batch_size should be a multiple of total no of images else error comes

PacktPublishing / Deep-Learning-for-Computer-Vision

Deep Learning for Computer Vision, by Packt

MIT License

213 stars 167 forks source link

batch_size should be a multiple of total no of images else error comes #7

Closed lokeshsoni closed 5 years ago

lokeshsoni commented 5 years ago

batch_size should be a multiple of total no of images else error comes while training as it does not match with total no of labels
sigmoid to be used when training a binary classifier
binary_cropssentropy instead of categorical_crossentropy

rajacheers commented 5 years ago

@lokeshsoni thanks for refactoring. But this level of change may disconnect users from the book. Will be happy to merge the binary cross entropy correction