July 25th, 20258 min read
Mixture-of-Recursions Learning Dynamic Recursive Depths for Adaptive Token-Level Computation
Explaining the Mixture-of-Recursions model, starting from kv-cache basics.
AI ResearchAdaptive Computation
Read full postWriting
Ideas I'm thinking about
A running log of essays on AI engineering, open-source work, and whatever else I think is interesting at the time. Each post links out to the canonical version in case you prefer reading in your app of choice.