I can't find a definitive answer on whether keras supports scipy sparse matrices as training / testing vectors. I've read this link, which seems to imply that it doesn't https://www.kaggle.com/c/walmart-recruiting-trip-type-classification/forums/t/18154/training-neural-network-with-scipy-sparse-matrix. This PR seems to imply that it does, though https://github.com/fchollet/keras/pull/1886. The docs don't state explicitly -- could someone shed some light?
Thanks
