Elamur Research⌕
☰
home / papers / 2025 / CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios

CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios

Paper · 2025 · Shiting Huang, Zhen Fang, Zehui Chen et al.

Summary

LLM agents that call external tools often run into problems during long, multi-step tasks. Existing tool-use benchmarks mostly compare final outputs with gold answers, so they do not show whether a model can notice a mistake, diagnose it and recover. The authors first examined failures on several established tool benchmarks and found that strong models rarely recover once an error occurs. They then built CRITICTOOL, which they describe as the first benchmark aimed specifically at self-critique in tool calling. The data comes from BFCL v3 and T-Eval. GPT-based simulators and repeated real API calls inject errors, which fall into two families: internal (model-driven) and external (environment). An evolutionary step then makes the scenarios harder. Models are scored on reflecting and correcting for internal errors, and on retrying, skipping or finishing for external ones. Results show that models differ in how they handle errors from each source. Self-critique scores also broadly track general tool-use accuracy on BFCL, which suggests that better error handling could improve overall agent performance.

Key points

  • In a preliminary study on NESTFUL, API-Bank, T-Eval and BFCL, models such as Qwen-turbo, Qwen2.5-72B, GPT-3.5 and GPT-4o recovered from errors rarely. GPT-4o's success ranged from about 17% (API-Bank) to about 54% (T-Eval).
  • The benchmark defines five error patterns:
    • tool selection
    • tool hallucination
    • parameter key
    • parameter value
    • environment errors (e.g., connection resets, timeouts)
  • Construction follows four phases: data collection, error diversification, tool response handling (via API cache, real execution or an API simulator), and data evolution. The evolution strategies add long context, extra functions, noisy queries and harder functions, followed by verification and human review.
  • The source data covers 203 real-world APIs across 23 tools. Combining evolution strategies degrades performance more than applying any single one.
  • Stated limitations include dependence on the quality of BFCL and T-Eval, possible GPT-4o bias in the synthetic data, and high construction costs.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Shiting Huang, Zhen Fang, Zehui Chen, Siyu Yuan, Junjie Ye, Yu Zeng, Lin Chen, Qi Mao and Feng Zhao. 2025. CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.emnlp-main.1355

@inproceedings{huang2025critictool,
  title = {{CRITICTOOL: Evaluating Self-Critique Capabilities of Large Language Models in Tool-Calling Error Scenarios}},
  author = {Shiting Huang, Zhen Fang, Zehui Chen, Siyu Yuan, Junjie Ye, Yu Zeng, Lin Chen, Qi Mao and Feng Zhao},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  year = {2025},
  doi = {10.18653/v1/2025.emnlp-main.1355},
  url = {https://doi.org/10.18653/v1/2025.emnlp-main.1355}
}

References

52 references, 23 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ibrahim Abdelaziz, Kinjal Basu, Mayank Agarwal, Sadhana Kumaravel, Matthew Stallone, Rameswar Panda, Yara Rizk, GP Bhargav, Maxwell Crouse, Chulaka Gunasekara, et al. 2024. Granite-function calling model: Introducing function calling abilities via multi-task learning of granular tasks. In Proceeding… find
  2. Mistral AI. 2024. Un ministral, des ministraux. find
  3. AI@Meta. 2024. Llama 3 model card. find
  4. Shengnan An, Zexiong Ma, Zeqi Lin, Nanning Zheng, Jian-Guang Lou, and Weizhu Chen. 2023. Learning from mistakes makes llm better reasoner. arXiv preprint arXiv:2310.20689. find
  5. Anthropic. 2024. Claude 3.5 sonnet. find
  6. Yushi Bai, Xin Lv, Jiajie Zhang, Hongchang Lyu, Jiankai Tang, Zhidian Huang, Zhengxiao Du, Xiao Liu, Aohan Zeng, Lei Hou, et al. 2023. Longbench: A bilingual, multitask benchmark for long context understanding. arXiv preprint arXiv:2308.14508.
  7. Kinjal Basu, Ibrahim Abdelaziz, Kelsey Bradford, Maxwell Crouse, Kiran Kate, Sadhana Kumaravel, Saurabh Goyal, Asim Munawar, Yara Rizk, Xin Wang, et al. 2024. Nestful: A benchmark for evaluating llms on nested sequences of api calls. arXiv preprint arXiv:2409.03797.
  8. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. In Advances in Neural Information Processing Systems, volume 33, pages 1877–1901.
  9. Lin Chen, Jisong Li, Xiaoyi Dong, Pan Zhang, Conghui He, Jiaqi Wang, Feng Zhao, and Dahua Lin. 2023. Sharegpt4v: Improving large multimodal models with better captions. arXiv preprint arXiv:2311.12793.
  10. Sijia Chen, Yibo Wang, Yi-Feng Wu, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang, and Lijun Zhang. 2024a. Advancing tool-augmented large language models: Integrating insights from errors in inference trees. arXiv preprint arXiv:2406.07115. — [arXiv:2406.07115](https://arxiv.org/abs/2406.07115) find
  11. Zehui Chen, Weihua Du, Wenwei Zhang, Kuikun Liu, Jiangning Liu, Miao Zheng, Jingming Zhuo, Songyang Zhang, Dahua Lin, Kai Chen, and Feng Zhao. 2024b. T-eval: Evaluating the tool utilization capability of large language models step by step. In Proceedings of the 62nd Annual Meeting of the Association… find
  12. Zehui Chen, Kuikun Liu, Qiuchen Wang, Wenwei Zhang, Jiangning Liu, Dahua Lin, Kai Chen, and Feng Zhao. 2024c. Agent-FLAN: Designing data and methods of effective agent tuning for large language models. In Findings of the Association for Computational Linguistics, pages 9354–9366.
  13. Team GLM, Aohan Zeng, Bin Xu, Bowen Wang, Chenhui Zhang, Da Yin, Dan Zhang, Diego Rojas, Guanyu Feng, Hanlin Zhao, et al. 2024. Chatglm: A family of large language models from glm-130b to glm-4 all tools. arXiv preprint arXiv:2406.12793.
  14. Zhibin Gou, Zhihong Shao, Yeyun Gong, Yelong Shen, Yujiu Yang, Nan Duan, and Weizhu Chen. 2023. Critic: Large language models can self-correct with tool-interactive critiquing. arXiv preprint arXiv:2305.11738.
  15. Zhicheng Guo, Sijie Cheng, Hao Wang, Shihao Liang, Yujia Qin, Peng Li, Zhiyuan Liu, Maosong Sun, and Yang Liu. 2024a. StableToolBench: Towards stable large-scale benchmarking on tool learning of large language models. In Findings of the Association for Computational Linguistics, pages 11143–11156. find
  16. Zishan Guo, Yufei Huang, and Deyi Xiong. 2024b. CToolEval: A Chinese benchmark for LLM-powered agent evaluation in real-world API interactions. In Findings of the Association for Computational Linguistics, pages 15711–15724. find
  17. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  18. Juyong Jiang, Fan Wang, Jiasi Shen, Sungju Kim, and Sunghun Kim. 2024. A survey on large language models for code generation. arXiv preprint arXiv:2406.00515.
  19. Pei Ke, Bosi Wen, Andrew Feng, Xiao Liu, Xuanyu Lei, Jiale Cheng, Shengyuan Wang, Aohan Zeng, Yuxiao Dong, Hongning Wang, Jie Tang, and Minlie Huang. 2024. CritiqueLLM: Towards an informative critique generation model for evaluation of large language model generation. In Proceedings of the 62nd Annu…
  20. Shirley Kokane, Ming Zhu, Tulika Awalgaonkar, Jianguo Zhang, Thai Hoang, Akshara Prabhakar, Zuxin Liu, Tian Lan, Liangwei Yang, Juntao Tan, et al. 2024. Spectool: A benchmark for characterizing errors in tool-use llms. arXiv preprint arXiv:2411.13547. — [arXiv:2411.13547](https://arxiv.org/abs/2411.13547) find
  21. Yilun Kong, Jingqing Ruan, YiHong Chen, Bin Zhang, Tianpeng Bao, Shi Shiwei, du Guo Qing, Xiaoru Hu, Hangyu Mao, Ziyue Li, Xingyu Zeng, Rui Zhao, and Xueqian Wang. 2024. TPTU-v2: Boosting task planning and tool usage of large language model-based agents in real-world industry systems. In Proceedings… find
  22. Tian Lan, Wenwei Zhang, Chen Xu, Heyan Huang, Dahua Lin, Kai Chen, and Xian-ling Mao. 2024. Criticeval: Evaluating large language models as critic. arXiv preprint arXiv:2402.13764. — [arXiv:2402.13764](https://arxiv.org/abs/2402.13764) find
  23. Minghao Li, Yingxiu Zhao, Bowen Yu, Feifan Song, Hangyu Li, Haiyang Yu, Zhoujun Li, Fei Huang, and Yongbin Li. 2023. API-bank: A comprehensive benchmark for tool-augmented LLMs. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 3102–3116.
  24. Zicheng Lin, Zhibin Gou, Tian Liang, Ruilin Luo, Haowei Liu, and Yujiu Yang. 2024. CriticBench: Benchmarking LLMs for critique-correct reasoning. In Findings of the Association for Computational Linguistics, pages 1552–1587. find
  25. Weiwen Liu, Xu Huang, Xingshan Zeng, Xinlong Hao, Shuai Yu, Dexun Li, Shuai Wang, Weinan Gan, Zhengying Liu, Yuanqing Yu, et al. 2024. Toolace: Winning the points of llm function calling. arXiv preprint arXiv:2409.00920. find
  26. Nat McAleese, Rai Michael Pokorny, Juan Felipe Ceron Uribe, Evgenia Nitishinskaya, Maja Trebacz, and Jan Leike. 2024. Llm critics help catch llm bugs. arXiv preprint arXiv:2407.00215. — [arXiv:2407.00215](https://arxiv.org/abs/2407.00215) find
  27. OpenAI. 2022. Introducing chatgpt. find
  28. OpenAI. 2024. Introducing openai o1. find
  29. Shishir G. Patil, Tianjun Zhang, Xin Wang, and Joseph E. Gonzalez. 2024. Gorilla: Large language model connected with massive apis. In Advances in Neural Information Processing Systems, volume 37, pages 126544–126565.
  30. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, et al. 2023. Toolllm: Facilitating large language models to master 16000+ real-world apis. arXiv preprint arXiv:2307.16789.
  31. Changle Qu, Sunhao Dai, Xiaochi Wei, Hengyi Cai, Shuaiqiang Wang, Dawei Yin, Jun Xu, and Ji-Rong Wen. 2024. From exploration to mastery: Enabling llms to master tools via self-driven interactions. arXiv preprint arXiv:2410.08197. — [arXiv:2410.08197](https://arxiv.org/abs/2410.08197) find
  32. Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. arXiv preprint arXiv:1908.10084.
  33. Yongliang Shen, Kaitao Song, Xu Tan, Wenqi Zhang, Kan Ren, Siyu Yuan, Weiming Lu, Dongsheng Li, and Yueting Zhuang. 2023. Taskbench: Benchmarking large language models for task automation. arXiv preprint arXiv:2311.18760. — [arXiv:2311.18760](https://arxiv.org/abs/2311.18760) find
  34. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. In Advances in Neural Information Processing Systems, volume 36, pages 8634–8652.
  35. Yifan Song, Da Yin, Xiang Yue, Jie Huang, Sujian Li, and Bill Yuchen Lin. 2024. Trial and error: Exploration-based trajectory optimization of LLM agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, pages 7584–7600.
  36. Jimin Sun, So Yeon Min, Yingshan Chang, and Yonatan Bisk. 2024. Tools fail: Detecting silent errors in faulty tools. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 14272–14289. find
  37. Qwen Team. 2024a. Qwen2 technical report. arXiv preprint arXiv:2407.10671. find
  38. Qwen Team. 2024b. Qwen2.5: A party of foundation models. find
  39. Ye Tian, Baolin Peng, Linfeng Song, Lifeng Jin, Dian Yu, Haitao Mi, and Dong Yu. 2024. Toward self-improvement of llms via imagination, searching, and criticizing. arXiv preprint arXiv:2404.12253. find
  40. Boshi Wang, Hao Fang, Jason Eisner, Benjamin Van Durme, and Yu Su. 2024a. LLMs in the imaginarium: Tool learning through simulated trial and error. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, pages 10583–10604. find
  41. Siyuan Wang, Zhuohan Long, Zhihao Fan, Zhongyu Wei, and Xuanjing Huang. 2024b. Benchmark self-evolving: A multi-agent framework for dynamic llm evaluation. arXiv preprint arXiv:2402.11443.
  42. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, brian ichter, Fei Xia, Ed Chi, Quoc V Le, and Denny Zhou. 2022. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems, volume 35, pages 24824–24837.
  43. Qiantong Xu, Fenglu Hong, Bo Li, Changran Hu, Zhengyu Chen, and Jian Zhang. 2023. On the tool manipulation capability of open-source large language models. arXiv preprint arXiv:2305.16504.
  44. Fanjia Yan, Huanzhi Mao, Charlie Cheng-Jie Ji, Tianjun Zhang, Shishir G. Patil, Ion Stoica, and Joseph E. Gonzalez. 2024. Berkeley function calling leaderboard. find
  45. Jihan Yao, Wenxuan Ding, Shangbin Feng, Lucy Lu Wang, and Yulia Tsvetkov. 2024. Varying shades of wrong: Aligning llms with wrong answers only. arXiv preprint arXiv:2410.11055. — [arXiv:2410.11055](https://arxiv.org/abs/2410.11055) find
  46. Junjie Ye, Guanyu Li, Songyang Gao, Caishuang Huang, Yilong Wu, Sixian Li, Xiaoran Fan, Shihan Dou, Qi Zhang, Tao Gui, et al. 2024a. Tooleyes: Fine-grained evaluation for tool learning capabilities of large language models in real-world scenarios. arXiv preprint arXiv:2401.00741.
  47. Junjie Ye, Yilong Wu, Songyang Gao, Caishuang Huang, Sixian Li, Guanyu Li, Xiaoran Fan, Qi Zhang, Tao Gui, and Xuanjing Huang. 2024b. RoTBench: A multi-level benchmark for evaluating the robustness of large language models in tool learning. In Proceedings of the 2024 Conference on Empirical Methods… find
  48. Jiahao Ying, Mingbao Lin, Yixin Cao, Wei Tang, Bo Wang, Qianru Sun, Xuanjing Huang, and Shuicheng Yan. 2024. LLMs-as-instructors: Learning from errors toward automating model improvement. In Findings of the Association for Computational Linguistics, pages 11185–11208. find
  49. Siyu Yuan, Kaitao Song, Jiangjie Chen, Xu Tan, Yongliang Shen, Ren Kan, Dongsheng Li, and Deqing Yang. 2024. Easytool: Enhancing llm-based agents with concise tool instruction. arXiv preprint arXiv:2401.06201. find
  50. Aohan Zeng, Mingdao Liu, Rui Lu, Bowen Wang, Xiao Liu, Yuxiao Dong, and Jie Tang. 2023. Agenttuning: Enabling generalized agent abilities for llms. arXiv preprint arXiv:2310.12823.
  51. Wenqi Zhang, Yongliang Shen, Linjuan Wu, Qiuying Peng, Jun Wang, Yueting Zhuang, and Weiming Lu. 2024. Self-contrast: Better reflection through inconsistent solving perspectives. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, pages 3602–3622.
  52. Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, Peiyu Liu, Jian-Yun Nie, and Ji-Rong Wen. 2023. A survey of l…

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/huang-2025-critictool-evaluating-self-critique-capabilities/