---
id: 20260913-T0-09
title: "Grokking何时发生？研究者量化记忆到泛化的转变规律"
title_en: "When Does Grokking Happen? Researchers Quantify Memorization-to-Generalization Transition"
url: https://ai.daily.yangsir.net/daily/20260913-T0-09
issue_date: 2026-09-13
publish_date: 2026-09-12T04:00:00.000Z
category: research
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2609.10657
---

# Grokking何时发生？研究者量化记忆到泛化的转变规律

神经网络训练超过记忆点后，常出现延迟的泛化转变，即grokking现象。已有理论解释了为什么会发生，但对其发生时间的定量规律尚不清晰。新研究通过缩放定律和相结构分析，量化了grokking转变的时间点与模型规模、数据量等变量的关系。这为预测和调控训练过程提供了可计算的理论框架。

## English Version

**When Does Grokking Happen? Researchers Quantify Memorization-to-Generalization Transition**

Neural networks often undergo a delayed transition to generalization after passing the memorization point, known as grokking. Existing theory explains why it happens but not when. A new study uses scaling laws and phase structure analysis to quantify how the timing of grokking relates to model size, data volume, and other variables. This provides a computable theoretical framework for predicting and controlling training dynamics.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2609.10657)

**详情页**：https://ai.daily.yangsir.net/daily/20260913-T0-09

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*