will-maclean / sb3-burn

Implementation of stable-baselines3 in rust with burn
MIT License
11 stars 0 forks source link

evaluate_policy doesn't use deterministic actions #25

Closed will-maclean closed 3 weeks ago

will-maclean commented 3 weeks ago

evaluate_policy shouldn't do e.g. action sampling or epsilong greedy. Instead, it should be deterministic/greedy.

will-maclean commented 3 weeks ago

evaluate_policy already uses deterministic actions