Gary Marcus

Rank 19 of 47
|
Score 64
Nikhil Prakash
@nikhil07prakash
·
17d
How do language models track mental states of each character in a story, often referred to as Theory of Mind?

Our recent work takes a step in demystifing it by reverse engineering how Llama-3-70B-Instruct solves a simple belief tracking task, and surprisingly found that it
6/25/2025, 12:28:33 AM
X

The statement and conversation focus on the technical exploration of how language models, specifically Llama-3-70B-Instruct, track mental states in narratives, a concept known as Theory of Mind. This involves reverse engineering to understand belief tracking tasks. The discussion is academic and technical, contributing to the understanding of AI capabilities.

  1. Principle 1:
    I will strive to do no harm with my words and actions.
    The statement does no harm and is focused on academic exploration. [+1]
  2. Principle 3:
    I will use my words and actions to promote understanding, empathy, and compassion.
    Promotes understanding of AI and its capabilities, fostering knowledge sharing. [+2]
  3. Principle 4:
    I will engage in constructive criticism and dialogue with those in disagreement and will not engage in personal attacks or ad hominem arguments.
    Engages in constructive dialogue about AI research without personal attacks. [+1]