DSCC-HS：用于大型语言模型幻觉抑制的动态自强化框架
DSCC-HS: A Dynamic Self-Reinforcing Framework for Hallucination Suppression in Large Language Models
AI总结：
提出DSCC-HS框架，通过FAP和HDP双代理在自回归解码时注入实时引导向量，无需修改目标模型即可主动抑制LLM幻觉，在TruthfulQA和BioGEN上达到SOTA。
摘要：
大型语言模型（LLM）的幻觉是其可靠部署的重大障碍。当前的方法如检索增强生成（RAG）通常是反应式的。我们引入了**用于幻觉抑制的动态自强化校准（DSCC-HS）**，这是一种在自回归解码期间进行干预的新型主动框架。受双过程认知理论的启发，DSCC-HS使用一个紧凑的代理模型，以对抗角色训练为事实对齐代理（FAP）和幻觉检测代理（HDP）。在推理过程中，这些代理通过在每个解码步骤注入实时引导向量（即FAP和HDP logits之间的差异）来动态引导大型目标模型。这种即插即用的方法无需修改目标模型。我们在TruthfulQA和BioGEN上的实验表明，DSCC-HS实现了最先进的性能。在TruthfulQA上，它达到了99.2%的事实一致性率（FCR）。在长文本BioGEN基准上，它获得了46.50的最高FActScore。这些结果验证了DSCC-HS作为增强LLM事实性的有原则且高效的解决方案。
英文摘要：
Large Language Model (LLM) hallucination is a significant barrier to their reliable deployment. Current methods like Retrieval-Augmented Generation (RAG) are often reactive. We introduce **Dynamic Self-reinforcing Calibration for Hallucination Suppression (DSCC-HS)**, a novel, proactive framework that intervenes during autoregressive decoding. Inspired by dual-process cognitive theory, DSCC-HS uses a compact proxy model, trained in adversarial roles as a Factual Alignment Proxy (FAP) and a Hallucination Detection Proxy (HDP). During inference, these proxies dynamically steer a large target model by injecting a real-time steering vector, which is the difference between FAP and HDP logits, at each decoding step. This plug-and-play approach requires no modification to the target model. Our experiments on TruthfulQA and BioGEN show DSCC-HS achieves state-of-the-art performance. On TruthfulQA, it reached a 99.2% Factual Consistency Rate (FCR). On the long-form BioGEN benchmark, it attained the highest FActScore of 46.50. These results validate DSCC-HS as a principled and efficient solution for enhancing LLM factuality.
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。
