History

Xuanlei Zhao 73aa406b96 update		2023-12-28 15:48:04 +08:00
..
colossal_moe	update pytest	2023-12-27 16:05:00 +08:00
tests	update pytest	2023-12-27 16:05:00 +08:00
README.md	update readme	2023-12-26 17:39:38 +08:00
infer.py	update	2023-12-26 17:32:59 +08:00
infer.sh	script	2023-12-26 17:33:32 +08:00
requirements.txt	init	2023-12-14 17:52:05 +08:00
setup.py	update	2023-12-15 16:32:32 +08:00
train.py	update	2023-12-28 15:48:04 +08:00
train.sh	script	2023-12-26 17:33:32 +08:00
version.txt	update	2023-12-15 16:32:32 +08:00

README.md

Mixtral

Usage

1. Installation

Please install the latest ColossalAI from source.

CUDA_EXT=1 pip install -U git+https://github.com/hpcaitech/ColossalAI

Then install dependencies.

cd ColossalAI/applications/ColossalMoE
pip install -e .

Additionally, we recommend you to use torch 1.13.1. We've tested our code on torch 1.13.1 and found it's compatible with our code.

2. Inference

Yon can use colossalai run to launch inference:

bash infer.sh

If you already have downloaded model weights, you can change name to your weights position in infer.sh.

3. Train

You first need to create ./hostfile, listing the ip address of all your devices, such as:

111.111.111.110
111.111.111.111

Then yon can use colossalai run to launch train:

bash train.sh

It requires 16 H100 (80G) to run the training. The number of GPUs should be divided by 8. If you already have downloaded model weights, you can change name to your weights position in train.sh.