nn- get training batches directly from database
In the CNNWhale class, the 'train()' method uses mini-batches to train the network. However, those batches are generated within this method and require the entire training set to be stored in the instance attributes 'train_x' and 'train_y'.
Improvement: have a method (or maybe a function/set of functions in int the data_handling module) to only read a batch from the database. This will allow the network to train on larger than memory datasets.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information