跳到正文
HuggingFace Daily Papers(社區熱門論文)·· 2026-10-14AI 評分43

CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation

CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation

AI 導讀

CoTranslator 將反向推理軌跡轉換為符合已驗證正向推理分佈的內容,支援高效、可遷移的學生模型推理蒸餾。在六個數學基準測試中,六個學生模型的平均準確率較語義抑制法最高提升 5.82 個百分點;兩個 4B 學生模型保留拒絕採樣平均準確率的 95.0%,估算運算量則為其 35.9%。

正文

Published on Oct 14

Authors:

,

,

,

,

,

,

Abstract

CoTranslator translates reverse reasoning traces to match verified forward reasoning distributions, enabling efficient and transferable reasoning distillation for student models.

Reasoning distillation requires chain-of-thought (CoT) data with correct answers, but obtaining such data efficiently remains challenging. Forward CoT generation (FCG) followed by rejection sampling (RS) produces verified, question-only responses, yet may require many teacher calls for difficult problems. Reverse CoT generation (RCG) is cheaper because it conditions on the known answer, but access to this answer can lead to post-hoc rationalization and a mismatch with the student's answer-free inference setting. We propose CoTranslator, which learns to translate reverse CoT responses toward the distribution of verified forward responses. Trained on a small set of paired reverse and rejection-sampled forward responses, it supports large-scale generation and verification of student training data. Across six student models evaluated on six mathematics benchmarks, CoTranslator improves average accuracy over semantic suppression by up to 5.82 percentage points. For two 4B students, it retains 95.0% of rejection sampling's average accuracy while using 35.9% of its estimated compute. Without retraining, the same translator transfers to a new mathematics dataset, a different teacher, and STEM problems, improving average STEM accuracy by 4.88 and 4.31 percentage points for 4B and 8B students, respectively. These results show that CoTranslator offers an efficient and transferable approach to reasoning distillation.

View arXiv page View PDF Add to collection

Get this paper in your agent:

hf papers read 2610.20828

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/2610.20828 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/2610.20828 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/2610.20828 in a Space README.md to link it from this page.

Collections including this paper 1

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