Flatness is Necessary, Neural Collapse is Not: Rethinking Generalization via Grokking

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Han, Ting, Adilova, Linara, Petzka, Henning, Kleesiek, Jens, Kamp, Michael
Format:	Preprint
Published:	2025
Subjects:	Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

When Flatness Does (Not) Guarantee Adversarial Robustness
by: Walter, Nils Philipp, et al.
Published: (2025)

The Uncanny Valley: Exploring Adversarial Robustness from a Flatness Perspective
by: Walter, Nils Philipp, et al.
Published: (2024)

Layer-wise Linear Mode Connectivity
by: Adilova, Linara, et al.
Published: (2023)

Landscaping Linear Mode Connectivity
by: Singh, Sidak Pal, et al.
Published: (2024)

Beyond Parameter Aggregation: Semantic Consensus for Federated Fine-Tuning of LLMs
by: Abourayya, Amr, et al.
Published: (2026)

On the regularization of Wasserstein GANs
by: Petzka, Henning, et al.
Published: (2017)

Whom to Trust? Adaptive Collaboration in Personalized Federated Learning
by: Abourayya, Amr, et al.
Published: (2025)

Explaining Grokking and Information Bottleneck through Neural Collapse Emergence
by: Sakamoto, Keitaro, et al.
Published: (2025)

Late-Stage Generalization Collapse in Grokking: Detecting anti-grokking with Weightwatcher
by: Prakash, Hari K, et al.
Published: (2026)

Grokking and Generalization Collapse: Insights from \texttt{HTSR} theory
by: Prakash, Hari K., et al.
Published: (2025)

Fisher information flow in artificial neural networks
by: Weimar, Maximilian, et al.
Published: (2025)

Little is Enough: Boosting Privacy by Sharing Only Hard Labels in Federated Semi-Supervised Learning
by: Abourayya, Amr, et al.
Published: (2023)

Deep Grokking: Would Deep Neural Networks Generalize Better?
by: Fan, Simin, et al.
Published: (2024)

Rethinking Continual Learning with Progressive Neural Collapse
by: Wang, Zheng, et al.
Published: (2025)

To Grok Grokking: Provable Grokking in Ridge Regression
by: Xu, Mingyue, et al.
Published: (2026)

NeuralGrok: Accelerate Grokking by Neural Gradient Transformation
by: Zhou, Xinyu, et al.
Published: (2025)

Grokking Beyond Neural Networks: An Empirical Exploration with Model Complexity
by: Miller, Jack, et al.
Published: (2023)

Rethinking PGD Attack: Is Sign Function Necessary?
by: Yang, Junjie, et al.
Published: (2023)

Position: Solve Layerwise Linear Models First to Understand Neural Dynamical Phenomena (Neural Collapse, Emergence, Lazy/Rich Regime, and Grokking)
by: Nam, Yoonsoo, et al.
Published: (2025)

The Complexity Dynamics of Grokking
by: DeMoss, Branton, et al.
Published: (2024)

Measuring Sharpness in Grokking
by: Miller, Jack, et al.
Published: (2024)

Bridging Lottery Ticket and Grokking: Understanding Grokking from Inner Structure of Networks
by: Minegishi, Gouki, et al.
Published: (2023)

Less Finetuning, Better Retrieval: Rethinking LLM Adaptation for Biomedical Retrievers via Synthetic Data and Model Merging
by: Khattab, Sameh, et al.
Published: (2026)

Grokked Models are Better Unlearners
by: Liang, Yuanbang, et al.
Published: (2025)

Grokking in LLM Pretraining? Monitor Memorization-to-Generalization without Test
by: Li, Ziyue, et al.
Published: (2025)

On the Robustness of Neural Collapse and the Neural Collapse of Robustness
by: Su, Jingtong, et al.
Published: (2023)

Topological Signatures of Grokking
by: Tang, Yifan, et al.
Published: (2026)

Exploring Grokking: Experimental and Mechanistic Investigations
by: Qiye, Hu, et al.
Published: (2024)

ILDR: Geometric Early Detection of Grokking
by: Golwala, Shreel
Published: (2026)

A Basin-Selection Perspective on Grokking via Singular Learning Theory
by: Cullen, Ben, et al.
Published: (2026)

GrokAlign: Geometric Characterisation and Acceleration of Grokking
by: Walker, Thomas, et al.
Published: (2025)

Distributional Spectral Diagnostics for Localizing Grokking Transitions
by: Wang, Ziyue, et al.
Published: (2026)

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

Preventing Model Collapse via Contraction-Conditioned Neural Filters
by: Han, Zongjian, et al.
Published: (2025)

A Pre-Training Analogue of Grokking in Language Models: Tracing Delayed Grammatical Generalization
by: Muckatira, Sherin, et al.
Published: (2026)

Rethinking Loss Reweighting for Imbalance Learning as an Inverse Problem: A Neural Collapse Point of View
by: Wang, Jinping, et al.
Published: (2026)

Grokking Explained: A Statistical Phenomenon
by: Carvalho, Breno W., et al.
Published: (2025)

Grokking in Linear Models for Logistic Regression
by: Das, Nataraj, et al.
Published: (2026)

Controlling Grokking with Nonlinearity and Data Symmetry
by: Salah, Ahmed, et al.
Published: (2024)

Beyond Unconstrained Features: Neural Collapse for Shallow Neural Networks with General Data
by: Hong, Wanli, et al.
Published: (2024)