myNLU Rasa in 3 minutes

You will need to have installed command line versions of: heroku node npm git

Required Tools Links

• HEROKU CLI Installation: https://devcenter.heroku.com/articles/heroku-cli
• Node & NPM Installers: https://nodejs.org/en/download/
• Git Installation: https://git-scm.com/book/en/v2/Getting-Started-Installing-Git

Get the myNLU Rasa quickstart and configure it

From the command line clone the repository into a mynlu-qs directory:

git clone https://github.com/AzumoHQ/mynlu_quickstart.git mynlu-qs

Go into that directory:

cd mynlu-qs

Install node dependencies:

npm install

Setup the application with the name mynlu-qs-test (you WILL need to choose another name):

npm run setup -- mynlu-qs-test

This final step will check that you have heroku, npm, node, git installed, create a Heroku app with the name mynlu-qs-test, provision the mynlu-rasa add-on, and create a local .env file with the settings for your mynlu-rasa access.

You’re good to go!

Starting your app locally

From the command line run:

npm run start

Now point your browser to: http://localhost:3000

You'll be able to see this:

You need to train your model, press "Choose File" and select a training file, you can find one of exame under model/demo_rasa.json, then press "Train" button.

After you've trained your model at least once, you can type on the sentence text box and click on the "Parse Sentence" button:

The app will call your mynlu-rasa instance and show the results below.

Grooming your model

You can edit the example training file: ./model/demo_rasa.json or create your own file and upload it using the "Train" section.

Final words

You can deploy your new app to Heroku via git, check out this tutorial here: https://devcenter.heroku.com/articles/git