Hallucination Detection & Interpretability

We all know that Large Language Models (LLMs) can confidently emit falsehoods, a phenomenon known as hallucination. Joshua Carpeggiani will tell us about some interpretability methods - peering into the insides of the model and making sense of what we see - that might help detect and correct hallucinations.

We welcome a variety of backgrounds, opinions and experience levels.

Getting here:

Enter the lobby at 100 University Ave (right next to St Andrew subway station), and message Annie Sorkin on the meetup app or call her on 437-577-2379 to be let up to the sixth floor.
As a backup contact please reach out to Mario Gibney at 416-786-5403.

Precise Schedule & Locations:

From 6:00pm to 6:45 we will have pizza and refreshments in the kitchenette area on the 6th floor. This is a perfect time for networking and discussion of the latest in AI news.

The main presentation will begin in room 6H at 6:45, and run until approximately 8pm.

After this (starting around 8:30), some of us will congregate at Neo Coffee Bar (King/Spadina) for more casual chit-chat. Look for the table with the blue robot if you're just joining us at the cafe after the main presentation.