通过细粒度自验证提升智能搜索模型的准确性与可解释性
FineVerify: Scaling Test-Time Compute with Fine-Grained Self-Verification for Agentic Search
- 将问题拆解为可验证子问题，逐项核验候选答案
- 4个样本下准确率提升8.2点，12个样本超越GPT-5
- 生成可审计的验证轨迹，适合系统调试与可靠性评估
智能搜索要求语言模型代理在多个来源中探索并回答复杂信息查询。扩大测试时计算量是提升代理性能的有效途径，但现有方法常因正确答案稀疏且评分依赖模型校准而失效。我们提出FineVerify，一种细粒度自验证框架：将每个问题分解为可验证的子问题，对采样的候选答案逐一验证，并选择综合得分最高的结果。该逐项核验机制将选择转化为更简单的局部判断，使评分标准保持一致。在四个智能搜索基准和两个模型上，FineVerify均优于标准扩展基线。仅用4条采样轨迹，便使GPT-5-mini准确率提升8.2个百分点，Gemini-3-flash平均提升5.6%。使用12个样本时，GPT-5-mini在BrowseComp-Plus上超越前沿GPT-5。除准确率外，FineVerify还能生成可解释的验证轨迹，有助于审计基准错误，揭示其在审查智能搜索系统中的广泛应用潜力。代码与数据已公开于https://github.com/XuZhao0/fineverify。
原文摘要 · Abstract (English)
Agentic search requires language model agents to explore many sources and answer complex information-seeking questions. Scaling test-time compute is a promising way to improve these agents, but current approaches can fail, because correct answers are often sparse and score-based selection depends on model calibration. We propose FineVerify, a fine-grained self-verification framework that decomposes each question into checkable sub-questions, verifies sampled candidates against each sub-question, and selects the candidate with the highest aggregated score. This per-check structure turns selection into simpler local judgments and produces scores under the same explicit criteria. Across four agentic search benchmarks and two models, FineVerify consistently outperforms standard scaling baselines. With only four sampled trajectories, it improves GPT-5-mini by 8.2 accuracy points and Gemini-3-flash by 5.6% on average. With 12 samples, FineVerify enables GPT-5-mini to surpass frontier GPT-5 on BrowseComp-Plus. Beyond accuracy, FineVerify produces interpretable verification traces that help audit benchmark errors, suggesting broader applications for inspecting agentic search systems. Code and data are available at https://github.com/XuZhao0/fineverify
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。