# Can In-context Learning Really Generalize to Out-of-distribution Tasks?

arXiv: https://arxiv.org/abs/2410.09695 · HTML: https://mlsanity.com/paper/2410.09695/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 38 references that are in the map.
- First posted on arXiv 13 Oct 2024.

## Builds on (most cited first, 30 of 34)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [An Explanation of In-context Learning as Implicit Bayesian Inference](https://mlsanity.com/paper/2111.02080/) (2021)
- [Transformers learn in-context by gradient descent](https://mlsanity.com/paper/2212.07677/) (2022)
- [What Can Transformers Learn In-Context? A Case Study of Simple Function Classes](https://mlsanity.com/paper/2208.01066/) (2022)
- [What learning algorithm is in-context learning? Investigations with linear models](https://mlsanity.com/paper/2211.15661/) (2022)
- [Trained Transformers Learn Linear Models In-Context](https://mlsanity.com/paper/2306.09927/) (2023)
- [Function Vectors in Large Language Models](https://mlsanity.com/paper/2310.15213/) (2023)
- [Transformers as Statisticians: Provable In-Context Learning with In-Context Algorithm Selection](https://mlsanity.com/paper/2306.04637/) (2023)
- [Transformers learn to implement preconditioned gradient descent for in-context learning](https://mlsanity.com/paper/2306.00297/) (2023)
- [Label Words are Anchors: An Information Flow Perspective for Understanding In-Context Learning](https://mlsanity.com/paper/2305.14160/) (2023)
- [What In-Context Learning "Learns" In-Context: Disentangling Task Recognition and Task Learning](https://mlsanity.com/paper/2305.09731/) (2023)
- [Pretraining task diversity and the emergence of non-Bayesian in-context learning for regression](https://mlsanity.com/paper/2306.15063/) (2023)
- [The Learnability of In-Context Learning](https://mlsanity.com/paper/2303.07895/) (2023)
- [What and How does In-Context Learning Learn? Bayesian Model Averaging, Parameterization, and Generalization](https://mlsanity.com/paper/2305.19420/) (2023)
- [The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains](https://mlsanity.com/paper/2402.11004/) (2024)
- [In-Context Convergence of Transformers](https://mlsanity.com/paper/2310.05249/) (2023)
- [What needs to go right for an induction head? A mechanistic study of in-context learning circuits and their formation](https://mlsanity.com/paper/2404.07129/) (2024)
- [From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples](https://mlsanity.com/paper/2404.07544/) (2024)
- [In-Context Learning Learns Label Relationships but Is Not Conventional Learning](https://mlsanity.com/paper/2307.12375/) (2023)
- [Identifying Semantic Induction Heads to Understand In-Context Learning](https://mlsanity.com/paper/2402.13055/) (2024)
- [Induction Heads as an Essential Mechanism for Pattern Matching in In-context Learning](https://mlsanity.com/paper/2407.07011/) (2024)
- [Pretraining Data Mixtures Enable Narrow Model Selection Capabilities in Transformer Models](https://mlsanity.com/paper/2311.00871/) (2023)
- [A Theoretical Understanding of Self-Correction through In-context Alignment](https://mlsanity.com/paper/2405.18634/) (2024)
- [Dual Operating Modes of In-Context Learning](https://mlsanity.com/paper/2402.18819/) (2024)
- [More Data Can Hurt for Linear Regression: Sample-wise Double Descent](https://mlsanity.com/paper/1912.07242/) (2019)
- [How Do Nonlinear Transformers Learn and Generalize in In-Context Learning?](https://mlsanity.com/paper/2402.15607/) (2024)
- [Out-of-distribution generalization via composition: a lens through induction heads in Transformers](https://mlsanity.com/paper/2408.09503/) (2024)

## Built on by (most cited first, 23 of 23)

- [Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens](https://mlsanity.com/paper/2508.01191/) (2025)
- [Train Once, Reuse Everywhere: Generalizable Implicit In-Context Learning by Routing Attention](https://mlsanity.com/paper/2509.22854/) (2025)
- [Out-of-Distribution Generalization of In-Context Learning: A Low-Dimensional Subspace Perspective](https://mlsanity.com/paper/2505.14808/) (2025)
- [Can GRPO Help LLMs Transcend Their Pretraining Origin?](https://mlsanity.com/paper/2510.15990/) (2025)
- [A Theoretical Framework for OOD Robustness in Transformers using Gevrey Classes](https://mlsanity.com/paper/2504.12991/) (2025)
- [Retrieval-Augmented Generation as Noisy In-Context Learning: A Unified Theory and Risk Bounds](https://mlsanity.com/paper/2506.03100/) (2025)
- [Harnessing Synthetic Data from Generative AI for Statistical Inference](https://mlsanity.com/paper/2603.05396/) (2026)
- [InPhyRe Discovers: Large Multimodal Models Struggle in Inductive Physical Reasoning](https://mlsanity.com/paper/2509.12263/) (2025)
- [Review of Hallucination Understanding in Large Language and Vision Models](https://mlsanity.com/paper/2510.00034/) (2025)
- [Bifrost: Steering Strategic Trajectories to Bridge Contextual Gaps for Self-Improving Agents](https://mlsanity.com/paper/2602.05810/) (2026)
- [SAE as a Crystal Ball: Interpretable Features Predict Cross-domain Transferability of LLMs without Training](https://mlsanity.com/paper/2603.02908/) (2026)
- [Large language models reorganize representational geometry during in-context learning](https://mlsanity.com/paper/2605.28854/) (2026)
- [Binary Autoencoder for Mechanistic Interpretability of Large Language Models](https://mlsanity.com/paper/2509.20997/) (2025)
- [Learning Linear Regression with Low-Rank Tasks in-Context](https://mlsanity.com/paper/2510.04548/) (2025)
- [Can AI Scientist Agents Learn from Lab-in-the-Loop Feedback? Evidence from Iterative Perturbation Discovery](https://mlsanity.com/paper/2603.26177/) (2026)
- [The Effect of Training Task Diversity on In-Context Learning through the Lens of Low-Dimensional Subspaces](https://mlsanity.com/paper/2606.06814/) (2026)
- [Can Transformers Break Encryption Schemes via In-Context Learning?](https://mlsanity.com/paper/2508.10235/) (2025)
- [When Looks Do Not Lie: Discourse Structure Guided In-Context Learning for Faithful Diagram Generation](https://mlsanity.com/paper/2601.20476/) (2026)
- [Toward Robust In-Context Learning: Leveraging Out-of-distribution Proxies for Target Inaccessible Demonstration Retrieval](https://mlsanity.com/paper/2606.00014/) (2026)
- [The Generalization Spectrum: A Chromatographic Approach to Evaluating Learning Algorithms](https://mlsanity.com/paper/2606.25450/) (2026)
- [A Unified Framework for In-Context Learning with Causal and Masked Language Models](https://mlsanity.com/paper/2607.04081/) (2026)
- [A Minimal Interpretable Architecture for Zero-Shot Reconstruction of Dynamical Systems](https://mlsanity.com/paper/2607.14937/) (2026)
- [Theory on Attention Dynamics for Out-of-Distribution In-Context Learning](https://mlsanity.com/paper/2609.36448/) (2026)

## Abstract (arXiv)

In this work, we explore the mechanism of in-context learning (ICL) on out-of-distribution (OOD) tasks that were not encountered during training. To achieve this, we conduct synthetic experiments where the objective is to learn OOD mathematical functions through ICL using a GPT-2 model. We reveal that Transformers may struggle to learn OOD task functions through ICL. Specifically, ICL performance resembles implementing a function within the pretraining hypothesis space and optimizing it with gradient descent based on the in-context examples. Additionally, we investigate ICL's well-documented ability to learn unseen abstract labels in context. We demonstrate that such ability only manifests in the scenarios without distributional shifts and, therefore, may not serve as evidence of new-task-learning ability. Furthermore, we assess ICL's performance on OOD tasks when the model is pretrained on multiple tasks. Both empirical and theoretical analyses demonstrate the existence of the low-test-error preference of ICL, where it tends to implement the pretraining function that yields low test error in the testing context. We validate this through numerical experiments. This new theoretical result, combined with our empirical findings, elucidates the mechanism of ICL in addressing OOD tasks.