Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).
Latest commits.
Builders behind this project.
