Search Results for author: Jacob Dunefsky

Found 1 papers, 1 papers with code

Observable Propagation: A Data-Efficient Approach to Uncover Feature Vectors in Transformers

1 code implementation26 Dec 2023 Jacob Dunefsky, Arman Cohan

Our results suggest that ObsProp surpasses traditional approaches for finding feature vectors in the low-data regime, and that ObsProp can be used to better understand the mechanisms responsible for bias in large language models.

Cannot find the paper you are looking for? You can Submit a new open access paper.