跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 2026-10-12AI 評分45

以語言正則化進行自我蒸餾

Self-Distillation with Verbal Regularization

AI 導讀

研究提出語言正則化自我蒸餾(VRSD),透過系統提示詞調節教師模型使用特權資訊,鼓勵重新思考與探索。在三個模型和四項任務中,VRSD 在 12 組模型與任務設定中有 11 組的 Pass@1 高於 OPSD,平均提升 3.9 個百分點,並普遍維持較高 token 熵。

正文

Published on Oct 12

·

Submitted by

J Li

on Oct 14

Authors:

,

,

,

,

,

,

,

,

Abstract

On-policy self-distillation (OPSD) trains a large language model using supervision from a teacher instantiated from the same model as the student. The teacher is conditioned on privileged information (PI), such as verified solutions. The student, which does not observe PI, learns from the teacher's next-token distributions along its own sampled responses. However, PI can bias supervision toward shortcut reasoning with limited verification or reconsideration, accompanied by a sharp drop in the entropy of the output distribution. To address this challenge, we introduce self-distillation with verbal regularization (VRSD), which uses system prompts to regulate how the teacher uses PI, thereby encouraging rethinking and exploration to counteract premature commitment. Theoretically, we model prompting as attenuating the influence of PI on the teacher distribution and derive conditions for entropy preservation. Across three models and four tasks, VRSD improves Pass@1 over OPSD in 11 of 12 model-task settings, by 3.9 percentage points on average, while generally maintaining higher token entropy.

View arXiv page View PDF GitHub Add to collection

Get this paper in your agent:

hf papers read 2610.16354

Don't have the latest CLI?

curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2610.16354 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2610.16354 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2610.16354 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.

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