跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 2026-10-14AI 評分31

FSPO:用於策略最佳化的函數空間時間自蒸餾

Function-Space Temporal Self-Distillation for Policy Optimization

AI 導讀

FSPO 提出一種結合當前特權適應與函數空間時間整合的單策略框架,在近期的自蒸餾策略方法 SDPO 上取得顯著改進。它將稀疏的歷史下一個 token 分佈存為固定的時間目標,無需維持持續運作的教師模型。

正文

The paper shows a significant improvement over the recent self-distillation policy methods, SDPO.

This work introduces Function-Space Policy Optimization (FSPO), a single-policy framework that combines current privileged adaptation with function-space temporal consolidation. FSPO stores sparse historical next-token distributions as fixed temporal targets, providing temporal separation without maintaining a persistent teacher model.
Screenshot 2026-10-26 175827

Please feel free to discuss the paper if you find it interesting.

來源:HuggingFace Daily Papers(社區熱門論文) · huggingface.co