Pith. sign in

REVIEW 5 cited by

Grokfast: Accelerated Grokking by Amplifying Slow Gradients

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2405.20233 v2 pith:FB5CS7X2 submitted 2024-05-30 cs.LG cs.AI

classification cs.LGcs.AI
keywords grokkinggeneralizationgradientsaccelerateartifactcodecomponentcomponents
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

One puzzling artifact in machine learning dubbed grokking is where delayed generalization is achieved tenfolds of iterations after near perfect overfitting to the training data. Focusing on the long delay itself on behalf of machine learning practitioners, our goal is to accelerate generalization of a model under grokking phenomenon. By regarding a series of gradients of a parameter over training iterations as a random signal over time, we can spectrally decompose the parameter trajectories under gradient descent into two components: the fast-varying, overfitting-yielding component and the slow-varying, generalization-inducing component. This analysis allows us to accelerate the grokking phenomenon more than $\times 50$ with only a few lines of code that amplifies the slow-varying components of gradients. The experiments show that our algorithm applies to diverse tasks involving images, languages, and graphs, enabling practical availability of this peculiar artifact of sudden generalization. Our code is available at https://github.com/ironjr/grokfast.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. How to Tame Grokking: Representation Geometry as a Control Signal

    cs.LG 2026-07 conditional novelty 6.5 of 10

    Dimensionality collapse precedes grokking; GeomDR, a spectral regularizer on hidden covariances, accelerates it up to 52× on modular and permutation tasks for MLPs and transformers.

  2. The Kinetics of Training: A Driven-Nucleation Rate Law for Emergence, Plasticity Loss, and Circuit Control in Language Models

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Capability formation in small transformers is claimed to obey a driven-nucleation rate law J = Nνσ(c)e^{−βK} − D, read forward as emergence, backward as plasticity loss, and completed as circuit control.

  3. The Active Ingredient in Muon's Grokking

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Orthogonalization—not spectral scaling—is the active ingredient in Muon's faster grokking, and reducing Newton-Schulz iterations trades first-crossing speed for solution stability.

  4. GrokAlign: Geometric Characterisation and Acceleration of Grokking

    cs.LG 2025-06 conditional novelty 5.0 of 10

    GrokAlign, a Jacobian-norm regularizer, accelerates grokking by aligning Jacobians with training data, and centroid alignment tracks when generalization and robustness emerge.

  5. Mechanistic Insights into Grokking from the Embedding Layer

    cs.LG 2025-05 conditional novelty 5.0 of 10

    Trainable embeddings in a simple MLP cause delayed generalization (grokking) on modular arithmetic, and a higher embedding learning rate plus balanced sampling accelerates it.

Pith tools