[Question] GLM5模型sglang rollout有概率出现首字token乱码/没有意义
Open
Nobody has claimed this yet.
question
- Dominant language
- Python
- Stars
- 8.5k
- Forks
- 1.3k
- Avg merge
- 5h 36m
- Merged PRs (30d)
- 22
Description
Your Question
GLM5模型sglang rollout有概率出现首字token乱码/没有意义
What I've Tried
使用Slime v0.2.4镜像，在H20上面使用scripts/run-glm5-744B-A40B.sh脚本训练glm5的rl，rollout-batch-size=128，模型sglang rollout回复首字容易出现乱码或者与问题不相关的内容，不知道大家是否遇到相同的问题。
Environment (if relevant)
- slime version:
- Python version:
- PyTorch version:
- CUDA/ROCm version:
- GPU type and count:
- OS:
Additional Context
No response
Pre-submission Checklist
- I have read the CONTRIBUTING.md and understand the collaboration scope.
- I have read the documentation and FAQ and my question is not answered there.
- I have searched for existing issues and my question has not been asked before.
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 rollout with scripts/run-glm5-744B-A40B.sh using the stated slime v0.2.4 image, H20 hardware, batch size 128, and GLM5 model. Record the malformed or unrelated first-token outputs and fill in the missing Python, PyTorch, CUDA/ROCm, GPU-count, and OS details. Done means identifying a reproducible cause or a confirmed fix and documenting the validation result.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 32/100