Keras: How to pass sample_weigths to validation data?

[rladeira]

In the method fit of a keras model, there is a parameter (sample_weights) which passes sample weights for training data. I think that is reasonable to also pass sample_weights for computing validation loss, so that overfitting can be tracked in these settings. Is this possible?

[fchollet]

Just pass as validation_data a triplet (samples, labels, sample_weights) instead of just (samples, labels).