java - Pass array of floats when training Stanford CRFClassifier -

for every word in document, i'm looking add in series of floating point numbers features stanford ner's crfclassifier train on. unfortunately, documentation on stanford ner's .prop files hasn't made clear how pass in custom features. in general, how 1 go adding custom features stanford ner training set?

you find answer inside 'nerfeaturefactory' class.