Commit Verlauf

Autor SHA1 Nachricht Datum
  Tobi Lütke 189916d6fb Move GRPO training out of default finetune pipeline vor 3 Monaten
  Tobi Lutke 599935754b finetune: remove orphaned files and abandoned experiments vor 3 Monaten
  Tobi Lütke 102ff861d3 fix: use Qwen3 recommended sampling params to prevent repetition loops vor 3 Monaten
  Tobi Lütke bf1b8fc90a lots of training stuff vor 3 Monaten