mlsanityData to 7 Oct 2026

SMART: Self-Aware Agent for Tool Overuse Mitigation

arXiv 2502.11435 ↗First posted 17 Feb 2025cs.AI cs.CL cs.LGcited by 52

On the map: LLM reasoning → Reasoning for agents, planning and code

also in 1 more
Abstract (arXiv)

Current Large Language Model (LLM) agents demonstrate strong reasoning and tool use capabilities, but often lack self-awareness, failing to balance these approaches effectively. This imbalance leads to Tool Overuse, where models unnecessarily rely on external tools for tasks solvable with parametric knowledge, increasing computational overhead. Inspired by human metacognition, we introduce SMART (Strategic Model-Aware Reasoning with Tools), a paradigm that enhances an agent's self-awareness to optimize task handling and reduce tool overuse. To support this paradigm, we introduce SMART-ER, a dataset spanning three domains, where reasoning alternates between parametric knowledge and tool-dependent steps, with each step enriched by rationales explaining when tools are necessary. Through supervised training, we develop SMARTAgent, a family of models that dynamically balance parametric knowledge and tool use. Evaluations show that SMARTAgent reduces tool use by 24% while improving performance by over 37%, enabling 7B-scale models to match its 70B counterpart and GPT-4o. Additionally, SMARTAgent generalizes to out-of-distribution test data like GSM8K and MINTQA, maintaining accuracy with just one-fifth the tool calls. These highlight the potential of strategic tool use to enhance reasoning, mitigate overuse, and bridge the gap between model size and performance, advancing intelligent and resource-efficient agent designs.

Key facts. Cited by 52 papers in the map (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 34 of its 54 references that are in the map. First posted on arXiv 17 Feb 2025.

Builds on

34 of 54 references.

References · 34

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  4. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
  7. 2023ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIscited by 1,880
  8. 2022Language Models (Mostly) Know What They Knowcited by 1,783
  9. 2023The Rise and Potential of Large Language Model Based Agents: A Surveycited by 1,422
  10. 2021Generalizing to Unseen Domains: A Survey on Domain Generalizationcited by 762
  11. 2023FreshLLMs: Refreshing Large Language Models with Search Engine Augmentationcited by 308
  12. 2023Large Language Models as Tool Makerscited by 293
22 more
  1. 2024Tool Learning with Large Language Models: A Surveycited by 221
  2. 2023MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Usecited by 209
  3. 2023Investigating the Factual Knowledge Boundary of Large Language Models with Retrieval Augmentationcited by 146
  4. 2023Domain Specialization as the Key to Make Large Language Models Disruptive: A Comprehensive Surveycited by 134
  5. 2022A Survey of Deep Learning for Mathematical Reasoningcited by 124
  6. 2023CRAFT: Customizing LLMs by Creating and Retrieving from Specialized Toolsetscited by 122
  7. 2024Tell Me More! Towards Implicit User Intention Understanding of Language Model Driven Agentscited by 71
  8. 2024Aligning LLMs with Individual Preferences via Interactioncited by 63
  9. 2024Benchmarking Knowledge Boundary for Large Language Models: A Different Perspective on Model Evaluationcited by 37
  10. 2024Knowledge Boundary of Large Language Models: A Surveycited by 36
  11. 2024Teaching Large Language Models to Express Knowledge Boundary from Their Own Signalscited by 32
  12. 2024Self-DC: When to Reason and When to Act? Self Divide-and-Conquer for Compositional Unknown Questionscited by 30
  13. 2023Toolink: Linking Toolkit Creation and Using through Chain-of-Solving on Open-Source Modelcited by 29
  14. 2024Investigate-Consolidate-Exploit: A General Strategy for Inter-Task Agent Self-Evolutioncited by 28
  15. 2024AppBench: Planning of Multiple APIs from Various APPs for Complex User Instructioncited by 24
  16. 2024MINTQA: A Multi-Hop Question Answering Benchmark for Evaluating LLMs on New and Tail Knowledgecited by 20
  17. 2024SMARTCAL: An Approach to Self-Aware Tool-Use Evaluation and Calibrationcited by 13
  18. 2023MinPrompt: Graph-based Minimal Prompt Data Augmentation for Few-shot Question Answeringcited by 12
  19. 2023"Merge Conflicts!" Exploring the Impacts of External Distractors to Parametric Knowledge Graphscited by 11
  20. 2023Information Association for Language Model Updating by Mitigating LM-Logical Discrepancycited by 10
  21. 2023Gotta: Generative Few-shot Question Answering by Prompt-based Cloze Data Augmentationcited by 5
  22. 2024UniRetriever: Multi-task Candidates Selection for Various Context-Adaptive Conversational Retrievalcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 25; 2026: 27.

52 papers.

Sort:

By area:

  1. 2025ToolRL: Reward is All Tool Learning Needscited by 342
  2. 2025Agentic Reinforced Policy Optimizationcited by 166
  3. 2025Acting Less is Reasoning More! Teaching Model to Act Efficientlycited by 80
  4. 2025UserBench: An Interactive Gym Environment for User-Centric Agentscited by 69
  5. 2025In-the-Flow Agentic System Optimization for Effective Planning and Tool Usecited by 67
  6. 2025Open Challenges in Multi-Agent Security: Towards Secure Systems of Interacting AI Agentscited by 66
  7. 2025TRAJECT-Bench:A Trajectory-Aware Benchmark for Evaluating Agentic Tool Usecited by 46
  8. 2025ToolOrchestra: Elevating Intelligence via Efficient Model and Tool Orchestrationcited by 44
  9. 2025CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agentscited by 32
  10. 2025UserRL: Training Interactive User-Centric Agent via Reinforcement Learningcited by 27
  11. 2025When Agents go Astray: Course-Correcting SWE Agents with PRMscited by 25
  12. 2025MedAgentGym: A Scalable Agentic Training Environment for Code-Centric Reasoning in Biomedical Data Sciencecited by 22
38 more
  1. 2025Toward Effective Tool-Integrated Reasoning via Self-Evolved Preference Learningcited by 17
  2. 2026A Comprehensive Survey on Agent Skills: Taxonomy, Techniques, and Applicationscited by 17
  3. 2025Search Wisely: Mitigating Sub-optimal Agentic Searches By Reducing Uncertaintycited by 16
  4. 2025ToolRM: Outcome Reward Models for Tool-Calling Large Language Modelscited by 16
  5. 2025xRouter: Training Cost-Aware LLMs Orchestration System via Reinforcement Learningcited by 16
  6. 2025Stop Wasting Your Tokens: Towards Efficient Runtime Multi-Agent Systemscited by 11
  7. 2026Toward Personalized LLM-Powered Agents: Foundations, Evaluation, and Future Directionscited by 10
  8. 2025A Desideratum for Conversational Agents: Capabilities, Challenges, and Future Directionscited by 9
  9. 2025HiPRAG: Hierarchical Process Rewards for Efficient Agentic Retrieval Augmented Generationcited by 9
  10. 2026Agentic Abstention: Do Agents Know When to Stop Instead of Act?cited by 9
  11. 2025Agentic Knowledgeable Self-awarenesscited by 8
  12. 2026ToolFailBench: Diagnosing Tool-Use Failures in LLM Agentscited by 8
  13. 2025ToolSpectrum : Towards Personalized Tool Utilization for Large Language Modelscited by 6
  14. 2026The Tool-Overuse Illusion: Why Does LLM Prefer External Tools over Internal Knowledge?cited by 6
  15. 2025AdaSearch: Balancing Parametric Knowledge and Search in Large Language Models via Reinforcement Learningcited by 5
  16. 2026Talk, Evaluate, Diagnose: User-aware Agent Evaluation with Automated Error Analysiscited by 5
  17. 2026Model-Adaptive Tool Necessity Reveals the Knowing-Doing Gap in LLM Tool Usecited by 5
  18. 2026Hallucinations Undermine Trust; Metacognition is a Way Forwardcited by 4
  19. 2026Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrievalcited by 4
  20. 2026Beyond Accuracy: Unveiling Inefficiency Patterns in Tool-Integrated Reasoningcited by 3
  21. 2026APPO: Agentic Procedural Policy Optimizationcited by 3
  22. 2025Agentic Metacognition: Designing a "Self-Aware" Low-Code Agent for Failure Prediction and Human Handoffcited by 2
  23. 2025Beyond Fast and Slow: Cognitive-Inspired Elastic Reasoning for Large Language Modelscited by 2
  24. 2026Controlling Tool Use with Heading-Specific Activation Steeringcited by 2
  25. 2026When Top-K Misses the Decision: Tool-Call Drift in Multi-Teacher On-Policy Distillationcited by 2
  26. 2026Scores Are Not Decisions: Cost-Aware Stopping for Tool Acquisition in LLM Agentscited by 2
  27. 2026From Passive Metric to Active Signal: The Evolving Role of Uncertainty Quantification in Large Language Modelscited by 1
  28. 2026The Workload-Router-Pool Architecture for LLM Inference Optimization: A Vision Paper from the vLLM Semantic Router Projectcited by 1
  29. 2026When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math Reasoningcited by 1
  30. 2026Efficient Agentic Reinforcement Learning with On-Policy Intrinsic Knowledge Boundary Enhancementcited by 1
  31. 2026Tunable Tool-Call Rates in LLM Agents via Representation Steeringcited by 1
  32. 2025LightSearcher: Efficient DeepSearch via Experiential Memorycited by 0
  33. 2026Know When to Trust the Skill: Delayed Appraisal and Epistemic Vigilance for Single-Agent LLMscited by 0
  34. 2026Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interactioncited by 0
  35. 2026Tool-Adaptive LLM Rerankercited by 0
  36. 2026Self-Aware Recursively Self-Improving Agents for Personal Singularity: A Goal-, Scope-, Tool-, and Benchmark-Driven Multi-Agent Architecturecited by 0
  37. 2026CRISP: Critical Step Perception for Training Efficient Deep Search Agentscited by 0
  38. 2026MetaRAG: Belief-Action Aligned Policy Optimization for Agentic RAGcited by 0

50 most cited of 52 All 52 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025MetaAgent: Toward Self-Evolving Agent via Tool Meta-Learning
  2. 2026ToolAtlas: Learning Once, Reusing Everywhere with Tool-Side Memory
  3. 2024SMART: Self-learning Meta-strategy Agent for Reasoning Tasks
  4. 2025ToolMind Technical Report: A Large-Scale, Reasoning-Enhanced Tool-Use Dataset
  5. 2024Learning to Use Tools via Cooperative and Interactive Agents
  6. 2026Internalizing Tool Knowledge in Small Language Models via QLoRA Fine-Tuning
  7. 2026HyperAgent: Planning and Acting over Tool-Schema Hypergraphs for Tool-Use LLM Agents
  8. 2025From Language to Action: A Review of Large Language Models as Autonomous Agents and Tool Users