[RLlib; Off-policy] Add sequence sampling to 'EpisodeReplayBuffer'.#48116

Merged

sven1977 merged 32 commits intoray-project:masterfrom simonsays1980:offpolicy-enable-sequence-sampling-in-episode-replay-buffer

Jan 6, 2025

+353-132

Commits on Oct 21, 2024

Added sequence sampling to 'EpisodeReplayBuffer' and changed episode construction to slicing. This is still incomplete b/c it needs the correct discounting of rewards in case a sequence is returned…
simonsays1980
committed

Commits on Oct 22, 2024

Removed states from 'EpisodeReplayBuffer._sample_episodes' and removed lookback. Furthermore, added 'get_initial_state' to 'DQNRainbowRLModule' and adapted module for stateful training.
simonsays1980
committed

Commits on Oct 23, 2024

Fixed multiple shape errors in DQN when training with LSTM.
simonsays1980
committed

Commits on Oct 29, 2024

Added next obs states to batch of the q-network when using the dueling architecture.
simonsays1980
committed

Commits on Oct 30, 2024

Commits on Nov 1, 2024

Added partial sequence sampling to add more terminal nodes to training.
simonsays1980
committed

Commits on Nov 11, 2024

Added stateless Cartpole example to tuned examples for DQN.
simonsays1980
committed

Commits on Nov 25, 2024

Merge branch 'master' into offpolicy-enable-sequence-sampling-in-episode-replay-buffer
simonsays1980
committed

Commits on Nov 28, 2024

Merge branch 'master' into offpolicy-enable-sequence-sampling-in-episode-replay-buffer
simonsays1980
committed

Commits on Dec 21, 2024

Merged master and resolved conflicts.
simonsays1980
committed

Commits on Dec 28, 2024

Commits on Jan 3, 2025