揭示大模型答題時如何讀取推理過程，並提升正確率
How Do Answer Tokens Read Reasoning Traces? Self-Reading Patterns in Thinking LLMs for Quantitative Reasoning
- 分析答案對推理過程的注意力模式，發現正確解題有明確聚焦與前向流動
- 錯誤解答注意力分散且不規則，反映模型內部不確定性
- 提出無需訓練的引導方法，利用自我閱讀質量提升推理可靠性
思考型大模型在回答前會生成推理過程。以往工作多著重於塑造這些推理過程，但對答案令牌如何讀取並整合推理以產生可靠結果的理解仍不足。本研究聚焦於定量推理，分析答案對推理過程的注意力機制，發現正確解題呈現良性自我閱讀模式：注意力沿推理路徑向前移動，並持續聚焦於關鍵語義錨點；而錯誤解題則表現為分散且不規則的注意力模式。我們將此解釋為答案解碼過程中的內部信心，即模型選擇可行的解決分支並整合關鍵證據。基於此，提出一種免訓練的引導方法，依賴自我閱讀品質（SRQ）得分，結合幾何度量進行流程控制與語義度量進行內容監控，選取數據構建引導向量，使推理朝向良性自我閱讀，遠離不確定與混亂的讀解。實驗表明該方法可穩定提升準確率。
原文摘要 · Abstract (English)
Thinking LLMs produce reasoning traces before answering. Prior activation steering work mainly targets on shaping these traces. It remains less understood how answer tokens actually read and integrate the reasoning to produce reliable outcomes. Focusing on quantitative reasoning, we analyze the answer-to-reasoning attention and observe a benign self-reading pattern aligned with correctness, characterized by a forward drift of the reading focus along the reasoning trace and a persistent concentration on key semantic anchors, whereas incorrect solutions exhibit diffuse and irregular attention pattern. We interpret this as internal certainty during answer decoding, where the model commits to a viable solution branch and integrates key evidence. Following this, we propose a training-free steering method driven by Self-Reading Quality (SRQ) scores combining geometric metrics for process control with semantic metrics for content monitoring. SRQ selects data to build steering vectors that guide inference toward benign self-reading and away from uncertain and disorganized reading. Experiments show that our method yields consistent accuracy gains.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。