1.Docker安装

由于Docker官网部署在国外, 可以采用通过国内的镜像下载. 这里可以使用阿里云的安装镜像.

对于10.10.3以下的用户推荐使用Docker Toolbox

http://mirrors.aliyun.com/docker-toolbox/mac/docker-toolbox/

对于10.10.3以上的用户推荐使用Docker for Mac

http://mirrors.aliyun.com/docker-toolbox/mac/docker-for-mac/

查看当前Docker版本信息

docker -version

2.拉取到本地的镜像

安装官方提供的CPU版本：

 docker pull mindspore/mindspore-cpu:0.1.0-alpha

查看拉取到本地的镜像：

docker images

docker images显示信息如下：REPOSITORY                TAG                 IMAGE ID            CREATED             SIZE
mindspore/mindspore-cpu   0.1.0-alpha         ef443be923bc        3 months ago        1.05GB信息说明：REPOSITORY： 表示当前镜像的仓库
TAG：镜像标签，一般用版本标识
IMAGE ID：镜像的唯一ID
CREATED： 镜像创建时间
SIZE： 镜像大小

创建并运行容器：

docker run -it mindspore/mindspore-cpu:0.1.0-alpha

查看容器启动情况：

docker ps

3.下载并运行MindSpore Lenet代码

克隆MindSpore Lenet代码

git clone https://github.com/mindspore-ai/docs.git

也可以去gitee上去克隆，克隆成功之后进入目录找到lenet.py文件：

root@6b7fbe26908d:/# git clone https://gitee.com/mindspore/docs.git
Cloning into 'docs'...
remote: Enumerating objects: 2240, done.
remote: Counting objects: 100% (2240/2240), done.
remote: Compressing objects: 100% (210/210), done.
remote: Total 7271 (delta 2118), reused 2059 (delta 2030), pack-reused 5031
Receiving objects: 100% (7271/7271), 21.54 MiB | 525.00 KiB/s, done.
Resolving deltas: 100% (5059/5059), done.
root@6b7fbe26908d:/# ls
bin  boot  dev  docs  etc  home  lib  lib64  media  mnt  opt  proc  root  run  sbin  srv  sys  tmp  usr  var
root@6b7fbe26908d:/# cd docs/
root@6b7fbe26908d:/docs# ls
CONTRIBUTING_DOC.md  CONTRIBUTING_DOC_CN.md  LICENSE  LICENSE-CC-BY-4.0  NOTICE  README.md  README_CN.md  api  docs  install  resource  tutorials
root@6b7fbe26908d:/docs# cd tutorials/
root@6b7fbe26908d:/docs/tutorials# ls
Makefile  notebook  requirements.txt  source_en  source_zh_cn  tutorial_code
root@6b7fbe26908d:/docs/tutorials# cd tutorial_code/
root@6b7fbe26908d:/docs/tutorials/tutorial_code# ls
distributed_training  lenet.py  model_safety  resnet  sample_for_cloud
root@6b7fbe26908d:/docs/tutorials/tutorial_code# vim lenet.py

196行的device_target default = "CPU"就可以跑起来：

if __name__ == "__main__":parser = argparse.ArgumentParser(description='MindSpore LeNet Example')parser.add_argument('--device_target', type=str, default="CPU", choices=['Ascend', 'GPU', 'CPU'],help='device where the code will be implemented (default: CPU)')

运行lenet.py

root@6b7fbe26908d:/docs/tutorials/tutorial_code# python lenet.py 
******Downloading the MNIST dataset******
============== Starting Training ==============
epoch: 1 step: 1, loss is 2.3070438
epoch: 1 step: 2, loss is 2.3044722
epoch: 1 step: 3, loss is 2.3040059
epoch: 1 step: 4, loss is 2.3056054
epoch: 1 step: 5, loss is 2.3035357

运行结束之后提示：

epoch: 1 step: 1873, loss is 0.056840077
epoch: 1 step: 1874, loss is 0.3317723
epoch: 1 step: 1875, loss is 0.013712672
============== Starting Testing ==============
============== Accuracy:{'Accuracy': 0.9628405448717948} ==============

Mac 快速上手 MindSpore Lenet代码

1.Docker安装

2.拉取到本地的镜像

3.下载并运行MindSpore Lenet代码