This is awesome and I really wanted to use it, however at some point it is converting the sparse numpy matrix to a dense matrix, which is unusable for my purposes because it causes a memory exception. Text classification in general uses pretty sparse feature vectors. The interface for this package is really nice otherwise.