Elamur Research⌕
☰
home / papers / 2025 / xLAM: A Family of Large Action Models to Empower AI Agent Systems

xLAM: A Family of Large Action Models to Empower AI Agent Systems

Paper · 2025 · Jianguo Zhang, Tian Lan, Ming Zhu et al.

Summary

Open-source LLMs lag behind proprietary systems on agent tasks such as tool use. A major cause is that public agent datasets are scarce, inconsistent in format, homogeneous in content, and often noisy, with problems like hallucinated actions or repeated turns. The authors (Salesforce AI) release xLAM, a family of five "large action models" with dense and mixture-of-experts architectures, ranging from 1B to 8x22B parameters. The smaller models target on-device use, and the larger ones target harder tasks.

Training relies on a data pipeline with several steps:

  • converting heterogeneous agent data into one modular, function-calling-style format;
  • augmenting it to add diversity;
  • checking and fixing quality issues;
  • synthesizing new function-calling data with strong LLMs;
  • generating preference data.

Models are trained with SFT and DPO, and evaluation results feed back into data improvement. On the Berkeley Function-Calling Leaderboard (BFCL-v2), the largest model ranked first, ahead of GPT-4 and Claude-3 variants. Compact xLAM models rivalled much bigger systems. The work shows that careful data engineering can give the community competitive open agent models.

Key points

  • Models and training. Five models are released: xLAM-8x22b-r, xLAM-8x7b-r, xLAM-7b-r, xLAM-7b-fc-r and xLAM-1b-fc-r. They are trained via SFT (full and LoRA) and DPO (LoRA).
  • Unified data format. Every sample is broken into separate modules: a description of the task, the set of tools the model may call, guidance on the expected output format, a handful of demonstration examples, the query, and the steps. Augmentation includes:
    • shuffling the order of tools, fields and prompt sections;
    • varying concatenation-token styles;
    • having LLMs rephrase task instructions, with verification.
  • BFCL-v2 results (cutoff 09/19/2024). xLAM-8x22b-r ranked 1st with 87.31 overall accuracy, versus 85.79 for GPT-4-0125-Preview (Prompt). xLAM models held 4 of the top 20 spots. xLAM-1b-fc-r scored 75.43, slightly above GPT-3.5-Turbo (FC).
  • Ablation on 7B models. Compared with raw data, augmentation improved ToolBench by 2.3%, Webshop by 5.8% and ToolQuery by 18.3%. Adding cleaning raised ToolQuery by a further 23.4%.
  • Limitations. Data synthesis depends on large LLMs (DeepSeek-V2, Mixtral-8x22b-Inst). Benchmark coverage was constrained by environment instability and the cost of commercial API calls.

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

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 9,086 worksArtificial Intelligence · 8,889 worksArts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksComputer Vision and Pattern Recognition · 1,282 works

Cite

Cite as Jianguo Zhang, Tian Lan, Ming Zhu, Zuxin Liu, Thai Hoang, Shirley Kokane, Weiran Yao, Juntao Tan, Zhiwei Liu, Yihao Feng, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Silvio Savarese and Caiming Xiong. 2025. xLAM: A Family of Large Action Models to Empower AI Agent Systems. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.naacl-long.578

@inproceedings{zhang2025xlam,
  title = {{xLAM: A Family of Large Action Models to Empower AI Agent Systems}},
  author = {Jianguo Zhang, Tian Lan, Ming Zhu, Zuxin Liu, Thai Hoang, Shirley Kokane, Weiran Yao, Juntao Tan, Zhiwei Liu, Yihao Feng, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Silvio Savarese and Caiming Xiong},
  booktitle = {Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)},
  year = {2025},
  doi = {10.18653/v1/2025.naacl-long.578},
  url = {https://doi.org/10.18653/v1/2025.naacl-long.578}
}

References

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

  1. AI@Meta. 2024. Llama 3 model card. find
  2. AI Anthropic. 2024. The claude 3 model family: Opus, sonnet, haiku. Claude-3 Model Card. find
  3. Xiao Bi, Deli Chen, Guanting Chen, Shanhuang Chen, Damai Dai, Chengqi Deng, Honghui Ding, Kai Dong, Qiushi Du, Zhe Fu, et al. 2024. Deepseek llm: Scaling open-source language models with longtermism. arXiv preprint arXiv:2401.02954. find
  4. 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.
  5. Zehui Chen, Kuikun Liu, Qiuchen Wang, Wenwei Zhang, Jiangning Liu, Dahua Lin, Kai Chen, and Feng Zhao. 2024. Agent-flan: Designing data and methods of effective agent tuning for large language models. arXiv preprint arXiv:2403.12881.
  6. Charlie Cheng-Jie Ji, Huanzhi Mao, Fanjia Yan, Shishir Patil, Tianjun Zhang, Ion Stoica, and Joseph Gonzalez. 2024. Gorilla openfunctions v2. In https://gorilla.cs.berkeley.edu//blogs/ 7_open_functions_v2.html. ↗
  7. DeepSeek-AI. 2024. Deepseek-v2: A strong, economical, and efficient mixture-of-experts language model. Preprint, arXiv:2405.04434. find
  8. Xiang Deng, Yu Gu, Boyuan Zheng, Shijie Chen, Samuel Stevens, Boshi Wang, Huan Sun, and Yu Su. 2023. Mind2web: Towards a generalist agent for the web. arXiv preprint arXiv:2306.06070.
  9. Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2023. Qlora: Efficient finetuning of quantized llms. Preprint, arXiv:2305.14314. find
  10. Yu Du, Fangyun Wei, and Hongyang Zhang. 2024. Anytool: Self-reflective, hierarchical agents for large-scale api calls. arXiv preprint arXiv:2402.04253. — [arXiv:2402.04253](https://arxiv.org/abs/2402.04253) find
  11. Sylvain Gugger, Lysandre Debut, Thomas Wolf, Philipp Schmid, Zachary Mueller, Sourab Mangrulkar, Marc Sun, and Benjamin Bossan. 2022. Accelerate: Training and inference at scale made simple, efficient and adaptable. https://github.com/huggingface/ accelerate. ↗
  12. Daya Guo, Qihao Zhu, Dejian Yang, Zhenda Xie, Kai Dong, Wentao Zhang, Guanting Chen, Xiao Bi, Y Wu, YK Li, et al. 2024. Deepseek-coder: When the large language model meets programming–the rise of code intelligence. arXiv preprint arXiv:2401.14196. find
  13. Sirui Hong, Mingchen Zhuge, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Ceyao Zhang, Jinlin Wang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu, and Jürgen Schmidhuber. 2023. Metagpt: Meta programming for a multi-agent collaborative framework. Preprint…
  14. Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685.
  15. Yue Huang, Jiawen Shi, Yuan Li, Chenrui Fan, Siyuan Wu, Qihui Zhang, Yixin Liu, Pan Zhou, Yao Wan, Neil Zhenqiang Gong, et al. 2023. Metatool benchmark for large language models: Deciding whether to use tools and which to use. arXiv preprint arXiv:2310.03128. find
  16. Albert Q Jiang, Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Emma Bou Hanna, Florian Bressand, et al. 2024. Mixtral of experts. arXiv preprint arXiv:2401.04088. find
  17. 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. In The 2023 Conference on Empirical Methods in Natural Language Processing.
  18. Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, et al. 2023a. Agentbench: Evaluating llms as agents. arXiv preprint arXiv:2308.03688.
  19. Zhiwei Liu, Weiran Yao, Jianguo Zhang, Le Xue, Shelby Heinecke, Rithesh Murthy, Yihao Feng, Zeyuan Chen, Juan Carlos Niebles, Devansh Arpit, et al. 2023b. Bolaa: Benchmarking and orchestrating llm-augmented autonomous agents. arXiv preprint arXiv:2308.05960. find
  20. Zhiwei Liu, Weiran Yao, Jianguo Zhang, Liangwei Yang, Zuxin Liu, Juntao Tan, Prafulla K Choubey, Tian Lan, Jason Wu, Huan Wang, et al. 2024a. Agentlite: A lightweight library for building and advancing task-oriented llm agent system. arXiv preprint arXiv:2402.15538. — [arXiv:2402.15538](https://arxiv.org/abs/2402.15538) find
  21. Zuxin Liu, Thai Hoang, Jianguo Zhang, Ming Zhu, Tian Lan, Shirley Kokane, Juntao Tan, Weiran Yao, Zhiwei Liu, Yihao Feng, Rithesh Murthy, Liangwei Yang, Silvio Savarese, Juan Carlos Niebles, Huan Wang, Shelby Heinecke, and Caiming Xiong. 2024b. Apigen: Automated pipeline for generating verifiable an… find
  22. Zuxin Liu, Jesse Zhang, Kavosh Asadi, Yao Liu, Ding Zhao, Shoham Sabach, and Rasool Fakoor. 2023c. Tail: Task-specific adapters for imitation learning with large pretrained models. arXiv preprint arXiv:2310.05905. — [arXiv:2310.05905](https://arxiv.org/abs/2310.05905) find
  23. Shayne Longpre, Robert Mahari, Anthony Chen, Naana Obeng-Marnu, Damien Sileo, William Brannon, Niklas Muennighoff, Nathan Khazam, Jad Kabbara, Kartik Perisetla, et al. 2023. The data provenance initiative: A large scale audit of dataset licensing & attribution in ai. arXiv preprint arXiv:2310.16787. — [arXiv:2310.16787](https://arxiv.org/abs/2310.16787) find
  24. Shayne Longpre, Robert Mahari, Ariel Lee, Campbell Lund, Hamidah Oderinwale, William Brannon, Nayan Saxena, Naana Obeng-Marnu, Tobin South, Cole Hunter, et al. 2024. Consent in crisis: The rapid decline of the ai data commons. arXiv preprint arXiv:2407.14933. — [arXiv:2407.14933](https://arxiv.org/abs/2407.14933) find
  25. Chang Ma, Junlei Zhang, Zhihao Zhu, Cheng Yang, Yujiu Yang, Yaohui Jin, Zhenzhong Lan, Lingpeng Kong, and Junxian He. 2024. Agentboard: An analytical evaluation board of multi-turn llm agents. arXiv preprint arXiv:2401.13178. find
  26. OpenAI. 2023. Gpt-4 technical report. ArXiv.
  27. Shishir G. Patil, Tianjun Zhang, Xin Wang, and Joseph E. Gonzalez. 2023. Gorilla: Large language model connected with massive apis. arXiv preprint arXiv:2305.15334.
  28. Yujia Qin, Shihao Liang, Yining Ye, Kunlun Zhu, Lan Yan, Yaxi Lu, Yankai Lin, Xin Cong, Xiangru Tang, Bill Qian, et al. 2024. Toolllm: Facilitating large language models to master 16000+ real-world apis. ICLR.
  29. Rafael Rafailov, Archit Sharma, Eric Mitchell, Stefano Ermon, Christopher D Manning, and Chelsea Finn. 2023. Direct preference optimization: Your language model is secretly a reward model. arXiv preprint arXiv:2305.18290.
  30. Machel Reid, Nikolay Savinov, Denis Teplyashin, Dmitry Lepikhin, Timothy Lillicrap, Jean-baptiste Alayrac, Radu Soricut, Angeliki Lazaridou, Orhan Firat, Julian Schrittwieser, et al. 2024. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context. arXiv preprint arXiv:2403.…
  31. Baptiste Roziere, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, et al. 2023. Code llama: Open foundation models for code. arXiv preprint arXiv:2308.12950.
  32. Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik R Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. In Thirty-seventh Conference on Neural Information Processing Systems.
  33. Venkat Krishna Srinivasan, Zhen Dong, Banghua Zhu, Brian Yu, Damon Mosk-Aoyama, Kurt Keutzer, Jiantao Jiao, and Jian Zhang. 2023. Nexusraven: a commercially-permissive language model for function calling. In NeurIPS 2023 Foundation Models for Decision Making Workshop. find
  34. Zhi Rui Tam, Cheng-Kuang Wu, Yi-Lin Tsai, Chieh-Yen Lin, Hung-yi Lee, and Yun-Nung Chen. 2024. Let me speak freely? a study on the impact of format restrictions on performance of large language models. arXiv preprint arXiv:2408.02442. — [arXiv:2408.02442](https://arxiv.org/abs/2408.02442) find
  35. Qiaoyu Tang, Ziliang Deng, Hongyu Lin, Xianpei Han, Qiao Liang, and Le Sun. 2023. Toolalpaca: Generalized tool learning for language models with 3000 simulated cases. arXiv preprint arXiv:2306.05301.
  36. XAgent Team. 2023. Xagent: An autonomous agent for complex task solving. find
  37. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971.
  38. Jize Wang, Zerun Ma, Yining Li, Songyang Zhang, Cailian Chen, Kai Chen, and Xinyi Le. 2024. Gta: A benchmark for general tool agents. arXiv preprint arXiv:2407.08713. — [arXiv:2407.08713](https://arxiv.org/abs/2407.08713) find
  39. Xingyao Wang, Zihan Wang, Jiateng Liu, Yangyi Chen, Lifan Yuan, Hao Peng, and Heng Ji. 2023. Mint: Evaluating llms in multi-turn interaction with tools and language feedback. arXiv preprint arXiv:2309.10691.
  40. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems, 35:24824–24837.
  41. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, et al. 2020. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 conference on empirical methods in natural language p…
  42. 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 framework. arXiv preprint arXiv:2308.08155.
  43. Tianbao Xie, Fan Zhou, Zhoujun Cheng, Peng Shi, Luoxuan Weng, Yitao Liu, Toh Jing Hua, Junning Zhao, Qian Liu, Che Liu, et al. 2023. Openagents: An open platform for language agents in the wild. arXiv preprint arXiv:2310.10634. find
  44. Yiheng Xu, Hongjin Su, Chen Xing, Boyu Mi, Qian Liu, Weijia Shi, Binyuan Hui, Fan Zhou, Yitao Liu, Tianbao Xie, et al. 2023. Lemur: Harmonizing natural language and code for language agents. arXiv preprint arXiv:2310.06830. — [arXiv:2310.06830](https://arxiv.org/abs/2310.06830) find
  45. Fanjia Yan, Huanzhi Mao, Charlie Cheng-Jie Ji, Tianjun Zhang, Shishir G. Patil, Ion Stoica, and Joseph E. Gonzalez. 2024. Berkeley function calling leaderboard. https://gorilla.cs.berkeley.edu/blogs/8_berkeley_function_calling_ leaderboard.html. ↗
  46. Nguyen, Geeta Chauhan, Yuchen Hao, Ajit Mathews, and Shen Li. 2023a. Pytorch fsdp: Experiences on scaling fully sharded data parallel. arXiv preprint arXiv:2304.11277. find
  47. Yanli Zhao, Andrew Gu, Rohan Varma, Liang Luo, Chien-Chin Huang, Min Xu, Less Wright, Hamid Shojanazeri, Myle Ott, Sam Shleifer, Alban Desmaison, Can Balioglu, Pritam Damania, Bernard Nguyen, Geeta Chauhan, Yuchen Hao, Ajit Mathews, and Shen Li. 2023b. Pytorch fsdp: Experiences on scaling fully shar… find
  48. Shuyan Zhou, Frank F Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Yonatan Bisk, Daniel Fried, Uri Alon, et al. 2023. Webarena: A realistic web environment for building autonomous agents. arXiv preprint arXiv:2307.13854.
  49. Shunyu Yao, Howard Chen, John Yang, and Karthik Narasimhan. 2022. Webshop: Towards scalable real-world web interaction with grounded language agents. Advances in Neural Information Processing Systems, 35:20744–20757.
  50. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. ReAct: Synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR).
  51. Da Yin, Faeze Brahman, Abhilasha Ravichander, Khyathi Chandu, Kai-Wei Chang, Yejin Choi, and Bill Yuchen Lin. 2023. Lumos: Learning Agents with Unified Data, Modular Design, and Open-Source LLMs. arXiv preprint arXiv:2311.05657. find
  52. 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.
  53. Jianguo Zhang, Tian Lan, Rithesh Murthy, Zhiwei Liu, Weiran Yao, Juntao Tan, Thai Hoang, Liangwei Yang, Yihao Feng, Zuxin Liu, et al. 2024. Agentohana: Design unified data and training pipeline for effective agent learning. arXiv preprint arXiv:2402.15506. — [arXiv:2402.15506](https://arxiv.org/abs/2402.15506) find
  54. Jianguo Zhang, Kun Qian, Zhiwei Liu, Shelby Heinecke, Rui Meng, Ye Liu, Zhou Yu, Silvio Savarese, and Caiming Xiong. 2023. Dialogstudio: Towards richest and most diverse unified dataset collection for conversational ai. arXiv preprint arXiv:2307.10172. find
  55. Yanli Zhao, Andrew Gu, Rohan Varma, Liang Luo, Chien-Chin Huang, Min Xu, Less Wright, Hamid Shojanazeri, Myle Ott, Sam Shleifer, Alban Desmaison, Can Balioglu, Pritam Damania, Bernard find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhang-2025-xlam-family-large-action-models/