Predicted action is not desirable while intent is classified correctly

When I start the conversation with intent “intent_A”, the coming action should be a form action “form_1” according to my stories. But the bot always return “utter_ask_if_check_detail”. What is the possible reason causing it? Is there any cache stored in location storage affecting the bot?

Story “inform 1” and “inform 2” are the only 2 stories contain “utter_ask_if_check_detail” in my data. Below are my stories file.

## inform 1
* inform{"entity": "placeholder"}
  - utter_ask_if_check_detail
* affirm
  - slot{"entity": "placeholder"}
  - action_reply
  - utter_check_what_detail
> check_ask_anything_else

## inform 2
* inform{"entity": "placeholder"}
  - utter_ask_if_check_detail
* deny
  - utter_help
> check_ask_anything_else

## main story 1 
* greet
  - utter_greet
  - utter_help
* intent_A
  - form_1
  - form{"name": "form_1"}
  - utter_anything_else
> check_ask_anything_else

## main story 2
* intent_A
  - form_1
  - form{"name": "form_1"}
  - utter_anything_else
> check_ask_anything_else

Does anyone have any idea?