Mind the Gap: Bridging Thought Leap for Improved Chain-of-Thought Tuning

TL;DR

This paper introduces a method to detect and fill reasoning gaps in Chain-of-Thought datasets for mathematical tasks, improving model performance and generalization by restoring reasoning completeness.

Contribution

We propose the CoT Thought Leap Bridge Task and construct ScaleQM+ dataset to automatically identify and generate missing reasoning steps, enhancing LLM training.

Findings

Models trained on bridged datasets outperform original datasets by up to +5.87% on NuminaMath.

Our approach improves data quality (+3.02%) and reinforcement learning starting points (+3.1%).

Enhanced reasoning completeness improves generalization to out-of-domain tasks.

Abstract

Large language models (LLMs) have achieved remarkable progress on mathematical tasks through Chain-of-Thought (CoT) reasoning. However, existing mathematical CoT datasets often suffer from Thought Leaps due to experts omitting intermediate steps, which negatively impacts model learning and generalization. We propose the CoT Thought Leap Bridge Task, which aims to automatically detect leaps and generate missing intermediate reasoning steps to restore the completeness and coherence of CoT. To facilitate this, we constructed a specialized training dataset called ScaleQM+, based on the structured ScaleQuestMath dataset, and trained CoT-Bridge to bridge thought leaps. Through comprehensive experiments on mathematical reasoning benchmarks, we demonstrate that models fine-tuned on bridged datasets consistently outperform those trained on original datasets, with improvements of up to +5.87% on NuminaMath. Our approach effectively enhances distilled data (+3.02%) and provides better starting points for reinforcement learning (+3.1%), functioning as a plug-and-play module compatible with existing optimization techniques. Furthermore, CoT-Bridge demonstrate improved generalization to out-of-domain logical reasoning tasks, confirming that enhancing reasoning completeness yields broadly applicable benefits.