Rasa.core.policies.ted_policy predicting wrong action

Getting Started with Rasa
1

I’m trying rasa with example on Generating NLU Data (rasa.com) and I’ve written appropriate nlu data and domain yml to go along with stories. So, intents and entities (there is only one for now: product) are correctly classified and identified, however, wrong action is being predicted by TEDPolicy.

Even though it says rasa.core.policies.ted_policy - TED predicted 'utter_migration_ibm' based on user intent. and intents are correctly classified and entities are also correctly classified and their value is correctly identified then why does Tedpolicy is making wrong prediction?

image
image1597×545 52.1 KB

in config.yml I’ve only mentioned names of policies

policies:
 - name: MemoizationPolicy
 - name: RulePolicy
 - name: TEDPolicy
 - name: EnsemblePolicy

However I was facing same problem when no policy was explicitly mentioned or uncommented. Please help. I also have “utter_ask_migration_product” and if I don’t mention IBM or dialogflow then that story gets correctly responded to. This means the second one “utter_migration_dialogflow” gets ignored and replaced with utter_migration_ibm whenever entity is identified.