modelscope / modelscope/ms-swift

How to do entropy regularization?

Open
#7,331 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

stale
Dominant language
Python
Stars
15.7k
Forks
1.7k
Avg merge
1d 16h
Merged PRs (30d)
136

Description

Hi:
I'm currently using ms-swift for GRPO (Group Relative Policy Optimization‌) training and would like to add entropy regularization to improve exploration and prevent premature convergence. However, I couldn't find clear documentation or examples on how to enable this feature, particularly when launching training via scripts.
Could you please provide guidance or implement support for entropy regularization in GRPO training?
Thanks.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the GRPO training entry points and launch scripts mentioned in the issue, and check whether entropy regularization is already supported but undocumented. Establish whether the requested outcome is usage documentation or new training support, then define completion as either a documented script invocation or tested GRPO behavior with the regularization enabled.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.