Query Circuits: Explaining How Language Models Answer User Prompts 文章

ArXiv CS.AI2026-06-02NEWSen作者: Tung-Yu Wu, Fazl Barez
查看原文 →
关系图谱

Query Circuits: Explaining How Language Models Answer User Prompts · 相关技术

相关技术

sparse autoencoderNormalized Deviation Faithfulnessindirect object identificationquery circuits