Elamur Research⌕
☰
home / papers / 2024 / Understanding the planning of LLM agents: A survey

Understanding the planning of LLM agents: A survey

Paper · 2024 · Xu Huang, Weiwen Liu, Xiaolong Chen et al.

Summary

This survey, by authors from USTC and Huawei Noah's Ark Lab, examines how large language models serve as the planning component of autonomous agents. Classical approaches such as PDDL-based symbolic planning and reinforcement learning need expert-built formal models or very large numbers of environment interactions. LLMs promise more flexible planning, but earlier surveys had not looked closely at this capability. The authors define planning as producing an action sequence from an environment, a goal, model parameters and prompts. They then group recent work into five overlapping families: task decomposition, multi-plan selection, external planner-aided planning, reflection and refinement, and memory-augmented planning. Each family gets a formal description and representative systems. They also run six prompt-based methods on four benchmarks. In those runs, spending more tokens generally improved results, few-shot examples mattered for complex tasks, and reflection helped most on hard interactive tasks. The survey closes with open problems: hallucination, plan feasibility and efficiency, multimodal feedback, and coarse evaluation. It offers a structured map of a fast-growing field.

Key points

  • Taxonomy and representative works:
    • Decomposition: CoT, ReAct, HuggingGPT, Plan-and-Solve, ProgPrompt.
    • Selection: ToT, GoT, CoT-SC.
    • External planners: LLM+P, LLM+PDDL.
    • Reflection: Reflexion, CRITIC, Self-Refine.
    • Memory: REMEMBER, MemoryBank.
  • Decomposition styles: task decomposition splits into "decomposition-first" methods (e.g., HuggingGPT) and "interleaved" methods (e.g., ReAct, PAL, PoT).
  • Experimental setup:
    • Benchmarks: ALFWorld, ScienceWorld, HotPotQA and FEVER.
    • Model: text-davinci-003.
    • Value function for SayCan: bge-small-en-v1.5.
    • Reflexion was allowed one retry.
  • Findings:
    • ZeroShot-CoT degraded sharply on the two QA benchmarks.
    • Reflexion used roughly twice ReAct's tokens but gave notable gains on ALFWorld and ScienceWorld.
  • Evaluation critique: existing environments mostly score only final success rate and usually have a single "golden" path. The authors call for fine-grained, step-wise evaluation.

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

 Institutions

🎓University of Science and Technology of ChinaHefei · China145 works · 114,409 citations🏢Huawei TechnologiesShenzhen · China96 works · 15,432 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 works

Cite

Cite as Xu Huang, Weiwen Liu, Xiaolong Chen, Xingmei Wang, Hao Wang, Defu Lian, Yasheng Wang, Ruiming Tang and Enhong Chen. 2024. Understanding the planning of LLM agents: A survey. arXiv:2402.02716.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2402.02716 · arXiv 2402.02716

@misc{huang2024understanding,
  title = {{Understanding the planning of LLM agents: A survey}},
  author = {Xu Huang, Weiwen Liu, Xiaolong Chen, Xingmei Wang, Hao Wang, Defu Lian, Yasheng Wang, Ruiming Tang and Enhong Chen},
  year = {2024},
  doi = {10.48550/arxiv.2402.02716},
  eprint = {2402.02716},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2402.02716}
}

References

65 references, 36 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Constructions Aeronautiques, Adele Howe, et al. Pddl— the planning domain definition language. Technical Report, Tech. Rep., 1998. find
  2. Shengnan An, Zexiong Ma, et al. Learning from mistakes makes llm better reasoner. arXiv preprint arXiv:2310.20689, 2023.
  3. Maciej Besta, Nils Blach, et al. Graph of thoughts: Solving elaborate problems with large language models. arXiv preprint arXiv:2308.09687, 2023. find
  4. Deng Cai, Yan Wang, Lemao Liu, and Shuming Shi. Recent advances in retrieval-augmented text generation. In SIGIR, pages 3417–3419, 2022. find
  5. Lili Chen, Kevin Lu, et al. Decision transformer: Reinforcement learning via sequence modeling. NeurIPS, 34:15084–15097, 2021. find
  6. Mark Chen, Jerry Tworek, et al. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374, 2021.
  7. Wenhu Chen, Xueguang Ma, et al. Program of thoughts prompting: Disentangling computation from reasoning for numerical reasoning tasks. arXiv preprint arXiv:2211.12588, 2022. find
  8. Gautier Dagan, Frank Keller, and Alex Lascarides. Dynamic planning with a llm. arXiv preprint arXiv:2308.06391, 2023.
  9. Xiang Deng, Yu Gu, et al. Mind2web: Towards a generalist agent for the web. arXiv preprint arXiv:2306.06070, 2023.
  10. Luyu Gao, Aman Madaan, et al. Pal: Program-aided language models. In ICML, pages 10764– 10799, 2023. find
  11. Alfonso Gerevini and Ivan Serina. Lpg: A planner based on local search for planning graphs with action costs. In Aips, volume 2, pages 281– 290, 2002. find
  12. Malik Ghallab, Dana Nau, et al. Automated Planning: theory and practice. Elsevier, 2004. find
  13. Zhibin Gou, Zhihong Shao, et al. Critic: Large language models can self-correct with tool-interactive critiquing. arXiv preprint arXiv:2305.11738, 2023.
  14. Lin Guan, Karthik Valmeekam, et al. Leveraging pre-trained large language models to construct and utilize world models for model-based task planning. arXiv preprint arXiv:2305.14909, 2023.
  15. Shibo Hao, Yi Gu, et al. Reasoning with language model is planning with world model. arXiv preprint arXiv:2305.14992, 2023. find
  16. Patrik Haslum, Nir Lipovetzky, et al. An introduction to the planning domain definition language, volume 13. Springer, 2019. find
  17. Ji He, Jianshu Chen, et al. Deep reinforcement learning with a natural language action space. arXiv preprint arXiv:1511.04636, 2015. find
  18. Lei Huang, Yu Weijiang, et al. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. arXiv preprint arXiv:2311.05232, 2023.
  19. Xu Huang, Jianxun Lian, et al. Recommender ai agent: Integrating large language models for interactive recommendations. arXiv preprint arXiv:2308.16505, 2023. — [arXiv:2308.16505](https://arxiv.org/abs/2308.16505) find
  20. Jeff Johnson, Matthijs Douze, and Hervé Jégou. Billion-scale similarity search with GPUs. IEEE Transactions on Big Data, 7(3):535–547, 2019.
  21. Geunwoo Kim et al. Language models can solve computer tasks. arXiv preprint arXiv:2303.17491, 2023.
  22. Takeshi Kojima, Shixiang Shane Gu, et al. Large language models are zero-shot reasoners. NeurIPS, 35:22199–22213, 2022.
  23. Patrick Lewis, Ethan Perez, et al. Retrieval-augmented generation for knowledge-intensive nlp tasks. NeurIPS, 33:9459–9474, 2020.
  24. Bill Yuchen Lin, Yicheng Fu, et al. Swiftsage: A generative agent with fast and slow thinking for complex interactive tasks. arXiv preprint arXiv:2305.17390, 2023.
  25. Nir Lipovetzky, Miquel Ramirez, et al. Width and inference based planners: Siw, bfs (f), and probe. IPC, page 43, 2014. find
  26. Bo Liu, Yuqian Jiang, et al. Llm+ p: Empowering large language models with optimal planning proficiency. arXiv preprint arXiv:2304.11477, 2023.
  27. Lei Liu, Xiaoyan Yang, et al. Think-in-memory: Recalling and post-thinking enable llms with long-term memory. arXiv preprint arXiv:2311.08719, 2023. — [arXiv:2311.08719](https://arxiv.org/abs/2311.08719) find
  28. Xiao Liu, Hao Yu, et al. Agentbench: Evaluating llms as agents. arXiv preprint arXiv:2308.03688, 2023.
  29. Aman Madaan, Niket Tandon, , et al. Self-refine: Iterative refinement with self-feedback. arXiv preprint arXiv:2303.17651, 2023. find
  30. Yuning Mao, Pengcheng He, Liu, et al. Generation-augmented retrieval for open-domain question answering. arXiv preprint arXiv:2009.08553, 2020.
  31. Charles Packer, Vivian Fang, et al. Memgpt: Towards llms as operating systems. arXiv preprint arXiv:2310.08560, 2023. find
  32. Shirui Pan, Linhao Luo, et al. Unifying large language models and knowledge graphs: A roadmap. TKDE, 2024.
  33. Joon Sung Park, Joseph O’Brien, et al. Generative agents: Interactive simulacra of human behavior. In SUIST, pages 1–22, 2023. find
  34. Yujia Qin, Shengding Hu, et al. Tool learning with foundation models. arXiv preprint arXiv:2304.08354, 2023.
  35. Jan Maarten Schraagen, Susan F Chipman, et al. Cognitive task analysis. Psychology Press, 2000. find
  36. Yongliang Shen, Kaitao Song, et al. Hugginggpt: Solving ai tasks with chatgpt and its friends in huggingface. arXiv preprint arXiv:2303.17580, 2023.
  37. Noah Shinn, Federico Cassano, et al. Reflexion: Language agents with verbal reinforcement learning. In NeurIPS, 2023.
  38. Mohit Shridhar, Xingdi Yuan, et al. Alfworld: Aligning text and embodied environments for interactive learning. arXiv preprint arXiv:2010.03768, 2020.
  39. Ishika Singh, Valts Blukis, et al. Progprompt: Generating situated robot task plans using large language models. In ICRA 2023, pages 11523–11530. IEEE, 2023.
  40. Jiankai Sun, Chuanyang Zheng, et al. A survey of reasoning with foundation models. arXiv preprint arXiv:2312.11562, 2023. — [arXiv:2312.11562](https://arxiv.org/abs/2312.11562) find
  41. James Thorne, Andreas Vlachos, et al. Fever: a large-scale dataset for fact extraction and verification. arXiv preprint arXiv:1803.05355, 2018.
  42. Hugo Touvron, Louis Martin, et al. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288, 2023.
  43. Ruoyao Wang, Peter Jansen, et al. Scienceworld: Is your agent smarter than a 5th grader? arXiv preprint arXiv:2203.07540, 2022. find
  44. Xuezhi Wang, Jason Wei, et al. Self-consistency improves chain of thought reasoning in language models. arXiv preprint arXiv:2203.11171, 2022.
  45. Lei Wang, Chen Ma, et al. A survey on large language model based autonomous agents. arXiv preprint arXiv:2308.11432, 2023.
  46. Lei Wang, Wanyu Xu, et al. Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models. arXiv preprint arXiv:2305.04091, 2023. find
  47. Yancheng Wang, Ziyan Jiang, et al. Recmind: Large language model powered agent for recommendation. arXiv preprint arXiv:2308.14296, 2023. find
  48. Jason Wei, Xuezhi Wang, et al. Chain-of-thought prompting elicits reasoning in large language models. NeurIPS, 35:24824–24837, 2022.
  49. Chenfei Wu, Shengming Yin, et al. Visual chatgpt: Talking, drawing and editing with visual foundation models. arXiv preprint arXiv:2303.04671, 2023.
  50. Shitao Xiao et al. C-pack: Packaged resources to advance general chinese embedding, 2023. find
  51. Hengjia Xiao and Peng Wang. Llm a*: Human in the loop large language models enabled a* search for robotics. arXiv preprint arXiv:2312.01797, 2023. — [arXiv:2312.01797](https://arxiv.org/abs/2312.01797) find
  52. Zhilin Yang, Peng Qi, et al. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. arXiv preprint arXiv:1809.09600, 2018.
  53. Sherry Yang, Nachum Ofir, et al. Foundation models for decision making: Problems, methods, and opportunities. arXiv preprint arXiv:2303.04129, 2023. find
  54. Zhun Yang, Adam Ishay, and Joohyung Lee. Coupling large language models with logic programming for robust and general reasoning from text. arXiv preprint arXiv:2307.07696, 2023. find
  55. Shunyu Yao, Rohan Rao, et al. Keep calm and explore: Language models for action generation in text-based games. arXiv preprint arXiv:2010.02903, 2020.
  56. Shunyu Yao, Rohan Rao, et al. Keep calm and explore: Language models for action generation in text-based games. arXiv preprint arXiv:2010.02903, 2020.
  57. Shunyu Yao, Jeffrey Zhao, et al. React: Synergizing reasoning and acting in language models. arXiv preprint arXiv:2210.03629, 2022.
  58. Shunyu Yao, Dian Yu, et al. Tree of thoughts: Deliberate problem solving with large language models. arXiv preprint arXiv:2305.10601, 2023. find
  59. Aohan Zeng, Mingdao Liu, et al. Agenttuning: Enabling generalized agent abilities for llms. arXiv preprint arXiv:2310.12823, 2023.
  60. Danyang Zhang, Lu Chen, et al. Large language model is semi-parametric reinforcement learning agent. arXiv preprint arXiv:2306.07929, 2023. — [arXiv:2306.07929](https://arxiv.org/abs/2306.07929) find
  61. Yue Zhang, Yafu Li, et al. Siren’s song in the ai ocean: A survey on hallucination in large language models. arXiv preprint arXiv:2309.01219, 2023.
  62. Wayne Xin Zhao, Kun Zhou, et al. A survey of large language models. arXiv preprint arXiv:2303.18223, 2023.
  63. Zirui Zhao, Wee Sun Lee, and David Hsu. Large language models as commonsense knowledge for large-scale task planning. arXiv preprint arXiv:2305.14078, 2023.
  64. Wanjun Zhong, Lianghong Guo, Qiqi Gao, and Yanlin Wang. Memorybank: Enhancing large language models with long-term memory. arXiv preprint arXiv:2305.10250, 2023. find
  65. Shuyan Zhou, Frank F Xu, et al. Webarena: A realistic web environment for building autonomous agents. arXiv preprint arXiv:2307.13854, 2023.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/huang-2024-understanding-planning-llm-agents-survey/