Pretraining on unlabled+labled data

[alenrooni]

I have a classification task with some labeled data and a large amount of unlabeled but related data. Is there a way i can use both of them in JRAE? I have used JRAE for text classification tasks and i'm trying to measure the effect of background text pretraining on accuracy.

[sancha]

As of right now, this is not supported. A little bit of plumbing is required.