Hacker News new | ask | show | jobs
by amelius 393 days ago
Why would you use RL if you're not going to control the environment, but just predict it?
1 comments
TOMDM 393 days ago
Because they're training a predictor, not an agent?
link