Repository for setting up the ABC-toolkit.
- Docker. For linux, I used this guide.
- nvidia-container-toolkit
On Windows, I recommend installing Windows Subsystem for Linux (WSL). This isn't required but I haven't tested without it and the commands below may be slightly different.
On Linux distributions, you will need to add the current user to the docker
group with sudo usermod -a -G docker USERNAME
.
-
Make a copy of
.env-default
and name it.env
. Update parameters if needed. For testing, default values should work, just create the following directories:./data/inputs
&./data/outputs
.INPUT_DIR
is mounted to/data/inputs
in all the containers. Your data should be (anywhere) inINPUT_DIR
.OUTPUT_DIR
will contain all container outputs.
-
Make the directory
backend/models
. In practice, this will contain all ML models but for now, leave it empty. -
Build the base image:
./build_base_image.sh
If so, all containers are running and the backend can detect the GPU!!