arXiv 2603.053082026-09-24cs.CLcs.AI

Med-V1：用于零样本和可扩展生物医学证据归因的小型语言模型

Med-V1: Small Language Models for Zero-shot and Scalable Biomedical Evidence Attribution

Division of Intramural Research, National Library of Medicine, National Institutes of Health（国家医学图书馆内部研究部，国立卫生研究院）；Department of Computer Science, University of Virginia（弗吉尼亚大学计算机科学系）；Center for Cancer Research, National Cancer Institute, National Institutes of Health（国家癌症研究所癌症研究中心，国立卫生研究院）；Department of Population Health Sciences, Weill Cornell Medicine Institute of AI for Digital Health, Weill Cornell Medicine（韦尔·科恩医学中心流行病学与健康科学系，韦尔·科恩医学中心人工智能与数字健康研究所）

Qiao Jin, Yin Fang, Lauren He, Yifan Yang, Guangzhi Xiong, Zhizheng Wang, Nicholas Wan, Joey Chan, Donald C. Comeau, Robert Leaman, Charalampos S. Floudas, Aido… 展开作者

Qiao Jin, Yin Fang, Lauren He, Yifan Yang, Guangzhi Xiong, Zhizheng Wang, Nicholas Wan, Joey Chan, Donald C. Comeau, Robert Leaman, Charalampos S. Floudas, Aidong Zhang, Michael F. Chiang, Yifan Peng, Zhiyong Lu

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出仅3B参数的小语言模型Med-V1，通过高质量合成数据训练，在生物医学证据归因任务上性能媲美GPT-5等前沿大模型，并用于量化LLM幻觉和识别临床指南中的证据错误归因。

摘要：

评估一篇文章是否支持某个断言对于幻觉检测和声明验证至关重要。虽然大型语言模型（LLM）有潜力自动化这一任务，但实现强性能需要如GPT-5这样的前沿模型，而这些模型在规模部署时成本过高。为了高效执行生物医学证据归因，我们提出了Med-V1，一个仅有三亿参数的小语言模型家族。在本研究中新开发的高质量合成数据上训练，Med-V1在统一为验证格式的五个生物医学基准上显著优于其基础模型（+27.0%至+71.3%）。尽管规模较小，Med-V1的性能与GPT-5等前沿LLM相当，并提供高质量的预测解释。我们使用Med-V1进行了首次用例研究，量化了不同引用指令下LLM生成答案中的幻觉。结果表明，格式指令强烈影响引文有效性和幻觉，GPT-5生成更多声明但表现出与GPT-4o相似的幻觉率。此外，我们展示了第二个用例，表明Med-V1可以自动识别临床实践指南中的高风险证据错误归因，揭示了否则难以大规模识别的潜在负面公共卫生影响。总体而言，Med-V1为生物医学证据归因和验证任务的实际应用提供了一种高效、准确的轻量级替代方案。Med-V1可在https://github.com/ncbi-nlp/Med-V1获取。

英文摘要：

Assessing whether an article supports an assertion is essential for hallucination detection and claim verification. While large language models (LLMs) have the potential to automate this task, achieving strong performance requires frontier models such as GPT-5 that are prohibitively expensive to deploy at scale. To efficiently perform biomedical evidence attribution, we present Med-V1, a family of small language models with only three billion parameters. Trained on high-quality synthetic data newly developed in this study, Med-V1 substantially outperforms (+27.0% to +71.3%) its base models on five biomedical benchmarks unified into a verification format. Despite its smaller size, Med-V1 performs comparably to frontier LLMs such as GPT-5, along with high-quality explanations for its predictions. We use Med-V1 to conduct a first-of-its-kind use case study that quantifies hallucinations in LLM-generated answers under different citation instructions. Results show that the format instruction strongly affects citation validity and hallucination, with GPT-5 generating more claims but exhibiting hallucination rates similar to GPT-4o. Additionally, we present a second use case showing that Med-V1 can automatically identify high-stakes evidence misattributions in clinical practice guidelines, revealing potentially negative public health impacts that are otherwise challenging to identify at scale. Overall, Med-V1 provides an efficient and accurate lightweight alternative to frontier LLMs for practical, real-world biomedical evidence attribution. Med-V1 is available at https://github.com/NLM-DIR/Med-V1.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。