"A Tale of Two Circuits: Grokking as Competition of Sparse and Dense ..."

William Merrill, Nikolaos Tsilivis, Aman Shukla (2023)

Details and statistics

DOI: 10.48550/ARXIV.2303.11873

access: open

type: Informal or Other Publication

metadata version: 2023-03-22

a service of  Schloss Dagstuhl - Leibniz Center for Informatics