conftrace_

loss hessian

5 papers

Explore in graph

Co-occurring keywords · papers mentioning both

learning rate (2) loss surface geometry (1) cross-entropy loss gradient (1) goldilocks zone (1) initialization norm (1) learning rate tuning (1) sgd stability (1) linear stability (1) optimization dynamics (1) model confidence (1)

Papers

A Precise Characterization of SGD Stability Using Loss Surface Geometry ICLR 2024 Deconstructing the Goldilocks Zone of Neural Network Initialization ICML 2024 Stepping on the Edge: Curvature Aware Learning Rate Tuners NeurIPS 2024 SAM operates far from home: eigenvalue regularization as a dynamical phenomenon ICML 2023 A Loss Curvature Perspective on Training Instabilities of Deep Learning Models ICLR 2022