Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models

Devs

Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models | Read Paper on Bytez