ReinforcementLearningAnIntroduction.jl icon indicating copy to clipboard operation
ReinforcementLearningAnIntroduction.jl copied to clipboard

batch update (figure 6.2)

Open baedan opened this issue 3 years ago • 0 comments

the notebook doesn't currently reproduce figure 6.2 which uses batch updating (replays all episodes in an experience buffer until convergence). as far as i know RL.jl doesn't currently support this out the gate since episode info isn't saved, but looking at RLTrajectories.jl, it should make this task a ton easier.

baedan avatar Jun 05 '22 08:06 baedan