Avery Griffin

neural networks

loss landscape degeneracy

effective parameter count interpretability

mechanistic interpretability

reverse engineering neural networks

sparse representations

dimensionality reduction

low-rank approximations interaction graphs

integrated gradients

modularity analysis