1 min readDec 29, 2018
- I looked up the code and explored how things work,You can also explore by setting different batch sizes, store the resultant array in a variable and view it’s size,order of the values,etc..
- For the train and validation generators it is not necessary to set batch size exactly to the number of samples in the respective sets because they will be sampled sooner or later in the next epoch,but it is always good to have a batch size that divides exactly the number of samples in the set, but for the test generator you should set batch size that divides your test set exactly, if you’re not sure set this to 1, because you would expect the predict generator to return exactly the same no. Of files in test set,no less no more.
