deepspeedai / deepspeedai/DeepSpeed

How to transfer Pre-trained Model Parameters to a New Model Using DeepSpeed MoE API

Open
#5,032 0 comments 2 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

I'm currently working on a project where I need to transfer parameters from a pre-trained model to a new model architecture that utilizes the DeepSpeed MoE (Mixture of Experts) API. I'm seeking advice on the best practices for this process.

How to efficiently migrate parameters from a huggingface pre-trained model to a MoE-structured model in a distributed training environment.(eg. ep_size=4, num_experts=4, gpus=8, experts=LlamaMLP)

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 repository files, tests, or entry points are identified. Start by reviewing the DeepSpeed MoE API and the Hugging Face pretrained-model integration described in the issue; the work would need a defined migration approach and a demonstrated distributed setup using ep_size=4, four experts, and eight GPUs.

Written by the indexing model from the issue text.

Assessment

Tech stack
huggingface, python, pytorch
Domain
distributed-systems, machine-learning
Issue type
Feature
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.