paper-with-me

홈 › Papers

Failure of contextual invariance in large language models

2026-03-24 · Sagar Kumar, Ariel Flint, Luca Maria Aiello, Andrea Baronchelli arxiv

Standard evaluation practices assume that large language model (LLM) outputs are stable when prompts are embedded in contextually equivalent discourses. Here, we test this assumption in the setting of gender inference. Using a controlled pronoun selection task, we introduce minimal, theoretically uninformative discourse context and find that this induces large, systematic shifts in model outputs. Correlations with cultural gender stereotypes, present in decontextualized settings, weaken or disappear once context is introduced, while theoretically irrelevant features, such as the gender of a pronoun for an unrelated referent, become the most informative predictors of model behavior. A Contextuality-by-Default analysis reveals that, in 19--52\% of cases across models, this dependence persists after accounting for all marginal effects of context on individual outputs and cannot be attributed to simple pronoun repetition. These findings show that LLM outputs violate contextual invariance even under near-identical syntactic formulations, with implications for bias benchmarking and deployment in high-stakes settings.

📄 PDF Abstract BibTeX arXiv:2603.23485

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Language-Guided Invariance Probing of Vision-Language Models

2025-11-17 · Jae Joong Lee arxiv

Recent vision-language models (VLMs) such as CLIP, OpenCLIP, EVA02-CLIP and SigLIP achieve strong zero-shot performance, but it is unclear how reliably they respond to controlled linguistic perturbations. We introduce La…

Image-text matching

Measuring Representation Robustness in Large Language Models for Geometry

2026-04-03 · Vedant Jawandhia, Yash Sinha, Murari Mandal, Ankan Pal 외 arxiv

Large language models (LLMs) are increasingly evaluated on mathematical reasoning, yet their robustness to equivalent problem representations remains poorly understood. In geometry, identical problems can be expressed in…

Mathematical Reasoning

Invariance Principle Meets Information Bottleneck for Out-of-Distribution Generalization

2021-06-11 · NeurIPS 2021 12 · Kartik Ahuja, Ethan Caballero, Dinghuai Zhang, Jean-Christophe Gagnon-Audet 외

The invariance principle from causality is at the heart of notable approaches such as invariant risk minimization (IRM) that seek to address out-of-distribution (OOD) generalization failures. Despite the promising theory…

Out-of-Distribution Generalizationregression

When Should Models Change Their Minds? Contextual Belief Management in Large Language Models

2026-05-28 · Haoming Xu, Weihong Xu, Zongrui Li, Mengru Wang 외 arxiv

Long-horizon interactions require language models to manage accumulating information: when to update their state, when to preserve their state, and what to ignore. We study this challenge as \textbf{Contextual Belief Man…

Reinforcement Learning

Inverse-Free Wilson Loops for Transformers: A Practical Diagnostic for Invariance and Order Sensitivity

2025-10-09 · Edward Y. Chang, Ethan Y. Chang arxiv

Large language models can change answers under harmless edits that matter in practice: RAG outputs flip when passages are reordered, fine-tuning erodes invariances learned at pretraining, debate or chain-of-thought promp…