modelscope / modelscope/ms-swift
Dora megatron相关
Open
Nobody has claimed this yet.
enhancement
- Dominant language
- Python
- Stars
- 15.7k
- Forks
- 1.7k
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 136
Description
请问是否有计划支持实现megatron的Dora训练呢？
当前megatron中的use_dora还没有实现功能：
if use_dora:
raise ValueError(f'{self.class.name} does not support DoRA yet, please set it to False')
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 Megatron integration containing the shown use_dora guard, then trace how this option is passed through training. Clarify the required DoRA behavior and validation before changing the guard. Done means Megatron DoRA training is supported instead of raising the current ValueError, with an appropriate validation path.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100