dsdr-dual-scale-diversity-regularization

【效率与协作】Improve LLM reasoning by promoting diversity at both trajectory and token levels simultaneously. Global (trajectory) scale rewards distinct correct solutions; local (token) scale applies entropy regularization per decision point. Dual-scale approach couples these via diversity-weighted allocation: solutions that are globally more distinctive receive …

独立 GitHub 快照 · 自动分类 · 以原作者文档为准