This app takes as input a Github issue body and predicts a title for it. Behind the scenes it uses a Tensor2Tensor TensorFlow model, served via TF-Serving .
(Thanks to Hamel Husain for the original concept and source data.)
Enter the body of a github issue or the url of a github issue and click on Submit. The model then tries to generate a title or summary of the issue.
Populate Random Issue
Generate Title
This demo is run using Kubeflow - a machine learning toolkit for Kubernetes. Kubeflow is dedicated to making deployment of machine learning on Kubernetes simple, portable and scalable.