Author name: Ming Liu

Decodable but Not Corrected by Fixed Residual-Stream Linear Steering: Evidence from Medical LLM Failure Regimes

Ming Liu / May 8, 2026

arXiv:2605.05715v1 Announce Type: cross
Abstract: Can linearly decodable failure signals in LLM hidden states be leveraged to correct those failures? We investigate this classification-correction gap via Overthinking (OT)–a stable behavioral regime (…

cs.AI, cs.CL

More Is Not Always Better: Cross-Component Interference in LLM Agent Scaffolding

Ming Liu / May 8, 2026

arXiv:2605.05716v1 Announce Type: cross
Abstract: LLM agent systems are built by stacking scaffolding components (planning, tools, memory, self-reflection, retrieval) assuming more is better. We study cross-component interference (CCI): degradation wh…