{ "step": 101, "losses": [ 2.1343601644039154, 2.0724628567695618, 1.8829140663146973, 1.5485129058361053, 2.0023295283317566, 1.9346746504306793, 2.039268136024475, 2.092041403055191, 1.945346713066101, 1.7569688260555267, 1.8735843896865845, 1.9090708792209625, 1.6151848435401917, 1.7625981569290161, 2.068992555141449, 1.7361219227313995, 2.0706606209278107, 1.9792323410511017, 1.9772961139678955, 1.8365838527679443, 1.6523617506027222, 1.9725398421287537, 1.9878025949001312, 1.7896310240030289, 2.046401262283325, 1.8547812402248383, 2.1061335802078247, 1.9161535799503326, 2.1761361956596375, 2.1191433668136597, 1.5494413673877716, 2.041987180709839, 1.799335092306137, 1.8639665395021439, 1.8064149618148804, 1.9582627415657043, 2.3036161065101624, 2.061281830072403, 1.7895398437976837, 2.1972492933273315, 1.8500069081783295, 2.01457941532135, 2.311652898788452, 2.0377022325992584, 2.1203809678554535, 1.9050646126270294, 2.1400992274284363, 2.0162739157676697, 2.0760363936424255, 1.9967831671237946, 2.199576199054718, 1.903381735086441, 1.9348146319389343, 1.565977357327938, 2.216809928417206, 2.279655873775482, 2.146350234746933, 1.8592103123664856, 2.0420128405094147, 2.202657461166382, 2.0355475544929504, 1.9122264087200165, 1.7364317774772644, 1.8870865404605865, 1.9341412782669067, 2.0909554958343506, 2.204548269510269, 2.0142270624637604, 1.6295936107635498, 1.788179874420166, 2.022219479084015, 2.15500944852829, 1.7791253924369812, 1.9474472403526306, 1.8155687749385834, 1.8396632373332977, 2.2242822349071503, 2.0792854130268097, 1.6167261078953743, 1.8821334838867188, 1.8196536004543304, 2.087564706802368, 1.9993097186088562, 2.1168552935123444, 2.0352262556552887, 2.2895873188972473, 2.1865938305854797, 1.8330432772636414, 2.10781130194664, 1.875508338212967, 1.6532813906669617, 2.096515417098999, 1.939268171787262, 1.852915108203888, 2.4114430844783783, 2.036993145942688, 2.056860029697418, 1.9688847362995148, 1.729723185300827, 1.9320075511932373, 2.1951059699058533 ], "dataset_offset": 0, "config": { "king_dir": "/root/king_165_hf", "optimizer": "muon_adamw", "lr": 5e-05, "muon_lr": 0.0001, "lora_r": 128, "steps": 5000, "grad_accum": 4, "seq_len": 2048, "unfreeze_head": false, "source_dirs": [ "/root/datasets/finewebedu" ], "source_weights": [ 1.0 ] } }