Forms not_intent() and TwoStageFallback

Hi, this is my configuration for TwoStageFallbackPolicy:

- name: TwoStageFallbackPolicy
  nlu_threshold: 0.3
  ambiguity_threshold: 0.1
  core_threshold: 0.3
  fallback_core_action_name: "action_default_ask_affirmation"
  fallback_nlu_action_name: "action_default_ask_affirmation"
  deny_suggestion_intent_name: "out_of_scope"

and this is part of my form implmentation:

class RegisterForm(FormAction):
    def name(self):
        return "register_form"

    def required_slots(tracker: Tracker) -> List[Text]:
        return ["first_name", "last_name", "email", "birth_date"]

    def slot_mappings(self) -> Dict[Text, Union[Dict, List[Dict]]]:
        return {
            "first_name": [self.from_text(not_intent="stop")],
            "last_name": [self.from_text(not_intent="stop")],
            "email": [self.from_text(not_intent="stop")],
            "birth_date": [self.from_text(not_intent="stop")]

I have created intent stop so user can exit register process. Everything works fine until some information passed by user in this form gets classified as stop intent with less than 0.3 confidence. I was testing my chatbot and after I passed name “Michał”, chatbot instead of asking me for next slot, started TwoStageFallback.

I checked NLU confidence for “Michał” message and top intent is indeed a stop intent, BUT with only 0.11 confidence! I have checked it multiple times for different messages and everytime I’m in a form and passed message is classified with stop intent with less than 0.3 confidence, TwoStageFallback is executed. It’s only happening when stop intent is #1 in intent ranking.

I know I can add more examples to stop intent, but with every user message some intent must be classified as #1 intent for this message, and sadly some names or maybe even emails might be classified as stop with less than 0.3 confidence. So, it’s not solving my problem.

Is there a way to “turn off” TwoStageFallback when in self.from_text()? (Or solve my problem in some other way)