Skip to content

[Feature] Add per-agent advantage normalization support for IPPO/MAPPO - #256

Open
lmBored wants to merge 2 commits into
facebookresearch:mainfrom
lmBored:adv_norm_per_agent
Open

[Feature] Add per-agent advantage normalization support for IPPO/MAPPO#256
lmBored wants to merge 2 commits into
facebookresearch:mainfrom
lmBored:adv_norm_per_agent

Conversation

@lmBored

@lmBored lmBored commented Aug 1, 2026

Copy link
Copy Markdown

Closes #255

@meta-cla meta-cla Bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Aug 1, 2026
@lmBored

lmBored commented Aug 1, 2026

Copy link
Copy Markdown
Author

@Xmaster6y it's a quick fix indeed. Can you review this change?

@Xmaster6y

Copy link
Copy Markdown
Contributor

Small focused tests could be good.

@lmBored

lmBored commented Aug 2, 2026

Copy link
Copy Markdown
Author

I added 2 small tests to check the advantage calculation

lmBored added a commit to Farama-Foundation/ViZDoom that referenced this pull request Aug 2, 2026
The feature is not implemented in BenchMARL yet. It's opened under this
PR: facebookresearch/BenchMARL#256
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Is per-agent advantage normalization intentionally left out?

2 participants