Writing
Essays, explorables, and research write-ups.
Why SAEs Fail (and Why CLTs Might Too)
InteractiveAn interactive explorable on sparse autoencoders, cross-layer transcoder features, and where both approaches break down.
InterpretabilitySparse AutoencodersMachine Learning
2025-05-15Seeing Through the Projection
8 min + interactiveWhy every 2D embedding of your data is a lie — and how to think about which lies are acceptable.
Dimensionality ReductionVisualizationResearch
2024-11-10The Case for Explorable Explanations
10 minStatic figures are a compromise forced by print. We can do better — and we should.
Research CommunicationInteractive MediaScience
2024-08-20