Judge feedback

Judge feedback is anonymous and shown in a random order.

• This is an excellent project and I appreciate the effort put into creating an easy-to-use and accessible web app for creating standard mechanistic interpretability plots. The fact that the code is available on GitHub for reproducibility makes it even more valuable. While the current version is limited, I believe that with some further development and hosting on a stronger server, it has the potential to become an even more powerful tool for researchers and practitioners alike. Keep up the great work! Also check out the less user-friendly http://interp-tools.redwoodresearch.org/.
• I love it. Very good hackathon vibes, and a fun tool. Would be nicer if "no effect" looked white on the heat map rather than dark blue (and even better if "full effect recovered" was 1 or -1). No idea how to compare it to others re a prize, since it's not really research per se, but I'm happy it exists and hope you enjoyed making it!