MindSpore Tutorial In Docker
This is a MindSpore tutorial in WSL2’s docker.
Install WSL2
First, install WSL2. We can refer to the installation tutorial in Windows 10 official website. I select Ubuntu 20.04 linux distribution.
Install Docker
$ curl -fsSL https://get.docker.com -o get-docker.sh
$ sudo sh get-docker.sh
$ sudo service docker start
$ sudo systemctl enable docker
$ sudo usermod -aG docker $USER
Install MindSpore
We select CPU version to install：
$ docker pull mindspore/mindspore-cpu:0.3.0-alpha
After installation, you can see the image we installed:
$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
mindspore/mindspore-cpu 0.3.0-alpha e9b3f15488e5 2 months ago 1.03GB
Then create our own container based on this image we just installed:
$ docker run -it mindspore/mindspore-cpu:0.3.0-alpha /bin/bash
Download MindSpore
Download MindSpore Lenet code:
$ git clone https://github.com/mindspore-ai/docs.git
Save this change so that there is no need to repeat the clone when starting the container later. Docker supports only submitting incremental modifications based on the original image to form a new image. Later, use this new image as a template to start the container, and the cloned files will exist in the container, so there is no need to clone repeatedly.
First, with docker ps -l to find the container ID of the cloned mindspore package. Then submit this container as a new image, which is named kanchenhao/mindspore here.
$ docker ps -l
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
25e14b1016ac mindspore/mindspore-cpu:0.3.0-alpha "/bin/bash" About a minute ago Exited (0) 5 seconds ago vibrant_goodall
$ docker commit 25e14b1016ac kanchenhao/mindspore
$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
kanchenhao/mindspore latest f0556d73f219 22 seconds ago 1.13GB
mindspore/mindspore-cpu 0.3.0-alpha e9b3f15488e5 2 months ago 1.03GB
Run Lenet
$ docker run -it kanchenhao/mindspore /bin/bash
$ cd /docs/tutorials/tutorial_code
$ python lenet.py
============== Starting Training ==============
epoch: 1 step: 1, loss is 2.3035789
epoch: 1 step: 2, loss is 2.3001177
epoch: 1 step: 3, loss is 2.2996633
epoch: 1 step: 4, loss is 2.3047967
...
epoch: 1 step: 1872, loss is 0.12087243
epoch: 1 step: 1873, loss is 0.074895434
epoch: 1 step: 1874, loss is 0.09245071
epoch: 1 step: 1875, loss is 0.07100312
============== Starting Testing ==============
============== Accuracy:{'Accuracy': 0.9629407051282052} ==============
I will continue to share my study notes about MindSpore in the follow-up, welcome everyone to give me suggestions and leave comments.