---
title: "Trajectory2Task — paper by Wang et al. (2026)"
description: "The authors propose Trajectory2Task, a pipeline that works trajectory-first, the reverse of the usual… 34 references, cited by 2 papers here."
authors: ["Ziyi Wang", "Yuxuan Lu", "Yimeng Zhang", "Pei Chen", "Ziwei Dong", "Jing Huang", "Jiri Gesi", "Xianfeng Tang", "Chen Luo", "Qun Liu", "Yisi Sang", "Hanqing Lu", "Manling Li", "Jin Lai", "Dakuo Wang"]
year: 2026
venue: "Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"
source: "https://aclanthology.org/2026.acl-long.2037/"
doi: "10.18653/v1/2026.acl-long.2037"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/wang-2026-trajectory2task-training-robust-tool-calling/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/wang-2026-trajectory2task-training-robust-tool-calling/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/9/) / Trajectory2Task: Training Robust Tool-Calling Agents with Synthesized Yet Verifiable Data for Complex User Intents

# Trajectory2Task: Training Robust Tool-Calling Agents with Synthesized Yet Verifiable Data for Complex User Intents

**Authors:** [Ziyi Wang](https://md.research.elamur.ai/free/en/authors/ziyi-wang/), [Yuxuan Lu](https://md.research.elamur.ai/free/en/authors/yuxuan-lu/), [Yimeng Zhang](https://md.research.elamur.ai/free/en/authors/yimeng-zhang/), [Pei Chen](https://md.research.elamur.ai/free/en/authors/pei-chen/), [Ziwei Dong](https://md.research.elamur.ai/free/en/authors/ziwei-dong/), [Jing Huang](https://md.research.elamur.ai/free/en/authors/jing-huang/), [Jiri Gesi](https://md.research.elamur.ai/free/en/authors/jiri-gesi/), [Xianfeng Tang](https://md.research.elamur.ai/free/en/authors/xianfeng-tang/), [Chen Luo](https://md.research.elamur.ai/free/en/authors/chen-luo/), [Qun Liu](https://md.research.elamur.ai/free/en/authors/qun-liu/), [Yisi Sang](https://md.research.elamur.ai/free/en/authors/yisi-sang/), [Hanqing Lu](https://md.research.elamur.ai/free/en/authors/hanqing-lu/), [Manling Li](https://md.research.elamur.ai/free/en/authors/manling-li/), Jin Lai and [Dakuo Wang](https://md.research.elamur.ai/free/en/authors/dakuo-wang/) · **Year:** 2026 · **Venue:** [Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)](https://md.research.elamur.ai/free/en/venues/acl/) ([ACL 2026](https://md.research.elamur.ai/free/en/venues/acl/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.acl-long.2037/) · [PDF](https://aclanthology.org/2026.acl-long.2037.pdf) · [DOI 10.18653/v1/2026.acl-long.2037](https://doi.org/10.18653/v1/2026.acl-long.2037) · [OpenAlex](https://openalex.org/W7166797679)

**Citations:** 0 citations (OpenAlex, 2026-10-14) · [2](https://md.research.elamur.ai/free/en/papers/wang-2026-trajectory2task-training-robust-tool-calling/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Tool-calling agents used in customer service often meet users whose requests lack key details, shift during the conversation, or break system policy. Existing benchmarks and training sets rarely cover these cases. The authors propose Trajectory2Task, a pipeline that works trajectory-first, the reverse of the usual task-first approach. A strong LLM agent first explores a tool environment and records valid multi-turn tool-call traces. The pipeline then rewrites those traces into user-facing tasks with controlled ambiguous, changing, or infeasible intents, and an LLM validates the results. Since every task is paired with an attainable environment state and a reference trajectory known to be correct, it can be checked automatically for both evaluation and training.

Built on the τ²-bench retail environment, the resulting Retail-3I data shows that current LLMs often fail under these intents, especially smaller models. Fine-tuning lightweight Qwen3 models on fewer than 3k successful trajectories improves robustness in all three scenarios. The gains also carry over to an unseen airline domain, which suggests the models learn general tool-use decisions rather than memorizing one domain.

## Key points
- **Three stress scenarios:** Ambiguous Intent (required slots are missing), Changing Intent (the goal drifts or branches into several issues), and Infeasible Intent (requests violate tool constraints or policy, including attempts to talk the agent past rules).
- **Benchmark failures:**
  - Moving from GENERAL to AMBIGUOUS, Qwen3-4B falls from 0.477 to 0.259 Pass^1, while Qwen3-235B-A22B barely changes (0.775 → 0.767).
  - In CHANGING, the best model reaches only 0.675 Pass^1.
  - In INFEASIBLE, even top models stay at or below 0.571 Pass^1 and 0.448 Pass^3, for example by calling `cancel_pending_order` on orders already delivered.
- **Training setup:** Qwen3-4B and Qwen3-8B were trained with SFT on 2,872 generated multi-turn trajectories. The largest gains came in CHANGING and INFEASIBLE, and the tuned 4B model matched or beat larger baseline models.
- **Results on the original τ²-bench (Pass^1):**
  - Retail: Qwen3-8B improves from 0.436 to 0.582, and Qwen3-4B from 0.363 to 0.532.
  - Airline (unseen during training): Qwen3-8B improves from 0.207 to 0.287, and Qwen3-4B from 0.187 to 0.313.
- **Limitations:** The simulated user is driven by an LLM through prompting, not by real conversation logs. The pipeline has so far been tested only on τ²-bench, mainly because of cost.

*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

- [Northeastern University](https://md.research.elamur.ai/free/en/universities/northeastern-university/) — University, Boston, United States · 57 works · 6,533 citations
- [Northwestern University](https://md.research.elamur.ai/free/en/universities/northwestern-university/) — University, Evanston, United States · 41 works · 2,194 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

## Cite

Cite as: Ziyi Wang, Yuxuan Lu, Yimeng Zhang, Pei Chen, Ziwei Dong, Jing Huang, Jiri Gesi, Xianfeng Tang, Chen Luo, Qun Liu, Yisi Sang, Hanqing Lu, Manling Li, Jin Lai and Dakuo Wang. 2026. Trajectory2Task: Training Robust Tool-Calling Agents with Synthesized Yet Verifiable Data for Complex User Intents. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{wang2026trajectory2task,
  title = {{Trajectory2Task: Training Robust Tool-Calling Agents with Synthesized Yet Verifiable Data for Complex User Intents}},
  author = {Ziyi Wang, Yuxuan Lu, Yimeng Zhang, Pei Chen, Ziwei Dong, Jing Huang, Jiri Gesi, Xianfeng Tang, Chen Luo, Qun Liu, Yisi Sang, Hanqing Lu, Manling Li, Jin Lai and Dakuo Wang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.2037},
  url = {https://doi.org/10.18653/v1/2026.acl-long.2037}
}
```

- [← previous Tracking the evolution of LLM capabilities for Belarusian with OpenAI Evals](https://md.research.elamur.ai/free/en/papers/poritski-2026-tracking-evolution-llm-capabilities-belarusian/)
- [next → UrBLiMP: A Benchmark for Evaluating the Linguistic Competence of Large Language Models in Urdu](https://md.research.elamur.ai/free/en/papers/adeeba-2026-urblimp-benchmark-evaluating-linguistic-competence/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/9/)

## References

34 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Mohammad Aliannejadi, Hamed Zamani, Fabio Crestani, and W Bruce Croft. 2019. Asking clarifying questions in open-domain information-seeking conversations. In Proceedings of the 42nd international acm sigir conference on research and development in information retrieval, pages 475–484. [find](https://scholar.google.com/scholar?q=Mohammad+Aliannejadi%2C+Hamed+Zamani%2C+Fabio+Crestani%2C+and+W+Bruce+Croft%2E+2019%2E+Asking+clarifying+questions+in+open%2Ddomain+information%2Dseeking+conversations%2E+In+Proceedings+of+the+42nd+international+acm+)
2. Victor Barres, Honghua Dong, Soham Ray, Xujie Si, and Karthik Narasimhan. 2025. τ 2 -bench: Evaluating conversational agents in a dual-control environment. arXiv preprint arXiv:2506.07982. [find](https://scholar.google.com/scholar?q=Victor+Barres%2C+Honghua+Dong%2C+Soham+Ray%2C+Xujie+Si%2C+and+Karthik+Narasimhan%2E+2025%2E+%CF%84+2+%2Dbench%3A+Evaluating+conversational+agents+in+a+dual%2Dcontrol+environment%2E+arXiv+preprint+arXiv%3A2506%2E07982%2E)
3. Andres M Bran, Sam Cox, Oliver Schilter, Carlo Baldassari, Andrew D White, and Philippe Schwaller. 2023. Chemcrow: Augmenting large-language models with chemistry tools. arXiv preprint arXiv:2304.05376. [find](https://scholar.google.com/scholar?q=Andres+M+Bran%2C+Sam+Cox%2C+Oliver+Schilter%2C+Carlo+Baldassari%2C+Andrew+D+White%2C+and+Philippe+Schwaller%2E+2023%2E+Chemcrow%3A+Augmenting+large%2Dlanguage+models+with+chemistry+tools%2E+arXiv+preprint+arXiv%3A2304%2E0537)
4. Carlos E Jimenez, John Yang, Alexander Wettig, Shunyu Yao, Kexin Pei, Ofir Press, and Karthik Narasimhan. 2023. [Swe-bench: Can language models resolve real-world github issues?](https://md.research.elamur.ai/free/en/papers/jimenez-2024-swe-bench-can-language-models/) arXiv preprint arXiv:2310.06770.
5. Stefan Larson, Anish Mahendran, Joseph J Peper, Christopher Clarke, Andrew Lee, Parker Hill, Jonathan K Kummerfeld, Kevin Leach, Michael A Laurenzano, Lingjia Tang, and 1 others. 2019. An evaluation dataset for intent classification and out-of-scope prediction. arXiv preprint arXiv:1909.02027. [find](https://scholar.google.com/scholar?q=Stefan+Larson%2C+Anish+Mahendran%2C+Joseph+J+Peper%2C+Christopher+Clarke%2C+Andrew+Lee%2C+Parker+Hill%2C+Jonathan+K+Kummerfeld%2C+Kevin+Leach%2C+Michael+A+Laurenzano%2C+Lingjia+Tang%2C+and+1+others%2E+2019%2E+An+evaluation+d)
6. 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](https://md.research.elamur.ai/free/en/papers/li-2023-api-bank-comprehensive-benchmark-tool/). arXiv preprint arXiv:2304.08244.
7. Haipeng Luo, Qingfeng Sun, Can Xu, Pu Zhao, Jian-guang Lou, Chongyang Tao, Xiubo Geng, Qingwei Lin, Shifeng Chen, and Dongmei Zhang. 2023. [Wizardmath: Empowering mathematical reasoning for large language models via reinforced evol-instruct](https://md.research.elamur.ai/free/en/papers/luo-2023-wizardmath-empowering-mathematical-reasoning-large/). arXiv preprint arXiv:2308.09583.
8. Ziyang Luo, Zhiqi Shen, Wenzhuo Yang, Zirui Zhao, Prathyusha Jwalapuram, Amrita Saha, Doyen Sahoo, Silvio Savarese, Caiming Xiong, and Junnan Li. 2025. Mcp-universe: Benchmarking large language models with real-world model context protocol servers. arXiv preprint arXiv:2508.14704. — [arXiv:2508.14704](https://arxiv.org/abs/2508.14704) [find](https://scholar.google.com/scholar?q=Ziyang+Luo%2C+Zhiqi+Shen%2C+Wenzhuo+Yang%2C+Zirui+Zhao%2C+Prathyusha+Jwalapuram%2C+Amrita+Saha%2C+Doyen+Sahoo%2C+Silvio+Savarese%2C+Caiming+Xiong%2C+and+Junnan+Li%2E+2025%2E+Mcp%2Duniverse%3A+Benchmarking+large+language+models)
9. Grégoire Mialon, Clémentine Fourrier, Thomas Wolf, Yann LeCun, and Thomas Scialom. 2023. [Gaia: a benchmark for general ai assistants](https://md.research.elamur.ai/free/en/papers/mialon-2023-gaia-benchmark-general-ai-assistants/). In The Twelfth International Conference on Learning Representations.
10. Guozhao Mo, Wenliang Zhong, Jiawei Chen, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, and Le Sun. 2025. Livemcpbench: Can agents navigate an ocean of mcp tools? arXiv preprint arXiv:2508.01780. — [arXiv:2508.01780](https://arxiv.org/abs/2508.01780) [find](https://scholar.google.com/scholar?q=Guozhao+Mo%2C+Wenliang+Zhong%2C+Jiawei+Chen%2C+Xuanang+Chen%2C+Yaojie+Lu%2C+Hongyu+Lin%2C+Ben+He%2C+Xianpei+Han%2C+and+Le+Sun%2E+2025%2E+Livemcpbench%3A+Can+agents+navigate+an+ocean+of+mcp+tools%3F+arXiv+preprint+arXiv%3A2508%2E)
11. Shishir G Patil, Huanzhi Mao, Fanjia Yan, Charlie Cheng-Jie Ji, Vishnu Suresh, Ion Stoica, and Joseph E Gonzalez. The berkeley function calling leaderboard (bfcl): From tool use to agentic evaluation of large language models. In Forty-second International Conference on Machine Learning. [find](https://scholar.google.com/scholar?q=Shishir+G+Patil%2C+Huanzhi+Mao%2C+Fanjia+Yan%2C+Charlie+Cheng%2DJie+Ji%2C+Vishnu+Suresh%2C+Ion+Stoica%2C+and+Joseph+E+Gonzalez%2E+The+berkeley+function+calling+leaderboard+%28bfcl%29%3A+From+tool+use+to+agentic+evaluation+)
12. Shishir G Patil, Tianjun Zhang, Xin Wang, and Joseph E Gonzalez. 2024. [Gorilla: Large language model connected with massive apis](https://md.research.elamur.ai/free/en/papers/patil-2023-gorilla-large-language-model-connected/). Advances in Neural Information Processing Systems, 37:126544–126565.
13. Akshara Prabhakar, Zuxin Liu, Ming Zhu, Jianguo Zhang, Tulika Awalgaonkar, Shiyu Wang, Zhiwei Liu, Haolin Chen, Thai Hoang, Juan Carlos Niebles, and 1 others. 2025. [Apigen-mt: Agentic pipeline for multi-turn data generation via simulated agent-human interplay](https://md.research.elamur.ai/free/en/papers/prabhakar-2025-apigen-mt-agentic-pipeline-multi/). arXiv preprint arXiv:2504.03601.
14. Cheng Qian, Bingxiang He, Zhong Zhuang, Jia Deng, Yujia Qin, Xin Cong, Zhong Zhang, Jie Zhou, Yankai Lin, Zhiyuan Liu, and 1 others. 2024. Tell me more! towards implicit user intention understanding of language model driven agents. arXiv preprint arXiv:2402.09205. [find](https://scholar.google.com/scholar?q=Cheng+Qian%2C+Bingxiang+He%2C+Zhong+Zhuang%2C+Jia+Deng%2C+Yujia+Qin%2C+Xin+Cong%2C+Zhong+Zhang%2C+Jie+Zhou%2C+Yankai+Lin%2C+Zhiyuan+Liu%2C+and+1+others%2E+2024%2E+Tell+me+more%21+towards+implicit+user+intention+understanding+o)
15. Cheng Qian, Zuxin Liu, Akshara Prabhakar, Zhiwei Liu, Jianguo Zhang, Haolin Chen, Heng Ji, Weiran Yao, Shelby Heinecke, Silvio Savarese, and 1 others. 2025. Userbench: An interactive gym environment for user-centric agents. arXiv preprint arXiv:2507.22034. — [arXiv:2507.22034](https://arxiv.org/abs/2507.22034) [find](https://scholar.google.com/scholar?q=Cheng+Qian%2C+Zuxin+Liu%2C+Akshara+Prabhakar%2C+Zhiwei+Liu%2C+Jianguo+Zhang%2C+Haolin+Chen%2C+Heng+Ji%2C+Weiran+Yao%2C+Shelby+Heinecke%2C+Silvio+Savarese%2C+and+1+others%2E+2025%2E+Userbench%3A+An+interactive+gym+environment+f)
16. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, and 1 others. 2023. [Toolllm: Facilitating large language models to master 16000+ real-world apis](https://md.research.elamur.ai/free/en/papers/qin-2023-toolllm-facilitating-large-language-models/). arXiv preprint arXiv:2307.16789.
17. Timo Schick, Jane Dwivedi-Yu, Roberto Dessì, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. 2023. [Toolformer: Language models can teach themselves to use tools](https://md.research.elamur.ai/free/en/papers/schick-2023-toolformer-language-models-can-teach/). Advances in Neural Information Processing Systems, 36:68539–68551.
18. Jeonghoon Shim, Woojung Song, Cheyon Jin, Seung-won KooK, and Yohan Jo. 2025. Non-collaborative user simulators for tool agents. arXiv preprint arXiv:2509.23124. — [arXiv:2509.23124](https://arxiv.org/abs/2509.23124) [find](https://scholar.google.com/scholar?q=Jeonghoon+Shim%2C+Woojung+Song%2C+Cheyon+Jin%2C+Seung%2Dwon+KooK%2C+and+Yohan+Jo%2E+2025%2E+Non%2Dcollaborative+user+simulators+for+tool+agents%2E+arXiv+preprint+arXiv%3A2509%2E23124%2E+%E2%80%94+%5BarXiv%3A2509%2E23124%5D%28https%3A%2F%2Farxiv%2Eorg)
19. Qiaoyu Tang, Ziliang Deng, Hongyu Lin, Xianpei Han, Qiao Liang, Boxi Cao, and Le Sun. 2023. [Toolalpaca: Generalized tool learning for language models with 3000 simulated cases](https://md.research.elamur.ai/free/en/papers/tang-2023-toolalpaca-generalized-tool-learning-language/). arXiv preprint arXiv:2306.05301.
20. Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yannic Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori Hashimoto. 2023. Alpaca: A strong, replicable instruction-following model. Stanford University Center for Research on Foundation Models (CRFM) Technical Report. [find](https://scholar.google.com/scholar?q=Rohan+Taori%2C+Ishaan+Gulrajani%2C+Tianyi+Zhang%2C+Yannic+Dubois%2C+Xuechen+Li%2C+Carlos+Guestrin%2C+Percy+Liang%2C+and+Tatsunori+Hashimoto%2E+2023%2E+Alpaca%3A+A+strong%2C+replicable+instruction%2Dfollowing+model%2E+Stanford+)
21. Harsh Trivedi, Tushar Khot, Mareike Hartmann, Ruskin Manku, Vinty Dong, Edward Li, Shashank Gupta, Ashish Sabharwal, and Niranjan Balasubramanian. 2024. [Appworld: A controllable world of apps and people for benchmarking interactive coding agents](https://md.research.elamur.ai/free/en/papers/trivedi-2024-appworld-controllable-world-apps-people/). arXiv preprint arXiv:2407.18901.
22. Wenhao Wang, Peizhi Niu, Zhao Xu, Zhaoyu Chen, Jian Du, Yaxin Du, Xianghe Pang, Keduan Huang, Yanfeng Wang, Qiang Yan, and 1 others. 2025a. [Mcp-flow: Facilitating llm agents to master real-world, diverse and scaling mcp tools](https://md.research.elamur.ai/free/en/papers/wang-2026-mcp-flow-facilitating-llm-agents/). arXiv preprint arXiv:2510.24284.
23. Yimeng Zhang, Tian Wang, Jiri Gesi, Ziyi Wang, Yuxuan Lu, Jiacheng Lin, Sinong Zhan, Vianne Gao, Ruochen Jiao, Junze Liu, and 1 others. 2025b. Shop-r1: Rewarding llms to simulate human behavior in online shopping via reinforcement learning. arXiv preprint arXiv:2507.17842. — [arXiv:2507.17842](https://arxiv.org/abs/2507.17842) [find](https://scholar.google.com/scholar?q=Yimeng+Zhang%2C+Tian+Wang%2C+Jiri+Gesi%2C+Ziyi+Wang%2C+Yuxuan+Lu%2C+Jiacheng+Lin%2C+Sinong+Zhan%2C+Vianne+Gao%2C+Ruochen+Jiao%2C+Junze+Liu%2C+and+1+others%2E+2025b%2E+Shop%2Dr1%3A+Rewarding+llms+to+simulate+human+behavior+in+onl)
24. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. 2022. [Self-consistency improves chain of thought reasoning in language models](https://md.research.elamur.ai/free/en/papers/wang-2022-self-consistency-improves-chain-thought/). arXiv preprint arXiv:2203.11171.
25. Zheng Zhang, Ryuichi Takanobu, Qi Zhu, MinLie Huang, and XiaoYan Zhu. 2020. Recent advances and challenges in task-oriented dialog systems. Science China Technological Sciences, 63(10):2011– 2027. [find](https://scholar.google.com/scholar?q=Zheng+Zhang%2C+Ryuichi+Takanobu%2C+Qi+Zhu%2C+MinLie+Huang%2C+and+XiaoYan+Zhu%2E+2020%2E+Recent+advances+and+challenges+in+task%2Doriented+dialog+systems%2E+Science+China+Technological+Sciences%2C+63%2810%29%3A2011%E2%80%93+2027%2E)
26. Ziyi Wang, Yuxuan Lu, Wenbo Li, Amirali Amini, Bo Sun, Yakov Bart, Weimin Lyu, Jiri Gesi, Tian Wang, Jing Huang, and 1 others. 2025b. [Opera: A dataset of observation, persona, rationale, and action for evaluating llms on human online shopping behavior simulation](https://md.research.elamur.ai/free/en/papers/wang-2026-opera-dataset-observation-persona-rationale/). arXiv preprint arXiv:2506.05606.
27. Ziyi Wang, Yuxuan Lu, Yimeng Zhang, Jing Huang, and Dakuo Wang. 2025c. Customer-r1: Personalized simulation of human behaviors via rl-based llm agent in online shopping. arXiv preprint arXiv:2510.07230. — [arXiv:2510.07230](https://arxiv.org/abs/2510.07230) [find](https://scholar.google.com/scholar?q=Ziyi+Wang%2C+Yuxuan+Lu%2C+Yimeng+Zhang%2C+Jing+Huang%2C+and+Dakuo+Wang%2E+2025c%2E+Customer%2Dr1%3A+Personalized+simulation+of+human+behaviors+via+rl%2Dbased+llm+agent+in+online+shopping%2E+arXiv+preprint+arXiv%3A2510%2E0723)
28. 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](https://md.research.elamur.ai/free/en/papers/wei-2022-chain-thought-prompting-elicits-reasoning/). Advances in neural information processing systems, 35:24824– 24837.
29. Chien-Sheng Wu, Andrea Madotto, Ehsan Hosseini-Asl, Caiming Xiong, Richard Socher, and Pascale Fung. 2019. [Transferable multi-domain state generator for task-oriented dialogue systems](https://md.research.elamur.ai/free/en/papers/wu-2019-transferable-multi-domain-state-generator/). arXiv preprint arXiv:1905.08743.
30. Can Xu, Qingfeng Sun, Kai Zheng, Xiubo Geng, Pu Zhao, Jiazhan Feng, Chongyang Tao, Qingwei Lin, and Daxin Jiang. 2024a. [Wizardlm: Empowering large pre-trained language models to follow complex instructions](https://md.research.elamur.ai/free/en/papers/xu-2023-wizardlm-empowering-large-pre-trained/). In The Twelfth International Conference on Learning Representations.
31. Xiaohan Xu, Ming Li, Chongyang Tao, Tao Shen, Reynold Cheng, Jinyang Li, Can Xu, Dacheng Tao, and Tianyi Zhou. 2024b. A survey on knowledge distillation of large language models. arXiv preprint arXiv:2402.13116. — [arXiv:2402.13116](https://arxiv.org/abs/2402.13116) [find](https://scholar.google.com/scholar?q=Xiaohan+Xu%2C+Ming+Li%2C+Chongyang+Tao%2C+Tao+Shen%2C+Reynold+Cheng%2C+Jinyang+Li%2C+Can+Xu%2C+Dacheng+Tao%2C+and+Tianyi+Zhou%2E+2024b%2E+A+survey+on+knowledge+distillation+of+large+language+models%2E+arXiv+preprint+arXiv%3A)
32. Zhangchen Xu, Adriana Meza Soria, Shawn Tan, Anurag Roy, Ashish Sunil Agrawal, Radha Poovendran, and Rameswar Panda. 2025. Toucan: Synthesizing 1.5 m tool-agentic data from real-world mcp environments. arXiv preprint arXiv:2510.01179. — [arXiv:2510.01179](https://arxiv.org/abs/2510.01179) [find](https://scholar.google.com/scholar?q=Zhangchen+Xu%2C+Adriana+Meza+Soria%2C+Shawn+Tan%2C+Anurag+Roy%2C+Ashish+Sunil+Agrawal%2C+Radha+Poovendran%2C+and+Rameswar+Panda%2E+2025%2E+Toucan%3A+Synthesizing+1%2E5+m+tool%2Dagentic+data+from+real%2Dworld+mcp+environments)
33. Shunyu Yao, Noah Shinn, Pedram Razavi, and Karthik Narasimhan. 2024. [τ -bench: A benchmark for tool-agent-user interaction in real-world domains](https://md.research.elamur.ai/free/en/papers/yao-2024-tau-bench-benchmark-tool/). arXiv preprint arXiv:2406.12045.
34. Yimeng Zhang, Jiri Gesi, Ran Xue, Tian Wang, Ziyi Wang, Yuxuan Lu, Sinong Zhan, Huimin Zeng, Qingjun Cui, Yufan Guo, and 1 others. 2025a. See, think, act: Online shopper behavior simulation with vlm agents. arXiv preprint arXiv:2510.19245. — [arXiv:2510.19245](https://arxiv.org/abs/2510.19245) [find](https://scholar.google.com/scholar?q=Yimeng+Zhang%2C+Jiri+Gesi%2C+Ran+Xue%2C+Tian+Wang%2C+Ziyi+Wang%2C+Yuxuan+Lu%2C+Sinong+Zhan%2C+Huimin+Zeng%2C+Qingjun+Cui%2C+Yufan+Guo%2C+and+1+others%2E+2025a%2E+See%2C+think%2C+act%3A+Online+shopper+behavior+simulation+with+vlm+a)

*Updated 31 October 2026*