DOE OSTI · 2584995
Low-Variance Hessian Estimation: Coordinating High-Dimensional Sparsification with Projective Integral Updates
Abstract
We can improve network pruning by leveraging the loss-topography extraction techniques used by projective integral updates for variational inference. Low-variance Hessians facilitate more aggressive pruning by providing better loss approximations when a parameter is removed.
Keep this discovery
Explore connections, maps & timelines
Duersch, Jed Alma. 2024-02-01. Low-Variance Hessian Estimation: Coordinating High-Dimensional Sparsification with Projective Integral Updates. https://doi.org/10.2172/2584995
Cite the original work for its findings. Save a collection to share your selection of sources.