Regularization

“What changes for most agents when you're doing multi-agent training is that, like, basically having lots of agents around makes your environment way more entropic. Like agents, agents are these huge generators of, like, entropy, because they're these big, comp…”