Knowledge check · 15 questions
Tests understanding of sparse autoencoders, features, circuits, and why mechanistic interpretability matters for AI safety
Mechanistic Interpretability: Reading the Model's Mind — Quick Check
15 questions
