Keras loss_weights and class_weights

HI, is there a way to input loss_weights (argument of compile in Keras), and class_weights (argument of fit)? (see https://keras.io/models/sequential/)

Thanks! The Keras integration is awsome!

Francesco

Hi Francesco,

This is unfortunately not possible using the Keras learner node at the moment. However, if scripting in Python works for you, you could use a DL Python Learner node to manually compile and fit the model.

Adding weighting to the Keras learner node is definitely on our list of planned features for the future. I can’t make any promises when it will make it into a release, though.

Thanks for your nice feedback, we really appreciate it! :slight_smile:

Marcel