Start your docker by running:
docker-compose up -d
from the rasa-docker directory.
Learn more about this from: https://rasa.com/docs/core/installation/
Run this to train with Markdown file:
curl --request POST --header 'content-type: application/x-yml' --data-binary @config_train_server_md.yml --url 'localhost:5000/train?project=test_model'
Then query like this:
curl 'http://localhost:5000/parse?q=bye&project=test_model'
It should reply with goodbye
intent high confidence score
curl 'http://localhost:5000/parse?q=I%20really%20need%20to%20know&project=test_model'
It should reply with learning
intent high confidence score
config_train_server_md.yml See also: https://rasa.com/docs/nlu/dataformat/ See also: https://github.com/RasaHQ/rasa-nlu-trainer
language: "en"
pipeline: "spacy_sklearn"
data: |
## intent:goodbye
- bye
- goodbye
## intent:learning
- how do I
- teach me about
- I need to know
- do you know how to
curl 'https://raw.githubusercontent.com/RasaHQ/rasa_nlu/master/sample_configs/config_train_server_json.yml' | \
curl --request POST --header 'content-type: application/x-yml' --data-binary @- --url 'localhost:5000/train?project=test_model'
See file attached for a local JSON file.
curl --request POST --header 'content-type: application/x-yml' --data-binary @config_train_server_json.yml --url 'localhost:5000/train?project=test_model'
See: https://rasa.com/docs/nlu/http/
-
delete models:
curl -X DELETE localhost:5000/models?project=test_model&model=model_XXXX
The server can't train more models right now!
See: RasaHQ/rasa#1323
https://hub.docker.com/r/rasa/duckling/
$ curl -XPOST http://0.0.0.0:8000/parse --data 'locale=en_GB&text=tomorrow at eight'
https://hub.docker.com/r/rasa/rasa_nlu/
If wget not installed on Mac:
brew install wget --with-libressl