Is Grokking a Loss of Normal Hyperbolicity of the Interpolation Manifold?

arXiv:2608.14803v1 Announce Type: new Abstract: A recent line of work recasts the post-memorization phase of grokking as constrained optimization: once a network interpolates the training set, weight decay drives a slow drift along the zero-loss manifold toward lower norm. In the language of dynamical systems, this is a fast-slow system in which the interpolation manifold plays the role of a slow manifold. We ask a question that this framing makes natural but the existing literature does not address: is the sharp generalization transition a loss of normal hyperbolicity of that manifold: a fold- or bifurcation-like event in which a normal restoring direction goes flat? Or does the manifold stay uniformly attracting while generalization happens by smooth drift? We propose a simple, optimizer-agnostic diagnostic: the smallest nonzero singular value $\sigma_{\min}^{+}(\mathbf J)$ of the residual Jacobian, which, for the squared loss, equals the slowest normal restoring rate of the manifold. On a two-layer ReLU network trained to grok modular addition under squared loss, $\sigma_{\min}^{+}(\mathbf J)$ does not collapse at the transition; it is near zero only before memorization and attains its largest values during the transition. The result holds across five seeds, and the six smallest singular values behave identically; there is no subspace-local collapse either. This is preliminary evidence against the bifurcation hypothesis and in favor of the smooth-contraction picture. We are explicit that a single-setting, gradual-transition experiment under Adam optimizer does not prove the absence of a bifurcation; it constrains where one could hide.

Title:Is Grokking a Loss of Normal Hyperbolicity of the Interpolation Manifold?

Abstract:A recent line of work recasts the post-memorization phase of grokking as constrained optimization: once a network interpolates the training set, weight decay drives a slow drift along the zero-loss manifold toward lower norm. In the language of dynamical systems, this is a fast-slow system in which the interpolation manifold plays the role of a slow manifold. We ask a question that this framing makes natural but the existing literature does not address: is the sharp generalization transition a loss of normal hyperbolicity of that manifold: a fold- or bifurcation-like event in which a normal restoring direction goes flat? Or does the manifold stay uniformly attracting while generalization happens by smooth drift? We propose a simple, optimizer-agnostic diagnostic: the smallest nonzero singular value $\sigma_{\min}^{+}(\mathbf J)$ of the residual Jacobian, which, for the squared loss, equals the slowest normal restoring rate of the manifold. On a two-layer ReLU network trained to grok modular addition under squared loss, $\sigma_{\min}^{+}(\mathbf J)$ does not collapse at the transition; it is near zero only before memorization and attains its largest values during the transition. The result holds across five seeds, and the six smallest singular values behave identically; there is no subspace-local collapse either. This is preliminary evidence against the bifurcation hypothesis and in favor of the smooth-contraction picture. We are explicit that a single-setting, gradual-transition experiment under Adam optimizer does not prove the absence of a bifurcation; it constrains where one could hide.

Machine Learning (cs.LG)

arXiv:2608.14803 [cs.LG]

(or arXiv:2608.14803v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2608.14803 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

aiqbench analysis

Is Grokking a Loss of Normal Hyperbolicity of the Interpolation Manifold?

Related coverage