Creates a docker image based on the official pytorch image with all you need for the fastai course. Also includes a convience script for people not familiar with docker.
- Nvidia Drivers installed on the host
- Install Docker
- Install Nvidia Docker Engine
- Clone repository:
git clone https://github.com/wAuner/fastai_docker.git cd fastai_docker- make script executable
chmod +x start - build and run:
./start
Next time you want to run it, just use ./start again. It checks if an image and a container exists.
If a container exists, it will reuse it.
It also makes the current directory available to the docker container under /workspace/mydata.
Note that if you create files (e.g. notebooks) from within docker, these will belong to root.