Show HN: HyperSAE – Hyperbolic Sparse Autoencoders for LLM Interpretability

3 points | by visha1v 7 hours ago