Rasa HTTP API model train request body format

Rasa Open Source
1

Hi, I have few issues with the request body format for model train HTTP API. According to the HTTP API documentation for model train , the config should be provided in plain text format. I don’t understand what do you mean by plain text and the example given is not suffice. { “config”: “language: en\npipeline: supervised_embeddings\npolicies:\n - name: MemoizationPolicy\n - name: TEDPolicy” }

Here, what is the format for providing parameters to your component? image

Eg: How to provide number of epochs to Customclassifier?

I tried the following: “- name: CustomClassifier\nepochs: 10” I was not able to receive the epochs in my Customclassifier.

2

@rctatman @Tobias_Wochinger @souvikg10 @akelad @Juste @Ghostvv

3

@Tobias_Wochinger @akelad Pls Help

4

when you say “not able to receive the epochs” - can you please show what exactly is being received? If you add a debug flag (e.g. rasa run --debug ) when you start the server you should get some helpful output

5

@ravichandran The payload should look like as you’d put all data in one file, e.g:


config:
- name: SpacyNLP
- name: SpaceTokenizer
# more NLU components

nlu:
- intent: greet
  example: |
    - Hi
    - Hello
    - Hey there

stories:
- story: My story
  steps:
  - intent: greent
  - action: utter_greet

Did you further specify an HTTP header with Content-Type application/x-yaml?