Failed to use AugmentedMemoizationPolicy to predict actions correctly

Rasa Open Source
1

How to design one story to make the following tracker state predict next action correctly?

2020-07-30 15:30:13 DEBUG    rasa.core.policies.memoization  - Launch DeLorean...
2020-07-30 15:30:13 DEBUG    rasa.core.policies.memoization  - Current tracker state [None, {}, {'intent_out_of_scope': 1.0, 'prev_action_utter_greet': 1.0}]
2020-07-30 15:30:27 DEBUG    rasa.core.policies.memoization  - There is no memorised next action

I used these stories but it failed. Anyone knows more?

## greet+out_of_scope
    - action_utter_greet
* out_of_scope
    - utter_default

## greet+out_of_scope
* greet
    - action_utter_greet
* out_of_scope
    - utter_default
2

Try a story with just the following:

* intent_out_of_scope
  - utter_default

The log shows an intent called intent_out_of_scope but the story shows intent named out_of_scope.

3

Hi, stephens

Thanks for the reply. However all intents have the prefix intent_ in the log. Seems it does not relate to this reason.

4

Turn debug mode on and send the full log along with your policies configuration.