Skip to content

Update README-EDiT.md #6

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
nash635 merged 2 commits into from
Feb 7, 2025
Merged

Update README-EDiT.md #6

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3ee7280
Update README-EDiT.md
JL-Cheng Jan 24, 2025
4e1cd26
Update README-EDiT.md
JL-Cheng Feb 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 4 additions & 3 deletions docs/README-EDiT.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,9 +1,10 @@
<h1 align="center"><b>EDiT</b></h1>
<h3 align="center"><b>EDiT: A Local-SGD-Based Efficient Distributed Training Method for Large Language Models</b></h3>

[![ICLR](https://img.shields.io/badge/ICLR-2025-blue)](https://arxiv.org/abs/2412.07210)
[![arXiv](https://img.shields.io/badge/arXiv-2412.07210-b31b1b.svg)](https://arxiv.org/abs/2412.07210)

We present PyTorch code for [EDiT: A Local-SGD-Based Efficient Distributed Training Method for Large Language Models](https://arxiv.org/abs/2412.07210).
We present PyTorch code for [EDiT: A Local-SGD-Based Efficient Distributed Training Method for Large Language Models](https://arxiv.org/abs/2412.07210), ICLR'25.

# Introduction

Expand Down Expand Up @@ -262,7 +263,7 @@ python -m atorch.distributed.run --nproc_per_node 8 train.py \
--outer_optim_class sgd
```

## 3D (Megatron-LM) Integratopm
## 3D (Megatron-LM) Integration
EDiT can also be integrated with 3D training through megatron. In atorch, we provide such functionality through patches and ATorchTrainerV2.

ATorchTrainerV2 handles model construction and argument handling. To enable EDiT, we need to inject several special arguments to trainer's input arguments, and before everything, patch the megatron.
Expand All @@ -289,4 +290,4 @@ cd examples/local_sgd/atorch_trainer_megatron

bash train.sh
```
Alternatively, you may reference to atorch/local_sgd/megatron, and migrate the patches directly into your Megatron-LM repo.
Alternatively, you may reference to atorch/local_sgd/megatron, and migrate the patches directly into your Megatron-LM repo.