arXiv:2608. 19762v1 Announce Type: cross Abstract: A minibatch can influence training beyond the update at which it is observed because AdamW stores past gradient information in its optimizer states.
arXiv:2608.24593v1 Announce Type: new Abstract: Adaptive optimizers retain gradient history in moment variables, allowing a local change in loss weighting to alter later updates. We examine whether t...
arXiv:2608. 01997v1 Announce Type: new Abstract: Single-optimizer training is a poor fit for the distinct phases of deep network optimization: adaptive methods handle noisy early gradients well but overshoot flat minima, while SGD with momentum generalizes better in the late phase but converges slowly early on.
arXiv:2606. 04476v1 Announce Type: new Abstract: In this paper, we study the gradient descent dynamics for jointly training both layers of a one-hidden-layer ReLU network to fit a linear target function.
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.
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.
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.
arXiv:2604. 22407v2 Announce Type: replace Abstract: Many continual-learning methods modify gradients upstream (e.
arXiv:2609.36738v1 Announce Type: new Abstract: Optimizer momentum is usually stored as a parameter-sized moving average of past gradients, which makes history costly and fixes each past signal in th...
arXiv:2610.16268v2 Announce Type: replace Abstract: Test-time-training (TTT) layers update fast-weight memory while processing an input sequence. A gradient-based update, however, does not by itself...
arXiv:2301. 06308v2 Announce Type: replace-cross Abstract: Sharpness-aware minimization (SAM) is a training method that seeks to find flat minima in deep learning, resulting in state-of-the-art performance across various domains.
The paper introduces Activation-Keyed Momentum (AK‑Momentum), a momentum update that uses the input activation of a linear layer as a key to apply a delta‑rule update, allowing each direction to decay at a rate proportional to its frequency of appearance. AK‑Momentum is proven to be a valid momentum, incorporates input‑side curvature correction without matrix inversion, and clears stale directions faster than traditional exponential moving average (EMA) under both fixed and drifting optima. It can replace the momentum buffer of any optimizer, scales with width under μP, adds only 22–25% extra compute, and demonstrates significant step‑count reductions in FineWeb‑Edu pretraining and other benchmarks. whyItMatters":"AK‑Momentum offers a principled, efficient way to adapt momentum decay to anisotropic training dynamics, improving convergence speed and stability across a range of models and datasets."
AYLA is a loss reparameterization framework that applies a sigmoid‑controlled power‑law transformation to the empirical loss, dynamically adjusting gradient magnitudes without changing stationary points or optimal solutions. By reshaping optimization trajectories, AYLA accelerates descent in flat or saddle‑dominated regions and stabilizes late‑stage training, leading to improved feature recovery in two‑layer tanh networks on synthetic Gaussian data. Experiments show enhanced weight alignment, neuron similarity, activation correlation, and richer internal representations, while mitigating rank collapse and promoting a transition from lazy to active feature‑learning regimes.