Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficient
17 OctThis paper introduces refined variants of the Local Learning Coefficient (LLC) to study the differentiation and specialization of attention heads in transformer language models during training. The findings reveal how attention heads evolve into distinct functional roles, analyze their specialization based on data types, and uncover a novel multigram circuit, contributing to a deeper understanding of model complexity and interpretability.
Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficient
17 OctThis paper introduces refined variants of the Local Learning Coefficient (LLC) to study the differentiation and specialization of attention heads in transformer language models during training. The findings reveal how attention heads evolve into distinct functional roles, analyze their specialization based on data types, and uncover a novel multigram circuit, contributing to a deeper understanding of model complexity and interpretability.