# Language Models "Grok" to Copy

arXiv: https://arxiv.org/abs/2409.09281 · HTML: https://mlsanity.com/paper/2409.09281/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Generation and context integration, one of 15 directions in Retrieval-augmented generation.
- Builds on 17 of its 24 references that are in the map.
- First posted on arXiv 14 Sep 2024.

## Placed in

- Retrieval-augmented generation / [Generation and context integration](https://mlsanity.com/rag/generation-and-context-integration/)

## Builds on (most cited first, 17 of 17)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity](https://mlsanity.com/paper/2104.08786/) (2021)
- [In-context Learning and Induction Heads](https://mlsanity.com/paper/2209.11895/) (2022)
- [Progress measures for grokking via mechanistic interpretability](https://mlsanity.com/paper/2301.05217/) (2023)
- [Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets](https://mlsanity.com/paper/2201.02177/) (2022)
- [Towards Understanding Grokking: An Effective Theory of Representation Learning](https://mlsanity.com/paper/2205.10343/) (2022)
- [How Do Large Language Models Acquire Factual Knowledge During Pretraining?](https://mlsanity.com/paper/2406.11813/) (2024)
- [Grokked Transformers are Implicit Reasoners: A Mechanistic Journey to the Edge of Generalization](https://mlsanity.com/paper/2405.15071/) (2024)
- [Grokfast: Accelerated Grokking by Amplifying Slow Gradients](https://mlsanity.com/paper/2405.20233/) (2024)
- [Fortify the Shortest Stave in Attention: Enhancing Context Awareness of Large Language Models for Effective Tool Use](https://mlsanity.com/paper/2312.04455/) (2023)
- [Batch-ICL: Effective, Efficient, and Order-Agnostic In-Context Learning](https://mlsanity.com/paper/2401.06469/) (2024)
- [Mixture of In-Context Experts Enhance LLMs' Long Context Awareness](https://mlsanity.com/paper/2406.19598/) (2024)
- [Deep Grokking: Would Deep Neural Networks Generalize Better?](https://mlsanity.com/paper/2405.19454/) (2024)
- [Grokking Beyond Neural Networks: An Empirical Exploration with Model Complexity](https://mlsanity.com/paper/2310.17247/) (2023)

## Built on by (most cited first, 5 of 5)

- [Grokking at the Edge of Numerical Stability](https://mlsanity.com/paper/2501.04697/) (2025)
- [Context parroting: A simple but tough-to-beat baseline for foundation models in scientific machine learning](https://mlsanity.com/paper/2505.11349/) (2025)
- [ResidualDroppath: Enhancing Feature Reuse over Residual Connections](https://mlsanity.com/paper/2411.09475/) (2024)
- [In-Context Learning Without Copying](https://mlsanity.com/paper/2511.05743/) (2025)
- [SafeLLM: Extraction as a Hallucination-Resistant Alternative to Rewriting in Safety-Critical Settings](https://mlsanity.com/paper/2606.12897/) (2026)

## Abstract (arXiv)

We examine the pre-training dynamics of language models, focusing on their ability to copy text from preceding context--a fundamental skill for various LLM applications, including in-context learning (ICL) and retrieval-augmented generation (RAG). We propose a novel perspective that Transformer-based language models develop copying abilities similarly to grokking, which refers to sudden generalization on test set long after the model fit to the training set. Our experiments yield three arguments: (1) The pre-training loss decreases rapidly, while the context copying ability of models initially lags and then abruptly saturates. (2) The speed of developing copying ability is independent of the number of tokens trained, similarly to how grokking speed is unaffected by dataset size as long as the data distribution is preserved. (3) Induction heads, the attention heads responsible for copying, form from shallow to deep layers during training, mirroring the development of circuits in deeper layers during grokking. We contend that the connection between grokking and context copying can provide valuable insights for more effective language model training, ultimately improving in-context performance. For example, we demonstrated that techniques that enhance grokking, such as regularization, either accelerate or enhance the development of context copying.