Rasa core does not execute correct action

Rasa Open Source
1

I am trying to integrate rasa core and nlu. NLU is able to predict the correct intent, but rasa core does not execute correct action. can anyone help me with this? I have tried hard but still unable to figure out, what’s going wrong. My code files are given below:

This is my stories.md

## greeting
    * greet
      - utter_greet

    ## sad path 1
    * mood_unhappy
      - utter_cheer_up

    ## say goodbye
    * bye
      - utter_goodbye

This is my domain.yml

intents:
  - greet
  - bye
  - mood_great
  - mood_unhappy

actions:
- utter_greet
- utter_cheer_up
- utter_goodbye

templates:

  utter_greet:
  - "Hello, how are you dude?"
  
  utter_cheer_up:
  - "Dont worry mate!!"

  utter_goodbye:
  - "Bye Bye"

This is my nlu.md

## intent:greet
- hey
- hello
- hi
- good morning
- good evening
- hey there

## intent:bye
- bye
- goodbye
- see you around
- see you later

## intent:mood_great
- perfect
- very good
- great
- amazing
- wonderful
- I am feeling very good
- I am great
- I'm good

## intent:mood_unhappy
- sad
- very sad
- unhappy
- bad
- very bad
- awful
- terrible
- not very good
- extremly sad
- so sad

This is my python code:

class rasaCore:
    
    def __init__(self, domain_path, model_path, stories_path):
        
        self.domain_path = domain_path
        self.model_path = model_path
        self.train_data_path = stories_path
        self.interpreter = RasaNLUInterpreter("./nluModels/default/rasa_nlu_model/")
        
    def trainCoreModel(self):
        
        agent = Agent(self.domain_path, policies = [MemoizationPolicy(), KerasPolicy()], 
                          interpreter=self.interpreter)
        data = agent.load_data(self.train_data_path)
        agent.train(data, augmentation_factor = 50,
                    epochs = 50,
                    batch_size = 10,
                    validation_split = 0.2)
        agent.persist(self.model_path)
    
    def generateIntentAndEntity(self, message):
        
        parsed_nlu_msg = self.interpreter.parse(message)
        print(json.dumps(parsed_nlu_msg, indent=2))
        
    def talkWithBot(self, message):
        
        self.generateIntentAndEntity(message)
        agent = Agent.load(self.model_path)
        print ("\n ============================= Core response ============================= \n")
        print (agent.handle_text(message))
        
# call rasa nlu 
training_data = "nlu.md"
conf_path = "nlu_config.yml"
train = trainModel(training_data, conf_path)
# #start training
train.startTraining()

# core 
print ("training rasa core started \n")
core = rasaCore("./domain.yml","./coreModels/dialogue","./stories.md")
core.trainCoreModel()
print ("training rasa core completed \n")

Even intent predicted by nlu is greet, but core gives response with other action or any empty list.

Could anyone please point out what i am doing wrong?

Thanks very much

2

Any reason you’re not just using the run/train scripst to run/train your bot? If you’re having issues with core prediction, I’d suggest using the --debug flag when running core to see what’s going on Debugging