Mindreader
Can we interpret what's happening inside of vision transformers? This is a simple project that explores the concepts being learned by individual neurons, as a step towards understanding more fully how these amazing models do what they do.
Made with ❤️ in San Francisco by Noah MacCallum and Sonia Joseph. Questions or feedback? DM us! Technical deep dive coming soon.