About CARE

Causality, Abstraction, Representation, and Extrapolation (CARE), is a new reading group organized by Jason Hartford, Chandler Squires, and Dhanya Sridhar.

Each week, we plan on inviting researchers to present their work on causal representation learning as well as related ideas like object-centric learning, causal abstraction, mechanistic interpretability, and compositionality in large models.

Use the Community Requests page to vote and submit ideas on what papers you’d like to see at upcoming talks.