Categorical Prior Lock-in: Why In-Context Learning Fails for Structured Data

• URL: http://arxiv.org/abs/2606.11961v1
• Date: Wed, 10 Jun 2026 11:41:13 GMT
• Title: Categorical Prior Lock-in: Why In-Context Learning Fails for Structured Data
• Abstract summary: We investigate the limits of in-context learning (ICL) for structured generation under distribution mismatches.<n>We identify a structural failure mode we term textitcategorical prior lock-in: the inability of ICL to update the model's prior over token distributions inherited from pre-training.
• Score: 2.327999580858199
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) are increasingly used as conditional generators for structured data, relying on in-context learning (ICL) to adapt to new distributions without parameter updates. We investigate the limits of ICL for structured generation under distribution mismatch, using high-cardinality tabular data as a controlled test case, and identify a structural failure mode we term \textit{categorical prior lock-in}: the inability of ICL to update the model's prior over token distributions inherited from pre-training. Across two 7B-parameter open-weight models, ICL improves numerical fidelity with additional examples but exhibits a sharp ceiling on categorical distributions, failing to reproduce rare classes entirely. Parameter-efficient fine-tuning (LoRA) overcomes these limitations but introduces measurable memorization risk and, in some cases, destabilizes structured output generation, highlighting a fundamental trade-off between adaptability and privacy.

Related papers

• On-the-go Forgetting without Explicit Unlearning via ERASE [2.6426981014417046]
  ERASE is a framework for on-the-go forgetting that suppresses the observable influence of private data without modifying model weights.<n>We rigorously characterize sufficient conditions when ERASE provably achieves functional forgetting of designated subclasses.<n> ERASE maintains the best observed balance between forgetting efficacy, computational efficiency, and retention fidelity over recent unlearning-based methods.
  arXiv  Detail & Related papers  (2026-09-05T08:09:27Z)
• CritICL: Inference-Time Weak-to-Strong Generalization from Small Language Model Failure Modes [8.494161214945963]
  We introduce CritICL, a novel inference-time framework that improves reasoning while maintaining high efficiency.<n>We utilize failure modes derived from weaker models and incorporate them into inference through critique-based in-context examples.<n>CRITICL consistently outperforms standard in-context learning and achieves performance competitive with or superior to test-time scaling methods.
  arXiv  Detail & Related papers  (2026-08-27T17:59:30Z)
• MIITA: Memory-Induced Inference-Time Adaptation for Continual Learning with Small Language Models [64.2820121997882]
  MIITA is a Memory-Induced Inference-Time Adaptation framework for supervised CL under constrained storage.<n>It stores supervised experiences as compact correction-direction prototypes with semantic anchors, and retrieves them at inference time using semantic and uncertainty-based cues.<n>Experiments show that MIITA consistently improves final performance and mitigates forgetting under fixed memory budgets.
  arXiv  Detail & Related papers  (2026-05-20T03:03:04Z)
• On Predicting the Post-training Potential of Pre-trained LLMs [60.07459271263409]
  We introduce a new task of predicting post-training potential - forecasting a base model's performance before post-training.<n>We propose RuDE, a unified framework that bypasses the generation gap of base models by leveraging response discrimination.<n>Experiments demonstrate a correlation greater than 90% with post-training performance.
  arXiv  Detail & Related papers  (2026-05-12T11:33:49Z)
• Excess Description Length of Learning Generalizable Predictors [7.527535569795127]
  We develop a formal information-theoretic framework for quantifying how much predictive structure fine-tuning extracts from a train dataset.<n>Our central quantity, Excess Description Length (EDL), is defined via prequential coding.<n>We establish that EDL is non-negative in expectation, converges to surplus description length in the infinite-data limit, and provides bounds on expected generalization gain.
  arXiv  Detail & Related papers  (2026-01-08T08:46:42Z)
• A Framework for Quantifying How Pre-Training and Context Benefit In-Context Learning [52.07397258423034]
  We propose a new framework to analyze the ICL performance in a class of realistic settings.<n>We derive the precise relationship between ICL performance, context length and the KL divergence between pre-train and query task distribution.
  arXiv  Detail & Related papers  (2025-10-26T09:21:29Z)
• SelfAug: Mitigating Catastrophic Forgetting in Retrieval-Augmented Generation via Distribution Self-Alignment [49.86376148975563]
  Large language models (LLMs) have revolutionized natural language processing through their capabilities in understanding and executing diverse tasks.<n> supervised fine-tuning, particularly in Retrieval-Augmented Generation (RAG) scenarios, often leads to catastrophic forgetting.<n>We propose SelfAug, a self-distribution alignment method that aligns input sequence logits to preserve the model's semantic distribution.
  arXiv  Detail & Related papers  (2025-09-04T06:50:47Z)
• Surprise Calibration for Better In-Context Learning [6.566285172635043]
  In-context learning (ICL) has emerged as a powerful paradigm for task adaptation in large language models.<n>Existing bias calibration methods apply fixed class priors across all inputs, limiting their efficacy in dynamic ICL settings.<n>We introduce a novel method-Surprise (SC), which captures the temporal dynamics of class priors.
  arXiv  Detail & Related papers  (2025-06-15T10:04:42Z)
• Singular Value Fine-tuning for Few-Shot Class-Incremental Learning [38.777602828340356]
  Class-Incremental Learning (CIL) aims to prevent catastrophic forgetting of previously learned classes while incorporating new ones.<n>We propose the Singular Value Finetuning for FSCIL (SVFCL)<n>SVFCL applies singular value decomposition to the foundation model weights, keeping the singular vectors fixed while fine-tuning the singular values for each task, and then merging them.
  arXiv  Detail & Related papers  (2025-03-13T09:57:28Z)
• EquiTabPFN: A Target-Permutation Equivariant Prior Fitted Networks [55.214444066134114]
  We design a fully target-equivariant architecture-ensuring permutation invariance via equivariant encoders, decoders, and a bi-attention mechanism.<n> Empirical evaluation on standard classification benchmarks shows that, on datasets with more classes than those seen during pre-training, our model matches or surpasses existing methods while incurring lower computational overhead.
  arXiv  Detail & Related papers  (2025-02-10T17:11:20Z)
• Continuous Knowledge-Preserving Decomposition with Adaptive Layer Selection for Few-Shot Class-Incremental Learning [73.59672160329296]
  CKPD-FSCIL is a unified framework that unlocks the underutilized capacity of pretrained weights.<n>Our method consistently outperforms state-of-the-art approaches in both adaptability and knowledge retention.
  arXiv  Detail & Related papers  (2025-01-09T07:18:48Z)
• Structured Language Generation Model for Robust Structure Prediction [6.4736137270915215]
  We propose a framework that reduces sequence-to-sequence problems to classification problems via methodologies in loss calibration and decoding method. Our experimental results show that SLGM is able to maintain performance without explicit dataset information, follow and potentially replace dataset-specific fine-tuning.
  arXiv  Detail & Related papers  (2024-02-14T06:33:22Z)
• When Does Confidence-Based Cascade Deferral Suffice? [69.28314307469381]
  Cascades are a classical strategy to enable inference cost to vary adaptively across samples. A deferral rule determines whether to invoke the next classifier in the sequence, or to terminate prediction. Despite being oblivious to the structure of the cascade, confidence-based deferral often works remarkably well in practice.
  arXiv  Detail & Related papers  (2023-07-06T04:13:57Z)
• Learning Prompt-Enhanced Context Features for Weakly-Supervised Video Anomaly Detection [37.99031842449251]
  Video anomaly detection under weak supervision presents significant challenges. We present a weakly supervised anomaly detection framework that focuses on efficient context modeling and enhanced semantic discriminability. Our approach significantly improves the detection accuracy of certain anomaly sub-classes, underscoring its practical value and efficacy.
  arXiv  Detail & Related papers  (2023-06-26T06:45:16Z)

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.