modelscope / modelscope/ms-swift
Support SAO (Single-Rollout Asynchronous Optimization)
Open
@hjh0119 is already working on this.
Since Jul 10, 2026.
enhancement
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
Checklist / 检查清单
- I have searched existing issues, and this is a new feature request. / 我已经搜索过现有的 issues，确认这是一个新的 Feature Request。
Feature Request Description / Feature Request 描述
As revealed by Z.ai, SAO (https://arxiv.org/pdf/2607.07508) is the secret ingredient behind the success of GLM-5.2. It employs a PPO-like method to address the issue that candidate trajectories within the same group become incomparable when they diverge significantly. SAO is expected to be better suited for long-horizon agentic training.
Pull Request / Pull Request 信息
No response
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Assessment
This issue has not been assessed yet.