Ted Policy is prediction is not executed since the policy_2_RulePolicy predicted policy_2_RulePolicy. How do I resolve this situation, I am unable to write a story, because everytime rule policy predicts fall back
2021-09-19 19:49:35 DEBUG rasa.core.policies.memoization - Current tracker state [{}, {'user': {'intent': 'route_map', 'entities': ('hotel_name',)}, 'prev_action': {'action_name': 'action_listen'}}]
2021-09-19 19:49:35 DEBUG rasa.core.policies.memoization - There is no memorised next action
2021-09-19 19:49:35 DEBUG rasa.core.policies.ted_policy - TED predicted 'action_get_direction' based on user intent.
2021-09-19 19:49:35 DEBUG rasa.core.policies.rule_policy - Current tracker state:
[state 1] user text: find me the direction to Festive Hotel | previous action name: action_listen
2021-09-19 19:49:35 DEBUG rasa.core.policies.rule_policy - There is no applicable rule.
2021-09-19 19:49:35 DEBUG rasa.core.policies.rule_policy - Current tracker state:
[state 1] user intent: route_map | user entities: ('hotel_name',) | previous action name: action_listen
2021-09-19 19:49:35 DEBUG rasa.core.policies.rule_policy - There is no applicable rule.
2021-09-19 19:49:35 DEBUG rasa.core.policies.ensemble - Made prediction using user intent.
2021-09-19 19:49:35 DEBUG rasa.core.policies.ensemble - Added `DefinePrevUserUtteredFeaturization(False)` event.
2021-09-19 19:49:35 DEBUG rasa.core.policies.ensemble - Predicted next action using policy_2_RulePolicy.
2021-09-19 19:49:35 DEBUG rasa.core.processor - Predicted next action 'action_default_fallback' with confidence 0.40.
version: "2.0"
language: en_core_web_md
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: RegexEntityExtractor
- name: EntitySynonymMapper
- name: ResponseSelector
epochs: 100
retrieval_intent: faq
- name: ResponseSelector
epochs: 100
retrieval_intent: chitchat
- name: FallbackClassifier
threshold: 0.7
- name: DucklingEntityExtractor
url: http://duckling.rasa.com:8000
dimensions:
- amount-of-money
- time
- number
- phone-number
- duration
- name: SpacyNLP
model: "en_core_web_md"
case_sensitive: false
- name: "SpacyEntityExtractor"
# Note: It is not possible to use the SpacyTokenizer + SpacyFeaturizer in
# combination with the WhitespaceTokenizer, and as a result the
# PERSON extraction by Spacy is not very robust.
# Because of this, the nlu training data is annotated as well, and the
# DIETClassifier will also extract PERSON entities .
dimensions: ["PERSON"]
policies:
- name: AugmentedMemoizationPolicy
- name: TEDPolicy
epochs: 40
- name: RulePolicy
core_fallback_threshold: 0.4
core_fallback_action_name: "action_default_fallback"
enable_fallback_prediction: True
moved TEDPolicy and AugmentedMemoizationPolicy from above to beneath Rule policy and increased the TEDPolicy Epoch to 100. Everything is working fine now, If someone could post a logical explanation, it would have been better.
@athulvingt Right! You are not using? Just checking.
- name: TEDPolicy
max_history: 5
@athulvingt Its only a matter of priority some time Policies and more the Epochs better will be the result and even some times examples also play a vital role. If you change the value of epochs again, it can again throw you an error. Do check that also
max_history : This parameter controls how much dialogue history the model looks at to decide which action to take next . Default max_history for this policy is None , which means that the complete dialogue history since session restart is taken into account.
If you want to limit the model to only see a certain number of previous dialogue turns, you can set max_history to a finite value. Please note that you should pick max_history carefully, so that the model has enough previous dialogue turns to create a correct prediction.
@athulvingt I’d encourage to use under RulePolicy Policies but as I said it depends much on priority and also recommended under that policy by rasa team.
Good Luck!