Elamur Research⌕
☰
home / papers / 2026 / From Coarse to Fine: Self-Adaptive Hierarchical Planning for LLM Agents

From Coarse to Fine: Self-Adaptive Hierarchical Planning for LLM Agents

Paper · 2026 · Haoran Tan, Zeyu Zhang, Chen Ma et al.

Summary

LLM agents that plan explicitly usually produce plans at a single, fixed level of detail. Such plans are too verbose for easy tasks and too shallow for hard ones. The authors propose AdaPlan-H, a planner inspired by the cognitive-science idea of progressive refinement. It first drafts a coarse global plan and then adds more detailed levels, each building on the previous one. It stops when the task's complexity no longer calls for more detail.

Training has two stages:

  • Imitation learning. GPT-4o turns successful trajectories into multi-level candidate plans. Monte Carlo rollouts with a Llama-3.1-8B-Instruct actor estimate how well each depth prefix works, and the best one is selected as supervision for fine-tuning.
  • Capability enhancement. Preference pairs built from plan quality further improve both the choice of depth and the plan content.

On two interactive benchmarks, one of which features ScienceWorld-style tasks, AdaPlan-H raises task success rates over fixed-granularity planners. It also cuts wasted planning tokens. This matters because it offers a modular way to match plan detail to task difficulty without retraining the actor.

Key points

  • A plan is built as stacked levels p1…pm. Each level is a complete plan, more detailed than the one above, and the depth m is chosen adaptively per task.
  • Supervision labels come from GPT-4o-generated hierarchical plans. Each depth prefix is scored by its average reward over K rollouts, following the Monte Carlo approach of Xiong et al. (2025), with Llama-3.1-8B-Instruct as the actor.
  • Experiments capped the hierarchy at 3 levels. Results hint that deeper hierarchies could help further.
  • Gains are larger when the actor used for exploration is the same as the actor used at test time. Only two plans per task were used to form preference pairs.
  • Code and data: github.com/import-myself/AHP (Findings of ACL 2026, Renmin University and Huawei Noah's Ark Lab).

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

🎓Renmin University of ChinaBeijing · China114 works · 16,118 citations🏢Huawei TechnologiesShenzhen · China75 works · 12,975 citations

 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

Cite

Cite as Haoran Tan, Zeyu Zhang, Chen Ma, Tianze Liu, Quanyu Dai and Xu Chen. 2026. From Coarse to Fine: Self-Adaptive Hierarchical Planning for LLM Agents. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.77

@inproceedings{tan2026coarse,
  title = {{From Coarse to Fine: Self-Adaptive Hierarchical Planning for LLM Agents}},
  author = {Haoran Tan, Zeyu Zhang, Chen Ma, Tianze Liu, Quanyu Dai and Xu Chen},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.77},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.77}
}

References

35 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Pengfei Cao, Tianyi Men, Wencan Liu, Jingwen Zhang, Xuzhao Li, Xixun Lin, Dianbo Sui, Yanan Cao, Kang Liu, and Jun Zhao. 2025. Large language models for planning: A comprehensive and systematic survey. arXiv preprint arXiv:2505.19683. — [arXiv:2505.19683](https://arxiv.org/abs/2505.19683) find
  3. Baian Chen, Chang Shu, Ehsan Shareghi, Nigel Collier, Karthik Narasimhan, and Shunyu Yao. 2023. Fireact: Toward language agent fine-tuning. arXiv preprint arXiv:2310.05915.
  4. Carlos G Correa, Sophia Sanborn, Mark K Ho, Frederick Callaway, Nathaniel D Daw, and Thomas L Griffiths. 2025. Exploring the hierarchical structure of human plans via program generation. Cognition, 255:105990. find
  5. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, and 1 others. 2024. The llama 3 herd of models. arXiv e-prints, pages arXiv–2407.
  6. Lutfi Eren Erdogan, Nicholas Lee, Sehoon Kim, Suhong Moon, Hiroki Furuta, Gopala Anumanchipalli, Kurt Keutzer, and Amir Gholami. 2025. Plan-and-act: Improving planning of agents for long-horizon tasks. arXiv preprint arXiv:2503.09572. — [arXiv:2503.09572](https://arxiv.org/abs/2503.09572) find
  7. Lang Feng, Zhenghai Xue, Tingcong Liu, and Bo An. 2025. Group-in-group policy optimization for llm agent training. arXiv preprint arXiv:2505.10978. find
  8. Team GLM, Aohan Zeng, Bin Xu, Bowen Wang, Chenhui Zhang, Da Yin, Diego Rojas, Guanyu Feng, Hanlin Zhao, Hanyu Lai, Hao Yu, Hongning Wang, Jiadai Sun, Jiajie Zhang, Jiale Cheng, Jiayi Gui, Jie Tang, Jing Zhang, Juanzi Li, and 37 others. 2024. Chatglm: A family of large language models from glm-130b t…
  9. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, and 1 others. 2022. Lora: Low-rank adaptation of large language models. ICLR, 1(2):3.
  10. Mengkang Hu, Yao Mu, Xinmiao Yu, Mingyu Ding, Shiguang Wu, Wenqi Shao, Qiguang Chen, Bin Wang, Yu Qiao, and Ping Luo. 2024. Tree-planner: Efficient close-loop task planning with large language models. Preprint, arXiv:2310.08582. — [arXiv:2310.08582](https://arxiv.org/abs/2310.08582) find
  11. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the 29th symposium on operating systems principles, pages 611–626.
  12. Bo Liu, Yuqian Jiang, Xiaohan Zhang, Qiang Liu, Shiqi Zhang, Joydeep Biswas, and Peter Stone. 2023. Llm+ p: Empowering large language models with optimal planning proficiency. arXiv preprint arXiv:2304.11477.
  13. Wei Liu, Yang Bai, Chengcheng Han, Rongxiang Weng, Jun Xu, Xuezhi Cao, Jingang Wang, and Xunliang Cai. 2024. Length desensitization in direct preference optimization. arXiv preprint arXiv:2409.06411. — [arXiv:2409.06411](https://arxiv.org/abs/2409.06411) find
  14. Arka Pal, Deep Karkhanis, Samuel Dooley, Manley Roberts, Siddartha Naidu, and Colin White. 2024. Smaug: Fixing failure modes of preference optimisation with dpo-positive. arXiv preprint arXiv:2402.13228. find
  15. Archiki Prasad, Alexander Koller, Mareike Hartmann, Peter Clark, Ashish Sabharwal, Mohit Bansal, and Tushar Khot. 2023. Adapt: As-needed decomposition and planning with language models. arXiv preprint arXiv:2311.05772.
  16. Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, and 1 others. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300.
  17. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. Advances in Neural Information Processing Systems, 36:8634–8652.
  18. Mohit Shridhar, Xingdi Yuan, Marc-Alexandre Côté, Yonatan Bisk, Adam Trischler, and Matthew Hausknecht. 2020. Alfworld: Aligning text and embodied environments for interactive learning. arXiv preprint arXiv:2010.03768.
  19. Yifan Song, Da Yin, Xiang Yue, Jie Huang, Sujian Li, and Bill Yuchen Lin. 2024. Trial and error: Exploration-based trajectory optimization for llm agents. arXiv preprint arXiv:2403.02502. — [arXiv:2403.02502](https://arxiv.org/abs/2403.02502) find
  20. Haoran Tan, Zeyu Zhang, Chen Ma, Xu Chen, Quanyu Dai, and Zhenhua Dong. 2025. Membench: Towards more comprehensive evaluation on the memory of llm-based agents. arXiv preprint arXiv:2506.21605. find
  21. Qwen Team. 2024. Qwen2.5: A party of foundation models. find
  22. Qwen Team. 2025. Qwen3 technical report. Preprint, arXiv:2505.09388.
  23. Yu-Min Tseng, Yu-Chao Huang, Teng-Yun Hsiao, Wei-Lin Chen, Chao-Wei Huang, Yu Meng, and Yun-Nung Chen. 2024. Two tales of persona in llms: A survey of role-playing and personalization. arXiv preprint arXiv:2406.01171.
  24. Lei Wang, Chen Ma, Xueyang Feng, Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai Tang, Xu Chen, Yankai Lin, and 1 others. 2024. A survey on large language model based autonomous agents. Frontiers of Computer Science, 18(6):186345.
  25. Lei Wang, Wanyu Xu, Yihuai Lan, Zhiqiang Hu, Yunshi Lan, Roy Ka-Wei Lee, and Ee-Peng Lim. 2023. Plan-and-solve prompting: Improving zero-shot chain-of-thought reasoning by large language models. arXiv preprint arXiv:2305.04091. find
  26. Ruoyao Wang, Peter Jansen, Marc-Alexandre Côté, and Prithviraj Ammanabrolu. 2022. Scienceworld: Is your agent smarter than a 5th grader? arXiv preprint arXiv:2203.07540. find
  27. Weimin Xiong, Yifan Song, Qingxiu Dong, Bingchan Zhao, Feifan Song, Xun Wang, and Sujian Li. 2025. Mpo: Boosting llm agents with meta plan optimization. arXiv preprint arXiv:2503.02682. find
  28. Weimin Xiong, Yifan Song, Xiutian Zhao, Wenhao Wu, Xun Wang, Ke Wang, Cheng Li, Wei Peng, and Sujian Li. 2024. Watch every step! llm agent learning via iterative step-level process refinement. arXiv preprint arXiv:2406.11176. find
  29. Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Tom Griffiths, Yuan Cao, and Karthik Narasimhan. 2023a. Tree of thoughts: Deliberate problem solving with large language models. Advances in neural information processing systems, 36:11809–11822. find
  30. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023b. React: Synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR).
  31. Jeffrey M Zacks, Barbara Tversky, and Gowri Iyer. 2001. Perceiving, remembering, and communicating structure in events. Journal of experimental psychology: General, 130(1):29. find
  32. 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.
  33. Zeyu Zhang, Xiaohe Bo, Chen Ma, Rui Li, Xu Chen, Quanyu Dai, Jieming Zhu, Zhenhua Dong, and Ji-Rong Wen. 2024. A survey on the memory mechanism of large language model based agents. arXiv preprint arXiv:2404.13501. find
  34. Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, and 1 others. 2023. A survey of large language models. arXiv preprint arXiv:2303.18223.
  35. Yaowei Zheng, Richong Zhang, Junhao Zhang, Yanhan Ye, Zheyan Luo, Zhangchi Feng, and Yongqiang Ma. 2024. Llamafactory: Unified efficient fine-tuning of 100+ language models. arXiv preprint arXiv:2403.13372.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/tan-2026-coarse-fine-self-adaptive-hierarchical/