Made sure BinaryCrossEntropy return the losses of the batch.

SMART-Lab / smartlearner

SMART-Learner is a machine learning library built with researchers in mind.

BSD 3-Clause "New" or "Revised" License

10 stars 4 forks source link

Made sure BinaryCrossEntropy return the losses of the batch. #63

Closed MarcCote closed 8 years ago

MarcCote commented 8 years ago

The BinaryCrossEntropy and ClassificationError were returning the loss of the whole batch instead of the loss of each individual example (as done with other Loss functions, e.g. L2Distance and L1Distance).

This PR fixes that and brings some unit tests.

ASalvail commented 8 years ago

For some reason, the perceptron test hangs. Will you have a look?