Rasa agent stop responding after one story has spoken

I wrote a simple bot using rasa. To handle messages I create the flask app and load the agent into that app. I receive user message and id from the request and put it to agent handle_text method then I got the response. The problem is that after I spoke one story that defined in my story.md my agent stops answering.

Here is my flask app

app = Flask(__name__)


# Define rasa interpreter
interpreter = None

# Define rasa agent
agent = None


@app.route('/')
def index():

    # Receive message from request
    message = request.args.get('msg')

    # Receive user id from request
    user_id = request.args.get('uid')

    #  Validation
    if not message:
        return 'No message specified in field \'msg\''
    if not user_id:
        return 'No user id specified in field \'uid\''

    # Put received message into rasa agent
    answers = agent.handle_text(message, sender_id=user_id)

    # Define text for the response
    text = None

    if len(answers) > 0:
        text = "User: {} | {}".format(user_id, answers[0].get('text'))
    else:
        text = "User: {} | Nothing to answer".format(user_id)

    return text


if __name__ == '__main__':

    # Load rasa interpreter
    interpreter = RasaNLUInterpreter(NLU_PATH)

    # Load rasa agent
    agent = Agent.load(CORE_PATH, interpreter=interpreter)

    app.run()

My stories.md is

## Simple flow
* greet
  - utter_greet
* bye 
  - utter_bye 
## Order pizza
* greet
  - utter_greet
* order_pizza_type
  - utter_finish_order_pizza
* bye
  - utter_bye
## Story
* order_pizza_type
  - utter_finish_order_pizza
## Generated Story -1054914010798310995
* greet
    - utter_greet
* order_pizza_type{"Country": "mexican"}
    - utter_finish_order_pizza
* bye
    - utter_bye
## New Story
* greet
  - utter_greet
* order_pizza_wish
  - utter_finish_order_pizza
* bye
    - utter_bye

and my config.yml

language: "en"

pipeline:
- name: "nlp_spacy"
- name: "tokenizer_spacy"
- name: "ner_crf"
- name: "tokenizer_whitespace"
- name: "intent_featurizer_count_vectors"
- name: "intent_classifier_tensorflow_embedding"
  intent_tokenization_flag: true
  intent_split_symbol: "+"

policies:
  - name: "KerasPolicy"
    featurizer:
    - name: MaxHistoryTrackerFeaturizer
      max_history: 5
      state_featurizer:
        - name: BinarySingleStateFeaturizer
  - name: "MemoizationPolicy"
    max_history: 5
  - name: "FallbackPolicy"
    nlu_threshold: 0.4
    core_threshold: 0.3

My expected result

$ curl -X GET "https://localhost?msg=hello&uid=1"
$ curl -X GET "https://localhost?msg=I want to order pizza&uid=1"
$ curl -X GET "https://localhost?msg=Bye&uid=1"
$ curl -X GET "https://localhost?msg=hello&uid=1"

Response

> User: 1 | Hey! How are you?
> User: 1 | Ok I will deliver pizza for you
> User: 1 | Bye
> User: 1 | Hey! How are you?

But my actual results are

$ curl -X GET "https://localhost?msg=hello&uid=1"
$ curl -X GET "https://localhost?msg=I want to order pizza&uid=1"
$ curl -X GET "https://localhost?msg=Bye&uid=1"
$ curl -X GET "https://localhost?msg=hello&uid=1"

Response

> User: 1 | Hey! How are you?
> User: 1 | Ok I will deliver pizza for you
> User: 1 | Bye
> User: 1 | Nothing to answer

As you can see no response for the second message "hello" after one storyline has been spoken.

Solution

Same as suggested in the comments I'd suggest to use interactive learning to debug your bot and create new training stories. Currently you have very sparse training data.

Did you use augmentation for the training? If you did not specify the parameter differently, the default augmentation is set to 20.

If you are using augmentation I'd suggest to also add another short story to handle a standalone greet:

## Simple flow
* greet
  - utter_greet

One more thing: It is recommended to use general intents and distinguish them by the recognized entities. Hence, instead of order_pizza_type and order_pizza_wish it would be better to have an intent order_pizza or even order and then slots for food_type, product_to_order (e.g. pizza) and so on. If you have very similar intents such as order_pizza_type and order_pizza_wish NLU will have a tough live to distinguish them.