Experiments to try and get the solo 8 up and standing.
As a lot of our training is gpu-dependant, we packaged our experiments to be
runnable on nvidia-docker. Therefore, to get the stack up and running, you
just need to build the included Dockerfiles (can be found in bin/). Note that
they don't require any files from the build context.
If you would like to run the stack using different versions of gym_solo
and/or learning_experiments, you will need to replace their respective
volume mounts from within the container. They are both located within
/sources.
There is also a convenience script available to run the container with custom packages and open a jupyter instance from within it.