What's the right config format for rasa server


(e8180kimo) #1

I am following a tutorial to build my rasa bot on my local webserver

This tutorial is using json format for the config file. However, my config is yml format. To build rasa onto a web server, do I have to use the json format? –if so, is there an easy way to convert my yml to json?

Does rasa_nlu.server only support json?

This is my yml: language: “en”

pipeline:

  • name: “nlp_spacy”
  • name: “tokenizer_spacy”
  • name: “ner_crf”
  • name: “intent_featurizer_count_vectors”
  • name: “intent_classifier_tensorflow_embedding”

(Akela Drissner) #2

This tutorial is old I think. The format is yaml now, so your config file is fine


(e8180kimo) #3

But when I run the server using python -m rasa_nlu.server -c config_spacy.yml --path /models and curl -XPOST localhost:5000/parse -d '{"q":"motivation"}' , it returns:

{
  "intent": {
    "name": null,
    "confidence": 1.0
  },
  "entities": [],
  "text": "motivation",
  "project": "default",
  "model": "fallback"

This is incorrect though. “motivation” is under intent:learner_typetopic in my nlu_data.md


(Akela Drissner) #4

What does the log say when you start the server? It looks like no models are being loaded from that folder.


(e8180kimo) #5

$ python -m rasa_nlu.server -c config_spacy.yml --path ./Models/

/Users/wisionlearning/anaconda3/lib/python3.6/site-packages/h5py/init.py:36: FutureWarning: Conversion of the second argument of issubdtype from float to np.floating is deprecated. In future, it will be treated as np.float64 == np.dtype(float).type.

from ._conv import register_converters as _register_converters

2018-08-31 13:58:35-0700 [-] Log opened.

2018-08-31 13:58:35-0700 [-] Site starting on 5000

2018-08-31 13:58:35-0700 [-] Starting factory <twisted.web.server.Site object at 0xb1514f6a0>


(Akela Drissner) #6

Run it with the --debug flag? What’s in the ./models/ directory?