Skip to content

Align epsilon help/docstring wording#6014

Open
qgallouedec wants to merge 1 commit into
mainfrom
align-async-grpo-epsilon-help
Open

Align epsilon help/docstring wording#6014
qgallouedec wants to merge 1 commit into
mainfrom
align-async-grpo-epsilon-help

Conversation

@qgallouedec

@qgallouedec qgallouedec commented Jun 11, 2026

Copy link
Copy Markdown
Member

GRPOConfig documents epsilon as "Epsilon value for clipping."; AsyncGRPOConfig adds a "Lower-bound" qualifier. Align the wording.


Note

Low Risk
Documentation-only edits in async_grpo_config.py with no runtime or configuration behavior impact.

Overview
Updates AsyncGRPOConfig documentation so the epsilon parameter matches GRPOConfig: the class docstring and the epsilon field metadata["help"] now say "Epsilon value for clipping." instead of "Lower-bound epsilon value for clipping."

No defaults, types, or training logic change—only wording for CLI/help and API docs.

Reviewed by Cursor Bugbot for commit 19f164b. Bugbot is set up for automated code reviews on this repo. Configure here.

@qgallouedec qgallouedec requested a review from AmineDiro June 11, 2026 22:32
@bot-ci-comment

Copy link
Copy Markdown

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant