| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.3704389701796629, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0009999978838190456, | |
| "loss": 2.9794, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0009999915352940948, | |
| "loss": 2.3885, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000999980954478887, | |
| "loss": 2.3057, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000999966141462985, | |
| "loss": 2.2692, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.000999947096371777, | |
| "loss": 2.2576, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009999238193664748, | |
| "loss": 2.2388, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009998963106441117, | |
| "loss": 2.2523, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0009998645704375414, | |
| "loss": 2.218, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000999828599015436, | |
| "loss": 2.2457, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009997883966822835, | |
| "loss": 2.198, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009997439637783859, | |
| "loss": 2.2013, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.000999695300679855, | |
| "loss": 2.1765, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0009996424077986109, | |
| "loss": 2.1741, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000999585285582377, | |
| "loss": 2.1898, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009995239345146772, | |
| "loss": 2.1466, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009994583551148314, | |
| "loss": 2.1423, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0009993885479379506, | |
| "loss": 2.1451, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000999314513574934, | |
| "loss": 2.202, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009992362526524616, | |
| "loss": 2.1208, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009991537658329906, | |
| "loss": 2.1591, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000999067053814749, | |
| "loss": 2.1788, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0009989761173317304, | |
| "loss": 2.147, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000998880957153687, | |
| "loss": 2.1249, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.000998781574086123, | |
| "loss": 2.165, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000998677968970289, | |
| "loss": 2.1428, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009985701426831735, | |
| "loss": 2.1384, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009984580961374964, | |
| "loss": 2.1585, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009983418302817008, | |
| "loss": 2.1156, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0009982213460999448, | |
| "loss": 2.0811, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000998096644612094, | |
| "loss": 2.1081, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009979677268737118, | |
| "loss": 2.1246, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009978345939760515, | |
| "loss": 2.1229, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000997697247046046, | |
| "loss": 2.1033, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0009975556872462994, | |
| "loss": 2.0931, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000997409915775076, | |
| "loss": 2.1206, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009972599338662915, | |
| "loss": 2.0537, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009971057427895012, | |
| "loss": 2.0762, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009969473438498897, | |
| "loss": 2.0883, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009967847383882604, | |
| "loss": 2.1174, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0009966179277810239, | |
| "loss": 2.1111, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009964469134401855, | |
| "loss": 2.1288, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009962716968133346, | |
| "loss": 2.0967, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009960922793836318, | |
| "loss": 2.1216, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009959086626697955, | |
| "loss": 2.0924, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0009957208482260908, | |
| "loss": 2.0809, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009955288376423152, | |
| "loss": 2.1082, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009953326325437852, | |
| "loss": 2.0885, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009951322345913224, | |
| "loss": 2.1133, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009949276454812408, | |
| "loss": 2.0844, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00099471886694533, | |
| "loss": 2.0796, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0009945059007508434, | |
| "loss": 2.1255, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009942887487004804, | |
| "loss": 2.0913, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009940674126323733, | |
| "loss": 2.1003, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009938418944200709, | |
| "loss": 2.0541, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009936121959725223, | |
| "loss": 2.0523, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.0009933783192340618, | |
| "loss": 2.1225, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009931402661843911, | |
| "loss": 2.0446, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.000992898038838564, | |
| "loss": 2.0921, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009926516392469674, | |
| "loss": 2.1081, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009924010694953064, | |
| "loss": 2.0734, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009921463317045843, | |
| "loss": 2.0652, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0009918874280310862, | |
| "loss": 2.0818, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009916243606663605, | |
| "loss": 2.0776, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009913571318371994, | |
| "loss": 2.1025, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009910857438056215, | |
| "loss": 2.066, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009908101988688512, | |
| "loss": 2.0575, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0009905304993593008, | |
| "loss": 2.1269, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009902466476445486, | |
| "loss": 2.0518, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009899586461273218, | |
| "loss": 2.0698, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.000989666497245473, | |
| "loss": 2.0988, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009893702034719624, | |
| "loss": 2.0986, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0009890697673148345, | |
| "loss": 2.0237, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009887651913171986, | |
| "loss": 2.0027, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009884564780572064, | |
| "loss": 2.0563, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009881436301480305, | |
| "loss": 2.0624, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000987826650237842, | |
| "loss": 2.0926, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.000987505541009788, | |
| "loss": 2.0585, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0009871803051819696, | |
| "loss": 2.0494, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009868509455074183, | |
| "loss": 2.0106, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009865174647740729, | |
| "loss": 2.0861, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009861798658047556, | |
| "loss": 2.0478, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.0009858381514571484, | |
| "loss": 2.0469, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.000985492324623769, | |
| "loss": 2.0671, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009851423882319458, | |
| "loss": 2.0808, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009847883452437937, | |
| "loss": 2.0331, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009844301986561893, | |
| "loss": 2.0295, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.000984067951500744, | |
| "loss": 2.0873, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00098370160684378, | |
| "loss": 2.1038, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0009833311677863042, | |
| "loss": 2.0337, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009829566374639801, | |
| "loss": 2.0407, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009825780190471042, | |
| "loss": 2.1049, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000982195315740576, | |
| "loss": 2.0475, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0009818085307838741, | |
| "loss": 2.0624, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.000981417667451026, | |
| "loss": 2.0714, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009810227290505816, | |
| "loss": 2.0947, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009806237189255859, | |
| "loss": 2.0591, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009802206404535489, | |
| "loss": 2.0301, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.000979813497046419, | |
| "loss": 2.0556, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0009794022921505523, | |
| "loss": 2.0753, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000978987029246685, | |
| "loss": 2.0898, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009785677118499029, | |
| "loss": 2.0464, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009781443435096116, | |
| "loss": 2.0828, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009777169278095074, | |
| "loss": 2.1137, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0009772854683675462, | |
| "loss": 2.0167, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.000976849968835913, | |
| "loss": 2.07, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009764104329009909, | |
| "loss": 2.0409, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009759668642833304, | |
| "loss": 2.015, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009755192667376173, | |
| "loss": 2.0175, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009750676440526411, | |
| "loss": 2.0773, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.0009746120000512632, | |
| "loss": 2.0245, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009741523385903841, | |
| "loss": 2.094, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009736886635609112, | |
| "loss": 2.0506, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009732209788877258, | |
| "loss": 2.0287, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009727492885296489, | |
| "loss": 2.1162, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009722735964794099, | |
| "loss": 2.1096, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0009717939067636099, | |
| "loss": 2.0621, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009713102234426903, | |
| "loss": 2.0796, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009708225506108965, | |
| "loss": 2.0565, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009703308923962447, | |
| "loss": 2.0669, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009698352529604857, | |
| "loss": 2.0638, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0009693356364990705, | |
| "loss": 2.0358, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009688320472411143, | |
| "loss": 2.0859, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009683244894493613, | |
| "loss": 2.0932, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009678129674201479, | |
| "loss": 2.0129, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009672974854833669, | |
| "loss": 2.055, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0009667780480024304, | |
| "loss": 2.0665, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009662546593742334, | |
| "loss": 2.0488, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009657273240291159, | |
| "loss": 2.0543, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009651960464308261, | |
| "loss": 2.0418, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009646608310764819, | |
| "loss": 2.033, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009641216824965338, | |
| "loss": 2.1034, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0009635786052547253, | |
| "loss": 2.0866, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0009630316039480556, | |
| "loss": 2.0607, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0009624806832067394, | |
| "loss": 2.0457, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0009619258476941686, | |
| "loss": 2.032, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000961367102106873, | |
| "loss": 2.0519, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0009608044511744791, | |
| "loss": 2.0449, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0009602378996596721, | |
| "loss": 1.9949, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0009596674523581539, | |
| "loss": 2.0394, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0009590931140986035, | |
| "loss": 2.0386, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0009585148897426354, | |
| "loss": 2.0254, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0009579327841847593, | |
| "loss": 2.0238, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.000957346802352338, | |
| "loss": 2.0509, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0009567569492055456, | |
| "loss": 2.0004, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0009561632297373263, | |
| "loss": 2.0203, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0009555656489733513, | |
| "loss": 2.0182, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.000954964211971977, | |
| "loss": 1.9754, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.0009543589238242012, | |
| "loss": 2.0374, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000953749789653621, | |
| "loss": 2.0367, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.000953136814616389, | |
| "loss": 2.0866, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0009525200039011694, | |
| "loss": 2.0083, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0009518993627290948, | |
| "loss": 2.0525, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.0009512748963537212, | |
| "loss": 2.0636, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.000950646610060984, | |
| "loss": 2.0522, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0009500145091691532, | |
| "loss": 2.05, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0009493785990287882, | |
| "loss": 1.9887, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0009487388850226926, | |
| "loss": 2.0309, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.000948095372565869, | |
| "loss": 1.9954, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.0009474480671054726, | |
| "loss": 2.0078, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0009467969741207652, | |
| "loss": 2.0395, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0009461420991230693, | |
| "loss": 2.0415, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0009454834476557207, | |
| "loss": 2.0308, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0009448210252940223, | |
| "loss": 2.0826, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0009441548376451963, | |
| "loss": 2.0424, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009434848903483373, | |
| "loss": 2.0125, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009428111890743639, | |
| "loss": 2.0139, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009421337395259717, | |
| "loss": 2.0682, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009414525474375837, | |
| "loss": 2.0577, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009407676185753029, | |
| "loss": 2.0262, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0009400789587368632, | |
| "loss": 2.0515, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0009393865737515794, | |
| "loss": 2.0398, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0009386904694802997, | |
| "loss": 2.0146, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0009379906518153543, | |
| "loss": 2.0438, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0009372871266805063, | |
| "loss": 2.0377, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.000936579900030902, | |
| "loss": 2.0789, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0009358689778530193, | |
| "loss": 2.0201, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0009351543661646185, | |
| "loss": 2.0114, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0009344360710146898, | |
| "loss": 2.0242, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0009337140984834034, | |
| "loss": 2.0436, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.0009329884546820572, | |
| "loss": 2.0452, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.000932259145753026, | |
| "loss": 2.0254, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0009315261778697083, | |
| "loss": 2.0409, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0009307895572364746, | |
| "loss": 2.0301, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0009300492900886154, | |
| "loss": 2.0078, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0009293053826922873, | |
| "loss": 1.9851, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.0009285578413444613, | |
| "loss": 1.9947, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0009278066723728682, | |
| "loss": 2.0331, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0009270518821359461, | |
| "loss": 2.0058, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0009262934770227858, | |
| "loss": 2.05, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0009255314634530771, | |
| "loss": 2.0444, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0009247658478770543, | |
| "loss": 2.0045, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.000923996636775442, | |
| "loss": 2.0211, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0009232238366593997, | |
| "loss": 2.0124, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0009224474540704671, | |
| "loss": 2.0067, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0009216674955805079, | |
| "loss": 2.0247, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0009208839677916557, | |
| "loss": 2.0314, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0009200968773362568, | |
| "loss": 2.067, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0009193062308768145, | |
| "loss": 2.0168, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0009185120351059326, | |
| "loss": 2.0649, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0009177142967462591, | |
| "loss": 2.0208, | |
| "step": 2000 | |
| } | |
| ], | |
| "max_steps": 10798, | |
| "num_train_epochs": 2, | |
| "total_flos": 1.8983090426321306e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |