Image de couverture de livre

The Geometry of Grokking: Norm Minimization on the Zero-Loss Manifold

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteur principal:	Musat, Tiberiu
Format:	Preprint
Publié:	2025
Sujets:	Machine Learning Artificial Intelligence
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Neural Weight Norm = Kolmogorov Complexity
par: Musat, Tiberiu
Publié: (2026)

Early-Warning Signals of Grokking via Loss-Landscape Geometry
par: Xu, Yongzhong
Publié: (2026)

Grokking Beyond the Euclidean Norm of Model Parameters
par: Notsawo, Pascal Jr Tikeng, et autres
Publié: (2025)

Clustering and Alignment: Understanding the Training Dynamics in Modular Addition
par: Musat, Tiberiu
Publié: (2024)

The Norm-Separation Delay Law of Grokking: A First-Principles Theory of Delayed Generalization
par: Khanh, Truong Xuan, et autres
Publié: (2026)

Mechanism and Emergence of Stacked Attention Heads in Multi-Layer Transformers
par: Musat, Tiberiu
Publié: (2024)

The Geometry of Multi-Task Grokking: Transverse Instability, Superposition, and Weight Decay Phase Structure
par: Xu, Yongzhong
Publié: (2026)

Topological Signatures of Grokking
par: Tang, Yifan, et autres
Publié: (2026)

VL Norm: Rethink Loss Aggregation in RLVR
par: He, Zhiyuan, et autres
Publié: (2025)

Grokking Explained: A Statistical Phenomenon
par: Carvalho, Breno W., et autres
Publié: (2025)

Grokking in Linear Models for Logistic Regression
par: Das, Nataraj, et autres
Publié: (2026)

Controlling Grokking with Nonlinearity and Data Symmetry
par: Salah, Ahmed, et autres
Publié: (2024)

Grokfast: Accelerated Grokking by Amplifying Slow Gradients
par: Lee, Jaerin, et autres
Publié: (2024)

Understanding Grokking Through A Robustness Viewpoint
par: Tan, Zhiquan, et autres
Publié: (2023)

Progress Measures for Grokking on Real-world Tasks
par: Golechha, Satvik
Publié: (2024)

Tracing the Path to Grokking: Embeddings, Dropout, and Network Activation
par: Salah, Ahmed, et autres
Publié: (2025)

NeuralGrok: Accelerate Grokking by Neural Gradient Transformation
par: Zhou, Xinyu, et autres
Publié: (2025)

Low-Dimensional and Transversely Curved Optimization Dynamics in Grokking
par: Xu, Yongzhong
Publié: (2026)

Muon Optimizer Accelerates Grokking
par: Tveit, Amund, et autres
Publié: (2025)

Grokking Finite-Dimensional Algebra
par: Notsawo, Pascal Jr Tikeng, et autres
Publié: (2026)

Grokking Group Multiplication with Cosets
par: Stander, Dashiell, et autres
Publié: (2023)

Unpacking the Implicit Norm Dynamics of Sharpness-Aware Minimization in Tensorized Models
par: Cao, Tianxiao, et autres
Publié: (2025)

Grokking and Generalization Collapse: Insights from \texttt{HTSR} theory
par: Prakash, Hari K., et autres
Publié: (2025)

Grokking as Structural Inference: Transformers Need Bayesian Lottery Tickets
par: Hidajat, Kai, et autres
Publié: (2026)

Global Minimizers of Sigmoid Contrastive Loss
par: Bangachev, Kiril, et autres
Publié: (2025)

Discovering Data Manifold Geometry via Non-Contracting Flows
par: Vigouroux, David, et autres
Publié: (2026)

Provable Scaling Laws of Feature Emergence from Learning Dynamics of Grokking
par: Tian, Yuandong
Publié: (2025)

Acceleration of Grokking in Learning Arithmetic Operations via Kolmogorov-Arnold Representation
par: Park, Yeachan, et autres
Publié: (2024)

Navigating Potholes with Geometry-Aware Sharpness Minimization
par: Dufort-Labbé, Simon, et autres
Publié: (2026)

Grokking as a Falsifiable Finite-Size Transition
par: Bi, Yuda, et autres
Publié: (2026)

Dichotomy of Early and Late Phase Implicit Biases Can Provably Induce Grokking
par: Lyu, Kaifeng, et autres
Publié: (2023)

Two Speeds of Learning: A Representation-Readout Decomposition of Grokking and Double Descent
par: Chou, Chi-Ning, et autres
Publié: (2026)

Towards Empirical Interpretation of Internal Circuits and Properties in Grokked Transformers on Modular Polynomials
par: Furuta, Hiroki, et autres
Publié: (2024)

Feature Repulsion and Spectral Lock-in: An Empirical Study of Two-Layer Network Grokking
par: Xu, Yongzhong
Publié: (2026)

Model Merging on Loss Landscape: A Geometry Perspective
par: Lu, Juanwu, et autres
Publié: (2026)

Grokking as a Variance-Limited Phase Transition: Spectral Gating and the Epsilon-Stability Threshold
par: Acharya, Pratyush, et autres
Publié: (2026)

Grokking at the Edge of Numerical Stability
par: Prieto, Lucas, et autres
Publié: (2025)

Uncovering the Representation Geometry of Minimal Cores in Overcomplete Reasoning Traces
par: Chowdhury, Sanjoy, et autres
Publié: (2026)

First-Passage Prediction of Grokking Delay: ACalibrated Law under AdamW with Causal Validation
par: Khanh, Truong Xuan, et autres
Publié: (2026)

Minimizing Surrogate Losses for Decision-Focused Learning using Differentiable Optimization
par: Mandi, Jayanta, et autres
Publié: (2025)