3 papers
cs.LG2026
Circuit Claims Depend on What Is Extracted and How It Is Compared
Yang Sheng, Jie Fu
Circuit extraction identifies a small set of model components whose presence preserves a target behavior under ablation, and the resulting circuit is often read as the mechanism be…
cs.AI2026
FormalJudge: A Neuro-Symbolic Paradigm for Agentic Oversight
Jiayi Zhou, Yang Sheng, Hantao Lou +2
As LLM-based agents increasingly operate in high-stakes domains with real-world consequences, ensuring their behavioral safety becomes paramount. The dominant oversight paradigm, L…
cs.CL2025
Large Language Models Can Better Understand Knowledge Graphs Than We Thought
Xinbang Dai, Yuncheng Hua, Tongtong Wu +3
When we integrate factual knowledge from knowledge graphs (KGs) into large language models (LLMs) to enhance their performance, the cost of injection through training increases wit…