推理显存沾满了后，看到显存自动释放了，且推理还在继续，是怎么做到
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 12.3k
- Forks
- 1.1k
- PR merge metrics
- No merged PRs in 30d
Description
推理显存沾满了后，通过nvidia-smi，看到显存自动释放了，且推理还在继续，就解决了不会因为问题的变多，显存沾满，程序崩了，想问下，moss是怎么处理的，具体怎么做的，有清楚的吗，谢谢
Contributor guide
No contributing guide indexed for this repository
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
The issue mentions MOSS inference and nvidia-smi but names no files, tests, or entry points. Start by locating the inference path and its GPU-memory handling, then trace what happens when memory usage reaches capacity. Done means documenting how memory is released while inference continues and how this prevents a crash.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100