You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have an LSTM model built with Keras framework. This model is for the next word prediction based on the previous text entered by the user.
So I have written a function in python which is loading this Keras model and a tokenizer file(for word-sequences) into it and using inbuilt functions like predict_classes and pad_sequences to make next word predictions.
I have to convert this function to C++, so can anyone help me out how to use this repo for my use. I tried using Cython but that didn't help as it converts a 52 line code in Python to 4000 line code in C++ which is not convenient. It is inconvenient to debug the 4000 line of code without knowing its relevance.
Activity
MatthewScholefield commentedon Sep 11, 2018
Of relevance: #1
rspadim commentedon Feb 4, 2019
Not sure but maybe numpy keras Backend?
https://github.com/keras-team/keras/blob/master/keras/layers/recurrent.py
https://github.com/keras-team/keras/blob/master/keras/backend/numpy_backend.py
sidharth1010 commentedon May 20, 2019
I have an LSTM model built with Keras framework. This model is for the next word prediction based on the previous text entered by the user.
So I have written a function in python which is loading this Keras model and a tokenizer file(for word-sequences) into it and using inbuilt functions like predict_classes and pad_sequences to make next word predictions.
I have to convert this function to C++, so can anyone help me out how to use this repo for my use. I tried using Cython but that didn't help as it converts a 52 line code in Python to 4000 line code in C++ which is not convenient. It is inconvenient to debug the 4000 line of code without knowing its relevance.