Exporting model for serving

[MarthinusBosman]

I'm trying to figure out if it's possible to export the estimator for serving inputs, and it's interesting that you have a gen_input_fn function instead of just an input_fn when predicting. Is the logic for the gen_input_fn part of the nmt repo or did you write it yourself? Asking because I can't figure out how to implement the serving_input_fn function for this model.

In another hope to get it right, do you maybe know how I could get the input tensor and output tensor info?

[davidhughhenrymack]

You should be able to serve in the normal way. Gen_input_fn is a function generator, it returns the standard input_fn

…

On Fri, Aug 17, 2018 at 5:35 PM MarthinusBosman ***@***.***> wrote: I'm trying to figure out if it's possible to export the estimator for serving inputs, and it's interesting that you have a gen_input_fn function instead of just an input_fn when predicting. Do you maybe know how I could get the input tensor and output tensor info? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#5>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AOUFIlqA-Jj6Z3Rrg-QgptsvDl6L-qbNks5uRvDmgaJpZM4WB214> .