I have a featureset for each token of the input, e.g. {\'ft1\': \'something\', \'ft2\':102} for a token. I need to vectorized it and pass into sklearn\'s SGDClassifier.