The performance of Glow is quite impressive to me with a first very basic MNIST train run. Now I wanted to do more advanced train runs using CNNs and am wondering, if I now need to implement the CNN by hand in Glow or simply use the image-classifier as described in the glow documentation. The documentation only tells about using the bundle for trained networks, or did I miss something?
You’re right – the bundles only support inference right now. We currently are more focused on inference over training. We have had questions about this before – see this issue on GH for a bit more details.