Evolutionary Perspectives on the Evaluation of LLM-Based AI Agents: A Comprehensive Survey

R. A. Bradley and M. E. Terry, “Rank analysis of incomplete block designs: I. the method of paired comparisons,” Biometrika , vol. 39, no. 3-4, pp. 324–345, 1952

1952

Earlier work this paper cites.

A. Vaswani, N. Shazeer et al. , “Attention is all you need,” in Advances in Neural Information Processing Systems , 2017, pp. 5998–6008. [Online]. Available: http://arxiv.org/abs/1706.03762

Original

2017

Earlier work this paper cites.

T. Shi, A. Karpathy et al. , “World of bits: An open-domain platform for web-based agents,” in Proceedings of the 34th International Conference on Machine Learning , ser. Proceedings of Machine Learning Research, D. Precup and Y. W. Teh, Eds., vol. 70. PMLR, 06–11 Aug 2017, pp. 3135–3144. [Online]. Available: https://proceedings.mlr.press/v70/shi17a.html

2017

Earlier work this paper cites.

W. Ling, D. Yogatama et al. , “Program induction by rationale generation: Learning to solve and explain algebraic word problems,” arXiv preprint arXiv:1705.04146 , 2017

Original

2017

Earlier work this paper cites.

E. Z. Liu, K. Guu et al. , “Reinforcement learning on web interfaces using workflow-guided exploration,” 2018. [Online]. Available: https://arxiv.org/abs/1802.08802

Original

2018

Earlier work this paper cites.

P. Clark, I. Cowhey et al. , “Think you have solved question answering? try arc, the ai2 reasoning challenge,” arXiv preprint arXiv:1803.05457 , 2018

Original

2018

Earlier work this paper cites.

T. Kočiskỳ, J. Schwarz et al. , “The narrativeqa reading comprehension challenge,” Transactions of the Association for Computational Linguistics , vol. 6, pp. 317–328, 2018

2018

Earlier work this paper cites.

Z. Yang, P. Qi et al. , “Hotpotqa: A dataset for diverse, explainable multi-hop question answering,” arXiv preprint arXiv:1809.09600 , 2018

Original

2018

Earlier work this paper cites.

M.-A. Côté, A. Kádár et al. , “Textworld: A learning environment for text-based games,” in Computer Games: 7th Workshop, CGW 2018, Held in Conjunction with the 27th International Conference on Artificial Intelligence, IJCAI 2018, Stockholm, Sweden, July 13, 2018, Revised Selected Papers 7 . Springer, 2019, pp. 41–75

2019

Earlier work this paper cites.

Y. Li, J. He et al. , “Mapping natural language instructions to mobile ui action sequences,” arXiv preprint arXiv:2005.03776 , 2020

Original

2020

Earlier work this paper cites.

P. Budzianowski, T.-H. Wen et al. , “Multiwoz – a large-scale multi-domain wizard-of-oz dataset for task-oriented dialogue modelling,” 2020. [Online]. Available: https://arxiv.org/abs/1810.00278

Original

2020

Earlier work this paper cites.

M. Chen, J. Tworek et al. , “Evaluating large language models trained on code,” arXiv preprint arXiv:2107.03374 , 2021

Original

2021

Earlier work this paper cites.

A. Burns, D. Arsan et al. , “Mobile app tasks with iterative feedback (motif): Addressing task feasibility in interactive visual environments,” arXiv preprint arXiv:2104.08560 , 2021

Original

2021

Earlier work this paper cites.

P. Dasigi, K. Lo et al. , “A dataset of information-seeking questions and answers anchored in research papers,” arXiv preprint arXiv:2105.03011 , 2021

Original

2021

Earlier work this paper cites.

J. DeYoung, I. Beltagy et al. , “Ms2: Multi-document summarization of medical studies,” arXiv preprint arXiv:2104.06486 , 2021

Original

2021

Earlier work this paper cites.

Y. Li et al. , “Revisiting, benchmarking and exploring api recommendation: How far are we?” arXiv , 2021. [Online]. Available: https://arxiv.org/abs/2112.12653

Original

2021

Earlier work this paper cites.

D. Chen, H. Chen, Y. Yang, A. Lin, and Z. Yu, “Action-based conversations dataset: A corpus for building more in-depth task-oriented dialogue systems,” 2021. [Online]. Available: https://arxiv.org/abs/2104.00783

Original

2021

Earlier work this paper cites.

M. Zhong, D. Yin et al. , “Qmsum: A new benchmark for query-based multi-domain meeting summarization,” arXiv preprint arXiv:2104.05938 , 2021

Original

2021

Earlier work this paper cites.

R. Y. Pang, A. Parrish et al. , “Quality: Question answering with long input texts, yes!” arXiv preprint arXiv:2112.08608 , 2021

Original

2021

Earlier work this paper cites.

D. Hafner, “Benchmarking the spectrum of agent capabilities,” arXiv preprint arXiv:2109.06780 , 2021

Original

2021

Earlier work this paper cites.

K. Cobbe, V. Kosaraju et al. , “Training verifiers to solve math word problems,” arXiv preprint arXiv:2110.14168 , 2021

Original

2021

Earlier work this paper cites.

D. Hendrycks, C. Burns et al. , “Measuring mathematical problem solving with the math dataset,” arXiv preprint arXiv:2103.03874 , 2021

Original

2021

Earlier work this paper cites.

A. Patel, S. Bhattamishra, and N. Goyal, “Are nlp models really able to solve simple math word problems?” arXiv preprint arXiv:2103.07191 , 2021

Original

2021

Earlier work this paper cites.

M. Geva, D. Khashabi et al. , “Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies,” Transactions of the Association for Computational Linguistics , vol. 9, pp. 346–361, 2021

2021

Earlier work this paper cites.

D. Hendrycks, S. Basart et al. , “Measuring coding challenge competence with apps,” 2021. [Online]. Available: https://arxiv.org/abs/2105.09938

Original

2021

Earlier work this paper cites.

P. Yin, W.-D. Li et al. , “Natural language to code generation in interactive data science notebooks,” arXiv preprint arXiv:2212.09248 , 2022

Original

2022

Earlier work this paper cites.

S. G. Venkatesh, P. Talukdar, and S. Narayanan, “Ugif: Ui grounded instruction following,” arXiv preprint arXiv:2211.07615 , 2022

Original

2022

Earlier work this paper cites.

P. Lu, S. Mishra et al. , “Learn to explain: Multimodal reasoning via thought chains for science question answering,” 2022. [Online]. Available: https://arxiv.org/abs/2209.09513

Original

2022

Earlier work this paper cites.

R. Wang, P. Jansen et al. , “Scienceworld: Is your agent smarter than a 5th grader?, 2022,” URL https://arxiv. org/abs/2203.07540 , 2022

Original

2022

Earlier work this paper cites.

J. Wei, X. Wang et al. , “Chain of thought prompting elicits reasoning in large language models,” in Advances in Neural Information Processing Systems , vol. 35, 2022, pp. 24 824–24 837. [Online]. Available: https://arxiv.org/abs/2201.11903

Original

2022

Earlier work this paper cites.

OpenAI, “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774 , 2023. [Online]. Available: https://doi.org/10.48550/arXiv.2303.08774

Original

2023

Earlier work this paper cites.

H. Touvron, T. Lavril et al. , “Llama: Open and efficient foundation language models,” CoRR , vol. abs/2302.13971, 2023. [Online]. Available: https://doi.org/10.48550/arXiv.2302.13971

Original

2023

Earlier work this paper cites.

R. Anil, S. Borgeaud et al. , “Gemini: A family of highly capable multimodal models,” arXiv preprint arXiv:2312.11805 , 2023. [Online]. Available: https://arxiv.org/abs/2312.11805

Original

2023

Earlier work this paper cites.

Y. Bai, Y. Lin et al. , “Qwen technical report,” arXiv preprint arXiv:2309.16609 , 2023. [Online]. Available: https://arxiv.org/abs/2309.16609

Original

2023

Earlier work this paper cites.

N. Shinn, F. Cassano et al. , “Reflexion: Language agents with verbal reinforcement learning,” arXiv preprint arXiv:2303.11366 , 2023. [Online]. Available: https://arxiv.org/abs/2303.11366

Original

2023

Earlier work this paper cites.

Y. Chang, X. Wang et al. , “A survey on evaluation of large language models,” 2023. [Online]. Available: https://arxiv.org/abs/2307.03109

Original

2023

Earlier work this paper cites.

Z. Guo, R. Jin et al. , “Evaluating large language models: A comprehensive survey,” 2023. [Online]. Available: https://arxiv.org/abs/2310.19736

Original

2023

Earlier work this paper cites.

X. Du, M. Liu et al. , “Classeval: A manually-crafted benchmark for evaluating llms on class-level code generation,” arXiv preprint arXiv:2308.01861 , 2023

Original

2023

Earlier work this paper cites.

Z. Yuan, J. Liu et al. , “Evaluating instruction-tuned large language models on code comprehension and generation,” arXiv preprint arXiv:2308.01240 , 2023

Original

2023

Earlier work this paper cites.

C. E. Jimenez, J. Yang et al. , “Swe-bench: Can language models resolve real-world github issues?” arXiv preprint arXiv:2310.06770 , 2023

Original

2023

Earlier work this paper cites.

T. Liu, C. Xu, and J. McAuley, “Repobench: Benchmarking repository-level code auto-completion systems,” arXiv preprint arXiv:2306.03091 , 2023

Original

2023

Earlier work this paper cites.

C. Qian, W. Liu et al. , “Chatdev: Communicative agents for software development,” arXiv preprint arXiv:2307.07924 , 2023

Original

2023

Earlier work this paper cites.

X. Tang, Y. Liu et al. , “Ml-bench: Evaluating large language models and agents for machine learning tasks on repository-level code,” arXiv preprint arXiv:2311.09835 , 2023

Original

2023

Earlier work this paper cites.

S. Yao, H. Chen et al. , “Webshop: Towards scalable real-world web interaction with grounded language agents,” 2023. [Online]. Available: https://arxiv.org/abs/2207.01206

Original

2023

Earlier work this paper cites.

X. Deng, Y. Gu et al. , “Mind2web: Towards a generalist agent for the web,” 2023. [Online]. Available: https://arxiv.org/abs/2306.06070

Original

2023

Earlier work this paper cites.

C. Rawles, A. Li et al. , “Androidinthewild: A large-scale dataset for android device control,” Advances in Neural Information Processing Systems , vol. 36, pp. 59 708–59 728, 2023

2023

Earlier work this paper cites.

D. Zhang, Z. Shen et al. , “Mobile-env: Building qualified evaluation benchmarks for llm-gui interaction,” arXiv preprint arXiv:2305.08144 , 2023

Original

2023

Earlier work this paper cites.

Y. Lee, K. Lee et al. , “Qasa: advanced question answering on scientific articles,” in International Conference on Machine Learning . PMLR, 2023, pp. 19 036–19 052

2023

Earlier work this paper cites.

J. Lála, O. O’Donoghue et al. , “Paperqa: Retrieval-augmented generative agent for scientific research,” arXiv preprint arXiv:2312.07559 , 2023

Original

2023

Earlier work this paper cites.

Y. Wu, X. Tang, T. M. Mitchell, and Y. Li, “Smartplay: A benchmark for llms as intelligent agents,” arXiv preprint arXiv:2310.01557 , 2023

Original

2023

Earlier work this paper cites.

K. Valmeekam, M. Marquez et al. , “Planbench: An extensible benchmark for evaluating large language models on planning and reasoning about change,” Advances in Neural Information Processing Systems , vol. 36, pp. 38 975–38 987, 2023

2023

Earlier work this paper cites.

K. Valmeekam, S. Sreedharan et al. , “On the planning abilities of large language models (a critical investigation with a proposed benchmark),” arXiv preprint arXiv:2302.06706 , 2023

Original

2023

Earlier work this paper cites.

K. Stein, D. Fišer et al. , “Autoplanbench: Automatically generating benchmarks for llm planners from pddl,” arXiv preprint arXiv:2311.09830 , 2023

Original

2023

Earlier work this paper cites.

C.-A. Cheng, A. Kolobov et al. , “Llf-bench: Benchmark for interactive learning from language feedback,” 2023. [Online]. Available: https://arxiv.org/abs/2312.06853

Original

2023

Earlier work this paper cites.

Q. Tang et al. , “Toolalpaca: Generalized tool learning for language models with 3000 simulated cases,” arXiv , 2023. [Online]. Available: https://arxiv.org/abs/2306.05301

Original

2023

Earlier work this paper cites.

S. Patil et al. , “Toolllm: Facilitating large language models to master 16000+ real-world apis,” arXiv , 2023. [Online]. Available: https://arxiv.org/abs/2305.16504

Original

2023

Earlier work this paper cites.

Y. Li et al. , “Api-bank: A comprehensive benchmark for tool-augmented llms,” EMNLP , 2023. [Online]. Available: https://aclanthology.org/2023.emnlp-main.187/

2023

Earlier work this paper cites.

R. Team, “Restgpt: Connecting llms with real-world restful apis,” RestGPT , 2023. [Online]. Available: https://restgpt.github.io/

2023

Earlier work this paper cites.

——, “Sotopia: Interactive evaluation for social intelligence in language agents,” arXiv preprint arXiv:2310.11667 , 2023. [Online]. Available: https://arxiv.org/abs/2310.11667

Original

2023

Earlier work this paper cites.

M. Alahi et al. , “Smarteval: Evaluation system for descriptive answers in examinations using iot-enabled technologies and artificial intelligence,” Sensors , vol. 23, no. 11, p. 5206, 2023. [Online]. Available: https://doi.org/10.3390/s23115206

2023

Earlier work this paper cites.

X. Liu, H. Yu et al. , “Agentbench: Evaluating llms as agents,” 2023. [Online]. Available: https://arxiv.org/abs/2308.03688

Original

2023

Earlier work this paper cites.

G. Mialon, C. Fourrier et al. , “Gaia: a benchmark for general ai assistants,” 2023. [Online]. Available: https://arxiv.org/abs/2311.12983

Original

2023

Earlier work this paper cites.

G. A. Team, “Galileo ai agent leaderboard,” 2023. [Online]. Available: https://huggingface.co/spaces/galileo-ai/agent-leaderboard

2023

Earlier work this paper cites.

F. Zhang, B. Chen et al. , “Repocoder: Repository-level code completion through iterative retrieval and generation,” arXiv preprint arXiv:2303.12570 , 2023

Original

2023

Earlier work this paper cites.

T. Carta, C. Romac et al. , “Grounding large language models in interactive environments with online reinforcement learning,” in International Conference on Machine Learning . PMLR, 2023, pp. 3676–3713

2023

Earlier work this paper cites.

X. Yao, Z. Ye et al. , “Tree of thoughts: Deliberate problem solving with large language models,” in Advances in Neural Information Processing Systems , vol. 36, 2023. [Online]. Available: https://arxiv.org/abs/2305.10601

Original

2023

Earlier work this paper cites.

Y. Zhang, J. Yang et al. , “Cumulative reasoning with large language models,” arXiv preprint arXiv:2308.04371 , 2023

Original

2023

Earlier work this paper cites.

W. Chen, Y. Su et al. , “Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors,” 2023. [Online]. Available: https://arxiv.org/abs/2308.10848

Original

2023

Cited alongside, same era.

T. Abuelsaad, D. Akkil et al. , “Agent-e: From autonomous web navigation to foundational design principles in agentic systems,” arXiv preprint arXiv:2407.13032 , 2024. [Online]. Available: https://arxiv.org/abs/2407.13032

Original

2024

Cited alongside, same era.

Y. Song, F. Xu et al. , “Beyond browsing: Api-based web agents,” arXiv preprint arXiv:2410.16464 , 2024. [Online]. Available: https://arxiv.org/abs/2410.16464

Original

2024

Cited alongside, same era.

M. T. R. Laskar, S. Alqahtani et al. , “A systematic survey and critical review on evaluating large language models: Challenges, limitations, and recommendations,” 2024. [Online]. Available: https://arxiv.org/abs/2407.04069

Original

2024

Cited alongside, same era.

G. Team, “Berkeley function calling leaderboard v3 (aka berkeley tool calling leaderboard v3),” Gorilla , 2024. [Online]. Available: https://gorilla.cs.berkeley.edu/leaderboard.html

2024

Later among the works it cites.

J. Lu et al. , “Toolsandbox: A stateful, conversational, interactive evaluation benchmark for llm tool use capabilities,” arXiv , 2024. [Online]. Available: https://arxiv.org/abs/2408.04682

Original

2024

Later among the works it cites.

F. Team, “Seal-tools: Self-instruct tool learning dataset for agent tuning and evaluation,” arXiv , 2024. [Online]. Available: https://arxiv.org/abs/2405.08355

Original

2024

Later among the works it cites.

N. Team, “Nexusraven-13b, a new sota open-source llm for function calling,” Nexusflow , 2024. [Online]. Available: https://github.com/nexusflowai/NexusRaven

2024

Later among the works it cites.