Unifying Central Flow, Rod Flow, and the Edge Coupling
stat.ML, cs.LG
Submitted: 2026-11-05
Updated: 2026-11-05
Code: https://github.com/alex-damian/EOS
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Sources
- Edge of Stochastic Stability: Revisiting the Edge of Stability for SGD
- Momentum Further Constrains Sharpness at the Edge of Stochastic Stability
- The Multiple Timescales of Gradient Descent on the Edge of Stability: A Perturbative Derivation of the Central Flow
- Better plain ViT baselines for ImageNet-1k
- Mean--Fluctuation Dynamics at the Edge of Stability
- Adaptive Gradient Methods at the Edge of Stability
- Adam at the Edge of Stability: Adaptive Feedback, Provable Oscillation, and Gradient Reversal
- The large learning rate phase of deep learning: the catapult mechanism
- The Origin of Edge of Stability
- Edge Flow: A Tractable and Predictive Continuous-Time Model for Gradient Descent at the Edge of Stability
- A Defense of the Quadratic Model
- Gradient Descent with Polyak's Momentum Finds Flatter Minima via Large Catapults
- A Rod Flow Model for Adam at the Edge of Stability
- Rod Flow: A Continuous-Time Model for Gradient Descent at the Edge of Stability
Related papers
- Uniform-in-Time Weak Propagation-of-Chaos in Shallow Neural Networks
- Diffusion Models Observe Only Gradients: A Geometric Perspective on Score Matching Errors
- Quantifying Error Propagation and Model Collapse in Diffusion Models
- Behavior of prediction performance metrics with rare events
- Statistical Inference and Learning for Shapley Additive Explanations (SHAP)
- Control Variate Design for Score-Function Gradient Estimators in Importance-Weighted Variational Inference