modelscope / modelscope/ms-swift
Qwen3-VL training is frozen
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
Describe the bug
When I LoRA finetuned Qwen3-VL 2B model with liger kernel, the process is frozen indefinitely. When I turned it off, things go back to normal.
Your hardware and system info
Write your system info like CUDA version/system/GPU/torch version here(在这里给出硬件信息和系统信息，如CUDA版本，系统，GPU型号和torch版本等)
transformers==4.57.0
torch==2.8.0
GPU: A40
Additional context
Add any other context about the problem here(在这里补充其他信息)
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 reproducing the reported LoRA fine-tuning freeze for Qwen3-VL 2B with Liger enabled, using the stated transformers 4.57.0, torch 2.8.0, and A40 environment. Compare the same run with Liger disabled and gather the missing CUDA, system, and complete training details. Done means the cause is identified and the affected configuration no longer freezes indefinitely.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100