Imagen de Portada

The Tool-Overuse Illusion: Why Does LLM Prefer External Tools over Internal Knowledge?

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Zeng, Yirong, You, Shen, Liu, Yufei, Du, Qunyao, Ding, Xiao, Hou, Yutai, Wang, Yuxian, Ning, Wu, Song, Haonan, Tu, Dandan, Cai, Bibo, Liu, Ting
Formato:	Preprint
Publicado:	2026
Materias:	Artificial Intelligence Software Engineering
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

AutoTool: Automatic Scaling of Tool-Use Capabilities in RL via Decoupled Entropy Constraints
por: Zeng, Yirong, et al.
Publicado: (2026)

Precision over Diversity: High-Precision Reward Generalizes to Robust Instruction Following
por: Zeng, Yirong, et al.
Publicado: (2026)

iTool: Reinforced Fine-Tuning with Dynamic Deficiency Calibration for Advanced Tool Use
por: Zeng, Yirong, et al.
Publicado: (2025)

Tool Zero: Training Tool-Augmented LLMs via Pure RL from Scratch
por: Zeng, Yirong, et al.
Publicado: (2025)

DeepTool: Scaling Interleaved Deliberation in Tool-Integrated Reasoning via Process-Supervised Reinforcement Learning
por: He, Yang, et al.
Publicado: (2026)

ToolACE-DEV: Self-Improving Tool Learning via Decomposition and EVolution
por: Huang, Xu, et al.
Publicado: (2025)

SMART: Self-Aware Agent for Tool Overuse Mitigation
por: Qian, Cheng, et al.
Publicado: (2025)

ToolVQA: A Dataset for Multi-step Reasoning VQA with External Tools
por: Yin, Shaofeng, et al.
Publicado: (2025)

ToolACE: Winning the Points of LLM Function Calling
por: Liu, Weiwen, et al.
Publicado: (2024)

Concise and Precise Context Compression for Tool-Using Language Models
por: Xu, Yang, et al.
Publicado: (2024)

The Tool Illusion: Rethinking Tool Use in Web Agents
por: Lou, Renze, et al.
Publicado: (2026)

Breaking the Illusion of Identity in LLM Tooling
por: Miller, Marek
Publicado: (2026)

The Evolution of Tool Use in LLM Agents: From Single-Tool Call to Multi-Tool Orchestration
por: Xu, Haoyuan, et al.
Publicado: (2026)

ToolBridge: An Open-Source Dataset to Equip LLMs with External Tool Capabilities
por: Jin, Zhenchao, et al.
Publicado: (2024)

Transparentize the Internal and External Knowledge Utilization in LLMs with Trustworthy Citation
por: Shen, Jiajun, et al.
Publicado: (2025)

Planning, Creation, Usage: Benchmarking LLMs for Comprehensive Tool Utilization in Real-World Complex Scenarios
por: Huang, Shijue, et al.
Publicado: (2024)

Relative Attention-based One-Class Adversarial Autoencoder for Continuous Authentication of Smartphone Users
por: Hu, Mingming, et al.
Publicado: (2022)

Tool-as-Interface: Learning Robot Policies from Observing Human Tool Use
por: Chen, Haonan, et al.
Publicado: (2025)

RefTool: Reference-Guided Tool Creation for Knowledge-Intensive Reasoning
por: Liu, Xiao, et al.
Publicado: (2025)

Do We Really Need External Tools to Mitigate Hallucinations? SIRA: Shared-Prefix Internal Reconstruction of Attribution
por: Qin, Tian, et al.
Publicado: (2026)

Internalizing Tools as Morphisms in Graded Transformers
por: Shaska, Tony
Publicado: (2025)

Reinforced Internal-External Knowledge Synergistic Reasoning for Efficient Adaptive Search Agent
por: Huang, Ziyang, et al.
Publicado: (2025)

Tool-RoCo: An Agent-as-Tool Self-organization Large Language Model Benchmark in Multi-robot Cooperation
por: Zhang, Ke, et al.
Publicado: (2025)

ToolGen: Unified Tool Retrieval and Calling via Generation
por: Wang, Renxi, et al.
Publicado: (2024)

Straggler-Resilient Federated Learning over A Hybrid Conventional and Pinching Antenna Network
por: Wu, Bibo, et al.
Publicado: (2025)

Direct Preference Knowledge Distillation for Large Language Models
por: Li, Yixing, et al.
Publicado: (2024)

Thinking Isn't an Illusion: Overcoming the Limitations of Reasoning Models via Tool Augmentations
por: Song, Zhao, et al.
Publicado: (2025)

Internalizing Tool Knowledge in Small Language Models via QLoRA Fine-Tuning
por: Shemla, Yuval, et al.
Publicado: (2026)

TInR: Exploring Tool-Internalized Reasoning in Large Language Models
por: Xu, Qiancheng, et al.
Publicado: (2026)

Why a Trade-Off? The Relationship between the External and Internal Validity of Experiments
por: María JIMÉNEZ-BUEDO
Publicado: (2010)

Tool Preferences in Agentic LLMs are Unreliable
por: Faghih, Kazem, et al.
Publicado: (2025)

PRISM: Preference Refinement via Implicit Scene Modeling for 3D Vision-Language Preference-Based Reinforcement Learning
por: Sun, Yirong, et al.
Publicado: (2025)

Evaluating the External and Parametric Knowledge Fusion of Large Language Models
por: Zhang, Hao, et al.
Publicado: (2024)

ToolRM: Towards Agentic Tool-Use Reward Modeling
por: Li, Renhao, et al.
Publicado: (2025)

Seeing the Evidence, Missing the Answer: Tool-Guided Vision-Language Models on Visual Illusions
por: Wang, Xuesong, et al.
Publicado: (2026)

WeMusic-Agent: Efficient Conversational Music Recommendation via Knowledge Internalization and Agentic Boundary Learning
por: Bi, Wendong, et al.
Publicado: (2025)

Internal and External Knowledge Interactive Refinement Framework for Knowledge-Intensive Question Answering
por: Du, Haowei, et al.
Publicado: (2024)

Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction
por: Guo, Jiahe, et al.
Publicado: (2026)

What Does Vision Tool-Use Reinforcement Learning Really Learn? Disentangling Tool-Induced and Intrinsic Effects for Crop-and-Zoom
por: Ma, Yan, et al.
Publicado: (2026)

CoVe: Training Interactive Tool-Use Agents via Constraint-Guided Verification
por: Chen, Jinpeng, et al.
Publicado: (2026)