用置信度指导搜索代理重试，提升效率与准确性。
BrowseConf: Confidence-Guided Test-Time Scaling for Web Agents
- 根据模型自述置信度动态决定是否重试任务。
- 高置信度时准确率显著提升，低置信度时接近零准确率。
- 减少令牌消耗，适合资源受限的在线推理场景。
大型语言模型的置信度是衡量模型不确定性和答案可靠性的重要指标。现有研究多集中于单轮交互场景，对复杂多轮交互中置信度的研究较为有限。本文探究基于LLM的搜索代理在长时间动作序列后，能否通过口语化置信度评分传达自身信心，这一任务相比单次交互更具挑战性。实验发现，模型在高置信度时任务准确率显著提升，而低置信度时准确率几乎为零。基于此观察，我们提出测试时缩放（Test-Time Scaling, TTS）方法，利用置信度分数判断答案质量，并引导模型持续重试直至达到满意置信水平。结果表明，所提方法显著降低令牌消耗，同时性能优于固定预算的基线TTS方法。
原文摘要 · Abstract (English)
Confidence in LLMs is a useful indicator of model uncertainty and answer reliability. Existing work mainly focused on single-turn scenarios, while research on confidence in complex multi-turn interactions is limited. In this paper, we investigate whether LLM-based search agents have the ability to communicate their own confidence through verbalized confidence scores after long sequences of actions, a significantly more challenging task compared to outputting confidence in a single interaction. Experimenting on open-source agentic models, we first find that models exhibit much higher task accuracy at high confidence while having near-zero accuracy when confidence is low. Based on this observation, we propose Test-Time Scaling (TTS) methods that use confidence scores to determine answer quality, encourage the model to try again until reaching a satisfactory confidence level. Results show that our proposed methods significantly reduce token consumption while demonstrating competitive performance compared to baseline fixed budget TTS methods.
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。