mindspore-ai / mindspore-ai/hyper-parallel
[Bug]: rl场景param_offload精度问题
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 53
- Forks
- 63
- Avg merge
- 23h 45m
- Merged PRs (30d)
- 63
Description
Checklist
- 1. I have searched the existing issues (https://gitcode.com/mindspore/hyper-parallel/issues)
- 2. I have read the relevant documentation.
- 3. I have created a minimal reproduction case that clearly demonstrates the issue, including a complete code example and the error message with full traceback and error logs.
🐛 Describe the bug
rl场景使能param_offload，会调用.to执行npu->cpu、cpu->npu的拷贝，发现存在精度问题（训练结果崩溃 梯度爆炸），不使能param_offload精度正常
Expected behavior
rl场景使能param_offload精度正常
Additional context
Environment info
Atlas A2/CANN 8.5
Thanks for contributing 🎉!
schema_version: 1
source: gitcode
gitcode_repo: mindspore/hyper-parallel
gitcode_issue: 237
source_url: https://gitcode.com/mindspore/hyper-parallel/issues/237
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
No source file, test, or runnable reproduction is included in the issue. Start by obtaining and running the reported RL reproduction with param_offload enabled on Atlas A2/CANN 8.5, then compare the .to-based NPU-to-CPU and CPU-to-NPU transfers with param_offload disabled. Done means training remains numerically stable and avoids the reported gradient explosion.
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
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100