One click button to deploy example application about rasa HOT 9 CLOSED

might make more sense to dockerise the project and use the docker cloud deploy button https://docs.docker.com/docker-cloud/apps/deploy-to-cloud-btn/ . thoughts?

from rasa.

An easy way to get started quickly would be to publish this project on https://hub.docker.com/. Then developers can download and run the image on their laptop, and hit it from their browser, or terminal with curl or anything else.

from rasa.

I think both idea would be good...either pulling from the hub on your laptop or running it over the cloud

from rasa.

@amn41 We have this working with Docker, SpaCy and Scikit. Currently, though, it pulls from API.ai to get all our agent data to train the model, and our api key is hardcoded in there. However, model training with that setup and our data takes about 5 seconds.

We'll try and get something up in the next few days.

from rasa.

@samhavens Your input on #50 would be nice :)

from rasa.

I think since the current setup is completely dockerized we might be able to use heroku again using one of the small spacy models (need to try it tough - even tough the language model on disk is only 50mb it might still be larger in mem).

from rasa.

@tmbo In version 2, they're bringing down the 50mb model to 15mb, so even more feasible for Heroku (see these comments)

from rasa.

Yes, very good point. I think that makes this even more feasible 👍

from rasa.

We do have the interactive sessions in the docs now which is even better

from rasa.