Transfer Learning Example lacks the depth required to showcase Flux

[logankilpatrick]

Right now, the TL example: https://fluxml.ai/tutorials/2020/10/18/transfer-learning.html really could use a lot more depth to it. TL is such a useful task and this example does not currently result in a very explainable result. I am working on an updated version which I will port back here but am running into issues: https://github.com/logankilpatrick/DeepLearningWithJulia/blob/main/src/transfer_learning.ipynb

[logankilpatrick]

Ideally, this would include at least some commentary on the way the data is being setup, some way of knowing the model trained (adding validation to training) or checking accuracy at the end with the test set, etc.

[mcabbott]

Comments above refer to the version on fluxml.github.io, which https://github.com/FluxML/fluxml.github.io/pull/156 has deleted. Moving the issue to the model zoo.