InterpretabilityResearch
Towards Monosemanticity explained: finding the concepts inside a neural network
Anthropic used sparse autoencoders to decompose a model's neurons into thousands of interpretable features. What superposition is, what they found and why it matters.
AI Desk3 min read