Jesse Hoogland

Structural Inference: Interpreting Small Language Models with Susceptibilities

Bayesian Influence Functions for Hessian-Free Data Attribution

Influence Dynamics and Stagewise Data Attribution

Compressibility measures Complexity: Minimum Description Length meets Singular Learning Theory

The Loss Kernel: A Geometric Probe for Deep Learning Interpretability

Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficient

Stagewise Development in Transformers and the Geometry of the Loss Landscape