MuZero's first step from research into the real world

https://deepmind.google/blog/muzeros-first-step-from-research-into-the-real-world/

MuZero-RC chose VP9 quantisation parameters per frame on part of YouTube live traffic; average 4% bitrate reduction without quality loss; objective framed as win/loss against the agent's own past performance.

Self-play planning with learned models (AlphaZero, MuZero, Dreamer)