Fetching the paper…

Quantifying Feature Space Universality Across Large Language Models via Sparse Autoencoders · Around