Hi,
I've been tinkering with this repo..
I noticed that whenever I submit any nonsenical, irrelevant input to the chat as a user, the bot pretty much always recognizes it at "greeting" with confidence of over 25%.
For example, as a user in the chat window, I submit "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
The response is "Good to see you again".
I would expect the model to find no valid intents and thus the filter should create an empty list. Instead, the filtered list "results" var always includes the "greeting" result.
What's causing this?
