| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.18521948508983144, | |
| "global_step": 1000, | |
| "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 | |
| } | |
| ], | |
| "max_steps": 10798, | |
| "num_train_epochs": 2, | |
| "total_flos": 9.498667895656284e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |