Composition Collapse: Stable Factual Knowledge Does Not Imply Compositional Reasoning

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Yu, Zhe, Xing, Wenpeng, Wei, Yunzhao, Chen, Jie, Wang, Hongzhi, Teng, Xuyang, Han, Meng
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Artificial Intelligence
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Cordon-MAS: Defending RAG against Knowledge Poisoning via Information-Flow Control
di: Yu, Zhe, et al.
Pubblicazione: (2026)

The Attribution Blind Spot: Detecting When Language Models Rely on Memory Rather Than Retrieved Context
di: Yu, Zhe, et al.
Pubblicazione: (2026)

From Retinal Evidence to Safe Decisions: RETINA-SAFE and ECRT for Hallucination Risk Triage in Medical LLMs
di: Yu, Zhe, et al.
Pubblicazione: (2026)

LatentAudit: Real-Time White-Box Faithfulness Monitoring for Retrieval-Augmented Generation with Verifiable Deployment
di: Yu, Zhe, et al.
Pubblicazione: (2026)

XDomainBench: Diagnosing Reasoning Collapse in High-Dimensional Scientific Knowledge Composition
di: Zhiren, Gong, et al.
Pubblicazione: (2026)

KnowledgeBerg: Evaluating Systematic Knowledge Coverage and Compositional Reasoning in Large Language Models
di: Zhang, Xiao, et al.
Pubblicazione: (2026)

HGMF: A Hierarchical Gaussian Mixture Framework for Scalable Tool Invocation within the Model Context Protocol
di: Xing, Wenpeng, et al.
Pubblicazione: (2025)

Silencing the Guardrails: Inference-Time Jailbreaking via Dynamic Contextual Representation Ablation
di: Xing, Wenpeng, et al.
Pubblicazione: (2026)

A Implies B: Circuit Analysis in LLMs for Propositional Logical Reasoning
di: Hong, Guan Zhe, et al.
Pubblicazione: (2024)

From Confidence to Collapse in LLM Factual Robustness
di: Fastowski, Alina, et al.
Pubblicazione: (2025)

Towards Robust and Secure Embodied AI: A Survey on Vulnerabilities and Attacks
di: Xing, Wenpeng, et al.
Pubblicazione: (2025)

Knowledge Authoring with Factual English, Rules, and Actions
di: Wang, Yuheng
Pubblicazione: (2024)

Reasoning Factual Knowledge in Structured Data with Large Language Models
di: Huang, Sirui, et al.
Pubblicazione: (2024)

DIAP: A Decentralized Agent Identity Protocol with Zero-Knowledge Proofs and a Hybrid P2P Stack
di: Liu, Yuanjie, et al.
Pubblicazione: (2025)

Does Reasoning Introduce Bias? A Study of Social Bias Evaluation and Mitigation in LLM Reasoning
di: Wu, Xuyang, et al.
Pubblicazione: (2025)

MedReason: Eliciting Factual Medical Reasoning Steps in LLMs via Knowledge Graphs
di: Wu, Juncheng, et al.
Pubblicazione: (2025)

Neural Probabilistic Circuits: Enabling Compositional and Interpretable Predictions through Logical Reasoning
di: Chen, Weixin, et al.
Pubblicazione: (2025)

CROP: Expert-Aligned Image Cropping via Compositional Reasoning and Optimizing Preference
di: Dong, Zhitong, et al.
Pubblicazione: (2026)

NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models
di: Zhou, Yi, et al.
Pubblicazione: (2025)

TriLens: Per-Layer Logit-Lens Entropy for White-Box Hallucination Detection
di: Yang, Bohan, et al.
Pubblicazione: (2026)

When Does Closeness in Distribution Imply Representational Similarity? An Identifiability Perspective
di: Nielsen, Beatrix M. G., et al.
Pubblicazione: (2025)

LLMs can be Dangerous Reasoners: Analyzing-based Jailbreak Attack on Large Language Models
di: Lin, Shi, et al.
Pubblicazione: (2024)

Compositional Neuro-Symbolic Reasoning
di: Das, Anugyan, et al.
Pubblicazione: (2026)

Empowering Small Language Models with Factual Hallucination-Aware Reasoning for Financial Classification
di: Yuan, Han, et al.
Pubblicazione: (2026)

WorkArena++: Towards Compositional Planning and Reasoning-based Common Knowledge Work Tasks
di: Boisvert, Léo, et al.
Pubblicazione: (2024)

Fine-Grained Self-Endorsement Improves Factuality and Reasoning
di: Wang, Ante, et al.
Pubblicazione: (2024)

Can LLMs Refuse Questions They Do Not Know? Measuring Knowledge-Aware Refusal in Factual Tasks
di: Pan, Wenbo, et al.
Pubblicazione: (2025)

Unveiling Factual Recall Behaviors of Large Language Models through Knowledge Neurons
di: Wang, Yifei, et al.
Pubblicazione: (2024)

Inconsistencies In Consistency Models: Better ODE Solving Does Not Imply Better Samples
di: Vouitsis, Noël, et al.
Pubblicazione: (2024)

Knowledgeable In-Context Tuning: Exploring and Exploiting Factual Knowledge for In-Context Learning
di: Wang, Jianing, et al.
Pubblicazione: (2023)

MAC-Tuning: LLM Multi-Compositional Problem Reasoning with Enhanced Knowledge Boundary Awareness
di: Huang, Junsheng, et al.
Pubblicazione: (2025)

SproutBench: A Benchmark for Safe and Ethical Large Language Models for Youth
di: Xing, Wenpeng, et al.
Pubblicazione: (2025)

Unveiling the Compositional Ability Gap in Vision-Language Reasoning Model
di: Li, Tianle, et al.
Pubblicazione: (2025)

UW-3DGS: Underwater 3D Reconstruction with Physics-Aware Gaussian Splatting
di: Xing, Wenpeng, et al.
Pubblicazione: (2025)

Cracking Factual Knowledge: A Comprehensive Analysis of Degenerate Knowledge Neurons in Large Language Models
di: Chen, Yuheng, et al.
Pubblicazione: (2024)

ReaRAG: Knowledge-guided Reasoning Enhances Factuality of Large Reasoning Models with Iterative Retrieval Augmented Generation
di: Lee, Zhicheng, et al.
Pubblicazione: (2025)

Local Success Does Not Compose: Benchmarking Large Language Models for Compositional Formal Verification
di: Xu, Xu, et al.
Pubblicazione: (2025)

T2I-FactualBench: Benchmarking the Factuality of Text-to-Image Models with Knowledge-Intensive Concepts
di: Huang, Ziwei, et al.
Pubblicazione: (2024)

MEraser: An Effective Fingerprint Erasure Approach for Large Language Models
di: Zhang, Jingxuan, et al.
Pubblicazione: (2025)

Spectral Logit Sculpting: Adaptive Low-Rank Logit Transformation for Controlled Text Generation
di: Li, Jin, et al.
Pubblicazione: (2025)