The paper studies how Adam’s two momentum timescales, β1 and β3, influence loss spikes during neural‑network training. By mapping training dynamics across the (β1,β3) plane, the authors find an approximately linear boundary, 1-β3 = C(1-β1), that separates spiky from non‑spiky behavior, with the coefficient C linked to the effective loss exponent in superquadratic loss functions. They also show that confident cross‑entropy losses create a core–wall landscape that behaves superquadratically at the scale of an optimizer update, explaining the observed spikes.
By Gaoxiang Tang, Huanran Chen, Ziming Liu
arXiv:2606. 15551v1 Announce Type: new Abstract: The Edge of Stability (EoS) phenomenon, where gradient descent operates with sharpness exceeding the classical convergence threshold yet the loss decreases over long timescales, is ubiquitous in modern deep learning but remains poorly understood in realistic settings.
By Eric Gan
arXiv:2607. 03519v1 Announce Type: new Abstract: We show that projected Adam for online optimization with arbitrary moment decay parameters $\beta_1,\beta_2\in[0,1)$ can have average regret bounded away from zero.
By Steven Heilman, Sampad Mohanty
The paper investigates why setting the two momentum parameters of Adam equal (β1=β2) has a special dynamic effect. By analysing Adam in continuous time, the authors show that the update decomposes into a sign component, a magnitude‑lag term proportional to the difference between the two memory times, and other terms. This lag term disappears exactly when β1=β2, making the diagonal the only regime where the mismatch‑induced response is structurally absent. Experiments on six vision and language tasks confirm that tied configurations are sign‑dominated, have smaller lag contributions, and exhibit smoother update‑norm trajectories.
By Alberto Fern\'andez-Hern\'andez, Cristian P\'erez-Corral, Jose I. Mestre, Manuel F. Dolz, Enrique S. Quintana-Ort\'i
arXiv:2605. 29547v2 Announce Type: replace-cross Abstract: Deep learning optimization relies heavily on the assumption of smooth loss landscapes, a condition systematically violated by modern architectures due to non-smooth components such as ReLU activations and quantization operators.
By Ruoran Xu, Borong She, Xiaobo Jin, Qiufeng Wang
arXiv:2604. 08742v2 Announce Type: replace-cross Abstract: Adam is widely used, but its convergence theory remains incomplete even in the deterministic full-batch setting because momentum and adaptive preconditioning are tightly coupled.
By Yaxin Yu, Long Chen, Zeyi Xu
arXiv:2606. 18080v1 Announce Type: new Abstract: Gradient descent in deep learning may operate at the edge of stability (EoS), a regime in which the largest eigenvalue of the loss Hessian hovers near the stability threshold $2/\eta$, where $\eta$ is the learning rate.
By Pierre Marion
arXiv:2607. 03998v1 Announce Type: new Abstract: The local sharpness of the loss, the top Hessian eigenvalue $\lambda_1$, determines the largest stable gradient step, but measuring it normally requires Lanczos or Hessian-vector iterations.
By Ashmitha R, J\"org Frochte
arXiv:2606. 04212v1 Announce Type: new Abstract: Existing analyses of the edge of stability (EoS) treat it as a global property of optimization.
By Shauna Kwag, Anakha Ganesh, Tomaso Poggio, Pierfrancesco Beneventano
arXiv:2505. 13196v3 Announce Type: replace-cross Abstract: We introduce Velocity-Regularized Adam (VRAdam), a physics-inspired optimizer for training deep neural networks that draws on ideas from quartic terms for kinetic energy with its stabilizing effects on various system dynamics.
By Pranav Vaidhyanathan, Lucas Schorling, Natalia Ares, Maike Osborne
The paper investigates the "edge of stability" phenomenon in deep learning, where Hessian eigenvalues remain stable above a classically predicted unstable threshold. It shows that many first‑order optimizers, including gradient descent, can violate this stability bound by up to a factor of 21.1, and that this deviation depends systematically on the optimizer used. The authors propose a new stability threshold based on the directional Hessian and gradient‑alignment score, which removes optimizer‑dependent offsets and offers consistent predictions while providing diagnostic tools to understand how optimizers balance temporal and spatial budgets.
By Jaerin Lee, Kyoung Mu Lee
arXiv:2608. 14803v1 Announce Type: new Abstract: A recent line of work recasts the post-memorization phase of grokking as constrained optimization: once a network interpolates the training set, weight decay drives a slow drift along the zero-loss manifold toward lower norm.
By Suvinava Basak