modelscope / modelscope/ms-swift
[RFC] support Black-Box On-Policy Distillation of Large Language Models(GAD)
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
Describe the feature
如题，我发现咱们已经有 rm 的训练了，但是还缺少后面 adversarial training 的部分，所以看看能不能复现一下🙏。（原版使用 Verl 我实在不喜欢）
Paste any useful information
论文地址：https://arxiv.org/pdf/2511.10643
Github： https://github.com/microsoft/LMOps/tree/main/gad
Additional context
NO
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.
Research direction
Read the issue's GAD paper and the linked Microsoft LMOps implementation first, then compare them with ms-swift's existing reward-model training. Define the integration scope for the missing adversarial-training stage and verify that ms-swift can reproduce the reported black-box on-policy distillation workflow without relying on Verl.
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
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100