Hello,
I have a handoff rule that is supposed to hand the issue to a live agent if the user tries to handoff in the middle of a form:
- rule: bot handoffs form_connect_general
condition:
- active_loop: form_connect_general
steps:
- intent: handoff
- action: action_deactivate_loop
- active_loop: null
- slot_was_set:
- requested_slot: null
- action: utter_handoff
- action: action_handoff
However, after testing it with rasa interactive, I get some weird behavior:
Chat History
# Bot You
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
1 action_listen
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
2 [cant](deny) [connect](connect)
intent: issue.connect 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
3 slot{"connect_error_type": "utc"}
form_connect_general 0.87
active_loop{"name": "form_connect_general"}
slot{"requested_slot": "any_device"}
To help in solving your connection problem I need to know what operating system you're using. Choose one from the list below or type your specific OS in the
message field!
Buttons:
1: Windows (windows)
2: macOS (macos)
3: Linux (linux)
4: Android (android)
5: iOS (ios)
Type out your own message...
action_listen 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
4 [live](human) [agent](human)
intent: handoff 1.00
─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
5 slot{"handoff_type": "h_human"}
And policy logs:
2022-12-19 13:33:21 DEBUG rasa.core.processor - Logged UserUtterance - tracker now has 26 events.
2022-12-19 13:33:21 DEBUG rasa.core.actions.action - Validating extracted slots:
2022-12-19 13:33:21 DEBUG rasa.core.actions.action - Calling action endpoint to run action 'action_validate_slot_mappings'.
2022-12-19 13:33:21 DEBUG rasa.core.processor - Default action 'action_extract_slots' was executed, resulting in 1 events: SlotSet(key: handoff_type, value: h_human)
2022-12-19 13:33:21 DEBUG rasa.core.lock_store - Deleted lock for conversation 'fdda6f19814b4127bd16810f833bfac0'.
2022-12-19 13:33:21 DEBUG rasa.core.tracker_store - Recreating tracker for id 'fdda6f19814b4127bd16810f833bfac0'
? Is the intent 'handoff' correct for '[live](human) [agent](human)' and are all entities labeled correctly? Yes
2022-12-19 13:33:25 DEBUG rasa.core.tracker_store - Recreating tracker for id 'fdda6f19814b4127bd16810f833bfac0'
2022-12-19 13:33:25 DEBUG rasa.core.tracker_store - Recreating tracker for id 'fdda6f19814b4127bd16810f833bfac0'
2022-12-19 13:33:25 DEBUG rasa.engine.runner.dask - Running graph with inputs: {'__tracker__': <rasa.shared.core.trackers.DialogueStateTracker object at 0x2883b8d00>}, targets: ['select_prediction'] and ExecutionContext(model_id='065e07a611aa48f58a1ae31a28410e59', should_add_diagnostic_data=False, is_finetuning=False, node_name=None).
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'rule_only_data_provider' running 'RuleOnlyDataProvider.provide'.
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'domain_provider' running 'DomainProvider.provide_inference'.
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'run_MemoizationPolicy0' running 'MemoizationPolicy.predict_action_probabilities'.
2022-12-19 13:33:25 DEBUG rasa.core.policies.memoization - Current tracker state:
[state 0] slots: {'is_restricted': (1.0, 0.0)}
[state 1] user intent: issue.connect | user entities: ('connect', 'deny') | previous action name: action_listen | slots: {'is_restricted': (1.0, 0.0)}
[state 2] user intent: issue.connect | user entities: ('connect', 'deny') | previous action name: form_connect_general | slots: {'is_restricted': (1.0, 0.0)}
[state 3] user intent: handoff | user entities: ('human',) | previous action name: action_listen | slots: {'is_restricted': (1.0, 0.0)}
2022-12-19 13:33:25 DEBUG rasa.core.policies.memoization - There is no memorised next action
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'run_RulePolicy1' running 'RulePolicy.predict_action_probabilities'.
2022-12-19 13:33:25 DEBUG rasa.core.policies.rule_policy - Current tracker state:
[state 0] slots: {'is_restricted': (1.0, 0.0)}
[state 1] user intent: issue.connect | user entities: ('connect', 'deny') | previous action name: action_listen | slots: {'is_restricted': (1.0, 0.0)}
[state 2] user intent: issue.connect | user entities: ('connect', 'deny') | previous action name: form_connect_general | active loop: {'name': 'form_connect_general'} | slots: {'is_restricted': (1.0, 0.0)}
[state 3] user text: live agent | previous action name: action_listen | active loop: {'name': 'form_connect_general'} | slots: {'is_restricted': (1.0, 0.0)}
2022-12-19 13:33:25 DEBUG rasa.core.policies.rule_policy - There is no applicable rule.
2022-12-19 13:33:25 DEBUG rasa.core.policies.rule_policy - Predicted loop 'form_connect_general'.
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'run_TEDPolicy2' running 'TEDPolicy.predict_action_probabilities'.
2022-12-19 13:33:25 DEBUG rasa.core.policies.ted_policy - TED predicted 'action_deactivate_loop' based on user intent.
2022-12-19 13:33:25 DEBUG rasa.engine.graph - Node 'select_prediction' running 'DefaultPolicyPredictionEnsemble.combine_predictions_from_kwargs'.
2022-12-19 13:33:25 DEBUG rasa.core.policies.ensemble - Made prediction using user intent.
2022-12-19 13:33:25 DEBUG rasa.core.policies.ensemble - Added `DefinePrevUserUtteredFeaturization(False)` event.
2022-12-19 13:33:25 DEBUG rasa.core.policies.ensemble - Predicted next action using RulePolicy.
2022-12-19 13:33:25 DEBUG rasa.core.tracker_store - Recreating tracker for id 'fdda6f19814b4127bd16810f833bfac0'
What i don’t understand is why it calls the form again instead of action_deactivate_loop and why the rule policy says there is no rule, but still predicts?