Remove unnecessary loading of validation data to memory

[angeloskath]

In ImportanceTraining classes we unnecessarily load all the data in memory.

importance-sampling/importance_sampling/training.py

Line 288 in 991230a

*dataset.test_data[:],

[MasterScrat]

This is problematic - it makes validation completely impossible in some cases, eg when using generators that load heavy images! Is there any way to get around this?

[angeloskath]

I agree it is problematic and it needs to be changed. The simplest solution I can think of is to do an external for loop and evaluate the unwrapped model. Something like the following:

model = .....
wrapped = ImportanceTraining(model)
for i in range(epochs):
    wrapped.fit(...)
    model.evaluate(...)