deepspeedai / deepspeedai/DeepSpeed
Deepspeed zero3 + qlora arise problem! Params didn't sharded first before load to each GPU!
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
When I using zero3+qlora, params first loaded to each GPU then sharded abnormally. But after I change it to zero3+lora, it worked successfully. So I wonder if zero3_init doesn't support qlora? @loadams @tjruwase @jeffra @jomayeri
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 DeepSpeed ZeRO-3 plus QLoRA behavior and compare it with the working ZeRO-3 plus LoRA setup described in the issue. Determine whether zero3_init supports QLoRA and verify whether parameters are sharded before or after loading; the issue is resolved when the behavior and required support or limitation are clearly established.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100