Skip to content

Latest commit

 

History

History
executable file
·
63 lines (44 loc) · 2.36 KB

README.md

File metadata and controls

executable file
·
63 lines (44 loc) · 2.36 KB
Raw

Snips NLU docker Python App

This docker container, runs a python application that contains the latest Snips NLU and allows to interact with it over MQTT

Setup

To setup, simply create a yaml file config.yaml and enter in your MQTT broker details following the examples below

MQTT_BROKER: xxx.xxx.xxx.xxx
MQTT_PORT: 1883
MQTT_USERNAME: username
MQTT_PASSWORD: password
MQTT_CLIENTNAME: snips-nlu

A volume config should be passed on to the docker container, within which the config and training files will reside The create two folders

  • training/entities: This is where the entities will reside
  • training/intents: This is where the intents will reside

Then using the training MQTT topic, instruct it to train.

The application accepts each intent and entity as a separate file in their respective directives. Documentation on writing entities and intents can be found on Snips documentation here

MQTT Topics

Outlined below are the topics used and their use cases

Topic Use Payload Type Example
snips/nlu/request/parse Used to send data to app to parse dict {"text": "turn on my light", "session_id": 1234}
snips/nlu/response/parse Response from parsing the text dict {"status": "success", "intent": "turnOnLights", "intent_slots": {}}
snips/nlu/request/train Used to instruct Snips to train str {"session_id": 1234}
snips/nlu/response/train Response from snips training dict {"session_id": 1234, "status": "success"}

Build

$ docker build -t "odianosen/snips-nlu-app" .

Deploying

docker run --name snips-nlu-app -d -v $PWD:/config snips-nlu-app:latest

Running in interactive DEBUG mode

docker run --name snips-nlu-app -it --rm -v $PWD:/config snips-nlu-app:latest -c /config -d DEBUG

License

View license information for the software contained in this image.