diliadis / deepmtp Goto Github PK

This is self-explanatory. More architecture should be made available as branch models. Even though users can easily incorporate their custom branch models, having more default options will help with ease of use.

Currently, the primary data object that holds the score matrix values and the potential instance and target features (for the training, validation, and test sets) is a dictionary. Until this point, this was a simple approach that worked well. Nevertheless, a custom python class should replace the dictionary, hopefully encouraging more functionality.