Introduces Label-NTK and Residual-NTK alignments to derive tighter NTK convergence bounds that track the full eigen-spectrum and match observed training speed.
Gradient descent on neural networks typically occurs at the edge of stability
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
Slingshot loss spikes are produced by low-precision arithmetic that breaks the zero-sum gradient constraint and drives exponential growth via Numerical Feature Inflation.
citing papers explorer
-
Label-NTK Alignments and A Tighter Convergence Bound in the NTK Regime
Introduces Label-NTK and Residual-NTK alignments to derive tighter NTK convergence bounds that track the full eigen-spectrum and match observed training speed.
-
Grokking or Glitching? How Low-Precision Drives Slingshot Loss Spikes
Slingshot loss spikes are produced by low-precision arithmetic that breaks the zero-sum gradient constraint and drives exponential growth via Numerical Feature Inflation.