Blog
hover to explore
Types as constraintsA playable sketch of type refinement: drag the precision dial and watch what gets forbidden.- Formal Methods in ML InterpretabilityWhat would it mean for an explanation to be correct, not merely persuasive?
What is an explanation?Climb an explanation ladder: saliency, counterfactuals, causal claims, formal guarantees.
Causal Structures in Neural NetworksDo-calculus as an epistemic tool for reading model decisions, not a dashboard feature.