arxiv
PublishedMay 28, 2026 at 4:00 AM
—neutral
GradientStabilizer:Fix the Norm, Not the Gradient
Publisher summary· verbatim
arXiv:2502.17055v4 Announce Type: replace-cross Abstract: Training instability in modern deep learning systems is frequently triggered by rare but extreme gradient-norm spikes, which can induce oversized parameter updates, corrupt optimizer state, and lead to slow recovery or divergence. Widely used
Stay posted· Newsletter
A 5-min weekly brief — top movers, price watch, story of the week.
Discussion
No replies yet. Be first.
Related coverage
More from ARXIV
arxivDiscovering a Zeta Map Algorithm on Dyck Paths via Mechanistic Interpretability15marxivImproving Relative Representations with Learned Anchors and Whitened Inner Products15marxivScaleMAP: Preserving Local Density and Neighborhood Structure in Low-Dimensional Embeddings15marxivThe Fast Mixing Mechanism for Differential Privacy15mThe Bubble Brief
WEEKLYRead AI insights every Tuesday — top movers, new releases, story of the week.
Originally published on arxiv ↗