cameronaaron / conversationai-models

A repository to house model building experiments and tools that are part of the Conversation AI effort.

ConversationAI Models

This repository is contains example code to train machine learning models for text classification as part of the Conversation AI project.

Outline of the codebase

• experiments/ contains the ML training framework.
• annotator-models/ contains a Dawid-Skene implementation for modelling rater quality to produce better annotations.
• attention-tutorial/ contains an introductory ipython notebook for RNNs with attention, as presented at Devoxx talk "Tensorflow, deep learning and modern RNN architectures, without a PhD by Martin Gorner"
• kaggle-classification/ early experiments with Keras and Estimator for training on the Jigsaw Toxicity Kaggle competition. Will be superceeded by experiments/ shortly.
• model_evaluation/ contains utilities to use a model deployed on cloud MLE, and some notebooks to illustrate typical evaluation metrics.

About this code

This repository contains example code to help experiment with models to improve conversations; it is not an official Google product.