推理提升平均准确率，但会损害高精度场景下的召回率。
Reasoning's Razor: Reasoning Improves Accuracy but Can Hurt Recall at Critical Operating Points in Safety and Hallucination Detection
- 对比有无推理生成，发现推理在低误报率下表现更差。
- 在严格低误报率条件下，不推理反而召回率更高。
- 适合对误报敏感的安全与幻觉检测任务使用。
推理已成为大语言模型的核心范式，在各类基准上持续提升准确率。然而其在高精度敏感任务中的适用性尚不明确。本文首次系统研究了在严格低误报率（FPR）条件下，分类任务中推理的作用。涵盖安全检测与幻觉检测两个任务，评估了微调和零样本设置下标准LLM与大型推理模型（LRM）的表现。结果揭示明显权衡：推理增强生成虽提升整体准确率，但在实际应用所需的低FPR阈值下表现不佳；相比之下，推理关闭模式（Think Off）在这些高精度场景中占据主导，仅当可接受较高FPR时，推理模式才更优。此外，基于标记的评分显著优于自我陈述置信度。最后，简单集成两种模式可兼顾两者优势。总体而言，推理是把双刃剑：有利于平均准确率，但常不适于要求严苛精度的应用。
原文摘要 · Abstract (English)
Reasoning has become a central paradigm for large language models (LLMs), consistently boosting accuracy across diverse benchmarks. Yet its suitability for precision-sensitive tasks remains unclear. We present the first systematic study of reasoning for classification tasks under strict low false positive rate (FPR) regimes. Our analysis covers two tasks--safety detection and hallucination detection--evaluated in both fine-tuned and zero-shot settings, using standard LLMs and Large Reasoning Models (LRMs). Our results reveal a clear trade-off: Think On (reasoning-augmented) generation improves overall accuracy, but underperforms at the low-FPR thresholds essential for practical use. In contrast, Think Off (no reasoning during inference) dominates in these precision-sensitive regimes, with Think On surpassing only when higher FPRs are acceptable. In addition, we find token-based scoring substantially outperforms self-verbalized confidence for precision-sensitive deployments. Finally, a simple ensemble of the two modes recovers the strengths of each. Taken together, our findings position reasoning as a double-edged tool: beneficial for average accuracy, but often ill-suited for applications requiring strict precision.
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。