Lost at the Beginning of Reasoning

Baohao Liao, Xinyi Chen, Sara Rajaee, Yuhui Xu, Christian Herold, Anders Søgaard, Maarten de Rijke, Christof Monz

Keywords: Reasoning, LLM, Chain-of-Thought, Self-correction

TL;DR: The first reasoning step is significantly important for long-CoT LLM.

Abstract: Recent advancements in large language models (LLMs) have significantly advanced complex reasoning capabilities, particularly through extended chain-of-thought (CoT) reasoning that incorporates mechanisms such as backtracking, self-reflection, and self-correction. Despite these developments, the self-correction abilities of LLMs during long CoT reasoning remain underexplored. And recent findings on overthinking suggest that such models often engage in unnecessarily redundant reasoning. In this work, we empirically show that the first reasoning step exerts a disproportionately large influence on the final prediction—errors introduced at this stage can substantially degrade subsequent reasoning quality. This phenomenon is consistently observed across state-of-the-art open- and closed-source reasoning models. Leveraging this insight, we propose an efficient sampling strategy that leverages a reward model to identify and retain high-quality first reasoning steps while discarding suboptimal ones, achieving up to a 70% reduction in inference cost without sacrificing accuracy.

Primary Area: foundation or frontier models, including LLMs

Submission Number: 11990