Imagen de Portada

Positional Fragility in LLMs: How Offset Effects Reshape Our Understanding of Memorization Risks

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Xu, Yixuan, Llaquet, Antoni-Joan Solergibert i, Bosselut, Antoine, Schlag, Imanol
Formato:	Preprint
Publicado:	2025
Materias:	Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Can Performant LLMs Be Ethical? Quantifying the Impact of Web Crawling Opt-Outs
por: Fan, Dongyang, et al.
Publicado: (2025)

Benchmarking Concept-Spilling Across Languages in LLMs
por: Badanin, Ilia, et al.
Publicado: (2026)

On the Effect of (Near) Duplicate Subwords in Language Modelling
por: Schäfer, Anton, et al.
Publicado: (2024)

From Language to Cognition: How LLMs Outgrow the Human Language Network
por: AlKhamissi, Badr, et al.
Publicado: (2025)

Tracking the Limits of Knowledge Propagation: How LLMs Fail at Multi-Step Reasoning with Conflicting Knowledge
por: Feng, Yiyang, et al.
Publicado: (2026)

Memorization or Reasoning? Exploring the Idiom Understanding of LLMs
por: Kim, Jisu, et al.
Publicado: (2025)

AbstRaL: Augmenting LLMs' Reasoning by Reinforcing Abstract Thinking
por: Gao, Silin, et al.
Publicado: (2025)

LLMs Are In-Context Bandit Reinforcement Learners
por: Monea, Giovanni, et al.
Publicado: (2024)

Understanding Verbatim Memorization in LLMs Through Circuit Discovery
por: Lasy, Ilya, et al.
Publicado: (2025)

Spurious Rewards Paradox: Mechanistically Understanding How RLVR Activates Memorization Shortcuts in LLMs
por: Yan, Lecheng, et al.
Publicado: (2026)

Instruction-tuning Aligns LLMs to the Human Brain
por: Aw, Khai Loong, et al.
Publicado: (2023)

Do LLMs Game Formalization? Evaluating Faithfulness in Logical Reasoning
por: Kim, Kyuhee, et al.
Publicado: (2026)

SFT-then-RL Outperforms Mixed-Policy Methods for LLM Reasoning
por: Limozin, Alexis, et al.
Publicado: (2026)

RLMEval: Evaluating Research-Level Neural Theorem Proving
por: Poiroux, Auguste, et al.
Publicado: (2025)

The Role of Language Imbalance in Cross-lingual Generalisation: Insights from Cloned Language Experiments
por: Schäfer, Anton, et al.
Publicado: (2024)

JOBSKAPE: A Framework for Generating Synthetic Job Postings to Enhance Skill Matching
por: Magron, Antoine, et al.
Publicado: (2024)

How Robust Are Router-LLMs? Analysis of the Fragility of LLM Routing Capabilities
por: Kassem, Aly M., et al.
Publicado: (2025)

Memorization and Knowledge Injection in Gated LLMs
por: Pan, Xu, et al.
Publicado: (2025)

How LLMs Distort Our Written Language
por: Abdulhai, Marwa, et al.
Publicado: (2026)

Be like a Goldfish, Don't Memorize! Mitigating Memorization in Generative LLMs
por: Hans, Abhimanyu, et al.
Publicado: (2024)

Let Me Teach You: Pedagogical Foundations of Feedback for Language Models
por: Borges, Beatriz, et al.
Publicado: (2023)

Making Reasoning Matter: Measuring and Improving Faithfulness of Chain-of-Thought Reasoning
por: Paul, Debjit, et al.
Publicado: (2024)

Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining
por: Bayazit, Deniz, et al.
Publicado: (2025)

Quantifying In-Context Reasoning Effects and Memorization Effects in LLMs
por: Lou, Siyu, et al.
Publicado: (2024)

An Engineering Journey Training Large Language Models at Scale on Alps: The Apertus Experience
por: Coles, Jonathan, et al.
Publicado: (2026)

Rethinking Skill Extraction in the Job Market Domain using Large Language Models
por: Nguyen, Khanh Cao, et al.
Publicado: (2024)

MedMT-Bench: Can LLMs Memorize and Understand Long Multi-Turn Conversations in Medical Scenarios?
por: Yang, Lin, et al.
Publicado: (2026)

Long-Term Ad Memorability: Understanding & Generating Memorable Ads
por: SI, Harini, et al.
Publicado: (2023)

Alpaca against Vicuna: Using LLMs to Uncover Memorization of LLMs
por: Kassem, Aly M., et al.
Publicado: (2024)

PERK: Long-Context Reasoning as Parameter-Efficient Test-Time Learning
por: Chen, Zeming, et al.
Publicado: (2025)

Complex Reasoning over Logical Queries on Commonsense Knowledge Graphs
por: Fang, Tianqing, et al.
Publicado: (2024)

PICLe: Pseudo-Annotations for In-Context Learning in Low-Resource Named Entity Detection
por: Mamooler, Sepideh, et al.
Publicado: (2024)

Mitigating Memorization in LLMs using Activation Steering
por: Suri, Manan, et al.
Publicado: (2025)

How Value Induction Reshapes LLM Behaviour
por: Arora, Arnav, et al.
Publicado: (2026)

Edited Media Understanding Frames: Reasoning About the Intent and Implications of Visual Misinformation
por: Da, Jeff, et al.
Publicado: (2020)

Deriving Activation Functions Using Integration
por: Huang, Allen Hao, et al.
Publicado: (2024)

Unveiling Over-Memorization in Finetuning LLMs for Reasoning Tasks
por: Ruan, Zhiwen, et al.
Publicado: (2025)

The LLM Language Network: A Neuroscientific Approach for Identifying Causally Task-Relevant Units
por: AlKhamissi, Badr, et al.
Publicado: (2024)

Brain-Like Language Processing via a Shallow Untrained Multihead Attention Network
por: AlKhamissi, Badr, et al.
Publicado: (2024)

Position: Privacy Is Not Just Memorization!
por: Mireshghallah, Niloofar, et al.
Publicado: (2025)