{ "step": 101, "losses": [ 2.1343601644039154, 2.072067230939865, 1.8817282915115356, 1.5482478141784668, 2.0044714510440826, 1.9346624314785004, 2.0394379794597626, 2.0922091901302338, 1.9468883574008942, 1.7588119804859161, 1.874794840812683, 1.911599040031433, 1.6168376505374908, 1.7635800540447235, 2.070902556180954, 1.7358599305152893, 2.070700913667679, 1.9784598052501678, 1.9782940447330475, 1.8401454091072083, 1.6551836282014847, 1.9741482436656952, 1.990296185016632, 1.8002875596284866, 2.04819917678833, 1.8555984199047089, 2.1058743596076965, 1.9181069433689117, 2.1770740747451782, 2.1200993955135345, 1.5497342869639397, 2.041239619255066, 1.7999589145183563, 1.8640441000461578, 1.8068368434906006, 1.958029717206955, 2.3048327565193176, 2.0624635815620422, 1.7916592359542847, 2.197334885597229, 1.8518704771995544, 2.014060288667679, 2.3126999735832214, 2.0366491973400116, 2.1218522787094116, 1.9050887823104858, 2.140252709388733, 2.016432911157608, 2.0771953761577606, 1.99711474776268, 2.2010857462882996, 1.906881034374237, 1.9355229437351227, 1.568751234561205, 2.218832314014435, 2.282701849937439, 2.1470184922218323, 1.8618396520614624, 2.0435074865818024, 2.2046265304088593, 2.036127895116806, 1.9144421517848969, 1.7361707091331482, 1.8860499560832977, 1.9349249005317688, 2.091381222009659, 2.2041552662849426, 2.0161964893341064, 1.6289199590682983, 1.7889264225959778, 2.021867960691452, 2.1548608541488647, 1.7794556617736816, 1.9493914544582367, 1.8160816133022308, 1.8409814834594727, 2.2246702015399933, 2.080227106809616, 1.6188245043158531, 1.8832258582115173, 1.818663239479065, 2.0883366763591766, 2.00100439786911, 2.1166227757930756, 2.0359020829200745, 2.2903840243816376, 2.1877321600914, 1.8351092636585236, 2.1081435680389404, 1.876007080078125, 1.6549921333789825, 2.0958774387836456, 1.93923220038414, 1.8541749119758606, 2.41263547539711, 2.037635773420334, 2.0588943660259247, 1.969245046377182, 1.7285726368427277, 1.9336746335029602, 2.1948340237140656 ], "dataset_offset": 0, "config": { "king_dir": "/root/king_165_hf", "optimizer": "adamw", "lr": 5e-05, "muon_lr": 0.02, "lora_r": 256, "steps": 5000, "grad_accum": 4, "seq_len": 2048, "unfreeze_head": false, "source_dirs": [ "/root/datasets/finewebedu" ], "source_weights": [ 1.0 ] } }