1 paper
Jonathn Chang, Arya Datla, Ziv Goldfeld
Causal abstraction offers a principled framework for mechanistic interpretability, aligning a high-level causal model with low-level neural computation through interchange interven…