This repository contains the code for the AllenNLP demo, a web application that demonstrates the capabilities of AllenNLP, an open source library for NLP research.
First, make sure you have Docker and Python 3 (3.7 or higher) installed. After that's complete, you can start things by running this command:
./demo start
This will start a local instance of the UI and a reverse proxy that forwards API requests to production. After this command succeeds, you should be able to open http://localhost:8080 in a browser and see a fully functional version of the demo.
If you're only working on the UI, you don't need to do anything else. As you make changes to the code they'll be automatically applied. While you're working you'll probably want to stream the logs from the UI container to your terminal, so you can see errors when they occur:
./demo logs --follow ui
By default the demo will send requests to API endpoints running in production. If you'd like to run an endpoint locally, just run the start command with the ID of the endpoint:
./demo start bidaf
This will build and start that endpoint and update the reverse proxy configuration so that requests for that model are routed to your local machine instead of the production site.
Change to the Python code related to that endpoint will automatically be applied. Again, you'll probably want to view the logs output by the container while working on it, which you can do like so:
./demo logs --follow bidaf
The ./demo command provides a bunch of other useful commands. You can see full usage information
like so:
./demo --help