# Investigating the Pre-Training Dynamics of In-Context Learning: Task Recognition vs. Task Learning

arXiv: https://arxiv.org/abs/2406.14022 · HTML: https://mlsanity.com/paper/2406.14022/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Robot manipulation imitation learning, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 22 of its 37 references that are in the map.
- First posted on arXiv 20 Jun 2024.

## Placed in

- Offline RL and imitation learning from data / [Robot manipulation imitation learning](https://mlsanity.com/offlinerl/robot-manipulation-imitation-learning/)

## Builds on (most cited first, 22 of 22)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [A large annotated corpus for learning natural language inference](https://mlsanity.com/paper/1508.05326/) (2015)
- [Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling](https://mlsanity.com/paper/2304.01373/) (2023)
- [Rethinking the Role of Demonstrations: What Makes In-Context Learning Work?](https://mlsanity.com/paper/2202.12837/) (2022)
- [StarCoder: may the source be with you!](https://mlsanity.com/paper/2305.06161/) (2023)
- [MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies](https://mlsanity.com/paper/2404.06395/) (2024)
- [A Survey on In-context Learning](https://mlsanity.com/paper/2301.00234/) (2022)
- [Baichuan 2: Open Large-scale Language Models](https://mlsanity.com/paper/2309.10305/) (2023)
- [OLMo: Accelerating the Science of Language Models](https://mlsanity.com/paper/2402.00838/) (2024)
- [Federated Learning with Matched Averaging](https://mlsanity.com/paper/2002.06440/) (2020)
- [Larger language models do in-context learning differently](https://mlsanity.com/paper/2303.03846/) (2023)
- [Good Debt or Bad Debt: Detecting Semantic Orientations in Economic Texts](https://mlsanity.com/paper/1307.5336/) (2013)
- [The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning](https://mlsanity.com/paper/2312.01552/) (2023)
- [Uniform convergence may be unable to explain generalization in deep learning](https://mlsanity.com/paper/1902.04742/) (2019)
- [What In-Context Learning "Learns" In-Context: Disentangling Task Recognition and Task Learning](https://mlsanity.com/paper/2305.09731/) (2023)
- [The mechanistic basis of data dependence and abrupt learning in an in-context classification task](https://mlsanity.com/paper/2312.03002/) (2023)
- [LLM360: Towards Fully Transparent Open-Source LLMs](https://mlsanity.com/paper/2312.06550/) (2023)
- [Optimizing Mode Connectivity via Neuron Alignment](https://mlsanity.com/paper/2009.02439/) (2020)
- [Investigating Societal Biases in a Poetry Composition System](https://mlsanity.com/paper/2011.02686/) (2020)
- [Dual Operating Modes of In-Context Learning](https://mlsanity.com/paper/2402.18819/) (2024)
- [Dual Process Learning: Controlling Use of In-Context vs. In-Weights Strategies with Weight Forgetting](https://mlsanity.com/paper/2406.00053/) (2024)

## Built on by (most cited first, 6 of 6)

- [DAWN-ICL: Strategic Planning of Problem-solving Trajectories for Zero-Shot In-Context Learning](https://mlsanity.com/paper/2410.20215/) (2024)
- [ICL CIPHERS: Quantifying "Learning" in In-Context Learning via Substitution Ciphers](https://mlsanity.com/paper/2504.19395/) (2025)
- [Illusion or Algorithm? Investigating Memorization, Emergence, and Symbolic Processing in In-Context Learning](https://mlsanity.com/paper/2505.11004/) (2025)
- [Review of Hallucination Understanding in Large Language and Vision Models](https://mlsanity.com/paper/2510.00034/) (2025)
- [Genomic Next-Token Predictors are In-Context Learners](https://mlsanity.com/paper/2511.12797/) (2025)
- [Decomposing Prediction Mechanisms for In-Context Recall](https://mlsanity.com/paper/2507.01414/) (2025)

## Abstract (arXiv)

The emergence of in-context learning (ICL) is potentially attributed to two major abilities: task recognition (TR) for recognizing the task from demonstrations and utilizing pre-trained priors, and task learning (TL) for learning from demonstrations. However, relationships between the two abilities and how such relationships affect the emergence of ICL is unclear. In this paper, we take the first step by examining the pre-training dynamics of the emergence of ICL. With carefully designed metrics, we find that these two abilities are, in fact, competitive during pre-training. Moreover, we observe a strong negative correlation between the competition and ICL performance. Further analysis of common pre-training factors (i.e., model size, dataset size, and data curriculum) demonstrates possible ways to manage the competition. Based on these insights, we propose a simple yet effective method to better integrate these two abilities for ICL at inference time. Through adaptive ensemble learning, the performance of ICL can be significantly boosted, enabling two small models to outperform a larger one with more than twice the parameters. The code is available at https://github.com/RUCAIBox/Competitive-ICL.