deepspeedai / deepspeedai/DeepSpeedExamples
7B bloom cuda oom run step3 with 80 T4(16G) when hybridEngine is on, but ok when hybridengine is off
Open
Nobody has claimed this yet.
deespeed chat
hybrid engine
- Dominant language
- Python
- Stars
- 6.8k
- Forks
- 1.1k
- Avg merge
- 2d 16h
- Merged PRs (30d)
- 1
Description
10 machines, 80 T4(16G)s, actor(7b bloom) with Zero3, critic(1.3b bloom) with Zero3, open both gradient_checkpointing
why HybridEngine using more gpu memory？
thank you in advance
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
Start by reproducing the step 3 run with the stated 7B and 1.3B BLOOM actors, ZeRO-3, gradient checkpointing, and 80 T4 GPUs. Compare GPU memory with HybridEngine enabled and disabled; done means identifying the cause of the additional memory use or documenting the relevant configuration constraint.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100