best_eval_3/train_state.json

{
  "step": 201,
  "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,
    1.4296438321471214,
    1.8817992806434631,
    1.8943344950675964,
    1.779104083776474,
    1.9108014702796936,
    1.805834949016571,
    1.834848165512085,
    1.9906627535820007,
    2.1963523626327515,
    2.1539515554904938,
    1.9684762954711914,
    1.910222202539444,
    2.019984722137451,
    1.9977181553840637,
    1.6560729146003723,
    1.9717278629541397,
    1.8783657252788544,
    1.8279190361499786,
    2.0361652076244354,
    1.8888441324234009,
    2.0143093168735504,
    1.7985135614871979,
    2.2679185271263123,
    1.9247187972068787,
    1.7891883850097656,
    2.112279176712036,
    2.047613203525543,
    2.1257645189762115,
    2.0965375006198883,
    1.8606078326702118,
    1.986148178577423,
    2.235414505004883,
    2.0115409195423126,
    1.8105634152889252,
    2.055979162454605,
    1.8827663362026215,
    2.228433668613434,
    2.168084293603897,
    1.9350906312465668,
    2.1963738203048706,
    1.9545983374118805,
    2.0042108595371246,
    2.076897233724594,
    2.164249151945114,
    2.2511336505413055,
    2.147927701473236,
    1.666139394044876,
    1.9886944591999054,
    1.8915810883045197,
    1.9184699952602386,
    1.9324964582920074,
    2.2565895915031433,
    1.8644168674945831,
    2.024057060480118,
    2.1561132073402405,
    1.831732153892517,
    2.0300653278827667,
    1.8685192167758942,
    1.9508120715618134,
    2.217629134654999,
    1.7884540557861328,
    1.9629022479057312,
    2.014945089817047,
    1.773240178823471,
    1.8480553328990936,
    2.174328625202179,
    2.0528362691402435,
    1.9184516966342926,
    1.9994118809700012,
    2.1629433631896973,
    1.956313133239746,
    1.4914195835590363,
    2.0376154482364655,
    1.9383164942264557,
    1.8460872173309326,
    2.1424327194690704,
    2.047647774219513,
    2.094540774822235,
    1.3891073316335678,
    1.9750277996063232,
    2.0804302990436554,
    1.9623984098434448,
    1.7242400497198105,
    1.6598125100135803,
    2.072585940361023,
    2.003816694021225,
    1.7694512009620667,
    2.2525570392608643,
    1.8560697436332703,
    2.7015225291252136,
    2.017684131860733,
    1.930792212486267,
    1.9603004455566406,
    2.167401760816574,
    2.0757872462272644,
    2.1265901923179626,
    2.067963659763336,
    1.9692361950874329,
    2.0600834786891937,
    2.0879141092300415
  ],
  "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
    ]
  }
}