modelscope / modelscope/ms-swift
GRPO for qwen3-vl-thinking models
Open
Nobody has claimed this yet.
stale
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
I am wondering if there is a example shell file for doing on qwen3-vl-4b-thinking models. As they are original-thinking models, it should be different from qwen2.5-vl-7b model. Including the dataset format, query prompt, answer extraction and reward scores.
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
Start by locating the existing qwen2.5-vl-7b GRPO example and compare how it defines the dataset format, query prompt, answer extraction, and reward scores. Add a shell example for qwen3-vl-4b-thinking models that covers those four areas, then verify that the documented command and configuration can run successfully.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, shell
- Domain
- documentation, machine-learning
- Issue type
- Feature
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100