Can i pass top 3 predicted responses to custom action?

Rasa Open Source
1

can i pass top 3 predicted responses to custom action?

i need highest top 3 confidence scored “responses”

2

Check out the tracker object, it’s the only thing that can help with stuff like that :slight_smile:

Reference: Tracker

3

@ChrisRahme i think tracker can only return One predicted response. I need 3 predicted responses. Please help me

4 
from typing import Any, Dict, List, Text

from rasa_sdk import Action, Tracker
from rasa_sdk.events import UserUtteranceReverted
from rasa_sdk.executor import CollectingDispatcher


class ActionDefaultFallback(Action):
    def name(self) -> Text:
        return "action_default_fallback"

    def run(
            self,
            dispatcher: CollectingDispatcher,
            tracker: Tracker,
            domain: Dict[Text, Any],
    ) -> List[Dict[Text, Any]]:
        # tell the user they are being passed to a customer service agent


        # assume there's a function to call customer service
        # pass the tracker so that the agent has a record of the conversation between the user
        # and the bot for context
        latest_message = tracker.current_state()['latest_message']
        intent_ranking = latest_message.get("intent_ranking")
        buttons = []
        if not intent_ranking:
            return None
        for i in range(5):
            if intent_ranking[i]['name'] == "nlu_fallback" or intent_ranking[i]['confidence']==0:
                continue
            else:
                buttons.append({"payload":"/{0}".format(intent_ranking[i]['name']),"title":intent_ranking[i]['name']})
        dispatcher.utter_message(buttons=buttons)
        # pause the tracker so that the bot stops responding to user input
        return []

Try this, rewrite the “action_default_fallback” action, return top 5 intent let user to choice

This is my config.yml, FallbackClassifier is necessary

language: zh
pipeline:
  - name: SpacyNLP                 
    model: "zh_core_web_lg"
  - name: SpacyTokenizer        
#  - name: SpacyEntityExtractor  
  - name: SpacyFeaturizer         
    pooling: mean
  - name: DIETClassifier              
    epochs: 1000
    learning_rate: 0.001
  - name: EntitySynonymMapper
  - name: FallbackClassifier
    threshold: 0.4
    ambiguity_threshold: 0.1
1 Like
5

@diorw thx.
i will try your code then tell you.

6

My rasa version is 3.0.2