Linear Latent World Models in Simple Transformers: A Case Study on Othello-GPT
Abstract
Foundation models exhibit significant capabilities in decision-making and logical deductions. Nonetheless, a continuing discourse persists regarding their genuine understanding of the world as opposed to mere stochastic mimicry. This paper meticulously examines a simple transformer trained for Othello, extending prior research to enhance comprehension of the emergent world model of Othello-GPT. The investigation reveals that Othello-GPT encapsulates a linear representation of opposing pieces, a factor that causally steers its decision-making process. This paper further elucidates the interplay between the linear world representation and causal decision-making, and their dependence on layer depth and model complexity. We have made the code public.
Cite
@article{arxiv.2310.07582,
title = {Linear Latent World Models in Simple Transformers: A Case Study on Othello-GPT},
author = {Dean S. Hazineh and Zechen Zhang and Jeffery Chiu},
journal= {arXiv preprint arXiv:2310.07582},
year = {2023}
}