Skip to content

A barebones Rasa NLU server installation using Docker

Notifications You must be signed in to change notification settings

just-ai/rasa-heroku-template

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Rasa NLU server template

This template contains all you need to deploy Rasa NLU server on Heroku cloud to make your Rasa project visible globally.

How to use

Click on the button below to deploy this template on your Heroku instance. Heroku will automatically build the Docker image and your project's NLU model.

Deploy to Heroku

It takes a couple of minutes to build and start the server. You can see the progress in Heroku logs.

How to make requests

Once your server is deployed, you can make requests to your NLU model via Rasa HTTP API For example:

curl https://<your Heroku application name>.herokuapp.com/model/parse -d '{"text":"hello"}'

How to change the model

Once you've deployed and tested your NLU server, you can then clone it to your machine and make changes in NLU model.

1. Clone the application

Install git and Heroku CLI. Run a terminal (or console) on your machine and type

heroku login
heroku git:clone -a <your Heroku application name>
cd <your Heroku application name>
git remote add origin https://github.com/just-ai/rasa-heroku-template
git pull origin master

You have to do these steps only once per project.

2. Make changes

Install Rasa

Install Rasa on your machine. Here is a great installation guide.

Train NLU model

Then go to the directory of your application (cloned on the previous step) and make some changes in the model. Please refer to the Rasa documentaion to learn how to build and evaluate NLU model.

Please note that you don't have to run rasa init command once your template project is already cloned from Heroku. Also note that NLU server doesn't run any actions - it only runs your NLU model. Thus you can use only rasa train nlu command.

Evaluate changes

To evaluate your changes on your local machine just run NLU server locally and make some HTTP requests to the Rasa HTTP API endpoint. You can also use shell command to try your mddel without running a server.

Push changes to Heroku

Once you've trained and evaluated your NLU model, you can push changes to Heroku.

Run the next commands in the terminal

git add .
git commit -am "some comments"
git push

Heroku will automatically handle the changes, re-build NLU model and re-start the server.

Please note that locally trained NLU models won't be pushed to the Heroku repository.

About

A barebones Rasa NLU server installation using Docker

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published