Provable Low-Frequency Bias of In-Context Learning of Representations

• URL: http://arxiv.org/abs/2507.13540v2
• Date: Wed, 30 Jul 2025 01:51:41 GMT
• Title: Provable Low-Frequency Bias of In-Context Learning of Representations
• Authors: Yongyi Yang, Hidenori Tanaka, Wei Hu,
• Abstract summary: In-context learning (ICL) enables large language models (LLMs) to acquire new behaviors from the input sequence alone without any parameter updates.<n>Recent studies have shown that ICL can surpass the original meaning learned in pretraining stage through internalizing the structure the data-generating process (DGP) of the prompt into the hidden representations.<n>We present the first rigorous explanation of such phenomena by introducing a unified framework of double convergence.<n>This double convergence process leads to an implicit bias towards smooth (low-frequency) representations, which we prove analytically and verify empirically.
• Score: 19.066378730056275
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: In-context learning (ICL) enables large language models (LLMs) to acquire new behaviors from the input sequence alone without any parameter updates. Recent studies have shown that ICL can surpass the original meaning learned in pretraining stage through internalizing the structure the data-generating process (DGP) of the prompt into the hidden representations. However, the mechanisms by which LLMs achieve this ability is left open. In this paper, we present the first rigorous explanation of such phenomena by introducing a unified framework of double convergence, where hidden representations converge both over context and across layers. This double convergence process leads to an implicit bias towards smooth (low-frequency) representations, which we prove analytically and verify empirically. Our theory explains several open empirical observations, including why learned representations exhibit globally structured but locally distorted geometry, and why their total energy decays without vanishing. Moreover, our theory predicts that ICL has an intrinsic robustness towards high-frequency noise, which we empirically confirm. These results provide new insights into the underlying mechanisms of ICL, and a theoretical foundation to study it that hopefully extends to more general data distributions and settings.

Related papers

• Generalization or Hallucination? Understanding Out-of-Context Reasoning in Transformers [76.42159902257677]
  We argue that both behaviors stem from a single mechanism known as out-of-context reasoning (OCR)<n>OCR drives both generalization and hallucination, depending on whether the associated concepts are causally related.<n>Our work provides a theoretical foundation for understanding the OCR phenomenon, offering a new lens for analyzing and mitigating undesirable behaviors from knowledge injection.
  arXiv  Detail & Related papers  (2025-06-12T16:50:45Z)
• How does Transformer Learn Implicit Reasoning? [41.315116538534106]
  We study how implicit multi-hop reasoning emerges by training transformers from scratch in a controlled symbolic environment.<n>We find that training with atomic triples is not necessary but accelerates learning, and that second-hop generalization relies on query-level exposure to specific compositional structures.
  arXiv  Detail & Related papers  (2025-05-29T17:02:49Z)
• From Compression to Expansion: A Layerwise Analysis of In-Context Learning [20.64102133977965]
  In-context learning (ICL) enables large language models to adapt to new tasks without weight updates by learning from demonstration sequences.<n>We conduct a statistical geometric analysis of ICL representations to investigate how task-specific information is captured across layers.<n>Our findings reveal an intriguing layerwise dynamic in ICL, highlight how structured representations emerge within LLMs, and showcase that analyzing internal representations can facilitate a deeper understanding of model behavior.
  arXiv  Detail & Related papers  (2025-05-22T22:22:03Z)
• The Curse of CoT: On the Limitations of Chain-of-Thought in In-Context Learning [39.613595533503144]
  Chain-of-Thought (CoT) prompting has been widely recognized for its ability to enhance reasoning capabilities in large language models.<n>We show that CoT consistently underperforms direct answering across varying model scales and benchmark complexities.<n>Our analysis uncovers a fundamental explicit-implicit duality driving CoT's performance in pattern-based ICL.
  arXiv  Detail & Related papers  (2025-04-07T13:51:06Z)
• Toward Understanding In-context vs. In-weight Learning [50.24035812301655]
  We identify simplified distributional properties that give rise to the emergence and disappearance of in-context learning.<n>We then extend the study to a full large language model, showing how fine-tuning on various collections of natural language prompts can elicit similar in-context and in-weight learning behaviour.
  arXiv  Detail & Related papers  (2024-10-30T14:09:00Z)
• Class-wise Activation Unravelling the Engima of Deep Double Descent [0.0]
  Double descent presents a counter-intuitive aspect within the machine learning domain. In this study, we revisited the phenomenon of double descent and discussed the conditions of its occurrence.
  arXiv  Detail & Related papers  (2024-05-13T12:07:48Z)
• A Theoretical Analysis of Self-Supervised Learning for Vision Transformers [66.08606211686339]
  Masked autoencoders (MAE) and contrastive learning (CL) capture different types of representations.<n>We study the training dynamics of one-layer softmax-based vision transformers (ViTs) on both MAE and CL objectives.
  arXiv  Detail & Related papers  (2024-03-04T17:24:03Z)
• Beyond DAGs: A Latent Partial Causal Model for Multimodal Learning [80.44084021062105]
  We propose a novel latent partial causal model for multimodal data, featuring two latent coupled variables, connected by an undirected edge, to represent the transfer of knowledge across modalities.<n>Under specific statistical assumptions, we establish an identifiability result, demonstrating that representations learned by multimodal contrastive learning correspond to the latent coupled variables up to a trivial transformation.<n>Experiments on a pre-trained CLIP model embodies disentangled representations, enabling few-shot learning and improving domain generalization across diverse real-world datasets.
  arXiv  Detail & Related papers  (2024-02-09T07:18:06Z)
• Sparsity-Guided Holistic Explanation for LLMs with Interpretable Inference-Time Intervention [53.896974148579346]
  Large Language Models (LLMs) have achieved unprecedented breakthroughs in various natural language processing domains. The enigmatic black-box'' nature of LLMs remains a significant challenge for interpretability, hampering transparent and accountable applications. We propose a novel methodology anchored in sparsity-guided techniques, aiming to provide a holistic interpretation of LLMs.
  arXiv  Detail & Related papers  (2023-12-22T19:55:58Z)
• A Theory of Emergent In-Context Learning as Implicit Structure Induction [8.17811111226145]
  Scaling large language models leads to an emergent capacity to learn in-context from example demonstrations. We argue that in-context learning relies on recombination of compositional operations found in natural language data. We show how in-context learning is supported by a representation of the input's compositional structure.
  arXiv  Detail & Related papers  (2023-03-14T15:24:05Z)
• A Free Lunch from the Noise: Provable and Practical Exploration for Representation Learning [55.048010996144036]
  We show that under some noise assumption, we can obtain the linear spectral feature of its corresponding Markov transition operator in closed-form for free. We propose Spectral Dynamics Embedding (SPEDE), which breaks the trade-off and completes optimistic exploration for representation learning by exploiting the structure of the noise.
  arXiv  Detail & Related papers  (2021-11-22T19:24:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.