Imagen de Portada

Is Model Editing Built on Sand? Revealing Its Illusory Success and Fragile Foundation

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Liu, Wei, Xu, Haomei, Liu, Bingqing, Deng, Zhiying, Wang, Haozhao, Wang, Jun, Li, Ruixuan, Teh, Yee Whye, Lee, Wee Sun
Formato:	Preprint
Publicado:	2025
Materias:	Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Are We Evaluating the Edit Locality of LLM Model Editing Properly?
por: Liu, Wei, et al.
Publicado: (2026)

From Backward Spreading to Forward Replay: Revisiting Target Construction in LLM Parameter Editing
por: Liu, Wei, et al.
Publicado: (2026)

LinTree: Improving LLM Reasoning with Explicitly Structured Search Histories
por: Kang, Liwei, et al.
Publicado: (2026)

SofT-GRPO: Surpassing Discrete-Token LLM Reinforcement Learning via Gumbel-Reparameterized Soft-Thinking Policy Optimization
por: Zheng, Zhi, et al.
Publicado: (2025)

Extending Epistemic Uncertainty Beyond Parameters Would Assist in Designing Reliable LLMs
por: Nguyen-Hien, T. Duy, et al.
Publicado: (2025)

Breaking Free from MMI: A New Frontier in Rationalization by Probing Input Utilization
por: Liu, Wei, et al.
Publicado: (2025)

Adversarial Cooperative Rationalization: The Risk of Spurious Correlations in Even Clean Datasets
por: Liu, Wei, et al.
Publicado: (2025)

Is the MMI Criterion Necessary for Interpretability? Degenerating Non-causal Features to Plain Noise for Self-Rationalization
por: Liu, Wei, et al.
Publicado: (2024)

NoProp: Training Neural Networks without Full Back-propagation or Full Forward-propagation
por: Li, Qinyu, et al.
Publicado: (2025)

Enhancing RAG with Active Learning on Conversation Records: Reject Incapables and Answer Capables
por: Geng, Xuzhao, et al.
Publicado: (2025)

Verifier-Backed Hard Problem Generation for Mathematical Reasoning
por: Lai, Yuhang, et al.
Publicado: (2026)

The Edge-of-Reach Problem in Offline Model-Based Reinforcement Learning
por: Sims, Anya, et al.
Publicado: (2024)

L3Ms -- Lagrange Large Language Models
por: Dhillon, Guneet S., et al.
Publicado: (2024)

Incorporating Unlabelled Data into Bayesian Neural Networks
por: Sharma, Mrinank, et al.
Publicado: (2023)

SymDiff: Equivariant Diffusion via Stochastic Symmetrisation
por: Zhang, Leo, et al.
Publicado: (2024)

Enhancing Large Language Model Reasoning with Reward Models: An Analytical Survey
por: Liu, Qiyuan, et al.
Publicado: (2025)

Foundation of Intelligence: Review of Math Word Problems from Human Cognition Perspective
por: Huang, Zhenya, et al.
Publicado: (2025)

Manifold Aware Denoising Score Matching (MAD)
por: Levy-Jurgenson, Alona, et al.
Publicado: (2026)

Amortized Probabilistic Detection of Communities in Graphs
por: Wang, Yueqi, et al.
Publicado: (2020)

BetaEdit: Null-Space Constrained Sequential Model Editing
por: Liu, Bingqing, et al.
Publicado: (2026)

Rao-Blackwellised Reparameterisation Gradients
por: Lam, Kevin H., et al.
Publicado: (2025)

Adversarial Attack for Explanation Robustness of Rationalization Models
por: Zhang, Yuankai, et al.
Publicado: (2024)

Learning to Contextualize Web Pages for Enhanced Decision Making by LLM Agents
por: Lee, Dongjun, et al.
Publicado: (2025)

Metropolis-Adjusted Diffusion Models
por: Lam, Kevin H., et al.
Publicado: (2026)

Selective Safety Steering via Value-Filtered Decoding
por: Einbinder, Bat-Sheva, et al.
Publicado: (2026)

Meta-Learning Objectives for Preference Optimization
por: Alfano, Carlo, et al.
Publicado: (2024)

EvIL: Evolution Strategies for Generalisable Imitation Learning
por: Sapora, Silvia, et al.
Publicado: (2024)

Unleashing the Power of Meta-tuning for Few-shot Generalization Through Sparse Interpolated Experts
por: Chen, Shengzhuang, et al.
Publicado: (2024)

When Is Enough Not Enough? Illusory Completion in Search Agents
por: Ko, Dayoon, et al.
Publicado: (2026)

SigmaDock: Untwisting Molecular Docking With Fragment-Based SE(3) Diffusion
por: Prat, Alvaro, et al.
Publicado: (2025)

Meta Flow Maps enable scalable reward alignment
por: Potaptchik, Peter, et al.
Publicado: (2026)

Prompting Strategies for Enabling Large Language Models to Infer Causation from Correlation
por: Sgouritsa, Eleni, et al.
Publicado: (2024)

Aortobronchial Fistula: A Case Report and Literature Review
por: Zhenghan Liu, et al.
Publicado: (2025)

Information Science: A House Built on Sand
por: Vagianos, Louis
Publicado: (1972)

Online Adaptation of Language Models with a Memory of Amortized Contexts
por: Tack, Jihoon, et al.
Publicado: (2024)

FedGIG: Graph Inversion from Gradient in Federated Learning
por: Xiao, Tianzhe, et al.
Publicado: (2024)

Variational Flow Maps: Make Some Noise for One-Step Conditional Generation
por: Mammadov, Abbas, et al.
Publicado: (2026)

Kalman Filter for Online Classification of Non-Stationary Data
por: Titsias, Michalis K., et al.
Publicado: (2023)

Context-Guided Diffusion for Out-of-Distribution Molecular and Protein Design
por: Klarner, Leo, et al.
Publicado: (2024)

The Illusory Normativity of Rights-Based AI Regulation
por: Mei, Yiyang, et al.
Publicado: (2025)