Austin Meek

Measuring Sparse Autoencoder Feature Space Similarities Across Large Language Models

Sparse Autoencoders Reveal Universal Feature Spaces Across Large Language Models