一篇新的研究论文探讨了大语言模型（LLM）中的“路由假设”，认为模型拥有知识但难以在内部路由以激活它。该研究在代码安全漏洞检测中重现了先前在数学推理任务中的发现，使用了GPT-OSS-120B、Llama-3.3-70B和Gemma-4-31B等模型。结果表明，结构化先验（备忘单）在合成数据上显著提高了性能，但在应用于真实CVE数据时导致准确率大幅下降，表明存在跨领域权衡。 AI
影响 表明面向分布的训练可能比提示校准更能有效提高LLM在真实世界数据上的可靠性。
排序理由 该集群包含一篇详细介绍LLM行为实验结果的研究论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2607.14628v1 Announce Type: new Abstract: Large language models (LLMs) exhibit a well-documented gap between latent capability and consistent activation: the router hypothesis posits that models possess the knowledge to solve a task but lack reliable internal routing to act…
Large language models (LLMs) exhibit a well-documented gap between latent capability and consistent activation: the router hypothesis posits that models possess the knowledge to solve a task but lack reliable internal routing to activate it. Prior work in formal mathematical reas…