I want to perform a multilabel image classification task for n classes. I\'ve got sparse label vectors for each image and each dimension of each label vector is currently encode
