Training rasa using http Api endpoint

Rasa Open Source
1

Hi Guys! i’m trying to use the api endpoint /model/train to train rasa using the below command

curl -X POST \
  http://localhost:5005/model/train \
  -H 'Accept-Encoding: gzip, deflate' \
  -H 'Content-Type: application/json' \
  -d '{
"config": "language: en\npipeline:\n - name: WhitespaceTokenizer\n - name: RegexFeaturizer\n - name: CRFEntityExtractor\n - name: EntitySynonymMapper\n - name:\ CountVectorsFeaturizer\n - name: EmbeddingIntentClassifier\npolicies:\n - name: MemoizationPolicy\n - name: KerasPolicy\n - name: MappingPolicy\n - name: FormPolicy\n",\
"nlu": "## intent:new_intent_to_train_on\n - new_utter_to_train_on for new_entity_to_train_on_value\n - new_utter_to_train_on for new_entity_to_train_on_value\n##\ intent:greet\n - hey\n - hello\n## intent:goodbye\n - cu\n - goodbye\n## intent:mood_great\n - Great!!\n - happy\n## intent:mood_unhappy\n - sad\n - Terrible",\
"stories": "##new_intent_to_train_on Path from model train without utter_slot_values\n* greet\n - utter_greet\n* new_intent_to_train_on\n -\ new_intent_to_train_on_form\n - form{\"name\": \"new_intent_to_train_on_form\"}\n - form{\"name\": null}\n\n## happy path\n * greet\n - utter_greet\n* mood_great\n -\ utter_happy\n\n## sad path 1\n * greet\n - utter_greet\n* mood_unhappy\n - utter_cheer_up\n",\
"domain": "entities:\n - new_entity_to_train_on\nintents:\n - new_intent_to_train_on\n - greet\n - goodbye\n - affirm\n - deny\n - mood_great\n -\ mood_unhappy\n\nactions:\n - utter_slots_values\n - utter_greet\n - utter_cheer_up\n - utter_did_that_help\n - utter_happy\n - utter_goodbye\ntemplates:\n\ utter_slots_values:\n - text: new_utter_output_to_send:\n\n- new_entity_to_train_message_values : {new_entity_to_train_on}\n\n\n utter_greet:\n - text: Hey! How are\  you?\n\n utter_cheer_up:\n - text: Here is something to cheer you up:\n image: https://i.imgur.com/nGF1K8f.jpg\\n\n utter_did_that_help:\n - text: Did that help\     you?\n\n utter_happy:\n - text: Great carry on!\n\n utter_goodbye:\n - text: Bye\n\nforms:\n - new_intent_to_train_on_form\nslots:\n new_entity_to_train_on:\n type:\ unfeaturized\n auto_fill: false",\
"out": "models",\
"force": true\
}'

I started rasa as an http server in one of the terminal before this using rasa run -m models --enable-api --cors “*” --debug.

The above data is just for testing and i have all the files including domain.yml file.

But i’m getting this warnings

/rasaaa/lib/python3.7/site-packages/rasa/utils/common.py:351: UserWarning: Loading domain from ‘/tmp/tmp1ns5z1bz/domain.yml’ failed. Using empty domain. Error: ‘The provided yaml file is invalid. You can use http://www.yamllint.com/ to validate the yaml syntax of your file.’

Core training was skipped because no valid domain file was found. Only an nlu-model was created.Please specify a valid domain using ‘–domain’ argument or check if the provided domain file exists.

Am i making some mistake? How can i train rasa using curl ? also is there any way i can directly add a new intent using any api endpoint? I’m using rasa 1.10.3.

Thanks in advance.

2

I did get this working in Python, but a cursory look gives this (I took you domain element and replaced \n with (in vi):

“domain”: "entities:

  • new_entity_to_train_on intents:
  • new_intent_to_train_on
  • greet
  • goodbye
  • affirm
  • deny
  • mood_great -\ mood_unhappy actions:
  • utter_slots_values
  • utter_greet
  • utter_cheer_up
  • utter_did_that_help
  • utter_happy
  • utter_goodbye templates: \ utter_slots_values:
  • text: new_utter_output_to_send:
  • new_entity_to_train_message_values : {new_entity_to_train_on} utter_greet:
  • text: Hey! How are\ you? utter_cheer_up:
  • text: Here is something to cheer you up: image: https://i.imgur.com/nGF1K8f.jpg
    utter_did_that_help:
  • text: Did that help\ you? utter_happy:
  • text: Great carry on! utter_goodbye:
  • text: Bye forms:
  • new_intent_to_train_on_form slots: new_entity_to_train_on: type:\ unfeaturized auto_fill: false",\

it appears to have several typos ?

I took the domain I wanted and then simply replaced with \n to get a single line (string).

I was thinking about this a while back and it would be nice to have a POJO (DAO type) object to use when building these to make sure it comes out syntactically correct when using API calls (granted I was using Python to make the call).

Other than the typos I see it should work as you are making the call the same way I am (headers/etc).

Jason

3

Hi Jason , Thanks for replying . I didn’t actually get what you modified to make it working .can you do some modification in the command i ran above to make me understand? Yes there may be many typos as I’m just trying to train model this way and just testing it with some random input.

4

this should work: (it is the base rasa init model), use the following as a single line. [HINT: spacing matters tremendously and you have to double the single quotes inside a string (double quotes)]

curl -X POST ‘http://localhost:5005/model/train’ -H ‘Accept-Encoding: gzip, deflate’ -H ‘Content-Type: application/json’ -d ‘{ “config”: “language: en\npipeline:\n - name: WhitespaceTokenizer\n - name: RegexFeaturizer\n - name: LexicalSyntacticFeaturizer\n - name: CountVectorsFeaturizer\n - name: CountVectorsFeaturizer\n analyzer: “char_wb”\n min_ngram: 1\n max_ngram: 4\n - name: DIETClassifier\n epochs: 100\n - name: EntitySynonymMapper\n - name: ResponseSelector\n epochs: 100\npolicies:\n - name: MemoizationPolicy\n - name: TEDPolicy\n max_history: 1\n epochs: 100\n - name: MappingPolicy\n\n”, “nlu”: "## intent:greet\n- hey\n- hello\n- hi\n- good morning\n- good evening\n- hey there\n\n## intent:goodbye\n- bye\n- goodbye\n- see you around\n- see you later\n\n## intent:affirm\n- yes\n- indeed\n- of course\n- that sounds good\n- correct\n\n## intent:deny\n- no\n- never\n- I don’‘t think so\n- don’‘t like that\n- no way\n- not really\n\n## intent:mood_great\n- perfect\n- very good\n- great\n- amazing\n- wonderful\n- I am feeling very good\n- I am great\n- I’‘m good\n\n## intent:mood_unhappy\n- sad\n- very sad\n- unhappy\n- bad\n- very bad\n- awful\n- terrible\n- not very good\n- extremely sad\n- so sad\n\n## intent:bot_challenge\n- are you a bot?\n- are you a human?\n- am I talking to a bot?\n- am I talking to a human?\n", “stories”: “## happy path\n* greet\n - utter_greet\n* mood_great\n - utter_happy\n\n## sad path 1\n* greet\n - utter_greet\n* mood_unhappy\n - utter_cheer_up\n - utter_did_that_help\n* affirm\n - utter_happy\n\n## sad path 2\n* greet\n - utter_greet\n* mood_unhappy\n - utter_cheer_up\n - utter_did_that_help\n* deny\n - utter_goodbye\n\n## say goodbye\n* goodbye\n - utter_goodbye\n\n## bot challenge\n* bot_challenge\n - utter_iamabot\n”, “domain”: “intents:\n - greet\n - goodbye\n - affirm\n - deny\n - mood_great\n - mood_unhappy\n - bot_challenge\n\nresponses:\n utter_greet:\n - text: “Hey! How are you?”\n\n utter_cheer_up:\n - text: “Here is something to cheer you up:”\n image: “https://i.imgur.com/nGF1K8f.jpg”\n\n utter_did_that_help:\n - text: “Did that help you?”\n\n utter_happy:\n - text: “Great, carry on!”\n\n utter_goodbye:\n - text: “Bye”\n\n utter_iamabot:\n - text: “I am a bot, powered by Rasa.”\n\nsession_config:\n session_expiration_time: 60\n carry_over_slots_to_new_session: true\n”, “out”: “models”, “force”: true }’

1 Like
5

Hi ! I was wondering if we can give the path of the files in .md/.json/.yml format instead of giving the data in the format discussed above as plain string. is that possible?

Thanks in advance.