hi, i’ve created my chatbot and everything was working fine, but now went i type a message it always repond from the default fallback action, and i don’t know how remove “rewind” from the bot can any one help?? (all stories and data are defined)
Hi,
There is some issue in RASA X it seems. I also keep on getting the same when a particular slot value is intentionally entered wrong, after the correct value entered is also taken into fallback policy.
One reason I found was to check the ambiguity confidence using rasa shell --debug or delete all previous model and rasa.db and try again!
@mishra-atul5001 i’ve tried rasa shell --debug but every message i type it respond with the action fallback
can you post the error log?
@mishra-atul5001 here is the log when i send “je veux avoir des informations” he answers from the action fallback even if i defined this sentence in nlu.md. i don’t know what is the problem !!
Your input -> je veux avoir des informations
2020-06-16 05:11:10 DEBUG rasa.core.tracker_store - Recreating tracker for id ‘4e971a06a1a146d09722933b2f6647d4’
2020-06-16 05:11:10 DEBUG rasa.nlu.classifiers.diet_classifier - There is no trained model for ‘ResponseSelector’: The component is either not trained or didn’t receive enough training data.
2020-06-16 05:11:10 DEBUG rasa.nlu.selectors.response_selector - Adding following selector key to message property: default
2020-06-16 05:11:10 DEBUG rasa.core.processor - Received user message ‘je veux avoir des informations’ with intent ‘{‘name’: ‘infos’, ‘confidence’: 0.9754295349121094}’ and entities ‘[]’
2020-06-16 05:11:10 DEBUG rasa.core.processor - Logged UserUtterance - tracker now has 42 events.
2020-06-16 05:11:10 DEBUG rasa.core.policies.memoization - Current tracker state [{‘intent_savoir_plus’: 1.0, ‘prev_action_listen’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation3’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation4’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation5’: 1.0}, {‘prev_action_listen’: 1.0, ‘intent_infos’: 1.0}]
2020-06-16 05:11:10 DEBUG rasa.core.policies.memoization - There is no memorised next action
2020-06-16 05:11:10 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.4).
2020-06-16 05:11:10 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_1_TEDPolicy
2020-06-16 05:11:10 DEBUG rasa.core.processor - Predicted next action ‘action_default_fallback’ with confidence 0.76.
2020-06-16 05:11:10 DEBUG rasa.core.processor - Action ‘action_default_fallback’ ended with events ‘[BotUttered(‘Excusez moi, je n’ai pas compris votre message. Veuillez le ressaisir s’il vous plait!’, {“elements”: null, “quick_replies”: null, “buttons”: null, “attachment”: null, “image”: null, “custom”: null}, {}, 1592309470.5808923), <rasa.core.events.UserUtteranceReverted object at 0x7fa6c1fd3f98>]’.
2020-06-16 05:11:10 DEBUG rasa.core.policies.memoization - Current tracker state [{‘intent_presentation1’: 1.0, ‘prev_utter_presentation2’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_action_listen’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation3’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation4’: 1.0}, {‘intent_savoir_plus’: 1.0, ‘prev_utter_presentation5’: 1.0}]
2020-06-16 05:11:10 DEBUG rasa.core.policies.memoization - There is a memorised next action ‘action_listen’
2020-06-16 05:11:10 DEBUG rasa.core.policies.mapping_policy - There is no mapped action for the predicted intent, ‘savoir_plus’.
2020-06-16 05:11:10 DEBUG rasa.core.policies.fallback - NLU confidence threshold met, confidence of fallback action set to core threshold (0.4).
2020-06-16 05:11:10 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_0_MemoizationPolicy
2020-06-16 05:11:10 DEBUG rasa.core.processor - Predicted next action ‘action_listen’ with confidence 1.00.
2020-06-16 05:11:10 DEBUG rasa.core.processor - Action ‘action_listen’ ended with events ‘[]’.
2020-06-16 05:11:10 DEBUG rasa.core.lock_store - Deleted lock for conversation ‘4e971a06a1a146d09722933b2f6647d4’.
Excusez moi, je n’ai pas compris votre message. Veuillez le ressaisir s’il vous plait!
Your input ->
Your FallBack is predicted with high confidence when it comes to core prediction. You can either increase the training data for the intents or Lower the nlu and core threshold in the FallBack Policy.
Could you please also attach the NLU data for FallBack/Chitchat and Greet data!
I do not understand anything anymore, when i click on a button it gives me also the same message!! can you help me please this is a part of the nlu.md:
intent:greet
- bonjour
- salut
- slt
- coucou
- bjr
- cc
- Cc
- psst
- saluut
- Bjr
- coucouu
- Bonjour
- salam
- hey
- hello
- hi
- Hi
- bonjour, est ce que vous pouvez m’aidez?
intent:infos
- infos
- informations
- Avoir des informations
- Je veux avoir des informations
- bonjour, je veux avoir des informations
- je veux avoir des informations
- je veux avoir des informations sur ESTS
- avoir des informations
intent:goodbye
- Au revoir
- au revoir
- bye
- bonne journée
- bonne journee
- goodbye
intent:thank
- merci
- merci infiniment
- je vous remercie
- ok merci
- merci pour les informations
- d’accord merci
- a vous aussi
- Merci
- d’accord, j’ai eu ce que je veux . merci
- merci beaucoup
this is a part of the domain.yml: intents:
- greet
- goodbye
- thank
- reclam
- infos
- LP
- DUT
- affirm
- presentation1
- presentation3
- presentation2
- savoir_plus
- formationIn
- formation
- Autres
- deny
- formationT
responses:
utter_greet:
- buttons:
- payload: Présentation de l’école title: Présentation de l’école
- payload: Faire une récalamation title: Faire une récalamation
- payload: Avoir des informations title: Avoir des informations text: Bonjour! Je suis un robot de ESTS, comment pourrais-je vous aidez?
utter_default:
- text: Excusez moi, je n’ai pas compris votre message. Veuillez le ressaisir s’il vous plait!
actions:
- utter_greet
- utter_goodbye
- utter_infos
- utter_default
- utter_thank
and this is the config.yml:
Configuration for Rasa NLU.
Components
language: fr pipeline:
- name: WhitespaceTokenizer
- name: RegexFeaturizer
- name: LexicalSyntacticFeaturizer
- name: CountVectorsFeaturizer
- name: CountVectorsFeaturizer analyzer: “char_wb” min_ngram: 1 max_ngram: 4
- name: DIETClassifier epochs: 100
- name: EntitySynonymMapper
- name: ResponseSelector epochs: 100
Configuration for Rasa Core.
Policies
policies:
- name: MemoizationPolicy
- name: TEDPolicy max_history: 5 epochs: 100
- name: MappingPolicy
- name: FallbackPolicy nlu_threshold: 0.4 core_threshold: 0.4 fallback_action_name: “action_default_fallback”
What about stories? Under which intent are you calling your fallback policy?
greet fallback
*greet
-action_default_fallback
any solution??
yes, you have denied the action fallback function in your intent greet only! That’s why.
Stories are most essential component of the RASA framework!
Try defining an intent called chitchat and then call the fall back in that!.
For greet call some utter_greet template.
Try something like this:
## new story
* greet
- action_greet_time
- utter_disclaimer
* mood_great
- utter_happy
- utter_disclaimer
* chitchat
- action_default_fallback
Give something like for response:
utter_disclaimer: - text: Wow!
For reference, you can refer this git repo: CoronaVirus-ChatBot
Hope this helps!
now i understand, i tried it and it worked. thanks for your help
Happy to help! Mark it as Solution so others can also be aware of this issue in future!
1 Like
Hi Mr. Mishra, I am trying to develop an Faq chatbot on covid when i gave some intents and responses my bot identifying the intents correctly but it selects the next possible action as “utter_fallback” with confidence 0.30 and rewind both in rasa x and rasa. For the same given input it provides correct response then later if i ask the same input after a set of other qns it gives me utter fallback. Out of all for every min of one intent’s response will be choosing feedback action it won’t be same intent every time.