Files for building the Docker Deep Learning pod.
For more information on what this is all about, see charlesreid1.com/wiki/Docker/Pods/Deep_Learning.
Dockerfile needs to be cleaned up a bit and tested out on production system.
Volume strategy will need to be determined. Need access to:
- Training data
- Pre-prepared notebooks/scripts
- Input files
- Place to write exported, finalized models
Things that need to be fixed first:
- Software modifications to
lfw_fuel
to preserve existing downloads - How to actually export (and import) neural network models in reproducible way (HDF5?)