Implementing the revised approach
In this section, we will cover each part of the implementation. You can find the code by using this GitHub link: https://github.com/jalajthanaki/Chatbot_tensorflow. Note that here, I'm using TensorFlow version 0.12.1. I perform training on a GeForce GTX 1060 6GB GPU for a few hours. In this implementation, we don't need to generate features because the seq2seq model generates its internal representation for sequences of words given in a sentence. Our implementation part has the following steps:
Data preparation
Implementing the seq2seq model
Let's begin our coding.
Data preparation
During this implementation, we will be using the Cornell movie-dialogs dataset. First of all, we need to prepare data in a format that we can use for training. There is a Python script that is used to perform data preparation. You can find the script at: https://github.com/jalajthanaki/Chatbot_tensorflow/blob/master/data/prepare_data_script/data.py.
Data preparation can be subdivided...
