Writing custom training loops

#1

Hello,

Is one able to customize what happens when calling fit on a custom DeepChem model, i.e. can one write custom training loops as you can in vanilla tensorflow.

#2

Yes for sure. I often do this by writing something like

for i in range(nb_epochs):
  # Fit one epoch
  model.fit(dataset, nb_epoch=1)
  # Do something (early stopping/logging/etc)