CoTranslator improves math reasoning training with efficient answer-based translation

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

Computation and LanguageArtificial IntelligenceMachine Learning

Summary

Getting computers to explain their math reasoning step-by-step is hard and slow using traditional methods. The authors show a new way called CoTranslator that teaches computers to turn reasoning that starts from the answer backward into a forward explanation, similar to how people solve problems. This approach requires fewer expensive checks and works well across different math and science tasks. It helps smaller models learn to reason better without needing a lot more computing power.

What this means in practice

  • For machine learning engineers: Generate verified reasoning steps for training smaller models efficiently on math problems by translating backward reasoning into forward explanations.
  • For ai platform developers: Reduce computational cost of producing quality reasoning data to improve AI services that rely on stepwise problem solving in STEM domains.
  • For automated customer service teams: Improve explainability of AI responses by translating internal answer-conditioned reasoning into step-by-step forward reasoning visible to users.$Commercial implications: This enables AI products with clearer, user-understandable explanations for replies, enhancing trust and usability.

Authors

Zinan Tang, Zhuoshi Pan, Kairui Chen, Ze Yan, Shaomian Zheng, Yukun Zhang, Gang Wang, Biqing Huang

Abstract

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.