paper-with-me

Papers

Grokking Is Conditional and Fragile: A Fully-Tractable, Multi-Seed Study at 12K Parameters

2026-07-06 · Yoshiyuki Ootani arxiv

Grokking -- the delayed onset of generalization long after a network has fit its training set - -is usually studied in models too large to read completely and reported from single training runs. We instead study a publicly released ~11,856-parameter Llama-style transformer (Glimmer-1-Base) on modular arithmetic, small enough to enumerate its weights, attention, and full input-output map, and we measure grokking as a multi-seed rate rather than a single outcome. In this fully-tractable regime grokking is a conditional, fragile phase transition. It is gated by training-set coverage, whose threshold tracks output cardinality (the modulus) more than task structure, an ordering that holds above the transition and across a ten-fold change in domain size. Weight decay reproduces the Omnigrok inverted-U at 12K parameters, a positive control on the rate measurement. Grokking also sits on a numerical knife-edge: two perturbations of the floating-point environment -- CPU thread count (reduction order) and CPU-versus-GPU execution -- each flip a minority of same-seed outcomes without a detectable shift in the aggregate rate. Decomposition into sub-task specialists helps chiefly by making coverage cheap rather than by adding supervision. Methodologically, multi-seed control under a fixed numerical environment overturns three dramatic single-run narratives in our own data, each a seed confound. The unit of evidence for grokking must therefore be a multi-seed rate under a pinned numerical environment, checked where possible against a direct reading of the model.

📄 PDF Abstract BibTeX arXiv:2607.05104

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

TD-Grokking: Learning from Zero-Reward Problems by Training-Time Decomposition

2026-06-03 · Ningyuan Xi, Hao Xu, Hongsheng Xin, Ning Miao arxiv

Large language models (LLMs) have made remarkable progress in reasoning tasks, largely driven by post-training paradigms, especially reinforcement learning with verifiable rewards (RLVR). However, a critical bottleneck p…

Reinforcement Learning

Grokking at the Edge of Linear Separability

2024-10-06 · Alon Beck, Noam Levi, Yohai Bar-Sinai

We study the generalization properties of binary logistic classification in a simplified setting, for which a "memorizing" and "generalizing" solution can always be strictly defined, and elucidate empirically and analyti…

Grokking modular arithmetic

2023-01-06 · Andrey Gromov

We present a simple neural network that can learn modular arithmetic tasks and exhibits a sudden jump in generalization known as ``grokking''. Concretely, we present (i) fully-connected two-layer networks that exhibit gr…

Beyond Progress Measures: Theoretical Insights into the Mechanism of Grokking

2025-04-04 · Zihan Gu, Ruoyu Chen, Hua Zhang, Yue Hu 외

Grokking, referring to the abrupt improvement in test accuracy after extended overfitting, offers valuable insights into the mechanisms of model generalization. Existing researches based on progress measures imply that g…

Grokking in the Wild: Data Augmentation for Real-World Multi-Hop Reasoning with Transformers

2025-04-29 · Roman Abramov, Felix Steinbauer, Gjergji Kasneci

Transformers have achieved great success in numerous NLP tasks but continue to exhibit notable gaps in multi-step factual reasoning, especially when real-world knowledge is sparse. Recent advances in grokking have demons…

Data AugmentationKnowledge GraphsMemorizationWorld Knowledge