Container code for data science ops.
I'm getting tired of messing around with CUDA and CUDNN versions on my local machine. Thus, I'm not going to deal with that anymore. Instead, I'm going to do driver maintainence on the local machine and run everything else from Docker. It's overhead intensive, but in deployment, this should jive with the server image.
In the event you haven't noticed, this is not professional by any means. This is convience code.