[gwern]

> Scaling laws for RL agents are discovered, similar to language models.

Still waiting for a followup to https://arxiv.org/abs/2104.03113 ...