LLM Microscope: What Model Internals Reveal About Answer Correctness and Context Use

Jiarui Liu*, Jivitesh Jain*, Mona Diab, Nishant Subramani
November, 2025
Arxiv