A Spectral Theory of Grokking: Weight Decay induces Feature Learning
cs.LG, cs.AI
Submitted: 2026-09-22
Updated: 2026-09-22
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Survival of the Fittest Representation: A Case Study with Modular Addition
- Verbalizable Representations Form a Global Workspace in Language Models
- On the Mechanism and Dynamics of Modular Addition: Fourier Features, Lottery Ticket, and Grokking
- Grokking on the Weight-Decay Clock: A Rate Hierarchy from Softly Broken Symmetries
- Feature Identification via the Empirical NTK
- Grokking as Compression: A Nonlinear Complexity Perspective
- Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets
- To Grok Grokking: Provable Grokking in Ridge Regression
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks