, "grad_norm": 1.223839282989502, "learning_rate": 0.0001, "loss": 0.0435, "step": 22690 }, { "epoch": 144.5859872611465, "grad_norm": 1.338431477546692, "learning_rate": 0.0001, "loss": 0.0414, "step": 22700 }, { "epoch": 144.64968152866243, "grad_norm": 1.3117352724075317, "learning_rate": 0.0001, "loss": 0.0456, "step": 22710 }, { "epoch": 144.71337579617835, "grad_norm": 1.2782429456710815, "learning_rate": 0.0001, "loss": 0.0423, "step": 22720 }, { "epoch": 144.77707006369425, "grad_norm": 1.3020769357681274, "learning_rate": 0.0001, "loss": 0.0432, "step": 22730 }, { "epoch": 144.84076433121018, "grad_norm": 1.0250300168991089, "learning_rate": 0.0001, "loss": 0.0405, "step": 22740 }, { "epoch": 144.9044585987261, "grad_norm": 1.0725499391555786, "learning_rate": 0.0001, "loss": 0.0392, "step": 22750 }, { "epoch": 144.96815286624204, "grad_norm": 1.49922513961792, "learning_rate": 0.0001, "loss": 0.0409, "step": 22760 }, { "epoch": 145.03184713375796, "grad_norm": 1.3055411577224731, "learning_rate": 0.0001, "loss": 0.0419, "step": 22770 }, { "epoch": 145.0955414012739, "grad_norm": 1.1849939823150635, "learning_rate": 0.0001, "loss": 0.0413, "step": 22780 }, { "epoch": 145.15923566878982, "grad_norm": 1.0431362390518188, "learning_rate": 0.0001, "loss": 0.0382, "step": 22790 }, { "epoch": 145.22292993630575, "grad_norm": 1.2323949337005615, "learning_rate": 0.0001, "loss": 0.0392, "step": 22800 }, { "epoch": 145.28662420382165, "grad_norm": 1.009100079536438, "learning_rate": 0.0001, "loss": 0.043, "step": 22810 }, { "epoch": 145.35031847133757, "grad_norm": 1.0391285419464111, "learning_rate": 0.0001, "loss": 0.0431, "step": 22820 }, { "epoch": 145.4140127388535, "grad_norm": 1.3155269622802734, "learning_rate": 0.0001, "loss": 0.0453, "step": 22830 }, { "epoch": 145.47770700636943, "grad_norm": 1.4609779119491577, "learning_rate": 0.0001, "loss": 0.0415, "step": 22840 }, { "epoch": 145.54140127388536, "grad_norm": 1.4144675731658936, "learning_rate": 0.0001, "loss": 0.0408, "step": 22850 }, { "epoch": 145.60509554140128, "grad_norm": 1.011778473854065, "learning_rate": 0.0001, "loss": 0.0401, "step": 22860 }, { "epoch": 145.6687898089172, "grad_norm": 1.088677167892456, "learning_rate": 0.0001, "loss": 0.045, "step": 22870 }, { "epoch": 145.7324840764331, "grad_norm": 1.4235870838165283, "learning_rate": 0.0001, "loss": 0.0401, "step": 22880 }, { "epoch": 145.79617834394904, "grad_norm": 1.4076781272888184, "learning_rate": 0.0001, "loss": 0.0429, "step": 22890 }, { "epoch": 145.85987261146497, "grad_norm": 1.4225860834121704, "learning_rate": 0.0001, "loss": 0.0451, "step": 22900 }, { "epoch": 145.9235668789809, "grad_norm": 1.0326751470565796, "learning_rate": 0.0001, "loss": 0.0414, "step": 22910 }, { "epoch": 145.98726114649682, "grad_norm": 1.3534477949142456, "learning_rate": 0.0001, "loss": 0.0435, "step": 22920 }, { "epoch": 146.05095541401275, "grad_norm": 1.148292899131775, "learning_rate": 0.0001, "loss": 0.041, "step": 22930 }, { "epoch": 146.11464968152868, "grad_norm": 1.1538774967193604, "learning_rate": 0.0001, "loss": 0.0388, "step": 22940 }, { "epoch": 146.17834394904457, "grad_norm": 1.0720783472061157, "learning_rate": 0.0001, "loss": 0.038, "step": 22950 }, { "epoch": 146.2420382165605, "grad_norm": 1.2162792682647705, "learning_rate": 0.0001, "loss": 0.0376, "step": 22960 }, { "epoch": 146.30573248407643, "grad_norm": 1.3609551191329956, "learning_rate": 0.0001, "loss": 0.0374, "step": 22970 }, { "epoch": 146.36942675159236, "grad_norm": 1.6826945543289185, "learning_rate": 0.0001, "loss": 0.0404, "step": 22980 }, { "epoch": 146.43312101910828, "grad_norm": 1.3101624250411987, "learning_rate": 0.0001, "loss": 0.0422, "step": 22990 }, { "epoch": 146.4968152866242, "grad_norm": 1.341106653213501, "learning_rate": 0.0001, "loss": 0.0439, "step": 23000 }, { "epoch": 146.56050955414014, "grad_norm": 1.3418002128601074, "learning_rate": 0.0001, "loss": 0.0413, "step": 23010 }, { "epoch": 146.62420382165604, "grad_norm": 1.277490258216858, "learning_rate": 0.0001, "loss": 0.042, "step": 23020 }, { "epoch": 146.68789808917197, "grad_norm": 1.1126214265823364, "learning_rate": 0.0001, "loss": 0.0418, "step": 23030 }, { "epoch": 146.7515923566879, "grad_norm": 1.0976314544677734, "learning_rate": 0.0001, "loss": 0.0437, "step": 23040 }, { "epoch": 146.81528662420382, "grad_norm": 1.5884140729904175, "learning_rate": 0.0001, "loss": 0.0458, "step": 23050 }, { "epoch": 146.87898089171975, "grad_norm": 1.2413733005523682, "learning_rate": 0.0001, "loss": 0.0394, "step": 23060 }, { "epoch": 146.94267515923568, "grad_norm": 1.5365949869155884, "learning_rate": 0.0001, "loss": 0.043, "step": 23070 }, { "epoch": 147.0063694267516, "grad_norm": 1.403793215751648, "learning_rate": 0.0001, "loss": 0.04, "step": 23080 }, { "epoch": 147.0700636942675, "grad_norm": 1.4962934255599976, "learning_rate": 0.0001, "loss": 0.0422, "step": 23090 }, { "epoch": 147.13375796178343, "grad_norm": 1.3011685609817505, "learning_rate": 0.0001, "loss": 0.0395, "step": 23100 }, { "epoch": 147.19745222929936, "grad_norm": 1.2333747148513794, "learning_rate": 0.0001, "loss": 0.0415, "step": 23110 }, { "epoch": 147.2611464968153, "grad_norm": 1.5599702596664429, "learning_rate": 0.0001, "loss": 0.0387, "step": 23120 }, { "epoch": 147.3248407643312, "grad_norm": 1.582851767539978, "learning_rate": 0.0001, "loss": 0.0416, "step": 23130 }, { "epoch": 147.38853503184714, "grad_norm": 1.106836199760437, "learning_rate": 0.0001, "loss": 0.0405, "step": 23140 }, { "epoch": 147.45222929936307, "grad_norm": 1.2367446422576904, "learning_rate": 0.0001, "loss": 0.0446, "step": 23150 }, { "epoch": 147.51592356687897, "grad_norm": 1.2843163013458252, "learning_rate": 0.0001, "loss": 0.0427, "step": 23160 }, { "epoch": 147.5796178343949, "grad_norm": 1.4917149543762207, "learning_rate": 0.0001, "loss": 0.0409, "step": 23170 }, { "epoch": 147.64331210191082, "grad_norm": 1.4322692155838013, "learning_rate": 0.0001, "loss": 0.0407, "step": 23180 }, { "epoch": 147.70700636942675, "grad_norm": 1.1429435014724731, "learning_rate": 0.0001, "loss": 0.0431, "step": 23190 }, { "epoch": 147.77070063694268, "grad_norm": 1.0270092487335205, "learning_rate": 0.0001, "loss": 0.0419, "step": 23200 }, { "epoch": 147.8343949044586, "grad_norm": 1.405318021774292, "learning_rate": 0.0001, "loss": 0.0409, "step": 23210 }, { "epoch": 147.89808917197453, "grad_norm": 1.3831537961959839, "learning_rate": 0.0001, "loss": 0.0411, "step": 23220 }, { "epoch": 147.96178343949043, "grad_norm": 1.2598471641540527, "learning_rate": 0.0001, "loss": 0.0425, "step": 23230 }, { "epoch": 148.02547770700636, "grad_norm": 1.3111273050308228, "learning_rate": 0.0001, "loss": 0.0368, "step": 23240 }, { "epoch": 148.0891719745223, "grad_norm": 1.2909523248672485, "learning_rate": 0.0001, "loss": 0.0439, "step": 23250 }, { "epoch": 148.15286624203821, "grad_norm": 1.7182209491729736, "learning_rate": 0.0001, "loss": 0.0426, "step": 23260 }, { "epoch": 148.21656050955414, "grad_norm": 1.1350971460342407, "learning_rate": 0.0001, "loss": 0.0445, "step": 23270 }, { "epoch": 148.28025477707007, "grad_norm": 1.1992568969726562, "learning_rate": 0.0001, "loss": 0.0422, "step": 23280 }, { "epoch": 148.343949044586, "grad_norm": 1.2901707887649536, "learning_rate": 0.0001, "loss": 0.0471, "step": 23290 }, { "epoch": 148.40764331210192, "grad_norm": 1.5449167490005493, "learning_rate": 0.0001, "loss": 0.0428, "step": 23300 }, { "epoch": 148.47133757961782, "grad_norm": 1.4885176420211792, "learning_rate": 0.0001, "loss": 0.0434, "step": 23310 }, { "epoch": 148.53503184713375, "grad_norm": 1.3406825065612793, "learning_rate": 0.0001, "loss": 0.045, "step": 23320 }, { "epoch": 148.59872611464968, "grad_norm": 1.1862071752548218, "learning_rate": 0.0001, "loss": 0.0428, "step": 23330 }, { "epoch": 148.6624203821656, "grad_norm": 1.194311499595642, "learning_rate": 0.0001, "loss": 0.0412, "step": 23340 }, { "epoch": 148.72611464968153, "grad_norm": 1.0305168628692627, "learning_rate": 0.0001, "loss": 0.0425, "step": 23350 }, { "epoch": 148.78980891719746, "grad_norm": 1.2172824144363403, "learning_rate": 0.0001, "loss": 0.0468, "step": 23360 }, { "epoch": 148.8535031847134, "grad_norm": 0.8877734541893005, "learning_rate": 0.0001, "loss": 0.0432, "step": 23370 }, { "epoch": 148.9171974522293, "grad_norm": 0.9375904202461243, "learning_rate": 0.0001, "loss": 0.0408, "step": 23380 }, { "epoch": 148.98089171974522, "grad_norm": 1.2367624044418335, "learning_rate": 0.0001, "loss": 0.0408, "step": 23390 }, { "epoch": 149.04458598726114, "grad_norm": 1.2861137390136719, "learning_rate": 0.0001, "loss": 0.04, "step": 23400 }, { "epoch": 149.10828025477707, "grad_norm": 1.2598673105239868, "learning_rate": 0.0001, "loss": 0.0404, "step": 23410 }, { "epoch": 149.171974522293, "grad_norm": 1.409626841545105, "learning_rate": 0.0001, "loss": 0.0403, "step": 23420 }, { "epoch": 149.23566878980893, "grad_norm": 1.0983492136001587, "learning_rate": 0.0001, "loss": 0.0445, "step": 23430 }, { "epoch": 149.29936305732485, "grad_norm": 1.155222773551941, "learning_rate": 0.0001, "loss": 0.0419, "step": 23440 }, { "epoch": 149.36305732484075, "grad_norm": 1.2112417221069336, "learning_rate": 0.0001, "loss": 0.0401, "step": 23450 }, { "epoch": 149.42675159235668, "grad_norm": 1.352385401725769, "learning_rate": 0.0001, "loss": 0.042, "step": 23460 }, { "epoch": 149.4904458598726, "grad_norm": 1.0877233743667603, "learning_rate": 0.0001, "loss": 0.0393, "step": 23470 }, { "epoch": 149.55414012738854, "grad_norm": 1.1147233247756958, "learning_rate": 0.0001, "loss": 0.0398, "step": 23480 }, { "epoch": 149.61783439490446, "grad_norm": 1.377508282661438, "learning_rate": 0.0001, "loss": 0.0403, "step": 23490 }, { "epoch": 149.6815286624204, "grad_norm": 1.4945098161697388, "learning_rate": 0.0001, "loss": 0.0415, "step": 23500 }, { "epoch": 149.74522292993632, "grad_norm": 1.2044917345046997, "learning_rate": 0.0001, "loss": 0.0469, "step": 23510 }, { "epoch": 149.80891719745222, "grad_norm": 1.4171329736709595, "learning_rate": 0.0001, "loss": 0.0402, "step": 23520 }, { "epoch": 149.87261146496814, "grad_norm": 1.5447747707366943, "learning_rate": 0.0001, "loss": 0.0447, "step": 23530 }, { "epoch": 149.93630573248407, "grad_norm": 1.2788276672363281, "learning_rate": 0.0001, "loss": 0.0404, "step": 23540 }, { "epoch": 150.0, "grad_norm": 1.2986900806427002, "learning_rate": 0.0001, "loss": 0.043, "step": 23550 }, { "epoch": 150.06369426751593, "grad_norm": 1.4466444253921509, "learning_rate": 0.0001, "loss": 0.0402, "step": 23560 }, { "epoch": 150.12738853503186, "grad_norm": 1.2912719249725342, "learning_rate": 0.0001, "loss": 0.0412, "step": 23570 }, { "epoch": 150.19108280254778, "grad_norm": 2.164609670639038, "learning_rate": 0.0001, "loss": 0.0477, "step": 23580 }, { "epoch": 150.25477707006368, "grad_norm": 1.9602553844451904, "learning_rate": 0.0001, "loss": 0.0425, "step": 23590 }, { "epoch": 150.3184713375796, "grad_norm": 1.9126790761947632, "learning_rate": 0.0001, "loss": 0.042, "step": 23600 }, { "epoch": 150.38216560509554, "grad_norm": 1.9422523975372314, "learning_rate": 0.0001, "loss": 0.0442, "step": 23610 }, { "epoch": 150.44585987261146, "grad_norm": 1.4904667139053345, "learning_rate": 0.0001, "loss": 0.0395, "step": 23620 }, { "epoch": 150.5095541401274, "grad_norm": 1.6396511793136597, "learning_rate": 0.0001, "loss": 0.0375, "step": 23630 }, { "epoch": 150.57324840764332, "grad_norm": 1.4071944952011108, "learning_rate": 0.0001, "loss": 0.0388, "step": 23640 }, { "epoch": 150.63694267515925, "grad_norm": 1.4646013975143433, "learning_rate": 0.0001, "loss": 0.0407, "step": 23650 }, { "epoch": 150.70063694267515, "grad_norm": 1.399474859237671, "learning_rate": 0.0001, "loss": 0.0377, "step": 23660 }, { "epoch": 150.76433121019107, "grad_norm": 1.2156884670257568, "learning_rate": 0.0001, "loss": 0.0381, "step": 23670 }, { "epoch": 150.828025477707, "grad_norm": 1.8310686349868774, "learning_rate": 0.0001, "loss": 0.0434, "step": 23680 }, { "epoch": 150.89171974522293, "grad_norm": 1.4532521963119507, "learning_rate": 0.0001, "loss": 0.0438, "step": 23690 }, { "epoch": 150.95541401273886, "grad_norm": 1.2975817918777466, "learning_rate": 0.0001, "loss": 0.0402, "step": 23700 }, { "epoch": 151.01910828025478, "grad_norm": 1.6516915559768677, "learning_rate": 0.0001, "loss": 0.0406, "step": 23710 }, { "epoch": 151.0828025477707, "grad_norm": 1.6906003952026367, "learning_rate": 0.0001, "loss": 0.0395, "step": 23720 }, { "epoch": 151.1464968152866, "grad_norm": 1.5646883249282837, "learning_rate": 0.0001, "loss": 0.0362, "step": 23730 }, { "epoch": 151.21019108280254, "grad_norm": 1.4517797231674194, "learning_rate": 0.0001, "loss": 0.0402, "step": 23740 }, { "epoch": 151.27388535031847, "grad_norm": 1.3119711875915527, "learning_rate": 0.0001, "loss": 0.0353, "step": 23750 }, { "epoch": 151.3375796178344, "grad_norm": 1.712907314300537, "learning_rate": 0.0001, "loss": 0.0373, "step": 23760 }, { "epoch": 151.40127388535032, "grad_norm": 1.3833366632461548, "learning_rate": 0.0001, "loss": 0.0425, "step": 23770 }, { "epoch": 151.46496815286625, "grad_norm": 1.415605068206787, "learning_rate": 0.0001, "loss": 0.0415, "step": 23780 }, { "epoch": 151.52866242038218, "grad_norm": 1.1499660015106201, "learning_rate": 0.0001, "loss": 0.0396, "step": 23790 }, { "epoch": 151.59235668789808, "grad_norm": 1.1327741146087646, "learning_rate": 0.0001, "loss": 0.0406, "step": 23800 }, { "epoch": 151.656050955414, "grad_norm": 1.1698484420776367, "learning_rate": 0.0001, "loss": 0.0388, "step": 23810 }, { "epoch": 151.71974522292993, "grad_norm": 1.0981069803237915, "learning_rate": 0.0001, "loss": 0.0394, "step": 23820 }, { "epoch": 151.78343949044586, "grad_norm": 1.037697196006775, "learning_rate": 0.0001, "loss": 0.0403, "step": 23830 }, { "epoch": 151.84713375796179, "grad_norm": 1.2339938879013062, "learning_rate": 0.0001, "loss": 0.0427, "step": 23840 }, { "epoch": 151.9108280254777, "grad_norm": 1.6200162172317505, "learning_rate": 0.0001, "loss": 0.0456, "step": 23850 }, { "epoch": 151.97452229299364, "grad_norm": 1.2143545150756836, "learning_rate": 0.0001, "loss": 0.0423, "step": 23860 }, { "epoch": 152.03821656050957, "grad_norm": 1.1828609704971313, "learning_rate": 0.0001, "loss": 0.0449, "step": 23870 }, { "epoch": 152.10191082802547, "grad_norm": 1.1335859298706055, "learning_rate": 0.0001, "loss": 0.0429, "step": 23880 }, { "epoch": 152.1656050955414, "grad_norm": 1.326183557510376, "learning_rate": 0.0001, "loss": 0.0448, "step": 23890 }, { "epoch": 152.22929936305732, "grad_norm": 1.1669327020645142, "learning_rate": 0.0001, "loss": 0.0427, "step": 23900 }, { "epoch": 152.29299363057325, "grad_norm": 1.119041085243225, "learning_rate": 0.0001, "loss": 0.0425, "step": 23910 }, { "epoch": 152.35668789808918, "grad_norm": 1.1089587211608887, "learning_rate": 0.0001, "loss": 0.0394, "step": 23920 }, { "epoch": 152.4203821656051, "grad_norm": 1.1945539712905884, "learning_rate": 0.0001, "loss": 0.0459, "step": 23930 }, { "epoch": 152.48407643312103, "grad_norm": 1.7660410404205322, "learning_rate": 0.0001, "loss": 0.0425, "step": 23940 }, { "epoch": 152.54777070063693, "grad_norm": 1.2984970808029175, "learning_rate": 0.0001, "loss": 0.0422, "step": 23950 }, { "epoch": 152.61146496815286, "grad_norm": 1.3853626251220703, "learning_rate": 0.0001, "loss": 0.0466, "step": 23960 }, { "epoch": 152.6751592356688, "grad_norm": 1.1597213745117188, "learning_rate": 0.0001, "loss": 0.0402, "step": 23970 }, { "epoch": 152.7388535031847, "grad_norm": 1.3137294054031372, "learning_rate": 0.0001, "loss": 0.0418, "step": 23980 }, { "epoch": 152.80254777070064, "grad_norm": 1.1779261827468872, "learning_rate": 0.0001, "loss": 0.0406, "step": 23990 }, { "epoch": 152.86624203821657, "grad_norm": 1.049983263015747, "learning_rate": 0.0001, "loss": 0.0421, "step": 24000 }, { "epoch": 152.9299363057325, "grad_norm": 1.3933231830596924, "learning_rate": 0.0001, "loss": 0.0421, "step": 24010 }, { "epoch": 152.9936305732484, "grad_norm": 1.2824467420578003, "learning_rate": 0.0001, "loss": 0.0392, "step": 24020 }, { "epoch": 153.05732484076432, "grad_norm": 0.8816484808921814, "learning_rate": 0.0001, "loss": 0.0409, "step": 24030 }, { "epoch": 153.12101910828025, "grad_norm": 1.223258376121521, "learning_rate": 0.0001, "loss": 0.0409, "step": 24040 }, { "epoch": 153.18471337579618, "grad_norm": 1.313758373260498, "learning_rate": 0.0001, "loss": 0.0389, "step": 24050 }, { "epoch": 153.2484076433121, "grad_norm": 1.4348137378692627, "learning_rate": 0.0001, "loss": 0.0407, "step": 24060 }, { "epoch": 153.31210191082803, "grad_norm": 1.13053297996521, "learning_rate": 0.0001, "loss": 0.041, "step": 24070 }, { "epoch": 153.37579617834396, "grad_norm": 1.004662036895752, "learning_rate": 0.0001, "loss": 0.041, "step": 24080 }, { "epoch": 153.43949044585986, "grad_norm": 1.1743638515472412, "learning_rate": 0.0001, "loss": 0.0417, "step": 24090 }, { "epoch": 153.5031847133758, "grad_norm": 1.3062245845794678, "learning_rate": 0.0001, "loss": 0.0424, "step": 24100 }, { "epoch": 153.56687898089172, "grad_norm": 0.9333073496818542, "learning_rate": 0.0001, "loss": 0.0389, "step": 24110 }, { "epoch": 153.63057324840764, "grad_norm": 1.1173516511917114, "learning_rate": 0.0001, "loss": 0.0404, "step": 24120 }, { "epoch": 153.69426751592357, "grad_norm": 1.3390679359436035, "learning_rate": 0.0001, "loss": 0.0403, "step": 24130 }, { "epoch": 153.7579617834395, "grad_norm": 1.1348854303359985, "learning_rate": 0.0001, "loss": 0.0415, "step": 24140 }, { "epoch": 153.82165605095543, "grad_norm": 0.9991518259048462, "learning_rate": 0.0001, "loss": 0.0412, "step": 24150 }, { "epoch": 153.88535031847132, "grad_norm": 1.4269582033157349, "learning_rate": 0.0001, "loss": 0.0483, "step": 24160 }, { "epoch": 153.94904458598725, "grad_norm": 1.4124714136123657, "learning_rate": 0.0001, "loss": 0.0471, "step": 24170 }, { "epoch": 154.01273885350318, "grad_norm": 1.4674975872039795, "learning_rate": 0.0001, "loss": 0.0451, "step": 24180 }, { "epoch": 154.0764331210191, "grad_norm": 1.454710602760315, "learning_rate": 0.0001, "loss": 0.0425, "step": 24190 }, { "epoch": 154.14012738853503, "grad_norm": 1.1825759410858154, "learning_rate": 0.0001, "loss": 0.0402, "step": 24200 }, { "epoch": 154.20382165605096, "grad_norm": 1.1918121576309204, "learning_rate": 0.0001, "loss": 0.041, "step": 24210 }, { "epoch": 154.2675159235669, "grad_norm": 1.3661644458770752, "learning_rate": 0.0001, "loss": 0.0417, "step": 24220 }, { "epoch": 154.3312101910828, "grad_norm": 1.4895235300064087, "learning_rate": 0.0001, "loss": 0.0411, "step": 24230 }, { "epoch": 154.39490445859872, "grad_norm": 1.279587745666504, "learning_rate": 0.0001, "loss": 0.0401, "step": 24240 }, { "epoch": 154.45859872611464, "grad_norm": 0.9469570517539978, "learning_rate": 0.0001, "loss": 0.0427, "step": 24250 }, { "epoch": 154.52229299363057, "grad_norm": 1.5025107860565186, "learning_rate": 0.0001, "loss": 0.0417, "step": 24260 }, { "epoch": 154.5859872611465, "grad_norm": 1.4646281003952026, "learning_rate": 0.0001, "loss": 0.0497, "step": 24270 }, { "epoch": 154.64968152866243, "grad_norm": 2.050090789794922, "learning_rate": 0.0001, "loss": 0.0435, "step": 24280 }, { "epoch": 154.71337579617835, "grad_norm": 1.6316157579421997, "learning_rate": 0.0001, "loss": 0.0422, "step": 24290 }, { "epoch": 154.77707006369425, "grad_norm": 1.2561861276626587, "learning_rate": 0.0001, "loss": 0.0453, "step": 24300 }, { "epoch": 154.84076433121018, "grad_norm": 1.2246601581573486, "learning_rate": 0.0001, "loss": 0.0435, "step": 24310 }, { "epoch": 154.9044585987261, "grad_norm": 1.1060340404510498, "learning_rate": 0.0001, "loss": 0.0434, "step": 24320 }, { "epoch": 154.96815286624204, "grad_norm": 1.3052364587783813, "learning_rate": 0.0001, "loss": 0.0457, "step": 24330 }, { "epoch": 155.03184713375796, "grad_norm": 1.397574782371521, "learning_rate": 0.0001, "loss": 0.0408, "step": 24340 }, { "epoch": 155.0955414012739, "grad_norm": 1.4378607273101807, "learning_rate": 0.0001, "loss": 0.0435, "step": 24350 }, { "epoch": 155.15923566878982, "grad_norm": 1.641574740409851, "learning_rate": 0.0001, "loss": 0.0389, "step": 24360 }, { "epoch": 155.22292993630575, "grad_norm": 1.6497501134872437, "learning_rate": 0.0001, "loss": 0.0416, "step": 24370 }, { "epoch": 155.28662420382165, "grad_norm": 1.3658382892608643, "learning_rate": 0.0001, "loss": 0.0395, "step": 24380 }, { "epoch": 155.35031847133757, "grad_norm": 1.6177762746810913, "learning_rate": 0.0001, "loss": 0.0398, "step": 24390 }, { "epoch": 155.4140127388535, "grad_norm": 1.643323302268982, "learning_rate": 0.0001, "loss": 0.0415, "step": 24400 }, { "epoch": 155.47770700636943, "grad_norm": 1.308790922164917, "learning_rate": 0.0001, "loss": 0.0419, "step": 24410 }, { "epoch": 155.54140127388536, "grad_norm": 1.4810349941253662, "learning_rate": 0.0001, "loss": 0.0421, "step": 24420 }, { "epoch": 155.60509554140128, "grad_norm": 1.3186718225479126, "learning_rate": 0.0001, "loss": 0.0443, "step": 24430 }, { "epoch": 155.6687898089172, "grad_norm": 1.3009299039840698, "learning_rate": 0.0001, "loss": 0.0367, "step": 24440 }, { "epoch": 155.7324840764331, "grad_norm": 1.1710878610610962, "learning_rate": 0.0001, "loss": 0.0397, "step": 24450 }, { "epoch": 155.79617834394904, "grad_norm": 1.1898648738861084, "learning_rate": 0.0001, "loss": 0.0404, "step": 24460 }, { "epoch": 155.85987261146497, "grad_norm": 1.5690186023712158, "learning_rate": 0.0001, "loss": 0.0417, "step": 24470 }, { "epoch": 155.9235668789809, "grad_norm": 1.4350321292877197, "learning_rate": 0.0001, "loss": 0.0432, "step": 24480 }, { "epoch": 155.98726114649682, "grad_norm": 1.2366423606872559, "learning_rate": 0.0001, "loss": 0.0394, "step": 24490 }, { "epoch": 156.05095541401275, "grad_norm": 1.216667652130127, "learning_rate": 0.0001, "loss": 0.0438, "step": 24500 }, { "epoch": 156.11464968152868, "grad_norm": 1.4550714492797852, "learning_rate": 0.0001, "loss": 0.0412, "step": 24510 }, { "epoch": 156.17834394904457, "grad_norm": 1.4664490222930908, "learning_rate": 0.0001, "loss": 0.0398, "step": 24520 }, { "epoch": 156.2420382165605, "grad_norm": 1.4314957857131958, "learning_rate": 0.0001, "loss": 0.0424, "step": 24530 }, { "epoch": 156.30573248407643, "grad_norm": 1.5013633966445923, "learning_rate": 0.0001, "loss": 0.0422, "step": 24540 }, { "epoch": 156.36942675159236, "grad_norm": 1.4780402183532715, "learning_rate": 0.0001, "loss": 0.0438, "step": 24550 }, { "epoch": 156.43312101910828, "grad_norm": 1.1221524477005005, "learning_rate": 0.0001, "loss": 0.0422, "step": 24560 }, { "epoch": 156.4968152866242, "grad_norm": 1.2680584192276, "learning_rate": 0.0001, "loss": 0.0423, "step": 24570 }, { "epoch": 156.56050955414014, "grad_norm": 1.248377799987793, "learning_rate": 0.0001, "loss": 0.0419, "step": 24580 }, { "epoch": 156.62420382165604, "grad_norm": 1.4844796657562256, "learning_rate": 0.0001, "loss": 0.0418, "step": 24590 }, { "epoch": 156.68789808917197, "grad_norm": 1.172333002090454, "learning_rate": 0.0001, "loss": 0.041, "step": 24600 }, { "epoch": 156.7515923566879, "grad_norm": 1.3098253011703491, "learning_rate": 0.0001, "loss": 0.0427, "step": 24610 }, { "epoch": 156.81528662420382, "grad_norm": 1.3481407165527344, "learning_rate": 0.0001, "loss": 0.0441, "step": 24620 }, { "epoch": 156.87898089171975, "grad_norm": 1.0495541095733643, "learning_rate": 0.0001, "loss": 0.0405, "step": 24630 }, { "epoch": 156.94267515923568, "grad_norm": 1.131656527519226, "learning_rate": 0.0001, "loss": 0.0424, "step": 24640 }, { "epoch": 157.0063694267516, "grad_norm": 1.4537160396575928, "learning_rate": 0.0001, "loss": 0.0447, "step": 24650 }, { "epoch": 157.0700636942675, "grad_norm": 1.3253560066223145, "learning_rate": 0.0001, "loss": 0.0477, "step": 24660 }, { "epoch": 157.13375796178343, "grad_norm": 1.1257190704345703, "learning_rate": 0.0001, "loss": 0.0402, "step": 24670 }, { "epoch": 157.19745222929936, "grad_norm": 1.2114185094833374, "learning_rate": 0.0001, "loss": 0.0458, "step": 24680 }, { "epoch": 157.2611464968153, "grad_norm": 1.4777450561523438, "learning_rate": 0.0001, "loss": 0.0433, "step": 24690 }, { "epoch": 157.3248407643312, "grad_norm": 1.093217134475708, "learning_rate": 0.0001, "loss": 0.0415, "step": 24700 }, { "epoch": 157.38853503184714, "grad_norm": 0.9879863262176514, "learning_rate": 0.0001, "loss": 0.042, "step": 24710 }, { "epoch": 157.45222929936307, "grad_norm": 1.5430659055709839, "learning_rate": 0.0001, "loss": 0.0465, "step": 24720 }, { "epoch": 157.51592356687897, "grad_norm": 1.2089976072311401, "learning_rate": 0.0001, "loss": 0.04, "step": 24730 }, { "epoch": 157.5796178343949, "grad_norm": 1.188377857208252, "learning_rate": 0.0001, "loss": 0.0442, "step": 24740 }, { "epoch": 157.64331210191082, "grad_norm": 1.1461690664291382, "learning_rate": 0.0001, "loss": 0.0451, "step": 24750 }, { "epoch": 157.70700636942675, "grad_norm": 1.2213969230651855, "learning_rate": 0.0001, "loss": 0.0427, "step": 24760 }, { "epoch": 157.77070063694268, "grad_norm": 1.2566920518875122, "learning_rate": 0.0001, "loss": 0.0412, "step": 24770 }, { "epoch": 157.8343949044586, "grad_norm": 1.151202917098999, "learning_rate": 0.0001, "loss": 0.0418, "step": 24780 }, { "epoch": 157.89808917197453, "grad_norm": 1.0909992456436157, "learning_rate": 0.0001, "loss": 0.0458, "step": 24790 }, { "epoch": 157.96178343949043, "grad_norm": 1.2069281339645386, "learning_rate": 0.0001, "loss": 0.0439, "step": 24800 }, { "epoch": 158.02547770700636, "grad_norm": 3.220517873764038, "learning_rate": 0.0001, "loss": 0.0418, "step": 24810 }, { "epoch": 158.0891719745223, "grad_norm": 3.4659812450408936, "learning_rate": 0.0001, "loss": 0.0414, "step": 24820 }, { "epoch": 158.15286624203821, "grad_norm": 2.334094762802124, "learning_rate": 0.0001, "loss": 0.0446, "step": 24830 }, { "epoch": 158.21656050955414, "grad_norm": 1.7982814311981201, "learning_rate": 0.0001, "loss": 0.0412, "step": 24840 }, { "epoch": 158.28025477707007, "grad_norm": 2.1393611431121826, "learning_rate": 0.0001, "loss": 0.0395, "step": 24850 }, { "epoch": 158.343949044586, "grad_norm": 1.875260829925537, "learning_rate": 0.0001, "loss": 0.0384, "step": 24860 }, { "epoch": 158.40764331210192, "grad_norm": 1.8515124320983887, "learning_rate": 0.0001, "loss": 0.0365, "step": 24870 }, { "epoch": 158.47133757961782, "grad_norm": 1.6600474119186401, "learning_rate": 0.0001, "loss": 0.0403, "step": 24880 }, { "epoch": 158.53503184713375, "grad_norm": 2.1158504486083984, "learning_rate": 0.0001, "loss": 0.0406, "step": 24890 }, { "epoch": 158.59872611464968, "grad_norm": 1.6463813781738281, "learning_rate": 0.0001, "loss": 0.0419, "step": 24900 }, { "epoch": 158.6624203821656, "grad_norm": 1.3109993934631348, "learning_rate": 0.0001, "loss": 0.0382, "step": 24910 }, { "epoch": 158.72611464968153, "grad_norm": 1.558479905128479, "learning_rate": 0.0001, "loss": 0.0379, "step": 24920 }, { "epoch": 158.78980891719746, "grad_norm": 1.3792321681976318, "learning_rate": 0.0001, "loss": 0.0397, "step": 24930 }, { "epoch": 158.8535031847134, "grad_norm": 1.3849902153015137, "learning_rate": 0.0001, "loss": 0.0379, "step": 24940 }, { "epoch": 158.9171974522293, "grad_norm": 1.226894497871399, "learning_rate": 0.0001, "loss": 0.0378, "step": 24950 }, { "epoch": 158.98089171974522, "grad_norm": 1.4493616819381714, "learning_rate": 0.0001, "loss": 0.0405, "step": 24960 }, { "epoch": 159.04458598726114, "grad_norm": 1.2942445278167725, "learning_rate": 0.0001, "loss": 0.0424, "step": 24970 }, { "epoch": 159.10828025477707, "grad_norm": 1.20154869556427, "learning_rate": 0.0001, "loss": 0.0398, "step": 24980 }, { "epoch": 159.171974522293, "grad_norm": 1.4153271913528442, "learning_rate": 0.0001, "loss": 0.0426, "step": 24990 }, { "epoch": 159.23566878980893, "grad_norm": 1.3335340023040771, "learning_rate": 0.0001, "loss": 0.0406, "step": 25000 }, { "epoch": 159.29936305732485, "grad_norm": 0.9652509093284607, "learning_rate": 0.0001, "loss": 0.0428, "step": 25010 }, { "epoch": 159.36305732484075, "grad_norm": 1.2719056606292725, "learning_rate": 0.0001, "loss": 0.0383, "step": 25020 }, { "epoch": 159.42675159235668, "grad_norm": 1.4470430612564087, "learning_rate": 0.0001, "loss": 0.041, "step": 25030 }, { "epoch": 159.4904458598726, "grad_norm": 1.2179313898086548, "learning_rate": 0.0001, "loss": 0.0421, "step": 25040 }, { "epoch": 159.55414012738854, "grad_norm": 1.1534720659255981, "learning_rate": 0.0001, "loss": 0.0418, "step": 25050 }, { "epoch": 159.61783439490446, "grad_norm": 1.0786340236663818, "learning_rate": 0.0001, "loss": 0.0377, "step": 25060 }, { "epoch": 159.6815286624204, "grad_norm": 1.3830900192260742, "learning_rate": 0.0001, "loss": 0.0408, "step": 25070 }, { "epoch": 159.74522292993632, "grad_norm": 1.2366207838058472, "learning_rate": 0.0001, "loss": 0.0415, "step": 25080 }, { "epoch": 159.80891719745222, "grad_norm": 1.695268154144287, "learning_rate": 0.0001, "loss": 0.0409, "step": 25090 }, { "epoch": 159.87261146496814, "grad_norm": 1.671838402748108, "learning_rate": 0.0001, "loss": 0.0436, "step": 25100 }, { "epoch": 159.93630573248407, "grad_norm": 2.0509116649627686, "learning_rate": 0.0001, "loss": 0.0386, "step": 25110 }, { "epoch": 160.0, "grad_norm": 2.416745185852051, "learning_rate": 0.0001, "loss": 0.0393, "step": 25120 }, { "epoch": 160.06369426751593, "grad_norm": 1.6840945482254028, "learning_rate": 0.0001, "loss": 0.04, "step": 25130 }, { "epoch": 160.12738853503186, "grad_norm": 1.7543840408325195, "learning_rate": 0.0001, "loss": 0.0406, "step": 25140 }, { "epoch": 160.19108280254778, "grad_norm": 1.9311878681182861, "learning_rate": 0.0001, "loss": 0.0423, "step": 25150 }, { "epoch": 160.25477707006368, "grad_norm": 1.705567717552185, "learning_rate": 0.0001, "loss": 0.041, "step": 25160 }, { "epoch": 160.3184713375796, "grad_norm": 1.6252483129501343, "learning_rate": 0.0001, "loss": 0.0376, "step": 25170 }, { "epoch": 160.38216560509554, "grad_norm": 1.9148813486099243, "learning_rate": 0.0001, "loss": 0.0406, "step": 25180 }, { "epoch": 160.44585987261146, "grad_norm": 1.65464186668396, "learning_rate": 0.0001, "loss": 0.0384, "step": 25190 }, { "epoch": 160.5095541401274, "grad_norm": 1.5073736906051636, "learning_rate": 0.0001, "loss": 0.0388, "step": 25200 }, { "epoch": 160.57324840764332, "grad_norm": 1.5822343826293945, "learning_rate": 0.0001, "loss": 0.0388, "step": 25210 }, { "epoch": 160.63694267515925, "grad_norm": 1.2700468301773071, "learning_rate": 0.0001, "loss": 0.0358, "step": 25220 }, { "epoch": 160.70063694267515, "grad_norm": 1.4719443321228027, "learning_rate": 0.0001, "loss": 0.0381, "step": 25230 }, { "epoch": 160.76433121019107, "grad_norm": 1.4049303531646729, "learning_rate": 0.0001, "loss": 0.0376, "step": 25240 }, { "epoch": 160.828025477707, "grad_norm": 1.4349926710128784, "learning_rate": 0.0001, "loss": 0.0416, "step": 25250 }, { "epoch": 160.89171974522293, "grad_norm": 1.4628790616989136, "learning_rate": 0.0001, "loss": 0.0412, "step": 25260 }, { "epoch": 160.95541401273886, "grad_norm": 1.4135618209838867, "learning_rate": 0.0001, "loss": 0.0372, "step": 25270 }, { "epoch": 161.01910828025478, "grad_norm": 1.5038278102874756, "learning_rate": 0.0001, "loss": 0.0423, "step": 25280 }, { "epoch": 161.0828025477707, "grad_norm": 1.1580928564071655, "learning_rate": 0.0001, "loss": 0.0439, "step": 25290 }, { "epoch": 161.1464968152866, "grad_norm": 1.2801655530929565, "learning_rate": 0.0001, "loss": 0.0378, "step": 25300 }, { "epoch": 161.21019108280254, "grad_norm": 1.4221186637878418, "learning_rate": 0.0001, "loss": 0.0427, "step": 25310 }, { "epoch": 161.27388535031847, "grad_norm": 1.1147404909133911, "learning_rate": 0.0001, "loss": 0.0444, "step": 25320 }, { "epoch": 161.3375796178344, "grad_norm": 1.631528615951538, "learning_rate": 0.0001, "loss": 0.043, "step": 25330 }, { "epoch": 161.40127388535032, "grad_norm": 2.39990234375, "learning_rate": 0.0001, "loss": 0.0437, "step": 25340 }, { "epoch": 161.46496815286625, "grad_norm": 1.8168717622756958, "learning_rate": 0.0001, "loss": 0.0431, "step": 25350 }, { "epoch": 161.52866242038218, "grad_norm": 1.561240792274475, "learning_rate": 0.0001, "loss": 0.0425, "step": 25360 }, { "epoch": 161.59235668789808, "grad_norm": 1.5665404796600342, "learning_rate": 0.0001, "loss": 0.0434, "step": 25370 }, { "epoch": 161.656050955414, "grad_norm": 1.9531586170196533, "learning_rate": 0.0001, "loss": 0.0448, "step": 25380 }, { "epoch": 161.71974522292993, "grad_norm": 1.5980523824691772, "learning_rate": 0.0001, "loss": 0.0396, "step": 25390 }, { "epoch": 161.78343949044586, "grad_norm": 1.6106913089752197, "learning_rate": 0.0001, "loss": 0.04, "step": 25400 }, { "epoch": 161.84713375796179, "grad_norm": 1.5993455648422241, "learning_rate": 0.0001, "loss": 0.0436, "step": 25410 }, { "epoch": 161.9108280254777, "grad_norm": 1.3508728742599487, "learning_rate": 0.0001, "loss": 0.0423, "step": 25420 }, { "epoch": 161.97452229299364, "grad_norm": 1.7286243438720703, "learning_rate": 0.0001, "loss": 0.0383, "step": 25430 }, { "epoch": 162.03821656050957, "grad_norm": 1.2496119737625122, "learning_rate": 0.0001, "loss": 0.043, "step": 25440 }, { "epoch": 162.10191082802547, "grad_norm": 1.3181946277618408, "learning_rate": 0.0001, "loss": 0.0384, "step": 25450 }, { "epoch": 162.1656050955414, "grad_norm": 1.163684606552124, "learning_rate": 0.0001, "loss": 0.0394, "step": 25460 }, { "epoch": 162.22929936305732, "grad_norm": 1.1693415641784668, "learning_rate": 0.0001, "loss": 0.0403, "step": 25470 }, { "epoch": 162.29299363057325, "grad_norm": 1.4698346853256226, "learning_rate": 0.0001, "loss": 0.0388, "step": 25480 }, { "epoch": 162.35668789808918, "grad_norm": 1.515838861465454, "learning_rate": 0.0001, "loss": 0.0422, "step": 25490 }, { "epoch": 162.4203821656051, "grad_norm": 1.3131831884384155, "learning_rate": 0.0001, "loss": 0.0394, "step": 25500 }, { "epoch": 162.48407643312103, "grad_norm": 1.274844765663147, "learning_rate": 0.0001, "loss": 0.0441, "step": 25510 }, { "epoch": 162.54777070063693, "grad_norm": 1.3511936664581299, "learning_rate": 0.0001, "loss": 0.0388, "step": 25520 }, { "epoch": 162.61146496815286, "grad_norm": 1.1314244270324707, "learning_rate": 0.0001, "loss": 0.0403, "step": 25530 }, { "epoch": 162.6751592356688, "grad_norm": 1.2114946842193604, "learning_rate": 0.0001, "loss": 0.0378, "step": 25540 }, { "epoch": 162.7388535031847, "grad_norm": 1.2530568838119507, "learning_rate": 0.0001, "loss": 0.0397, "step": 25550 }, { "epoch": 162.80254777070064, "grad_norm": 1.2873040437698364, "learning_rate": 0.0001, "loss": 0.0428, "step": 25560 }, { "epoch": 162.86624203821657, "grad_norm": 1.6302363872528076, "learning_rate": 0.0001, "loss": 0.0436, "step": 25570 }, { "epoch": 162.9299363057325, "grad_norm": 1.1816763877868652, "learning_rate": 0.0001, "loss": 0.0378, "step": 25580 }, { "epoch": 162.9936305732484, "grad_norm": 1.347872018814087, "learning_rate": 0.0001, "loss": 0.0414, "step": 25590 }, { "epoch": 163.05732484076432, "grad_norm": 2.0239837169647217, "learning_rate": 0.0001, "loss": 0.0413, "step": 25600 }, { "epoch": 163.12101910828025, "grad_norm": 2.0131728649139404, "learning_rate": 0.0001, "loss": 0.04, "step": 25610 }, { "epoch": 163.18471337579618, "grad_norm": 1.5618644952774048, "learning_rate": 0.0001, "loss": 0.0388, "step": 25620 }, { "epoch": 163.2484076433121, "grad_norm": 1.837222695350647, "learning_rate": 0.0001, "loss": 0.0399, "step": 25630 }, { "epoch": 163.31210191082803, "grad_norm": 1.2818734645843506, "learning_rate": 0.0001, "loss": 0.0392, "step": 25640 }, { "epoch": 163.37579617834396, "grad_norm": 1.55012845993042, "learning_rate": 0.0001, "loss": 0.0373, "step": 25650 }, { "epoch": 163.43949044585986, "grad_norm": 1.492782473564148, "learning_rate": 0.0001, "loss": 0.0385, "step": 25660 }, { "epoch": 163.5031847133758, "grad_norm": 1.1240006685256958, "learning_rate": 0.0001, "loss": 0.0414, "step": 25670 }, { "epoch": 163.56687898089172, "grad_norm": 1.3381246328353882, "learning_rate": 0.0001, "loss": 0.0426, "step": 25680 }, { "epoch": 163.63057324840764, "grad_norm": 1.2075941562652588, "learning_rate": 0.0001, "loss": 0.0392, "step": 25690 }, { "epoch": 163.69426751592357, "grad_norm": 1.3267219066619873, "learning_rate": 0.0001, "loss": 0.042, "step": 25700 }, { "epoch": 163.7579617834395, "grad_norm": 1.4787123203277588, "learning_rate": 0.0001, "loss": 0.0413, "step": 25710 }, { "epoch": 163.82165605095543, "grad_norm": 1.3090670108795166, "learning_rate": 0.0001, "loss": 0.04, "step": 25720 }, { "epoch": 163.88535031847132, "grad_norm": 1.352474331855774, "learning_rate": 0.0001, "loss": 0.039, "step": 25730 }, { "epoch": 163.94904458598725, "grad_norm": 1.129274606704712, "learning_rate": 0.0001, "loss": 0.0371, "step": 25740 }, { "epoch": 164.01273885350318, "grad_norm": 1.4053664207458496, "learning_rate": 0.0001, "loss": 0.0399, "step": 25750 }, { "epoch": 164.0764331210191, "grad_norm": 1.3793877363204956, "learning_rate": 0.0001, "loss": 0.0407, "step": 25760 }, { "epoch": 164.14012738853503, "grad_norm": 1.2400795221328735, "learning_rate": 0.0001, "loss": 0.0423, "step": 25770 }, { "epoch": 164.20382165605096, "grad_norm": 1.197780966758728, "learning_rate": 0.0001, "loss": 0.0457, "step": 25780 }, { "epoch": 164.2675159235669, "grad_norm": 1.1878834962844849, "learning_rate": 0.0001, "loss": 0.0437, "step": 25790 }, { "epoch": 164.3312101910828, "grad_norm": 2.572580575942993, "learning_rate": 0.0001, "loss": 0.061, "step": 25800 }, { "epoch": 164.39490445859872, "grad_norm": 2.0653934478759766, "learning_rate": 0.0001, "loss": 0.0455, "step": 25810 }, { "epoch": 164.45859872611464, "grad_norm": 1.7009754180908203, "learning_rate": 0.0001, "loss": 0.0412, "step": 25820 }, { "epoch": 164.52229299363057, "grad_norm": 2.2358460426330566, "learning_rate": 0.0001, "loss": 0.0418, "step": 25830 }, { "epoch": 164.5859872611465, "grad_norm": 2.515097141265869, "learning_rate": 0.0001, "loss": 0.0409, "step": 25840 }, { "epoch": 164.64968152866243, "grad_norm": 2.1716525554656982, "learning_rate": 0.0001, "loss": 0.0403, "step": 25850 }, { "epoch": 164.71337579617835, "grad_norm": 2.1485376358032227, "learning_rate": 0.0001, "loss": 0.0413, "step": 25860 }, { "epoch": 164.77707006369425, "grad_norm": 3.887179136276245, "learning_rate": 0.0001, "loss": 0.0391, "step": 25870 }, { "epoch": 164.84076433121018, "grad_norm": 2.490983724594116, "learning_rate": 0.0001, "loss": 0.0385, "step": 25880 }, { "epoch": 164.9044585987261, "grad_norm": 1.62213933467865, "learning_rate": 0.0001, "loss": 0.0372, "step": 25890 }, { "epoch": 164.96815286624204, "grad_norm": 1.899932861328125, "learning_rate": 0.0001, "loss": 0.0358, "step": 25900 }, { "epoch": 165.03184713375796, "grad_norm": 1.8176660537719727, "learning_rate": 0.0001, "loss": 0.0376, "step": 25910 }, { "epoch": 165.0955414012739, "grad_norm": 1.4636518955230713, "learning_rate": 0.0001, "loss": 0.0378, "step": 25920 }, { "epoch": 165.15923566878982, "grad_norm": 1.503072738647461, "learning_rate": 0.0001, "loss": 0.0358, "step": 25930 }, { "epoch": 165.22292993630575, "grad_norm": 1.3665357828140259, "learning_rate": 0.0001, "loss": 0.0376, "step": 25940 }, { "epoch": 165.28662420382165, "grad_norm": 1.1499816179275513, "learning_rate": 0.0001, "loss": 0.0343, "step": 25950 }, { "epoch": 165.35031847133757, "grad_norm": 1.51056969165802, "learning_rate": 0.0001, "loss": 0.0378, "step": 25960 }, { "epoch": 165.4140127388535, "grad_norm": 1.5637305974960327, "learning_rate": 0.0001, "loss": 0.0369, "step": 25970 }, { "epoch": 165.47770700636943, "grad_norm": 2.218721628189087, "learning_rate": 0.0001, "loss": 0.0364, "step": 25980 }, { "epoch": 165.54140127388536, "grad_norm": 1.4360579252243042, "learning_rate": 0.0001, "loss": 0.0366, "step": 25990 }, { "epoch": 165.60509554140128, "grad_norm": 1.0983659029006958, "learning_rate": 0.0001, "loss": 0.0385, "step": 26000 }, { "epoch": 165.6687898089172, "grad_norm": 1.4479960203170776, "learning_rate": 0.0001, "loss": 0.0452, "step": 26010 }, { "epoch": 165.7324840764331, "grad_norm": 2.470059871673584, "learning_rate": 0.0001, "loss": 0.0404, "step": 26020 }, { "epoch": 165.79617834394904, "grad_norm": 1.6594293117523193, "learning_rate": 0.0001, "loss": 0.0434, "step": 26030 }, { "epoch": 165.85987261146497, "grad_norm": 1.5401031970977783, "learning_rate": 0.0001, "loss": 0.0385, "step": 26040 }, { "epoch": 165.9235668789809, "grad_norm": 1.7732815742492676, "learning_rate": 0.0001, "loss": 0.038, "step": 26050 }, { "epoch": 165.98726114649682, "grad_norm": 1.3656411170959473, "learning_rate": 0.0001, "loss": 0.0405, "step": 26060 }, { "epoch": 166.05095541401275, "grad_norm": 1.7433961629867554, "learning_rate": 0.0001, "loss": 0.0399, "step": 26070 }, { "epoch": 166.11464968152868, "grad_norm": 1.5516161918640137, "learning_rate": 0.0001, "loss": 0.0365, "step": 26080 }, { "epoch": 166.17834394904457, "grad_norm": 1.540628433227539, "learning_rate": 0.0001, "loss": 0.0378, "step": 26090 }, { "epoch": 166.2420382165605, "grad_norm": 1.5442771911621094, "learning_rate": 0.0001, "loss": 0.0387, "step": 26100 }, { "epoch": 166.30573248407643, "grad_norm": 1.8679898977279663, "learning_rate": 0.0001, "loss": 0.0414, "step": 26110 }, { "epoch": 166.36942675159236, "grad_norm": 1.235515832901001, "learning_rate": 0.0001, "loss": 0.0396, "step": 26120 }, { "epoch": 166.43312101910828, "grad_norm": 1.3286247253417969, "learning_rate": 0.0001, "loss": 0.0408, "step": 26130 }, { "epoch": 166.4968152866242, "grad_norm": 1.7231615781784058, "learning_rate": 0.0001, "loss": 0.0411, "step": 26140 }, { "epoch": 166.56050955414014, "grad_norm": 1.103092908859253, "learning_rate": 0.0001, "loss": 0.0392, "step": 26150 }, { "epoch": 166.62420382165604, "grad_norm": 1.3253971338272095, "learning_rate": 0.0001, "loss": 0.0401, "step": 26160 }, { "epoch": 166.68789808917197, "grad_norm": 1.2611347436904907, "learning_rate": 0.0001, "loss": 0.0388, "step": 26170 }, { "epoch": 166.7515923566879, "grad_norm": 1.4998639822006226, "learning_rate": 0.0001, "loss": 0.0407, "step": 26180 }, { "epoch": 166.81528662420382, "grad_norm": 1.21586012840271, "learning_rate": 0.0001, "loss": 0.0421, "step": 26190 }, { "epoch": 166.87898089171975, "grad_norm": 1.4635716676712036, "learning_rate": 0.0001, "loss": 0.0399, "step": 26200 }, { "epoch": 166.94267515923568, "grad_norm": 1.2546398639678955, "learning_rate": 0.0001, "loss": 0.0384, "step": 26210 }, { "epoch": 167.0063694267516, "grad_norm": 1.315240502357483, "learning_rate": 0.0001, "loss": 0.0387, "step": 26220 }, { "epoch": 167.0700636942675, "grad_norm": 1.0332545042037964, "learning_rate": 0.0001, "loss": 0.0395, "step": 26230 }, { "epoch": 167.13375796178343, "grad_norm": 1.1266324520111084, "learning_rate": 0.0001, "loss": 0.0369, "step": 26240 }, { "epoch": 167.19745222929936, "grad_norm": 1.2887262105941772, "learning_rate": 0.0001, "loss": 0.0377, "step": 26250 }, { "epoch": 167.2611464968153, "grad_norm": 1.148904800415039, "learning_rate": 0.0001, "loss": 0.0436, "step": 26260 }, { "epoch": 167.3248407643312, "grad_norm": 1.4013324975967407, "learning_rate": 0.0001, "loss": 0.0419, "step": 26270 }, { "epoch": 167.38853503184714, "grad_norm": 1.286721110343933, "learning_rate": 0.0001, "loss": 0.0407, "step": 26280 }, { "epoch": 167.45222929936307, "grad_norm": 1.1419994831085205, "learning_rate": 0.0001, "loss": 0.0385, "step": 26290 }, { "epoch": 167.51592356687897, "grad_norm": 1.3651407957077026, "learning_rate": 0.0001, "loss": 0.0434, "step": 26300 }, { "epoch": 167.5796178343949, "grad_norm": 1.4151825904846191, "learning_rate": 0.0001, "loss": 0.0428, "step": 26310 }, { "epoch": 167.64331210191082, "grad_norm": 1.2939258813858032, "learning_rate": 0.0001, "loss": 0.0452, "step": 26320 }, { "epoch": 167.70700636942675, "grad_norm": 1.06137216091156, "learning_rate": 0.0001, "loss": 0.0411, "step": 26330 }, { "epoch": 167.77070063694268, "grad_norm": 1.4554097652435303, "learning_rate": 0.0001, "loss": 0.0417, "step": 26340 }, { "epoch": 167.8343949044586, "grad_norm": 1.435455083847046, "learning_rate": 0.0001, "loss": 0.0409, "step": 26350 }, { "epoch": 167.89808917197453, "grad_norm": 1.4666752815246582, "learning_rate": 0.0001, "loss": 0.0413, "step": 26360 }, { "epoch": 167.96178343949043, "grad_norm": 0.944307804107666, "learning_rate": 0.0001, "loss": 0.0381, "step": 26370 }, { "epoch": 168.02547770700636, "grad_norm": 1.318549394607544, "learning_rate": 0.0001, "loss": 0.0404, "step": 26380 }, { "epoch": 168.0891719745223, "grad_norm": 1.1841094493865967, "learning_rate": 0.0001, "loss": 0.0402, "step": 26390 }, { "epoch": 168.15286624203821, "grad_norm": 1.3456629514694214, "learning_rate": 0.0001, "loss": 0.0379, "step": 26400 }, { "epoch": 168.21656050955414, "grad_norm": 1.1305735111236572, "learning_rate": 0.0001, "loss": 0.0413, "step": 26410 }, { "epoch": 168.28025477707007, "grad_norm": 1.3521744012832642, "learning_rate": 0.0001, "loss": 0.0393, "step": 26420 }, { "epoch": 168.343949044586, "grad_norm": 1.581177830696106, "learning_rate": 0.0001, "loss": 0.0391, "step": 26430 }, { "epoch": 168.40764331210192, "grad_norm": 1.1022067070007324, "learning_rate": 0.0001, "loss": 0.0422, "step": 26440 }, { "epoch": 168.47133757961782, "grad_norm": 1.2478073835372925, "learning_rate": 0.0001, "loss": 0.0395, "step": 26450 }, { "epoch": 168.53503184713375, "grad_norm": 1.124142050743103, "learning_rate": 0.0001, "loss": 0.0398, "step": 26460 }, { "epoch": 168.59872611464968, "grad_norm": 1.042655110359192, "learning_rate": 0.0001, "loss": 0.0408, "step": 26470 }, { "epoch": 168.6624203821656, "grad_norm": 1.299103021621704, "learning_rate": 0.0001, "loss": 0.0412, "step": 26480 }, { "epoch": 168.72611464968153, "grad_norm": 1.5161244869232178, "learning_rate": 0.0001, "loss": 0.0397, "step": 26490 }, { "epoch": 168.78980891719746, "grad_norm": 1.0887621641159058, "learning_rate": 0.0001, "loss": 0.0424, "step": 26500 }, { "epoch": 168.8535031847134, "grad_norm": 0.9225420951843262, "learning_rate": 0.0001, "loss": 0.0387, "step": 26510 }, { "epoch": 168.9171974522293, "grad_norm": 1.0004398822784424, "learning_rate": 0.0001, "loss": 0.042, "step": 26520 }, { "epoch": 168.98089171974522, "grad_norm": 1.0282377004623413, "learning_rate": 0.0001, "loss": 0.0441, "step": 26530 }, { "epoch": 169.04458598726114, "grad_norm": 1.201341152191162, "learning_rate": 0.0001, "loss": 0.0414, "step": 26540 }, { "epoch": 169.10828025477707, "grad_norm": 1.3775709867477417, "learning_rate": 0.0001, "loss": 0.0444, "step": 26550 }, { "epoch": 169.171974522293, "grad_norm": 1.3140430450439453, "learning_rate": 0.0001, "loss": 0.0416, "step": 26560 }, { "epoch": 169.23566878980893, "grad_norm": 1.3119462728500366, "learning_rate": 0.0001, "loss": 0.0412, "step": 26570 }, { "epoch": 169.29936305732485, "grad_norm": 0.9658041000366211, "learning_rate": 0.0001, "loss": 0.0422, "step": 26580 }, { "epoch": 169.36305732484075, "grad_norm": 1.446026086807251, "learning_rate": 0.0001, "loss": 0.0412, "step": 26590 }, { "epoch": 169.42675159235668, "grad_norm": 1.4173517227172852, "learning_rate": 0.0001, "loss": 0.0391, "step": 26600 }, { "epoch": 169.4904458598726, "grad_norm": 1.0266181230545044, "learning_rate": 0.0001, "loss": 0.0422, "step": 26610 }, { "epoch": 169.55414012738854, "grad_norm": 1.3256518840789795, "learning_rate": 0.0001, "loss": 0.042, "step": 26620 }, { "epoch": 169.61783439490446, "grad_norm": 1.184558629989624, "learning_rate": 0.0001, "loss": 0.0405, "step": 26630 }, { "epoch": 169.6815286624204, "grad_norm": 1.186049461364746, "learning_rate": 0.0001, "loss": 0.0405, "step": 26640 }, { "epoch": 169.74522292993632, "grad_norm": 1.2143104076385498, "learning_rate": 0.0001, "loss": 0.0402, "step": 26650 }, { "epoch": 169.80891719745222, "grad_norm": 1.1673262119293213, "learning_rate": 0.0001, "loss": 0.0445, "step": 26660 }, { "epoch": 169.87261146496814, "grad_norm": 1.3193249702453613, "learning_rate": 0.0001, "loss": 0.0397, "step": 26670 }, { "epoch": 169.93630573248407, "grad_norm": 1.1279054880142212, "learning_rate": 0.0001, "loss": 0.0438, "step": 26680 }, { "epoch": 170.0, "grad_norm": 1.3315712213516235, "learning_rate": 0.0001, "loss": 0.0403, "step": 26690 }, { "epoch": 170.06369426751593, "grad_norm": 1.2509377002716064, "learning_rate": 0.0001, "loss": 0.0398, "step": 26700 }, { "epoch": 170.12738853503186, "grad_norm": 1.2707734107971191, "learning_rate": 0.0001, "loss": 0.0443, "step": 26710 }, { "epoch": 170.19108280254778, "grad_norm": 1.286431074142456, "learning_rate": 0.0001, "loss": 0.0438, "step": 26720 }, { "epoch": 170.25477707006368, "grad_norm": 1.4239487648010254, "learning_rate": 0.0001, "loss": 0.0449, "step": 26730 }, { "epoch": 170.3184713375796, "grad_norm": 1.2424075603485107, "learning_rate": 0.0001, "loss": 0.0416, "step": 26740 }, { "epoch": 170.38216560509554, "grad_norm": 1.098090410232544, "learning_rate": 0.0001, "loss": 0.0426, "step": 26750 }, { "epoch": 170.44585987261146, "grad_norm": 1.2069289684295654, "learning_rate": 0.0001, "loss": 0.0371, "step": 26760 }, { "epoch": 170.5095541401274, "grad_norm": 1.150944709777832, "learning_rate": 0.0001, "loss": 0.0434, "step": 26770 }, { "epoch": 170.57324840764332, "grad_norm": 0.985156774520874, "learning_rate": 0.0001, "loss": 0.0415, "step": 26780 }, { "epoch": 170.63694267515925, "grad_norm": 1.098341464996338, "learning_rate": 0.0001, "loss": 0.0416, "step": 26790 }, { "epoch": 170.70063694267515, "grad_norm": 1.3370184898376465, "learning_rate": 0.0001, "loss": 0.044, "step": 26800 }, { "epoch": 170.76433121019107, "grad_norm": 1.284399390220642, "learning_rate": 0.0001, "loss": 0.0469, "step": 26810 }, { "epoch": 170.828025477707, "grad_norm": 1.1947078704833984, "learning_rate": 0.0001, "loss": 0.0476, "step": 26820 }, { "epoch": 170.89171974522293, "grad_norm": 1.2138065099716187, "learning_rate": 0.0001, "loss": 0.0467, "step": 26830 }, { "epoch": 170.95541401273886, "grad_norm": 1.1772713661193848, "learning_rate": 0.0001, "loss": 0.0456, "step": 26840 }, { "epoch": 171.01910828025478, "grad_norm": 1.038590669631958, "learning_rate": 0.0001, "loss": 0.0459, "step": 26850 }, { "epoch": 171.0828025477707, "grad_norm": 1.2464863061904907, "learning_rate": 0.0001, "loss": 0.0482, "step": 26860 }, { "epoch": 171.1464968152866, "grad_norm": 1.1012500524520874, "learning_rate": 0.0001, "loss": 0.0496, "step": 26870 }, { "epoch": 171.21019108280254, "grad_norm": 1.109938144683838, "learning_rate": 0.0001, "loss": 0.0447, "step": 26880 }, { "epoch": 171.27388535031847, "grad_norm": 1.2287417650222778, "learning_rate": 0.0001, "loss": 0.0444, "step": 26890 }, { "epoch": 171.3375796178344, "grad_norm": 0.8876630067825317, "learning_rate": 0.0001, "loss": 0.043, "step": 26900 }, { "epoch": 171.40127388535032, "grad_norm": 1.509400486946106, "learning_rate": 0.0001, "loss": 0.0456, "step": 26910 }, { "epoch": 171.46496815286625, "grad_norm": 1.1540380716323853, "learning_rate": 0.0001, "loss": 0.0438, "step": 26920 }, { "epoch": 171.52866242038218, "grad_norm": 1.4533026218414307, "learning_rate": 0.0001, "loss": 0.0455, "step": 26930 }, { "epoch": 171.59235668789808, "grad_norm": 1.1866062879562378, "learning_rate": 0.0001, "loss": 0.0441, "step": 26940 }, { "epoch": 171.656050955414, "grad_norm": 1.1270403861999512, "learning_rate": 0.0001, "loss": 0.0429, "step": 26950 }, { "epoch": 171.71974522292993, "grad_norm": 1.1631407737731934, "learning_rate": 0.0001, "loss": 0.0445, "step": 26960 }, { "epoch": 171.78343949044586, "grad_norm": 1.351057529449463, "learning_rate": 0.0001, "loss": 0.0424, "step": 26970 }, { "epoch": 171.84713375796179, "grad_norm": 1.1602582931518555, "learning_rate": 0.0001, "loss": 0.0438, "step": 26980 }, { "epoch": 171.9108280254777, "grad_norm": 1.238795518875122, "learning_rate": 0.0001, "loss": 0.0395, "step": 26990 }, { "epoch": 171.97452229299364, "grad_norm": 1.0594886541366577, "learning_rate": 0.0001, "loss": 0.0423, "step": 27000 }, { "epoch": 172.03821656050957, "grad_norm": 1.2388973236083984, "learning_rate": 0.0001, "loss": 0.0416, "step": 27010 }, { "epoch": 172.10191082802547, "grad_norm": 1.3692820072174072, "learning_rate": 0.0001, "loss": 0.0444, "step": 27020 }, { "epoch": 172.1656050955414, "grad_norm": 1.184633731842041, "learning_rate": 0.0001, "loss": 0.0448, "step": 27030 }, { "epoch": 172.22929936305732, "grad_norm": 1.0745805501937866, "learning_rate": 0.0001, "loss": 0.047, "step": 27040 }, { "epoch": 172.29299363057325, "grad_norm": 0.9755812287330627, "learning_rate": 0.0001, "loss": 0.0425, "step": 27050 }, { "epoch": 172.35668789808918, "grad_norm": 1.6168073415756226, "learning_rate": 0.0001, "loss": 0.0428, "step": 27060 }, { "epoch": 172.4203821656051, "grad_norm": 1.0660924911499023, "learning_rate": 0.0001, "loss": 0.0423, "step": 27070 }, { "epoch": 172.48407643312103, "grad_norm": 1.3322921991348267, "learning_rate": 0.0001, "loss": 0.0414, "step": 27080 }, { "epoch": 172.54777070063693, "grad_norm": 1.4371017217636108, "learning_rate": 0.0001, "loss": 0.0418, "step": 27090 }, { "epoch": 172.61146496815286, "grad_norm": 1.0973165035247803, "learning_rate": 0.0001, "loss": 0.0452, "step": 27100 }, { "epoch": 172.6751592356688, "grad_norm": 0.7574190497398376, "learning_rate": 0.0001, "loss": 0.0405, "step": 27110 }, { "epoch": 172.7388535031847, "grad_norm": 1.2838088274002075, "learning_rate": 0.0001, "loss": 0.0419, "step": 27120 }, { "epoch": 172.80254777070064, "grad_norm": 0.92631995677948, "learning_rate": 0.0001, "loss": 0.0427, "step": 27130 }, { "epoch": 172.86624203821657, "grad_norm": 1.3462578058242798, "learning_rate": 0.0001, "loss": 0.043, "step": 27140 }, { "epoch": 172.9299363057325, "grad_norm": 1.3867928981781006, "learning_rate": 0.0001, "loss": 0.0449, "step": 27150 }, { "epoch": 172.9936305732484, "grad_norm": 1.0245195627212524, "learning_rate": 0.0001, "loss": 0.0438, "step": 27160 }, { "epoch": 173.05732484076432, "grad_norm": 1.115623116493225, "learning_rate": 0.0001, "loss": 0.0421, "step": 27170 }, { "epoch": 173.12101910828025, "grad_norm": 1.0020779371261597, "learning_rate": 0.0001, "loss": 0.0443, "step": 27180 }, { "epoch": 173.18471337579618, "grad_norm": 1.1798030138015747, "learning_rate": 0.0001, "loss": 0.0415, "step": 27190 }, { "epoch": 173.2484076433121, "grad_norm": 1.0928598642349243, "learning_rate": 0.0001, "loss": 0.0442, "step": 27200 }, { "epoch": 173.31210191082803, "grad_norm": 1.085828423500061, "learning_rate": 0.0001, "loss": 0.0426, "step": 27210 }, { "epoch": 173.37579617834396, "grad_norm": 1.281373381614685, "learning_rate": 0.0001, "loss": 0.0428, "step": 27220 }, { "epoch": 173.43949044585986, "grad_norm": 1.1029655933380127, "learning_rate": 0.0001, "loss": 0.0421, "step": 27230 }, { "epoch": 173.5031847133758, "grad_norm": 1.2002276182174683, "learning_rate": 0.0001, "loss": 0.0415, "step": 27240 }, { "epoch": 173.56687898089172, "grad_norm": 1.0270057916641235, "learning_rate": 0.0001, "loss": 0.0451, "step": 27250 }, { "epoch": 173.63057324840764, "grad_norm": 1.2861648797988892, "learning_rate": 0.0001, "loss": 0.0463, "step": 27260 }, { "epoch": 173.69426751592357, "grad_norm": 1.2846711874008179, "learning_rate": 0.0001, "loss": 0.0405, "step": 27270 }, { "epoch": 173.7579617834395, "grad_norm": 1.0945512056350708, "learning_rate": 0.0001, "loss": 0.0382, "step": 27280 }, { "epoch": 173.82165605095543, "grad_norm": 1.305539846420288, "learning_rate": 0.0001, "loss": 0.0451, "step": 27290 }, { "epoch": 173.88535031847132, "grad_norm": 1.207916498184204, "learning_rate": 0.0001, "loss": 0.0399, "step": 27300 }, { "epoch": 173.94904458598725, "grad_norm": 1.3871763944625854, "learning_rate": 0.0001, "loss": 0.0408, "step": 27310 }, { "epoch": 174.01273885350318, "grad_norm": 1.3509833812713623, "learning_rate": 0.0001, "loss": 0.0421, "step": 27320 }, { "epoch": 174.0764331210191, "grad_norm": 1.4691298007965088, "learning_rate": 0.0001, "loss": 0.0399, "step": 27330 }, { "epoch": 174.14012738853503, "grad_norm": 1.3073296546936035, "learning_rate": 0.0001, "loss": 0.0388, "step": 27340 }, { "epoch": 174.20382165605096, "grad_norm": 1.2725536823272705, "learning_rate": 0.0001, "loss": 0.0387, "step": 27350 }, { "epoch": 174.2675159235669, "grad_norm": 0.9117061495780945, "learning_rate": 0.0001, "loss": 0.0423, "step": 27360 }, { "epoch": 174.3312101910828, "grad_norm": 1.1556227207183838, "learning_rate": 0.0001, "loss": 0.0354, "step": 27370 }, { "epoch": 174.39490445859872, "grad_norm": 1.0132596492767334, "learning_rate": 0.0001, "loss": 0.0394, "step": 27380 }, { "epoch": 174.45859872611464, "grad_norm": 1.0453369617462158, "learning_rate": 0.0001, "loss": 0.038, "step": 27390 }, { "epoch": 174.52229299363057, "grad_norm": 1.2817437648773193, "learning_rate": 0.0001, "loss": 0.04, "step": 27400 }, { "epoch": 174.5859872611465, "grad_norm": 0.9832063913345337, "learning_rate": 0.0001, "loss": 0.0419, "step": 27410 }, { "epoch": 174.64968152866243, "grad_norm": 1.4307115077972412, "learning_rate": 0.0001, "loss": 0.0398, "step": 27420 }, { "epoch": 174.71337579617835, "grad_norm": 1.2359397411346436, "learning_rate": 0.0001, "loss": 0.0413, "step": 27430 }, { "epoch": 174.77707006369425, "grad_norm": 1.0873208045959473, "learning_rate": 0.0001, "loss": 0.0409, "step": 27440 }, { "epoch": 174.84076433121018, "grad_norm": 1.2529611587524414, "learning_rate": 0.0001, "loss": 0.0372, "step": 27450 }, { "epoch": 174.9044585987261, "grad_norm": 1.359026551246643, "learning_rate": 0.0001, "loss": 0.0419, "step": 27460 }, { "epoch": 174.96815286624204, "grad_norm": 1.3663288354873657, "learning_rate": 0.0001, "loss": 0.0384, "step": 27470 }, { "epoch": 175.03184713375796, "grad_norm": 1.3026119470596313, "learning_rate": 0.0001, "loss": 0.0375, "step": 27480 }, { "epoch": 175.0955414012739, "grad_norm": 1.3242886066436768, "learning_rate": 0.0001, "loss": 0.0377, "step": 27490 }, { "epoch": 175.15923566878982, "grad_norm": 1.3016964197158813, "learning_rate": 0.0001, "loss": 0.0381, "step": 27500 }, { "epoch": 175.22292993630575, "grad_norm": 1.2981458902359009, "learning_rate": 0.0001, "loss": 0.0412, "step": 27510 }, { "epoch": 175.28662420382165, "grad_norm": 1.2078138589859009, "learning_rate": 0.0001, "loss": 0.0395, "step": 27520 }, { "epoch": 175.35031847133757, "grad_norm": 1.350228190422058, "learning_rate": 0.0001, "loss": 0.0407, "step": 27530 }, { "epoch": 175.4140127388535, "grad_norm": 1.1330409049987793, "learning_rate": 0.0001, "loss": 0.0369, "step": 27540 }, { "epoch": 175.47770700636943, "grad_norm": 1.3226470947265625, "learning_rate": 0.0001, "loss": 0.037, "step": 27550 }, { "epoch": 175.54140127388536, "grad_norm": 0.8587093353271484, "learning_rate": 0.0001, "loss": 0.038, "step": 27560 }, { "epoch": 175.60509554140128, "grad_norm": 1.1406182050704956, "learning_rate": 0.0001, "loss": 0.0411, "step": 27570 }, { "epoch": 175.6687898089172, "grad_norm": 1.1899775266647339, "learning_rate": 0.0001, "loss": 0.0387, "step": 27580 }, { "epoch": 175.7324840764331, "grad_norm": 1.555887222290039, "learning_rate": 0.0001, "loss": 0.0423, "step": 27590 }, { "epoch": 175.79617834394904, "grad_norm": 1.20456862449646, "learning_rate": 0.0001, "loss": 0.0375, "step": 27600 }, { "epoch": 175.85987261146497, "grad_norm": 1.2808033227920532, "learning_rate": 0.0001, "loss": 0.0374, "step": 27610 }, { "epoch": 175.9235668789809, "grad_norm": 1.0320719480514526, "learning_rate": 0.0001, "loss": 0.041, "step": 27620 }, { "epoch": 175.98726114649682, "grad_norm": 1.1039716005325317, "learning_rate": 0.0001, "loss": 0.0419, "step": 27630 }, { "epoch": 176.05095541401275, "grad_norm": 1.061352252960205, "learning_rate": 0.0001, "loss": 0.0392, "step": 27640 }, { "epoch": 176.11464968152868, "grad_norm": 1.1100679636001587, "learning_rate": 0.0001, "loss": 0.0391, "step": 27650 }, { "epoch": 176.17834394904457, "grad_norm": 0.9597607851028442, "learning_rate": 0.0001, "loss": 0.0402, "step": 27660 }, { "epoch": 176.2420382165605, "grad_norm": 1.3692336082458496, "learning_rate": 0.0001, "loss": 0.0412, "step": 27670 }, { "epoch": 176.30573248407643, "grad_norm": 1.313267707824707, "learning_rate": 0.0001, "loss": 0.0421, "step": 27680 }, { "epoch": 176.36942675159236, "grad_norm": 1.3886327743530273, "learning_rate": 0.0001, "loss": 0.0383, "step": 27690 }, { "epoch": 176.43312101910828, "grad_norm": 0.9382830858230591, "learning_rate": 0.0001, "loss": 0.0394, "step": 27700 }, { "epoch": 176.4968152866242, "grad_norm": 1.3715250492095947, "learning_rate": 0.0001, "loss": 0.0398, "step": 27710 }, { "epoch": 176.56050955414014, "grad_norm": 1.1132737398147583, "learning_rate": 0.0001, "loss": 0.0418, "step": 27720 }, { "epoch": 176.62420382165604, "grad_norm": 1.1000609397888184, "learning_rate": 0.0001, "loss": 0.04, "step": 27730 }, { "epoch": 176.68789808917197, "grad_norm": 0.7890186309814453, "learning_rate": 0.0001, "loss": 0.0421, "step": 27740 }, { "epoch": 176.7515923566879, "grad_norm": 0.9523730278015137, "learning_rate": 0.0001, "loss": 0.0391, "step": 27750 }, { "epoch": 176.81528662420382, "grad_norm": 1.2351354360580444, "learning_rate": 0.0001, "loss": 0.0403, "step": 27760 }, { "epoch": 176.87898089171975, "grad_norm": 1.0298023223876953, "learning_rate": 0.0001, "loss": 0.0406, "step": 27770 }, { "epoch": 176.94267515923568, "grad_norm": 1.0587148666381836, "learning_rate": 0.0001, "loss": 0.0419, "step": 27780 }, { "epoch": 177.0063694267516, "grad_norm": 1.1526753902435303, "learning_rate": 0.0001, "loss": 0.0414, "step": 27790 }, { "epoch": 177.0700636942675, "grad_norm": 1.1323052644729614, "learning_rate": 0.0001, "loss": 0.0418, "step": 27800 }, { "epoch": 177.13375796178343, "grad_norm": 1.092228889465332, "learning_rate": 0.0001, "loss": 0.044, "step": 27810 }, { "epoch": 177.19745222929936, "grad_norm": 1.4680229425430298, "learning_rate": 0.0001, "loss": 0.0396, "step": 27820 }, { "epoch": 177.2611464968153, "grad_norm": 1.0889805555343628, "learning_rate": 0.0001, "loss": 0.0403, "step": 27830 }, { "epoch": 177.3248407643312, "grad_norm": 0.9758105278015137, "learning_rate": 0.0001, "loss": 0.0416, "step": 27840 }, { "epoch": 177.38853503184714, "grad_norm": 1.222868800163269, "learning_rate": 0.0001, "loss": 0.0386, "step": 27850 }, { "epoch": 177.45222929936307, "grad_norm": 0.9305564165115356, "learning_rate": 0.0001, "loss": 0.0379, "step": 27860 }, { "epoch": 177.51592356687897, "grad_norm": 1.3661898374557495, "learning_rate": 0.0001, "loss": 0.0419, "step": 27870 }, { "epoch": 177.5796178343949, "grad_norm": 1.2020865678787231, "learning_rate": 0.0001, "loss": 0.0428, "step": 27880 }, { "epoch": 177.64331210191082, "grad_norm": 1.4377483129501343, "learning_rate": 0.0001, "loss": 0.0407, "step": 27890 }, { "epoch": 177.70700636942675, "grad_norm": 1.1283162832260132, "learning_rate": 0.0001, "loss": 0.0369, "step": 27900 }, { "epoch": 177.77070063694268, "grad_norm": 1.4635212421417236, "learning_rate": 0.0001, "loss": 0.0402, "step": 27910 }, { "epoch": 177.8343949044586, "grad_norm": 1.320654034614563, "learning_rate": 0.0001, "loss": 0.0415, "step": 27920 }, { "epoch": 177.89808917197453, "grad_norm": 1.3256644010543823, "learning_rate": 0.0001, "loss": 0.0409, "step": 27930 }, { "epoch": 177.96178343949043, "grad_norm": 1.4172810316085815, "learning_rate": 0.0001, "loss": 0.0435, "step": 27940 }, { "epoch": 178.02547770700636, "grad_norm": 1.4686444997787476, "learning_rate": 0.0001, "loss": 0.0418, "step": 27950 }, { "epoch": 178.0891719745223, "grad_norm": 1.4916799068450928, "learning_rate": 0.0001, "loss": 0.0378, "step": 27960 }, { "epoch": 178.15286624203821, "grad_norm": 1.4036619663238525, "learning_rate": 0.0001, "loss": 0.0419, "step": 27970 }, { "epoch": 178.21656050955414, "grad_norm": 1.3803871870040894, "learning_rate": 0.0001, "loss": 0.0427, "step": 27980 }, { "epoch": 178.28025477707007, "grad_norm": 1.3439949750900269, "learning_rate": 0.0001, "loss": 0.0413, "step": 27990 }, { "epoch": 178.343949044586, "grad_norm": 1.5502773523330688, "learning_rate": 0.0001, "loss": 0.0423, "step": 28000 }, { "epoch": 178.40764331210192, "grad_norm": 1.417075514793396, "learning_rate": 0.0001, "loss": 0.0413, "step": 28010 }, { "epoch": 178.47133757961782, "grad_norm": 1.4388283491134644, "learning_rate": 0.0001, "loss": 0.0396, "step": 28020 }, { "epoch": 178.53503184713375, "grad_norm": 1.2522529363632202, "learning_rate": 0.0001, "loss": 0.0363, "step": 28030 }, { "epoch": 178.59872611464968, "grad_norm": 1.5700647830963135, "learning_rate": 0.0001, "loss": 0.0419, "step": 28040 }, { "epoch": 178.6624203821656, "grad_norm": 1.1416901350021362, "learning_rate": 0.0001, "loss": 0.0419, "step": 28050 }, { "epoch": 178.72611464968153, "grad_norm": 1.252306580543518, "learning_rate": 0.0001, "loss": 0.0379, "step": 28060 }, { "epoch": 178.78980891719746, "grad_norm": 1.0787714719772339, "learning_rate": 0.0001, "loss": 0.0402, "step": 28070 }, { "epoch": 178.8535031847134, "grad_norm": 1.6872787475585938, "learning_rate": 0.0001, "loss": 0.0369, "step": 28080 }, { "epoch": 178.9171974522293, "grad_norm": 1.083227276802063, "learning_rate": 0.0001, "loss": 0.0373, "step": 28090 }, { "epoch": 178.98089171974522, "grad_norm": 1.2707674503326416, "learning_rate": 0.0001, "loss": 0.0415, "step": 28100 }, { "epoch": 179.04458598726114, "grad_norm": 1.1461601257324219, "learning_rate": 0.0001, "loss": 0.0414, "step": 28110 }, { "epoch": 179.10828025477707, "grad_norm": 1.034867286682129, "learning_rate": 0.0001, "loss": 0.0379, "step": 28120 }, { "epoch": 179.171974522293, "grad_norm": 1.2111843824386597, "learning_rate": 0.0001, "loss": 0.0369, "step": 28130 }, { "epoch": 179.23566878980893, "grad_norm": 1.1937381029129028, "learning_rate": 0.0001, "loss": 0.0425, "step": 28140 }, { "epoch": 179.29936305732485, "grad_norm": 1.493621826171875, "learning_rate": 0.0001, "loss": 0.0407, "step": 28150 }, { "epoch": 179.36305732484075, "grad_norm": 1.3387782573699951, "learning_rate": 0.0001, "loss": 0.0392, "step": 28160 }, { "epoch": 179.42675159235668, "grad_norm": 1.4262067079544067, "learning_rate": 0.0001, "loss": 0.0382, "step": 28170 }, { "epoch": 179.4904458598726, "grad_norm": 1.1009641885757446, "learning_rate": 0.0001, "loss": 0.0385, "step": 28180 }, { "epoch": 179.55414012738854, "grad_norm": 1.341638445854187, "learning_rate": 0.0001, "loss": 0.0366, "step": 28190 }, { "epoch": 179.61783439490446, "grad_norm": 1.1314433813095093, "learning_rate": 0.0001, "loss": 0.0382, "step": 28200 }, { "epoch": 179.6815286624204, "grad_norm": 0.9736745357513428, "learning_rate": 0.0001, "loss": 0.0407, "step": 28210 }, { "epoch": 179.74522292993632, "grad_norm": 1.399791955947876, "learning_rate": 0.0001, "loss": 0.0388, "step": 28220 }, { "epoch": 179.80891719745222, "grad_norm": 1.348896861076355, "learning_rate": 0.0001, "loss": 0.0379, "step": 28230 }, { "epoch": 179.87261146496814, "grad_norm": 1.2279342412948608, "learning_rate": 0.0001, "loss": 0.0395, "step": 28240 }, { "epoch": 179.93630573248407, "grad_norm": 1.3260265588760376, "learning_rate": 0.0001, "loss": 0.0404, "step": 28250 }, { "epoch": 180.0, "grad_norm": 1.13275146484375, "learning_rate": 0.0001, "loss": 0.0462, "step": 28260 }, { "epoch": 180.06369426751593, "grad_norm": 1.2856526374816895, "learning_rate": 0.0001, "loss": 0.0373, "step": 28270 }, { "epoch": 180.12738853503186, "grad_norm": 1.2428916692733765, "learning_rate": 0.0001, "loss": 0.0401, "step": 28280 }, { "epoch": 180.19108280254778, "grad_norm": 1.213640809059143, "learning_rate": 0.0001, "loss": 0.0415, "step": 28290 }, { "epoch": 180.25477707006368, "grad_norm": 1.129056692123413, "learning_rate": 0.0001, "loss": 0.0366, "step": 28300 }, { "epoch": 180.3184713375796, "grad_norm": 1.2163691520690918, "learning_rate": 0.0001, "loss": 0.0394, "step": 28310 }, { "epoch": 180.38216560509554, "grad_norm": 1.17877197265625, "learning_rate": 0.0001, "loss": 0.0393, "step": 28320 }, { "epoch": 180.44585987261146, "grad_norm": 1.60802161693573, "learning_rate": 0.0001, "loss": 0.0378, "step": 28330 }, { "epoch": 180.5095541401274, "grad_norm": 1.2360820770263672, "learning_rate": 0.0001, "loss": 0.0414, "step": 28340 }, { "epoch": 180.57324840764332, "grad_norm": 1.2843979597091675, "learning_rate": 0.0001, "loss": 0.0418, "step": 28350 }, { "epoch": 180.63694267515925, "grad_norm": 1.2933907508850098, "learning_rate": 0.0001, "loss": 0.0389, "step": 28360 }, { "epoch": 180.70063694267515, "grad_norm": 1.1653871536254883, "learning_rate": 0.0001, "loss": 0.0387, "step": 28370 }, { "epoch": 180.76433121019107, "grad_norm": 0.8834797143936157, "learning_rate": 0.0001, "loss": 0.0374, "step": 28380 }, { "epoch": 180.828025477707, "grad_norm": 1.340494990348816, "learning_rate": 0.0001, "loss": 0.0416, "step": 28390 }, { "epoch": 180.89171974522293, "grad_norm": 1.1351171731948853, "learning_rate": 0.0001, "loss": 0.0394, "step": 28400 }, { "epoch": 180.95541401273886, "grad_norm": 1.1681863069534302, "learning_rate": 0.0001, "loss": 0.0408, "step": 28410 }, { "epoch": 181.01910828025478, "grad_norm": 1.0287246704101562, "learning_rate": 0.0001, "loss": 0.0427, "step": 28420 }, { "epoch": 181.0828025477707, "grad_norm": 1.2387866973876953, "learning_rate": 0.0001, "loss": 0.0443, "step": 28430 }, { "epoch": 181.1464968152866, "grad_norm": 1.4623594284057617, "learning_rate": 0.0001, "loss": 0.0401, "step": 28440 }, { "epoch": 181.21019108280254, "grad_norm": 1.4812171459197998, "learning_rate": 0.0001, "loss": 0.0428, "step": 28450 }, { "epoch": 181.27388535031847, "grad_norm": 1.2184154987335205, "learning_rate": 0.0001, "loss": 0.0372, "step": 28460 }, { "epoch": 181.3375796178344, "grad_norm": 1.300281286239624, "learning_rate": 0.0001, "loss": 0.0384, "step": 28470 }, { "epoch": 181.40127388535032, "grad_norm": 1.1575673818588257, "learning_rate": 0.0001, "loss": 0.0387, "step": 28480 }, { "epoch": 181.46496815286625, "grad_norm": 1.4423770904541016, "learning_rate": 0.0001, "loss": 0.0394, "step": 28490 }, { "epoch": 181.52866242038218, "grad_norm": 1.4819447994232178, "learning_rate": 0.0001, "loss": 0.0391, "step": 28500 }, { "epoch": 181.59235668789808, "grad_norm": 1.1941454410552979, "learning_rate": 0.0001, "loss": 0.0412, "step": 28510 }, { "epoch": 181.656050955414, "grad_norm": 1.2021304368972778, "learning_rate": 0.0001, "loss": 0.0405, "step": 28520 }, { "epoch": 181.71974522292993, "grad_norm": 1.3605358600616455, "learning_rate": 0.0001, "loss": 0.039, "step": 28530 }, { "epoch": 181.78343949044586, "grad_norm": 1.2624249458312988, "learning_rate": 0.0001, "loss": 0.0381, "step": 28540 }, { "epoch": 181.84713375796179, "grad_norm": 0.9344484806060791, "learning_rate": 0.0001, "loss": 0.0401, "step": 28550 }, { "epoch": 181.9108280254777, "grad_norm": 0.9739436507225037, "learning_rate": 0.0001, "loss": 0.039, "step": 28560 }, { "epoch": 181.97452229299364, "grad_norm": 1.3892474174499512, "learning_rate": 0.0001, "loss": 0.0392, "step": 28570 }, { "epoch": 182.03821656050957, "grad_norm": 1.1415321826934814, "learning_rate": 0.0001, "loss": 0.0389, "step": 28580 }, { "epoch": 182.10191082802547, "grad_norm": 1.260800838470459, "learning_rate": 0.0001, "loss": 0.0359, "step": 28590 }, { "epoch": 182.1656050955414, "grad_norm": 1.240836262702942, "learning_rate": 0.0001, "loss": 0.0369, "step": 28600 }, { "epoch": 182.22929936305732, "grad_norm": 1.017594814300537, "learning_rate": 0.0001, "loss": 0.0333, "step": 28610 }, { "epoch": 182.29299363057325, "grad_norm": 1.1513885259628296, "learning_rate": 0.0001, "loss": 0.0387, "step": 28620 }, { "epoch": 182.35668789808918, "grad_norm": 1.2033592462539673, "learning_rate": 0.0001, "loss": 0.0383, "step": 28630 }, { "epoch": 182.4203821656051, "grad_norm": 1.212739109992981, "learning_rate": 0.0001, "loss": 0.0403, "step": 28640 }, { "epoch": 182.48407643312103, "grad_norm": 1.4185130596160889, "learning_rate": 0.0001, "loss": 0.0393, "step": 28650 }, { "epoch": 182.54777070063693, "grad_norm": 1.0242481231689453, "learning_rate": 0.0001, "loss": 0.042, "step": 28660 }, { "epoch": 182.61146496815286, "grad_norm": 1.080989956855774, "learning_rate": 0.0001, "loss": 0.041, "step": 28670 }, { "epoch": 182.6751592356688, "grad_norm": 1.0556470155715942, "learning_rate": 0.0001, "loss": 0.0421, "step": 28680 }, { "epoch": 182.7388535031847, "grad_norm": 0.9175068736076355, "learning_rate": 0.0001, "loss": 0.0371, "step": 28690 }, { "epoch": 182.80254777070064, "grad_norm": 1.133894920349121, "learning_rate": 0.0001, "loss": 0.0419, "step": 28700 }, { "epoch": 182.86624203821657, "grad_norm": 0.9041225910186768, "learning_rate": 0.0001, "loss": 0.0396, "step": 28710 }, { "epoch": 182.9299363057325, "grad_norm": 0.8762444853782654, "learning_rate": 0.0001, "loss": 0.0376, "step": 28720 }, { "epoch": 182.9936305732484, "grad_norm": 1.2266799211502075, "learning_rate": 0.0001, "loss": 0.0392, "step": 28730 }, { "epoch": 183.05732484076432, "grad_norm": 1.022120714187622, "learning_rate": 0.0001, "loss": 0.0395, "step": 28740 }, { "epoch": 183.12101910828025, "grad_norm": 0.941301703453064, "learning_rate": 0.0001, "loss": 0.0406, "step": 28750 }, { "epoch": 183.18471337579618, "grad_norm": 1.016165852546692, "learning_rate": 0.0001, "loss": 0.0436, "step": 28760 }, { "epoch": 183.2484076433121, "grad_norm": 1.1180089712142944, "learning_rate": 0.0001, "loss": 0.0385, "step": 28770 }, { "epoch": 183.31210191082803, "grad_norm": 0.948436975479126, "learning_rate": 0.0001, "loss": 0.0361, "step": 28780 }, { "epoch": 183.37579617834396, "grad_norm": 1.010602355003357, "learning_rate": 0.0001, "loss": 0.0385, "step": 28790 }, { "epoch": 183.43949044585986, "grad_norm": 1.342373251914978, "learning_rate": 0.0001, "loss": 0.0388, "step": 28800 }, { "epoch": 183.5031847133758, "grad_norm": 0.9900673031806946, "learning_rate": 0.0001, "loss": 0.0418, "step": 28810 }, { "epoch": 183.56687898089172, "grad_norm": 1.2544835805892944, "learning_rate": 0.0001, "loss": 0.043, "step": 28820 }, { "epoch": 183.63057324840764, "grad_norm": 1.0332496166229248, "learning_rate": 0.0001, "loss": 0.0395, "step": 28830 }, { "epoch": 183.69426751592357, "grad_norm": 0.9654817581176758, "learning_rate": 0.0001, "loss": 0.0371, "step": 28840 }, { "epoch": 183.7579617834395, "grad_norm": 1.058855652809143, "learning_rate": 0.0001, "loss": 0.0416, "step": 28850 }, { "epoch": 183.82165605095543, "grad_norm": 1.0760524272918701, "learning_rate": 0.0001, "loss": 0.0387, "step": 28860 }, { "epoch": 183.88535031847132, "grad_norm": 1.2122528553009033, "learning_rate": 0.0001, "loss": 0.037, "step": 28870 }, { "epoch": 183.94904458598725, "grad_norm": 1.4050918817520142, "learning_rate": 0.0001, "loss": 0.0388, "step": 28880 }, { "epoch": 184.01273885350318, "grad_norm": 1.1008667945861816, "learning_rate": 0.0001, "loss": 0.0389, "step": 28890 }, { "epoch": 184.0764331210191, "grad_norm": 1.1355342864990234, "learning_rate": 0.0001, "loss": 0.0407, "step": 28900 }, { "epoch": 184.14012738853503, "grad_norm": 1.375720500946045, "learning_rate": 0.0001, "loss": 0.0399, "step": 28910 }, { "epoch": 184.20382165605096, "grad_norm": 1.5118138790130615, "learning_rate": 0.0001, "loss": 0.0408, "step": 28920 }, { "epoch": 184.2675159235669, "grad_norm": 1.6296851634979248, "learning_rate": 0.0001, "loss": 0.0366, "step": 28930 }, { "epoch": 184.3312101910828, "grad_norm": 1.3748072385787964, "learning_rate": 0.0001, "loss": 0.0381, "step": 28940 }, { "epoch": 184.39490445859872, "grad_norm": 1.6597249507904053, "learning_rate": 0.0001, "loss": 0.0403, "step": 28950 }, { "epoch": 184.45859872611464, "grad_norm": 1.32137131690979, "learning_rate": 0.0001, "loss": 0.0415, "step": 28960 }, { "epoch": 184.52229299363057, "grad_norm": 1.3911018371582031, "learning_rate": 0.0001, "loss": 0.0379, "step": 28970 }, { "epoch": 184.5859872611465, "grad_norm": 1.2566313743591309, "learning_rate": 0.0001, "loss": 0.0385, "step": 28980 }, { "epoch": 184.64968152866243, "grad_norm": 1.347359538078308, "learning_rate": 0.0001, "loss": 0.0379, "step": 28990 }, { "epoch": 184.71337579617835, "grad_norm": 1.0616509914398193, "learning_rate": 0.0001, "loss": 0.0357, "step": 29000 }, { "epoch": 184.77707006369425, "grad_norm": 1.1297980546951294, "learning_rate": 0.0001, "loss": 0.0382, "step": 29010 }, { "epoch": 184.84076433121018, "grad_norm": 1.7067506313323975, "learning_rate": 0.0001, "loss": 0.0407, "step": 29020 }, { "epoch": 184.9044585987261, "grad_norm": 1.3019959926605225, "learning_rate": 0.0001, "loss": 0.0392, "step": 29030 }, { "epoch": 184.96815286624204, "grad_norm": 1.4575566053390503, "learning_rate": 0.0001, "loss": 0.0385, "step": 29040 }, { "epoch": 185.03184713375796, "grad_norm": 1.3656929731369019, "learning_rate": 0.0001, "loss": 0.0353, "step": 29050 }, { "epoch": 185.0955414012739, "grad_norm": 1.2784122228622437, "learning_rate": 0.0001, "loss": 0.0379, "step": 29060 }, { "epoch": 185.15923566878982, "grad_norm": 1.191662311553955, "learning_rate": 0.0001, "loss": 0.0379, "step": 29070 }, { "epoch": 185.22292993630575, "grad_norm": 1.160449743270874, "learning_rate": 0.0001, "loss": 0.0395, "step": 29080 }, { "epoch": 185.28662420382165, "grad_norm": 1.248860478401184, "learning_rate": 0.0001, "loss": 0.0376, "step": 29090 }, { "epoch": 185.35031847133757, "grad_norm": 1.2548590898513794, "learning_rate": 0.0001, "loss": 0.0385, "step": 29100 }, { "epoch": 185.4140127388535, "grad_norm": 1.3346754312515259, "learning_rate": 0.0001, "loss": 0.0365, "step": 29110 }, { "epoch": 185.47770700636943, "grad_norm": 1.6456656455993652, "learning_rate": 0.0001, "loss": 0.0379, "step": 29120 }, { "epoch": 185.54140127388536, "grad_norm": 1.7478750944137573, "learning_rate": 0.0001, "loss": 0.0412, "step": 29130 }, { "epoch": 185.60509554140128, "grad_norm": 1.8883826732635498, "learning_rate": 0.0001, "loss": 0.0395, "step": 29140 }, { "epoch": 185.6687898089172, "grad_norm": 1.681461215019226, "learning_rate": 0.0001, "loss": 0.0394, "step": 29150 }, { "epoch": 185.7324840764331, "grad_norm": 1.4162949323654175, "learning_rate": 0.0001, "loss": 0.0386, "step": 29160 }, { "epoch": 185.79617834394904, "grad_norm": 1.362223505973816, "learning_rate": 0.0001, "loss": 0.0376, "step": 29170 }, { "epoch": 185.85987261146497, "grad_norm": 1.4726401567459106, "learning_rate": 0.0001, "loss": 0.0372, "step": 29180 }, { "epoch": 185.9235668789809, "grad_norm": 1.681881070137024, "learning_rate": 0.0001, "loss": 0.0379, "step": 29190 }, { "epoch": 185.98726114649682, "grad_norm": 1.3844432830810547, "learning_rate": 0.0001, "loss": 0.0359, "step": 29200 }, { "epoch": 186.05095541401275, "grad_norm": 1.371648907661438, "learning_rate": 0.0001, "loss": 0.0363, "step": 29210 }, { "epoch": 186.11464968152868, "grad_norm": 1.3851096630096436, "learning_rate": 0.0001, "loss": 0.0352, "step": 29220 }, { "epoch": 186.17834394904457, "grad_norm": 1.335100531578064, "learning_rate": 0.0001, "loss": 0.0367, "step": 29230 }, { "epoch": 186.2420382165605, "grad_norm": 1.5366061925888062, "learning_rate": 0.0001, "loss": 0.0381, "step": 29240 }, { "epoch": 186.30573248407643, "grad_norm": 2.3087072372436523, "learning_rate": 0.0001, "loss": 0.0371, "step": 29250 }, { "epoch": 186.36942675159236, "grad_norm": 1.472855567932129, "learning_rate": 0.0001, "loss": 0.041, "step": 29260 }, { "epoch": 186.43312101910828, "grad_norm": 1.6458349227905273, "learning_rate": 0.0001, "loss": 0.0361, "step": 29270 }, { "epoch": 186.4968152866242, "grad_norm": 1.4345064163208008, "learning_rate": 0.0001, "loss": 0.0372, "step": 29280 }, { "epoch": 186.56050955414014, "grad_norm": 1.247711181640625, "learning_rate": 0.0001, "loss": 0.0369, "step": 29290 }, { "epoch": 186.62420382165604, "grad_norm": 1.0004384517669678, "learning_rate": 0.0001, "loss": 0.036, "step": 29300 }, { "epoch": 186.68789808917197, "grad_norm": 1.2691227197647095, "learning_rate": 0.0001, "loss": 0.0395, "step": 29310 }, { "epoch": 186.7515923566879, "grad_norm": 1.253364086151123, "learning_rate": 0.0001, "loss": 0.0384, "step": 29320 }, { "epoch": 186.81528662420382, "grad_norm": 1.4148814678192139, "learning_rate": 0.0001, "loss": 0.0382, "step": 29330 }, { "epoch": 186.87898089171975, "grad_norm": 1.3054192066192627, "learning_rate": 0.0001, "loss": 0.0401, "step": 29340 }, { "epoch": 186.94267515923568, "grad_norm": 1.2333908081054688, "learning_rate": 0.0001, "loss": 0.0409, "step": 29350 }, { "epoch": 187.0063694267516, "grad_norm": 1.281765341758728, "learning_rate": 0.0001, "loss": 0.0395, "step": 29360 }, { "epoch": 187.0700636942675, "grad_norm": 1.0807627439498901, "learning_rate": 0.0001, "loss": 0.0389, "step": 29370 }, { "epoch": 187.13375796178343, "grad_norm": 1.0689653158187866, "learning_rate": 0.0001, "loss": 0.0377, "step": 29380 }, { "epoch": 187.19745222929936, "grad_norm": 1.1184011697769165, "learning_rate": 0.0001, "loss": 0.0421, "step": 29390 }, { "epoch": 187.2611464968153, "grad_norm": 1.1315314769744873, "learning_rate": 0.0001, "loss": 0.0369, "step": 29400 }, { "epoch": 187.3248407643312, "grad_norm": 1.0055714845657349, "learning_rate": 0.0001, "loss": 0.0365, "step": 29410 }, { "epoch": 187.38853503184714, "grad_norm": 0.9172794222831726, "learning_rate": 0.0001, "loss": 0.0403, "step": 29420 }, { "epoch": 187.45222929936307, "grad_norm": 1.2071651220321655, "learning_rate": 0.0001, "loss": 0.0402, "step": 29430 }, { "epoch": 187.51592356687897, "grad_norm": 1.2824705839157104, "learning_rate": 0.0001, "loss": 0.042, "step": 29440 }, { "epoch": 187.5796178343949, "grad_norm": 1.3816717863082886, "learning_rate": 0.0001, "loss": 0.0394, "step": 29450 }, { "epoch": 187.64331210191082, "grad_norm": 1.4384416341781616, "learning_rate": 0.0001, "loss": 0.038, "step": 29460 }, { "epoch": 187.70700636942675, "grad_norm": 1.4576616287231445, "learning_rate": 0.0001, "loss": 0.0382, "step": 29470 }, { "epoch": 187.77070063694268, "grad_norm": 1.379384160041809, "learning_rate": 0.0001, "loss": 0.0396, "step": 29480 }, { "epoch": 187.8343949044586, "grad_norm": 1.037794828414917, "learning_rate": 0.0001, "loss": 0.0379, "step": 29490 }, { "epoch": 187.89808917197453, "grad_norm": 1.3423106670379639, "learning_rate": 0.0001, "loss": 0.0401, "step": 29500 }, { "epoch": 187.96178343949043, "grad_norm": 1.4240047931671143, "learning_rate": 0.0001, "loss": 0.0432, "step": 29510 }, { "epoch": 188.02547770700636, "grad_norm": 0.9500240087509155, "learning_rate": 0.0001, "loss": 0.0415, "step": 29520 }, { "epoch": 188.0891719745223, "grad_norm": 1.3472676277160645, "learning_rate": 0.0001, "loss": 0.0386, "step": 29530 }, { "epoch": 188.15286624203821, "grad_norm": 1.1035850048065186, "learning_rate": 0.0001, "loss": 0.0367, "step": 29540 }, { "epoch": 188.21656050955414, "grad_norm": 1.540549874305725, "learning_rate": 0.0001, "loss": 0.043, "step": 29550 }, { "epoch": 188.28025477707007, "grad_norm": 0.9996964931488037, "learning_rate": 0.0001, "loss": 0.0369, "step": 29560 }, { "epoch": 188.343949044586, "grad_norm": 1.1795777082443237, "learning_rate": 0.0001, "loss": 0.0401, "step": 29570 }, { "epoch": 188.40764331210192, "grad_norm": 1.6337063312530518, "learning_rate": 0.0001, "loss": 0.0406, "step": 29580 }, { "epoch": 188.47133757961782, "grad_norm": 1.3726274967193604, "learning_rate": 0.0001, "loss": 0.0381, "step": 29590 }, { "epoch": 188.53503184713375, "grad_norm": 1.4340685606002808, "learning_rate": 0.0001, "loss": 0.0385, "step": 29600 }, { "epoch": 188.59872611464968, "grad_norm": 1.2615129947662354, "learning_rate": 0.0001, "loss": 0.0391, "step": 29610 }, { "epoch": 188.6624203821656, "grad_norm": 1.139945387840271, "learning_rate": 0.0001, "loss": 0.038, "step": 29620 }, { "epoch": 188.72611464968153, "grad_norm": 1.652138113975525, "learning_rate": 0.0001, "loss": 0.0369, "step": 29630 }, { "epoch": 188.78980891719746, "grad_norm": 1.6041374206542969, "learning_rate": 0.0001, "loss": 0.0403, "step": 29640 }, { "epoch": 188.8535031847134, "grad_norm": 1.7945141792297363, "learning_rate": 0.0001, "loss": 0.041, "step": 29650 }, { "epoch": 188.9171974522293, "grad_norm": 1.3952025175094604, "learning_rate": 0.0001, "loss": 0.0395, "step": 29660 }, { "epoch": 188.98089171974522, "grad_norm": 1.1097499132156372, "learning_rate": 0.0001, "loss": 0.0408, "step": 29670 }, { "epoch": 189.04458598726114, "grad_norm": 1.6564316749572754, "learning_rate": 0.0001, "loss": 0.0413, "step": 29680 }, { "epoch": 189.10828025477707, "grad_norm": 1.2588731050491333, "learning_rate": 0.0001, "loss": 0.0383, "step": 29690 }, { "epoch": 189.171974522293, "grad_norm": 1.8140603303909302, "learning_rate": 0.0001, "loss": 0.0397, "step": 29700 }, { "epoch": 189.23566878980893, "grad_norm": 1.4849945306777954, "learning_rate": 0.0001, "loss": 0.0396, "step": 29710 }, { "epoch": 189.29936305732485, "grad_norm": 1.6732988357543945, "learning_rate": 0.0001, "loss": 0.038, "step": 29720 }, { "epoch": 189.36305732484075, "grad_norm": 1.9260339736938477, "learning_rate": 0.0001, "loss": 0.0394, "step": 29730 }, { "epoch": 189.42675159235668, "grad_norm": 1.6890064477920532, "learning_rate": 0.0001, "loss": 0.0398, "step": 29740 }, { "epoch": 189.4904458598726, "grad_norm": 1.2672864198684692, "learning_rate": 0.0001, "loss": 0.0355, "step": 29750 }, { "epoch": 189.55414012738854, "grad_norm": 1.3695805072784424, "learning_rate": 0.0001, "loss": 0.0374, "step": 29760 }, { "epoch": 189.61783439490446, "grad_norm": 1.3142062425613403, "learning_rate": 0.0001, "loss": 0.037, "step": 29770 }, { "epoch": 189.6815286624204, "grad_norm": 1.5822160243988037, "learning_rate": 0.0001, "loss": 0.0361, "step": 29780 }, { "epoch": 189.74522292993632, "grad_norm": 1.2394400835037231, "learning_rate": 0.0001, "loss": 0.0381, "step": 29790 }, { "epoch": 189.80891719745222, "grad_norm": 1.6653004884719849, "learning_rate": 0.0001, "loss": 0.0381, "step": 29800 }, { "epoch": 189.87261146496814, "grad_norm": 1.6157829761505127, "learning_rate": 0.0001, "loss": 0.0355, "step": 29810 }, { "epoch": 189.93630573248407, "grad_norm": 1.3630292415618896, "learning_rate": 0.0001, "loss": 0.0394, "step": 29820 }, { "epoch": 190.0, "grad_norm": 1.2595937252044678, "learning_rate": 0.0001, "loss": 0.0395, "step": 29830 }, { "epoch": 190.06369426751593, "grad_norm": 1.36641263961792, "learning_rate": 0.0001, "loss": 0.0396, "step": 29840 }, { "epoch": 190.12738853503186, "grad_norm": 1.2994872331619263, "learning_rate": 0.0001, "loss": 0.0344, "step": 29850 }, { "epoch": 190.19108280254778, "grad_norm": 1.3295466899871826, "learning_rate": 0.0001, "loss": 0.0374, "step": 29860 }, { "epoch": 190.25477707006368, "grad_norm": 1.178858995437622, "learning_rate": 0.0001, "loss": 0.0369, "step": 29870 }, { "epoch": 190.3184713375796, "grad_norm": 1.5599777698516846, "learning_rate": 0.0001, "loss": 0.0385, "step": 29880 }, { "epoch": 190.38216560509554, "grad_norm": 1.3609580993652344, "learning_rate": 0.0001, "loss": 0.0367, "step": 29890 }, { "epoch": 190.44585987261146, "grad_norm": 1.236720323562622, "learning_rate": 0.0001, "loss": 0.037, "step": 29900 }, { "epoch": 190.5095541401274, "grad_norm": 1.6708694696426392, "learning_rate": 0.0001, "loss": 0.0397, "step": 29910 }, { "epoch": 190.57324840764332, "grad_norm": 1.1640938520431519, "learning_rate": 0.0001, "loss": 0.0365, "step": 29920 }, { "epoch": 190.63694267515925, "grad_norm": 1.1613903045654297, "learning_rate": 0.0001, "loss": 0.0397, "step": 29930 }, { "epoch": 190.70063694267515, "grad_norm": 1.5740258693695068, "learning_rate": 0.0001, "loss": 0.0395, "step": 29940 }, { "epoch": 190.76433121019107, "grad_norm": 1.574310064315796, "learning_rate": 0.0001, "loss": 0.0389, "step": 29950 }, { "epoch": 190.828025477707, "grad_norm": 1.331788182258606, "learning_rate": 0.0001, "loss": 0.0412, "step": 29960 }, { "epoch": 190.89171974522293, "grad_norm": 1.0475136041641235, "learning_rate": 0.0001, "loss": 0.0389, "step": 29970 }, { "epoch": 190.95541401273886, "grad_norm": 0.9208523035049438, "learning_rate": 0.0001, "loss": 0.0414, "step": 29980 }, { "epoch": 191.01910828025478, "grad_norm": 1.201279878616333, "learning_rate": 0.0001, "loss": 0.0355, "step": 29990 }, { "epoch": 191.0828025477707, "grad_norm": 1.3505276441574097, "learning_rate": 0.0001, "loss": 0.0393, "step": 30000 }, { "epoch": 191.1464968152866, "grad_norm": 1.153049111366272, "learning_rate": 0.0001, "loss": 0.0388, "step": 30010 }, { "epoch": 191.21019108280254, "grad_norm": 0.9338540434837341, "learning_rate": 0.0001, "loss": 0.0365, "step": 30020 }, { "epoch": 191.27388535031847, "grad_norm": 0.9516999125480652, "learning_rate": 0.0001, "loss": 0.0386, "step": 30030 }, { "epoch": 191.3375796178344, "grad_norm": 1.532299280166626, "learning_rate": 0.0001, "loss": 0.0378, "step": 30040 }, { "epoch": 191.40127388535032, "grad_norm": 1.6472572088241577, "learning_rate": 0.0001, "loss": 0.0404, "step": 30050 }, { "epoch": 191.46496815286625, "grad_norm": 1.2208566665649414, "learning_rate": 0.0001, "loss": 0.0413, "step": 30060 }, { "epoch": 191.52866242038218, "grad_norm": 1.1361027956008911, "learning_rate": 0.0001, "loss": 0.0421, "step": 30070 }, { "epoch": 191.59235668789808, "grad_norm": 1.1664724349975586, "learning_rate": 0.0001, "loss": 0.0357, "step": 30080 }, { "epoch": 191.656050955414, "grad_norm": 0.9896093010902405, "learning_rate": 0.0001, "loss": 0.0396, "step": 30090 }, { "epoch": 191.71974522292993, "grad_norm": 1.1265963315963745, "learning_rate": 0.0001, "loss": 0.0335, "step": 30100 }, { "epoch": 191.78343949044586, "grad_norm": 1.240195870399475, "learning_rate": 0.0001, "loss": 0.0432, "step": 30110 }, { "epoch": 191.84713375796179, "grad_norm": 0.9545515775680542, "learning_rate": 0.0001, "loss": 0.0375, "step": 30120 }, { "epoch": 191.9108280254777, "grad_norm": 0.9983978271484375, "learning_rate": 0.0001, "loss": 0.0423, "step": 30130 }, { "epoch": 191.97452229299364, "grad_norm": 0.9119243025779724, "learning_rate": 0.0001, "loss": 0.0415, "step": 30140 }, { "epoch": 192.03821656050957, "grad_norm": 1.1292387247085571, "learning_rate": 0.0001, "loss": 0.0388, "step": 30150 }, { "epoch": 192.10191082802547, "grad_norm": 1.595593810081482, "learning_rate": 0.0001, "loss": 0.0389, "step": 30160 }, { "epoch": 192.1656050955414, "grad_norm": 1.0034037828445435, "learning_rate": 0.0001, "loss": 0.0395, "step": 30170 }, { "epoch": 192.22929936305732, "grad_norm": 1.7669800519943237, "learning_rate": 0.0001, "loss": 0.0401, "step": 30180 }, { "epoch": 192.29299363057325, "grad_norm": 0.9602085947990417, "learning_rate": 0.0001, "loss": 0.0415, "step": 30190 }, { "epoch": 192.35668789808918, "grad_norm": 0.9665617942810059, "learning_rate": 0.0001, "loss": 0.0387, "step": 30200 }, { "epoch": 192.4203821656051, "grad_norm": 1.1232694387435913, "learning_rate": 0.0001, "loss": 0.0378, "step": 30210 }, { "epoch": 192.48407643312103, "grad_norm": 0.8711768984794617, "learning_rate": 0.0001, "loss": 0.045, "step": 30220 }, { "epoch": 192.54777070063693, "grad_norm": 1.2056331634521484, "learning_rate": 0.0001, "loss": 0.0423, "step": 30230 }, { "epoch": 192.61146496815286, "grad_norm": 0.9683982729911804, "learning_rate": 0.0001, "loss": 0.0414, "step": 30240 }, { "epoch": 192.6751592356688, "grad_norm": 1.1629849672317505, "learning_rate": 0.0001, "loss": 0.0433, "step": 30250 }, { "epoch": 192.7388535031847, "grad_norm": 1.1420708894729614, "learning_rate": 0.0001, "loss": 0.0432, "step": 30260 }, { "epoch": 192.80254777070064, "grad_norm": 1.2877910137176514, "learning_rate": 0.0001, "loss": 0.0423, "step": 30270 }, { "epoch": 192.86624203821657, "grad_norm": 1.223312258720398, "learning_rate": 0.0001, "loss": 0.0417, "step": 30280 }, { "epoch": 192.9299363057325, "grad_norm": 1.0409001111984253, "learning_rate": 0.0001, "loss": 0.0373, "step": 30290 }, { "epoch": 192.9936305732484, "grad_norm": 1.2361739873886108, "learning_rate": 0.0001, "loss": 0.0416, "step": 30300 }, { "epoch": 193.05732484076432, "grad_norm": 1.1293675899505615, "learning_rate": 0.0001, "loss": 0.0436, "step": 30310 }, { "epoch": 193.12101910828025, "grad_norm": 0.9782614707946777, "learning_rate": 0.0001, "loss": 0.042, "step": 30320 }, { "epoch": 193.18471337579618, "grad_norm": 1.165056824684143, "learning_rate": 0.0001, "loss": 0.0422, "step": 30330 }, { "epoch": 193.2484076433121, "grad_norm": 1.2391263246536255, "learning_rate": 0.0001, "loss": 0.0408, "step": 30340 }, { "epoch": 193.31210191082803, "grad_norm": 1.496337652206421, "learning_rate": 0.0001, "loss": 0.0428, "step": 30350 }, { "epoch": 193.37579617834396, "grad_norm": 1.3156101703643799, "learning_rate": 0.0001, "loss": 0.0403, "step": 30360 }, { "epoch": 193.43949044585986, "grad_norm": 1.2842307090759277, "learning_rate": 0.0001, "loss": 0.0385, "step": 30370 }, { "epoch": 193.5031847133758, "grad_norm": 1.3062889575958252, "learning_rate": 0.0001, "loss": 0.0403, "step": 30380 }, { "epoch": 193.56687898089172, "grad_norm": 1.3258154392242432, "learning_rate": 0.0001, "loss": 0.0405, "step": 30390 }, { "epoch": 193.63057324840764, "grad_norm": 1.312279224395752, "learning_rate": 0.0001, "loss": 0.0389, "step": 30400 }, { "epoch": 193.69426751592357, "grad_norm": 1.3830115795135498, "learning_rate": 0.0001, "loss": 0.0358, "step": 30410 }, { "epoch": 193.7579617834395, "grad_norm": 1.176771640777588, "learning_rate": 0.0001, "loss": 0.0397, "step": 30420 }, { "epoch": 193.82165605095543, "grad_norm": 1.5105870962142944, "learning_rate": 0.0001, "loss": 0.0394, "step": 30430 }, { "epoch": 193.88535031847132, "grad_norm": 1.323082685470581, "learning_rate": 0.0001, "loss": 0.0394, "step": 30440 }, { "epoch": 193.94904458598725, "grad_norm": 1.3793610334396362, "learning_rate": 0.0001, "loss": 0.0401, "step": 30450 }, { "epoch": 194.01273885350318, "grad_norm": 1.2603574991226196, "learning_rate": 0.0001, "loss": 0.0352, "step": 30460 }, { "epoch": 194.0764331210191, "grad_norm": 1.1363389492034912, "learning_rate": 0.0001, "loss": 0.0377, "step": 30470 }, { "epoch": 194.14012738853503, "grad_norm": 1.440557837486267, "learning_rate": 0.0001, "loss": 0.038, "step": 30480 }, { "epoch": 194.20382165605096, "grad_norm": 1.2849828004837036, "learning_rate": 0.0001, "loss": 0.0379, "step": 30490 }, { "epoch": 194.2675159235669, "grad_norm": 1.5465353727340698, "learning_rate": 0.0001, "loss": 0.0361, "step": 30500 }, { "epoch": 194.3312101910828, "grad_norm": 1.4232208728790283, "learning_rate": 0.0001, "loss": 0.0374, "step": 30510 }, { "epoch": 194.39490445859872, "grad_norm": 1.4426549673080444, "learning_rate": 0.0001, "loss": 0.0413, "step": 30520 }, { "epoch": 194.45859872611464, "grad_norm": 1.1907291412353516, "learning_rate": 0.0001, "loss": 0.0371, "step": 30530 }, { "epoch": 194.52229299363057, "grad_norm": 1.441036343574524, "learning_rate": 0.0001, "loss": 0.0383, "step": 30540 }, { "epoch": 194.5859872611465, "grad_norm": 1.3865214586257935, "learning_rate": 0.0001, "loss": 0.0378, "step": 30550 }, { "epoch": 194.64968152866243, "grad_norm": 1.2172904014587402, "learning_rate": 0.0001, "loss": 0.0394, "step": 30560 }, { "epoch": 194.71337579617835, "grad_norm": 1.4048956632614136, "learning_rate": 0.0001, "loss": 0.0414, "step": 30570 }, { "epoch": 194.77707006369425, "grad_norm": 1.2165932655334473, "learning_rate": 0.0001, "loss": 0.0406, "step": 30580 }, { "epoch": 194.84076433121018, "grad_norm": 1.3762054443359375, "learning_rate": 0.0001, "loss": 0.0403, "step": 30590 }, { "epoch": 194.9044585987261, "grad_norm": 1.3413491249084473, "learning_rate": 0.0001, "loss": 0.0392, "step": 30600 }, { "epoch": 194.96815286624204, "grad_norm": 1.259851336479187, "learning_rate": 0.0001, "loss": 0.0412, "step": 30610 }, { "epoch": 195.03184713375796, "grad_norm": 1.1722190380096436, "learning_rate": 0.0001, "loss": 0.0407, "step": 30620 }, { "epoch": 195.0955414012739, "grad_norm": 1.0736509561538696, "learning_rate": 0.0001, "loss": 0.0416, "step": 30630 }, { "epoch": 195.15923566878982, "grad_norm": 1.1030274629592896, "learning_rate": 0.0001, "loss": 0.0402, "step": 30640 }, { "epoch": 195.22292993630575, "grad_norm": 1.2144708633422852, "learning_rate": 0.0001, "loss": 0.0411, "step": 30650 }, { "epoch": 195.28662420382165, "grad_norm": 1.4002349376678467, "learning_rate": 0.0001, "loss": 0.0436, "step": 30660 }, { "epoch": 195.35031847133757, "grad_norm": 1.279742956161499, "learning_rate": 0.0001, "loss": 0.0451, "step": 30670 }, { "epoch": 195.4140127388535, "grad_norm": 1.0116530656814575, "learning_rate": 0.0001, "loss": 0.0411, "step": 30680 }, { "epoch": 195.47770700636943, "grad_norm": 1.0642750263214111, "learning_rate": 0.0001, "loss": 0.0412, "step": 30690 }, { "epoch": 195.54140127388536, "grad_norm": 1.2979053258895874, "learning_rate": 0.0001, "loss": 0.0456, "step": 30700 }, { "epoch": 195.60509554140128, "grad_norm": 1.1634206771850586, "learning_rate": 0.0001, "loss": 0.0407, "step": 30710 }, { "epoch": 195.6687898089172, "grad_norm": 1.3750640153884888, "learning_rate": 0.0001, "loss": 0.0415, "step": 30720 }, { "epoch": 195.7324840764331, "grad_norm": 1.3999735116958618, "learning_rate": 0.0001, "loss": 0.0409, "step": 30730 }, { "epoch": 195.79617834394904, "grad_norm": 1.1802892684936523, "learning_rate": 0.0001, "loss": 0.0417, "step": 30740 }, { "epoch": 195.85987261146497, "grad_norm": 1.0939621925354004, "learning_rate": 0.0001, "loss": 0.039, "step": 30750 }, { "epoch": 195.9235668789809, "grad_norm": 1.3235255479812622, "learning_rate": 0.0001, "loss": 0.0412, "step": 30760 }, { "epoch": 195.98726114649682, "grad_norm": 1.3233615159988403, "learning_rate": 0.0001, "loss": 0.0375, "step": 30770 }, { "epoch": 196.05095541401275, "grad_norm": 0.9408957958221436, "learning_rate": 0.0001, "loss": 0.0393, "step": 30780 }, { "epoch": 196.11464968152868, "grad_norm": 1.3995678424835205, "learning_rate": 0.0001, "loss": 0.0402, "step": 30790 }, { "epoch": 196.17834394904457, "grad_norm": 1.279699683189392, "learning_rate": 0.0001, "loss": 0.036, "step": 30800 }, { "epoch": 196.2420382165605, "grad_norm": 1.142619013786316, "learning_rate": 0.0001, "loss": 0.0427, "step": 30810 }, { "epoch": 196.30573248407643, "grad_norm": 1.320800542831421, "learning_rate": 0.0001, "loss": 0.0395, "step": 30820 }, { "epoch": 196.36942675159236, "grad_norm": 1.5406473875045776, "learning_rate": 0.0001, "loss": 0.0379, "step": 30830 }, { "epoch": 196.43312101910828, "grad_norm": 2.1672701835632324, "learning_rate": 0.0001, "loss": 0.0394, "step": 30840 }, { "epoch": 196.4968152866242, "grad_norm": 1.6804364919662476, "learning_rate": 0.0001, "loss": 0.039, "step": 30850 }, { "epoch": 196.56050955414014, "grad_norm": 1.8836538791656494, "learning_rate": 0.0001, "loss": 0.0412, "step": 30860 }, { "epoch": 196.62420382165604, "grad_norm": 1.4830455780029297, "learning_rate": 0.0001, "loss": 0.0377, "step": 30870 }, { "epoch": 196.68789808917197, "grad_norm": 0.99431973695755, "learning_rate": 0.0001, "loss": 0.0362, "step": 30880 }, { "epoch": 196.7515923566879, "grad_norm": 1.5594602823257446, "learning_rate": 0.0001, "loss": 0.0334, "step": 30890 }, { "epoch": 196.81528662420382, "grad_norm": 1.4091806411743164, "learning_rate": 0.0001, "loss": 0.0412, "step": 30900 }, { "epoch": 196.87898089171975, "grad_norm": 1.266850471496582, "learning_rate": 0.0001, "loss": 0.037, "step": 30910 }, { "epoch": 196.94267515923568, "grad_norm": 1.5250871181488037, "learning_rate": 0.0001, "loss": 0.0392, "step": 30920 }, { "epoch": 197.0063694267516, "grad_norm": 1.4553189277648926, "learning_rate": 0.0001, "loss": 0.0383, "step": 30930 }, { "epoch": 197.0700636942675, "grad_norm": 1.4473998546600342, "learning_rate": 0.0001, "loss": 0.0374, "step": 30940 }, { "epoch": 197.13375796178343, "grad_norm": 1.2245738506317139, "learning_rate": 0.0001, "loss": 0.0405, "step": 30950 }, { "epoch": 197.19745222929936, "grad_norm": 1.164402961730957, "learning_rate": 0.0001, "loss": 0.0367, "step": 30960 }, { "epoch": 197.2611464968153, "grad_norm": 1.1188262701034546, "learning_rate": 0.0001, "loss": 0.0379, "step": 30970 }, { "epoch": 197.3248407643312, "grad_norm": 1.1863760948181152, "learning_rate": 0.0001, "loss": 0.0351, "step": 30980 }, { "epoch": 197.38853503184714, "grad_norm": 1.0835152864456177, "learning_rate": 0.0001, "loss": 0.0391, "step": 30990 }, { "epoch": 197.45222929936307, "grad_norm": 1.2631272077560425, "learning_rate": 0.0001, "loss": 0.036, "step": 31000 }, { "epoch": 197.51592356687897, "grad_norm": 1.3075958490371704, "learning_rate": 0.0001, "loss": 0.0388, "step": 31010 }, { "epoch": 197.5796178343949, "grad_norm": 1.4392507076263428, "learning_rate": 0.0001, "loss": 0.0411, "step": 31020 }, { "epoch": 197.64331210191082, "grad_norm": 1.428548812866211, "learning_rate": 0.0001, "loss": 0.0363, "step": 31030 }, { "epoch": 197.70700636942675, "grad_norm": 1.7614153623580933, "learning_rate": 0.0001, "loss": 0.0398, "step": 31040 }, { "epoch": 197.77070063694268, "grad_norm": 1.5615723133087158, "learning_rate": 0.0001, "loss": 0.0406, "step": 31050 }, { "epoch": 197.8343949044586, "grad_norm": 1.9067742824554443, "learning_rate": 0.0001, "loss": 0.0385, "step": 31060 }, { "epoch": 197.89808917197453, "grad_norm": 1.3763774633407593, "learning_rate": 0.0001, "loss": 0.0388, "step": 31070 }, { "epoch": 197.96178343949043, "grad_norm": 1.6388776302337646, "learning_rate": 0.0001, "loss": 0.0379, "step": 31080 }, { "epoch": 198.02547770700636, "grad_norm": 1.6214971542358398, "learning_rate": 0.0001, "loss": 0.0359, "step": 31090 }, { "epoch": 198.0891719745223, "grad_norm": 1.541113257408142, "learning_rate": 0.0001, "loss": 0.0362, "step": 31100 }, { "epoch": 198.15286624203821, "grad_norm": 1.4273598194122314, "learning_rate": 0.0001, "loss": 0.0362, "step": 31110 }, { "epoch": 198.21656050955414, "grad_norm": 1.3761423826217651, "learning_rate": 0.0001, "loss": 0.0358, "step": 31120 }, { "epoch": 198.28025477707007, "grad_norm": 1.198242425918579, "learning_rate": 0.0001, "loss": 0.0353, "step": 31130 }, { "epoch": 198.343949044586, "grad_norm": 1.1443171501159668, "learning_rate": 0.0001, "loss": 0.0371, "step": 31140 }, { "epoch": 198.40764331210192, "grad_norm": 1.061008095741272, "learning_rate": 0.0001, "loss": 0.0358, "step": 31150 }, { "epoch": 198.47133757961782, "grad_norm": 1.1833891868591309, "learning_rate": 0.0001, "loss": 0.0404, "step": 31160 }, { "epoch": 198.53503184713375, "grad_norm": 1.2825343608856201, "learning_rate": 0.0001, "loss": 0.0384, "step": 31170 }, { "epoch": 198.59872611464968, "grad_norm": 1.0114703178405762, "learning_rate": 0.0001, "loss": 0.0399, "step": 31180 }, { "epoch": 198.6624203821656, "grad_norm": 1.1701241731643677, "learning_rate": 0.0001, "loss": 0.0402, "step": 31190 }, { "epoch": 198.72611464968153, "grad_norm": 1.3151874542236328, "learning_rate": 0.0001, "loss": 0.0377, "step": 31200 }, { "epoch": 198.78980891719746, "grad_norm": 1.0915164947509766, "learning_rate": 0.0001, "loss": 0.0343, "step": 31210 }, { "epoch": 198.8535031847134, "grad_norm": 0.9437780380249023, "learning_rate": 0.0001, "loss": 0.042, "step": 31220 }, { "epoch": 198.9171974522293, "grad_norm": 1.0188149213790894, "learning_rate": 0.0001, "loss": 0.0405, "step": 31230 }, { "epoch": 198.98089171974522, "grad_norm": 1.0938233137130737, "learning_rate": 0.0001, "loss": 0.0371, "step": 31240 }, { "epoch": 199.04458598726114, "grad_norm": 1.1780885457992554, "learning_rate": 0.0001, "loss": 0.0387, "step": 31250 }, { "epoch": 199.10828025477707, "grad_norm": 0.9412981867790222, "learning_rate": 0.0001, "loss": 0.0359, "step": 31260 }, { "epoch": 199.171974522293, "grad_norm": 1.2324005365371704, "learning_rate": 0.0001, "loss": 0.0377, "step": 31270 }, { "epoch": 199.23566878980893, "grad_norm": 1.4699127674102783, "learning_rate": 0.0001, "loss": 0.0396, "step": 31280 }, { "epoch": 199.29936305732485, "grad_norm": 1.0777356624603271, "learning_rate": 0.0001, "loss": 0.0371, "step": 31290 }, { "epoch": 199.36305732484075, "grad_norm": 0.9331629276275635, "learning_rate": 0.0001, "loss": 0.0361, "step": 31300 }, { "epoch": 199.42675159235668, "grad_norm": 1.5903513431549072, "learning_rate": 0.0001, "loss": 0.0364, "step": 31310 }, { "epoch": 199.4904458598726, "grad_norm": 0.9572700262069702, "learning_rate": 0.0001, "loss": 0.0341, "step": 31320 }, { "epoch": 199.55414012738854, "grad_norm": 1.4313803911209106, "learning_rate": 0.0001, "loss": 0.0385, "step": 31330 }, { "epoch": 199.61783439490446, "grad_norm": 1.118453025817871, "learning_rate": 0.0001, "loss": 0.0398, "step": 31340 }, { "epoch": 199.6815286624204, "grad_norm": 1.4428298473358154, "learning_rate": 0.0001, "loss": 0.0388, "step": 31350 }, { "epoch": 199.74522292993632, "grad_norm": 1.294424057006836, "learning_rate": 0.0001, "loss": 0.0427, "step": 31360 }, { "epoch": 199.80891719745222, "grad_norm": 1.394228458404541, "learning_rate": 0.0001, "loss": 0.0381, "step": 31370 }, { "epoch": 199.87261146496814, "grad_norm": 1.3453199863433838, "learning_rate": 0.0001, "loss": 0.0422, "step": 31380 }, { "epoch": 199.93630573248407, "grad_norm": 1.5819547176361084, "learning_rate": 0.0001, "loss": 0.0372, "step": 31390 }, { "epoch": 200.0, "grad_norm": 1.455633521080017, "learning_rate": 0.0001, "loss": 0.0432, "step": 31400 }, { "epoch": 200.06369426751593, "grad_norm": 1.3475197553634644, "learning_rate": 0.0001, "loss": 0.0374, "step": 31410 }, { "epoch": 200.12738853503186, "grad_norm": 1.3173516988754272, "learning_rate": 0.0001, "loss": 0.037, "step": 31420 }, { "epoch": 200.19108280254778, "grad_norm": 1.2071382999420166, "learning_rate": 0.0001, "loss": 0.0437, "step": 31430 }, { "epoch": 200.25477707006368, "grad_norm": 1.4156559705734253, "learning_rate": 0.0001, "loss": 0.0381, "step": 31440 }, { "epoch": 200.3184713375796, "grad_norm": 1.5116679668426514, "learning_rate": 0.0001, "loss": 0.0377, "step": 31450 }, { "epoch": 200.38216560509554, "grad_norm": 1.360504150390625, "learning_rate": 0.0001, "loss": 0.0407, "step": 31460 }, { "epoch": 200.44585987261146, "grad_norm": 1.3780006170272827, "learning_rate": 0.0001, "loss": 0.0358, "step": 31470 }, { "epoch": 200.5095541401274, "grad_norm": 1.2662951946258545, "learning_rate": 0.0001, "loss": 0.04, "step": 31480 }, { "epoch": 200.57324840764332, "grad_norm": 1.4881353378295898, "learning_rate": 0.0001, "loss": 0.0374, "step": 31490 }, { "epoch": 200.63694267515925, "grad_norm": 1.0562597513198853, "learning_rate": 0.0001, "loss": 0.0397, "step": 31500 }, { "epoch": 200.70063694267515, "grad_norm": 1.2676736116409302, "learning_rate": 0.0001, "loss": 0.0369, "step": 31510 }, { "epoch": 200.76433121019107, "grad_norm": 1.1131120920181274, "learning_rate": 0.0001, "loss": 0.0367, "step": 31520 }, { "epoch": 200.828025477707, "grad_norm": 1.1586833000183105, "learning_rate": 0.0001, "loss": 0.0397, "step": 31530 }, { "epoch": 200.89171974522293, "grad_norm": 1.0662219524383545, "learning_rate": 0.0001, "loss": 0.0384, "step": 31540 }, { "epoch": 200.95541401273886, "grad_norm": 1.0737093687057495, "learning_rate": 0.0001, "loss": 0.0391, "step": 31550 }, { "epoch": 201.01910828025478, "grad_norm": 2.0638680458068848, "learning_rate": 0.0001, "loss": 0.044, "step": 31560 }, { "epoch": 201.0828025477707, "grad_norm": 3.3678324222564697, "learning_rate": 0.0001, "loss": 0.0381, "step": 31570 }, { "epoch": 201.1464968152866, "grad_norm": 2.0775318145751953, "learning_rate": 0.0001, "loss": 0.0379, "step": 31580 }, { "epoch": 201.21019108280254, "grad_norm": 2.0469284057617188, "learning_rate": 0.0001, "loss": 0.0373, "step": 31590 }, { "epoch": 201.27388535031847, "grad_norm": 1.459092378616333, "learning_rate": 0.0001, "loss": 0.038, "step": 31600 }, { "epoch": 201.3375796178344, "grad_norm": 1.5063393115997314, "learning_rate": 0.0001, "loss": 0.0353, "step": 31610 }, { "epoch": 201.40127388535032, "grad_norm": 1.1744693517684937, "learning_rate": 0.0001, "loss": 0.0382, "step": 31620 }, { "epoch": 201.46496815286625, "grad_norm": 1.6139277219772339, "learning_rate": 0.0001, "loss": 0.0354, "step": 31630 }, { "epoch": 201.52866242038218, "grad_norm": 1.2319034337997437, "learning_rate": 0.0001, "loss": 0.0366, "step": 31640 }, { "epoch": 201.59235668789808, "grad_norm": 1.7676138877868652, "learning_rate": 0.0001, "loss": 0.0352, "step": 31650 }, { "epoch": 201.656050955414, "grad_norm": 1.3634642362594604, "learning_rate": 0.0001, "loss": 0.0401, "step": 31660 }, { "epoch": 201.71974522292993, "grad_norm": 1.5985932350158691, "learning_rate": 0.0001, "loss": 0.0395, "step": 31670 }, { "epoch": 201.78343949044586, "grad_norm": 1.1848751306533813, "learning_rate": 0.0001, "loss": 0.0374, "step": 31680 }, { "epoch": 201.84713375796179, "grad_norm": 1.3552436828613281, "learning_rate": 0.0001, "loss": 0.0382, "step": 31690 }, { "epoch": 201.9108280254777, "grad_norm": 1.4430491924285889, "learning_rate": 0.0001, "loss": 0.0374, "step": 31700 }, { "epoch": 201.97452229299364, "grad_norm": 1.3431897163391113, "learning_rate": 0.0001, "loss": 0.0369, "step": 31710 }, { "epoch": 202.03821656050957, "grad_norm": 0.835646390914917, "learning_rate": 0.0001, "loss": 0.0315, "step": 31720 }, { "epoch": 202.10191082802547, "grad_norm": 1.1966474056243896, "learning_rate": 0.0001, "loss": 0.0362, "step": 31730 }, { "epoch": 202.1656050955414, "grad_norm": 1.5055004358291626, "learning_rate": 0.0001, "loss": 0.0372, "step": 31740 }, { "epoch": 202.22929936305732, "grad_norm": 1.5647480487823486, "learning_rate": 0.0001, "loss": 0.0394, "step": 31750 }, { "epoch": 202.29299363057325, "grad_norm": 1.0968800783157349, "learning_rate": 0.0001, "loss": 0.0392, "step": 31760 }, { "epoch": 202.35668789808918, "grad_norm": 1.5528323650360107, "learning_rate": 0.0001, "loss": 0.04, "step": 31770 }, { "epoch": 202.4203821656051, "grad_norm": 1.0501762628555298, "learning_rate": 0.0001, "loss": 0.0396, "step": 31780 }, { "epoch": 202.48407643312103, "grad_norm": 1.4192005395889282, "learning_rate": 0.0001, "loss": 0.0359, "step": 31790 }, { "epoch": 202.54777070063693, "grad_norm": 1.0395748615264893, "learning_rate": 0.0001, "loss": 0.0364, "step": 31800 }, { "epoch": 202.61146496815286, "grad_norm": 1.4094828367233276, "learning_rate": 0.0001, "loss": 0.0382, "step": 31810 }, { "epoch": 202.6751592356688, "grad_norm": 1.3492820262908936, "learning_rate": 0.0001, "loss": 0.0395, "step": 31820 }, { "epoch": 202.7388535031847, "grad_norm": 1.2947336435317993, "learning_rate": 0.0001, "loss": 0.0378, "step": 31830 }, { "epoch": 202.80254777070064, "grad_norm": 1.363209843635559, "learning_rate": 0.0001, "loss": 0.0343, "step": 31840 }, { "epoch": 202.86624203821657, "grad_norm": 2.1559031009674072, "learning_rate": 0.0001, "loss": 0.0388, "step": 31850 }, { "epoch": 202.9299363057325, "grad_norm": 2.6023809909820557, "learning_rate": 0.0001, "loss": 0.0385, "step": 31860 }, { "epoch": 202.9936305732484, "grad_norm": 2.0685648918151855, "learning_rate": 0.0001, "loss": 0.038, "step": 31870 }, { "epoch": 203.05732484076432, "grad_norm": 1.5425788164138794, "learning_rate": 0.0001, "loss": 0.0385, "step": 31880 }, { "epoch": 203.12101910828025, "grad_norm": 1.2819983959197998, "learning_rate": 0.0001, "loss": 0.0357, "step": 31890 }, { "epoch": 203.18471337579618, "grad_norm": 1.515013575553894, "learning_rate": 0.0001, "loss": 0.0341, "step": 31900 }, { "epoch": 203.2484076433121, "grad_norm": 1.7473214864730835, "learning_rate": 0.0001, "loss": 0.0368, "step": 31910 }, { "epoch": 203.31210191082803, "grad_norm": 1.6436519622802734, "learning_rate": 0.0001, "loss": 0.0348, "step": 31920 }, { "epoch": 203.37579617834396, "grad_norm": 1.2461966276168823, "learning_rate": 0.0001, "loss": 0.0414, "step": 31930 }, { "epoch": 203.43949044585986, "grad_norm": 1.0842393636703491, "learning_rate": 0.0001, "loss": 0.0355, "step": 31940 }, { "epoch": 203.5031847133758, "grad_norm": 1.391852617263794, "learning_rate": 0.0001, "loss": 0.0398, "step": 31950 }, { "epoch": 203.56687898089172, "grad_norm": 2.2876296043395996, "learning_rate": 0.0001, "loss": 0.04, "step": 31960 }, { "epoch": 203.63057324840764, "grad_norm": 2.01747727394104, "learning_rate": 0.0001, "loss": 0.0396, "step": 31970 }, { "epoch": 203.69426751592357, "grad_norm": 1.7828278541564941, "learning_rate": 0.0001, "loss": 0.0399, "step": 31980 }, { "epoch": 203.7579617834395, "grad_norm": 1.5832936763763428, "learning_rate": 0.0001, "loss": 0.034, "step": 31990 }, { "epoch": 203.82165605095543, "grad_norm": 2.4744555950164795, "learning_rate": 0.0001, "loss": 0.0359, "step": 32000 }, { "epoch": 203.88535031847132, "grad_norm": 1.686309814453125, "learning_rate": 0.0001, "loss": 0.0365, "step": 32010 }, { "epoch": 203.94904458598725, "grad_norm": 1.366796851158142, "learning_rate": 0.0001, "loss": 0.0363, "step": 32020 }, { "epoch": 204.01273885350318, "grad_norm": 1.2280523777008057, "learning_rate": 0.0001, "loss": 0.0362, "step": 32030 }, { "epoch": 204.0764331210191, "grad_norm": 1.405163288116455, "learning_rate": 0.0001, "loss": 0.0352, "step": 32040 }, { "epoch": 204.14012738853503, "grad_norm": 1.2587742805480957, "learning_rate": 0.0001, "loss": 0.0362, "step": 32050 }, { "epoch": 204.20382165605096, "grad_norm": 1.2533652782440186, "learning_rate": 0.0001, "loss": 0.0366, "step": 32060 }, { "epoch": 204.2675159235669, "grad_norm": 1.314207673072815, "learning_rate": 0.0001, "loss": 0.0379, "step": 32070 }, { "epoch": 204.3312101910828, "grad_norm": 1.425071120262146, "learning_rate": 0.0001, "loss": 0.0336, "step": 32080 }, { "epoch": 204.39490445859872, "grad_norm": 1.4265319108963013, "learning_rate": 0.0001, "loss": 0.0363, "step": 32090 }, { "epoch": 204.45859872611464, "grad_norm": 1.4327075481414795, "learning_rate": 0.0001, "loss": 0.0402, "step": 32100 }, { "epoch": 204.52229299363057, "grad_norm": 1.3853281736373901, "learning_rate": 0.0001, "loss": 0.041, "step": 32110 }, { "epoch": 204.5859872611465, "grad_norm": 1.078947901725769, "learning_rate": 0.0001, "loss": 0.0382, "step": 32120 }, { "epoch": 204.64968152866243, "grad_norm": 1.0103532075881958, "learning_rate": 0.0001, "loss": 0.0403, "step": 32130 }, { "epoch": 204.71337579617835, "grad_norm": 1.6044495105743408, "learning_rate": 0.0001, "loss": 0.04, "step": 32140 }, { "epoch": 204.77707006369425, "grad_norm": 1.2725425958633423, "learning_rate": 0.0001, "loss": 0.0369, "step": 32150 }, { "epoch": 204.84076433121018, "grad_norm": 1.144245982170105, "learning_rate": 0.0001, "loss": 0.0366, "step": 32160 }, { "epoch": 204.9044585987261, "grad_norm": 1.1575924158096313, "learning_rate": 0.0001, "loss": 0.0391, "step": 32170 }, { "epoch": 204.96815286624204, "grad_norm": 1.063045620918274, "learning_rate": 0.0001, "loss": 0.0391, "step": 32180 }, { "epoch": 205.03184713375796, "grad_norm": 1.4883577823638916, "learning_rate": 0.0001, "loss": 0.0377, "step": 32190 }, { "epoch": 205.0955414012739, "grad_norm": 1.5128039121627808, "learning_rate": 0.0001, "loss": 0.0358, "step": 32200 }, { "epoch": 205.15923566878982, "grad_norm": 1.3630733489990234, "learning_rate": 0.0001, "loss": 0.0394, "step": 32210 }, { "epoch": 205.22292993630575, "grad_norm": 1.6538875102996826, "learning_rate": 0.0001, "loss": 0.0399, "step": 32220 }, { "epoch": 205.28662420382165, "grad_norm": 1.6160049438476562, "learning_rate": 0.0001, "loss": 0.0394, "step": 32230 }, { "epoch": 205.35031847133757, "grad_norm": 1.6533313989639282, "learning_rate": 0.0001, "loss": 0.0358, "step": 32240 }, { "epoch": 205.4140127388535, "grad_norm": 1.3957695960998535, "learning_rate": 0.0001, "loss": 0.0377, "step": 32250 }, { "epoch": 205.47770700636943, "grad_norm": 1.214178204536438, "learning_rate": 0.0001, "loss": 0.0352, "step": 32260 }, { "epoch": 205.54140127388536, "grad_norm": 1.3810524940490723, "learning_rate": 0.0001, "loss": 0.037, "step": 32270 }, { "epoch": 205.60509554140128, "grad_norm": 1.1169325113296509, "learning_rate": 0.0001, "loss": 0.0378, "step": 32280 }, { "epoch": 205.6687898089172, "grad_norm": 1.1969465017318726, "learning_rate": 0.0001, "loss": 0.0387, "step": 32290 }, { "epoch": 205.7324840764331, "grad_norm": 1.3175653219223022, "learning_rate": 0.0001, "loss": 0.0381, "step": 32300 }, { "epoch": 205.79617834394904, "grad_norm": 1.503699541091919, "learning_rate": 0.0001, "loss": 0.0352, "step": 32310 }, { "epoch": 205.85987261146497, "grad_norm": 1.124083399772644, "learning_rate": 0.0001, "loss": 0.0406, "step": 32320 }, { "epoch": 205.9235668789809, "grad_norm": 1.3274762630462646, "learning_rate": 0.0001, "loss": 0.0385, "step": 32330 }, { "epoch": 205.98726114649682, "grad_norm": 1.5753265619277954, "learning_rate": 0.0001, "loss": 0.0375, "step": 32340 }, { "epoch": 206.05095541401275, "grad_norm": 1.3607674837112427, "learning_rate": 0.0001, "loss": 0.0373, "step": 32350 }, { "epoch": 206.11464968152868, "grad_norm": 0.9501575827598572, "learning_rate": 0.0001, "loss": 0.0366, "step": 32360 }, { "epoch": 206.17834394904457, "grad_norm": 1.2678688764572144, "learning_rate": 0.0001, "loss": 0.0398, "step": 32370 }, { "epoch": 206.2420382165605, "grad_norm": 1.2368000745773315, "learning_rate": 0.0001, "loss": 0.0378, "step": 32380 }, { "epoch": 206.30573248407643, "grad_norm": 1.018023133277893, "learning_rate": 0.0001, "loss": 0.0391, "step": 32390 }, { "epoch": 206.36942675159236, "grad_norm": 1.0186299085617065, "learning_rate": 0.0001, "loss": 0.0367, "step": 32400 }, { "epoch": 206.43312101910828, "grad_norm": 1.5644053220748901, "learning_rate": 0.0001, "loss": 0.0408, "step": 32410 }, { "epoch": 206.4968152866242, "grad_norm": 1.371147871017456, "learning_rate": 0.0001, "loss": 0.0399, "step": 32420 }, { "epoch": 206.56050955414014, "grad_norm": 1.328029990196228, "learning_rate": 0.0001, "loss": 0.0422, "step": 32430 }, { "epoch": 206.62420382165604, "grad_norm": 1.3132193088531494, "learning_rate": 0.0001, "loss": 0.0401, "step": 32440 }, { "epoch": 206.68789808917197, "grad_norm": 1.1110198497772217, "learning_rate": 0.0001, "loss": 0.039, "step": 32450 }, { "epoch": 206.7515923566879, "grad_norm": 1.290074348449707, "learning_rate": 0.0001, "loss": 0.0382, "step": 32460 }, { "epoch": 206.81528662420382, "grad_norm": 1.165427565574646, "learning_rate": 0.0001, "loss": 0.0361, "step": 32470 }, { "epoch": 206.87898089171975, "grad_norm": 1.3054157495498657, "learning_rate": 0.0001, "loss": 0.0416, "step": 32480 }, { "epoch": 206.94267515923568, "grad_norm": 0.9838351607322693, "learning_rate": 0.0001, "loss": 0.0435, "step": 32490 }, { "epoch": 207.0063694267516, "grad_norm": 1.1383174657821655, "learning_rate": 0.0001, "loss": 0.0422, "step": 32500 }, { "epoch": 207.0700636942675, "grad_norm": 1.040458083152771, "learning_rate": 0.0001, "loss": 0.0408, "step": 32510 }, { "epoch": 207.13375796178343, "grad_norm": 0.9558162093162537, "learning_rate": 0.0001, "loss": 0.0419, "step": 32520 }, { "epoch": 207.19745222929936, "grad_norm": 0.9042818546295166, "learning_rate": 0.0001, "loss": 0.0413, "step": 32530 }, { "epoch": 207.2611464968153, "grad_norm": 0.9135211110115051, "learning_rate": 0.0001, "loss": 0.0364, "step": 32540 }, { "epoch": 207.3248407643312, "grad_norm": 1.2681710720062256, "learning_rate": 0.0001, "loss": 0.0393, "step": 32550 }, { "epoch": 207.38853503184714, "grad_norm": 1.285109281539917, "learning_rate": 0.0001, "loss": 0.0411, "step": 32560 }, { "epoch": 207.45222929936307, "grad_norm": 1.2297110557556152, "learning_rate": 0.0001, "loss": 0.038, "step": 32570 }, { "epoch": 207.51592356687897, "grad_norm": 1.237902283668518, "learning_rate": 0.0001, "loss": 0.039, "step": 32580 }, { "epoch": 207.5796178343949, "grad_norm": 1.250104308128357, "learning_rate": 0.0001, "loss": 0.04, "step": 32590 }, { "epoch": 207.64331210191082, "grad_norm": 1.1201218366622925, "learning_rate": 0.0001, "loss": 0.0431, "step": 32600 }, { "epoch": 207.70700636942675, "grad_norm": 1.2526764869689941, "learning_rate": 0.0001, "loss": 0.0457, "step": 32610 }, { "epoch": 207.77070063694268, "grad_norm": 1.0889972448349, "learning_rate": 0.0001, "loss": 0.0444, "step": 32620 }, { "epoch": 207.8343949044586, "grad_norm": 1.0506101846694946, "learning_rate": 0.0001, "loss": 0.0358, "step": 32630 }, { "epoch": 207.89808917197453, "grad_norm": 1.1281707286834717, "learning_rate": 0.0001, "loss": 0.0408, "step": 32640 }, { "epoch": 207.96178343949043, "grad_norm": 1.276333212852478, "learning_rate": 0.0001, "loss": 0.0404, "step": 32650 }, { "epoch": 208.02547770700636, "grad_norm": 0.8912331461906433, "learning_rate": 0.0001, "loss": 0.0373, "step": 32660 }, { "epoch": 208.0891719745223, "grad_norm": 1.108154296875, "learning_rate": 0.0001, "loss": 0.038, "step": 32670 }, { "epoch": 208.15286624203821, "grad_norm": 1.2120827436447144, "learning_rate": 0.0001, "loss": 0.0379, "step": 32680 }, { "epoch": 208.21656050955414, "grad_norm": 0.8309304714202881, "learning_rate": 0.0001, "loss": 0.0369, "step": 32690 }, { "epoch": 208.28025477707007, "grad_norm": 0.8681199550628662, "learning_rate": 0.0001, "loss": 0.0412, "step": 32700 }, { "epoch": 208.343949044586, "grad_norm": 1.2001006603240967, "learning_rate": 0.0001, "loss": 0.0382, "step": 32710 }, { "epoch": 208.40764331210192, "grad_norm": 1.1698673963546753, "learning_rate": 0.0001, "loss": 0.0399, "step": 32720 }, { "epoch": 208.47133757961782, "grad_norm": 1.0914419889450073, "learning_rate": 0.0001, "loss": 0.0427, "step": 32730 }, { "epoch": 208.53503184713375, "grad_norm": 1.062771201133728, "learning_rate": 0.0001, "loss": 0.0403, "step": 32740 }, { "epoch": 208.59872611464968, "grad_norm": 1.0985647439956665, "learning_rate": 0.0001, "loss": 0.0407, "step": 32750 }, { "epoch": 208.6624203821656, "grad_norm": 1.3695896863937378, "learning_rate": 0.0001, "loss": 0.0406, "step": 32760 }, { "epoch": 208.72611464968153, "grad_norm": 0.9608553647994995, "learning_rate": 0.0001, "loss": 0.0428, "step": 32770 }, { "epoch": 208.78980891719746, "grad_norm": 0.9660821557044983, "learning_rate": 0.0001, "loss": 0.0395, "step": 32780 }, { "epoch": 208.8535031847134, "grad_norm": 1.2507669925689697, "learning_rate": 0.0001, "loss": 0.039, "step": 32790 }, { "epoch": 208.9171974522293, "grad_norm": 1.4384294748306274, "learning_rate": 0.0001, "loss": 0.0364, "step": 32800 }, { "epoch": 208.98089171974522, "grad_norm": 1.2644312381744385, "learning_rate": 0.0001, "loss": 0.0381, "step": 32810 }, { "epoch": 209.04458598726114, "grad_norm": 1.181681513786316, "learning_rate": 0.0001, "loss": 0.0396, "step": 32820 }, { "epoch": 209.10828025477707, "grad_norm": 1.3099188804626465, "learning_rate": 0.0001, "loss": 0.0364, "step": 32830 }, { "epoch": 209.171974522293, "grad_norm": 0.9756727814674377, "learning_rate": 0.0001, "loss": 0.0377, "step": 32840 }, { "epoch": 209.23566878980893, "grad_norm": 1.4192818403244019, "learning_rate": 0.0001, "loss": 0.0375, "step": 32850 }, { "epoch": 209.29936305732485, "grad_norm": 1.5119730234146118, "learning_rate": 0.0001, "loss": 0.034, "step": 32860 }, { "epoch": 209.36305732484075, "grad_norm": 1.1037497520446777, "learning_rate": 0.0001, "loss": 0.036, "step": 32870 }, { "epoch": 209.42675159235668, "grad_norm": 1.4023284912109375, "learning_rate": 0.0001, "loss": 0.0371, "step": 32880 }, { "epoch": 209.4904458598726, "grad_norm": 1.2039812803268433, "learning_rate": 0.0001, "loss": 0.038, "step": 32890 }, { "epoch": 209.55414012738854, "grad_norm": 1.1229835748672485, "learning_rate": 0.0001, "loss": 0.0391, "step": 32900 }, { "epoch": 209.61783439490446, "grad_norm": 1.2284495830535889, "learning_rate": 0.0001, "loss": 0.0377, "step": 32910 }, { "epoch": 209.6815286624204, "grad_norm": 1.1087111234664917, "learning_rate": 0.0001, "loss": 0.0407, "step": 32920 }, { "epoch": 209.74522292993632, "grad_norm": 0.8735319972038269, "learning_rate": 0.0001, "loss": 0.0373, "step": 32930 }, { "epoch": 209.80891719745222, "grad_norm": 1.2303470373153687, "learning_rate": 0.0001, "loss": 0.0398, "step": 32940 }, { "epoch": 209.87261146496814, "grad_norm": 1.065474510192871, "learning_rate": 0.0001, "loss": 0.0437, "step": 32950 }, { "epoch": 209.93630573248407, "grad_norm": 0.9121845364570618, "learning_rate": 0.0001, "loss": 0.0402, "step": 32960 }, { "epoch": 210.0, "grad_norm": 0.9825683832168579, "learning_rate": 0.0001, "loss": 0.0359, "step": 32970 }, { "epoch": 210.06369426751593, "grad_norm": 1.178066372871399, "learning_rate": 0.0001, "loss": 0.0413, "step": 32980 }, { "epoch": 210.12738853503186, "grad_norm": 1.1145116090774536, "learning_rate": 0.0001, "loss": 0.0347, "step": 32990 }, { "epoch": 210.19108280254778, "grad_norm": 1.1834787130355835, "learning_rate": 0.0001, "loss": 0.037, "step": 33000 }, { "epoch": 210.25477707006368, "grad_norm": 1.0260658264160156, "learning_rate": 0.0001, "loss": 0.0369, "step": 33010 }, { "epoch": 210.3184713375796, "grad_norm": 1.0984398126602173, "learning_rate": 0.0001, "loss": 0.0405, "step": 33020 }, { "epoch": 210.38216560509554, "grad_norm": 1.119694471359253, "learning_rate": 0.0001, "loss": 0.0411, "step": 33030 }, { "epoch": 210.44585987261146, "grad_norm": 0.9487443566322327, "learning_rate": 0.0001, "loss": 0.0401, "step": 33040 }, { "epoch": 210.5095541401274, "grad_norm": 1.0752348899841309, "learning_rate": 0.0001, "loss": 0.0388, "step": 33050 }, { "epoch": 210.57324840764332, "grad_norm": 0.774002730846405, "learning_rate": 0.0001, "loss": 0.0391, "step": 33060 }, { "epoch": 210.63694267515925, "grad_norm": 1.0870723724365234, "learning_rate": 0.0001, "loss": 0.0369, "step": 33070 }, { "epoch": 210.70063694267515, "grad_norm": 0.9641359448432922, "learning_rate": 0.0001, "loss": 0.0391, "step": 33080 }, { "epoch": 210.76433121019107, "grad_norm": 0.8194343447685242, "learning_rate": 0.0001, "loss": 0.0416, "step": 33090 }, { "epoch": 210.828025477707, "grad_norm": 0.9527749419212341, "learning_rate": 0.0001, "loss": 0.0435, "step": 33100 }, { "epoch": 210.89171974522293, "grad_norm": 1.367424488067627, "learning_rate": 0.0001, "loss": 0.0376, "step": 33110 }, { "epoch": 210.95541401273886, "grad_norm": 0.9443729519844055, "learning_rate": 0.0001, "loss": 0.0395, "step": 33120 }, { "epoch": 211.01910828025478, "grad_norm": 1.3153634071350098, "learning_rate": 0.0001, "loss": 0.0378, "step": 33130 }, { "epoch": 211.0828025477707, "grad_norm": 1.1072232723236084, "learning_rate": 0.0001, "loss": 0.0381, "step": 33140 }, { "epoch": 211.1464968152866, "grad_norm": 1.0254931449890137, "learning_rate": 0.0001, "loss": 0.0387, "step": 33150 }, { "epoch": 211.21019108280254, "grad_norm": 1.4022209644317627, "learning_rate": 0.0001, "loss": 0.036, "step": 33160 }, { "epoch": 211.27388535031847, "grad_norm": 1.4189989566802979, "learning_rate": 0.0001, "loss": 0.0387, "step": 33170 }, { "epoch": 211.3375796178344, "grad_norm": 1.5277377367019653, "learning_rate": 0.0001, "loss": 0.0383, "step": 33180 }, { "epoch": 211.40127388535032, "grad_norm": 1.1393043994903564, "learning_rate": 0.0001, "loss": 0.0363, "step": 33190 }, { "epoch": 211.46496815286625, "grad_norm": 1.1746846437454224, "learning_rate": 0.0001, "loss": 0.0372, "step": 33200 }, { "epoch": 211.52866242038218, "grad_norm": 1.322127342224121, "learning_rate": 0.0001, "loss": 0.0393, "step": 33210 }, { "epoch": 211.59235668789808, "grad_norm": 0.9058648943901062, "learning_rate": 0.0001, "loss": 0.0383, "step": 33220 }, { "epoch": 211.656050955414, "grad_norm": 1.0282562971115112, "learning_rate": 0.0001, "loss": 0.0356, "step": 33230 }, { "epoch": 211.71974522292993, "grad_norm": 1.4246822595596313, "learning_rate": 0.0001, "loss": 0.0375, "step": 33240 }, { "epoch": 211.78343949044586, "grad_norm": 1.0727250576019287, "learning_rate": 0.0001, "loss": 0.0371, "step": 33250 }, { "epoch": 211.84713375796179, "grad_norm": 1.3461252450942993, "learning_rate": 0.0001, "loss": 0.0352, "step": 33260 }, { "epoch": 211.9108280254777, "grad_norm": 1.058933138847351, "learning_rate": 0.0001, "loss": 0.0355, "step": 33270 }, { "epoch": 211.97452229299364, "grad_norm": 1.1239421367645264, "learning_rate": 0.0001, "loss": 0.0399, "step": 33280 }, { "epoch": 212.03821656050957, "grad_norm": 1.5336681604385376, "learning_rate": 0.0001, "loss": 0.0379, "step": 33290 }, { "epoch": 212.10191082802547, "grad_norm": 1.2022324800491333, "learning_rate": 0.0001, "loss": 0.0413, "step": 33300 }, { "epoch": 212.1656050955414, "grad_norm": 1.6990982294082642, "learning_rate": 0.0001, "loss": 0.0394, "step": 33310 }, { "epoch": 212.22929936305732, "grad_norm": 1.4266780614852905, "learning_rate": 0.0001, "loss": 0.038, "step": 33320 }, { "epoch": 212.29299363057325, "grad_norm": 1.3347718715667725, "learning_rate": 0.0001, "loss": 0.0331, "step": 33330 }, { "epoch": 212.35668789808918, "grad_norm": 1.130900263786316, "learning_rate": 0.0001, "loss": 0.0378, "step": 33340 }, { "epoch": 212.4203821656051, "grad_norm": 1.1033202409744263, "learning_rate": 0.0001, "loss": 0.0365, "step": 33350 }, { "epoch": 212.48407643312103, "grad_norm": 0.9900699853897095, "learning_rate": 0.0001, "loss": 0.0368, "step": 33360 }, { "epoch": 212.54777070063693, "grad_norm": 1.3043619394302368, "learning_rate": 0.0001, "loss": 0.0371, "step": 33370 }, { "epoch": 212.61146496815286, "grad_norm": 1.0653421878814697, "learning_rate": 0.0001, "loss": 0.0373, "step": 33380 }, { "epoch": 212.6751592356688, "grad_norm": 0.8584771752357483, "learning_rate": 0.0001, "loss": 0.0385, "step": 33390 }, { "epoch": 212.7388535031847, "grad_norm": 0.8849217295646667, "learning_rate": 0.0001, "loss": 0.0399, "step": 33400 }, { "epoch": 212.80254777070064, "grad_norm": 1.078558087348938, "learning_rate": 0.0001, "loss": 0.042, "step": 33410 }, { "epoch": 212.86624203821657, "grad_norm": 1.1876811981201172, "learning_rate": 0.0001, "loss": 0.0404, "step": 33420 }, { "epoch": 212.9299363057325, "grad_norm": 1.0193474292755127, "learning_rate": 0.0001, "loss": 0.0417, "step": 33430 }, { "epoch": 212.9936305732484, "grad_norm": 1.3403781652450562, "learning_rate": 0.0001, "loss": 0.045, "step": 33440 }, { "epoch": 213.05732484076432, "grad_norm": 1.4703983068466187, "learning_rate": 0.0001, "loss": 0.042, "step": 33450 }, { "epoch": 213.12101910828025, "grad_norm": 1.1113498210906982, "learning_rate": 0.0001, "loss": 0.039, "step": 33460 }, { "epoch": 213.18471337579618, "grad_norm": 1.1375246047973633, "learning_rate": 0.0001, "loss": 0.0411, "step": 33470 }, { "epoch": 213.2484076433121, "grad_norm": 0.9994776248931885, "learning_rate": 0.0001, "loss": 0.0384, "step": 33480 }, { "epoch": 213.31210191082803, "grad_norm": 1.1728557348251343, "learning_rate": 0.0001, "loss": 0.0392, "step": 33490 }, { "epoch": 213.37579617834396, "grad_norm": 1.1076784133911133, "learning_rate": 0.0001, "loss": 0.0395, "step": 33500 }, { "epoch": 213.43949044585986, "grad_norm": 1.3529729843139648, "learning_rate": 0.0001, "loss": 0.0402, "step": 33510 }, { "epoch": 213.5031847133758, "grad_norm": 1.2467448711395264, "learning_rate": 0.0001, "loss": 0.0361, "step": 33520 }, { "epoch": 213.56687898089172, "grad_norm": 1.2322348356246948, "learning_rate": 0.0001, "loss": 0.0369, "step": 33530 }, { "epoch": 213.63057324840764, "grad_norm": 1.1937118768692017, "learning_rate": 0.0001, "loss": 0.0357, "step": 33540 }, { "epoch": 213.69426751592357, "grad_norm": 1.4286952018737793, "learning_rate": 0.0001, "loss": 0.0369, "step": 33550 }, { "epoch": 213.7579617834395, "grad_norm": 0.8971046209335327, "learning_rate": 0.0001, "loss": 0.0382, "step": 33560 }, { "epoch": 213.82165605095543, "grad_norm": 1.0921176671981812, "learning_rate": 0.0001, "loss": 0.0371, "step": 33570 }, { "epoch": 213.88535031847132, "grad_norm": 1.0268373489379883, "learning_rate": 0.0001, "loss": 0.0358, "step": 33580 }, { "epoch": 213.94904458598725, "grad_norm": 1.5148357152938843, "learning_rate": 0.0001, "loss": 0.0413, "step": 33590 }, { "epoch": 214.01273885350318, "grad_norm": 1.3581281900405884, "learning_rate": 0.0001, "loss": 0.0393, "step": 33600 }, { "epoch": 214.0764331210191, "grad_norm": 1.1861177682876587, "learning_rate": 0.0001, "loss": 0.0355, "step": 33610 }, { "epoch": 214.14012738853503, "grad_norm": 1.2243609428405762, "learning_rate": 0.0001, "loss": 0.0367, "step": 33620 }, { "epoch": 214.20382165605096, "grad_norm": 1.4785832166671753, "learning_rate": 0.0001, "loss": 0.0404, "step": 33630 }, { "epoch": 214.2675159235669, "grad_norm": 1.3964241743087769, "learning_rate": 0.0001, "loss": 0.0398, "step": 33640 }, { "epoch": 214.3312101910828, "grad_norm": 1.4281744956970215, "learning_rate": 0.0001, "loss": 0.0354, "step": 33650 }, { "epoch": 214.39490445859872, "grad_norm": 1.1424652338027954, "learning_rate": 0.0001, "loss": 0.0382, "step": 33660 }, { "epoch": 214.45859872611464, "grad_norm": 1.3166109323501587, "learning_rate": 0.0001, "loss": 0.0372, "step": 33670 }, { "epoch": 214.52229299363057, "grad_norm": 1.0088744163513184, "learning_rate": 0.0001, "loss": 0.0373, "step": 33680 }, { "epoch": 214.5859872611465, "grad_norm": 0.9313046336174011, "learning_rate": 0.0001, "loss": 0.0383, "step": 33690 }, { "epoch": 214.64968152866243, "grad_norm": 1.0359766483306885, "learning_rate": 0.0001, "loss": 0.0396, "step": 33700 }, { "epoch": 214.71337579617835, "grad_norm": 1.1151752471923828, "learning_rate": 0.0001, "loss": 0.0344, "step": 33710 }, { "epoch": 214.77707006369425, "grad_norm": 1.1460193395614624, "learning_rate": 0.0001, "loss": 0.0376, "step": 33720 }, { "epoch": 214.84076433121018, "grad_norm": 1.1246838569641113, "learning_rate": 0.0001, "loss": 0.0384, "step": 33730 }, { "epoch": 214.9044585987261, "grad_norm": 0.8511585593223572, "learning_rate": 0.0001, "loss": 0.0351, "step": 33740 }, { "epoch": 214.96815286624204, "grad_norm": 1.2852450609207153, "learning_rate": 0.0001, "loss": 0.038, "step": 33750 }, { "epoch": 215.03184713375796, "grad_norm": 1.1809024810791016, "learning_rate": 0.0001, "loss": 0.0392, "step": 33760 }, { "epoch": 215.0955414012739, "grad_norm": 1.1909950971603394, "learning_rate": 0.0001, "loss": 0.0373, "step": 33770 }, { "epoch": 215.15923566878982, "grad_norm": 1.0873844623565674, "learning_rate": 0.0001, "loss": 0.0389, "step": 33780 }, { "epoch": 215.22292993630575, "grad_norm": 0.8915159106254578, "learning_rate": 0.0001, "loss": 0.0378, "step": 33790 }, { "epoch": 215.28662420382165, "grad_norm": 1.2451962232589722, "learning_rate": 0.0001, "loss": 0.0369, "step": 33800 }, { "epoch": 215.35031847133757, "grad_norm": 1.1901905536651611, "learning_rate": 0.0001, "loss": 0.0392, "step": 33810 }, { "epoch": 215.4140127388535, "grad_norm": 0.9805779457092285, "learning_rate": 0.0001, "loss": 0.037, "step": 33820 }, { "epoch": 215.47770700636943, "grad_norm": 1.0663748979568481, "learning_rate": 0.0001, "loss": 0.0392, "step": 33830 }, { "epoch": 215.54140127388536, "grad_norm": 1.0367205142974854, "learning_rate": 0.0001, "loss": 0.0366, "step": 33840 }, { "epoch": 215.60509554140128, "grad_norm": 1.5065743923187256, "learning_rate": 0.0001, "loss": 0.0379, "step": 33850 }, { "epoch": 215.6687898089172, "grad_norm": 1.2490063905715942, "learning_rate": 0.0001, "loss": 0.0367, "step": 33860 }, { "epoch": 215.7324840764331, "grad_norm": 1.2161319255828857, "learning_rate": 0.0001, "loss": 0.0374, "step": 33870 }, { "epoch": 215.79617834394904, "grad_norm": 1.2283717393875122, "learning_rate": 0.0001, "loss": 0.0352, "step": 33880 }, { "epoch": 215.85987261146497, "grad_norm": 1.0892059803009033, "learning_rate": 0.0001, "loss": 0.0382, "step": 33890 }, { "epoch": 215.9235668789809, "grad_norm": 1.0045192241668701, "learning_rate": 0.0001, "loss": 0.0367, "step": 33900 }, { "epoch": 215.98726114649682, "grad_norm": 1.2494127750396729, "learning_rate": 0.0001, "loss": 0.0393, "step": 33910 }, { "epoch": 216.05095541401275, "grad_norm": 1.320367455482483, "learning_rate": 0.0001, "loss": 0.0374, "step": 33920 }, { "epoch": 216.11464968152868, "grad_norm": 1.3254932165145874, "learning_rate": 0.0001, "loss": 0.0348, "step": 33930 }, { "epoch": 216.17834394904457, "grad_norm": 1.3284021615982056, "learning_rate": 0.0001, "loss": 0.036, "step": 33940 }, { "epoch": 216.2420382165605, "grad_norm": 1.2583905458450317, "learning_rate": 0.0001, "loss": 0.0358, "step": 33950 }, { "epoch": 216.30573248407643, "grad_norm": 1.7136766910552979, "learning_rate": 0.0001, "loss": 0.038, "step": 33960 }, { "epoch": 216.36942675159236, "grad_norm": 1.005869746208191, "learning_rate": 0.0001, "loss": 0.0345, "step": 33970 }, { "epoch": 216.43312101910828, "grad_norm": 1.274662971496582, "learning_rate": 0.0001, "loss": 0.0338, "step": 33980 }, { "epoch": 216.4968152866242, "grad_norm": 1.1519001722335815, "learning_rate": 0.0001, "loss": 0.0416, "step": 33990 }, { "epoch": 216.56050955414014, "grad_norm": 1.2084294557571411, "learning_rate": 0.0001, "loss": 0.0373, "step": 34000 }, { "epoch": 216.62420382165604, "grad_norm": 1.329668641090393, "learning_rate": 0.0001, "loss": 0.0348, "step": 34010 }, { "epoch": 216.68789808917197, "grad_norm": 1.307163953781128, "learning_rate": 0.0001, "loss": 0.0394, "step": 34020 }, { "epoch": 216.7515923566879, "grad_norm": 1.1637675762176514, "learning_rate": 0.0001, "loss": 0.0404, "step": 34030 }, { "epoch": 216.81528662420382, "grad_norm": 1.1961065530776978, "learning_rate": 0.0001, "loss": 0.0416, "step": 34040 }, { "epoch": 216.87898089171975, "grad_norm": 1.3367812633514404, "learning_rate": 0.0001, "loss": 0.037, "step": 34050 }, { "epoch": 216.94267515923568, "grad_norm": 1.235418438911438, "learning_rate": 0.0001, "loss": 0.04, "step": 34060 }, { "epoch": 217.0063694267516, "grad_norm": 1.3440849781036377, "learning_rate": 0.0001, "loss": 0.0402, "step": 34070 }, { "epoch": 217.0700636942675, "grad_norm": 1.2553550004959106, "learning_rate": 0.0001, "loss": 0.0367, "step": 34080 }, { "epoch": 217.13375796178343, "grad_norm": 1.1368674039840698, "learning_rate": 0.0001, "loss": 0.0364, "step": 34090 }, { "epoch": 217.19745222929936, "grad_norm": 1.244038462638855, "learning_rate": 0.0001, "loss": 0.0366, "step": 34100 }, { "epoch": 217.2611464968153, "grad_norm": 1.2840453386306763, "learning_rate": 0.0001, "loss": 0.0377, "step": 34110 }, { "epoch": 217.3248407643312, "grad_norm": 1.230649471282959, "learning_rate": 0.0001, "loss": 0.0379, "step": 34120 }, { "epoch": 217.38853503184714, "grad_norm": 1.0768157243728638, "learning_rate": 0.0001, "loss": 0.0371, "step": 34130 }, { "epoch": 217.45222929936307, "grad_norm": 1.425881266593933, "learning_rate": 0.0001, "loss": 0.0379, "step": 34140 }, { "epoch": 217.51592356687897, "grad_norm": 1.3131017684936523, "learning_rate": 0.0001, "loss": 0.0385, "step": 34150 }, { "epoch": 217.5796178343949, "grad_norm": 1.0843976736068726, "learning_rate": 0.0001, "loss": 0.0366, "step": 34160 }, { "epoch": 217.64331210191082, "grad_norm": 1.2979406118392944, "learning_rate": 0.0001, "loss": 0.0375, "step": 34170 }, { "epoch": 217.70700636942675, "grad_norm": 1.306558609008789, "learning_rate": 0.0001, "loss": 0.0391, "step": 34180 }, { "epoch": 217.77070063694268, "grad_norm": 1.4235169887542725, "learning_rate": 0.0001, "loss": 0.0382, "step": 34190 }, { "epoch": 217.8343949044586, "grad_norm": 0.9449760913848877, "learning_rate": 0.0001, "loss": 0.0366, "step": 34200 }, { "epoch": 217.89808917197453, "grad_norm": 1.2448503971099854, "learning_rate": 0.0001, "loss": 0.0379, "step": 34210 }, { "epoch": 217.96178343949043, "grad_norm": 1.07317054271698, "learning_rate": 0.0001, "loss": 0.0343, "step": 34220 }, { "epoch": 218.02547770700636, "grad_norm": 1.3940385580062866, "learning_rate": 0.0001, "loss": 0.0359, "step": 34230 }, { "epoch": 218.0891719745223, "grad_norm": 1.0365091562271118, "learning_rate": 0.0001, "loss": 0.0408, "step": 34240 }, { "epoch": 218.15286624203821, "grad_norm": 1.2326017618179321, "learning_rate": 0.0001, "loss": 0.0365, "step": 34250 }, { "epoch": 218.21656050955414, "grad_norm": 1.1504554748535156, "learning_rate": 0.0001, "loss": 0.0376, "step": 34260 }, { "epoch": 218.28025477707007, "grad_norm": 1.2562566995620728, "learning_rate": 0.0001, "loss": 0.0366, "step": 34270 }, { "epoch": 218.343949044586, "grad_norm": 1.5363149642944336, "learning_rate": 0.0001, "loss": 0.0362, "step": 34280 }, { "epoch": 218.40764331210192, "grad_norm": 1.3760106563568115, "learning_rate": 0.0001, "loss": 0.036, "step": 34290 }, { "epoch": 218.47133757961782, "grad_norm": 1.1441152095794678, "learning_rate": 0.0001, "loss": 0.0409, "step": 34300 }, { "epoch": 218.53503184713375, "grad_norm": 1.1084964275360107, "learning_rate": 0.0001, "loss": 0.0329, "step": 34310 }, { "epoch": 218.59872611464968, "grad_norm": 1.2208516597747803, "learning_rate": 0.0001, "loss": 0.0352, "step": 34320 }, { "epoch": 218.6624203821656, "grad_norm": 1.2333216667175293, "learning_rate": 0.0001, "loss": 0.0358, "step": 34330 }, { "epoch": 218.72611464968153, "grad_norm": 1.1956578493118286, "learning_rate": 0.0001, "loss": 0.0384, "step": 34340 }, { "epoch": 218.78980891719746, "grad_norm": 1.2694761753082275, "learning_rate": 0.0001, "loss": 0.0383, "step": 34350 }, { "epoch": 218.8535031847134, "grad_norm": 1.6207195520401, "learning_rate": 0.0001, "loss": 0.0373, "step": 34360 }, { "epoch": 218.9171974522293, "grad_norm": 1.1080431938171387, "learning_rate": 0.0001, "loss": 0.0351, "step": 34370 }, { "epoch": 218.98089171974522, "grad_norm": 1.0287612676620483, "learning_rate": 0.0001, "loss": 0.035, "step": 34380 }, { "epoch": 219.04458598726114, "grad_norm": 1.2064188718795776, "learning_rate": 0.0001, "loss": 0.0372, "step": 34390 }, { "epoch": 219.10828025477707, "grad_norm": 1.5146559476852417, "learning_rate": 0.0001, "loss": 0.0334, "step": 34400 }, { "epoch": 219.171974522293, "grad_norm": 1.0100630521774292, "learning_rate": 0.0001, "loss": 0.0364, "step": 34410 }, { "epoch": 219.23566878980893, "grad_norm": 1.4515116214752197, "learning_rate": 0.0001, "loss": 0.0415, "step": 34420 }, { "epoch": 219.29936305732485, "grad_norm": 1.3945237398147583, "learning_rate": 0.0001, "loss": 0.0379, "step": 34430 }, { "epoch": 219.36305732484075, "grad_norm": 0.9920321702957153, "learning_rate": 0.0001, "loss": 0.0373, "step": 34440 }, { "epoch": 219.42675159235668, "grad_norm": 1.0630916357040405, "learning_rate": 0.0001, "loss": 0.0386, "step": 34450 }, { "epoch": 219.4904458598726, "grad_norm": 1.1584967374801636, "learning_rate": 0.0001, "loss": 0.0355, "step": 34460 }, { "epoch": 219.55414012738854, "grad_norm": 0.8908206820487976, "learning_rate": 0.0001, "loss": 0.0384, "step": 34470 }, { "epoch": 219.61783439490446, "grad_norm": 1.0391285419464111, "learning_rate": 0.0001, "loss": 0.0391, "step": 34480 }, { "epoch": 219.6815286624204, "grad_norm": 1.1975020170211792, "learning_rate": 0.0001, "loss": 0.0386, "step": 34490 }, { "epoch": 219.74522292993632, "grad_norm": 1.1744617223739624, "learning_rate": 0.0001, "loss": 0.0386, "step": 34500 }, { "epoch": 219.80891719745222, "grad_norm": 1.1119794845581055, "learning_rate": 0.0001, "loss": 0.0392, "step": 34510 }, { "epoch": 219.87261146496814, "grad_norm": 1.2937242984771729, "learning_rate": 0.0001, "loss": 0.039, "step": 34520 }, { "epoch": 219.93630573248407, "grad_norm": 1.3072876930236816, "learning_rate": 0.0001, "loss": 0.0339, "step": 34530 }, { "epoch": 220.0, "grad_norm": 1.7743135690689087, "learning_rate": 0.0001, "loss": 0.0398, "step": 34540 }, { "epoch": 220.06369426751593, "grad_norm": 1.5877419710159302, "learning_rate": 0.0001, "loss": 0.0384, "step": 34550 }, { "epoch": 220.12738853503186, "grad_norm": 1.6837949752807617, "learning_rate": 0.0001, "loss": 0.0404, "step": 34560 }, { "epoch": 220.19108280254778, "grad_norm": 1.304896593093872, "learning_rate": 0.0001, "loss": 0.0389, "step": 34570 }, { "epoch": 220.25477707006368, "grad_norm": 1.3834651708602905, "learning_rate": 0.0001, "loss": 0.039, "step": 34580 }, { "epoch": 220.3184713375796, "grad_norm": 1.5504786968231201, "learning_rate": 0.0001, "loss": 0.0389, "step": 34590 }, { "epoch": 220.38216560509554, "grad_norm": 1.3195146322250366, "learning_rate": 0.0001, "loss": 0.0362, "step": 34600 }, { "epoch": 220.44585987261146, "grad_norm": 1.5444369316101074, "learning_rate": 0.0001, "loss": 0.0369, "step": 34610 }, { "epoch": 220.5095541401274, "grad_norm": 1.3147720098495483, "learning_rate": 0.0001, "loss": 0.0351, "step": 34620 }, { "epoch": 220.57324840764332, "grad_norm": 1.4125890731811523, "learning_rate": 0.0001, "loss": 0.0352, "step": 34630 }, { "epoch": 220.63694267515925, "grad_norm": 1.3380968570709229, "learning_rate": 0.0001, "loss": 0.035, "step": 34640 }, { "epoch": 220.70063694267515, "grad_norm": 1.1634955406188965, "learning_rate": 0.0001, "loss": 0.0348, "step": 34650 }, { "epoch": 220.76433121019107, "grad_norm": 1.2401655912399292, "learning_rate": 0.0001, "loss": 0.0384, "step": 34660 }, { "epoch": 220.828025477707, "grad_norm": 1.1198344230651855, "learning_rate": 0.0001, "loss": 0.0372, "step": 34670 }, { "epoch": 220.89171974522293, "grad_norm": 1.1624298095703125, "learning_rate": 0.0001, "loss": 0.0396, "step": 34680 }, { "epoch": 220.95541401273886, "grad_norm": 1.0920010805130005, "learning_rate": 0.0001, "loss": 0.0373, "step": 34690 }, { "epoch": 221.01910828025478, "grad_norm": 1.4376391172409058, "learning_rate": 0.0001, "loss": 0.0401, "step": 34700 }, { "epoch": 221.0828025477707, "grad_norm": 1.435752272605896, "learning_rate": 0.0001, "loss": 0.0343, "step": 34710 }, { "epoch": 221.1464968152866, "grad_norm": 1.2081164121627808, "learning_rate": 0.0001, "loss": 0.0365, "step": 34720 }, { "epoch": 221.21019108280254, "grad_norm": 1.3214213848114014, "learning_rate": 0.0001, "loss": 0.0344, "step": 34730 }, { "epoch": 221.27388535031847, "grad_norm": 1.2877249717712402, "learning_rate": 0.0001, "loss": 0.0362, "step": 34740 }, { "epoch": 221.3375796178344, "grad_norm": 1.102467656135559, "learning_rate": 0.0001, "loss": 0.0385, "step": 34750 }, { "epoch": 221.40127388535032, "grad_norm": 1.2473591566085815, "learning_rate": 0.0001, "loss": 0.0359, "step": 34760 }, { "epoch": 221.46496815286625, "grad_norm": 0.9177541732788086, "learning_rate": 0.0001, "loss": 0.0426, "step": 34770 }, { "epoch": 221.52866242038218, "grad_norm": 1.156457543373108, "learning_rate": 0.0001, "loss": 0.0404, "step": 34780 }, { "epoch": 221.59235668789808, "grad_norm": 0.9671953320503235, "learning_rate": 0.0001, "loss": 0.0369, "step": 34790 }, { "epoch": 221.656050955414, "grad_norm": 1.2830098867416382, "learning_rate": 0.0001, "loss": 0.0409, "step": 34800 }, { "epoch": 221.71974522292993, "grad_norm": 1.078117847442627, "learning_rate": 0.0001, "loss": 0.0395, "step": 34810 }, { "epoch": 221.78343949044586, "grad_norm": 1.3273437023162842, "learning_rate": 0.0001, "loss": 0.0364, "step": 34820 }, { "epoch": 221.84713375796179, "grad_norm": 0.9872651100158691, "learning_rate": 0.0001, "loss": 0.0381, "step": 34830 }, { "epoch": 221.9108280254777, "grad_norm": 1.1984680891036987, "learning_rate": 0.0001, "loss": 0.0364, "step": 34840 }, { "epoch": 221.97452229299364, "grad_norm": 1.0591274499893188, "learning_rate": 0.0001, "loss": 0.0372, "step": 34850 }, { "epoch": 222.03821656050957, "grad_norm": 1.2881008386611938, "learning_rate": 0.0001, "loss": 0.0404, "step": 34860 }, { "epoch": 222.10191082802547, "grad_norm": 1.3484337329864502, "learning_rate": 0.0001, "loss": 0.0397, "step": 34870 }, { "epoch": 222.1656050955414, "grad_norm": 1.1573104858398438, "learning_rate": 0.0001, "loss": 0.0354, "step": 34880 }, { "epoch": 222.22929936305732, "grad_norm": 1.5683220624923706, "learning_rate": 0.0001, "loss": 0.0393, "step": 34890 }, { "epoch": 222.29299363057325, "grad_norm": 1.3881950378417969, "learning_rate": 0.0001, "loss": 0.0377, "step": 34900 }, { "epoch": 222.35668789808918, "grad_norm": 1.315365195274353, "learning_rate": 0.0001, "loss": 0.0386, "step": 34910 }, { "epoch": 222.4203821656051, "grad_norm": 1.145278811454773, "learning_rate": 0.0001, "loss": 0.0391, "step": 34920 }, { "epoch": 222.48407643312103, "grad_norm": 1.5095152854919434, "learning_rate": 0.0001, "loss": 0.0359, "step": 34930 }, { "epoch": 222.54777070063693, "grad_norm": 1.3267889022827148, "learning_rate": 0.0001, "loss": 0.0339, "step": 34940 }, { "epoch": 222.61146496815286, "grad_norm": 1.1485929489135742, "learning_rate": 0.0001, "loss": 0.0363, "step": 34950 }, { "epoch": 222.6751592356688, "grad_norm": 0.9991099834442139, "learning_rate": 0.0001, "loss": 0.037, "step": 34960 }, { "epoch": 222.7388535031847, "grad_norm": 0.9591273665428162, "learning_rate": 0.0001, "loss": 0.0369, "step": 34970 }, { "epoch": 222.80254777070064, "grad_norm": 0.9095736145973206, "learning_rate": 0.0001, "loss": 0.0373, "step": 34980 }, { "epoch": 222.86624203821657, "grad_norm": 0.9539541602134705, "learning_rate": 0.0001, "loss": 0.0397, "step": 34990 }, { "epoch": 222.9299363057325, "grad_norm": 1.1604855060577393, "learning_rate": 0.0001, "loss": 0.039, "step": 35000 }, { "epoch": 222.9936305732484, "grad_norm": 1.3614381551742554, "learning_rate": 0.0001, "loss": 0.0392, "step": 35010 }, { "epoch": 223.05732484076432, "grad_norm": 1.206765055656433, "learning_rate": 0.0001, "loss": 0.0374, "step": 35020 }, { "epoch": 223.12101910828025, "grad_norm": 1.1923637390136719, "learning_rate": 0.0001, "loss": 0.0397, "step": 35030 }, { "epoch": 223.18471337579618, "grad_norm": 1.2550679445266724, "learning_rate": 0.0001, "loss": 0.036, "step": 35040 }, { "epoch": 223.2484076433121, "grad_norm": 1.2869428396224976, "learning_rate": 0.0001, "loss": 0.039, "step": 35050 }, { "epoch": 223.31210191082803, "grad_norm": 1.5117400884628296, "learning_rate": 0.0001, "loss": 0.0363, "step": 35060 }, { "epoch": 223.37579617834396, "grad_norm": 1.2154688835144043, "learning_rate": 0.0001, "loss": 0.0395, "step": 35070 }, { "epoch": 223.43949044585986, "grad_norm": 1.3037278652191162, "learning_rate": 0.0001, "loss": 0.0356, "step": 35080 }, { "epoch": 223.5031847133758, "grad_norm": 1.2618730068206787, "learning_rate": 0.0001, "loss": 0.0397, "step": 35090 }, { "epoch": 223.56687898089172, "grad_norm": 1.024621605873108, "learning_rate": 0.0001, "loss": 0.0373, "step": 35100 }, { "epoch": 223.63057324840764, "grad_norm": 1.4472191333770752, "learning_rate": 0.0001, "loss": 0.0374, "step": 35110 }, { "epoch": 223.69426751592357, "grad_norm": 1.3284426927566528, "learning_rate": 0.0001, "loss": 0.0374, "step": 35120 }, { "epoch": 223.7579617834395, "grad_norm": 1.1153159141540527, "learning_rate": 0.0001, "loss": 0.0382, "step": 35130 }, { "epoch": 223.82165605095543, "grad_norm": 1.2116389274597168, "learning_rate": 0.0001, "loss": 0.0375, "step": 35140 }, { "epoch": 223.88535031847132, "grad_norm": 1.1020087003707886, "learning_rate": 0.0001, "loss": 0.0354, "step": 35150 }, { "epoch": 223.94904458598725, "grad_norm": 1.2529253959655762, "learning_rate": 0.0001, "loss": 0.0349, "step": 35160 }, { "epoch": 224.01273885350318, "grad_norm": 1.2428643703460693, "learning_rate": 0.0001, "loss": 0.0394, "step": 35170 }, { "epoch": 224.0764331210191, "grad_norm": 0.9337192177772522, "learning_rate": 0.0001, "loss": 0.0355, "step": 35180 }, { "epoch": 224.14012738853503, "grad_norm": 1.1117817163467407, "learning_rate": 0.0001, "loss": 0.0383, "step": 35190 }, { "epoch": 224.20382165605096, "grad_norm": 1.1084517240524292, "learning_rate": 0.0001, "loss": 0.0365, "step": 35200 }, { "epoch": 224.2675159235669, "grad_norm": 1.1531281471252441, "learning_rate": 0.0001, "loss": 0.0389, "step": 35210 }, { "epoch": 224.3312101910828, "grad_norm": 1.0448635816574097, "learning_rate": 0.0001, "loss": 0.0355, "step": 35220 }, { "epoch": 224.39490445859872, "grad_norm": 1.2953529357910156, "learning_rate": 0.0001, "loss": 0.0377, "step": 35230 }, { "epoch": 224.45859872611464, "grad_norm": 1.1253985166549683, "learning_rate": 0.0001, "loss": 0.0372, "step": 35240 }, { "epoch": 224.52229299363057, "grad_norm": 0.8725929260253906, "learning_rate": 0.0001, "loss": 0.036, "step": 35250 }, { "epoch": 224.5859872611465, "grad_norm": 1.1832205057144165, "learning_rate": 0.0001, "loss": 0.0381, "step": 35260 }, { "epoch": 224.64968152866243, "grad_norm": 0.9351279735565186, "learning_rate": 0.0001, "loss": 0.0361, "step": 35270 }, { "epoch": 224.71337579617835, "grad_norm": 1.2700871229171753, "learning_rate": 0.0001, "loss": 0.0363, "step": 35280 }, { "epoch": 224.77707006369425, "grad_norm": 1.245864987373352, "learning_rate": 0.0001, "loss": 0.0364, "step": 35290 }, { "epoch": 224.84076433121018, "grad_norm": 1.1364097595214844, "learning_rate": 0.0001, "loss": 0.0367, "step": 35300 }, { "epoch": 224.9044585987261, "grad_norm": 0.8892900943756104, "learning_rate": 0.0001, "loss": 0.0388, "step": 35310 }, { "epoch": 224.96815286624204, "grad_norm": 0.9981647729873657, "learning_rate": 0.0001, "loss": 0.0401, "step": 35320 }, { "epoch": 225.03184713375796, "grad_norm": 1.2687710523605347, "learning_rate": 0.0001, "loss": 0.0366, "step": 35330 }, { "epoch": 225.0955414012739, "grad_norm": 0.8559130430221558, "learning_rate": 0.0001, "loss": 0.0356, "step": 35340 }, { "epoch": 225.15923566878982, "grad_norm": 1.2313696146011353, "learning_rate": 0.0001, "loss": 0.0402, "step": 35350 }, { "epoch": 225.22292993630575, "grad_norm": 1.048885464668274, "learning_rate": 0.0001, "loss": 0.0382, "step": 35360 }, { "epoch": 225.28662420382165, "grad_norm": 1.2058621644973755, "learning_rate": 0.0001, "loss": 0.0373, "step": 35370 }, { "epoch": 225.35031847133757, "grad_norm": 1.2721463441848755, "learning_rate": 0.0001, "loss": 0.0367, "step": 35380 }, { "epoch": 225.4140127388535, "grad_norm": 1.2000668048858643, "learning_rate": 0.0001, "loss": 0.0367, "step": 35390 }, { "epoch": 225.47770700636943, "grad_norm": 1.076486587524414, "learning_rate": 0.0001, "loss": 0.036, "step": 35400 }, { "epoch": 225.54140127388536, "grad_norm": 1.2282081842422485, "learning_rate": 0.0001, "loss": 0.0386, "step": 35410 }, { "epoch": 225.60509554140128, "grad_norm": 1.0816006660461426, "learning_rate": 0.0001, "loss": 0.038, "step": 35420 }, { "epoch": 225.6687898089172, "grad_norm": 1.024766445159912, "learning_rate": 0.0001, "loss": 0.0379, "step": 35430 }, { "epoch": 225.7324840764331, "grad_norm": 1.3582886457443237, "learning_rate": 0.0001, "loss": 0.0375, "step": 35440 }, { "epoch": 225.79617834394904, "grad_norm": 1.011779546737671, "learning_rate": 0.0001, "loss": 0.0379, "step": 35450 }, { "epoch": 225.85987261146497, "grad_norm": 1.7880879640579224, "learning_rate": 0.0001, "loss": 0.0352, "step": 35460 }, { "epoch": 225.9235668789809, "grad_norm": 0.9835379719734192, "learning_rate": 0.0001, "loss": 0.0357, "step": 35470 }, { "epoch": 225.98726114649682, "grad_norm": 1.11488938331604, "learning_rate": 0.0001, "loss": 0.0392, "step": 35480 }, { "epoch": 226.05095541401275, "grad_norm": 1.074530839920044, "learning_rate": 0.0001, "loss": 0.0372, "step": 35490 }, { "epoch": 226.11464968152868, "grad_norm": 1.1858272552490234, "learning_rate": 0.0001, "loss": 0.0345, "step": 35500 }, { "epoch": 226.17834394904457, "grad_norm": 1.0084596872329712, "learning_rate": 0.0001, "loss": 0.0345, "step": 35510 }, { "epoch": 226.2420382165605, "grad_norm": 1.0425504446029663, "learning_rate": 0.0001, "loss": 0.0402, "step": 35520 }, { "epoch": 226.30573248407643, "grad_norm": 1.1644315719604492, "learning_rate": 0.0001, "loss": 0.0393, "step": 35530 }, { "epoch": 226.36942675159236, "grad_norm": 1.3601127862930298, "learning_rate": 0.0001, "loss": 0.0376, "step": 35540 }, { "epoch": 226.43312101910828, "grad_norm": 1.1276054382324219, "learning_rate": 0.0001, "loss": 0.0364, "step": 35550 }, { "epoch": 226.4968152866242, "grad_norm": 1.4365603923797607, "learning_rate": 0.0001, "loss": 0.0355, "step": 35560 }, { "epoch": 226.56050955414014, "grad_norm": 1.06871497631073, "learning_rate": 0.0001, "loss": 0.0383, "step": 35570 }, { "epoch": 226.62420382165604, "grad_norm": 1.2552393674850464, "learning_rate": 0.0001, "loss": 0.0375, "step": 35580 }, { "epoch": 226.68789808917197, "grad_norm": 1.1176425218582153, "learning_rate": 0.0001, "loss": 0.0377, "step": 35590 }, { "epoch": 226.7515923566879, "grad_norm": 1.2893611192703247, "learning_rate": 0.0001, "loss": 0.0403, "step": 35600 }, { "epoch": 226.81528662420382, "grad_norm": 1.2035247087478638, "learning_rate": 0.0001, "loss": 0.0374, "step": 35610 }, { "epoch": 226.87898089171975, "grad_norm": 1.2216148376464844, "learning_rate": 0.0001, "loss": 0.0366, "step": 35620 }, { "epoch": 226.94267515923568, "grad_norm": 1.1591886281967163, "learning_rate": 0.0001, "loss": 0.0381, "step": 35630 }, { "epoch": 227.0063694267516, "grad_norm": 1.1926630735397339, "learning_rate": 0.0001, "loss": 0.0366, "step": 35640 }, { "epoch": 227.0700636942675, "grad_norm": 1.4374502897262573, "learning_rate": 0.0001, "loss": 0.0413, "step": 35650 }, { "epoch": 227.13375796178343, "grad_norm": 1.235823631286621, "learning_rate": 0.0001, "loss": 0.036, "step": 35660 }, { "epoch": 227.19745222929936, "grad_norm": 1.4529318809509277, "learning_rate": 0.0001, "loss": 0.0369, "step": 35670 }, { "epoch": 227.2611464968153, "grad_norm": 1.465067744255066, "learning_rate": 0.0001, "loss": 0.0361, "step": 35680 }, { "epoch": 227.3248407643312, "grad_norm": 1.2922719717025757, "learning_rate": 0.0001, "loss": 0.0351, "step": 35690 }, { "epoch": 227.38853503184714, "grad_norm": 1.9498533010482788, "learning_rate": 0.0001, "loss": 0.038, "step": 35700 }, { "epoch": 227.45222929936307, "grad_norm": 3.4151053428649902, "learning_rate": 0.0001, "loss": 0.0428, "step": 35710 }, { "epoch": 227.51592356687897, "grad_norm": 2.242095947265625, "learning_rate": 0.0001, "loss": 0.0362, "step": 35720 }, { "epoch": 227.5796178343949, "grad_norm": 2.2206780910491943, "learning_rate": 0.0001, "loss": 0.0329, "step": 35730 }, { "epoch": 227.64331210191082, "grad_norm": 1.8054449558258057, "learning_rate": 0.0001, "loss": 0.0312, "step": 35740 }, { "epoch": 227.70700636942675, "grad_norm": 1.4532297849655151, "learning_rate": 0.0001, "loss": 0.0354, "step": 35750 }, { "epoch": 227.77070063694268, "grad_norm": 1.4806281328201294, "learning_rate": 0.0001, "loss": 0.0341, "step": 35760 }, { "epoch": 227.8343949044586, "grad_norm": 1.4027023315429688, "learning_rate": 0.0001, "loss": 0.0395, "step": 35770 }, { "epoch": 227.89808917197453, "grad_norm": 1.2566760778427124, "learning_rate": 0.0001, "loss": 0.0391, "step": 35780 }, { "epoch": 227.96178343949043, "grad_norm": 1.427185297012329, "learning_rate": 0.0001, "loss": 0.0374, "step": 35790 }, { "epoch": 228.02547770700636, "grad_norm": 1.3455698490142822, "learning_rate": 0.0001, "loss": 0.0359, "step": 35800 }, { "epoch": 228.0891719745223, "grad_norm": 1.2067991495132446, "learning_rate": 0.0001, "loss": 0.0348, "step": 35810 }, { "epoch": 228.15286624203821, "grad_norm": 1.2348952293395996, "learning_rate": 0.0001, "loss": 0.0363, "step": 35820 }, { "epoch": 228.21656050955414, "grad_norm": 1.2945250272750854, "learning_rate": 0.0001, "loss": 0.0375, "step": 35830 }, { "epoch": 228.28025477707007, "grad_norm": 1.8432481288909912, "learning_rate": 0.0001, "loss": 0.0324, "step": 35840 }, { "epoch": 228.343949044586, "grad_norm": 1.6169838905334473, "learning_rate": 0.0001, "loss": 0.0387, "step": 35850 }, { "epoch": 228.40764331210192, "grad_norm": 1.5047752857208252, "learning_rate": 0.0001, "loss": 0.0381, "step": 35860 }, { "epoch": 228.47133757961782, "grad_norm": 1.399518609046936, "learning_rate": 0.0001, "loss": 0.0372, "step": 35870 }, { "epoch": 228.53503184713375, "grad_norm": 1.3082762956619263, "learning_rate": 0.0001, "loss": 0.0331, "step": 35880 }, { "epoch": 228.59872611464968, "grad_norm": 1.1141036748886108, "learning_rate": 0.0001, "loss": 0.0334, "step": 35890 }, { "epoch": 228.6624203821656, "grad_norm": 1.2692159414291382, "learning_rate": 0.0001, "loss": 0.0364, "step": 35900 }, { "epoch": 228.72611464968153, "grad_norm": 1.411794662475586, "learning_rate": 0.0001, "loss": 0.0391, "step": 35910 }, { "epoch": 228.78980891719746, "grad_norm": 1.1993858814239502, "learning_rate": 0.0001, "loss": 0.0337, "step": 35920 }, { "epoch": 228.8535031847134, "grad_norm": 1.3169771432876587, "learning_rate": 0.0001, "loss": 0.0326, "step": 35930 }, { "epoch": 228.9171974522293, "grad_norm": 1.2478078603744507, "learning_rate": 0.0001, "loss": 0.0365, "step": 35940 }, { "epoch": 228.98089171974522, "grad_norm": 1.1061545610427856, "learning_rate": 0.0001, "loss": 0.0379, "step": 35950 }, { "epoch": 229.04458598726114, "grad_norm": 1.2344844341278076, "learning_rate": 0.0001, "loss": 0.0369, "step": 35960 }, { "epoch": 229.10828025477707, "grad_norm": 1.1509974002838135, "learning_rate": 0.0001, "loss": 0.0323, "step": 35970 }, { "epoch": 229.171974522293, "grad_norm": 0.93095463514328, "learning_rate": 0.0001, "loss": 0.0352, "step": 35980 }, { "epoch": 229.23566878980893, "grad_norm": 1.3012117147445679, "learning_rate": 0.0001, "loss": 0.0383, "step": 35990 }, { "epoch": 229.29936305732485, "grad_norm": 1.0434585809707642, "learning_rate": 0.0001, "loss": 0.038, "step": 36000 }, { "epoch": 229.36305732484075, "grad_norm": 1.5452648401260376, "learning_rate": 0.0001, "loss": 0.0379, "step": 36010 }, { "epoch": 229.42675159235668, "grad_norm": 1.4534764289855957, "learning_rate": 0.0001, "loss": 0.0379, "step": 36020 }, { "epoch": 229.4904458598726, "grad_norm": 1.4736281633377075, "learning_rate": 0.0001, "loss": 0.0392, "step": 36030 }, { "epoch": 229.55414012738854, "grad_norm": 1.2569262981414795, "learning_rate": 0.0001, "loss": 0.0354, "step": 36040 }, { "epoch": 229.61783439490446, "grad_norm": 1.2871760129928589, "learning_rate": 0.0001, "loss": 0.0356, "step": 36050 }, { "epoch": 229.6815286624204, "grad_norm": 1.1280183792114258, "learning_rate": 0.0001, "loss": 0.0343, "step": 36060 }, { "epoch": 229.74522292993632, "grad_norm": 0.9457000494003296, "learning_rate": 0.0001, "loss": 0.0368, "step": 36070 }, { "epoch": 229.80891719745222, "grad_norm": 1.392842411994934, "learning_rate": 0.0001, "loss": 0.0386, "step": 36080 }, { "epoch": 229.87261146496814, "grad_norm": 0.8799054026603699, "learning_rate": 0.0001, "loss": 0.0413, "step": 36090 }, { "epoch": 229.93630573248407, "grad_norm": 1.1883089542388916, "learning_rate": 0.0001, "loss": 0.0385, "step": 36100 }, { "epoch": 230.0, "grad_norm": 1.0176373720169067, "learning_rate": 0.0001, "loss": 0.0376, "step": 36110 }, { "epoch": 230.06369426751593, "grad_norm": 1.2252906560897827, "learning_rate": 0.0001, "loss": 0.0388, "step": 36120 }, { "epoch": 230.12738853503186, "grad_norm": 1.3187698125839233, "learning_rate": 0.0001, "loss": 0.0396, "step": 36130 }, { "epoch": 230.19108280254778, "grad_norm": 1.0445849895477295, "learning_rate": 0.0001, "loss": 0.0399, "step": 36140 }, { "epoch": 230.25477707006368, "grad_norm": 1.426711916923523, "learning_rate": 0.0001, "loss": 0.037, "step": 36150 }, { "epoch": 230.3184713375796, "grad_norm": 1.3893810510635376, "learning_rate": 0.0001, "loss": 0.0363, "step": 36160 }, { "epoch": 230.38216560509554, "grad_norm": 1.2316498756408691, "learning_rate": 0.0001, "loss": 0.0369, "step": 36170 }, { "epoch": 230.44585987261146, "grad_norm": 1.4368312358856201, "learning_rate": 0.0001, "loss": 0.0352, "step": 36180 }, { "epoch": 230.5095541401274, "grad_norm": 1.2467104196548462, "learning_rate": 0.0001, "loss": 0.0371, "step": 36190 }, { "epoch": 230.57324840764332, "grad_norm": 1.0731298923492432, "learning_rate": 0.0001, "loss": 0.0382, "step": 36200 }, { "epoch": 230.63694267515925, "grad_norm": 1.6169719696044922, "learning_rate": 0.0001, "loss": 0.0376, "step": 36210 }, { "epoch": 230.70063694267515, "grad_norm": 1.3501325845718384, "learning_rate": 0.0001, "loss": 0.0369, "step": 36220 }, { "epoch": 230.76433121019107, "grad_norm": 0.9952942728996277, "learning_rate": 0.0001, "loss": 0.0372, "step": 36230 }, { "epoch": 230.828025477707, "grad_norm": 1.0269075632095337, "learning_rate": 0.0001, "loss": 0.0363, "step": 36240 }, { "epoch": 230.89171974522293, "grad_norm": 1.1657150983810425, "learning_rate": 0.0001, "loss": 0.0379, "step": 36250 }, { "epoch": 230.95541401273886, "grad_norm": 0.7916203737258911, "learning_rate": 0.0001, "loss": 0.0395, "step": 36260 }, { "epoch": 231.01910828025478, "grad_norm": 0.7217150926589966, "learning_rate": 0.0001, "loss": 0.0355, "step": 36270 }, { "epoch": 231.0828025477707, "grad_norm": 1.2926435470581055, "learning_rate": 0.0001, "loss": 0.0367, "step": 36280 }, { "epoch": 231.1464968152866, "grad_norm": 1.4064974784851074, "learning_rate": 0.0001, "loss": 0.0414, "step": 36290 }, { "epoch": 231.21019108280254, "grad_norm": 1.3601983785629272, "learning_rate": 0.0001, "loss": 0.0372, "step": 36300 }, { "epoch": 231.27388535031847, "grad_norm": 1.0859745740890503, "learning_rate": 0.0001, "loss": 0.0394, "step": 36310 }, { "epoch": 231.3375796178344, "grad_norm": 1.0886807441711426, "learning_rate": 0.0001, "loss": 0.04, "step": 36320 }, { "epoch": 231.40127388535032, "grad_norm": 0.9648420214653015, "learning_rate": 0.0001, "loss": 0.0446, "step": 36330 }, { "epoch": 231.46496815286625, "grad_norm": 1.340224266052246, "learning_rate": 0.0001, "loss": 0.0399, "step": 36340 }, { "epoch": 231.52866242038218, "grad_norm": 1.2635457515716553, "learning_rate": 0.0001, "loss": 0.035, "step": 36350 }, { "epoch": 231.59235668789808, "grad_norm": 1.1624586582183838, "learning_rate": 0.0001, "loss": 0.0409, "step": 36360 }, { "epoch": 231.656050955414, "grad_norm": 1.0120651721954346, "learning_rate": 0.0001, "loss": 0.0415, "step": 36370 }, { "epoch": 231.71974522292993, "grad_norm": 0.9355180859565735, "learning_rate": 0.0001, "loss": 0.0375, "step": 36380 }, { "epoch": 231.78343949044586, "grad_norm": 1.0191519260406494, "learning_rate": 0.0001, "loss": 0.0411, "step": 36390 }, { "epoch": 231.84713375796179, "grad_norm": 1.0054514408111572, "learning_rate": 0.0001, "loss": 0.036, "step": 36400 }, { "epoch": 231.9108280254777, "grad_norm": 1.1956568956375122, "learning_rate": 0.0001, "loss": 0.0379, "step": 36410 }, { "epoch": 231.97452229299364, "grad_norm": 1.3460952043533325, "learning_rate": 0.0001, "loss": 0.0396, "step": 36420 }, { "epoch": 232.03821656050957, "grad_norm": 1.2391098737716675, "learning_rate": 0.0001, "loss": 0.0401, "step": 36430 }, { "epoch": 232.10191082802547, "grad_norm": 1.2776521444320679, "learning_rate": 0.0001, "loss": 0.0369, "step": 36440 }, { "epoch": 232.1656050955414, "grad_norm": 1.005661964416504, "learning_rate": 0.0001, "loss": 0.0387, "step": 36450 }, { "epoch": 232.22929936305732, "grad_norm": 0.9778310656547546, "learning_rate": 0.0001, "loss": 0.0352, "step": 36460 }, { "epoch": 232.29299363057325, "grad_norm": 1.1286321878433228, "learning_rate": 0.0001, "loss": 0.0393, "step": 36470 }, { "epoch": 232.35668789808918, "grad_norm": 1.1743730306625366, "learning_rate": 0.0001, "loss": 0.038, "step": 36480 }, { "epoch": 232.4203821656051, "grad_norm": 1.0073760747909546, "learning_rate": 0.0001, "loss": 0.0365, "step": 36490 }, { "epoch": 232.48407643312103, "grad_norm": 1.2045271396636963, "learning_rate": 0.0001, "loss": 0.0367, "step": 36500 }, { "epoch": 232.54777070063693, "grad_norm": 1.5735965967178345, "learning_rate": 0.0001, "loss": 0.037, "step": 36510 }, { "epoch": 232.61146496815286, "grad_norm": 1.3165006637573242, "learning_rate": 0.0001, "loss": 0.0381, "step": 36520 }, { "epoch": 232.6751592356688, "grad_norm": 1.3727672100067139, "learning_rate": 0.0001, "loss": 0.0377, "step": 36530 }, { "epoch": 232.7388535031847, "grad_norm": 1.0359222888946533, "learning_rate": 0.0001, "loss": 0.0366, "step": 36540 }, { "epoch": 232.80254777070064, "grad_norm": 0.9055234789848328, "learning_rate": 0.0001, "loss": 0.0344, "step": 36550 }, { "epoch": 232.86624203821657, "grad_norm": 1.018726110458374, "learning_rate": 0.0001, "loss": 0.0381, "step": 36560 }, { "epoch": 232.9299363057325, "grad_norm": 0.9724273681640625, "learning_rate": 0.0001, "loss": 0.0368, "step": 36570 }, { "epoch": 232.9936305732484, "grad_norm": 1.1195895671844482, "learning_rate": 0.0001, "loss": 0.0382, "step": 36580 }, { "epoch": 233.05732484076432, "grad_norm": 0.8693147897720337, "learning_rate": 0.0001, "loss": 0.0364, "step": 36590 }, { "epoch": 233.12101910828025, "grad_norm": 1.2441985607147217, "learning_rate": 0.0001, "loss": 0.041, "step": 36600 }, { "epoch": 233.18471337579618, "grad_norm": 1.3867590427398682, "learning_rate": 0.0001, "loss": 0.0381, "step": 36610 }, { "epoch": 233.2484076433121, "grad_norm": 1.2833584547042847, "learning_rate": 0.0001, "loss": 0.0381, "step": 36620 }, { "epoch": 233.31210191082803, "grad_norm": 1.0377310514450073, "learning_rate": 0.0001, "loss": 0.0367, "step": 36630 }, { "epoch": 233.37579617834396, "grad_norm": 1.519026517868042, "learning_rate": 0.0001, "loss": 0.0401, "step": 36640 }, { "epoch": 233.43949044585986, "grad_norm": 1.2409478425979614, "learning_rate": 0.0001, "loss": 0.0361, "step": 36650 }, { "epoch": 233.5031847133758, "grad_norm": 1.2665379047393799, "learning_rate": 0.0001, "loss": 0.0367, "step": 36660 }, { "epoch": 233.56687898089172, "grad_norm": 1.4187514781951904, "learning_rate": 0.0001, "loss": 0.0354, "step": 36670 }, { "epoch": 233.63057324840764, "grad_norm": 1.383379340171814, "learning_rate": 0.0001, "loss": 0.0372, "step": 36680 }, { "epoch": 233.69426751592357, "grad_norm": 1.315983533859253, "learning_rate": 0.0001, "loss": 0.0356, "step": 36690 }, { "epoch": 233.7579617834395, "grad_norm": 1.337899088859558, "learning_rate": 0.0001, "loss": 0.0386, "step": 36700 }, { "epoch": 233.82165605095543, "grad_norm": 1.3755278587341309, "learning_rate": 0.0001, "loss": 0.034, "step": 36710 }, { "epoch": 233.88535031847132, "grad_norm": 1.257706880569458, "learning_rate": 0.0001, "loss": 0.0344, "step": 36720 }, { "epoch": 233.94904458598725, "grad_norm": 1.2221990823745728, "learning_rate": 0.0001, "loss": 0.0407, "step": 36730 }, { "epoch": 234.01273885350318, "grad_norm": 1.1787115335464478, "learning_rate": 0.0001, "loss": 0.0375, "step": 36740 }, { "epoch": 234.0764331210191, "grad_norm": 1.309579849243164, "learning_rate": 0.0001, "loss": 0.0378, "step": 36750 }, { "epoch": 234.14012738853503, "grad_norm": 1.4614367485046387, "learning_rate": 0.0001, "loss": 0.0379, "step": 36760 }, { "epoch": 234.20382165605096, "grad_norm": 1.4778765439987183, "learning_rate": 0.0001, "loss": 0.0384, "step": 36770 }, { "epoch": 234.2675159235669, "grad_norm": 1.4658387899398804, "learning_rate": 0.0001, "loss": 0.0386, "step": 36780 }, { "epoch": 234.3312101910828, "grad_norm": 1.2378418445587158, "learning_rate": 0.0001, "loss": 0.0415, "step": 36790 }, { "epoch": 234.39490445859872, "grad_norm": 1.2588400840759277, "learning_rate": 0.0001, "loss": 0.0366, "step": 36800 }, { "epoch": 234.45859872611464, "grad_norm": 1.5963842868804932, "learning_rate": 0.0001, "loss": 0.0381, "step": 36810 }, { "epoch": 234.52229299363057, "grad_norm": 1.206799864768982, "learning_rate": 0.0001, "loss": 0.035, "step": 36820 }, { "epoch": 234.5859872611465, "grad_norm": 0.9936434626579285, "learning_rate": 0.0001, "loss": 0.0357, "step": 36830 }, { "epoch": 234.64968152866243, "grad_norm": 1.0293997526168823, "learning_rate": 0.0001, "loss": 0.0401, "step": 36840 }, { "epoch": 234.71337579617835, "grad_norm": 1.3093801736831665, "learning_rate": 0.0001, "loss": 0.0331, "step": 36850 }, { "epoch": 234.77707006369425, "grad_norm": 1.0616830587387085, "learning_rate": 0.0001, "loss": 0.0337, "step": 36860 }, { "epoch": 234.84076433121018, "grad_norm": 1.4679762125015259, "learning_rate": 0.0001, "loss": 0.0384, "step": 36870 }, { "epoch": 234.9044585987261, "grad_norm": 1.2414621114730835, "learning_rate": 0.0001, "loss": 0.0387, "step": 36880 }, { "epoch": 234.96815286624204, "grad_norm": 1.4153037071228027, "learning_rate": 0.0001, "loss": 0.0406, "step": 36890 }, { "epoch": 235.03184713375796, "grad_norm": 1.2811189889907837, "learning_rate": 0.0001, "loss": 0.0373, "step": 36900 }, { "epoch": 235.0955414012739, "grad_norm": 1.064914584159851, "learning_rate": 0.0001, "loss": 0.0361, "step": 36910 }, { "epoch": 235.15923566878982, "grad_norm": 1.1148275136947632, "learning_rate": 0.0001, "loss": 0.0367, "step": 36920 }, { "epoch": 235.22292993630575, "grad_norm": 1.0600470304489136, "learning_rate": 0.0001, "loss": 0.0399, "step": 36930 }, { "epoch": 235.28662420382165, "grad_norm": 1.2230085134506226, "learning_rate": 0.0001, "loss": 0.0345, "step": 36940 }, { "epoch": 235.35031847133757, "grad_norm": 1.3594125509262085, "learning_rate": 0.0001, "loss": 0.036, "step": 36950 }, { "epoch": 235.4140127388535, "grad_norm": 1.4233297109603882, "learning_rate": 0.0001, "loss": 0.0373, "step": 36960 }, { "epoch": 235.47770700636943, "grad_norm": 1.168590784072876, "learning_rate": 0.0001, "loss": 0.034, "step": 36970 }, { "epoch": 235.54140127388536, "grad_norm": 1.1759713888168335, "learning_rate": 0.0001, "loss": 0.0365, "step": 36980 }, { "epoch": 235.60509554140128, "grad_norm": 1.3791998624801636, "learning_rate": 0.0001, "loss": 0.0344, "step": 36990 }, { "epoch": 235.6687898089172, "grad_norm": 1.4404200315475464, "learning_rate": 0.0001, "loss": 0.0355, "step": 37000 }, { "epoch": 235.7324840764331, "grad_norm": 1.3892136812210083, "learning_rate": 0.0001, "loss": 0.0372, "step": 37010 }, { "epoch": 235.79617834394904, "grad_norm": 1.0873730182647705, "learning_rate": 0.0001, "loss": 0.036, "step": 37020 }, { "epoch": 235.85987261146497, "grad_norm": 0.8961618542671204, "learning_rate": 0.0001, "loss": 0.0364, "step": 37030 }, { "epoch": 235.9235668789809, "grad_norm": 0.9130296111106873, "learning_rate": 0.0001, "loss": 0.0367, "step": 37040 }, { "epoch": 235.98726114649682, "grad_norm": 1.2331202030181885, "learning_rate": 0.0001, "loss": 0.0346, "step": 37050 }, { "epoch": 236.05095541401275, "grad_norm": 1.201324462890625, "learning_rate": 0.0001, "loss": 0.0362, "step": 37060 }, { "epoch": 236.11464968152868, "grad_norm": 1.3701984882354736, "learning_rate": 0.0001, "loss": 0.036, "step": 37070 }, { "epoch": 236.17834394904457, "grad_norm": 1.2185769081115723, "learning_rate": 0.0001, "loss": 0.0387, "step": 37080 }, { "epoch": 236.2420382165605, "grad_norm": 1.0725207328796387, "learning_rate": 0.0001, "loss": 0.0363, "step": 37090 }, { "epoch": 236.30573248407643, "grad_norm": 1.546636939048767, "learning_rate": 0.0001, "loss": 0.0367, "step": 37100 }, { "epoch": 236.36942675159236, "grad_norm": 1.511220932006836, "learning_rate": 0.0001, "loss": 0.0377, "step": 37110 }, { "epoch": 236.43312101910828, "grad_norm": 1.0286017656326294, "learning_rate": 0.0001, "loss": 0.0359, "step": 37120 }, { "epoch": 236.4968152866242, "grad_norm": 1.197819471359253, "learning_rate": 0.0001, "loss": 0.0369, "step": 37130 }, { "epoch": 236.56050955414014, "grad_norm": 1.2227574586868286, "learning_rate": 0.0001, "loss": 0.0351, "step": 37140 }, { "epoch": 236.62420382165604, "grad_norm": 1.14378821849823, "learning_rate": 0.0001, "loss": 0.0392, "step": 37150 }, { "epoch": 236.68789808917197, "grad_norm": 1.4355815649032593, "learning_rate": 0.0001, "loss": 0.0352, "step": 37160 }, { "epoch": 236.7515923566879, "grad_norm": 1.1780827045440674, "learning_rate": 0.0001, "loss": 0.0343, "step": 37170 }, { "epoch": 236.81528662420382, "grad_norm": 0.8396813273429871, "learning_rate": 0.0001, "loss": 0.0432, "step": 37180 }, { "epoch": 236.87898089171975, "grad_norm": 1.0508098602294922, "learning_rate": 0.0001, "loss": 0.0382, "step": 37190 }, { "epoch": 236.94267515923568, "grad_norm": 1.158918857574463, "learning_rate": 0.0001, "loss": 0.0314, "step": 37200 }, { "epoch": 237.0063694267516, "grad_norm": 1.1115894317626953, "learning_rate": 0.0001, "loss": 0.0381, "step": 37210 }, { "epoch": 237.0700636942675, "grad_norm": 1.0242609977722168, "learning_rate": 0.0001, "loss": 0.0372, "step": 37220 }, { "epoch": 237.13375796178343, "grad_norm": 0.9324741959571838, "learning_rate": 0.0001, "loss": 0.0327, "step": 37230 }, { "epoch": 237.19745222929936, "grad_norm": 1.125645399093628, "learning_rate": 0.0001, "loss": 0.0366, "step": 37240 }, { "epoch": 237.2611464968153, "grad_norm": 1.1389379501342773, "learning_rate": 0.0001, "loss": 0.0379, "step": 37250 }, { "epoch": 237.3248407643312, "grad_norm": 1.0023903846740723, "learning_rate": 0.0001, "loss": 0.0381, "step": 37260 }, { "epoch": 237.38853503184714, "grad_norm": 1.2564688920974731, "learning_rate": 0.0001, "loss": 0.0359, "step": 37270 }, { "epoch": 237.45222929936307, "grad_norm": 1.309394121170044, "learning_rate": 0.0001, "loss": 0.0393, "step": 37280 }, { "epoch": 237.51592356687897, "grad_norm": 1.3125602006912231, "learning_rate": 0.0001, "loss": 0.0379, "step": 37290 }, { "epoch": 237.5796178343949, "grad_norm": 1.2738348245620728, "learning_rate": 0.0001, "loss": 0.0401, "step": 37300 }, { "epoch": 237.64331210191082, "grad_norm": 1.1472101211547852, "learning_rate": 0.0001, "loss": 0.0383, "step": 37310 }, { "epoch": 237.70700636942675, "grad_norm": 1.1225981712341309, "learning_rate": 0.0001, "loss": 0.0383, "step": 37320 }, { "epoch": 237.77070063694268, "grad_norm": 1.0716512203216553, "learning_rate": 0.0001, "loss": 0.0351, "step": 37330 }, { "epoch": 237.8343949044586, "grad_norm": 1.204351782798767, "learning_rate": 0.0001, "loss": 0.0355, "step": 37340 }, { "epoch": 237.89808917197453, "grad_norm": 1.0981342792510986, "learning_rate": 0.0001, "loss": 0.0356, "step": 37350 }, { "epoch": 237.96178343949043, "grad_norm": 1.002315878868103, "learning_rate": 0.0001, "loss": 0.0415, "step": 37360 }, { "epoch": 238.02547770700636, "grad_norm": 1.172223687171936, "learning_rate": 0.0001, "loss": 0.0397, "step": 37370 }, { "epoch": 238.0891719745223, "grad_norm": 0.9457066059112549, "learning_rate": 0.0001, "loss": 0.0385, "step": 37380 }, { "epoch": 238.15286624203821, "grad_norm": 0.8481797575950623, "learning_rate": 0.0001, "loss": 0.0375, "step": 37390 }, { "epoch": 238.21656050955414, "grad_norm": 1.0663888454437256, "learning_rate": 0.0001, "loss": 0.0403, "step": 37400 }, { "epoch": 238.28025477707007, "grad_norm": 1.115404486656189, "learning_rate": 0.0001, "loss": 0.0386, "step": 37410 }, { "epoch": 238.343949044586, "grad_norm": 1.0839614868164062, "learning_rate": 0.0001, "loss": 0.0381, "step": 37420 }, { "epoch": 238.40764331210192, "grad_norm": 0.9844870567321777, "learning_rate": 0.0001, "loss": 0.0427, "step": 37430 }, { "epoch": 238.47133757961782, "grad_norm": 1.4015437364578247, "learning_rate": 0.0001, "loss": 0.0409, "step": 37440 }, { "epoch": 238.53503184713375, "grad_norm": 1.1272248029708862, "learning_rate": 0.0001, "loss": 0.0344, "step": 37450 }, { "epoch": 238.59872611464968, "grad_norm": 0.9271261692047119, "learning_rate": 0.0001, "loss": 0.0367, "step": 37460 }, { "epoch": 238.6624203821656, "grad_norm": 1.313252568244934, "learning_rate": 0.0001, "loss": 0.0352, "step": 37470 }, { "epoch": 238.72611464968153, "grad_norm": 1.2105528116226196, "learning_rate": 0.0001, "loss": 0.0349, "step": 37480 }, { "epoch": 238.78980891719746, "grad_norm": 1.1740500926971436, "learning_rate": 0.0001, "loss": 0.0367, "step": 37490 }, { "epoch": 238.8535031847134, "grad_norm": 1.302181363105774, "learning_rate": 0.0001, "loss": 0.0402, "step": 37500 }, { "epoch": 238.9171974522293, "grad_norm": 0.8243229985237122, "learning_rate": 0.0001, "loss": 0.032, "step": 37510 }, { "epoch": 238.98089171974522, "grad_norm": 1.0522212982177734, "learning_rate": 0.0001, "loss": 0.0338, "step": 37520 }, { "epoch": 239.04458598726114, "grad_norm": 1.0405008792877197, "learning_rate": 0.0001, "loss": 0.0379, "step": 37530 }, { "epoch": 239.10828025477707, "grad_norm": 1.0616925954818726, "learning_rate": 0.0001, "loss": 0.0405, "step": 37540 }, { "epoch": 239.171974522293, "grad_norm": 1.2537585496902466, "learning_rate": 0.0001, "loss": 0.0335, "step": 37550 }, { "epoch": 239.23566878980893, "grad_norm": 1.3326773643493652, "learning_rate": 0.0001, "loss": 0.0354, "step": 37560 }, { "epoch": 239.29936305732485, "grad_norm": 1.3960000276565552, "learning_rate": 0.0001, "loss": 0.0372, "step": 37570 }, { "epoch": 239.36305732484075, "grad_norm": 1.3373116254806519, "learning_rate": 0.0001, "loss": 0.0355, "step": 37580 }, { "epoch": 239.42675159235668, "grad_norm": 1.7213186025619507, "learning_rate": 0.0001, "loss": 0.0389, "step": 37590 }, { "epoch": 239.4904458598726, "grad_norm": 1.8800990581512451, "learning_rate": 0.0001, "loss": 0.0389, "step": 37600 }, { "epoch": 239.55414012738854, "grad_norm": 1.9200228452682495, "learning_rate": 0.0001, "loss": 0.0374, "step": 37610 }, { "epoch": 239.61783439490446, "grad_norm": 2.380587577819824, "learning_rate": 0.0001, "loss": 0.0407, "step": 37620 }, { "epoch": 239.6815286624204, "grad_norm": 1.6294316053390503, "learning_rate": 0.0001, "loss": 0.0398, "step": 37630 }, { "epoch": 239.74522292993632, "grad_norm": 1.5317310094833374, "learning_rate": 0.0001, "loss": 0.0374, "step": 37640 }, { "epoch": 239.80891719745222, "grad_norm": 1.1019470691680908, "learning_rate": 0.0001, "loss": 0.0395, "step": 37650 }, { "epoch": 239.87261146496814, "grad_norm": 1.1694936752319336, "learning_rate": 0.0001, "loss": 0.0435, "step": 37660 }, { "epoch": 239.93630573248407, "grad_norm": 0.8743157386779785, "learning_rate": 0.0001, "loss": 0.0395, "step": 37670 }, { "epoch": 240.0, "grad_norm": 1.1402453184127808, "learning_rate": 0.0001, "loss": 0.0403, "step": 37680 }, { "epoch": 240.06369426751593, "grad_norm": 1.2525731325149536, "learning_rate": 0.0001, "loss": 0.039, "step": 37690 }, { "epoch": 240.12738853503186, "grad_norm": 1.2821747064590454, "learning_rate": 0.0001, "loss": 0.0421, "step": 37700 }, { "epoch": 240.19108280254778, "grad_norm": 1.2467116117477417, "learning_rate": 0.0001, "loss": 0.0398, "step": 37710 }, { "epoch": 240.25477707006368, "grad_norm": 1.3031026124954224, "learning_rate": 0.0001, "loss": 0.038, "step": 37720 }, { "epoch": 240.3184713375796, "grad_norm": 1.2696959972381592, "learning_rate": 0.0001, "loss": 0.036, "step": 37730 }, { "epoch": 240.38216560509554, "grad_norm": 1.2195409536361694, "learning_rate": 0.0001, "loss": 0.0414, "step": 37740 }, { "epoch": 240.44585987261146, "grad_norm": 1.2428081035614014, "learning_rate": 0.0001, "loss": 0.0385, "step": 37750 }, { "epoch": 240.5095541401274, "grad_norm": 1.251268982887268, "learning_rate": 0.0001, "loss": 0.0405, "step": 37760 }, { "epoch": 240.57324840764332, "grad_norm": 1.1190954446792603, "learning_rate": 0.0001, "loss": 0.0352, "step": 37770 }, { "epoch": 240.63694267515925, "grad_norm": 1.0977758169174194, "learning_rate": 0.0001, "loss": 0.0363, "step": 37780 }, { "epoch": 240.70063694267515, "grad_norm": 1.1666903495788574, "learning_rate": 0.0001, "loss": 0.0395, "step": 37790 }, { "epoch": 240.76433121019107, "grad_norm": 1.2371225357055664, "learning_rate": 0.0001, "loss": 0.039, "step": 37800 }, { "epoch": 240.828025477707, "grad_norm": 1.1898325681686401, "learning_rate": 0.0001, "loss": 0.0394, "step": 37810 }, { "epoch": 240.89171974522293, "grad_norm": 1.198706865310669, "learning_rate": 0.0001, "loss": 0.0374, "step": 37820 }, { "epoch": 240.95541401273886, "grad_norm": 1.1891937255859375, "learning_rate": 0.0001, "loss": 0.0363, "step": 37830 }, { "epoch": 241.01910828025478, "grad_norm": 1.1958959102630615, "learning_rate": 0.0001, "loss": 0.039, "step": 37840 }, { "epoch": 241.0828025477707, "grad_norm": 1.4195466041564941, "learning_rate": 0.0001, "loss": 0.0365, "step": 37850 }, { "epoch": 241.1464968152866, "grad_norm": 1.2213205099105835, "learning_rate": 0.0001, "loss": 0.0409, "step": 37860 }, { "epoch": 241.21019108280254, "grad_norm": 1.3260822296142578, "learning_rate": 0.0001, "loss": 0.0388, "step": 37870 }, { "epoch": 241.27388535031847, "grad_norm": 1.2005621194839478, "learning_rate": 0.0001, "loss": 0.0367, "step": 37880 }, { "epoch": 241.3375796178344, "grad_norm": 1.2626508474349976, "learning_rate": 0.0001, "loss": 0.0377, "step": 37890 }, { "epoch": 241.40127388535032, "grad_norm": 1.410179615020752, "learning_rate": 0.0001, "loss": 0.0342, "step": 37900 }, { "epoch": 241.46496815286625, "grad_norm": 1.317521572113037, "learning_rate": 0.0001, "loss": 0.0374, "step": 37910 }, { "epoch": 241.52866242038218, "grad_norm": 1.393587350845337, "learning_rate": 0.0001, "loss": 0.0359, "step": 37920 }, { "epoch": 241.59235668789808, "grad_norm": 1.1570240259170532, "learning_rate": 0.0001, "loss": 0.0419, "step": 37930 }, { "epoch": 241.656050955414, "grad_norm": 1.4754844903945923, "learning_rate": 0.0001, "loss": 0.0347, "step": 37940 }, { "epoch": 241.71974522292993, "grad_norm": 1.3224889039993286, "learning_rate": 0.0001, "loss": 0.0382, "step": 37950 }, { "epoch": 241.78343949044586, "grad_norm": 1.2584998607635498, "learning_rate": 0.0001, "loss": 0.0358, "step": 37960 }, { "epoch": 241.84713375796179, "grad_norm": 1.0728317499160767, "learning_rate": 0.0001, "loss": 0.0386, "step": 37970 }, { "epoch": 241.9108280254777, "grad_norm": 1.2075536251068115, "learning_rate": 0.0001, "loss": 0.0368, "step": 37980 }, { "epoch": 241.97452229299364, "grad_norm": 1.251961588859558, "learning_rate": 0.0001, "loss": 0.0374, "step": 37990 }, { "epoch": 242.03821656050957, "grad_norm": 1.2657880783081055, "learning_rate": 0.0001, "loss": 0.0417, "step": 38000 }, { "epoch": 242.10191082802547, "grad_norm": 1.2771165370941162, "learning_rate": 0.0001, "loss": 0.0359, "step": 38010 }, { "epoch": 242.1656050955414, "grad_norm": 1.3204354047775269, "learning_rate": 0.0001, "loss": 0.0402, "step": 38020 }, { "epoch": 242.22929936305732, "grad_norm": 1.258922815322876, "learning_rate": 0.0001, "loss": 0.0375, "step": 38030 }, { "epoch": 242.29299363057325, "grad_norm": 1.3939753770828247, "learning_rate": 0.0001, "loss": 0.04, "step": 38040 }, { "epoch": 242.35668789808918, "grad_norm": 1.1580673456192017, "learning_rate": 0.0001, "loss": 0.0357, "step": 38050 }, { "epoch": 242.4203821656051, "grad_norm": 1.2607835531234741, "learning_rate": 0.0001, "loss": 0.0373, "step": 38060 }, { "epoch": 242.48407643312103, "grad_norm": 1.3215267658233643, "learning_rate": 0.0001, "loss": 0.0304, "step": 38070 }, { "epoch": 242.54777070063693, "grad_norm": 1.330223798751831, "learning_rate": 0.0001, "loss": 0.0361, "step": 38080 }, { "epoch": 242.61146496815286, "grad_norm": 1.3845665454864502, "learning_rate": 0.0001, "loss": 0.0373, "step": 38090 }, { "epoch": 242.6751592356688, "grad_norm": 1.1693421602249146, "learning_rate": 0.0001, "loss": 0.0373, "step": 38100 }, { "epoch": 242.7388535031847, "grad_norm": 1.1554416418075562, "learning_rate": 0.0001, "loss": 0.0329, "step": 38110 }, { "epoch": 242.80254777070064, "grad_norm": 1.190109133720398, "learning_rate": 0.0001, "loss": 0.037, "step": 38120 }, { "epoch": 242.86624203821657, "grad_norm": 1.0397346019744873, "learning_rate": 0.0001, "loss": 0.0356, "step": 38130 }, { "epoch": 242.9299363057325, "grad_norm": 1.097022294998169, "learning_rate": 0.0001, "loss": 0.037, "step": 38140 }, { "epoch": 242.9936305732484, "grad_norm": 1.2302005290985107, "learning_rate": 0.0001, "loss": 0.0389, "step": 38150 }, { "epoch": 243.05732484076432, "grad_norm": 1.6437727212905884, "learning_rate": 0.0001, "loss": 0.0403, "step": 38160 }, { "epoch": 243.12101910828025, "grad_norm": 1.4099280834197998, "learning_rate": 0.0001, "loss": 0.0373, "step": 38170 }, { "epoch": 243.18471337579618, "grad_norm": 1.2914222478866577, "learning_rate": 0.0001, "loss": 0.0393, "step": 38180 }, { "epoch": 243.2484076433121, "grad_norm": 1.229018211364746, "learning_rate": 0.0001, "loss": 0.0377, "step": 38190 }, { "epoch": 243.31210191082803, "grad_norm": 1.2758491039276123, "learning_rate": 0.0001, "loss": 0.0367, "step": 38200 }, { "epoch": 243.37579617834396, "grad_norm": 1.3169094324111938, "learning_rate": 0.0001, "loss": 0.0341, "step": 38210 }, { "epoch": 243.43949044585986, "grad_norm": 1.3025810718536377, "learning_rate": 0.0001, "loss": 0.0347, "step": 38220 }, { "epoch": 243.5031847133758, "grad_norm": 1.1348801851272583, "learning_rate": 0.0001, "loss": 0.0386, "step": 38230 }, { "epoch": 243.56687898089172, "grad_norm": 0.9762815237045288, "learning_rate": 0.0001, "loss": 0.0418, "step": 38240 }, { "epoch": 243.63057324840764, "grad_norm": 1.352552890777588, "learning_rate": 0.0001, "loss": 0.0351, "step": 38250 }, { "epoch": 243.69426751592357, "grad_norm": 1.3132089376449585, "learning_rate": 0.0001, "loss": 0.0361, "step": 38260 }, { "epoch": 243.7579617834395, "grad_norm": 1.109779715538025, "learning_rate": 0.0001, "loss": 0.0382, "step": 38270 }, { "epoch": 243.82165605095543, "grad_norm": 1.350961446762085, "learning_rate": 0.0001, "loss": 0.0405, "step": 38280 }, { "epoch": 243.88535031847132, "grad_norm": 1.5059821605682373, "learning_rate": 0.0001, "loss": 0.0404, "step": 38290 }, { "epoch": 243.94904458598725, "grad_norm": 1.2673898935317993, "learning_rate": 0.0001, "loss": 0.0439, "step": 38300 }, { "epoch": 244.01273885350318, "grad_norm": 1.4705147743225098, "learning_rate": 0.0001, "loss": 0.035, "step": 38310 }, { "epoch": 244.0764331210191, "grad_norm": 1.3874760866165161, "learning_rate": 0.0001, "loss": 0.0346, "step": 38320 }, { "epoch": 244.14012738853503, "grad_norm": 1.5200510025024414, "learning_rate": 0.0001, "loss": 0.0373, "step": 38330 }, { "epoch": 244.20382165605096, "grad_norm": 1.479053020477295, "learning_rate": 0.0001, "loss": 0.0363, "step": 38340 }, { "epoch": 244.2675159235669, "grad_norm": 1.0688518285751343, "learning_rate": 0.0001, "loss": 0.04, "step": 38350 }, { "epoch": 244.3312101910828, "grad_norm": 0.9754378795623779, "learning_rate": 0.0001, "loss": 0.0388, "step": 38360 }, { "epoch": 244.39490445859872, "grad_norm": 1.3140039443969727, "learning_rate": 0.0001, "loss": 0.0382, "step": 38370 }, { "epoch": 244.45859872611464, "grad_norm": 1.061177372932434, "learning_rate": 0.0001, "loss": 0.0355, "step": 38380 }, { "epoch": 244.52229299363057, "grad_norm": 1.237644076347351, "learning_rate": 0.0001, "loss": 0.0403, "step": 38390 }, { "epoch": 244.5859872611465, "grad_norm": 1.3356153964996338, "learning_rate": 0.0001, "loss": 0.0426, "step": 38400 }, { "epoch": 244.64968152866243, "grad_norm": 1.3169410228729248, "learning_rate": 0.0001, "loss": 0.0376, "step": 38410 }, { "epoch": 244.71337579617835, "grad_norm": 1.2123373746871948, "learning_rate": 0.0001, "loss": 0.0399, "step": 38420 }, { "epoch": 244.77707006369425, "grad_norm": 1.3768290281295776, "learning_rate": 0.0001, "loss": 0.0365, "step": 38430 }, { "epoch": 244.84076433121018, "grad_norm": 1.4034032821655273, "learning_rate": 0.0001, "loss": 0.039, "step": 38440 }, { "epoch": 244.9044585987261, "grad_norm": 1.2410112619400024, "learning_rate": 0.0001, "loss": 0.037, "step": 38450 }, { "epoch": 244.96815286624204, "grad_norm": 1.1338787078857422, "learning_rate": 0.0001, "loss": 0.0391, "step": 38460 }, { "epoch": 245.03184713375796, "grad_norm": 0.8917012810707092, "learning_rate": 0.0001, "loss": 0.0345, "step": 38470 }, { "epoch": 245.0955414012739, "grad_norm": 1.231847882270813, "learning_rate": 0.0001, "loss": 0.032, "step": 38480 }, { "epoch": 245.15923566878982, "grad_norm": 1.0399856567382812, "learning_rate": 0.0001, "loss": 0.0366, "step": 38490 }, { "epoch": 245.22292993630575, "grad_norm": 1.319430947303772, "learning_rate": 0.0001, "loss": 0.0371, "step": 38500 }, { "epoch": 245.28662420382165, "grad_norm": 1.1136637926101685, "learning_rate": 0.0001, "loss": 0.036, "step": 38510 }, { "epoch": 245.35031847133757, "grad_norm": 1.5151187181472778, "learning_rate": 0.0001, "loss": 0.0384, "step": 38520 }, { "epoch": 245.4140127388535, "grad_norm": 1.227916955947876, "learning_rate": 0.0001, "loss": 0.0405, "step": 38530 }, { "epoch": 245.47770700636943, "grad_norm": 1.4242594242095947, "learning_rate": 0.0001, "loss": 0.0377, "step": 38540 }, { "epoch": 245.54140127388536, "grad_norm": 1.5913708209991455, "learning_rate": 0.0001, "loss": 0.0379, "step": 38550 }, { "epoch": 245.60509554140128, "grad_norm": 1.3021389245986938, "learning_rate": 0.0001, "loss": 0.0381, "step": 38560 }, { "epoch": 245.6687898089172, "grad_norm": 1.4622331857681274, "learning_rate": 0.0001, "loss": 0.037, "step": 38570 }, { "epoch": 245.7324840764331, "grad_norm": 1.4294910430908203, "learning_rate": 0.0001, "loss": 0.0372, "step": 38580 }, { "epoch": 245.79617834394904, "grad_norm": 2.0074005126953125, "learning_rate": 0.0001, "loss": 0.0345, "step": 38590 }, { "epoch": 245.85987261146497, "grad_norm": 1.562698483467102, "learning_rate": 0.0001, "loss": 0.0346, "step": 38600 }, { "epoch": 245.9235668789809, "grad_norm": 1.5355429649353027, "learning_rate": 0.0001, "loss": 0.0363, "step": 38610 }, { "epoch": 245.98726114649682, "grad_norm": 1.4150171279907227, "learning_rate": 0.0001, "loss": 0.0352, "step": 38620 }, { "epoch": 246.05095541401275, "grad_norm": 1.0679646730422974, "learning_rate": 0.0001, "loss": 0.0328, "step": 38630 }, { "epoch": 246.11464968152868, "grad_norm": 1.2016615867614746, "learning_rate": 0.0001, "loss": 0.0333, "step": 38640 }, { "epoch": 246.17834394904457, "grad_norm": 1.6751015186309814, "learning_rate": 0.0001, "loss": 0.0345, "step": 38650 }, { "epoch": 246.2420382165605, "grad_norm": 1.0370818376541138, "learning_rate": 0.0001, "loss": 0.0341, "step": 38660 }, { "epoch": 246.30573248407643, "grad_norm": 0.8928934931755066, "learning_rate": 0.0001, "loss": 0.0348, "step": 38670 }, { "epoch": 246.36942675159236, "grad_norm": 1.0263293981552124, "learning_rate": 0.0001, "loss": 0.037, "step": 38680 }, { "epoch": 246.43312101910828, "grad_norm": 1.053449273109436, "learning_rate": 0.0001, "loss": 0.0411, "step": 38690 }, { "epoch": 246.4968152866242, "grad_norm": 0.8633420467376709, "learning_rate": 0.0001, "loss": 0.0348, "step": 38700 }, { "epoch": 246.56050955414014, "grad_norm": 1.1320809125900269, "learning_rate": 0.0001, "loss": 0.0389, "step": 38710 }, { "epoch": 246.62420382165604, "grad_norm": 1.1553854942321777, "learning_rate": 0.0001, "loss": 0.0408, "step": 38720 }, { "epoch": 246.68789808917197, "grad_norm": 1.196958303451538, "learning_rate": 0.0001, "loss": 0.0389, "step": 38730 }, { "epoch": 246.7515923566879, "grad_norm": 1.1651240587234497, "learning_rate": 0.0001, "loss": 0.0371, "step": 38740 }, { "epoch": 246.81528662420382, "grad_norm": 1.195574402809143, "learning_rate": 0.0001, "loss": 0.0389, "step": 38750 }, { "epoch": 246.87898089171975, "grad_norm": 1.2917758226394653, "learning_rate": 0.0001, "loss": 0.0381, "step": 38760 }, { "epoch": 246.94267515923568, "grad_norm": 1.0362112522125244, "learning_rate": 0.0001, "loss": 0.036, "step": 38770 }, { "epoch": 247.0063694267516, "grad_norm": 1.06684410572052, "learning_rate": 0.0001, "loss": 0.0396, "step": 38780 }, { "epoch": 247.0700636942675, "grad_norm": 1.0433917045593262, "learning_rate": 0.0001, "loss": 0.0355, "step": 38790 }, { "epoch": 247.13375796178343, "grad_norm": 1.3473765850067139, "learning_rate": 0.0001, "loss": 0.0395, "step": 38800 }, { "epoch": 247.19745222929936, "grad_norm": 1.188442587852478, "learning_rate": 0.0001, "loss": 0.0384, "step": 38810 }, { "epoch": 247.2611464968153, "grad_norm": 1.3953254222869873, "learning_rate": 0.0001, "loss": 0.0354, "step": 38820 }, { "epoch": 247.3248407643312, "grad_norm": 1.0856415033340454, "learning_rate": 0.0001, "loss": 0.0399, "step": 38830 }, { "epoch": 247.38853503184714, "grad_norm": 0.9632884860038757, "learning_rate": 0.0001, "loss": 0.0357, "step": 38840 }, { "epoch": 247.45222929936307, "grad_norm": 1.4895765781402588, "learning_rate": 0.0001, "loss": 0.0392, "step": 38850 }, { "epoch": 247.51592356687897, "grad_norm": 1.4208263158798218, "learning_rate": 0.0001, "loss": 0.0372, "step": 38860 }, { "epoch": 247.5796178343949, "grad_norm": 1.0997153520584106, "learning_rate": 0.0001, "loss": 0.0362, "step": 38870 }, { "epoch": 247.64331210191082, "grad_norm": 1.3060553073883057, "learning_rate": 0.0001, "loss": 0.0335, "step": 38880 }, { "epoch": 247.70700636942675, "grad_norm": 1.1926926374435425, "learning_rate": 0.0001, "loss": 0.0372, "step": 38890 }, { "epoch": 247.77070063694268, "grad_norm": 1.0853298902511597, "learning_rate": 0.0001, "loss": 0.0384, "step": 38900 }, { "epoch": 247.8343949044586, "grad_norm": 1.2627431154251099, "learning_rate": 0.0001, "loss": 0.0376, "step": 38910 }, { "epoch": 247.89808917197453, "grad_norm": 1.0340842008590698, "learning_rate": 0.0001, "loss": 0.0365, "step": 38920 }, { "epoch": 247.96178343949043, "grad_norm": 1.1423312425613403, "learning_rate": 0.0001, "loss": 0.0373, "step": 38930 }, { "epoch": 248.02547770700636, "grad_norm": 1.1178392171859741, "learning_rate": 0.0001, "loss": 0.0367, "step": 38940 }, { "epoch": 248.0891719745223, "grad_norm": 0.8518730401992798, "learning_rate": 0.0001, "loss": 0.0344, "step": 38950 }, { "epoch": 248.15286624203821, "grad_norm": 0.9991711378097534, "learning_rate": 0.0001, "loss": 0.0375, "step": 38960 }, { "epoch": 248.21656050955414, "grad_norm": 1.0895887613296509, "learning_rate": 0.0001, "loss": 0.0377, "step": 38970 }, { "epoch": 248.28025477707007, "grad_norm": 0.8476464152336121, "learning_rate": 0.0001, "loss": 0.0339, "step": 38980 }, { "epoch": 248.343949044586, "grad_norm": 1.437153935432434, "learning_rate": 0.0001, "loss": 0.036, "step": 38990 }, { "epoch": 248.40764331210192, "grad_norm": 0.8590802550315857, "learning_rate": 0.0001, "loss": 0.0359, "step": 39000 }, { "epoch": 248.47133757961782, "grad_norm": 1.0137271881103516, "learning_rate": 0.0001, "loss": 0.0388, "step": 39010 }, { "epoch": 248.53503184713375, "grad_norm": 1.0142929553985596, "learning_rate": 0.0001, "loss": 0.0362, "step": 39020 }, { "epoch": 248.59872611464968, "grad_norm": 1.293901801109314, "learning_rate": 0.0001, "loss": 0.0372, "step": 39030 }, { "epoch": 248.6624203821656, "grad_norm": 0.8441070318222046, "learning_rate": 0.0001, "loss": 0.0388, "step": 39040 }, { "epoch": 248.72611464968153, "grad_norm": 1.160766839981079, "learning_rate": 0.0001, "loss": 0.0377, "step": 39050 }, { "epoch": 248.78980891719746, "grad_norm": 1.0360950231552124, "learning_rate": 0.0001, "loss": 0.0373, "step": 39060 }, { "epoch": 248.8535031847134, "grad_norm": 1.4813148975372314, "learning_rate": 0.0001, "loss": 0.0369, "step": 39070 }, { "epoch": 248.9171974522293, "grad_norm": 1.002224087715149, "learning_rate": 0.0001, "loss": 0.0346, "step": 39080 }, { "epoch": 248.98089171974522, "grad_norm": 1.1095322370529175, "learning_rate": 0.0001, "loss": 0.0415, "step": 39090 }, { "epoch": 249.04458598726114, "grad_norm": 0.7800636291503906, "learning_rate": 0.0001, "loss": 0.0369, "step": 39100 }, { "epoch": 249.10828025477707, "grad_norm": 1.0409716367721558, "learning_rate": 0.0001, "loss": 0.0404, "step": 39110 }, { "epoch": 249.171974522293, "grad_norm": 1.0049842596054077, "learning_rate": 0.0001, "loss": 0.0355, "step": 39120 }, { "epoch": 249.23566878980893, "grad_norm": 1.8813443183898926, "learning_rate": 0.0001, "loss": 0.0368, "step": 39130 }, { "epoch": 249.29936305732485, "grad_norm": 1.2721819877624512, "learning_rate": 0.0001, "loss": 0.0357, "step": 39140 }, { "epoch": 249.36305732484075, "grad_norm": 1.1097452640533447, "learning_rate": 0.0001, "loss": 0.0379, "step": 39150 }, { "epoch": 249.42675159235668, "grad_norm": 1.2229541540145874, "learning_rate": 0.0001, "loss": 0.0374, "step": 39160 }, { "epoch": 249.4904458598726, "grad_norm": 1.0266590118408203, "learning_rate": 0.0001, "loss": 0.0388, "step": 39170 }, { "epoch": 249.55414012738854, "grad_norm": 0.9982840418815613, "learning_rate": 0.0001, "loss": 0.0373, "step": 39180 }, { "epoch": 249.61783439490446, "grad_norm": 1.17172110080719, "learning_rate": 0.0001, "loss": 0.0342, "step": 39190 }, { "epoch": 249.6815286624204, "grad_norm": 1.3709553480148315, "learning_rate": 0.0001, "loss": 0.0356, "step": 39200 }, { "epoch": 249.74522292993632, "grad_norm": 1.0520275831222534, "learning_rate": 0.0001, "loss": 0.0366, "step": 39210 }, { "epoch": 249.80891719745222, "grad_norm": 1.1917165517807007, "learning_rate": 0.0001, "loss": 0.0359, "step": 39220 }, { "epoch": 249.87261146496814, "grad_norm": 1.36899995803833, "learning_rate": 0.0001, "loss": 0.0347, "step": 39230 }, { "epoch": 249.93630573248407, "grad_norm": 0.8348183035850525, "learning_rate": 0.0001, "loss": 0.0374, "step": 39240 }, { "epoch": 250.0, "grad_norm": 1.1149948835372925, "learning_rate": 0.0001, "loss": 0.0364, "step": 39250 }, { "epoch": 250.06369426751593, "grad_norm": 1.148521900177002, "learning_rate": 0.0001, "loss": 0.0334, "step": 39260 }, { "epoch": 250.12738853503186, "grad_norm": 1.1407711505889893, "learning_rate": 0.0001, "loss": 0.0356, "step": 39270 }, { "epoch": 250.19108280254778, "grad_norm": 0.9370103478431702, "learning_rate": 0.0001, "loss": 0.0372, "step": 39280 }, { "epoch": 250.25477707006368, "grad_norm": 1.1347886323928833, "learning_rate": 0.0001, "loss": 0.0388, "step": 39290 }, { "epoch": 250.3184713375796, "grad_norm": 1.4655194282531738, "learning_rate": 0.0001, "loss": 0.0364, "step": 39300 }, { "epoch": 250.38216560509554, "grad_norm": 1.303299069404602, "learning_rate": 0.0001, "loss": 0.0381, "step": 39310 }, { "epoch": 250.44585987261146, "grad_norm": 1.2536325454711914, "learning_rate": 0.0001, "loss": 0.0355, "step": 39320 }, { "epoch": 250.5095541401274, "grad_norm": 1.1255990266799927, "learning_rate": 0.0001, "loss": 0.0365, "step": 39330 }, { "epoch": 250.57324840764332, "grad_norm": 0.8566361665725708, "learning_rate": 0.0001, "loss": 0.0392, "step": 39340 }, { "epoch": 250.63694267515925, "grad_norm": 0.9439602494239807, "learning_rate": 0.0001, "loss": 0.0361, "step": 39350 }, { "epoch": 250.70063694267515, "grad_norm": 0.9017433524131775, "learning_rate": 0.0001, "loss": 0.0367, "step": 39360 }, { "epoch": 250.76433121019107, "grad_norm": 1.0080907344818115, "learning_rate": 0.0001, "loss": 0.0384, "step": 39370 }, { "epoch": 250.828025477707, "grad_norm": 1.0528770685195923, "learning_rate": 0.0001, "loss": 0.0366, "step": 39380 }, { "epoch": 250.89171974522293, "grad_norm": 1.3471490144729614, "learning_rate": 0.0001, "loss": 0.0391, "step": 39390 }, { "epoch": 250.95541401273886, "grad_norm": 1.070737361907959, "learning_rate": 0.0001, "loss": 0.0387, "step": 39400 }, { "epoch": 251.01910828025478, "grad_norm": 1.047183632850647, "learning_rate": 0.0001, "loss": 0.0358, "step": 39410 }, { "epoch": 251.0828025477707, "grad_norm": 1.3335965871810913, "learning_rate": 0.0001, "loss": 0.0378, "step": 39420 }, { "epoch": 251.1464968152866, "grad_norm": 1.3517805337905884, "learning_rate": 0.0001, "loss": 0.0381, "step": 39430 }, { "epoch": 251.21019108280254, "grad_norm": 0.9605425000190735, "learning_rate": 0.0001, "loss": 0.0342, "step": 39440 }, { "epoch": 251.27388535031847, "grad_norm": 1.0991289615631104, "learning_rate": 0.0001, "loss": 0.0356, "step": 39450 }, { "epoch": 251.3375796178344, "grad_norm": 1.1115819215774536, "learning_rate": 0.0001, "loss": 0.0356, "step": 39460 }, { "epoch": 251.40127388535032, "grad_norm": 1.039502501487732, "learning_rate": 0.0001, "loss": 0.035, "step": 39470 }, { "epoch": 251.46496815286625, "grad_norm": 1.0881743431091309, "learning_rate": 0.0001, "loss": 0.0351, "step": 39480 }, { "epoch": 251.52866242038218, "grad_norm": 0.9524617791175842, "learning_rate": 0.0001, "loss": 0.0396, "step": 39490 }, { "epoch": 251.59235668789808, "grad_norm": 0.8530215620994568, "learning_rate": 0.0001, "loss": 0.0358, "step": 39500 }, { "epoch": 251.656050955414, "grad_norm": 1.3539470434188843, "learning_rate": 0.0001, "loss": 0.0396, "step": 39510 }, { "epoch": 251.71974522292993, "grad_norm": 1.3242079019546509, "learning_rate": 0.0001, "loss": 0.0375, "step": 39520 }, { "epoch": 251.78343949044586, "grad_norm": 1.0728343725204468, "learning_rate": 0.0001, "loss": 0.0365, "step": 39530 }, { "epoch": 251.84713375796179, "grad_norm": 0.9900240302085876, "learning_rate": 0.0001, "loss": 0.0369, "step": 39540 }, { "epoch": 251.9108280254777, "grad_norm": 0.9332537651062012, "learning_rate": 0.0001, "loss": 0.0347, "step": 39550 }, { "epoch": 251.97452229299364, "grad_norm": 1.1535468101501465, "learning_rate": 0.0001, "loss": 0.0367, "step": 39560 }, { "epoch": 252.03821656050957, "grad_norm": 1.228428602218628, "learning_rate": 0.0001, "loss": 0.0401, "step": 39570 }, { "epoch": 252.10191082802547, "grad_norm": 1.1441463232040405, "learning_rate": 0.0001, "loss": 0.0346, "step": 39580 }, { "epoch": 252.1656050955414, "grad_norm": 1.1306768655776978, "learning_rate": 0.0001, "loss": 0.0344, "step": 39590 }, { "epoch": 252.22929936305732, "grad_norm": 0.9543322324752808, "learning_rate": 0.0001, "loss": 0.0343, "step": 39600 }, { "epoch": 252.29299363057325, "grad_norm": 0.9552854895591736, "learning_rate": 0.0001, "loss": 0.0379, "step": 39610 }, { "epoch": 252.35668789808918, "grad_norm": 1.24187433719635, "learning_rate": 0.0001, "loss": 0.0371, "step": 39620 }, { "epoch": 252.4203821656051, "grad_norm": 1.1260238885879517, "learning_rate": 0.0001, "loss": 0.0399, "step": 39630 }, { "epoch": 252.48407643312103, "grad_norm": 1.338300108909607, "learning_rate": 0.0001, "loss": 0.0375, "step": 39640 }, { "epoch": 252.54777070063693, "grad_norm": 0.9027591943740845, "learning_rate": 0.0001, "loss": 0.0399, "step": 39650 }, { "epoch": 252.61146496815286, "grad_norm": 1.160433053970337, "learning_rate": 0.0001, "loss": 0.0343, "step": 39660 }, { "epoch": 252.6751592356688, "grad_norm": 0.968633770942688, "learning_rate": 0.0001, "loss": 0.0357, "step": 39670 }, { "epoch": 252.7388535031847, "grad_norm": 1.3513246774673462, "learning_rate": 0.0001, "loss": 0.0376, "step": 39680 }, { "epoch": 252.80254777070064, "grad_norm": 1.555954933166504, "learning_rate": 0.0001, "loss": 0.0364, "step": 39690 }, { "epoch": 252.86624203821657, "grad_norm": 1.176641583442688, "learning_rate": 0.0001, "loss": 0.0349, "step": 39700 }, { "epoch": 252.9299363057325, "grad_norm": 1.5806512832641602, "learning_rate": 0.0001, "loss": 0.0325, "step": 39710 }, { "epoch": 252.9936305732484, "grad_norm": 1.2016791105270386, "learning_rate": 0.0001, "loss": 0.037, "step": 39720 }, { "epoch": 253.05732484076432, "grad_norm": 1.0624375343322754, "learning_rate": 0.0001, "loss": 0.0342, "step": 39730 }, { "epoch": 253.12101910828025, "grad_norm": 1.484440565109253, "learning_rate": 0.0001, "loss": 0.0386, "step": 39740 }, { "epoch": 253.18471337579618, "grad_norm": 1.3611104488372803, "learning_rate": 0.0001, "loss": 0.035, "step": 39750 }, { "epoch": 253.2484076433121, "grad_norm": 1.531968355178833, "learning_rate": 0.0001, "loss": 0.0341, "step": 39760 }, { "epoch": 253.31210191082803, "grad_norm": 1.1460241079330444, "learning_rate": 0.0001, "loss": 0.0343, "step": 39770 }, { "epoch": 253.37579617834396, "grad_norm": 1.0233923196792603, "learning_rate": 0.0001, "loss": 0.0413, "step": 39780 }, { "epoch": 253.43949044585986, "grad_norm": 1.2973779439926147, "learning_rate": 0.0001, "loss": 0.0375, "step": 39790 }, { "epoch": 253.5031847133758, "grad_norm": 0.8861666321754456, "learning_rate": 0.0001, "loss": 0.036, "step": 39800 }, { "epoch": 253.56687898089172, "grad_norm": 1.4456570148468018, "learning_rate": 0.0001, "loss": 0.0372, "step": 39810 }, { "epoch": 253.63057324840764, "grad_norm": 1.2855803966522217, "learning_rate": 0.0001, "loss": 0.0357, "step": 39820 }, { "epoch": 253.69426751592357, "grad_norm": 1.297797679901123, "learning_rate": 0.0001, "loss": 0.037, "step": 39830 }, { "epoch": 253.7579617834395, "grad_norm": 1.2935699224472046, "learning_rate": 0.0001, "loss": 0.0396, "step": 39840 }, { "epoch": 253.82165605095543, "grad_norm": 1.258784294128418, "learning_rate": 0.0001, "loss": 0.0349, "step": 39850 }, { "epoch": 253.88535031847132, "grad_norm": 1.2249183654785156, "learning_rate": 0.0001, "loss": 0.0384, "step": 39860 }, { "epoch": 253.94904458598725, "grad_norm": 1.298838496208191, "learning_rate": 0.0001, "loss": 0.0332, "step": 39870 }, { "epoch": 254.01273885350318, "grad_norm": 1.1641769409179688, "learning_rate": 0.0001, "loss": 0.0347, "step": 39880 }, { "epoch": 254.0764331210191, "grad_norm": 1.2098495960235596, "learning_rate": 0.0001, "loss": 0.0348, "step": 39890 }, { "epoch": 254.14012738853503, "grad_norm": 1.61612069606781, "learning_rate": 0.0001, "loss": 0.0357, "step": 39900 }, { "epoch": 254.20382165605096, "grad_norm": 1.3464027643203735, "learning_rate": 0.0001, "loss": 0.0353, "step": 39910 }, { "epoch": 254.2675159235669, "grad_norm": 1.2497451305389404, "learning_rate": 0.0001, "loss": 0.0355, "step": 39920 }, { "epoch": 254.3312101910828, "grad_norm": 1.3347822427749634, "learning_rate": 0.0001, "loss": 0.0361, "step": 39930 }, { "epoch": 254.39490445859872, "grad_norm": 1.2438023090362549, "learning_rate": 0.0001, "loss": 0.0317, "step": 39940 }, { "epoch": 254.45859872611464, "grad_norm": 0.9288575053215027, "learning_rate": 0.0001, "loss": 0.0358, "step": 39950 }, { "epoch": 254.52229299363057, "grad_norm": 1.1724246740341187, "learning_rate": 0.0001, "loss": 0.0351, "step": 39960 }, { "epoch": 254.5859872611465, "grad_norm": 1.2542874813079834, "learning_rate": 0.0001, "loss": 0.035, "step": 39970 }, { "epoch": 254.64968152866243, "grad_norm": 1.2109510898590088, "learning_rate": 0.0001, "loss": 0.0353, "step": 39980 }, { "epoch": 254.71337579617835, "grad_norm": 1.1813298463821411, "learning_rate": 0.0001, "loss": 0.0364, "step": 39990 }, { "epoch": 254.77707006369425, "grad_norm": 1.0539896488189697, "learning_rate": 0.0001, "loss": 0.0328, "step": 40000 }, { "epoch": 254.84076433121018, "grad_norm": 1.032413363456726, "learning_rate": 0.0001, "loss": 0.0414, "step": 40010 }, { "epoch": 254.9044585987261, "grad_norm": 0.8608676195144653, "learning_rate": 0.0001, "loss": 0.0357, "step": 40020 }, { "epoch": 254.96815286624204, "grad_norm": 1.0754095315933228, "learning_rate": 0.0001, "loss": 0.0345, "step": 40030 }, { "epoch": 255.03184713375796, "grad_norm": 0.9391030669212341, "learning_rate": 0.0001, "loss": 0.0382, "step": 40040 }, { "epoch": 255.0955414012739, "grad_norm": 0.9750571250915527, "learning_rate": 0.0001, "loss": 0.0357, "step": 40050 }, { "epoch": 255.15923566878982, "grad_norm": 1.0396004915237427, "learning_rate": 0.0001, "loss": 0.038, "step": 40060 }, { "epoch": 255.22292993630575, "grad_norm": 0.9939588904380798, "learning_rate": 0.0001, "loss": 0.0372, "step": 40070 }, { "epoch": 255.28662420382165, "grad_norm": 1.140607476234436, "learning_rate": 0.0001, "loss": 0.0351, "step": 40080 }, { "epoch": 255.35031847133757, "grad_norm": 1.0620851516723633, "learning_rate": 0.0001, "loss": 0.037, "step": 40090 }, { "epoch": 255.4140127388535, "grad_norm": 1.1248798370361328, "learning_rate": 0.0001, "loss": 0.0373, "step": 40100 }, { "epoch": 255.47770700636943, "grad_norm": 1.2419238090515137, "learning_rate": 0.0001, "loss": 0.034, "step": 40110 }, { "epoch": 255.54140127388536, "grad_norm": 1.2845141887664795, "learning_rate": 0.0001, "loss": 0.0371, "step": 40120 }, { "epoch": 255.60509554140128, "grad_norm": 1.1156795024871826, "learning_rate": 0.0001, "loss": 0.0342, "step": 40130 }, { "epoch": 255.6687898089172, "grad_norm": 1.0549962520599365, "learning_rate": 0.0001, "loss": 0.0353, "step": 40140 }, { "epoch": 255.7324840764331, "grad_norm": 1.1968787908554077, "learning_rate": 0.0001, "loss": 0.0367, "step": 40150 }, { "epoch": 255.79617834394904, "grad_norm": 1.343434453010559, "learning_rate": 0.0001, "loss": 0.0345, "step": 40160 }, { "epoch": 255.85987261146497, "grad_norm": 1.4189624786376953, "learning_rate": 0.0001, "loss": 0.034, "step": 40170 }, { "epoch": 255.9235668789809, "grad_norm": 1.1227983236312866, "learning_rate": 0.0001, "loss": 0.0355, "step": 40180 }, { "epoch": 255.98726114649682, "grad_norm": 1.061169147491455, "learning_rate": 0.0001, "loss": 0.0365, "step": 40190 }, { "epoch": 256.0509554140127, "grad_norm": 1.352393388748169, "learning_rate": 0.0001, "loss": 0.0363, "step": 40200 }, { "epoch": 256.11464968152865, "grad_norm": 1.4288982152938843, "learning_rate": 0.0001, "loss": 0.0403, "step": 40210 }, { "epoch": 256.1783439490446, "grad_norm": 1.1119682788848877, "learning_rate": 0.0001, "loss": 0.0351, "step": 40220 }, { "epoch": 256.2420382165605, "grad_norm": 1.4129903316497803, "learning_rate": 0.0001, "loss": 0.0385, "step": 40230 }, { "epoch": 256.30573248407643, "grad_norm": 1.4923968315124512, "learning_rate": 0.0001, "loss": 0.0376, "step": 40240 }, { "epoch": 256.36942675159236, "grad_norm": 1.771908164024353, "learning_rate": 0.0001, "loss": 0.0379, "step": 40250 }, { "epoch": 256.4331210191083, "grad_norm": 1.320591926574707, "learning_rate": 0.0001, "loss": 0.0366, "step": 40260 }, { "epoch": 256.4968152866242, "grad_norm": 1.2099783420562744, "learning_rate": 0.0001, "loss": 0.0341, "step": 40270 }, { "epoch": 256.56050955414014, "grad_norm": 1.4712110757827759, "learning_rate": 0.0001, "loss": 0.0337, "step": 40280 }, { "epoch": 256.62420382165607, "grad_norm": 1.3025492429733276, "learning_rate": 0.0001, "loss": 0.0343, "step": 40290 }, { "epoch": 256.687898089172, "grad_norm": 1.0235086679458618, "learning_rate": 0.0001, "loss": 0.0341, "step": 40300 }, { "epoch": 256.7515923566879, "grad_norm": 1.1006501913070679, "learning_rate": 0.0001, "loss": 0.0333, "step": 40310 }, { "epoch": 256.81528662420385, "grad_norm": 1.2582913637161255, "learning_rate": 0.0001, "loss": 0.0408, "step": 40320 }, { "epoch": 256.8789808917197, "grad_norm": 1.2466845512390137, "learning_rate": 0.0001, "loss": 0.0364, "step": 40330 }, { "epoch": 256.94267515923565, "grad_norm": 1.1319836378097534, "learning_rate": 0.0001, "loss": 0.0356, "step": 40340 }, { "epoch": 257.0063694267516, "grad_norm": 1.6147215366363525, "learning_rate": 0.0001, "loss": 0.0406, "step": 40350 }, { "epoch": 257.0700636942675, "grad_norm": 1.4422791004180908, "learning_rate": 0.0001, "loss": 0.0332, "step": 40360 }, { "epoch": 257.13375796178343, "grad_norm": 1.259582281112671, "learning_rate": 0.0001, "loss": 0.0349, "step": 40370 }, { "epoch": 257.19745222929936, "grad_norm": 1.529337763786316, "learning_rate": 0.0001, "loss": 0.0346, "step": 40380 }, { "epoch": 257.2611464968153, "grad_norm": 1.4069291353225708, "learning_rate": 0.0001, "loss": 0.0347, "step": 40390 }, { "epoch": 257.3248407643312, "grad_norm": 1.214743971824646, "learning_rate": 0.0001, "loss": 0.0395, "step": 40400 }, { "epoch": 257.38853503184714, "grad_norm": 1.3197388648986816, "learning_rate": 0.0001, "loss": 0.0337, "step": 40410 }, { "epoch": 257.45222929936307, "grad_norm": 1.254143476486206, "learning_rate": 0.0001, "loss": 0.0364, "step": 40420 }, { "epoch": 257.515923566879, "grad_norm": 0.9803391098976135, "learning_rate": 0.0001, "loss": 0.0395, "step": 40430 }, { "epoch": 257.5796178343949, "grad_norm": 0.8937922716140747, "learning_rate": 0.0001, "loss": 0.0362, "step": 40440 }, { "epoch": 257.64331210191085, "grad_norm": 1.1459063291549683, "learning_rate": 0.0001, "loss": 0.0353, "step": 40450 }, { "epoch": 257.7070063694268, "grad_norm": 0.9591116309165955, "learning_rate": 0.0001, "loss": 0.0361, "step": 40460 }, { "epoch": 257.77070063694265, "grad_norm": 1.2572208642959595, "learning_rate": 0.0001, "loss": 0.0369, "step": 40470 }, { "epoch": 257.8343949044586, "grad_norm": 1.3055620193481445, "learning_rate": 0.0001, "loss": 0.0361, "step": 40480 }, { "epoch": 257.8980891719745, "grad_norm": 1.0289416313171387, "learning_rate": 0.0001, "loss": 0.0373, "step": 40490 }, { "epoch": 257.96178343949043, "grad_norm": 1.315490961074829, "learning_rate": 0.0001, "loss": 0.0383, "step": 40500 }, { "epoch": 258.02547770700636, "grad_norm": 1.4568434953689575, "learning_rate": 0.0001, "loss": 0.0391, "step": 40510 }, { "epoch": 258.0891719745223, "grad_norm": 1.6988950967788696, "learning_rate": 0.0001, "loss": 0.0366, "step": 40520 }, { "epoch": 258.1528662420382, "grad_norm": 1.4318190813064575, "learning_rate": 0.0001, "loss": 0.0375, "step": 40530 }, { "epoch": 258.21656050955414, "grad_norm": 1.4592267274856567, "learning_rate": 0.0001, "loss": 0.0398, "step": 40540 }, { "epoch": 258.28025477707007, "grad_norm": 1.4465161561965942, "learning_rate": 0.0001, "loss": 0.0371, "step": 40550 }, { "epoch": 258.343949044586, "grad_norm": 1.438842535018921, "learning_rate": 0.0001, "loss": 0.0337, "step": 40560 }, { "epoch": 258.4076433121019, "grad_norm": 1.1985069513320923, "learning_rate": 0.0001, "loss": 0.0334, "step": 40570 }, { "epoch": 258.47133757961785, "grad_norm": 1.1427385807037354, "learning_rate": 0.0001, "loss": 0.0361, "step": 40580 }, { "epoch": 258.5350318471338, "grad_norm": 1.2578494548797607, "learning_rate": 0.0001, "loss": 0.0346, "step": 40590 }, { "epoch": 258.5987261146497, "grad_norm": 0.9351394772529602, "learning_rate": 0.0001, "loss": 0.0354, "step": 40600 }, { "epoch": 258.6624203821656, "grad_norm": 1.2462968826293945, "learning_rate": 0.0001, "loss": 0.0387, "step": 40610 }, { "epoch": 258.7261146496815, "grad_norm": 1.458530068397522, "learning_rate": 0.0001, "loss": 0.0342, "step": 40620 }, { "epoch": 258.78980891719743, "grad_norm": 1.1030333042144775, "learning_rate": 0.0001, "loss": 0.0343, "step": 40630 }, { "epoch": 258.85350318471336, "grad_norm": 0.9754841327667236, "learning_rate": 0.0001, "loss": 0.0337, "step": 40640 }, { "epoch": 258.9171974522293, "grad_norm": 1.1921379566192627, "learning_rate": 0.0001, "loss": 0.0346, "step": 40650 }, { "epoch": 258.9808917197452, "grad_norm": 1.0579555034637451, "learning_rate": 0.0001, "loss": 0.0356, "step": 40660 }, { "epoch": 259.04458598726114, "grad_norm": 1.2193256616592407, "learning_rate": 0.0001, "loss": 0.0364, "step": 40670 }, { "epoch": 259.10828025477707, "grad_norm": 1.0966907739639282, "learning_rate": 0.0001, "loss": 0.0343, "step": 40680 }, { "epoch": 259.171974522293, "grad_norm": 0.9206891059875488, "learning_rate": 0.0001, "loss": 0.0342, "step": 40690 }, { "epoch": 259.2356687898089, "grad_norm": 1.0358755588531494, "learning_rate": 0.0001, "loss": 0.0354, "step": 40700 }, { "epoch": 259.29936305732485, "grad_norm": 1.3031980991363525, "learning_rate": 0.0001, "loss": 0.0359, "step": 40710 }, { "epoch": 259.3630573248408, "grad_norm": 1.3820966482162476, "learning_rate": 0.0001, "loss": 0.0361, "step": 40720 }, { "epoch": 259.4267515923567, "grad_norm": 0.831355631351471, "learning_rate": 0.0001, "loss": 0.0356, "step": 40730 }, { "epoch": 259.49044585987264, "grad_norm": 1.394894003868103, "learning_rate": 0.0001, "loss": 0.0358, "step": 40740 }, { "epoch": 259.5541401273885, "grad_norm": 1.1588490009307861, "learning_rate": 0.0001, "loss": 0.039, "step": 40750 }, { "epoch": 259.61783439490443, "grad_norm": 1.060334324836731, "learning_rate": 0.0001, "loss": 0.0371, "step": 40760 }, { "epoch": 259.68152866242036, "grad_norm": 1.085038661956787, "learning_rate": 0.0001, "loss": 0.0361, "step": 40770 }, { "epoch": 259.7452229299363, "grad_norm": 1.2378358840942383, "learning_rate": 0.0001, "loss": 0.0341, "step": 40780 }, { "epoch": 259.8089171974522, "grad_norm": 1.0876519680023193, "learning_rate": 0.0001, "loss": 0.037, "step": 40790 }, { "epoch": 259.87261146496814, "grad_norm": 1.0899673700332642, "learning_rate": 0.0001, "loss": 0.0376, "step": 40800 }, { "epoch": 259.9363057324841, "grad_norm": 0.8612358570098877, "learning_rate": 0.0001, "loss": 0.0381, "step": 40810 }, { "epoch": 260.0, "grad_norm": 1.2971384525299072, "learning_rate": 0.0001, "loss": 0.0377, "step": 40820 }, { "epoch": 260.0636942675159, "grad_norm": 1.4082956314086914, "learning_rate": 0.0001, "loss": 0.0352, "step": 40830 }, { "epoch": 260.12738853503186, "grad_norm": 1.6485306024551392, "learning_rate": 0.0001, "loss": 0.034, "step": 40840 }, { "epoch": 260.1910828025478, "grad_norm": 1.1781524419784546, "learning_rate": 0.0001, "loss": 0.0354, "step": 40850 }, { "epoch": 260.2547770700637, "grad_norm": 1.5280039310455322, "learning_rate": 0.0001, "loss": 0.0349, "step": 40860 }, { "epoch": 260.31847133757964, "grad_norm": 1.1339553594589233, "learning_rate": 0.0001, "loss": 0.0384, "step": 40870 }, { "epoch": 260.38216560509557, "grad_norm": 1.0624908208847046, "learning_rate": 0.0001, "loss": 0.0355, "step": 40880 }, { "epoch": 260.4458598726115, "grad_norm": 1.0587377548217773, "learning_rate": 0.0001, "loss": 0.0363, "step": 40890 }, { "epoch": 260.50955414012736, "grad_norm": 1.449224829673767, "learning_rate": 0.0001, "loss": 0.0365, "step": 40900 }, { "epoch": 260.5732484076433, "grad_norm": 1.2189059257507324, "learning_rate": 0.0001, "loss": 0.0367, "step": 40910 }, { "epoch": 260.6369426751592, "grad_norm": 1.1664115190505981, "learning_rate": 0.0001, "loss": 0.0361, "step": 40920 }, { "epoch": 260.70063694267515, "grad_norm": 1.104088544845581, "learning_rate": 0.0001, "loss": 0.0357, "step": 40930 }, { "epoch": 260.7643312101911, "grad_norm": 1.3404933214187622, "learning_rate": 0.0001, "loss": 0.0375, "step": 40940 }, { "epoch": 260.828025477707, "grad_norm": 1.077052354812622, "learning_rate": 0.0001, "loss": 0.0328, "step": 40950 }, { "epoch": 260.89171974522293, "grad_norm": 1.1205425262451172, "learning_rate": 0.0001, "loss": 0.0389, "step": 40960 }, { "epoch": 260.95541401273886, "grad_norm": 1.2096598148345947, "learning_rate": 0.0001, "loss": 0.0385, "step": 40970 }, { "epoch": 261.0191082802548, "grad_norm": 0.8393043875694275, "learning_rate": 0.0001, "loss": 0.0359, "step": 40980 }, { "epoch": 261.0828025477707, "grad_norm": 1.1736856698989868, "learning_rate": 0.0001, "loss": 0.0386, "step": 40990 }, { "epoch": 261.14649681528664, "grad_norm": 1.0650978088378906, "learning_rate": 0.0001, "loss": 0.033, "step": 41000 }, { "epoch": 261.21019108280257, "grad_norm": 1.0244630575180054, "learning_rate": 0.0001, "loss": 0.0381, "step": 41010 }, { "epoch": 261.2738853503185, "grad_norm": 1.0137009620666504, "learning_rate": 0.0001, "loss": 0.0352, "step": 41020 }, { "epoch": 261.3375796178344, "grad_norm": 1.2236757278442383, "learning_rate": 0.0001, "loss": 0.0346, "step": 41030 }, { "epoch": 261.4012738853503, "grad_norm": 1.081122875213623, "learning_rate": 0.0001, "loss": 0.038, "step": 41040 }, { "epoch": 261.4649681528662, "grad_norm": 1.1872209310531616, "learning_rate": 0.0001, "loss": 0.0387, "step": 41050 }, { "epoch": 261.52866242038215, "grad_norm": 1.3290133476257324, "learning_rate": 0.0001, "loss": 0.0345, "step": 41060 }, { "epoch": 261.5923566878981, "grad_norm": 1.3208553791046143, "learning_rate": 0.0001, "loss": 0.0375, "step": 41070 }, { "epoch": 261.656050955414, "grad_norm": 1.0892447233200073, "learning_rate": 0.0001, "loss": 0.0349, "step": 41080 }, { "epoch": 261.71974522292993, "grad_norm": 1.4140607118606567, "learning_rate": 0.0001, "loss": 0.0377, "step": 41090 }, { "epoch": 261.78343949044586, "grad_norm": 1.130845069885254, "learning_rate": 0.0001, "loss": 0.0348, "step": 41100 }, { "epoch": 261.8471337579618, "grad_norm": 1.3417372703552246, "learning_rate": 0.0001, "loss": 0.039, "step": 41110 }, { "epoch": 261.9108280254777, "grad_norm": 1.5777103900909424, "learning_rate": 0.0001, "loss": 0.0336, "step": 41120 }, { "epoch": 261.97452229299364, "grad_norm": 1.178314208984375, "learning_rate": 0.0001, "loss": 0.0343, "step": 41130 }, { "epoch": 262.03821656050957, "grad_norm": 1.1094969511032104, "learning_rate": 0.0001, "loss": 0.0365, "step": 41140 }, { "epoch": 262.1019108280255, "grad_norm": 1.0582515001296997, "learning_rate": 0.0001, "loss": 0.0328, "step": 41150 }, { "epoch": 262.1656050955414, "grad_norm": 1.1566035747528076, "learning_rate": 0.0001, "loss": 0.0379, "step": 41160 }, { "epoch": 262.22929936305735, "grad_norm": 1.3901227712631226, "learning_rate": 0.0001, "loss": 0.0342, "step": 41170 }, { "epoch": 262.2929936305732, "grad_norm": 1.4202693700790405, "learning_rate": 0.0001, "loss": 0.037, "step": 41180 }, { "epoch": 262.35668789808915, "grad_norm": 1.4792617559432983, "learning_rate": 0.0001, "loss": 0.0375, "step": 41190 }, { "epoch": 262.4203821656051, "grad_norm": 1.43850576877594, "learning_rate": 0.0001, "loss": 0.0356, "step": 41200 }, { "epoch": 262.484076433121, "grad_norm": 1.1997050046920776, "learning_rate": 0.0001, "loss": 0.0348, "step": 41210 }, { "epoch": 262.54777070063693, "grad_norm": 1.3334983587265015, "learning_rate": 0.0001, "loss": 0.0367, "step": 41220 }, { "epoch": 262.61146496815286, "grad_norm": 1.0785316228866577, "learning_rate": 0.0001, "loss": 0.0321, "step": 41230 }, { "epoch": 262.6751592356688, "grad_norm": 1.097367286682129, "learning_rate": 0.0001, "loss": 0.0368, "step": 41240 }, { "epoch": 262.7388535031847, "grad_norm": 0.9903227686882019, "learning_rate": 0.0001, "loss": 0.038, "step": 41250 }, { "epoch": 262.80254777070064, "grad_norm": 1.2051888704299927, "learning_rate": 0.0001, "loss": 0.0381, "step": 41260 }, { "epoch": 262.86624203821657, "grad_norm": 1.203722357749939, "learning_rate": 0.0001, "loss": 0.036, "step": 41270 }, { "epoch": 262.9299363057325, "grad_norm": 1.1353344917297363, "learning_rate": 0.0001, "loss": 0.0355, "step": 41280 }, { "epoch": 262.9936305732484, "grad_norm": 1.1861594915390015, "learning_rate": 0.0001, "loss": 0.0355, "step": 41290 }, { "epoch": 263.05732484076435, "grad_norm": 1.114586591720581, "learning_rate": 0.0001, "loss": 0.0332, "step": 41300 }, { "epoch": 263.1210191082803, "grad_norm": 1.155036449432373, "learning_rate": 0.0001, "loss": 0.0332, "step": 41310 }, { "epoch": 263.18471337579615, "grad_norm": 0.963544487953186, "learning_rate": 0.0001, "loss": 0.0366, "step": 41320 }, { "epoch": 263.2484076433121, "grad_norm": 0.8189671039581299, "learning_rate": 0.0001, "loss": 0.0357, "step": 41330 }, { "epoch": 263.312101910828, "grad_norm": 1.006394863128662, "learning_rate": 0.0001, "loss": 0.0337, "step": 41340 }, { "epoch": 263.37579617834393, "grad_norm": 1.0319392681121826, "learning_rate": 0.0001, "loss": 0.0362, "step": 41350 }, { "epoch": 263.43949044585986, "grad_norm": 1.3904144763946533, "learning_rate": 0.0001, "loss": 0.0403, "step": 41360 }, { "epoch": 263.5031847133758, "grad_norm": 1.3573861122131348, "learning_rate": 0.0001, "loss": 0.0379, "step": 41370 }, { "epoch": 263.5668789808917, "grad_norm": 1.0372722148895264, "learning_rate": 0.0001, "loss": 0.0361, "step": 41380 }, { "epoch": 263.63057324840764, "grad_norm": 0.9946988821029663, "learning_rate": 0.0001, "loss": 0.0363, "step": 41390 }, { "epoch": 263.69426751592357, "grad_norm": 1.3063592910766602, "learning_rate": 0.0001, "loss": 0.0385, "step": 41400 }, { "epoch": 263.7579617834395, "grad_norm": 1.0376213788986206, "learning_rate": 0.0001, "loss": 0.035, "step": 41410 }, { "epoch": 263.8216560509554, "grad_norm": 1.1034560203552246, "learning_rate": 0.0001, "loss": 0.0409, "step": 41420 }, { "epoch": 263.88535031847135, "grad_norm": 1.0050134658813477, "learning_rate": 0.0001, "loss": 0.0358, "step": 41430 }, { "epoch": 263.9490445859873, "grad_norm": 1.157002568244934, "learning_rate": 0.0001, "loss": 0.0386, "step": 41440 }, { "epoch": 264.0127388535032, "grad_norm": 1.2368427515029907, "learning_rate": 0.0001, "loss": 0.0394, "step": 41450 }, { "epoch": 264.07643312101914, "grad_norm": 1.2322436571121216, "learning_rate": 0.0001, "loss": 0.0426, "step": 41460 }, { "epoch": 264.140127388535, "grad_norm": 1.0874234437942505, "learning_rate": 0.0001, "loss": 0.0414, "step": 41470 }, { "epoch": 264.20382165605093, "grad_norm": 1.319815754890442, "learning_rate": 0.0001, "loss": 0.0374, "step": 41480 }, { "epoch": 264.26751592356686, "grad_norm": 1.3276795148849487, "learning_rate": 0.0001, "loss": 0.0393, "step": 41490 }, { "epoch": 264.3312101910828, "grad_norm": 1.1859794855117798, "learning_rate": 0.0001, "loss": 0.0336, "step": 41500 }, { "epoch": 264.3949044585987, "grad_norm": 1.4010268449783325, "learning_rate": 0.0001, "loss": 0.0366, "step": 41510 }, { "epoch": 264.45859872611464, "grad_norm": 1.3486671447753906, "learning_rate": 0.0001, "loss": 0.0376, "step": 41520 }, { "epoch": 264.5222929936306, "grad_norm": 1.0949841737747192, "learning_rate": 0.0001, "loss": 0.0324, "step": 41530 }, { "epoch": 264.5859872611465, "grad_norm": 1.29159414768219, "learning_rate": 0.0001, "loss": 0.0346, "step": 41540 }, { "epoch": 264.6496815286624, "grad_norm": 1.236178994178772, "learning_rate": 0.0001, "loss": 0.0373, "step": 41550 }, { "epoch": 264.71337579617835, "grad_norm": 0.9748308062553406, "learning_rate": 0.0001, "loss": 0.0367, "step": 41560 }, { "epoch": 264.7770700636943, "grad_norm": 1.079221248626709, "learning_rate": 0.0001, "loss": 0.0352, "step": 41570 }, { "epoch": 264.8407643312102, "grad_norm": 1.1225643157958984, "learning_rate": 0.0001, "loss": 0.0361, "step": 41580 }, { "epoch": 264.90445859872614, "grad_norm": 1.3501662015914917, "learning_rate": 0.0001, "loss": 0.0372, "step": 41590 }, { "epoch": 264.96815286624206, "grad_norm": 1.2766447067260742, "learning_rate": 0.0001, "loss": 0.0367, "step": 41600 }, { "epoch": 265.03184713375794, "grad_norm": 1.4640973806381226, "learning_rate": 0.0001, "loss": 0.0368, "step": 41610 }, { "epoch": 265.09554140127386, "grad_norm": 1.1997907161712646, "learning_rate": 0.0001, "loss": 0.038, "step": 41620 }, { "epoch": 265.1592356687898, "grad_norm": 1.1170296669006348, "learning_rate": 0.0001, "loss": 0.0357, "step": 41630 }, { "epoch": 265.2229299363057, "grad_norm": 1.143489122390747, "learning_rate": 0.0001, "loss": 0.0339, "step": 41640 }, { "epoch": 265.28662420382165, "grad_norm": 1.0804650783538818, "learning_rate": 0.0001, "loss": 0.0353, "step": 41650 }, { "epoch": 265.3503184713376, "grad_norm": 0.9693114757537842, "learning_rate": 0.0001, "loss": 0.0368, "step": 41660 }, { "epoch": 265.4140127388535, "grad_norm": 0.9183288812637329, "learning_rate": 0.0001, "loss": 0.0366, "step": 41670 }, { "epoch": 265.4777070063694, "grad_norm": 1.0983431339263916, "learning_rate": 0.0001, "loss": 0.0352, "step": 41680 }, { "epoch": 265.54140127388536, "grad_norm": 0.9339079260826111, "learning_rate": 0.0001, "loss": 0.0384, "step": 41690 }, { "epoch": 265.6050955414013, "grad_norm": 0.7745937705039978, "learning_rate": 0.0001, "loss": 0.0363, "step": 41700 }, { "epoch": 265.6687898089172, "grad_norm": 1.065065860748291, "learning_rate": 0.0001, "loss": 0.0368, "step": 41710 }, { "epoch": 265.73248407643314, "grad_norm": 1.177975058555603, "learning_rate": 0.0001, "loss": 0.0346, "step": 41720 }, { "epoch": 265.79617834394907, "grad_norm": 1.0432804822921753, "learning_rate": 0.0001, "loss": 0.0386, "step": 41730 }, { "epoch": 265.859872611465, "grad_norm": 1.1709144115447998, "learning_rate": 0.0001, "loss": 0.0391, "step": 41740 }, { "epoch": 265.92356687898086, "grad_norm": 1.2911566495895386, "learning_rate": 0.0001, "loss": 0.0409, "step": 41750 }, { "epoch": 265.9872611464968, "grad_norm": 1.1695177555084229, "learning_rate": 0.0001, "loss": 0.0348, "step": 41760 }, { "epoch": 266.0509554140127, "grad_norm": 1.2939857244491577, "learning_rate": 0.0001, "loss": 0.0349, "step": 41770 }, { "epoch": 266.11464968152865, "grad_norm": 1.0917294025421143, "learning_rate": 0.0001, "loss": 0.0351, "step": 41780 }, { "epoch": 266.1783439490446, "grad_norm": 1.2485785484313965, "learning_rate": 0.0001, "loss": 0.0372, "step": 41790 }, { "epoch": 266.2420382165605, "grad_norm": 1.6118338108062744, "learning_rate": 0.0001, "loss": 0.0361, "step": 41800 }, { "epoch": 266.30573248407643, "grad_norm": 1.4788111448287964, "learning_rate": 0.0001, "loss": 0.0376, "step": 41810 }, { "epoch": 266.36942675159236, "grad_norm": 1.2995959520339966, "learning_rate": 0.0001, "loss": 0.0343, "step": 41820 }, { "epoch": 266.4331210191083, "grad_norm": 1.7478834390640259, "learning_rate": 0.0001, "loss": 0.0412, "step": 41830 }, { "epoch": 266.4968152866242, "grad_norm": 1.448765754699707, "learning_rate": 0.0001, "loss": 0.0381, "step": 41840 }, { "epoch": 266.56050955414014, "grad_norm": 1.4678527116775513, "learning_rate": 0.0001, "loss": 0.0383, "step": 41850 }, { "epoch": 266.62420382165607, "grad_norm": 1.3792979717254639, "learning_rate": 0.0001, "loss": 0.0339, "step": 41860 }, { "epoch": 266.687898089172, "grad_norm": 1.3987376689910889, "learning_rate": 0.0001, "loss": 0.0324, "step": 41870 }, { "epoch": 266.7515923566879, "grad_norm": 1.2138363122940063, "learning_rate": 0.0001, "loss": 0.033, "step": 41880 }, { "epoch": 266.81528662420385, "grad_norm": 1.9482084512710571, "learning_rate": 0.0001, "loss": 0.035, "step": 41890 }, { "epoch": 266.8789808917197, "grad_norm": 1.5547943115234375, "learning_rate": 0.0001, "loss": 0.0345, "step": 41900 }, { "epoch": 266.94267515923565, "grad_norm": 1.2293944358825684, "learning_rate": 0.0001, "loss": 0.0349, "step": 41910 }, { "epoch": 267.0063694267516, "grad_norm": 1.3084596395492554, "learning_rate": 0.0001, "loss": 0.034, "step": 41920 }, { "epoch": 267.0700636942675, "grad_norm": 1.1380929946899414, "learning_rate": 0.0001, "loss": 0.036, "step": 41930 }, { "epoch": 267.13375796178343, "grad_norm": 1.1168432235717773, "learning_rate": 0.0001, "loss": 0.0327, "step": 41940 }, { "epoch": 267.19745222929936, "grad_norm": 1.2341219186782837, "learning_rate": 0.0001, "loss": 0.0355, "step": 41950 }, { "epoch": 267.2611464968153, "grad_norm": 1.855569839477539, "learning_rate": 0.0001, "loss": 0.0348, "step": 41960 }, { "epoch": 267.3248407643312, "grad_norm": 1.293509602546692, "learning_rate": 0.0001, "loss": 0.0355, "step": 41970 }, { "epoch": 267.38853503184714, "grad_norm": 1.424421787261963, "learning_rate": 0.0001, "loss": 0.0356, "step": 41980 }, { "epoch": 267.45222929936307, "grad_norm": 1.5532640218734741, "learning_rate": 0.0001, "loss": 0.0372, "step": 41990 }, { "epoch": 267.515923566879, "grad_norm": 1.3986773490905762, "learning_rate": 0.0001, "loss": 0.0345, "step": 42000 }, { "epoch": 267.5796178343949, "grad_norm": 1.1807239055633545, "learning_rate": 0.0001, "loss": 0.034, "step": 42010 }, { "epoch": 267.64331210191085, "grad_norm": 1.3786509037017822, "learning_rate": 0.0001, "loss": 0.0378, "step": 42020 }, { "epoch": 267.7070063694268, "grad_norm": 1.0782362222671509, "learning_rate": 0.0001, "loss": 0.0336, "step": 42030 }, { "epoch": 267.77070063694265, "grad_norm": 1.1753889322280884, "learning_rate": 0.0001, "loss": 0.0338, "step": 42040 }, { "epoch": 267.8343949044586, "grad_norm": 1.4751590490341187, "learning_rate": 0.0001, "loss": 0.0351, "step": 42050 }, { "epoch": 267.8980891719745, "grad_norm": 1.1646952629089355, "learning_rate": 0.0001, "loss": 0.0356, "step": 42060 }, { "epoch": 267.96178343949043, "grad_norm": 1.2643789052963257, "learning_rate": 0.0001, "loss": 0.0363, "step": 42070 }, { "epoch": 268.02547770700636, "grad_norm": 1.1898013353347778, "learning_rate": 0.0001, "loss": 0.0337, "step": 42080 }, { "epoch": 268.0891719745223, "grad_norm": 1.1963155269622803, "learning_rate": 0.0001, "loss": 0.0369, "step": 42090 }, { "epoch": 268.1528662420382, "grad_norm": 1.348332405090332, "learning_rate": 0.0001, "loss": 0.0376, "step": 42100 }, { "epoch": 268.21656050955414, "grad_norm": 1.1147171258926392, "learning_rate": 0.0001, "loss": 0.0331, "step": 42110 }, { "epoch": 268.28025477707007, "grad_norm": 1.0956354141235352, "learning_rate": 0.0001, "loss": 0.0351, "step": 42120 }, { "epoch": 268.343949044586, "grad_norm": 1.551871418952942, "learning_rate": 0.0001, "loss": 0.0357, "step": 42130 }, { "epoch": 268.4076433121019, "grad_norm": 1.0832804441452026, "learning_rate": 0.0001, "loss": 0.0326, "step": 42140 }, { "epoch": 268.47133757961785, "grad_norm": 1.1016018390655518, "learning_rate": 0.0001, "loss": 0.0347, "step": 42150 }, { "epoch": 268.5350318471338, "grad_norm": 1.1153035163879395, "learning_rate": 0.0001, "loss": 0.0359, "step": 42160 }, { "epoch": 268.5987261146497, "grad_norm": 1.1724807024002075, "learning_rate": 0.0001, "loss": 0.0354, "step": 42170 }, { "epoch": 268.6624203821656, "grad_norm": 1.1544467210769653, "learning_rate": 0.0001, "loss": 0.0344, "step": 42180 }, { "epoch": 268.7261146496815, "grad_norm": 0.9096255898475647, "learning_rate": 0.0001, "loss": 0.0367, "step": 42190 }, { "epoch": 268.78980891719743, "grad_norm": 1.0616328716278076, "learning_rate": 0.0001, "loss": 0.0374, "step": 42200 }, { "epoch": 268.85350318471336, "grad_norm": 1.1638422012329102, "learning_rate": 0.0001, "loss": 0.0363, "step": 42210 }, { "epoch": 268.9171974522293, "grad_norm": 1.0865498781204224, "learning_rate": 0.0001, "loss": 0.0354, "step": 42220 }, { "epoch": 268.9808917197452, "grad_norm": 1.2595208883285522, "learning_rate": 0.0001, "loss": 0.0359, "step": 42230 }, { "epoch": 269.04458598726114, "grad_norm": 1.1066268682479858, "learning_rate": 0.0001, "loss": 0.035, "step": 42240 }, { "epoch": 269.10828025477707, "grad_norm": 1.180991291999817, "learning_rate": 0.0001, "loss": 0.0409, "step": 42250 }, { "epoch": 269.171974522293, "grad_norm": 1.0404473543167114, "learning_rate": 0.0001, "loss": 0.0366, "step": 42260 }, { "epoch": 269.2356687898089, "grad_norm": 1.1242390871047974, "learning_rate": 0.0001, "loss": 0.0388, "step": 42270 }, { "epoch": 269.29936305732485, "grad_norm": 1.2727631330490112, "learning_rate": 0.0001, "loss": 0.0389, "step": 42280 }, { "epoch": 269.3630573248408, "grad_norm": 0.9853754043579102, "learning_rate": 0.0001, "loss": 0.0358, "step": 42290 }, { "epoch": 269.4267515923567, "grad_norm": 1.0679224729537964, "learning_rate": 0.0001, "loss": 0.0399, "step": 42300 }, { "epoch": 269.49044585987264, "grad_norm": 0.8277201652526855, "learning_rate": 0.0001, "loss": 0.0404, "step": 42310 }, { "epoch": 269.5541401273885, "grad_norm": 1.1178659200668335, "learning_rate": 0.0001, "loss": 0.0385, "step": 42320 }, { "epoch": 269.61783439490443, "grad_norm": 1.0422252416610718, "learning_rate": 0.0001, "loss": 0.0353, "step": 42330 }, { "epoch": 269.68152866242036, "grad_norm": 0.9037112593650818, "learning_rate": 0.0001, "loss": 0.038, "step": 42340 }, { "epoch": 269.7452229299363, "grad_norm": 0.9932235479354858, "learning_rate": 0.0001, "loss": 0.0369, "step": 42350 }, { "epoch": 269.8089171974522, "grad_norm": 0.8170825242996216, "learning_rate": 0.0001, "loss": 0.0357, "step": 42360 }, { "epoch": 269.87261146496814, "grad_norm": 1.1722221374511719, "learning_rate": 0.0001, "loss": 0.0373, "step": 42370 }, { "epoch": 269.9363057324841, "grad_norm": 1.1758480072021484, "learning_rate": 0.0001, "loss": 0.0364, "step": 42380 }, { "epoch": 270.0, "grad_norm": 1.0548934936523438, "learning_rate": 0.0001, "loss": 0.0379, "step": 42390 }, { "epoch": 270.0636942675159, "grad_norm": 1.172040343284607, "learning_rate": 0.0001, "loss": 0.0376, "step": 42400 }, { "epoch": 270.12738853503186, "grad_norm": 1.1506582498550415, "learning_rate": 0.0001, "loss": 0.0336, "step": 42410 }, { "epoch": 270.1910828025478, "grad_norm": 1.149198055267334, "learning_rate": 0.0001, "loss": 0.0374, "step": 42420 }, { "epoch": 270.2547770700637, "grad_norm": 1.0450284481048584, "learning_rate": 0.0001, "loss": 0.0353, "step": 42430 }, { "epoch": 270.31847133757964, "grad_norm": 1.0451959371566772, "learning_rate": 0.0001, "loss": 0.0373, "step": 42440 }, { "epoch": 270.38216560509557, "grad_norm": 0.9476402401924133, "learning_rate": 0.0001, "loss": 0.0348, "step": 42450 }, { "epoch": 270.4458598726115, "grad_norm": 1.0170619487762451, "learning_rate": 0.0001, "loss": 0.0361, "step": 42460 }, { "epoch": 270.50955414012736, "grad_norm": 1.0504766702651978, "learning_rate": 0.0001, "loss": 0.0347, "step": 42470 }, { "epoch": 270.5732484076433, "grad_norm": 1.3412259817123413, "learning_rate": 0.0001, "loss": 0.033, "step": 42480 }, { "epoch": 270.6369426751592, "grad_norm": 1.3114560842514038, "learning_rate": 0.0001, "loss": 0.0359, "step": 42490 }, { "epoch": 270.70063694267515, "grad_norm": 1.3639932870864868, "learning_rate": 0.0001, "loss": 0.0349, "step": 42500 }, { "epoch": 270.7643312101911, "grad_norm": 1.225223183631897, "learning_rate": 0.0001, "loss": 0.0352, "step": 42510 }, { "epoch": 270.828025477707, "grad_norm": 1.5885882377624512, "learning_rate": 0.0001, "loss": 0.0358, "step": 42520 }, { "epoch": 270.89171974522293, "grad_norm": 1.0665132999420166, "learning_rate": 0.0001, "loss": 0.0381, "step": 42530 }, { "epoch": 270.95541401273886, "grad_norm": 1.1357611417770386, "learning_rate": 0.0001, "loss": 0.0352, "step": 42540 }, { "epoch": 271.0191082802548, "grad_norm": 1.212393045425415, "learning_rate": 0.0001, "loss": 0.0342, "step": 42550 }, { "epoch": 271.0828025477707, "grad_norm": 1.1770343780517578, "learning_rate": 0.0001, "loss": 0.0358, "step": 42560 }, { "epoch": 271.14649681528664, "grad_norm": 1.0111671686172485, "learning_rate": 0.0001, "loss": 0.0327, "step": 42570 }, { "epoch": 271.21019108280257, "grad_norm": 1.078924298286438, "learning_rate": 0.0001, "loss": 0.0366, "step": 42580 }, { "epoch": 271.2738853503185, "grad_norm": 1.160529613494873, "learning_rate": 0.0001, "loss": 0.0352, "step": 42590 }, { "epoch": 271.3375796178344, "grad_norm": 1.2797297239303589, "learning_rate": 0.0001, "loss": 0.038, "step": 42600 }, { "epoch": 271.4012738853503, "grad_norm": 1.2786548137664795, "learning_rate": 0.0001, "loss": 0.036, "step": 42610 }, { "epoch": 271.4649681528662, "grad_norm": 1.0959022045135498, "learning_rate": 0.0001, "loss": 0.0345, "step": 42620 }, { "epoch": 271.52866242038215, "grad_norm": 1.159274935722351, "learning_rate": 0.0001, "loss": 0.0373, "step": 42630 }, { "epoch": 271.5923566878981, "grad_norm": 1.2667843103408813, "learning_rate": 0.0001, "loss": 0.0339, "step": 42640 }, { "epoch": 271.656050955414, "grad_norm": 1.4221832752227783, "learning_rate": 0.0001, "loss": 0.0358, "step": 42650 }, { "epoch": 271.71974522292993, "grad_norm": 1.1540274620056152, "learning_rate": 0.0001, "loss": 0.038, "step": 42660 }, { "epoch": 271.78343949044586, "grad_norm": 1.261274814605713, "learning_rate": 0.0001, "loss": 0.0394, "step": 42670 }, { "epoch": 271.8471337579618, "grad_norm": 1.2717167139053345, "learning_rate": 0.0001, "loss": 0.036, "step": 42680 }, { "epoch": 271.9108280254777, "grad_norm": 1.220115303993225, "learning_rate": 0.0001, "loss": 0.0361, "step": 42690 }, { "epoch": 271.97452229299364, "grad_norm": 1.044256567955017, "learning_rate": 0.0001, "loss": 0.035, "step": 42700 }, { "epoch": 272.03821656050957, "grad_norm": 0.9948313236236572, "learning_rate": 0.0001, "loss": 0.0348, "step": 42710 }, { "epoch": 272.1019108280255, "grad_norm": 1.3001606464385986, "learning_rate": 0.0001, "loss": 0.0378, "step": 42720 }, { "epoch": 272.1656050955414, "grad_norm": 1.024067759513855, "learning_rate": 0.0001, "loss": 0.0362, "step": 42730 }, { "epoch": 272.22929936305735, "grad_norm": 0.953614354133606, "learning_rate": 0.0001, "loss": 0.037, "step": 42740 }, { "epoch": 272.2929936305732, "grad_norm": 1.4615201950073242, "learning_rate": 0.0001, "loss": 0.0421, "step": 42750 }, { "epoch": 272.35668789808915, "grad_norm": 1.0046385526657104, "learning_rate": 0.0001, "loss": 0.0401, "step": 42760 }, { "epoch": 272.4203821656051, "grad_norm": 1.2055983543395996, "learning_rate": 0.0001, "loss": 0.0419, "step": 42770 }, { "epoch": 272.484076433121, "grad_norm": 1.0911413431167603, "learning_rate": 0.0001, "loss": 0.0363, "step": 42780 }, { "epoch": 272.54777070063693, "grad_norm": 1.234527826309204, "learning_rate": 0.0001, "loss": 0.036, "step": 42790 }, { "epoch": 272.61146496815286, "grad_norm": 0.889971911907196, "learning_rate": 0.0001, "loss": 0.0343, "step": 42800 }, { "epoch": 272.6751592356688, "grad_norm": 1.0099434852600098, "learning_rate": 0.0001, "loss": 0.0344, "step": 42810 }, { "epoch": 272.7388535031847, "grad_norm": 0.9875357747077942, "learning_rate": 0.0001, "loss": 0.036, "step": 42820 }, { "epoch": 272.80254777070064, "grad_norm": 0.9319187998771667, "learning_rate": 0.0001, "loss": 0.0376, "step": 42830 }, { "epoch": 272.86624203821657, "grad_norm": 1.1737709045410156, "learning_rate": 0.0001, "loss": 0.0392, "step": 42840 }, { "epoch": 272.9299363057325, "grad_norm": 1.0742828845977783, "learning_rate": 0.0001, "loss": 0.0409, "step": 42850 }, { "epoch": 272.9936305732484, "grad_norm": 1.0961254835128784, "learning_rate": 0.0001, "loss": 0.0387, "step": 42860 }, { "epoch": 273.05732484076435, "grad_norm": 1.21340012550354, "learning_rate": 0.0001, "loss": 0.0386, "step": 42870 }, { "epoch": 273.1210191082803, "grad_norm": 1.162640929222107, "learning_rate": 0.0001, "loss": 0.0384, "step": 42880 }, { "epoch": 273.18471337579615, "grad_norm": 1.0844014883041382, "learning_rate": 0.0001, "loss": 0.0364, "step": 42890 }, { "epoch": 273.2484076433121, "grad_norm": 1.2888518571853638, "learning_rate": 0.0001, "loss": 0.0362, "step": 42900 }, { "epoch": 273.312101910828, "grad_norm": 0.9037073850631714, "learning_rate": 0.0001, "loss": 0.0342, "step": 42910 }, { "epoch": 273.37579617834393, "grad_norm": 1.3914045095443726, "learning_rate": 0.0001, "loss": 0.0361, "step": 42920 }, { "epoch": 273.43949044585986, "grad_norm": 1.4002925157546997, "learning_rate": 0.0001, "loss": 0.0406, "step": 42930 }, { "epoch": 273.5031847133758, "grad_norm": 0.9656053185462952, "learning_rate": 0.0001, "loss": 0.0391, "step": 42940 }, { "epoch": 273.5668789808917, "grad_norm": 1.4634919166564941, "learning_rate": 0.0001, "loss": 0.0316, "step": 42950 }, { "epoch": 273.63057324840764, "grad_norm": 1.438966989517212, "learning_rate": 0.0001, "loss": 0.0374, "step": 42960 }, { "epoch": 273.69426751592357, "grad_norm": 1.1152071952819824, "learning_rate": 0.0001, "loss": 0.0363, "step": 42970 }, { "epoch": 273.7579617834395, "grad_norm": 1.342861294746399, "learning_rate": 0.0001, "loss": 0.0353, "step": 42980 }, { "epoch": 273.8216560509554, "grad_norm": 1.4852344989776611, "learning_rate": 0.0001, "loss": 0.0358, "step": 42990 }, { "epoch": 273.88535031847135, "grad_norm": 1.2211552858352661, "learning_rate": 0.0001, "loss": 0.0347, "step": 43000 }, { "epoch": 273.9490445859873, "grad_norm": 0.9353947043418884, "learning_rate": 0.0001, "loss": 0.0352, "step": 43010 }, { "epoch": 274.0127388535032, "grad_norm": 1.1681313514709473, "learning_rate": 0.0001, "loss": 0.0342, "step": 43020 }, { "epoch": 274.07643312101914, "grad_norm": 0.9419155120849609, "learning_rate": 0.0001, "loss": 0.0334, "step": 43030 }, { "epoch": 274.140127388535, "grad_norm": 1.0775197744369507, "learning_rate": 0.0001, "loss": 0.0342, "step": 43040 }, { "epoch": 274.20382165605093, "grad_norm": 1.3677889108657837, "learning_rate": 0.0001, "loss": 0.0354, "step": 43050 }, { "epoch": 274.26751592356686, "grad_norm": 1.4099345207214355, "learning_rate": 0.0001, "loss": 0.0314, "step": 43060 }, { "epoch": 274.3312101910828, "grad_norm": 1.3413026332855225, "learning_rate": 0.0001, "loss": 0.0371, "step": 43070 }, { "epoch": 274.3949044585987, "grad_norm": 1.22281014919281, "learning_rate": 0.0001, "loss": 0.0348, "step": 43080 }, { "epoch": 274.45859872611464, "grad_norm": 1.2305679321289062, "learning_rate": 0.0001, "loss": 0.0354, "step": 43090 }, { "epoch": 274.5222929936306, "grad_norm": 1.252339243888855, "learning_rate": 0.0001, "loss": 0.0345, "step": 43100 }, { "epoch": 274.5859872611465, "grad_norm": 1.404598593711853, "learning_rate": 0.0001, "loss": 0.0369, "step": 43110 }, { "epoch": 274.6496815286624, "grad_norm": 1.2815181016921997, "learning_rate": 0.0001, "loss": 0.037, "step": 43120 }, { "epoch": 274.71337579617835, "grad_norm": 1.378035545349121, "learning_rate": 0.0001, "loss": 0.0335, "step": 43130 }, { "epoch": 274.7770700636943, "grad_norm": 1.3566014766693115, "learning_rate": 0.0001, "loss": 0.0345, "step": 43140 }, { "epoch": 274.8407643312102, "grad_norm": 1.4531959295272827, "learning_rate": 0.0001, "loss": 0.0398, "step": 43150 }, { "epoch": 274.90445859872614, "grad_norm": 1.3926578760147095, "learning_rate": 0.0001, "loss": 0.0357, "step": 43160 }, { "epoch": 274.96815286624206, "grad_norm": 1.7300584316253662, "learning_rate": 0.0001, "loss": 0.0379, "step": 43170 }, { "epoch": 275.03184713375794, "grad_norm": 1.3308717012405396, "learning_rate": 0.0001, "loss": 0.0357, "step": 43180 }, { "epoch": 275.09554140127386, "grad_norm": 1.173810362815857, "learning_rate": 0.0001, "loss": 0.0334, "step": 43190 }, { "epoch": 275.1592356687898, "grad_norm": 1.3923869132995605, "learning_rate": 0.0001, "loss": 0.0346, "step": 43200 }, { "epoch": 275.2229299363057, "grad_norm": 1.5564002990722656, "learning_rate": 0.0001, "loss": 0.0377, "step": 43210 }, { "epoch": 275.28662420382165, "grad_norm": 1.3708455562591553, "learning_rate": 0.0001, "loss": 0.0342, "step": 43220 }, { "epoch": 275.3503184713376, "grad_norm": 1.1598389148712158, "learning_rate": 0.0001, "loss": 0.0326, "step": 43230 }, { "epoch": 275.4140127388535, "grad_norm": 1.3342294692993164, "learning_rate": 0.0001, "loss": 0.0348, "step": 43240 }, { "epoch": 275.4777070063694, "grad_norm": 1.3119187355041504, "learning_rate": 0.0001, "loss": 0.0338, "step": 43250 }, { "epoch": 275.54140127388536, "grad_norm": 1.2165501117706299, "learning_rate": 0.0001, "loss": 0.0335, "step": 43260 }, { "epoch": 275.6050955414013, "grad_norm": 1.111997127532959, "learning_rate": 0.0001, "loss": 0.0384, "step": 43270 }, { "epoch": 275.6687898089172, "grad_norm": 1.4193241596221924, "learning_rate": 0.0001, "loss": 0.0359, "step": 43280 }, { "epoch": 275.73248407643314, "grad_norm": 2.4954309463500977, "learning_rate": 0.0001, "loss": 0.0367, "step": 43290 }, { "epoch": 275.79617834394907, "grad_norm": 1.6218295097351074, "learning_rate": 0.0001, "loss": 0.0346, "step": 43300 }, { "epoch": 275.859872611465, "grad_norm": 1.4813251495361328, "learning_rate": 0.0001, "loss": 0.0359, "step": 43310 }, { "epoch": 275.92356687898086, "grad_norm": 1.51995050907135, "learning_rate": 0.0001, "loss": 0.0351, "step": 43320 }, { "epoch": 275.9872611464968, "grad_norm": 1.274019479751587, "learning_rate": 0.0001, "loss": 0.0336, "step": 43330 }, { "epoch": 276.0509554140127, "grad_norm": 1.5140258073806763, "learning_rate": 0.0001, "loss": 0.0326, "step": 43340 }, { "epoch": 276.11464968152865, "grad_norm": 1.3450520038604736, "learning_rate": 0.0001, "loss": 0.0327, "step": 43350 }, { "epoch": 276.1783439490446, "grad_norm": 1.3490824699401855, "learning_rate": 0.0001, "loss": 0.0352, "step": 43360 }, { "epoch": 276.2420382165605, "grad_norm": 1.1642521619796753, "learning_rate": 0.0001, "loss": 0.0374, "step": 43370 }, { "epoch": 276.30573248407643, "grad_norm": 1.1848996877670288, "learning_rate": 0.0001, "loss": 0.0358, "step": 43380 }, { "epoch": 276.36942675159236, "grad_norm": 1.2004694938659668, "learning_rate": 0.0001, "loss": 0.0336, "step": 43390 }, { "epoch": 276.4331210191083, "grad_norm": 1.1324703693389893, "learning_rate": 0.0001, "loss": 0.0343, "step": 43400 }, { "epoch": 276.4968152866242, "grad_norm": 1.0144710540771484, "learning_rate": 0.0001, "loss": 0.0367, "step": 43410 }, { "epoch": 276.56050955414014, "grad_norm": 1.0643551349639893, "learning_rate": 0.0001, "loss": 0.0374, "step": 43420 }, { "epoch": 276.62420382165607, "grad_norm": 1.139322280883789, "learning_rate": 0.0001, "loss": 0.0349, "step": 43430 }, { "epoch": 276.687898089172, "grad_norm": 1.178084373474121, "learning_rate": 0.0001, "loss": 0.0346, "step": 43440 }, { "epoch": 276.7515923566879, "grad_norm": 1.4199572801589966, "learning_rate": 0.0001, "loss": 0.0369, "step": 43450 }, { "epoch": 276.81528662420385, "grad_norm": 0.9612724781036377, "learning_rate": 0.0001, "loss": 0.0327, "step": 43460 }, { "epoch": 276.8789808917197, "grad_norm": 0.9493834972381592, "learning_rate": 0.0001, "loss": 0.0353, "step": 43470 }, { "epoch": 276.94267515923565, "grad_norm": 1.1346145868301392, "learning_rate": 0.0001, "loss": 0.0344, "step": 43480 }, { "epoch": 277.0063694267516, "grad_norm": 1.1999173164367676, "learning_rate": 0.0001, "loss": 0.0369, "step": 43490 }, { "epoch": 277.0700636942675, "grad_norm": 1.1325973272323608, "learning_rate": 0.0001, "loss": 0.0365, "step": 43500 }, { "epoch": 277.13375796178343, "grad_norm": 1.2360800504684448, "learning_rate": 0.0001, "loss": 0.0327, "step": 43510 }, { "epoch": 277.19745222929936, "grad_norm": 1.4315769672393799, "learning_rate": 0.0001, "loss": 0.034, "step": 43520 }, { "epoch": 277.2611464968153, "grad_norm": 1.1524970531463623, "learning_rate": 0.0001, "loss": 0.0347, "step": 43530 }, { "epoch": 277.3248407643312, "grad_norm": 0.9499521255493164, "learning_rate": 0.0001, "loss": 0.0373, "step": 43540 }, { "epoch": 277.38853503184714, "grad_norm": 0.9928136467933655, "learning_rate": 0.0001, "loss": 0.0359, "step": 43550 }, { "epoch": 277.45222929936307, "grad_norm": 1.1344939470291138, "learning_rate": 0.0001, "loss": 0.0399, "step": 43560 }, { "epoch": 277.515923566879, "grad_norm": 1.1643462181091309, "learning_rate": 0.0001, "loss": 0.0361, "step": 43570 }, { "epoch": 277.5796178343949, "grad_norm": 1.0533946752548218, "learning_rate": 0.0001, "loss": 0.0388, "step": 43580 }, { "epoch": 277.64331210191085, "grad_norm": 0.861634373664856, "learning_rate": 0.0001, "loss": 0.0381, "step": 43590 }, { "epoch": 277.7070063694268, "grad_norm": 0.9700922966003418, "learning_rate": 0.0001, "loss": 0.0343, "step": 43600 }, { "epoch": 277.77070063694265, "grad_norm": 1.1828765869140625, "learning_rate": 0.0001, "loss": 0.0364, "step": 43610 }, { "epoch": 277.8343949044586, "grad_norm": 1.0859293937683105, "learning_rate": 0.0001, "loss": 0.0396, "step": 43620 }, { "epoch": 277.8980891719745, "grad_norm": 1.158491611480713, "learning_rate": 0.0001, "loss": 0.0371, "step": 43630 }, { "epoch": 277.96178343949043, "grad_norm": 1.0128037929534912, "learning_rate": 0.0001, "loss": 0.0359, "step": 43640 }, { "epoch": 278.02547770700636, "grad_norm": 1.277807593345642, "learning_rate": 0.0001, "loss": 0.0333, "step": 43650 }, { "epoch": 278.0891719745223, "grad_norm": 1.361131191253662, "learning_rate": 0.0001, "loss": 0.0363, "step": 43660 }, { "epoch": 278.1528662420382, "grad_norm": 1.1404985189437866, "learning_rate": 0.0001, "loss": 0.0346, "step": 43670 }, { "epoch": 278.21656050955414, "grad_norm": 0.9133114218711853, "learning_rate": 0.0001, "loss": 0.0352, "step": 43680 }, { "epoch": 278.28025477707007, "grad_norm": 1.0280283689498901, "learning_rate": 0.0001, "loss": 0.0355, "step": 43690 }, { "epoch": 278.343949044586, "grad_norm": 0.9359015822410583, "learning_rate": 0.0001, "loss": 0.0346, "step": 43700 }, { "epoch": 278.4076433121019, "grad_norm": 0.9360371828079224, "learning_rate": 0.0001, "loss": 0.0356, "step": 43710 }, { "epoch": 278.47133757961785, "grad_norm": 1.2504301071166992, "learning_rate": 0.0001, "loss": 0.0352, "step": 43720 }, { "epoch": 278.5350318471338, "grad_norm": 1.2449568510055542, "learning_rate": 0.0001, "loss": 0.0386, "step": 43730 }, { "epoch": 278.5987261146497, "grad_norm": 0.9668094515800476, "learning_rate": 0.0001, "loss": 0.0381, "step": 43740 }, { "epoch": 278.6624203821656, "grad_norm": 1.3358653783798218, "learning_rate": 0.0001, "loss": 0.0365, "step": 43750 }, { "epoch": 278.7261146496815, "grad_norm": 1.076244592666626, "learning_rate": 0.0001, "loss": 0.0348, "step": 43760 }, { "epoch": 278.78980891719743, "grad_norm": 1.033848524093628, "learning_rate": 0.0001, "loss": 0.0404, "step": 43770 }, { "epoch": 278.85350318471336, "grad_norm": 1.1198419332504272, "learning_rate": 0.0001, "loss": 0.0375, "step": 43780 }, { "epoch": 278.9171974522293, "grad_norm": 0.9278236031532288, "learning_rate": 0.0001, "loss": 0.0367, "step": 43790 }, { "epoch": 278.9808917197452, "grad_norm": 1.2731590270996094, "learning_rate": 0.0001, "loss": 0.0352, "step": 43800 }, { "epoch": 279.04458598726114, "grad_norm": 1.1037293672561646, "learning_rate": 0.0001, "loss": 0.0337, "step": 43810 }, { "epoch": 279.10828025477707, "grad_norm": 1.143509030342102, "learning_rate": 0.0001, "loss": 0.0373, "step": 43820 }, { "epoch": 279.171974522293, "grad_norm": 1.1272472143173218, "learning_rate": 0.0001, "loss": 0.0365, "step": 43830 }, { "epoch": 279.2356687898089, "grad_norm": 1.1248914003372192, "learning_rate": 0.0001, "loss": 0.035, "step": 43840 }, { "epoch": 279.29936305732485, "grad_norm": 1.3280314207077026, "learning_rate": 0.0001, "loss": 0.0365, "step": 43850 }, { "epoch": 279.3630573248408, "grad_norm": 0.9171236157417297, "learning_rate": 0.0001, "loss": 0.0332, "step": 43860 }, { "epoch": 279.4267515923567, "grad_norm": 0.9205530285835266, "learning_rate": 0.0001, "loss": 0.036, "step": 43870 }, { "epoch": 279.49044585987264, "grad_norm": 1.016737937927246, "learning_rate": 0.0001, "loss": 0.0377, "step": 43880 }, { "epoch": 279.5541401273885, "grad_norm": 0.9768182039260864, "learning_rate": 0.0001, "loss": 0.0364, "step": 43890 }, { "epoch": 279.61783439490443, "grad_norm": 0.956591784954071, "learning_rate": 0.0001, "loss": 0.0367, "step": 43900 }, { "epoch": 279.68152866242036, "grad_norm": 1.2489410638809204, "learning_rate": 0.0001, "loss": 0.0373, "step": 43910 }, { "epoch": 279.7452229299363, "grad_norm": 1.313222050666809, "learning_rate": 0.0001, "loss": 0.0371, "step": 43920 }, { "epoch": 279.8089171974522, "grad_norm": 0.7744364142417908, "learning_rate": 0.0001, "loss": 0.0372, "step": 43930 }, { "epoch": 279.87261146496814, "grad_norm": 1.0760700702667236, "learning_rate": 0.0001, "loss": 0.0384, "step": 43940 }, { "epoch": 279.9363057324841, "grad_norm": 1.3452250957489014, "learning_rate": 0.0001, "loss": 0.0394, "step": 43950 }, { "epoch": 280.0, "grad_norm": 1.1383167505264282, "learning_rate": 0.0001, "loss": 0.0402, "step": 43960 }, { "epoch": 280.0636942675159, "grad_norm": 1.3161146640777588, "learning_rate": 0.0001, "loss": 0.0361, "step": 43970 }, { "epoch": 280.12738853503186, "grad_norm": 1.2585593461990356, "learning_rate": 0.0001, "loss": 0.035, "step": 43980 }, { "epoch": 280.1910828025478, "grad_norm": 0.9860112071037292, "learning_rate": 0.0001, "loss": 0.0332, "step": 43990 }, { "epoch": 280.2547770700637, "grad_norm": 1.4319769144058228, "learning_rate": 0.0001, "loss": 0.0343, "step": 44000 }, { "epoch": 280.31847133757964, "grad_norm": 1.240887999534607, "learning_rate": 0.0001, "loss": 0.0347, "step": 44010 }, { "epoch": 280.38216560509557, "grad_norm": 1.0701842308044434, "learning_rate": 0.0001, "loss": 0.0343, "step": 44020 }, { "epoch": 280.4458598726115, "grad_norm": 1.2081607580184937, "learning_rate": 0.0001, "loss": 0.0382, "step": 44030 }, { "epoch": 280.50955414012736, "grad_norm": 1.2432738542556763, "learning_rate": 0.0001, "loss": 0.0379, "step": 44040 }, { "epoch": 280.5732484076433, "grad_norm": 1.0416839122772217, "learning_rate": 0.0001, "loss": 0.0358, "step": 44050 }, { "epoch": 280.6369426751592, "grad_norm": 0.9202220439910889, "learning_rate": 0.0001, "loss": 0.0339, "step": 44060 }, { "epoch": 280.70063694267515, "grad_norm": 1.0472546815872192, "learning_rate": 0.0001, "loss": 0.037, "step": 44070 }, { "epoch": 280.7643312101911, "grad_norm": 1.2817422151565552, "learning_rate": 0.0001, "loss": 0.0363, "step": 44080 }, { "epoch": 280.828025477707, "grad_norm": 1.5080642700195312, "learning_rate": 0.0001, "loss": 0.037, "step": 44090 }, { "epoch": 280.89171974522293, "grad_norm": 1.2433043718338013, "learning_rate": 0.0001, "loss": 0.0354, "step": 44100 }, { "epoch": 280.95541401273886, "grad_norm": 1.5411279201507568, "learning_rate": 0.0001, "loss": 0.0377, "step": 44110 }, { "epoch": 281.0191082802548, "grad_norm": 1.114504098892212, "learning_rate": 0.0001, "loss": 0.0361, "step": 44120 }, { "epoch": 281.0828025477707, "grad_norm": 0.8765749931335449, "learning_rate": 0.0001, "loss": 0.0316, "step": 44130 }, { "epoch": 281.14649681528664, "grad_norm": 1.0860843658447266, "learning_rate": 0.0001, "loss": 0.0343, "step": 44140 }, { "epoch": 281.21019108280257, "grad_norm": 1.2273205518722534, "learning_rate": 0.0001, "loss": 0.0358, "step": 44150 }, { "epoch": 281.2738853503185, "grad_norm": 1.1715985536575317, "learning_rate": 0.0001, "loss": 0.0361, "step": 44160 }, { "epoch": 281.3375796178344, "grad_norm": 0.8709027767181396, "learning_rate": 0.0001, "loss": 0.0326, "step": 44170 }, { "epoch": 281.4012738853503, "grad_norm": 1.154150128364563, "learning_rate": 0.0001, "loss": 0.0389, "step": 44180 }, { "epoch": 281.4649681528662, "grad_norm": 0.9389222860336304, "learning_rate": 0.0001, "loss": 0.0337, "step": 44190 }, { "epoch": 281.52866242038215, "grad_norm": 1.1225155591964722, "learning_rate": 0.0001, "loss": 0.0364, "step": 44200 }, { "epoch": 281.5923566878981, "grad_norm": 1.0920051336288452, "learning_rate": 0.0001, "loss": 0.0361, "step": 44210 }, { "epoch": 281.656050955414, "grad_norm": 1.1205543279647827, "learning_rate": 0.0001, "loss": 0.0424, "step": 44220 }, { "epoch": 281.71974522292993, "grad_norm": 0.8762069940567017, "learning_rate": 0.0001, "loss": 0.0385, "step": 44230 }, { "epoch": 281.78343949044586, "grad_norm": 1.0423755645751953, "learning_rate": 0.0001, "loss": 0.0352, "step": 44240 }, { "epoch": 281.8471337579618, "grad_norm": 1.100990891456604, "learning_rate": 0.0001, "loss": 0.0377, "step": 44250 }, { "epoch": 281.9108280254777, "grad_norm": 1.0274600982666016, "learning_rate": 0.0001, "loss": 0.0368, "step": 44260 }, { "epoch": 281.97452229299364, "grad_norm": 1.095402479171753, "learning_rate": 0.0001, "loss": 0.0372, "step": 44270 }, { "epoch": 282.03821656050957, "grad_norm": 1.188027262687683, "learning_rate": 0.0001, "loss": 0.0331, "step": 44280 }, { "epoch": 282.1019108280255, "grad_norm": 1.0368901491165161, "learning_rate": 0.0001, "loss": 0.0331, "step": 44290 }, { "epoch": 282.1656050955414, "grad_norm": 0.9748605489730835, "learning_rate": 0.0001, "loss": 0.0324, "step": 44300 }, { "epoch": 282.22929936305735, "grad_norm": 1.2132242918014526, "learning_rate": 0.0001, "loss": 0.0339, "step": 44310 }, { "epoch": 282.2929936305732, "grad_norm": 1.1423362493515015, "learning_rate": 0.0001, "loss": 0.0334, "step": 44320 }, { "epoch": 282.35668789808915, "grad_norm": 0.922899603843689, "learning_rate": 0.0001, "loss": 0.0338, "step": 44330 }, { "epoch": 282.4203821656051, "grad_norm": 1.0938568115234375, "learning_rate": 0.0001, "loss": 0.0375, "step": 44340 }, { "epoch": 282.484076433121, "grad_norm": 1.094982624053955, "learning_rate": 0.0001, "loss": 0.0383, "step": 44350 }, { "epoch": 282.54777070063693, "grad_norm": 1.124700665473938, "learning_rate": 0.0001, "loss": 0.0362, "step": 44360 }, { "epoch": 282.61146496815286, "grad_norm": 1.1684857606887817, "learning_rate": 0.0001, "loss": 0.0371, "step": 44370 }, { "epoch": 282.6751592356688, "grad_norm": 1.1649283170700073, "learning_rate": 0.0001, "loss": 0.0384, "step": 44380 }, { "epoch": 282.7388535031847, "grad_norm": 1.2268637418746948, "learning_rate": 0.0001, "loss": 0.038, "step": 44390 }, { "epoch": 282.80254777070064, "grad_norm": 1.2884384393692017, "learning_rate": 0.0001, "loss": 0.0343, "step": 44400 }, { "epoch": 282.86624203821657, "grad_norm": 0.8679237961769104, "learning_rate": 0.0001, "loss": 0.0354, "step": 44410 }, { "epoch": 282.9299363057325, "grad_norm": 0.9741628170013428, "learning_rate": 0.0001, "loss": 0.0369, "step": 44420 }, { "epoch": 282.9936305732484, "grad_norm": 0.776404619216919, "learning_rate": 0.0001, "loss": 0.039, "step": 44430 }, { "epoch": 283.05732484076435, "grad_norm": 1.0487397909164429, "learning_rate": 0.0001, "loss": 0.0324, "step": 44440 }, { "epoch": 283.1210191082803, "grad_norm": 0.83967125415802, "learning_rate": 0.0001, "loss": 0.0394, "step": 44450 }, { "epoch": 283.18471337579615, "grad_norm": 1.2918636798858643, "learning_rate": 0.0001, "loss": 0.0385, "step": 44460 }, { "epoch": 283.2484076433121, "grad_norm": 1.1960738897323608, "learning_rate": 0.0001, "loss": 0.0369, "step": 44470 }, { "epoch": 283.312101910828, "grad_norm": 1.0860790014266968, "learning_rate": 0.0001, "loss": 0.0414, "step": 44480 }, { "epoch": 283.37579617834393, "grad_norm": 1.228253960609436, "learning_rate": 0.0001, "loss": 0.0432, "step": 44490 }, { "epoch": 283.43949044585986, "grad_norm": 1.3048975467681885, "learning_rate": 0.0001, "loss": 0.0384, "step": 44500 }, { "epoch": 283.5031847133758, "grad_norm": 0.9904842376708984, "learning_rate": 0.0001, "loss": 0.0397, "step": 44510 }, { "epoch": 283.5668789808917, "grad_norm": 1.054785966873169, "learning_rate": 0.0001, "loss": 0.0371, "step": 44520 }, { "epoch": 283.63057324840764, "grad_norm": 0.8634565472602844, "learning_rate": 0.0001, "loss": 0.0402, "step": 44530 }, { "epoch": 283.69426751592357, "grad_norm": 1.104156255722046, "learning_rate": 0.0001, "loss": 0.0372, "step": 44540 }, { "epoch": 283.7579617834395, "grad_norm": 1.3261520862579346, "learning_rate": 0.0001, "loss": 0.0367, "step": 44550 }, { "epoch": 283.8216560509554, "grad_norm": 1.1062259674072266, "learning_rate": 0.0001, "loss": 0.0386, "step": 44560 }, { "epoch": 283.88535031847135, "grad_norm": 1.0963363647460938, "learning_rate": 0.0001, "loss": 0.0385, "step": 44570 }, { "epoch": 283.9490445859873, "grad_norm": 1.2185121774673462, "learning_rate": 0.0001, "loss": 0.0348, "step": 44580 }, { "epoch": 284.0127388535032, "grad_norm": 1.2626385688781738, "learning_rate": 0.0001, "loss": 0.035, "step": 44590 }, { "epoch": 284.07643312101914, "grad_norm": 1.2622215747833252, "learning_rate": 0.0001, "loss": 0.0394, "step": 44600 }, { "epoch": 284.140127388535, "grad_norm": 0.8895390629768372, "learning_rate": 0.0001, "loss": 0.0379, "step": 44610 }, { "epoch": 284.20382165605093, "grad_norm": 1.1323559284210205, "learning_rate": 0.0001, "loss": 0.0348, "step": 44620 }, { "epoch": 284.26751592356686, "grad_norm": 0.848147988319397, "learning_rate": 0.0001, "loss": 0.0367, "step": 44630 }, { "epoch": 284.3312101910828, "grad_norm": 1.192083477973938, "learning_rate": 0.0001, "loss": 0.0331, "step": 44640 }, { "epoch": 284.3949044585987, "grad_norm": 0.9159651398658752, "learning_rate": 0.0001, "loss": 0.0375, "step": 44650 }, { "epoch": 284.45859872611464, "grad_norm": 0.8347543478012085, "learning_rate": 0.0001, "loss": 0.0369, "step": 44660 }, { "epoch": 284.5222929936306, "grad_norm": 1.074648141860962, "learning_rate": 0.0001, "loss": 0.0394, "step": 44670 }, { "epoch": 284.5859872611465, "grad_norm": 1.0881580114364624, "learning_rate": 0.0001, "loss": 0.0379, "step": 44680 }, { "epoch": 284.6496815286624, "grad_norm": 0.922605574131012, "learning_rate": 0.0001, "loss": 0.0372, "step": 44690 }, { "epoch": 284.71337579617835, "grad_norm": 1.0410199165344238, "learning_rate": 0.0001, "loss": 0.0374, "step": 44700 }, { "epoch": 284.7770700636943, "grad_norm": 0.9655277729034424, "learning_rate": 0.0001, "loss": 0.0369, "step": 44710 }, { "epoch": 284.8407643312102, "grad_norm": 0.9567978978157043, "learning_rate": 0.0001, "loss": 0.0354, "step": 44720 }, { "epoch": 284.90445859872614, "grad_norm": 1.4627861976623535, "learning_rate": 0.0001, "loss": 0.0382, "step": 44730 }, { "epoch": 284.96815286624206, "grad_norm": 1.8039450645446777, "learning_rate": 0.0001, "loss": 0.0372, "step": 44740 }, { "epoch": 285.03184713375794, "grad_norm": 1.3925288915634155, "learning_rate": 0.0001, "loss": 0.0377, "step": 44750 }, { "epoch": 285.09554140127386, "grad_norm": 1.6182067394256592, "learning_rate": 0.0001, "loss": 0.0357, "step": 44760 }, { "epoch": 285.1592356687898, "grad_norm": 1.5097450017929077, "learning_rate": 0.0001, "loss": 0.0355, "step": 44770 }, { "epoch": 285.2229299363057, "grad_norm": 1.451165795326233, "learning_rate": 0.0001, "loss": 0.0326, "step": 44780 }, { "epoch": 285.28662420382165, "grad_norm": 1.4841830730438232, "learning_rate": 0.0001, "loss": 0.0354, "step": 44790 }, { "epoch": 285.3503184713376, "grad_norm": 1.6146024465560913, "learning_rate": 0.0001, "loss": 0.033, "step": 44800 }, { "epoch": 285.4140127388535, "grad_norm": 1.1461505889892578, "learning_rate": 0.0001, "loss": 0.0357, "step": 44810 }, { "epoch": 285.4777070063694, "grad_norm": 1.1141712665557861, "learning_rate": 0.0001, "loss": 0.0328, "step": 44820 }, { "epoch": 285.54140127388536, "grad_norm": 1.1938540935516357, "learning_rate": 0.0001, "loss": 0.0337, "step": 44830 }, { "epoch": 285.6050955414013, "grad_norm": 1.367857813835144, "learning_rate": 0.0001, "loss": 0.038, "step": 44840 }, { "epoch": 285.6687898089172, "grad_norm": 1.6394901275634766, "learning_rate": 0.0001, "loss": 0.0389, "step": 44850 }, { "epoch": 285.73248407643314, "grad_norm": 2.1308364868164062, "learning_rate": 0.0001, "loss": 0.0339, "step": 44860 }, { "epoch": 285.79617834394907, "grad_norm": 2.1096949577331543, "learning_rate": 0.0001, "loss": 0.0354, "step": 44870 }, { "epoch": 285.859872611465, "grad_norm": 1.1539332866668701, "learning_rate": 0.0001, "loss": 0.0303, "step": 44880 }, { "epoch": 285.92356687898086, "grad_norm": 1.4240739345550537, "learning_rate": 0.0001, "loss": 0.0363, "step": 44890 }, { "epoch": 285.9872611464968, "grad_norm": 1.8088120222091675, "learning_rate": 0.0001, "loss": 0.0339, "step": 44900 }, { "epoch": 286.0509554140127, "grad_norm": 2.450813055038452, "learning_rate": 0.0001, "loss": 0.0372, "step": 44910 }, { "epoch": 286.11464968152865, "grad_norm": 1.6003942489624023, "learning_rate": 0.0001, "loss": 0.0318, "step": 44920 }, { "epoch": 286.1783439490446, "grad_norm": 1.3973393440246582, "learning_rate": 0.0001, "loss": 0.0324, "step": 44930 }, { "epoch": 286.2420382165605, "grad_norm": 1.424452781677246, "learning_rate": 0.0001, "loss": 0.0314, "step": 44940 }, { "epoch": 286.30573248407643, "grad_norm": 0.961066722869873, "learning_rate": 0.0001, "loss": 0.0332, "step": 44950 }, { "epoch": 286.36942675159236, "grad_norm": 1.229058861732483, "learning_rate": 0.0001, "loss": 0.031, "step": 44960 }, { "epoch": 286.4331210191083, "grad_norm": 1.4941835403442383, "learning_rate": 0.0001, "loss": 0.0351, "step": 44970 }, { "epoch": 286.4968152866242, "grad_norm": 1.1390705108642578, "learning_rate": 0.0001, "loss": 0.0364, "step": 44980 }, { "epoch": 286.56050955414014, "grad_norm": 1.525409460067749, "learning_rate": 0.0001, "loss": 0.0348, "step": 44990 }, { "epoch": 286.62420382165607, "grad_norm": 1.4153002500534058, "learning_rate": 0.0001, "loss": 0.0354, "step": 45000 }, { "epoch": 286.687898089172, "grad_norm": 1.1235291957855225, "learning_rate": 0.0001, "loss": 0.0353, "step": 45010 }, { "epoch": 286.7515923566879, "grad_norm": 1.1104594469070435, "learning_rate": 0.0001, "loss": 0.0358, "step": 45020 }, { "epoch": 286.81528662420385, "grad_norm": 1.0595121383666992, "learning_rate": 0.0001, "loss": 0.0326, "step": 45030 }, { "epoch": 286.8789808917197, "grad_norm": 1.256129264831543, "learning_rate": 0.0001, "loss": 0.0367, "step": 45040 }, { "epoch": 286.94267515923565, "grad_norm": 1.2272473573684692, "learning_rate": 0.0001, "loss": 0.0379, "step": 45050 }, { "epoch": 287.0063694267516, "grad_norm": 1.2711127996444702, "learning_rate": 0.0001, "loss": 0.034, "step": 45060 }, { "epoch": 287.0700636942675, "grad_norm": 1.2848252058029175, "learning_rate": 0.0001, "loss": 0.0336, "step": 45070 }, { "epoch": 287.13375796178343, "grad_norm": 1.4026614427566528, "learning_rate": 0.0001, "loss": 0.0359, "step": 45080 }, { "epoch": 287.19745222929936, "grad_norm": 1.1834582090377808, "learning_rate": 0.0001, "loss": 0.0339, "step": 45090 }, { "epoch": 287.2611464968153, "grad_norm": 1.1167091131210327, "learning_rate": 0.0001, "loss": 0.0314, "step": 45100 }, { "epoch": 287.3248407643312, "grad_norm": 1.1208134889602661, "learning_rate": 0.0001, "loss": 0.034, "step": 45110 }, { "epoch": 287.38853503184714, "grad_norm": 1.0402405261993408, "learning_rate": 0.0001, "loss": 0.0383, "step": 45120 }, { "epoch": 287.45222929936307, "grad_norm": 1.1601752042770386, "learning_rate": 0.0001, "loss": 0.034, "step": 45130 }, { "epoch": 287.515923566879, "grad_norm": 1.2098785638809204, "learning_rate": 0.0001, "loss": 0.0348, "step": 45140 }, { "epoch": 287.5796178343949, "grad_norm": 0.8887012600898743, "learning_rate": 0.0001, "loss": 0.0355, "step": 45150 }, { "epoch": 287.64331210191085, "grad_norm": 1.2331479787826538, "learning_rate": 0.0001, "loss": 0.0389, "step": 45160 }, { "epoch": 287.7070063694268, "grad_norm": 1.2905447483062744, "learning_rate": 0.0001, "loss": 0.04, "step": 45170 }, { "epoch": 287.77070063694265, "grad_norm": 1.1720672845840454, "learning_rate": 0.0001, "loss": 0.0368, "step": 45180 }, { "epoch": 287.8343949044586, "grad_norm": 1.0756138563156128, "learning_rate": 0.0001, "loss": 0.0353, "step": 45190 }, { "epoch": 287.8980891719745, "grad_norm": 1.3297829627990723, "learning_rate": 0.0001, "loss": 0.0376, "step": 45200 }, { "epoch": 287.96178343949043, "grad_norm": 0.9974574446678162, "learning_rate": 0.0001, "loss": 0.0361, "step": 45210 }, { "epoch": 288.02547770700636, "grad_norm": 1.1463783979415894, "learning_rate": 0.0001, "loss": 0.0349, "step": 45220 }, { "epoch": 288.0891719745223, "grad_norm": 0.900465190410614, "learning_rate": 0.0001, "loss": 0.0371, "step": 45230 }, { "epoch": 288.1528662420382, "grad_norm": 1.165389060974121, "learning_rate": 0.0001, "loss": 0.035, "step": 45240 }, { "epoch": 288.21656050955414, "grad_norm": 1.1833956241607666, "learning_rate": 0.0001, "loss": 0.0346, "step": 45250 }, { "epoch": 288.28025477707007, "grad_norm": 0.897674024105072, "learning_rate": 0.0001, "loss": 0.0336, "step": 45260 }, { "epoch": 288.343949044586, "grad_norm": 0.9867950081825256, "learning_rate": 0.0001, "loss": 0.0371, "step": 45270 }, { "epoch": 288.4076433121019, "grad_norm": 1.0691981315612793, "learning_rate": 0.0001, "loss": 0.0368, "step": 45280 }, { "epoch": 288.47133757961785, "grad_norm": 1.0799325704574585, "learning_rate": 0.0001, "loss": 0.037, "step": 45290 }, { "epoch": 288.5350318471338, "grad_norm": 1.2571431398391724, "learning_rate": 0.0001, "loss": 0.0361, "step": 45300 }, { "epoch": 288.5987261146497, "grad_norm": 1.1072733402252197, "learning_rate": 0.0001, "loss": 0.0378, "step": 45310 }, { "epoch": 288.6624203821656, "grad_norm": 1.0335721969604492, "learning_rate": 0.0001, "loss": 0.0368, "step": 45320 }, { "epoch": 288.7261146496815, "grad_norm": 1.059165120124817, "learning_rate": 0.0001, "loss": 0.0345, "step": 45330 }, { "epoch": 288.78980891719743, "grad_norm": 1.460705041885376, "learning_rate": 0.0001, "loss": 0.0396, "step": 45340 }, { "epoch": 288.85350318471336, "grad_norm": 1.1726692914962769, "learning_rate": 0.0001, "loss": 0.0319, "step": 45350 }, { "epoch": 288.9171974522293, "grad_norm": 1.3234169483184814, "learning_rate": 0.0001, "loss": 0.0342, "step": 45360 }, { "epoch": 288.9808917197452, "grad_norm": 1.1953760385513306, "learning_rate": 0.0001, "loss": 0.0351, "step": 45370 }, { "epoch": 289.04458598726114, "grad_norm": 1.1398710012435913, "learning_rate": 0.0001, "loss": 0.034, "step": 45380 }, { "epoch": 289.10828025477707, "grad_norm": 1.3252860307693481, "learning_rate": 0.0001, "loss": 0.0362, "step": 45390 }, { "epoch": 289.171974522293, "grad_norm": 1.125174880027771, "learning_rate": 0.0001, "loss": 0.0345, "step": 45400 }, { "epoch": 289.2356687898089, "grad_norm": 0.9995195269584656, "learning_rate": 0.0001, "loss": 0.0338, "step": 45410 }, { "epoch": 289.29936305732485, "grad_norm": 1.2484087944030762, "learning_rate": 0.0001, "loss": 0.0367, "step": 45420 }, { "epoch": 289.3630573248408, "grad_norm": 1.2294191122055054, "learning_rate": 0.0001, "loss": 0.0367, "step": 45430 }, { "epoch": 289.4267515923567, "grad_norm": 1.2595106363296509, "learning_rate": 0.0001, "loss": 0.0341, "step": 45440 }, { "epoch": 289.49044585987264, "grad_norm": 1.397071361541748, "learning_rate": 0.0001, "loss": 0.0379, "step": 45450 }, { "epoch": 289.5541401273885, "grad_norm": 1.1038715839385986, "learning_rate": 0.0001, "loss": 0.0375, "step": 45460 }, { "epoch": 289.61783439490443, "grad_norm": 1.2760542631149292, "learning_rate": 0.0001, "loss": 0.0373, "step": 45470 }, { "epoch": 289.68152866242036, "grad_norm": 0.960760235786438, "learning_rate": 0.0001, "loss": 0.0353, "step": 45480 }, { "epoch": 289.7452229299363, "grad_norm": 1.1412304639816284, "learning_rate": 0.0001, "loss": 0.039, "step": 45490 }, { "epoch": 289.8089171974522, "grad_norm": 1.256427526473999, "learning_rate": 0.0001, "loss": 0.0342, "step": 45500 }, { "epoch": 289.87261146496814, "grad_norm": 1.4177285432815552, "learning_rate": 0.0001, "loss": 0.0324, "step": 45510 }, { "epoch": 289.9363057324841, "grad_norm": 1.472516417503357, "learning_rate": 0.0001, "loss": 0.0343, "step": 45520 }, { "epoch": 290.0, "grad_norm": 1.3647425174713135, "learning_rate": 0.0001, "loss": 0.0343, "step": 45530 }, { "epoch": 290.0636942675159, "grad_norm": 1.2212821245193481, "learning_rate": 0.0001, "loss": 0.0373, "step": 45540 }, { "epoch": 290.12738853503186, "grad_norm": 1.1484074592590332, "learning_rate": 0.0001, "loss": 0.0326, "step": 45550 }, { "epoch": 290.1910828025478, "grad_norm": 0.9446930885314941, "learning_rate": 0.0001, "loss": 0.0335, "step": 45560 }, { "epoch": 290.2547770700637, "grad_norm": 0.8844623565673828, "learning_rate": 0.0001, "loss": 0.0375, "step": 45570 }, { "epoch": 290.31847133757964, "grad_norm": 1.0428314208984375, "learning_rate": 0.0001, "loss": 0.033, "step": 45580 }, { "epoch": 290.38216560509557, "grad_norm": 1.0005360841751099, "learning_rate": 0.0001, "loss": 0.0345, "step": 45590 }, { "epoch": 290.4458598726115, "grad_norm": 1.3928552865982056, "learning_rate": 0.0001, "loss": 0.0354, "step": 45600 }, { "epoch": 290.50955414012736, "grad_norm": 0.9504624009132385, "learning_rate": 0.0001, "loss": 0.0363, "step": 45610 }, { "epoch": 290.5732484076433, "grad_norm": 1.061607003211975, "learning_rate": 0.0001, "loss": 0.0341, "step": 45620 }, { "epoch": 290.6369426751592, "grad_norm": 1.120059609413147, "learning_rate": 0.0001, "loss": 0.0368, "step": 45630 }, { "epoch": 290.70063694267515, "grad_norm": 1.2262295484542847, "learning_rate": 0.0001, "loss": 0.0368, "step": 45640 }, { "epoch": 290.7643312101911, "grad_norm": 1.2056422233581543, "learning_rate": 0.0001, "loss": 0.0334, "step": 45650 }, { "epoch": 290.828025477707, "grad_norm": 1.042881965637207, "learning_rate": 0.0001, "loss": 0.0322, "step": 45660 }, { "epoch": 290.89171974522293, "grad_norm": 1.2565863132476807, "learning_rate": 0.0001, "loss": 0.0415, "step": 45670 }, { "epoch": 290.95541401273886, "grad_norm": 1.236675500869751, "learning_rate": 0.0001, "loss": 0.0332, "step": 45680 }, { "epoch": 291.0191082802548, "grad_norm": 1.3746179342269897, "learning_rate": 0.0001, "loss": 0.0347, "step": 45690 }, { "epoch": 291.0828025477707, "grad_norm": 1.9767720699310303, "learning_rate": 0.0001, "loss": 0.0323, "step": 45700 }, { "epoch": 291.14649681528664, "grad_norm": 1.3148441314697266, "learning_rate": 0.0001, "loss": 0.0331, "step": 45710 }, { "epoch": 291.21019108280257, "grad_norm": 1.2294303178787231, "learning_rate": 0.0001, "loss": 0.033, "step": 45720 }, { "epoch": 291.2738853503185, "grad_norm": 1.6239006519317627, "learning_rate": 0.0001, "loss": 0.0356, "step": 45730 }, { "epoch": 291.3375796178344, "grad_norm": 1.3223844766616821, "learning_rate": 0.0001, "loss": 0.0348, "step": 45740 }, { "epoch": 291.4012738853503, "grad_norm": 1.2532706260681152, "learning_rate": 0.0001, "loss": 0.0366, "step": 45750 }, { "epoch": 291.4649681528662, "grad_norm": 1.1771845817565918, "learning_rate": 0.0001, "loss": 0.0313, "step": 45760 }, { "epoch": 291.52866242038215, "grad_norm": 1.3910146951675415, "learning_rate": 0.0001, "loss": 0.0346, "step": 45770 }, { "epoch": 291.5923566878981, "grad_norm": 1.249388575553894, "learning_rate": 0.0001, "loss": 0.034, "step": 45780 }, { "epoch": 291.656050955414, "grad_norm": 0.9249176383018494, "learning_rate": 0.0001, "loss": 0.0356, "step": 45790 }, { "epoch": 291.71974522292993, "grad_norm": 1.3157418966293335, "learning_rate": 0.0001, "loss": 0.0342, "step": 45800 }, { "epoch": 291.78343949044586, "grad_norm": 0.9764295816421509, "learning_rate": 0.0001, "loss": 0.035, "step": 45810 }, { "epoch": 291.8471337579618, "grad_norm": 1.0099252462387085, "learning_rate": 0.0001, "loss": 0.0356, "step": 45820 }, { "epoch": 291.9108280254777, "grad_norm": 0.9718540906906128, "learning_rate": 0.0001, "loss": 0.0367, "step": 45830 }, { "epoch": 291.97452229299364, "grad_norm": 0.89156174659729, "learning_rate": 0.0001, "loss": 0.0369, "step": 45840 }, { "epoch": 292.03821656050957, "grad_norm": 1.0048847198486328, "learning_rate": 0.0001, "loss": 0.0362, "step": 45850 }, { "epoch": 292.1019108280255, "grad_norm": 0.7625566124916077, "learning_rate": 0.0001, "loss": 0.0353, "step": 45860 }, { "epoch": 292.1656050955414, "grad_norm": 1.1627241373062134, "learning_rate": 0.0001, "loss": 0.0365, "step": 45870 }, { "epoch": 292.22929936305735, "grad_norm": 1.1571664810180664, "learning_rate": 0.0001, "loss": 0.0386, "step": 45880 }, { "epoch": 292.2929936305732, "grad_norm": 1.041471004486084, "learning_rate": 0.0001, "loss": 0.0372, "step": 45890 }, { "epoch": 292.35668789808915, "grad_norm": 1.1702582836151123, "learning_rate": 0.0001, "loss": 0.0397, "step": 45900 }, { "epoch": 292.4203821656051, "grad_norm": 1.294015884399414, "learning_rate": 0.0001, "loss": 0.0367, "step": 45910 }, { "epoch": 292.484076433121, "grad_norm": 0.9624034762382507, "learning_rate": 0.0001, "loss": 0.0372, "step": 45920 }, { "epoch": 292.54777070063693, "grad_norm": 1.5555548667907715, "learning_rate": 0.0001, "loss": 0.035, "step": 45930 }, { "epoch": 292.61146496815286, "grad_norm": 1.238620638847351, "learning_rate": 0.0001, "loss": 0.0377, "step": 45940 }, { "epoch": 292.6751592356688, "grad_norm": 1.2116658687591553, "learning_rate": 0.0001, "loss": 0.0341, "step": 45950 }, { "epoch": 292.7388535031847, "grad_norm": 0.9636592864990234, "learning_rate": 0.0001, "loss": 0.0378, "step": 45960 }, { "epoch": 292.80254777070064, "grad_norm": 0.9468685388565063, "learning_rate": 0.0001, "loss": 0.0413, "step": 45970 }, { "epoch": 292.86624203821657, "grad_norm": 1.0234829187393188, "learning_rate": 0.0001, "loss": 0.0384, "step": 45980 }, { "epoch": 292.9299363057325, "grad_norm": 1.3461953401565552, "learning_rate": 0.0001, "loss": 0.0344, "step": 45990 }, { "epoch": 292.9936305732484, "grad_norm": 1.3401614427566528, "learning_rate": 0.0001, "loss": 0.0351, "step": 46000 }, { "epoch": 293.05732484076435, "grad_norm": 0.8736355304718018, "learning_rate": 0.0001, "loss": 0.0373, "step": 46010 }, { "epoch": 293.1210191082803, "grad_norm": 0.836724579334259, "learning_rate": 0.0001, "loss": 0.0345, "step": 46020 }, { "epoch": 293.18471337579615, "grad_norm": 1.2048605680465698, "learning_rate": 0.0001, "loss": 0.0334, "step": 46030 }, { "epoch": 293.2484076433121, "grad_norm": 1.1349681615829468, "learning_rate": 0.0001, "loss": 0.0376, "step": 46040 }, { "epoch": 293.312101910828, "grad_norm": 1.3366310596466064, "learning_rate": 0.0001, "loss": 0.0337, "step": 46050 }, { "epoch": 293.37579617834393, "grad_norm": 1.081712007522583, "learning_rate": 0.0001, "loss": 0.036, "step": 46060 }, { "epoch": 293.43949044585986, "grad_norm": 1.1636725664138794, "learning_rate": 0.0001, "loss": 0.0405, "step": 46070 }, { "epoch": 293.5031847133758, "grad_norm": 1.141408920288086, "learning_rate": 0.0001, "loss": 0.0356, "step": 46080 }, { "epoch": 293.5668789808917, "grad_norm": 0.8394553661346436, "learning_rate": 0.0001, "loss": 0.0338, "step": 46090 }, { "epoch": 293.63057324840764, "grad_norm": 0.9791378974914551, "learning_rate": 0.0001, "loss": 0.0348, "step": 46100 }, { "epoch": 293.69426751592357, "grad_norm": 1.0999194383621216, "learning_rate": 0.0001, "loss": 0.0356, "step": 46110 }, { "epoch": 293.7579617834395, "grad_norm": 1.0220917463302612, "learning_rate": 0.0001, "loss": 0.0357, "step": 46120 }, { "epoch": 293.8216560509554, "grad_norm": 1.2311331033706665, "learning_rate": 0.0001, "loss": 0.0346, "step": 46130 }, { "epoch": 293.88535031847135, "grad_norm": 1.1174674034118652, "learning_rate": 0.0001, "loss": 0.0357, "step": 46140 }, { "epoch": 293.9490445859873, "grad_norm": 0.9410785436630249, "learning_rate": 0.0001, "loss": 0.035, "step": 46150 }, { "epoch": 294.0127388535032, "grad_norm": 1.0542203187942505, "learning_rate": 0.0001, "loss": 0.0354, "step": 46160 }, { "epoch": 294.07643312101914, "grad_norm": 1.342514991760254, "learning_rate": 0.0001, "loss": 0.0319, "step": 46170 }, { "epoch": 294.140127388535, "grad_norm": 0.9326585531234741, "learning_rate": 0.0001, "loss": 0.0337, "step": 46180 }, { "epoch": 294.20382165605093, "grad_norm": 0.7697176337242126, "learning_rate": 0.0001, "loss": 0.0331, "step": 46190 }, { "epoch": 294.26751592356686, "grad_norm": 0.9877646565437317, "learning_rate": 0.0001, "loss": 0.038, "step": 46200 }, { "epoch": 294.3312101910828, "grad_norm": 1.1587224006652832, "learning_rate": 0.0001, "loss": 0.0357, "step": 46210 }, { "epoch": 294.3949044585987, "grad_norm": 0.7220644354820251, "learning_rate": 0.0001, "loss": 0.035, "step": 46220 }, { "epoch": 294.45859872611464, "grad_norm": 1.1561535596847534, "learning_rate": 0.0001, "loss": 0.0351, "step": 46230 }, { "epoch": 294.5222929936306, "grad_norm": 0.9013060927391052, "learning_rate": 0.0001, "loss": 0.0348, "step": 46240 }, { "epoch": 294.5859872611465, "grad_norm": 1.073850393295288, "learning_rate": 0.0001, "loss": 0.0364, "step": 46250 }, { "epoch": 294.6496815286624, "grad_norm": 1.233127474784851, "learning_rate": 0.0001, "loss": 0.0339, "step": 46260 }, { "epoch": 294.71337579617835, "grad_norm": 1.0443358421325684, "learning_rate": 0.0001, "loss": 0.0365, "step": 46270 }, { "epoch": 294.7770700636943, "grad_norm": 1.1024444103240967, "learning_rate": 0.0001, "loss": 0.0311, "step": 46280 }, { "epoch": 294.8407643312102, "grad_norm": 1.282793402671814, "learning_rate": 0.0001, "loss": 0.0353, "step": 46290 }, { "epoch": 294.90445859872614, "grad_norm": 1.3501722812652588, "learning_rate": 0.0001, "loss": 0.0347, "step": 46300 }, { "epoch": 294.96815286624206, "grad_norm": 0.9902558326721191, "learning_rate": 0.0001, "loss": 0.0367, "step": 46310 }, { "epoch": 295.03184713375794, "grad_norm": 1.1095466613769531, "learning_rate": 0.0001, "loss": 0.0369, "step": 46320 }, { "epoch": 295.09554140127386, "grad_norm": 1.4392274618148804, "learning_rate": 0.0001, "loss": 0.0382, "step": 46330 }, { "epoch": 295.1592356687898, "grad_norm": 1.2473170757293701, "learning_rate": 0.0001, "loss": 0.0339, "step": 46340 }, { "epoch": 295.2229299363057, "grad_norm": 1.303056240081787, "learning_rate": 0.0001, "loss": 0.0319, "step": 46350 }, { "epoch": 295.28662420382165, "grad_norm": 1.2540146112442017, "learning_rate": 0.0001, "loss": 0.033, "step": 46360 }, { "epoch": 295.3503184713376, "grad_norm": 1.033745288848877, "learning_rate": 0.0001, "loss": 0.0365, "step": 46370 }, { "epoch": 295.4140127388535, "grad_norm": 1.321531891822815, "learning_rate": 0.0001, "loss": 0.0351, "step": 46380 }, { "epoch": 295.4777070063694, "grad_norm": 1.1775898933410645, "learning_rate": 0.0001, "loss": 0.0348, "step": 46390 }, { "epoch": 295.54140127388536, "grad_norm": 1.017320156097412, "learning_rate": 0.0001, "loss": 0.03, "step": 46400 }, { "epoch": 295.6050955414013, "grad_norm": 0.8965663313865662, "learning_rate": 0.0001, "loss": 0.0376, "step": 46410 }, { "epoch": 295.6687898089172, "grad_norm": 1.4088455438613892, "learning_rate": 0.0001, "loss": 0.0336, "step": 46420 }, { "epoch": 295.73248407643314, "grad_norm": 1.063539981842041, "learning_rate": 0.0001, "loss": 0.0301, "step": 46430 }, { "epoch": 295.79617834394907, "grad_norm": 0.9126721024513245, "learning_rate": 0.0001, "loss": 0.0388, "step": 46440 }, { "epoch": 295.859872611465, "grad_norm": 1.1134536266326904, "learning_rate": 0.0001, "loss": 0.0344, "step": 46450 }, { "epoch": 295.92356687898086, "grad_norm": 1.235303282737732, "learning_rate": 0.0001, "loss": 0.0314, "step": 46460 }, { "epoch": 295.9872611464968, "grad_norm": 1.1775763034820557, "learning_rate": 0.0001, "loss": 0.0343, "step": 46470 }, { "epoch": 296.0509554140127, "grad_norm": 1.0674512386322021, "learning_rate": 0.0001, "loss": 0.0353, "step": 46480 }, { "epoch": 296.11464968152865, "grad_norm": 1.3645247220993042, "learning_rate": 0.0001, "loss": 0.0367, "step": 46490 }, { "epoch": 296.1783439490446, "grad_norm": 1.529172658920288, "learning_rate": 0.0001, "loss": 0.038, "step": 46500 }, { "epoch": 296.2420382165605, "grad_norm": 1.2496042251586914, "learning_rate": 0.0001, "loss": 0.0345, "step": 46510 }, { "epoch": 296.30573248407643, "grad_norm": 1.588055968284607, "learning_rate": 0.0001, "loss": 0.0395, "step": 46520 }, { "epoch": 296.36942675159236, "grad_norm": 1.966866135597229, "learning_rate": 0.0001, "loss": 0.0353, "step": 46530 }, { "epoch": 296.4331210191083, "grad_norm": 0.981648325920105, "learning_rate": 0.0001, "loss": 0.0337, "step": 46540 }, { "epoch": 296.4968152866242, "grad_norm": 1.1169226169586182, "learning_rate": 0.0001, "loss": 0.0312, "step": 46550 }, { "epoch": 296.56050955414014, "grad_norm": 1.0304183959960938, "learning_rate": 0.0001, "loss": 0.0339, "step": 46560 }, { "epoch": 296.62420382165607, "grad_norm": 0.9227129220962524, "learning_rate": 0.0001, "loss": 0.0345, "step": 46570 }, { "epoch": 296.687898089172, "grad_norm": 1.4222588539123535, "learning_rate": 0.0001, "loss": 0.0357, "step": 46580 }, { "epoch": 296.7515923566879, "grad_norm": 0.9788175225257874, "learning_rate": 0.0001, "loss": 0.0371, "step": 46590 }, { "epoch": 296.81528662420385, "grad_norm": 0.923803985118866, "learning_rate": 0.0001, "loss": 0.0352, "step": 46600 }, { "epoch": 296.8789808917197, "grad_norm": 0.9612699151039124, "learning_rate": 0.0001, "loss": 0.0355, "step": 46610 }, { "epoch": 296.94267515923565, "grad_norm": 1.0367647409439087, "learning_rate": 0.0001, "loss": 0.0376, "step": 46620 }, { "epoch": 297.0063694267516, "grad_norm": 1.021781086921692, "learning_rate": 0.0001, "loss": 0.0339, "step": 46630 }, { "epoch": 297.0700636942675, "grad_norm": 0.9812472462654114, "learning_rate": 0.0001, "loss": 0.0358, "step": 46640 }, { "epoch": 297.13375796178343, "grad_norm": 0.9852846264839172, "learning_rate": 0.0001, "loss": 0.0348, "step": 46650 }, { "epoch": 297.19745222929936, "grad_norm": 0.9852179288864136, "learning_rate": 0.0001, "loss": 0.0345, "step": 46660 }, { "epoch": 297.2611464968153, "grad_norm": 0.9588971734046936, "learning_rate": 0.0001, "loss": 0.0357, "step": 46670 }, { "epoch": 297.3248407643312, "grad_norm": 1.0587732791900635, "learning_rate": 0.0001, "loss": 0.0342, "step": 46680 }, { "epoch": 297.38853503184714, "grad_norm": 1.4860389232635498, "learning_rate": 0.0001, "loss": 0.0366, "step": 46690 }, { "epoch": 297.45222929936307, "grad_norm": 1.2985950708389282, "learning_rate": 0.0001, "loss": 0.0333, "step": 46700 }, { "epoch": 297.515923566879, "grad_norm": 1.255226969718933, "learning_rate": 0.0001, "loss": 0.0345, "step": 46710 }, { "epoch": 297.5796178343949, "grad_norm": 1.2541046142578125, "learning_rate": 0.0001, "loss": 0.0343, "step": 46720 }, { "epoch": 297.64331210191085, "grad_norm": 1.108913540840149, "learning_rate": 0.0001, "loss": 0.0335, "step": 46730 }, { "epoch": 297.7070063694268, "grad_norm": 0.9577028155326843, "learning_rate": 0.0001, "loss": 0.0396, "step": 46740 }, { "epoch": 297.77070063694265, "grad_norm": 1.0436091423034668, "learning_rate": 0.0001, "loss": 0.0363, "step": 46750 }, { "epoch": 297.8343949044586, "grad_norm": 1.4176102876663208, "learning_rate": 0.0001, "loss": 0.0347, "step": 46760 }, { "epoch": 297.8980891719745, "grad_norm": 1.1826629638671875, "learning_rate": 0.0001, "loss": 0.0388, "step": 46770 }, { "epoch": 297.96178343949043, "grad_norm": 0.9778571724891663, "learning_rate": 0.0001, "loss": 0.0357, "step": 46780 }, { "epoch": 298.02547770700636, "grad_norm": 0.9944453239440918, "learning_rate": 0.0001, "loss": 0.0373, "step": 46790 }, { "epoch": 298.0891719745223, "grad_norm": 0.9602555632591248, "learning_rate": 0.0001, "loss": 0.0334, "step": 46800 }, { "epoch": 298.1528662420382, "grad_norm": 1.3472695350646973, "learning_rate": 0.0001, "loss": 0.0334, "step": 46810 }, { "epoch": 298.21656050955414, "grad_norm": 1.1809178590774536, "learning_rate": 0.0001, "loss": 0.0358, "step": 46820 }, { "epoch": 298.28025477707007, "grad_norm": 1.3819297552108765, "learning_rate": 0.0001, "loss": 0.0357, "step": 46830 }, { "epoch": 298.343949044586, "grad_norm": 1.2943118810653687, "learning_rate": 0.0001, "loss": 0.035, "step": 46840 }, { "epoch": 298.4076433121019, "grad_norm": 1.2062207460403442, "learning_rate": 0.0001, "loss": 0.0324, "step": 46850 }, { "epoch": 298.47133757961785, "grad_norm": 1.1283020973205566, "learning_rate": 0.0001, "loss": 0.0404, "step": 46860 }, { "epoch": 298.5350318471338, "grad_norm": 1.3950250148773193, "learning_rate": 0.0001, "loss": 0.0376, "step": 46870 }, { "epoch": 298.5987261146497, "grad_norm": 1.181159257888794, "learning_rate": 0.0001, "loss": 0.0363, "step": 46880 }, { "epoch": 298.6624203821656, "grad_norm": 1.343131422996521, "learning_rate": 0.0001, "loss": 0.0359, "step": 46890 }, { "epoch": 298.7261146496815, "grad_norm": 1.1481035947799683, "learning_rate": 0.0001, "loss": 0.0367, "step": 46900 }, { "epoch": 298.78980891719743, "grad_norm": 1.375125527381897, "learning_rate": 0.0001, "loss": 0.0346, "step": 46910 }, { "epoch": 298.85350318471336, "grad_norm": 0.9401555061340332, "learning_rate": 0.0001, "loss": 0.0329, "step": 46920 }, { "epoch": 298.9171974522293, "grad_norm": 0.9395368695259094, "learning_rate": 0.0001, "loss": 0.0333, "step": 46930 }, { "epoch": 298.9808917197452, "grad_norm": 1.1747198104858398, "learning_rate": 0.0001, "loss": 0.0308, "step": 46940 }, { "epoch": 299.04458598726114, "grad_norm": 1.2521270513534546, "learning_rate": 0.0001, "loss": 0.0323, "step": 46950 }, { "epoch": 299.10828025477707, "grad_norm": 1.044223427772522, "learning_rate": 0.0001, "loss": 0.0332, "step": 46960 }, { "epoch": 299.171974522293, "grad_norm": 1.3313097953796387, "learning_rate": 0.0001, "loss": 0.0347, "step": 46970 }, { "epoch": 299.2356687898089, "grad_norm": 1.0436036586761475, "learning_rate": 0.0001, "loss": 0.0299, "step": 46980 }, { "epoch": 299.29936305732485, "grad_norm": 1.067168116569519, "learning_rate": 0.0001, "loss": 0.0344, "step": 46990 }, { "epoch": 299.3630573248408, "grad_norm": 1.2042415142059326, "learning_rate": 0.0001, "loss": 0.0348, "step": 47000 }, { "epoch": 299.4267515923567, "grad_norm": 1.1399540901184082, "learning_rate": 0.0001, "loss": 0.0354, "step": 47010 }, { "epoch": 299.49044585987264, "grad_norm": 1.102791666984558, "learning_rate": 0.0001, "loss": 0.0324, "step": 47020 }, { "epoch": 299.5541401273885, "grad_norm": 1.4291596412658691, "learning_rate": 0.0001, "loss": 0.0343, "step": 47030 }, { "epoch": 299.61783439490443, "grad_norm": 1.3367365598678589, "learning_rate": 0.0001, "loss": 0.04, "step": 47040 }, { "epoch": 299.68152866242036, "grad_norm": 0.9168161749839783, "learning_rate": 0.0001, "loss": 0.038, "step": 47050 }, { "epoch": 299.7452229299363, "grad_norm": 1.2150237560272217, "learning_rate": 0.0001, "loss": 0.0337, "step": 47060 }, { "epoch": 299.8089171974522, "grad_norm": 0.9768970012664795, "learning_rate": 0.0001, "loss": 0.0373, "step": 47070 }, { "epoch": 299.87261146496814, "grad_norm": 1.1120942831039429, "learning_rate": 0.0001, "loss": 0.0339, "step": 47080 }, { "epoch": 299.9363057324841, "grad_norm": 1.0430455207824707, "learning_rate": 0.0001, "loss": 0.0345, "step": 47090 }, { "epoch": 300.0, "grad_norm": 1.1116300821304321, "learning_rate": 0.0001, "loss": 0.0325, "step": 47100 }, { "epoch": 300.0636942675159, "grad_norm": 1.0305347442626953, "learning_rate": 0.0001, "loss": 0.0372, "step": 47110 }, { "epoch": 300.12738853503186, "grad_norm": 0.9036486148834229, "learning_rate": 0.0001, "loss": 0.035, "step": 47120 }, { "epoch": 300.1910828025478, "grad_norm": 1.1828255653381348, "learning_rate": 0.0001, "loss": 0.0352, "step": 47130 }, { "epoch": 300.2547770700637, "grad_norm": 1.0326695442199707, "learning_rate": 0.0001, "loss": 0.0365, "step": 47140 }, { "epoch": 300.31847133757964, "grad_norm": 1.0615900754928589, "learning_rate": 0.0001, "loss": 0.0327, "step": 47150 }, { "epoch": 300.38216560509557, "grad_norm": 0.9610784649848938, "learning_rate": 0.0001, "loss": 0.0377, "step": 47160 }, { "epoch": 300.4458598726115, "grad_norm": 1.0157966613769531, "learning_rate": 0.0001, "loss": 0.0376, "step": 47170 }, { "epoch": 300.50955414012736, "grad_norm": 0.9023928642272949, "learning_rate": 0.0001, "loss": 0.0364, "step": 47180 }, { "epoch": 300.5732484076433, "grad_norm": 0.8867015242576599, "learning_rate": 0.0001, "loss": 0.0381, "step": 47190 }, { "epoch": 300.6369426751592, "grad_norm": 1.1114941835403442, "learning_rate": 0.0001, "loss": 0.0364, "step": 47200 }, { "epoch": 300.70063694267515, "grad_norm": 1.405709981918335, "learning_rate": 0.0001, "loss": 0.0343, "step": 47210 }, { "epoch": 300.7643312101911, "grad_norm": 1.356150507926941, "learning_rate": 0.0001, "loss": 0.0362, "step": 47220 }, { "epoch": 300.828025477707, "grad_norm": 1.268028736114502, "learning_rate": 0.0001, "loss": 0.0374, "step": 47230 }, { "epoch": 300.89171974522293, "grad_norm": 1.1418602466583252, "learning_rate": 0.0001, "loss": 0.0358, "step": 47240 }, { "epoch": 300.95541401273886, "grad_norm": 1.0366029739379883, "learning_rate": 0.0001, "loss": 0.0345, "step": 47250 }, { "epoch": 301.0191082802548, "grad_norm": 0.9871407747268677, "learning_rate": 0.0001, "loss": 0.0381, "step": 47260 }, { "epoch": 301.0828025477707, "grad_norm": 0.9111390709877014, "learning_rate": 0.0001, "loss": 0.0314, "step": 47270 }, { "epoch": 301.14649681528664, "grad_norm": 1.419487714767456, "learning_rate": 0.0001, "loss": 0.0332, "step": 47280 }, { "epoch": 301.21019108280257, "grad_norm": 1.1123313903808594, "learning_rate": 0.0001, "loss": 0.0307, "step": 47290 }, { "epoch": 301.2738853503185, "grad_norm": 1.2041008472442627, "learning_rate": 0.0001, "loss": 0.0347, "step": 47300 }, { "epoch": 301.3375796178344, "grad_norm": 1.0298658609390259, "learning_rate": 0.0001, "loss": 0.0359, "step": 47310 }, { "epoch": 301.4012738853503, "grad_norm": 1.1538044214248657, "learning_rate": 0.0001, "loss": 0.0356, "step": 47320 }, { "epoch": 301.4649681528662, "grad_norm": 1.2592946290969849, "learning_rate": 0.0001, "loss": 0.0343, "step": 47330 }, { "epoch": 301.52866242038215, "grad_norm": 1.092138409614563, "learning_rate": 0.0001, "loss": 0.0383, "step": 47340 }, { "epoch": 301.5923566878981, "grad_norm": 1.2161915302276611, "learning_rate": 0.0001, "loss": 0.0336, "step": 47350 }, { "epoch": 301.656050955414, "grad_norm": 1.6076486110687256, "learning_rate": 0.0001, "loss": 0.0354, "step": 47360 }, { "epoch": 301.71974522292993, "grad_norm": 1.0695769786834717, "learning_rate": 0.0001, "loss": 0.0411, "step": 47370 }, { "epoch": 301.78343949044586, "grad_norm": 1.3006231784820557, "learning_rate": 0.0001, "loss": 0.0368, "step": 47380 }, { "epoch": 301.8471337579618, "grad_norm": 1.2884327173233032, "learning_rate": 0.0001, "loss": 0.0419, "step": 47390 }, { "epoch": 301.9108280254777, "grad_norm": 1.217578411102295, "learning_rate": 0.0001, "loss": 0.0336, "step": 47400 }, { "epoch": 301.97452229299364, "grad_norm": 0.9321362972259521, "learning_rate": 0.0001, "loss": 0.0337, "step": 47410 }, { "epoch": 302.03821656050957, "grad_norm": 1.0909852981567383, "learning_rate": 0.0001, "loss": 0.034, "step": 47420 }, { "epoch": 302.1019108280255, "grad_norm": 1.0290369987487793, "learning_rate": 0.0001, "loss": 0.0329, "step": 47430 }, { "epoch": 302.1656050955414, "grad_norm": 0.9294818639755249, "learning_rate": 0.0001, "loss": 0.0359, "step": 47440 }, { "epoch": 302.22929936305735, "grad_norm": 1.104468584060669, "learning_rate": 0.0001, "loss": 0.0373, "step": 47450 }, { "epoch": 302.2929936305732, "grad_norm": 1.3479137420654297, "learning_rate": 0.0001, "loss": 0.0323, "step": 47460 }, { "epoch": 302.35668789808915, "grad_norm": 1.3341991901397705, "learning_rate": 0.0001, "loss": 0.0323, "step": 47470 }, { "epoch": 302.4203821656051, "grad_norm": 0.840078592300415, "learning_rate": 0.0001, "loss": 0.0327, "step": 47480 }, { "epoch": 302.484076433121, "grad_norm": 1.1433727741241455, "learning_rate": 0.0001, "loss": 0.0384, "step": 47490 }, { "epoch": 302.54777070063693, "grad_norm": 1.1694775819778442, "learning_rate": 0.0001, "loss": 0.0323, "step": 47500 }, { "epoch": 302.61146496815286, "grad_norm": 1.1758345365524292, "learning_rate": 0.0001, "loss": 0.0357, "step": 47510 }, { "epoch": 302.6751592356688, "grad_norm": 1.0548443794250488, "learning_rate": 0.0001, "loss": 0.0355, "step": 47520 }, { "epoch": 302.7388535031847, "grad_norm": 1.0118440389633179, "learning_rate": 0.0001, "loss": 0.0368, "step": 47530 }, { "epoch": 302.80254777070064, "grad_norm": 1.0920729637145996, "learning_rate": 0.0001, "loss": 0.0347, "step": 47540 }, { "epoch": 302.86624203821657, "grad_norm": 1.2672810554504395, "learning_rate": 0.0001, "loss": 0.0362, "step": 47550 }, { "epoch": 302.9299363057325, "grad_norm": 1.239977240562439, "learning_rate": 0.0001, "loss": 0.0373, "step": 47560 }, { "epoch": 302.9936305732484, "grad_norm": 1.2086493968963623, "learning_rate": 0.0001, "loss": 0.0355, "step": 47570 }, { "epoch": 303.05732484076435, "grad_norm": 0.8762528896331787, "learning_rate": 0.0001, "loss": 0.0321, "step": 47580 }, { "epoch": 303.1210191082803, "grad_norm": 1.2911731004714966, "learning_rate": 0.0001, "loss": 0.035, "step": 47590 }, { "epoch": 303.18471337579615, "grad_norm": 1.1773478984832764, "learning_rate": 0.0001, "loss": 0.038, "step": 47600 }, { "epoch": 303.2484076433121, "grad_norm": 1.431656002998352, "learning_rate": 0.0001, "loss": 0.0372, "step": 47610 }, { "epoch": 303.312101910828, "grad_norm": 1.2295665740966797, "learning_rate": 0.0001, "loss": 0.041, "step": 47620 }, { "epoch": 303.37579617834393, "grad_norm": 1.4248936176300049, "learning_rate": 0.0001, "loss": 0.0369, "step": 47630 }, { "epoch": 303.43949044585986, "grad_norm": 1.1206456422805786, "learning_rate": 0.0001, "loss": 0.0339, "step": 47640 }, { "epoch": 303.5031847133758, "grad_norm": 1.1738200187683105, "learning_rate": 0.0001, "loss": 0.0363, "step": 47650 }, { "epoch": 303.5668789808917, "grad_norm": 0.9723765850067139, "learning_rate": 0.0001, "loss": 0.0351, "step": 47660 }, { "epoch": 303.63057324840764, "grad_norm": 1.2040776014328003, "learning_rate": 0.0001, "loss": 0.0346, "step": 47670 }, { "epoch": 303.69426751592357, "grad_norm": 1.1124205589294434, "learning_rate": 0.0001, "loss": 0.0368, "step": 47680 }, { "epoch": 303.7579617834395, "grad_norm": 1.244495153427124, "learning_rate": 0.0001, "loss": 0.0369, "step": 47690 }, { "epoch": 303.8216560509554, "grad_norm": 1.25987708568573, "learning_rate": 0.0001, "loss": 0.0351, "step": 47700 }, { "epoch": 303.88535031847135, "grad_norm": 1.0343005657196045, "learning_rate": 0.0001, "loss": 0.0365, "step": 47710 }, { "epoch": 303.9490445859873, "grad_norm": 1.1158092021942139, "learning_rate": 0.0001, "loss": 0.036, "step": 47720 }, { "epoch": 304.0127388535032, "grad_norm": 1.2512322664260864, "learning_rate": 0.0001, "loss": 0.0347, "step": 47730 }, { "epoch": 304.07643312101914, "grad_norm": 1.0461679697036743, "learning_rate": 0.0001, "loss": 0.0361, "step": 47740 }, { "epoch": 304.140127388535, "grad_norm": 1.1413750648498535, "learning_rate": 0.0001, "loss": 0.0357, "step": 47750 }, { "epoch": 304.20382165605093, "grad_norm": 1.3367133140563965, "learning_rate": 0.0001, "loss": 0.0351, "step": 47760 }, { "epoch": 304.26751592356686, "grad_norm": 1.1472598314285278, "learning_rate": 0.0001, "loss": 0.0371, "step": 47770 }, { "epoch": 304.3312101910828, "grad_norm": 1.4771578311920166, "learning_rate": 0.0001, "loss": 0.0368, "step": 47780 }, { "epoch": 304.3949044585987, "grad_norm": 1.294313669204712, "learning_rate": 0.0001, "loss": 0.035, "step": 47790 }, { "epoch": 304.45859872611464, "grad_norm": 1.2305539846420288, "learning_rate": 0.0001, "loss": 0.0335, "step": 47800 }, { "epoch": 304.5222929936306, "grad_norm": 1.1527963876724243, "learning_rate": 0.0001, "loss": 0.0361, "step": 47810 }, { "epoch": 304.5859872611465, "grad_norm": 1.2757527828216553, "learning_rate": 0.0001, "loss": 0.0381, "step": 47820 }, { "epoch": 304.6496815286624, "grad_norm": 1.1857426166534424, "learning_rate": 0.0001, "loss": 0.0399, "step": 47830 }, { "epoch": 304.71337579617835, "grad_norm": 1.0344887971878052, "learning_rate": 0.0001, "loss": 0.035, "step": 47840 }, { "epoch": 304.7770700636943, "grad_norm": 1.006388783454895, "learning_rate": 0.0001, "loss": 0.038, "step": 47850 }, { "epoch": 304.8407643312102, "grad_norm": 1.4530593156814575, "learning_rate": 0.0001, "loss": 0.0402, "step": 47860 }, { "epoch": 304.90445859872614, "grad_norm": 1.2591325044631958, "learning_rate": 0.0001, "loss": 0.0366, "step": 47870 }, { "epoch": 304.96815286624206, "grad_norm": 1.2994801998138428, "learning_rate": 0.0001, "loss": 0.0379, "step": 47880 }, { "epoch": 305.03184713375794, "grad_norm": 1.5258467197418213, "learning_rate": 0.0001, "loss": 0.0379, "step": 47890 }, { "epoch": 305.09554140127386, "grad_norm": 1.2959849834442139, "learning_rate": 0.0001, "loss": 0.0349, "step": 47900 }, { "epoch": 305.1592356687898, "grad_norm": 0.9066084027290344, "learning_rate": 0.0001, "loss": 0.0351, "step": 47910 }, { "epoch": 305.2229299363057, "grad_norm": 1.095852255821228, "learning_rate": 0.0001, "loss": 0.0327, "step": 47920 }, { "epoch": 305.28662420382165, "grad_norm": 1.2716748714447021, "learning_rate": 0.0001, "loss": 0.0321, "step": 47930 }, { "epoch": 305.3503184713376, "grad_norm": 1.0282198190689087, "learning_rate": 0.0001, "loss": 0.0352, "step": 47940 }, { "epoch": 305.4140127388535, "grad_norm": 1.104235291481018, "learning_rate": 0.0001, "loss": 0.0352, "step": 47950 }, { "epoch": 305.4777070063694, "grad_norm": 1.0737333297729492, "learning_rate": 0.0001, "loss": 0.0363, "step": 47960 }, { "epoch": 305.54140127388536, "grad_norm": 1.212580919265747, "learning_rate": 0.0001, "loss": 0.0312, "step": 47970 }, { "epoch": 305.6050955414013, "grad_norm": 1.045650839805603, "learning_rate": 0.0001, "loss": 0.0344, "step": 47980 }, { "epoch": 305.6687898089172, "grad_norm": 1.1025422811508179, "learning_rate": 0.0001, "loss": 0.0322, "step": 47990 }, { "epoch": 305.73248407643314, "grad_norm": 0.8958613872528076, "learning_rate": 0.0001, "loss": 0.0338, "step": 48000 }, { "epoch": 305.79617834394907, "grad_norm": 1.0599108934402466, "learning_rate": 0.0001, "loss": 0.0327, "step": 48010 }, { "epoch": 305.859872611465, "grad_norm": 1.1078335046768188, "learning_rate": 0.0001, "loss": 0.035, "step": 48020 }, { "epoch": 305.92356687898086, "grad_norm": 4.204675674438477, "learning_rate": 0.0001, "loss": 0.0394, "step": 48030 }, { "epoch": 305.9872611464968, "grad_norm": 1.236045241355896, "learning_rate": 0.0001, "loss": 0.0358, "step": 48040 }, { "epoch": 306.0509554140127, "grad_norm": 1.8702837228775024, "learning_rate": 0.0001, "loss": 0.0357, "step": 48050 }, { "epoch": 306.11464968152865, "grad_norm": 1.7252693176269531, "learning_rate": 0.0001, "loss": 0.0346, "step": 48060 }, { "epoch": 306.1783439490446, "grad_norm": 1.4821860790252686, "learning_rate": 0.0001, "loss": 0.0349, "step": 48070 }, { "epoch": 306.2420382165605, "grad_norm": 1.415016531944275, "learning_rate": 0.0001, "loss": 0.0316, "step": 48080 }, { "epoch": 306.30573248407643, "grad_norm": 1.3335000276565552, "learning_rate": 0.0001, "loss": 0.0335, "step": 48090 }, { "epoch": 306.36942675159236, "grad_norm": 1.450937032699585, "learning_rate": 0.0001, "loss": 0.0336, "step": 48100 }, { "epoch": 306.4331210191083, "grad_norm": 1.3859387636184692, "learning_rate": 0.0001, "loss": 0.0329, "step": 48110 }, { "epoch": 306.4968152866242, "grad_norm": 1.2422223091125488, "learning_rate": 0.0001, "loss": 0.0331, "step": 48120 }, { "epoch": 306.56050955414014, "grad_norm": 1.2955352067947388, "learning_rate": 0.0001, "loss": 0.0336, "step": 48130 }, { "epoch": 306.62420382165607, "grad_norm": 1.2765369415283203, "learning_rate": 0.0001, "loss": 0.0335, "step": 48140 }, { "epoch": 306.687898089172, "grad_norm": 1.7132853269577026, "learning_rate": 0.0001, "loss": 0.0319, "step": 48150 }, { "epoch": 306.7515923566879, "grad_norm": 1.5218452215194702, "learning_rate": 0.0001, "loss": 0.0345, "step": 48160 }, { "epoch": 306.81528662420385, "grad_norm": 1.2290292978286743, "learning_rate": 0.0001, "loss": 0.034, "step": 48170 }, { "epoch": 306.8789808917197, "grad_norm": 1.0981260538101196, "learning_rate": 0.0001, "loss": 0.0345, "step": 48180 }, { "epoch": 306.94267515923565, "grad_norm": 1.446760654449463, "learning_rate": 0.0001, "loss": 0.0366, "step": 48190 }, { "epoch": 307.0063694267516, "grad_norm": 1.2620530128479004, "learning_rate": 0.0001, "loss": 0.0327, "step": 48200 }, { "epoch": 307.0700636942675, "grad_norm": 1.228833794593811, "learning_rate": 0.0001, "loss": 0.0327, "step": 48210 }, { "epoch": 307.13375796178343, "grad_norm": 1.3371745347976685, "learning_rate": 0.0001, "loss": 0.0337, "step": 48220 }, { "epoch": 307.19745222929936, "grad_norm": 1.166153073310852, "learning_rate": 0.0001, "loss": 0.0344, "step": 48230 }, { "epoch": 307.2611464968153, "grad_norm": 1.02849280834198, "learning_rate": 0.0001, "loss": 0.0312, "step": 48240 }, { "epoch": 307.3248407643312, "grad_norm": 1.4104081392288208, "learning_rate": 0.0001, "loss": 0.0326, "step": 48250 }, { "epoch": 307.38853503184714, "grad_norm": 1.4119073152542114, "learning_rate": 0.0001, "loss": 0.0347, "step": 48260 }, { "epoch": 307.45222929936307, "grad_norm": 1.4535290002822876, "learning_rate": 0.0001, "loss": 0.0343, "step": 48270 }, { "epoch": 307.515923566879, "grad_norm": 1.6694965362548828, "learning_rate": 0.0001, "loss": 0.0336, "step": 48280 }, { "epoch": 307.5796178343949, "grad_norm": 1.7213605642318726, "learning_rate": 0.0001, "loss": 0.0324, "step": 48290 }, { "epoch": 307.64331210191085, "grad_norm": 1.3948047161102295, "learning_rate": 0.0001, "loss": 0.0355, "step": 48300 }, { "epoch": 307.7070063694268, "grad_norm": 1.4420909881591797, "learning_rate": 0.0001, "loss": 0.0353, "step": 48310 }, { "epoch": 307.77070063694265, "grad_norm": 1.195412278175354, "learning_rate": 0.0001, "loss": 0.0331, "step": 48320 }, { "epoch": 307.8343949044586, "grad_norm": 1.3075296878814697, "learning_rate": 0.0001, "loss": 0.0319, "step": 48330 }, { "epoch": 307.8980891719745, "grad_norm": 1.0050346851348877, "learning_rate": 0.0001, "loss": 0.0334, "step": 48340 }, { "epoch": 307.96178343949043, "grad_norm": 1.1327838897705078, "learning_rate": 0.0001, "loss": 0.0325, "step": 48350 }, { "epoch": 308.02547770700636, "grad_norm": 1.1480188369750977, "learning_rate": 0.0001, "loss": 0.0333, "step": 48360 }, { "epoch": 308.0891719745223, "grad_norm": 1.035345196723938, "learning_rate": 0.0001, "loss": 0.031, "step": 48370 }, { "epoch": 308.1528662420382, "grad_norm": 1.363608479499817, "learning_rate": 0.0001, "loss": 0.0334, "step": 48380 }, { "epoch": 308.21656050955414, "grad_norm": 1.656591534614563, "learning_rate": 0.0001, "loss": 0.0359, "step": 48390 }, { "epoch": 308.28025477707007, "grad_norm": 1.0461459159851074, "learning_rate": 0.0001, "loss": 0.0371, "step": 48400 }, { "epoch": 308.343949044586, "grad_norm": 1.470081090927124, "learning_rate": 0.0001, "loss": 0.0339, "step": 48410 }, { "epoch": 308.4076433121019, "grad_norm": 1.3069369792938232, "learning_rate": 0.0001, "loss": 0.0412, "step": 48420 }, { "epoch": 308.47133757961785, "grad_norm": 1.7521291971206665, "learning_rate": 0.0001, "loss": 0.033, "step": 48430 }, { "epoch": 308.5350318471338, "grad_norm": 1.255363941192627, "learning_rate": 0.0001, "loss": 0.0362, "step": 48440 }, { "epoch": 308.5987261146497, "grad_norm": 1.6075910329818726, "learning_rate": 0.0001, "loss": 0.0344, "step": 48450 }, { "epoch": 308.6624203821656, "grad_norm": 1.2553014755249023, "learning_rate": 0.0001, "loss": 0.0321, "step": 48460 }, { "epoch": 308.7261146496815, "grad_norm": 1.045088529586792, "learning_rate": 0.0001, "loss": 0.0323, "step": 48470 }, { "epoch": 308.78980891719743, "grad_norm": 1.4154258966445923, "learning_rate": 0.0001, "loss": 0.0337, "step": 48480 }, { "epoch": 308.85350318471336, "grad_norm": 1.1551027297973633, "learning_rate": 0.0001, "loss": 0.0352, "step": 48490 }, { "epoch": 308.9171974522293, "grad_norm": 1.300678014755249, "learning_rate": 0.0001, "loss": 0.0372, "step": 48500 }, { "epoch": 308.9808917197452, "grad_norm": 1.3895182609558105, "learning_rate": 0.0001, "loss": 0.0373, "step": 48510 }, { "epoch": 309.04458598726114, "grad_norm": 1.4432920217514038, "learning_rate": 0.0001, "loss": 0.033, "step": 48520 }, { "epoch": 309.10828025477707, "grad_norm": 1.0412057638168335, "learning_rate": 0.0001, "loss": 0.0326, "step": 48530 }, { "epoch": 309.171974522293, "grad_norm": 1.1883296966552734, "learning_rate": 0.0001, "loss": 0.0382, "step": 48540 }, { "epoch": 309.2356687898089, "grad_norm": 0.8670355081558228, "learning_rate": 0.0001, "loss": 0.0342, "step": 48550 }, { "epoch": 309.29936305732485, "grad_norm": 1.3442203998565674, "learning_rate": 0.0001, "loss": 0.0349, "step": 48560 }, { "epoch": 309.3630573248408, "grad_norm": 1.0714792013168335, "learning_rate": 0.0001, "loss": 0.0364, "step": 48570 }, { "epoch": 309.4267515923567, "grad_norm": 1.302787184715271, "learning_rate": 0.0001, "loss": 0.035, "step": 48580 }, { "epoch": 309.49044585987264, "grad_norm": 1.0723320245742798, "learning_rate": 0.0001, "loss": 0.0356, "step": 48590 }, { "epoch": 309.5541401273885, "grad_norm": 0.9095292687416077, "learning_rate": 0.0001, "loss": 0.0314, "step": 48600 }, { "epoch": 309.61783439490443, "grad_norm": 0.8456424474716187, "learning_rate": 0.0001, "loss": 0.0373, "step": 48610 }, { "epoch": 309.68152866242036, "grad_norm": 1.2786270380020142, "learning_rate": 0.0001, "loss": 0.032, "step": 48620 }, { "epoch": 309.7452229299363, "grad_norm": 1.334264874458313, "learning_rate": 0.0001, "loss": 0.0353, "step": 48630 }, { "epoch": 309.8089171974522, "grad_norm": 0.9341813921928406, "learning_rate": 0.0001, "loss": 0.0393, "step": 48640 }, { "epoch": 309.87261146496814, "grad_norm": 1.0123891830444336, "learning_rate": 0.0001, "loss": 0.0339, "step": 48650 }, { "epoch": 309.9363057324841, "grad_norm": 1.1442644596099854, "learning_rate": 0.0001, "loss": 0.0369, "step": 48660 }, { "epoch": 310.0, "grad_norm": 1.3304336071014404, "learning_rate": 0.0001, "loss": 0.0367, "step": 48670 }, { "epoch": 310.0636942675159, "grad_norm": 1.2431581020355225, "learning_rate": 0.0001, "loss": 0.0354, "step": 48680 }, { "epoch": 310.12738853503186, "grad_norm": 1.247581958770752, "learning_rate": 0.0001, "loss": 0.0319, "step": 48690 }, { "epoch": 310.1910828025478, "grad_norm": 1.0470401048660278, "learning_rate": 0.0001, "loss": 0.0331, "step": 48700 }, { "epoch": 310.2547770700637, "grad_norm": 1.0660183429718018, "learning_rate": 0.0001, "loss": 0.0323, "step": 48710 }, { "epoch": 310.31847133757964, "grad_norm": 1.3303393125534058, "learning_rate": 0.0001, "loss": 0.0367, "step": 48720 }, { "epoch": 310.38216560509557, "grad_norm": 1.0360931158065796, "learning_rate": 0.0001, "loss": 0.0383, "step": 48730 }, { "epoch": 310.4458598726115, "grad_norm": 1.139817476272583, "learning_rate": 0.0001, "loss": 0.0381, "step": 48740 }, { "epoch": 310.50955414012736, "grad_norm": 1.2348335981369019, "learning_rate": 0.0001, "loss": 0.0323, "step": 48750 }, { "epoch": 310.5732484076433, "grad_norm": 1.4372522830963135, "learning_rate": 0.0001, "loss": 0.0375, "step": 48760 }, { "epoch": 310.6369426751592, "grad_norm": 1.0911014080047607, "learning_rate": 0.0001, "loss": 0.0353, "step": 48770 }, { "epoch": 310.70063694267515, "grad_norm": 1.0122169256210327, "learning_rate": 0.0001, "loss": 0.0337, "step": 48780 }, { "epoch": 310.7643312101911, "grad_norm": 1.4188947677612305, "learning_rate": 0.0001, "loss": 0.0347, "step": 48790 }, { "epoch": 310.828025477707, "grad_norm": 1.1632888317108154, "learning_rate": 0.0001, "loss": 0.0375, "step": 48800 }, { "epoch": 310.89171974522293, "grad_norm": 0.9111810326576233, "learning_rate": 0.0001, "loss": 0.0337, "step": 48810 }, { "epoch": 310.95541401273886, "grad_norm": 1.1691757440567017, "learning_rate": 0.0001, "loss": 0.0324, "step": 48820 }, { "epoch": 311.0191082802548, "grad_norm": 1.0939053297042847, "learning_rate": 0.0001, "loss": 0.0367, "step": 48830 }, { "epoch": 311.0828025477707, "grad_norm": 1.1819531917572021, "learning_rate": 0.0001, "loss": 0.033, "step": 48840 }, { "epoch": 311.14649681528664, "grad_norm": 1.0932427644729614, "learning_rate": 0.0001, "loss": 0.0316, "step": 48850 }, { "epoch": 311.21019108280257, "grad_norm": 1.067057728767395, "learning_rate": 0.0001, "loss": 0.0336, "step": 48860 }, { "epoch": 311.2738853503185, "grad_norm": 1.6025292873382568, "learning_rate": 0.0001, "loss": 0.036, "step": 48870 }, { "epoch": 311.3375796178344, "grad_norm": 1.1858031749725342, "learning_rate": 0.0001, "loss": 0.0365, "step": 48880 }, { "epoch": 311.4012738853503, "grad_norm": 2.3019299507141113, "learning_rate": 0.0001, "loss": 0.0374, "step": 48890 }, { "epoch": 311.4649681528662, "grad_norm": 1.4602935314178467, "learning_rate": 0.0001, "loss": 0.0384, "step": 48900 }, { "epoch": 311.52866242038215, "grad_norm": 1.4051129817962646, "learning_rate": 0.0001, "loss": 0.0373, "step": 48910 }, { "epoch": 311.5923566878981, "grad_norm": 1.1651278734207153, "learning_rate": 0.0001, "loss": 0.035, "step": 48920 }, { "epoch": 311.656050955414, "grad_norm": 0.9352720379829407, "learning_rate": 0.0001, "loss": 0.0327, "step": 48930 }, { "epoch": 311.71974522292993, "grad_norm": 1.3282408714294434, "learning_rate": 0.0001, "loss": 0.036, "step": 48940 }, { "epoch": 311.78343949044586, "grad_norm": 1.0498236417770386, "learning_rate": 0.0001, "loss": 0.0353, "step": 48950 }, { "epoch": 311.8471337579618, "grad_norm": 1.0363422632217407, "learning_rate": 0.0001, "loss": 0.0315, "step": 48960 }, { "epoch": 311.9108280254777, "grad_norm": 1.279074788093567, "learning_rate": 0.0001, "loss": 0.0316, "step": 48970 }, { "epoch": 311.97452229299364, "grad_norm": 1.0561537742614746, "learning_rate": 0.0001, "loss": 0.0342, "step": 48980 }, { "epoch": 312.03821656050957, "grad_norm": 1.2100006341934204, "learning_rate": 0.0001, "loss": 0.0358, "step": 48990 }, { "epoch": 312.1019108280255, "grad_norm": 1.1512647867202759, "learning_rate": 0.0001, "loss": 0.0306, "step": 49000 }, { "epoch": 312.1656050955414, "grad_norm": 1.2147923707962036, "learning_rate": 0.0001, "loss": 0.0329, "step": 49010 }, { "epoch": 312.22929936305735, "grad_norm": 1.1334359645843506, "learning_rate": 0.0001, "loss": 0.0336, "step": 49020 }, { "epoch": 312.2929936305732, "grad_norm": 1.0030288696289062, "learning_rate": 0.0001, "loss": 0.0365, "step": 49030 }, { "epoch": 312.35668789808915, "grad_norm": 1.1298242807388306, "learning_rate": 0.0001, "loss": 0.0329, "step": 49040 }, { "epoch": 312.4203821656051, "grad_norm": 0.9809176921844482, "learning_rate": 0.0001, "loss": 0.0372, "step": 49050 }, { "epoch": 312.484076433121, "grad_norm": 0.9794339537620544, "learning_rate": 0.0001, "loss": 0.033, "step": 49060 }, { "epoch": 312.54777070063693, "grad_norm": 1.4330012798309326, "learning_rate": 0.0001, "loss": 0.0347, "step": 49070 }, { "epoch": 312.61146496815286, "grad_norm": 1.0071946382522583, "learning_rate": 0.0001, "loss": 0.0352, "step": 49080 }, { "epoch": 312.6751592356688, "grad_norm": 0.8900026082992554, "learning_rate": 0.0001, "loss": 0.0381, "step": 49090 }, { "epoch": 312.7388535031847, "grad_norm": 1.007394552230835, "learning_rate": 0.0001, "loss": 0.0341, "step": 49100 }, { "epoch": 312.80254777070064, "grad_norm": 0.8989470601081848, "learning_rate": 0.0001, "loss": 0.0323, "step": 49110 }, { "epoch": 312.86624203821657, "grad_norm": 1.221390724182129, "learning_rate": 0.0001, "loss": 0.0327, "step": 49120 }, { "epoch": 312.9299363057325, "grad_norm": 1.2223471403121948, "learning_rate": 0.0001, "loss": 0.0384, "step": 49130 }, { "epoch": 312.9936305732484, "grad_norm": 1.039230465888977, "learning_rate": 0.0001, "loss": 0.0363, "step": 49140 }, { "epoch": 313.05732484076435, "grad_norm": 0.8568423986434937, "learning_rate": 0.0001, "loss": 0.0333, "step": 49150 }, { "epoch": 313.1210191082803, "grad_norm": 1.1016603708267212, "learning_rate": 0.0001, "loss": 0.0334, "step": 49160 }, { "epoch": 313.18471337579615, "grad_norm": 1.0080175399780273, "learning_rate": 0.0001, "loss": 0.0351, "step": 49170 }, { "epoch": 313.2484076433121, "grad_norm": 1.1871936321258545, "learning_rate": 0.0001, "loss": 0.0375, "step": 49180 }, { "epoch": 313.312101910828, "grad_norm": 1.147308588027954, "learning_rate": 0.0001, "loss": 0.0354, "step": 49190 }, { "epoch": 313.37579617834393, "grad_norm": 0.8012629151344299, "learning_rate": 0.0001, "loss": 0.0312, "step": 49200 }, { "epoch": 313.43949044585986, "grad_norm": 1.199102759361267, "learning_rate": 0.0001, "loss": 0.033, "step": 49210 }, { "epoch": 313.5031847133758, "grad_norm": 1.3595844507217407, "learning_rate": 0.0001, "loss": 0.0331, "step": 49220 }, { "epoch": 313.5668789808917, "grad_norm": 1.226365327835083, "learning_rate": 0.0001, "loss": 0.0347, "step": 49230 }, { "epoch": 313.63057324840764, "grad_norm": 0.873022198677063, "learning_rate": 0.0001, "loss": 0.0365, "step": 49240 }, { "epoch": 313.69426751592357, "grad_norm": 1.0970237255096436, "learning_rate": 0.0001, "loss": 0.0354, "step": 49250 }, { "epoch": 313.7579617834395, "grad_norm": 1.1516809463500977, "learning_rate": 0.0001, "loss": 0.0375, "step": 49260 }, { "epoch": 313.8216560509554, "grad_norm": 0.7763618230819702, "learning_rate": 0.0001, "loss": 0.038, "step": 49270 }, { "epoch": 313.88535031847135, "grad_norm": 0.9229810237884521, "learning_rate": 0.0001, "loss": 0.0379, "step": 49280 }, { "epoch": 313.9490445859873, "grad_norm": 0.8201013803482056, "learning_rate": 0.0001, "loss": 0.033, "step": 49290 }, { "epoch": 314.0127388535032, "grad_norm": 1.0097019672393799, "learning_rate": 0.0001, "loss": 0.0351, "step": 49300 }, { "epoch": 314.07643312101914, "grad_norm": 1.1692396402359009, "learning_rate": 0.0001, "loss": 0.0384, "step": 49310 }, { "epoch": 314.140127388535, "grad_norm": 1.2315562963485718, "learning_rate": 0.0001, "loss": 0.034, "step": 49320 }, { "epoch": 314.20382165605093, "grad_norm": 1.0522371530532837, "learning_rate": 0.0001, "loss": 0.0348, "step": 49330 }, { "epoch": 314.26751592356686, "grad_norm": 0.9874677062034607, "learning_rate": 0.0001, "loss": 0.0319, "step": 49340 }, { "epoch": 314.3312101910828, "grad_norm": 0.9937770962715149, "learning_rate": 0.0001, "loss": 0.0356, "step": 49350 }, { "epoch": 314.3949044585987, "grad_norm": 1.066222906112671, "learning_rate": 0.0001, "loss": 0.0334, "step": 49360 }, { "epoch": 314.45859872611464, "grad_norm": 1.1111458539962769, "learning_rate": 0.0001, "loss": 0.0374, "step": 49370 }, { "epoch": 314.5222929936306, "grad_norm": 1.0088576078414917, "learning_rate": 0.0001, "loss": 0.0353, "step": 49380 }, { "epoch": 314.5859872611465, "grad_norm": 0.9628540873527527, "learning_rate": 0.0001, "loss": 0.0327, "step": 49390 }, { "epoch": 314.6496815286624, "grad_norm": 0.8684588074684143, "learning_rate": 0.0001, "loss": 0.0296, "step": 49400 }, { "epoch": 314.71337579617835, "grad_norm": 1.0356645584106445, "learning_rate": 0.0001, "loss": 0.037, "step": 49410 }, { "epoch": 314.7770700636943, "grad_norm": 0.8464203476905823, "learning_rate": 0.0001, "loss": 0.0366, "step": 49420 }, { "epoch": 314.8407643312102, "grad_norm": 1.0669236183166504, "learning_rate": 0.0001, "loss": 0.0312, "step": 49430 }, { "epoch": 314.90445859872614, "grad_norm": 1.3918564319610596, "learning_rate": 0.0001, "loss": 0.0356, "step": 49440 }, { "epoch": 314.96815286624206, "grad_norm": 1.315972924232483, "learning_rate": 0.0001, "loss": 0.0359, "step": 49450 }, { "epoch": 315.03184713375794, "grad_norm": 1.1071991920471191, "learning_rate": 0.0001, "loss": 0.0336, "step": 49460 }, { "epoch": 315.09554140127386, "grad_norm": 1.1564644575119019, "learning_rate": 0.0001, "loss": 0.0349, "step": 49470 }, { "epoch": 315.1592356687898, "grad_norm": 1.0838284492492676, "learning_rate": 0.0001, "loss": 0.0333, "step": 49480 }, { "epoch": 315.2229299363057, "grad_norm": 1.0941531658172607, "learning_rate": 0.0001, "loss": 0.037, "step": 49490 }, { "epoch": 315.28662420382165, "grad_norm": 1.2790648937225342, "learning_rate": 0.0001, "loss": 0.034, "step": 49500 }, { "epoch": 315.3503184713376, "grad_norm": 1.4616940021514893, "learning_rate": 0.0001, "loss": 0.035, "step": 49510 }, { "epoch": 315.4140127388535, "grad_norm": 1.245163917541504, "learning_rate": 0.0001, "loss": 0.0364, "step": 49520 }, { "epoch": 315.4777070063694, "grad_norm": 1.2621515989303589, "learning_rate": 0.0001, "loss": 0.0352, "step": 49530 }, { "epoch": 315.54140127388536, "grad_norm": 1.5803333520889282, "learning_rate": 0.0001, "loss": 0.0302, "step": 49540 }, { "epoch": 315.6050955414013, "grad_norm": 1.2366626262664795, "learning_rate": 0.0001, "loss": 0.033, "step": 49550 }, { "epoch": 315.6687898089172, "grad_norm": 1.173293113708496, "learning_rate": 0.0001, "loss": 0.0318, "step": 49560 }, { "epoch": 315.73248407643314, "grad_norm": 1.126599669456482, "learning_rate": 0.0001, "loss": 0.0317, "step": 49570 }, { "epoch": 315.79617834394907, "grad_norm": 1.38130521774292, "learning_rate": 0.0001, "loss": 0.0336, "step": 49580 }, { "epoch": 315.859872611465, "grad_norm": 1.069881796836853, "learning_rate": 0.0001, "loss": 0.0311, "step": 49590 }, { "epoch": 315.92356687898086, "grad_norm": 1.0378636121749878, "learning_rate": 0.0001, "loss": 0.0321, "step": 49600 }, { "epoch": 315.9872611464968, "grad_norm": 1.1320176124572754, "learning_rate": 0.0001, "loss": 0.0363, "step": 49610 }, { "epoch": 316.0509554140127, "grad_norm": 0.9722546935081482, "learning_rate": 0.0001, "loss": 0.0332, "step": 49620 }, { "epoch": 316.11464968152865, "grad_norm": 1.6010421514511108, "learning_rate": 0.0001, "loss": 0.0334, "step": 49630 }, { "epoch": 316.1783439490446, "grad_norm": 1.2439757585525513, "learning_rate": 0.0001, "loss": 0.0333, "step": 49640 }, { "epoch": 316.2420382165605, "grad_norm": 1.4671342372894287, "learning_rate": 0.0001, "loss": 0.0335, "step": 49650 }, { "epoch": 316.30573248407643, "grad_norm": 1.0604040622711182, "learning_rate": 0.0001, "loss": 0.0346, "step": 49660 }, { "epoch": 316.36942675159236, "grad_norm": 1.0519325733184814, "learning_rate": 0.0001, "loss": 0.0342, "step": 49670 }, { "epoch": 316.4331210191083, "grad_norm": 1.4003177881240845, "learning_rate": 0.0001, "loss": 0.0343, "step": 49680 }, { "epoch": 316.4968152866242, "grad_norm": 1.188480257987976, "learning_rate": 0.0001, "loss": 0.0342, "step": 49690 }, { "epoch": 316.56050955414014, "grad_norm": 1.2306877374649048, "learning_rate": 0.0001, "loss": 0.0352, "step": 49700 }, { "epoch": 316.62420382165607, "grad_norm": 1.237248182296753, "learning_rate": 0.0001, "loss": 0.0351, "step": 49710 }, { "epoch": 316.687898089172, "grad_norm": 1.189093828201294, "learning_rate": 0.0001, "loss": 0.0379, "step": 49720 }, { "epoch": 316.7515923566879, "grad_norm": 1.410282850265503, "learning_rate": 0.0001, "loss": 0.0342, "step": 49730 }, { "epoch": 316.81528662420385, "grad_norm": 1.1686707735061646, "learning_rate": 0.0001, "loss": 0.0326, "step": 49740 }, { "epoch": 316.8789808917197, "grad_norm": 1.6622321605682373, "learning_rate": 0.0001, "loss": 0.0335, "step": 49750 }, { "epoch": 316.94267515923565, "grad_norm": 1.2048001289367676, "learning_rate": 0.0001, "loss": 0.0343, "step": 49760 }, { "epoch": 317.0063694267516, "grad_norm": 1.3014142513275146, "learning_rate": 0.0001, "loss": 0.0329, "step": 49770 }, { "epoch": 317.0700636942675, "grad_norm": 1.2162941694259644, "learning_rate": 0.0001, "loss": 0.0341, "step": 49780 }, { "epoch": 317.13375796178343, "grad_norm": 1.302886724472046, "learning_rate": 0.0001, "loss": 0.0321, "step": 49790 }, { "epoch": 317.19745222929936, "grad_norm": 1.053026556968689, "learning_rate": 0.0001, "loss": 0.0338, "step": 49800 }, { "epoch": 317.2611464968153, "grad_norm": 0.7771125435829163, "learning_rate": 0.0001, "loss": 0.0329, "step": 49810 }, { "epoch": 317.3248407643312, "grad_norm": 1.4267258644104004, "learning_rate": 0.0001, "loss": 0.0334, "step": 49820 }, { "epoch": 317.38853503184714, "grad_norm": 1.0625871419906616, "learning_rate": 0.0001, "loss": 0.0389, "step": 49830 }, { "epoch": 317.45222929936307, "grad_norm": 1.2870330810546875, "learning_rate": 0.0001, "loss": 0.0306, "step": 49840 }, { "epoch": 317.515923566879, "grad_norm": 1.1287494897842407, "learning_rate": 0.0001, "loss": 0.0328, "step": 49850 }, { "epoch": 317.5796178343949, "grad_norm": 1.1679432392120361, "learning_rate": 0.0001, "loss": 0.0319, "step": 49860 }, { "epoch": 317.64331210191085, "grad_norm": 1.306862711906433, "learning_rate": 0.0001, "loss": 0.0335, "step": 49870 }, { "epoch": 317.7070063694268, "grad_norm": 0.8149302005767822, "learning_rate": 0.0001, "loss": 0.0378, "step": 49880 }, { "epoch": 317.77070063694265, "grad_norm": 1.1681735515594482, "learning_rate": 0.0001, "loss": 0.0358, "step": 49890 }, { "epoch": 317.8343949044586, "grad_norm": 0.9246296286582947, "learning_rate": 0.0001, "loss": 0.0337, "step": 49900 }, { "epoch": 317.8980891719745, "grad_norm": 1.327435851097107, "learning_rate": 0.0001, "loss": 0.0378, "step": 49910 }, { "epoch": 317.96178343949043, "grad_norm": 1.361509919166565, "learning_rate": 0.0001, "loss": 0.0359, "step": 49920 }, { "epoch": 318.02547770700636, "grad_norm": 1.177109718322754, "learning_rate": 0.0001, "loss": 0.0338, "step": 49930 }, { "epoch": 318.0891719745223, "grad_norm": 1.1442642211914062, "learning_rate": 0.0001, "loss": 0.0316, "step": 49940 }, { "epoch": 318.1528662420382, "grad_norm": 1.2398285865783691, "learning_rate": 0.0001, "loss": 0.0347, "step": 49950 }, { "epoch": 318.21656050955414, "grad_norm": 1.0016403198242188, "learning_rate": 0.0001, "loss": 0.0358, "step": 49960 }, { "epoch": 318.28025477707007, "grad_norm": 1.0298974514007568, "learning_rate": 0.0001, "loss": 0.034, "step": 49970 }, { "epoch": 318.343949044586, "grad_norm": 1.551386833190918, "learning_rate": 0.0001, "loss": 0.0307, "step": 49980 }, { "epoch": 318.4076433121019, "grad_norm": 1.1612610816955566, "learning_rate": 0.0001, "loss": 0.0364, "step": 49990 }, { "epoch": 318.47133757961785, "grad_norm": 1.3166825771331787, "learning_rate": 0.0001, "loss": 0.0347, "step": 50000 }, { "epoch": 318.47133757961785, "step": 50000, "total_flos": 0.0, "train_loss": 0.07033285598695278, "train_runtime": 14593.99, "train_samples_per_second": 438.537, "train_steps_per_second": 3.426 } ], "logging_steps": 10, "max_steps": 50000, "num_input_tokens_seen": 0, "num_train_epochs": 319, "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": 128, "trial_name": null, "trial_params": null }