Hi
My model currently has many faq intents, eg.faq/about_company
, faq/about_services
etc. along with some other intents like query_status
etc.
The problem is I have many faq intents and I notice that the intent classifier will just predict a generic faq intent for all the faq intents.
Question:
- Does this mean all the examples for the different faq intents are lumped together, causing the faq intent to have way more examples than all the other intents?
- I understand that class imbalance is inherently a bad thing as it causes bias in the model’s prediction of intent. So will the scenario above cause an issue with the model prediction?
Thank you in advance.