4 ms·Show HN: HyperSAE – Hyperbolic Sparse Autoencoders for LLM Interpretability3 points by visha1v 2mo ago