deepspeedai / deepspeedai/DeepSpeed

Deepspeed zero3 memory leak with rlhf [BUG]

Open
#3,590 0 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

bug deepspeed-chat
Dominant language
Python
Stars
43.1k
Forks
5k
Avg merge
4d 15h
Merged PRs (30d)
112

Description

Describe the bug
At the beginning, the model was able to train, but when it reached 600+steps, the memory of the main process skyrocketed. I didn't save the checkpoint
Memory usage at the beginning
image
Memory explosion
image

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

No files, tests, or entry points are named. Start by reproducing the RLHF run with DeepSpeed ZeRO-3 and monitor main-process memory around step 600; done means identifying the leak and confirming that memory remains stable afterward.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
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

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.