I am using a generator to create batches for my model and therefore use the model.fit_generator() Keras function. Using verbose=1 (default option) I get a progress bar and an ETA during the training procedure which is quite helpful to evaluate how much time takes one epoch and how many epochs to run during the night/week-end.
The problem is that during the validation procedure, I do not have any visible information about how long it will take. The progress bar for the training stays at 100%, the ETA at 1s and that's it. Moreover, I noticed that the waiting time is not simply proportional to the amount of data. If I have 90% for training and 10% for testing, I should expect the waiting time for the validation to be 9 times shorter, but that's not the case.
Is there any way to display the progress of this validation procedure (or evaluate it based on the training and the amount of data)?
