5 Commits

Author SHA1 Message Date
Kye Gomez
eae0f04b8e fix training 2026-04-20 09:43:25 -04:00
Kye Gomez
7ba690797b [improvement][loguru-logging][replace print with loguru in training script][feat][ckpt-logging][add checkpoint start and success log events][docs][readme-optimizer][remove muon optimizer
reference][feat][train-requirements][add requirements txt to training folder]
2026-04-20 08:25:00 -04:00
Kye Gomez
18cca894dd [fix][rope Every decode token was stuck at position 0, so <q_decoded, k_cached> lost the (n - m) term entirely] 2026-04-20 08:19:14 -04:00
Kye Gomez
537b116b3e just use adam for now in training maybe add muon later 2026-04-19 23:34:58 -04:00
Kye Gomez
5ffb897dcf [feat][training-script][add 3b fineweb-edu training
script][feat][tokenizer][add MythosTokenizer class with encode
  decode][improvement][deps][add transformers and datasets
  dependencies][docs][readme-training][add training section with run
  commands][improvement][pyproject][pin torch and add new deps]
2026-04-19 22:48:30 -04:00