Commit 1f6077bd authored by chenzk's avatar chenzk
Browse files

v1.1

parent ee7bfade
......@@ -10,7 +10,20 @@ Vision Transformer先将图像用卷积进行分块以降低计算量,再对
Transformer的核心思想是利用注意力模块attention提取特征:
![img](./images/attention.png)
## 环境配置
### Docker
```
docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:1.10.0-centos7.6-dtk-23.04-py38-latest
docker run --shm-size 10g --network=host --name=megatron --privileged --device=/dev/kfd --device=/dev/dri --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v $PWD/megatron-deepspeed-vit:/home/megatron-deepspeed-vit -it <your IMAGE ID> bash
pip install -r requirements.txt
```
### Dockerfile
```
cd megatron-deepspeed-vit/docker
docker build --no-cache -t megatron-:latest .
docker run --rm --shm-size 10g --network=host --name=megatron --privileged --device=/dev/kfd --device=/dev/dri --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v $PWD/../../megatron-deepspeed-vit:/home/megatron-deepspeed-vit -it megatron bash
# 若遇到Dockerfile启动的方式安装环境需要长时间等待,可注释掉里面的pip安装,启动容器后再安装python库:pip install -r requirements.txt
```
### Anaconda
1、关于本项目DCU显卡所需的特殊深度学习库可从光合开发者社区下载安装:
https://developer.hpccube.com/tool/
```
......
# 模型编码
modelCode=360
modelCode=362
# 模型名称
modelName=megatron-deepspeed-vit_pytorch
# 模型描述
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment