How Does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding.

Xi Chen, Aske Plaat, Niki van Stein

Browse the full AAAI paper archive.