This blog post offers an interpretability deep dive, examining the most important concepts emerging in one of today’s central vision foundation models, DINOv2. This blogpost is the first of a […]

The superposition hypothesis states that a single neuron within a population may participate in the representation of multiple features in order for the population to represent more features than the ...

Modern methods for explainable machine learning are designed to describe how models map inputs to outputs--without deep consideration of how these explanations will be used in practice. This paper arg...

Motivated by the hypothesis that neural network representations encode abstract, interpretable features as linearly accessible, approximately orthogonal directions, sparse autoencoders (SAEs) have bec...

First Workshop on Interpreting Cognition in Deep Learning Models (NeurIPS 2025)

Using sparse autoencoders, the authors show that vision-language embeddings boil down to a small, stable dictionary of single-modality concepts that snap together into cross-modal bridges. This resear...

Neural firing-rate responses to sensory stimuli show progressive changes both within and across sessions, raising the question of how the brain mainta…

ArXiv link for AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders

Kempner Graduate Fellowships support a dynamic and diverse community of PhD students across a number of graduate programs at Harvard, seeding new and innovative scientific discoveries in labs across t...

Welcome to the OpenReview homepage for ICLR 2025 Workshop XAI4Science