Notes on reinforcement learning, efficient AI systems, and research ideas that are useful enough to write down.
May 31, 2026
read