Categories
Interpretability
3 posts

Capital-of Is Not a Single SAE Feature. So I Built a Mutation Loop to Find What Is.
SAE features can't isolate relations in Gemma-2-2B. I built a mutation-selection loop that can. The bottleneck was tokenization.
aiinterpretabilityllmsparse-autoencoders
Sparse Autoencoders Can't Measure Generation-Time Behavior. That's Not a Bug.
Why sycophancy SAE features have Cohen's d=9.9 but hallucination detection fails. The answer turned out to be deeper than measurement timing.
aiinterpretabilitysparse-autoencoders
Stop Designing Your AI System. Design Its Environment.
Self-evolving AI harnesses fail when they optimize a fixed evaluator. The biological model is right: what needs to evolve is the selection pressure, not just the genome.
aiagentsarchitectureinterpretability