Mechanistic interpretability has proven transformative in understanding Large Language Models (LLMs), revealing intricate computational patterns and internal mechanisms. However, this powerful approach remains underutilized in computer vision, despite its potential to uncover fundamental properties of visual processing in neural networks. In this workshop, we will discuss methods to advance our understanding of vision models through mechanistic interpretability, potentially revealing novel insights about learned visual representations and emergent algorithms.
Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and Video
Mila Quebec / Meta
9:10-10:00 AM
Northeastern University
10:20-10:50 AM
UC Berkeley
10:50-11:20 AM
MIT / OpenAI
12:00-12:30 PM
MIT
3:30-4:00 PM
Reading Minds & Machines
Weizmann Institute of Science
4:00-4:30 PM
MIT
Reve
MIT
Northeastern University
UC Berkeley
Oxford
Adobe
Oxford