| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.12008838505139782, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999999855889116, | |
| "loss": 3.4871, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999999357728057, | |
| "loss": 2.9969, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999998503737748, | |
| "loss": 2.9734, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999997293918308, | |
| "loss": 2.9203, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999995728269912, | |
| "loss": 2.7908, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999993806792782, | |
| "loss": 2.7973, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000499999152948719, | |
| "loss": 2.8511, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999988896353463, | |
| "loss": 2.8423, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999985907391973, | |
| "loss": 2.8411, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999982562603146, | |
| "loss": 2.838, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000499997886198746, | |
| "loss": 2.8751, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999974805545439, | |
| "loss": 2.7868, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999970393277663, | |
| "loss": 2.8003, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999965625184758, | |
| "loss": 2.8343, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999960501267404, | |
| "loss": 2.8053, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999955021526329, | |
| "loss": 2.7828, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999949185962313, | |
| "loss": 2.7822, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999942994576189, | |
| "loss": 2.7426, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999936447368836, | |
| "loss": 2.7808, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004999929544341185, | |
| "loss": 2.7522, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999922285494221, | |
| "loss": 2.8008, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999914670828975, | |
| "loss": 2.748, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999906700346533, | |
| "loss": 2.783, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999898374048027, | |
| "loss": 2.7879, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999889691934643, | |
| "loss": 2.696, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999880654007619, | |
| "loss": 2.7354, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999871260268238, | |
| "loss": 2.7336, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999861510717839, | |
| "loss": 2.7077, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999851405357809, | |
| "loss": 2.7408, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999840944189586, | |
| "loss": 2.7115, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000499983012721466, | |
| "loss": 2.6973, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999818954434569, | |
| "loss": 2.6644, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999807425850907, | |
| "loss": 2.7442, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000499979554146531, | |
| "loss": 2.7007, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999783301279471, | |
| "loss": 2.7284, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999770705295135, | |
| "loss": 2.6649, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999757753514091, | |
| "loss": 2.7269, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999744445938185, | |
| "loss": 2.64, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999730782569309, | |
| "loss": 2.6886, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000499971676340941, | |
| "loss": 2.6911, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999702388460482, | |
| "loss": 2.6814, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999687657724571, | |
| "loss": 2.7235, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999672571203774, | |
| "loss": 2.6898, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999657128900239, | |
| "loss": 2.6222, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999641330816162, | |
| "loss": 2.6827, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999625176953794, | |
| "loss": 2.6634, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999608667315432, | |
| "loss": 2.6976, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999591801903428, | |
| "loss": 2.6856, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000499957458072018, | |
| "loss": 2.624, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000499955700376814, | |
| "loss": 2.6693, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999539071049812, | |
| "loss": 2.7058, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999520782567746, | |
| "loss": 2.6878, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999502138324544, | |
| "loss": 2.6674, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999483138322863, | |
| "loss": 2.634, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999463782565405, | |
| "loss": 2.6242, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999444071054925, | |
| "loss": 2.636, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999424003794229, | |
| "loss": 2.6926, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999403580786175, | |
| "loss": 2.6564, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999382802033666, | |
| "loss": 2.6474, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999361667539663, | |
| "loss": 2.6462, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999340177307172, | |
| "loss": 2.6069, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004999318331339252, | |
| "loss": 2.632, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999296129639014, | |
| "loss": 2.6533, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999273572209616, | |
| "loss": 2.6524, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000499925065905427, | |
| "loss": 2.6229, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999227390176237, | |
| "loss": 2.6587, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999203765578828, | |
| "loss": 2.6175, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999179785265406, | |
| "loss": 2.6214, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999155449239384, | |
| "loss": 2.6338, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999130757504227, | |
| "loss": 2.6364, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999105710063449, | |
| "loss": 2.6509, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999080306920613, | |
| "loss": 2.6465, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999054548079338, | |
| "loss": 2.6452, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999028433543286, | |
| "loss": 2.6858, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004999001963316179, | |
| "loss": 2.6624, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998975137401781, | |
| "loss": 2.5747, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000499894795580391, | |
| "loss": 2.6239, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998920418526438, | |
| "loss": 2.612, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998892525573282, | |
| "loss": 2.6312, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998864276948413, | |
| "loss": 2.6321, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000499883567265585, | |
| "loss": 2.6128, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998806712699667, | |
| "loss": 2.6299, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998777397083982, | |
| "loss": 2.6444, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998747725812971, | |
| "loss": 2.6058, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998717698890857, | |
| "loss": 2.5646, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998687316321911, | |
| "loss": 2.6089, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998656578110461, | |
| "loss": 2.6257, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998625484260881, | |
| "loss": 2.6794, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998594034777594, | |
| "loss": 2.6585, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998562229665079, | |
| "loss": 2.6161, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998530068927862, | |
| "loss": 2.5684, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000499849755257052, | |
| "loss": 2.6284, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998464680597682, | |
| "loss": 2.5965, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998431453014028, | |
| "loss": 2.5825, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998397869824284, | |
| "loss": 2.6056, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998363931033231, | |
| "loss": 2.5434, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998329636645702, | |
| "loss": 2.6535, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998294986666576, | |
| "loss": 2.7084, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998259981100785, | |
| "loss": 2.6648, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998224619953312, | |
| "loss": 2.6241, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000499818890322919, | |
| "loss": 2.5672, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998152830933501, | |
| "loss": 2.6968, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998116403071381, | |
| "loss": 2.5603, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004998079619648014, | |
| "loss": 2.6031, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004998042480668638, | |
| "loss": 2.5779, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004998004986138534, | |
| "loss": 2.6792, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997967136063043, | |
| "loss": 2.6038, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997928930447551, | |
| "loss": 2.6469, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997890369297496, | |
| "loss": 2.62, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997851452618364, | |
| "loss": 2.6517, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997812180415697, | |
| "loss": 2.6313, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997772552695085, | |
| "loss": 2.62, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997732569462167, | |
| "loss": 2.6205, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997692230722633, | |
| "loss": 2.5727, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997651536482227, | |
| "loss": 2.5626, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997610486746738, | |
| "loss": 2.6196, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997569081522012, | |
| "loss": 2.6219, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997527320813939, | |
| "loss": 2.6443, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997485204628466, | |
| "loss": 2.6027, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997442732971584, | |
| "loss": 2.6495, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997399905849342, | |
| "loss": 2.6099, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997356723267833, | |
| "loss": 2.6074, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997313185233204, | |
| "loss": 2.6089, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997269291751651, | |
| "loss": 2.6539, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997225042829423, | |
| "loss": 2.6044, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997180438472816, | |
| "loss": 2.5768, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997135478688181, | |
| "loss": 2.5801, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997090163481916, | |
| "loss": 2.6145, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004997044492860471, | |
| "loss": 2.6338, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996998466830345, | |
| "loss": 2.6048, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996952085398091, | |
| "loss": 2.6023, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000499690534857031, | |
| "loss": 2.6037, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996858256353653, | |
| "loss": 2.5876, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996810808754824, | |
| "loss": 2.5932, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996763005780576, | |
| "loss": 2.6014, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996714847437712, | |
| "loss": 2.6211, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996666333733088, | |
| "loss": 2.6615, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996617464673609, | |
| "loss": 2.6057, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996568240266228, | |
| "loss": 2.5722, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996518660517954, | |
| "loss": 2.5588, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996468725435843, | |
| "loss": 2.612, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996418435027002, | |
| "loss": 2.5926, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996367789298589, | |
| "loss": 2.5635, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996316788257812, | |
| "loss": 2.5991, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004996265431911932, | |
| "loss": 2.6328, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004996213720268257, | |
| "loss": 2.5715, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004996161653334147, | |
| "loss": 2.5659, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004996109231117013, | |
| "loss": 2.5627, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004996056453624317, | |
| "loss": 2.5869, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004996003320863571, | |
| "loss": 2.6017, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995949832842337, | |
| "loss": 2.532, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995895989568227, | |
| "loss": 2.6129, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995841791048906, | |
| "loss": 2.6216, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995787237292089, | |
| "loss": 2.5878, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995732328305538, | |
| "loss": 2.5592, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000499567706409707, | |
| "loss": 2.637, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995621444674552, | |
| "loss": 2.571, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995565470045898, | |
| "loss": 2.5604, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995509140219076, | |
| "loss": 2.6172, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995452455202105, | |
| "loss": 2.6257, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995395415003051, | |
| "loss": 2.5337, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995338019630033, | |
| "loss": 2.5631, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000499528026909122, | |
| "loss": 2.5511, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995222163394834, | |
| "loss": 2.609, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995163702549142, | |
| "loss": 2.5627, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995104886562466, | |
| "loss": 2.5877, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004995045715443178, | |
| "loss": 2.5274, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994986189199701, | |
| "loss": 2.545, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994926307840505, | |
| "loss": 2.5648, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994866071374115, | |
| "loss": 2.5424, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994805479809103, | |
| "loss": 2.5759, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994744533154094, | |
| "loss": 2.6146, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994683231417762, | |
| "loss": 2.589, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994621574608833, | |
| "loss": 2.6288, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994559562736083, | |
| "loss": 2.5964, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994497195808336, | |
| "loss": 2.5697, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994434473834472, | |
| "loss": 2.5815, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994371396823416, | |
| "loss": 2.5621, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994307964784147, | |
| "loss": 2.5517, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994244177725693, | |
| "loss": 2.6157, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994180035657132, | |
| "loss": 2.5651, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994115538587595, | |
| "loss": 2.5868, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004994050686526261, | |
| "loss": 2.6103, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004993985479482362, | |
| "loss": 2.6364, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004993919917465178, | |
| "loss": 2.6651, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004993854000484038, | |
| "loss": 2.5453, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000499378772854833, | |
| "loss": 2.5657, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000499372110166748, | |
| "loss": 2.5375, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993654119850975, | |
| "loss": 2.5654, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993586783108349, | |
| "loss": 2.6298, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993519091449184, | |
| "loss": 2.5405, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993451044883116, | |
| "loss": 2.6205, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993382643419831, | |
| "loss": 2.5851, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993313887069062, | |
| "loss": 2.5528, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993244775840596, | |
| "loss": 2.5741, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993175309744273, | |
| "loss": 2.5482, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993105488789976, | |
| "loss": 2.5643, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004993035312987645, | |
| "loss": 2.5748, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992964782347269, | |
| "loss": 2.5976, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992893896878884, | |
| "loss": 2.5567, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000499282265659258, | |
| "loss": 2.516, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00049927510614985, | |
| "loss": 2.5344, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992679111606829, | |
| "loss": 2.6098, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992606806927813, | |
| "loss": 2.567, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992534147471738, | |
| "loss": 2.5139, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992461133248951, | |
| "loss": 2.6271, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992387764269841, | |
| "loss": 2.6127, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992314040544851, | |
| "loss": 2.5665, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992239962084475, | |
| "loss": 2.5802, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992165528899256, | |
| "loss": 2.5765, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000499209074099979, | |
| "loss": 2.5625, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004992015598396718, | |
| "loss": 2.5349, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991940101100739, | |
| "loss": 2.5717, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991864249122598, | |
| "loss": 2.5886, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991788042473088, | |
| "loss": 2.5742, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000499171148116306, | |
| "loss": 2.5869, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991634565203407, | |
| "loss": 2.5709, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000499155729460508, | |
| "loss": 2.6118, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991479669379076, | |
| "loss": 2.5576, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991401689536442, | |
| "loss": 2.5306, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991323355088279, | |
| "loss": 2.5807, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991244666045735, | |
| "loss": 2.5359, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991165622420011, | |
| "loss": 2.5796, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991086224222358, | |
| "loss": 2.5182, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004991006471464075, | |
| "loss": 2.6564, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004990926364156513, | |
| "loss": 2.638, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004990845902311077, | |
| "loss": 2.5452, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004990765085939216, | |
| "loss": 2.5622, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990683915052436, | |
| "loss": 2.5423, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990602389662287, | |
| "loss": 2.5891, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990520509780374, | |
| "loss": 2.5968, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990438275418351, | |
| "loss": 2.5594, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990355686587923, | |
| "loss": 2.5458, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990272743300843, | |
| "loss": 2.5304, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000499018944556892, | |
| "loss": 2.5505, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990105793404008, | |
| "loss": 2.6207, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004990021786818013, | |
| "loss": 2.6038, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000498993742582289, | |
| "loss": 2.5597, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989852710430651, | |
| "loss": 2.5344, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000498976764065335, | |
| "loss": 2.6021, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989682216503097, | |
| "loss": 2.5924, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989596437992049, | |
| "loss": 2.5742, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989510305132416, | |
| "loss": 2.5709, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989423817936458, | |
| "loss": 2.6258, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989336976416484, | |
| "loss": 2.5664, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989249780584853, | |
| "loss": 2.5551, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989162230453978, | |
| "loss": 2.5816, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004989074326036319, | |
| "loss": 2.5314, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988986067344388, | |
| "loss": 2.6238, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988897454390747, | |
| "loss": 2.6155, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988808487188009, | |
| "loss": 2.6048, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988719165748835, | |
| "loss": 2.5906, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000498862949008594, | |
| "loss": 2.5826, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988539460212088, | |
| "loss": 2.5686, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988449076140091, | |
| "loss": 2.5948, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988358337882816, | |
| "loss": 2.5445, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988267245453175, | |
| "loss": 2.5417, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988175798864137, | |
| "loss": 2.5712, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004988083998128714, | |
| "loss": 2.5913, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987991843259975, | |
| "loss": 2.5529, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987899334271035, | |
| "loss": 2.5581, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987806471175063, | |
| "loss": 2.5334, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987713253985273, | |
| "loss": 2.5911, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987619682714936, | |
| "loss": 2.5962, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987525757377368, | |
| "loss": 2.5038, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987431477985938, | |
| "loss": 2.5732, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987336844554066, | |
| "loss": 2.5667, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987241857095221, | |
| "loss": 2.5632, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987146515622921, | |
| "loss": 2.545, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0004987050820150739, | |
| "loss": 2.5234, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986954770692293, | |
| "loss": 2.5294, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986858367261256, | |
| "loss": 2.6312, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986761609871347, | |
| "loss": 2.5496, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986664498536339, | |
| "loss": 2.5203, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986567033270055, | |
| "loss": 2.6495, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986469214086364, | |
| "loss": 2.592, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986371040999192, | |
| "loss": 2.532, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986272514022512, | |
| "loss": 2.591, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986173633170346, | |
| "loss": 2.5353, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004986074398456769, | |
| "loss": 2.565, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985974809895903, | |
| "loss": 2.6095, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985874867501927, | |
| "loss": 2.5576, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985774571289062, | |
| "loss": 2.5559, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985673921271584, | |
| "loss": 2.5174, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985572917463821, | |
| "loss": 2.5949, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985471559880146, | |
| "loss": 2.5327, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985369848534988, | |
| "loss": 2.6106, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985267783442823, | |
| "loss": 2.6212, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985165364618176, | |
| "loss": 2.6182, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004985062592075627, | |
| "loss": 2.5897, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984959465829804, | |
| "loss": 2.5322, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984855985895382, | |
| "loss": 2.4541, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984752152287093, | |
| "loss": 2.5546, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984647965019714, | |
| "loss": 2.5689, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984543424108075, | |
| "loss": 2.5218, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984438529567054, | |
| "loss": 2.6078, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984333281411582, | |
| "loss": 2.5947, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000498422767965664, | |
| "loss": 2.5687, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984121724317256, | |
| "loss": 2.5415, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004984015415408514, | |
| "loss": 2.5362, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983908752945542, | |
| "loss": 2.521, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983801736943524, | |
| "loss": 2.5902, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000498369436741769, | |
| "loss": 2.565, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983586644383322, | |
| "loss": 2.5585, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983478567855754, | |
| "loss": 2.5787, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983370137850369, | |
| "loss": 2.5266, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983261354382598, | |
| "loss": 2.4995, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983152217467924, | |
| "loss": 2.5792, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004983042727121883, | |
| "loss": 2.5772, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004982932883360058, | |
| "loss": 2.4929, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004982822686198083, | |
| "loss": 2.508, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004982712135651643, | |
| "loss": 2.5573, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004982601231736472, | |
| "loss": 2.6541, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004982489974468355, | |
| "loss": 2.4863, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000498237836386313, | |
| "loss": 2.5646, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004982266399936678, | |
| "loss": 2.5657, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000498215408270494, | |
| "loss": 2.5708, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00049820414121839, | |
| "loss": 2.5032, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981928388389594, | |
| "loss": 2.5778, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000498181501133811, | |
| "loss": 2.5989, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981701281045585, | |
| "loss": 2.5047, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981587197528205, | |
| "loss": 2.5285, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000498147276080221, | |
| "loss": 2.5799, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981357970883886, | |
| "loss": 2.5478, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981242827789572, | |
| "loss": 2.5617, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981138897052764, | |
| "loss": 2.5479, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004981023082969261, | |
| "loss": 2.5556, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980906915757434, | |
| "loss": 2.5612, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980790395433817, | |
| "loss": 2.4746, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980673522014993, | |
| "loss": 2.5154, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00049805562955176, | |
| "loss": 2.5829, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980438715958318, | |
| "loss": 2.5546, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980320783353888, | |
| "loss": 2.6171, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980202497721091, | |
| "loss": 2.5568, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004980083859076766, | |
| "loss": 2.6162, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979964867437797, | |
| "loss": 2.5808, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979845522821121, | |
| "loss": 2.5665, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979725825243724, | |
| "loss": 2.5723, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979605774722644, | |
| "loss": 2.5039, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979485371274967, | |
| "loss": 2.5566, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979364614917831, | |
| "loss": 2.5481, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979243505668421, | |
| "loss": 2.545, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004979122043543979, | |
| "loss": 2.5013, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.000497900022856179, | |
| "loss": 2.5336, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978878060739191, | |
| "loss": 2.5577, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978755540093573, | |
| "loss": 2.5775, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978632666642374, | |
| "loss": 2.5011, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978509440403083, | |
| "loss": 2.5834, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978385861393237, | |
| "loss": 2.5143, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978261929630427, | |
| "loss": 2.574, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978137645132294, | |
| "loss": 2.5587, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004978013007916523, | |
| "loss": 2.5767, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0004977888018000858, | |
| "loss": 2.5945, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977762675403087, | |
| "loss": 2.5834, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977636980141051, | |
| "loss": 2.6018, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977510932232639, | |
| "loss": 2.5104, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977384531695794, | |
| "loss": 2.4966, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977257778548506, | |
| "loss": 2.5948, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977130672808815, | |
| "loss": 2.5277, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004977003214494813, | |
| "loss": 2.5563, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976875403624641, | |
| "loss": 2.5506, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976747240216491, | |
| "loss": 2.5865, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976618724288604, | |
| "loss": 2.583, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976489855859272, | |
| "loss": 2.5752, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976360634946838, | |
| "loss": 2.5702, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976231061569694, | |
| "loss": 2.6195, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004976101135746281, | |
| "loss": 2.5673, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975970857495093, | |
| "loss": 2.5204, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975840226834672, | |
| "loss": 2.5061, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975709243783612, | |
| "loss": 2.538, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975577908360554, | |
| "loss": 2.536, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975446220584195, | |
| "loss": 2.554, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975314180473274, | |
| "loss": 2.5227, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975181788046585, | |
| "loss": 2.5565, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004975049043322976, | |
| "loss": 2.5408, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004974915946321335, | |
| "loss": 2.5228, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000497478249706061, | |
| "loss": 2.6042, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004974648695559794, | |
| "loss": 2.5345, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000497451454183793, | |
| "loss": 2.5326, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004974380035914114, | |
| "loss": 2.5002, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004974245177807489, | |
| "loss": 2.5726, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004974109967537251, | |
| "loss": 2.5553, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973974405122644, | |
| "loss": 2.5504, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973838490582962, | |
| "loss": 2.5844, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973702223937552, | |
| "loss": 2.5305, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973565605205807, | |
| "loss": 2.5156, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973428634407174, | |
| "loss": 2.5365, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973291311561146, | |
| "loss": 2.519, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973153636687271, | |
| "loss": 2.5673, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004973015609805142, | |
| "loss": 2.5224, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004972877230934407, | |
| "loss": 2.4995, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.000497273850009476, | |
| "loss": 2.5978, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004972599417305948, | |
| "loss": 2.5886, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004972459982587765, | |
| "loss": 2.6197, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004972320195960059, | |
| "loss": 2.5968, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004972180057442726, | |
| "loss": 2.5245, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004972039567055711, | |
| "loss": 2.5286, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971898724819011, | |
| "loss": 2.5388, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971757530752671, | |
| "loss": 2.5169, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971615984876789, | |
| "loss": 2.5923, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971474087211511, | |
| "loss": 2.5335, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971331837777033, | |
| "loss": 2.5831, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971189236593603, | |
| "loss": 2.5268, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004971046283681515, | |
| "loss": 2.5427, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970902979061119, | |
| "loss": 2.5611, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000497075932275281, | |
| "loss": 2.6082, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970615314777035, | |
| "loss": 2.5046, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970470955154291, | |
| "loss": 2.5643, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970326243905126, | |
| "loss": 2.5789, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970181181050134, | |
| "loss": 2.4757, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004970035766609965, | |
| "loss": 2.4911, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969890000605315, | |
| "loss": 2.5294, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969743883056932, | |
| "loss": 2.4996, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969597413985612, | |
| "loss": 2.5432, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969450593412202, | |
| "loss": 2.5731, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969303421357601, | |
| "loss": 2.6248, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969155897842754, | |
| "loss": 2.5902, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004969008022888659, | |
| "loss": 2.4852, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004968859796516365, | |
| "loss": 2.4911, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004968711218746967, | |
| "loss": 2.5582, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004968562289601613, | |
| "loss": 2.5825, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004968413009101502, | |
| "loss": 2.5981, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.000496826337726788, | |
| "loss": 2.5896, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004968113394122044, | |
| "loss": 2.5474, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967963059685341, | |
| "loss": 2.5784, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967812373979169, | |
| "loss": 2.5398, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967661337024976, | |
| "loss": 2.5089, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967509948844258, | |
| "loss": 2.519, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967358209458564, | |
| "loss": 2.5411, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967206118889491, | |
| "loss": 2.5732, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004967053677158686, | |
| "loss": 2.5164, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966900884287845, | |
| "loss": 2.5663, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966747740298717, | |
| "loss": 2.5126, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966594245213098, | |
| "loss": 2.5704, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966440399052836, | |
| "loss": 2.5488, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966286201839828, | |
| "loss": 2.5444, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0004966131653596021, | |
| "loss": 2.5005, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004965976754343413, | |
| "loss": 2.5497, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000496582150410405, | |
| "loss": 2.6044, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004965665902900029, | |
| "loss": 2.6103, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004965509950753498, | |
| "loss": 2.4993, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004965353647686653, | |
| "loss": 2.4838, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004965196993721743, | |
| "loss": 2.4869, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000496503998888106, | |
| "loss": 2.5232, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964882633186955, | |
| "loss": 2.514, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964724926661825, | |
| "loss": 2.5312, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964566869328115, | |
| "loss": 2.578, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964408461208319, | |
| "loss": 2.557, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964249702324989, | |
| "loss": 2.4945, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004964090592700718, | |
| "loss": 2.5046, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004963931132358153, | |
| "loss": 2.5283, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000496377132131999, | |
| "loss": 2.5303, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004963611159608975, | |
| "loss": 2.4897, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004963450647247906, | |
| "loss": 2.5238, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004963289784259627, | |
| "loss": 2.5362, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004963128570667034, | |
| "loss": 2.5355, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962967006493074, | |
| "loss": 2.5849, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962805091760741, | |
| "loss": 2.5662, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962642826493082, | |
| "loss": 2.4892, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962480210713193, | |
| "loss": 2.5777, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962317244444217, | |
| "loss": 2.5342, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004962153927709353, | |
| "loss": 2.5293, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961990260531843, | |
| "loss": 2.5612, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961826242934983, | |
| "loss": 2.4989, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961661874942117, | |
| "loss": 2.557, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961497156576641, | |
| "loss": 2.542, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961332087862, | |
| "loss": 2.5551, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961166668821687, | |
| "loss": 2.5445, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004961000899479247, | |
| "loss": 2.5366, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004960834779858275, | |
| "loss": 2.5445, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004960668309982415, | |
| "loss": 2.5115, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000496050148987536, | |
| "loss": 2.562, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004960334319560854, | |
| "loss": 2.5237, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004960166799062691, | |
| "loss": 2.5659, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004959998928404715, | |
| "loss": 2.5704, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004959830707610819, | |
| "loss": 2.5654, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004959662136704945, | |
| "loss": 2.557, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004959493215711087, | |
| "loss": 2.4897, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004959323944653288, | |
| "loss": 2.572, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000495915432355564, | |
| "loss": 2.5766, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004958984352442286, | |
| "loss": 2.5121, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004958814031337419, | |
| "loss": 2.4934, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004958643360265279, | |
| "loss": 2.6128, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000495847233925016, | |
| "loss": 2.5246, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004958300968316402, | |
| "loss": 2.5577, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004958129247488397, | |
| "loss": 2.484, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004957957176790587, | |
| "loss": 2.5616, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004957784756247463, | |
| "loss": 2.4936, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004957611985883566, | |
| "loss": 2.58, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004957438865723486, | |
| "loss": 2.5083, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004957265395791865, | |
| "loss": 2.4819, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.000495709157611339, | |
| "loss": 2.5151, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956917406712805, | |
| "loss": 2.4644, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956742887614897, | |
| "loss": 2.5624, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956568018844507, | |
| "loss": 2.5754, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956392800426525, | |
| "loss": 2.5833, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956217232385888, | |
| "loss": 2.5116, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004956041314747587, | |
| "loss": 2.4878, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004955865047536658, | |
| "loss": 2.627, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004955688430778193, | |
| "loss": 2.4976, | |
| "step": 5000 | |
| } | |
| ], | |
| "max_steps": 83272, | |
| "num_train_epochs": 2, | |
| "total_flos": 4.13118422807321e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |