deepspeedai / deepspeedai/DeepSpeed

llama-2-70b stage 2 optimization solution

Open
#4,233 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

I have five 8-card A800s with 1T memory. I use the following code to load 130G llama-2-70b. Why is it loaded to the memory every time, and then it hangs when the memory is full? Is there any optimization solution?Added the offload_optimizer parameter is also loaded to the full memory.
image
image
image
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

Start by reviewing the loading configuration and the screenshots attached to the issue, focusing on how the 130G Llama-2-70B model and offload_optimizer setting use the five A800 cards. Reproduce the memory growth and hang if the environment and code can be obtained; done would mean identifying the cause and documenting or validating an optimization.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.