huggingface / huggingface/accelerate

Why does accelerate's MegatronLMPlugin use its own megatron?

Open
#2,390 19 comments 0 reactions 0 assignees View on GitHub
wip
Dominant language
Python
Stars
9.9k
Forks
1.5k
Avg merge
5d 2h
Merged PRs (30d)
27

Description

I want to use the MegatronLMPlugin to train a megatron model, but [the guide](https://huggingface.co/docs/accelerate/main/en/usage_guides/megatron_lm) tell me the Megatron is https://github.com/huggingface/Megatron-LM, not the https://github.com/NVIDIA/Megatron-LM.

I want to ask why does accelerate use its own megatron? What differents between these two megatron?

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.