Using for production purpose

[ghost]

I'm trying to use my trained model as an API that input is text and the output is the labeled text. However, it's seem like every time i want to deploy ,it's need to initialize the model again. Can i separate the initialization and the predicting part so that I just have to load the model once then I can use it for predicting several inputs at different times.
Thank you for your amazing model :)

I have read the thread #5 for this issue but i didn't quite get the solution for this problem.

[Franck-Dernoncourt]

There is no solution yet. However we are working on it :-)

[Franck-Dernoncourt]

Separating the initialization and the predicting part was done on the series of commits on Jul 11, 2017.