fix: correct SAC target entropy to -action_dim#31
Open
Mr-Neutr0n wants to merge 1 commit into
Open
Conversation
The default target entropy was set to -action_dim / 2 instead of -action_dim as specified in the original SAC paper (Haarnoja et al., 2018). This causes under-exploration by targeting a lower entropy than intended. The DRQ learner already used the correct formula.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bug
The default
target_entropyis set to-action_dim / 2in the SAC, SAC v1, and REDQ learners. The original SAC paper (Haarnoja et al., 2018) specifies-action_dimas the target entropy heuristic. Using half the correct value causes under-exploration by targeting lower entropy than intended.Note that the DRQ learner already used the correct
-action_dimformula, making this an inconsistency as well.Fix
Changed the default
target_entropyfrom-action_dim / 2to-action_dimin:jaxrl/agents/sac/sac_learner.pyjaxrl/agents/sac_v1/sac_v1_learner.pyjaxrl/agents/redq/redq_learner.pyThis matches the paper and is consistent with the existing DRQ implementation.
Reference
Haarnoja, T., Zhou, A., Abbeel, P., & Levine, S. (2018). Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor. ICML 2018.