Elamur Research⌕
☰
home / papers / 2026 / EvoHyper: Evolving Hypergraph Topologies for Unified Collaboration in Multi-Agent Communication

EvoHyper: Evolving Hypergraph Topologies for Unified Collaboration in Multi-Agent Communication

Paper · 2026 · Heng Zhang, Yihao Zhong, Lubin Gan et al.

Summary

LLM-based multi-agent systems usually pair a fixed communication graph with a separately managed memory module. The authors analysed GSM8K execution logs of G-Designer, GPTSwarm, MetaGPT and AgentVerse. They found that this split causes key facts to be duplicated across messages, memory and prompts, and most of the duplication comes from agents sharing one subtask. EvoHyper replaces the pairwise graph with a single hypergraph. Each hyperedge is a collaboration unit that groups several agents together with their shared memory. A variational encoder-decoder derives a task-adapted hypergraph from an anchor topology. A learnable evolution controller then reshapes it during execution through Update, Spawn and Merge operations, while a hierarchical, periodically compressed memory keeps context bounded. On four reasoning and code-generation benchmarks, EvoHyper outperforms prior topology-design methods by 3.2–7.8% and cuts tokens by up to 23.5%. It also adjusts structural complexity to task difficulty. The work suggests that representing collaboration and memory jointly is a practical way to make multi-agent pipelines cheaper and more focused.

Key points

  • Log analysis on GSM8K shows each key fact appears 2.8–4.1 times across messages, memory and prompts, and 67% of the repeats come from agents working on the same subtask.
  • Benchmarks are MMLU, GSM8K, MultiArith and HumanEval. EvoHyper averages 91.66%, ahead of G-Designer (88.78%) and AgentDropout (88.24%). It scores 96.49% on GSM8K, 99.18% on MultiArith and 92.28% pass@1 on HumanEval.
  • In the ablation, taking out the evolution controller hurts performance more than any other component (3.32 points). Removing the hypergraph costs 2.92, Merge 2.48, the VAE framework 1.69 and hierarchical memory 0.99.
  • Best hyperparameters are hyperedge size k=2, sparsity ζ=0.1, K=3 interaction rounds and operation threshold τ_op=0.5.
  • Across rounds, Update and Spawn dominate early and Merge grows later. Stated limitations are the assumption of reliable agents, a predefined operation set, and evaluation limited to reasoning and code tasks.

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 · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 works

Cite

Cite as Heng Zhang, Yihao Zhong, Lubin Gan, Zhihe Chen, Jiajun Wu, Yuling Shi, Xiaodong Gu, Hao Zhang, Haochen You and Jin Huang. 2026. EvoHyper: Evolving Hypergraph Topologies for Unified Collaboration in Multi-Agent Communication. 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.1258

@inproceedings{zhang2026evohyper,
  title = {{EvoHyper: Evolving Hypergraph Topologies for Unified Collaboration in Multi-Agent Communication}},
  author = {Heng Zhang, Yihao Zhong, Lubin Gan, Zhihe Chen, Jiajun Wu, Yuling Shi, Xiaodong Gu, Hao Zhang, Haochen You and Jin Huang},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1258},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1258}
}

References

35 references, 24 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Anthropic. 2024. The claude 3 model family: Opus, sonnet, haiku. Technical Report. find
  2. Maciej Besta, Nils Blach, Ales Kubicek, Robert Gerstenberger, Lukas Gianinazzi, Joanna Gajda, Tomasz Lehmann, Michal Podstawski, Hubert Niewiadomski, Piotr Nyczyk, and Torsten Hoefler. 2023. Graph of thoughts: Solving elaborate problems with large language models. arXiv preprint arXiv:2308.09687. find
  3. Chi-Min Chan, Weize Chen, Yusheng Su, Jianxuan Yu, Wei Xue, Shanghang Zhang, Jie Fu, and Zhiyuan Liu. 2023. Chateval: Towards better llm-based evaluators through multi-agent debate. arXiv preprint arXiv:2308.07201.
  4. Guangyao Chen, Siwei Dong, Yu Shu, Ge Zhang, Jaward Sesay, Börje F Karlsson, Jie Fu, and Yemin Shi. 2024a. Autoagents: A framework for automatic agent generation. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI), pages 21–29.
  5. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, and 39 others. 202… — Evaluating Large Language Models Trained on Code
  6. Weize Chen, Yusheng Su, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chen Qian, Chi-Min Chan, Yujia Qin, Yaxi Lu, Ruobing Xie, Zhiyuan Liu, Maosong Sun, and Jie Zhou. 2024b. Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors. In Proceedings of the 12th International Confer…
  7. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  8. Yilun Du, Shuang Li, Antonio Torralba, Joshua B Tenenbaum, and Igor Mordatch. 2024. Improving factuality and reasoning in language models through multiagent debate. In Proceedings of the 41st International Conference on Machine Learning.
  9. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2021. Measuring massive multitask language understanding. In Proceedings of the International Conference on Learning Representations (ICLR).
  10. Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Jinlin Wang, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu, and Jürgen Schmidhuber. 2024. Metagpt: Meta programming for a multi-agent collaborative framework. In The T…
  11. Dongfu Jiang, Xiang Ren, and Bill Yuchen Lin. 2023. Llm-blender: Ensembling large language models with pairwise ranking and generative fusion. arXiv preprint arXiv:2306.02561.
  12. Guohao Li, Hasan Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. 2023. Camel: Communicative agents for "mind" exploration of large language model society. In Advances in Neural Information Processing Systems (NeurIPS).
  13. Wang Ling, Dani Yogatama, Chris Dyer, and Phil Blunsom. 2017. Program induction by rationale generation: Learning to solve and explain algebraic word problems. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics, pages 158–167. find
  14. Zijun Liu, Yanzhe Zhang, Peng Li, Yang Liu, and Diyi Yang. 2024. A dynamic llm-powered agent network for task-oriented agent collaboration. Preprint, arXiv:2310.02170.
  15. OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, and 1 others. 2024. Gpt-4 technical report. Preprint, arXiv:2303.08774.
  16. Arkil Patel, Satwik Bhattamishra, and Navin Goyal. 2021. Are nlp models really able to solve simple math word problems? In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2080–2094.
  17. Chen Qian, Wei Liu, Hongzhang Liu, Nuo Chen, Yufan Dang, Jiahao Li, Cheng Yang, Weize Chen, Yusheng Su, Xin Cong, Juyuan Xu, Dahai Li, Zhiyuan Liu, and Maosong Sun. 2024. Chatdev: Communicative agents for software development. Preprint, arXiv:2307.07924.
  18. Chen Qian, Zihao Xie, YiFei Wang, Wei Liu, Kunlun Zhu, Hanchen Xia, Yufan Dang, Zhuoyun Du, Weize Chen, Cheng Yang, Zhiyuan Liu, and Maosong Sun. 2025. Scaling large language model-based multi-agent collaboration. Preprint, arXiv:2406.07155.
  19. Subhro Roy and Dan Roth. 2016. Solving general arithmetic word problems. arXiv preprint arXiv:1608.01413.
  20. Significant Gravitas. 2026. Autogpt. https://github.com/Significant-Gravitas/AutoGPT. Software, accessed April 20, 2026. ↗
  21. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2023a. Self-consistency improves chain of thought reasoning in language models. In Proceedings of the International Conference on Learning Representations (ICLR).
  22. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2023b. Self-consistency improves chain of thought reasoning in language models. Preprint, arXiv:2203.11171.
  23. Zhexuan Wang, Yutong Wang, Xuebo Liu, Liang Ding, Miao Zhang, Jie Liu, and Min Zhang. 2025. Agentdropout: Dynamic agent elimination for token-efficient and high-performance llm-based multi-agent collaboration. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics…
  24. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, and 1 others. 2022. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems, volume 35, pages 24824–24837.
  25. Qingyun Wu, Gagan Bansal, Jieyu Zhang, Yiran Wu, Shaokun Zhang, Erkang Zhu, Beibin Li, Li Jiang, Xiaoyun Zhang, and Chi Wang. 2023. Autogen: Enabling next-gen llm applications via multi-agent conversation. arXiv preprint arXiv:2308.08155.
  26. Xixi Wu, Yifei Shen, Caihua Shan, Kaitao Song, Siwei Wang, Bohang Zhang, Jiarui Feng, Hong Cheng, Wei Chen, Yun Xiong, and Dongsheng Li. 2024. Can graph learning improve planning in llm-based agents? Preprint, arXiv:2405.19119. — [arXiv:2405.19119](https://arxiv.org/abs/2405.19119) find
  27. Shunyu Yao, Dian Yu, Jeffrey Zhao, Izhak Shafran, Thomas L Griffiths, Yuan Cao, and Karthik Narasimhan. 2023a. Tree of thoughts: Deliberate problem solving with large language models. In Advances in Neural Information Processing Systems. find
  28. 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 Proceedings of the International Conference on Learning Representations (ICLR).
  29. Ceyao Zhang, Kaijie Yang, Siyi Hu, Zihao Wang, Guanghe Li, Yihang Sun, Cheng Zhang, Zhaowei Zhang, Anji Liu, Song-Chun Zhu, Xiaojun Chang, Junge Zhang, Feng Yin, Yitao Liang, and Yaodong Yang. 2024. Proagent: Building proactive cooperative agents with large language models. In Proceedings of the AAA… find
  30. Guibin Zhang, Haotian Ren, Chong Zhan, Zhenhong Zhou, Junhao Wang, He Zhu, Wangchunshu Zhou, and Shuicheng Yan. 2025a. Memevolve: Meta-evolution of agent memory systems. arXiv preprint arXiv:2512.18746. — [arXiv:2512.18746](https://arxiv.org/abs/2512.18746) find
  31. Guibin Zhang, Yanwei Yue, Zhixun Li, Sukwon Yun, Guancheng Wan, Kun Wang, Dawei Cheng, Jeffrey Xu Yu, and Tianlong Chen. 2025b. Cut the crap: An economical communication pipeline for llm-based multi-agent systems. In The Thirteenth International Conference on Learning Representations (ICLR). find
  32. Guibin Zhang, Yanwei Yue, Xiangguo Sun, Guancheng Wan, Miao Yu, Junfeng Fang, Kun Wang, Dawei Cheng, and Tianlong Chen. 2025c. G-designer: Architecting multi-agent communication topologies via graph neural networks. In Proceedings of the 42nd International Conference on Machine Learning. find
  33. Jiayi Zhang, Jinyu Xiang, Zhaoyang Yu, Fengwei Teng, Xionghui Chen, Jiaqi Chen, Mingchen Zhuge, Xiao Cheng, Sirui Hong, Jinlin Wang, and 1 others. 2025d. Aflow: Automating agentic workflow generation. In The Thirteenth International Conference on Learning Representations (ICLR). Oral Presentation.
  34. Chulun Zhou, Chunkang Zhang, Guoxin Yu, Fandong Meng, Jie Zhou, Wai Lam, and Mo Yu. 2025. Improving multi-step rag with hypergraph-based memory for long-context complex relational modeling. arXiv preprint arXiv:2512.23959. — [arXiv:2512.23959](https://arxiv.org/abs/2512.23959) find
  35. Mingchen Zhuge, Wenyi Wang, Louis Kirsch, Francesco Faccio, Dmitrii Khizbullin, and Jürgen Schmidhuber. 2024. Gptswarm: Language agents as optimizable graphs. In Proceedings of the 41st International Conference on Machine Learning, PMLR, pages 62743–62767.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhang-2026-evohyper-evolving-hypergraph-topologies-unified/