I have only defined rules and provided them with sufficient training data. My problem now is that RulePolicy, although an intent has been recognised with, for example, 95%, still triggers the action_default_fallback every now and then.
For example, I have a rule “Deny”. If I now write “No” 3 times in a row in the chat, then I get the correct answer behind the intent “Deny” but the fourth time the action_default_fallback is suddenly triggered.
If I continue with typing “No” in the chat then at some point just an action_listen is triggered instead of action_default_fallback.
Is it possible to switch off this behaviour somehow?
recipe: default.v1
language: de
pipeline:
- name: "SpacyNLP"
model: "de_core_news_lg"
case_sensitive: False
- name: "SpacyTokenizer"
"intent_tokenization_flag": False
"intent_split_symbol": "_"
"token_pattern": None
- name: "SpacyFeaturizer"
- name: "RegexFeaturizer"
- name: LexicalSyntacticFeaturizer
- name: "CountVectorsFeaturizer"
- name: "CountVectorsFeaturizer"
"analyzer": "char_wb"
"min_ngram": 1
"max_ngram": 4
- name: DIETClassifier
epochs: 100
constrain_similarities: true
- name: EntitySynonymMapper
- name: FallbackClassifier
threshold: 0.6
ambiguity_threshold: 0.05
policies:
- name: RulePolicy