deepspeedai / deepspeedai/DeepSpeed
[BUG] pipline engine's training stucked when zero=1
Open
Nobody has claimed this yet.
bug
compression
- Dominant language
- Python
- Stars
- 43.1k
- Forks
- 5k
- Avg merge
- 4d 15h
- Merged PRs (30d)
- 112
Description
pp_size = 8
stage 0 contains a vision encoder of 45 layers
stage 1~7 contain 56 layers of decoder
zero 0 is well but zero 1 and bf16/fp16 failed
much more GPU memory will be saved if zero 1 runs well
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
No files, tests, logs, model details, or exact reproduction command are provided. Start by reproducing the pp_size=8 configuration with the stated stage layer split under zero=0 and zero=1 with bf16/fp16, then compare where training stops. Done means zero=1 completes training in the failing configuration while retaining the expected GPU-memory savings.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- distributed-systems, machine-learning, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100