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.
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.jldoesn't currently support this out the gate since episode info isn't saved, but looking atRLTrajectories.jl, it should make this task a ton easier.