Beyond Confidence: Stability-Aware Test-Time Adaptation for LLM Reasoning
This work proposes Test-Time Adaptation via Stability-Aware Confidence Optimization (TASCO), a framework that incorporates local stability into confidence-based test-time adaptation while keeping the LLM frozen and improves reasoning accuracy and token efficiency across diverse LLMs and reasoning benchmarks.
Bin-Cheng Gu, Min Gao, Zong-Wei Wang et al.
· 0 citations