Elamur Research⌕
☰
home / papers / 2023 / Agent Lumos: Unified and Modular Training for Open-Source Language Agents

Agent Lumos: Unified and Modular Training for Open-Source Language Agents

Paper · 2023 · Da Yin, Faeze Brahman, Abhilasha Ravichander et al.

Summary

Most language agents rely on closed APIs such as GPT-4. These are costly, opaque and hard to reproduce, especially for long-context interactive tasks. Lumos (ACL 2024) is an open framework for training agents on top of LLAMA-2-7B/13B. It divides the work among three components:

  • a planning module that breaks a task into natural-language subgoals;
  • a grounding module that maps each subgoal to executable tool calls;
  • an execution module that runs those calls with APIs, neural models or simulators.

The modules can be used in two ways. Lumos-O produces the whole plan in one pass. Lumos-I plans step by step, using execution feedback. For training data, the authors had strong LLMs rewrite the existing ground-truth reasoning steps of many benchmarks into a single shared format. This produced about 56K annotations covering QA, web, math and multimodal tasks. Across 9 datasets, Lumos beats larger open-source agents and in some cases GPT-based agents. It also transfers to unseen tasks. Because the modules are separable, each one can be upgraded independently, and the released data can be used to add agent skills to other open models.

Key points

  • Released resources: about 56K unified agent training annotations, plus models, on Hugging Face (ai2lumos). The code is at github.com/allenai/lumos.
  • Web and QA results: Lumos scores 5.0% higher than GPT-4 on Mind2Web. On HotpotQA it gains 4.1% (LLM accuracy) over ReAct and 3.5% over ReWOO, both built on GPT-3.5-turbo.
  • Unseen tasks: On WebShop and InterCode SQL, Lumos outperforms 30B–33B-scale agents. On WebShop the margin is close to 20 reward points.
  • Training formulations (LLAMA-2-7B): The modular setups usually beat chain-of-thought and single-module "integrated" training. Examples:
    • Lumos-I reaches 45.9 on HotpotQA, against 39.6 (integrated) and 22.1 (CoT).
    • Lumos-O reaches 65.5 on SVAMP.
  • General instruction following: When Alpaca data is mixed into training, Lumos scores 39.3 ROUGE-L on Super-NaturalInstruction, close to Alpaca-only LLAMA-2-7B at 39.8.
  • Limitations: Lumos cannot backtrack or replan after errors. Some QA experiments still use GPT models as sub-question tools.

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 WashingtonSeattle · United States422 works · 213,539 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksArts and Humanities · 17,959 worksLanguage and Linguistics · 17,789 works

Cite

Cite as Da Yin, Faeze Brahman, Abhilasha Ravichander, Khyathi Chandu, Kai-Wei Chang, Yejin Choi and Bill Yuchen Lin. 2023. Agent Lumos: Unified and Modular Training for Open-Source Language Agents. arXiv:2311.05657.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2311.05657 · arXiv 2311.05657

@misc{yin2023agent,
  title = {{Agent Lumos: Unified and Modular Training for Open-Source Language Agents}},
  author = {Da Yin, Faeze Brahman, Abhilasha Ravichander, Khyathi Chandu, Kai-Wei Chang, Yejin Choi and Bill Yuchen Lin},
  year = {2023},
  doi = {10.48550/arxiv.2311.05657},
  eprint = {2311.05657},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2311.05657}
}

References

50 references, 37 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, et al. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chaitanya Malaviya, Asli Celikyilmaz, and Yejin Choi. 2019. COMET: Commonsense transformers for automatic knowledge graph construction. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 4762–4779, Florence…
  3. Faeze Brahman, Chandra Bhagavatula, Valentina Pyatkin, Jena D Hwang, Xiang Lorraine Li, Hirona J Arai, Soumya Sanyal, Keisuke Sakaguchi, Xiang Ren, and Yejin Choi. 2023. Plasma: Making small language models better procedural knowledge models for (counterfactual) planning. arXiv preprint arXiv:2305.1…
  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. Vladimir Karpukhin, Barlas Oguz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open-domain question answering. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 6769–6781, Onli…
  6. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, et al. 2021. Evaluating large language models trained on code. arXiv preprint arXiv:2107.03374.
  7. Ariel N. Lee, Cole J. Hunter, Nataniel Ruiz, Bleys Goodson, Wing Lian, Guan Wang, Eugene Pentland, Austin Cook, Chanvichet Vong, and "Teknium". 2023. Openorcaplatypus: Llama2-13b model instruct-tuned on filtered openorcav1 gpt-4 dataset and merged with divergent stem and logic dataset model. https:/… find
  8. Wei-Lin Chiang, Zhuohan Li, Zi Lin, Ying Sheng, Zhanghao Wu, Hao Zhang, Lianmin Zheng, Siyuan Zhuang, Yonghao Zhuang, Joseph E. Gonzalez, Ion Stoica, and Eric P. Xing. 2023. Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality. find
  9. Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. 2023. Let’s verify step by step. arXiv preprint arXiv:2305.20050.
  10. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  11. Bill Yuchen Lin, Yicheng Fu, Karina Yang, Prithviraj Ammanabrolu, Faeze Brahman, Shiyu Huang, Chandra Bhagavatula, Yejin Choi, and Xiang Ren. 2023. Swiftsage: A generative agent with fast and slow thinking for complex interactive tasks. ArXiv preprint, abs/2305.17390.
  12. 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.
  13. Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. 2021. Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies. Transactions of the Association for Computational Linguistics, 9:346– 361. find
  14. Yu Gu, Sue Kase, Michelle Vanni, Brian Sadler, Percy Liang, Xifeng Yan, and Yu Su. 2021. Beyond iid: three levels of generalization for question answering on knowledge bases. In Proceedings of the Web Conference 2021, pages 3477–3488. find
  15. Cheng-Yu Hsieh, Chun-Liang Li, Chih-kuan Yeh, Hootan Nakhost, Yasuhisa Fujii, Alex Ratner, Ranjay Krishna, Chen-Yu Lee, and Tomas Pfister. 2023. Distilling step-by-step! outperforming larger language models with less training data and smaller model sizes. In Findings of the Association for Computati…
  16. Mandar Joshi, Eunsol Choi, Daniel Weld, and Luke Zettlemoyer. 2017. TriviaQA: A large scale distantly supervised challenge dataset for reading comprehension. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1601–1611, Vancouver…
  17. Xi Victoria Lin, Chenglong Wang, Luke Zettlemoyer, and Michael D Ernst. 2018. Nl2bash: A corpus and semantic parser for natural language interface to the linux operating system. arXiv preprint arXiv:1802.08979. find
  18. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2023a. Improved baselines with visual instruction tuning. arXiv preprint arXiv:2310.03744.
  19. Xiao Liu, Hao Yu, Hanchen Zhang, Yifan Xu, Xuanyu Lei, Hanyu Lai, Yu Gu, Hangliang Ding, Kaiwen Men, Kejuan Yang, et al. 2023b. Agentbench: Evaluating llms as agents. arXiv preprint arXiv:2308.03688.
  20. Zhiwei Liu, Weiran Yao, Jianguo Zhang, Le Xue, Shelby Heinecke, Rithesh Murthy, Yihao Feng, Zeyuan Chen, Juan Carlos Niebles, Devansh Arpit, et al. 2023c. Bolaa: Benchmarking and orchestrating llm-augmented autonomous agents. arXiv preprint arXiv:2308.05960.
  21. Pan Lu, Swaroop Mishra, Tony Xia, Liang Qiu, Kai-Wei Chang, Song-Chun Zhu, Oyvind Tafjord, Peter Clark, and Ashwin Kalyan. 2022. Learn to explain: Multimodal reasoning via thought chains for science question answering. In The 36th Conference on Neural Information Processing Systems (NeurIPS). find
  22. Pan Lu, Baolin Peng, Hao Cheng, Michel Galley, Kai-Wei Chang, Ying Nian Wu, Song-Chun Zhu, and Jianfeng Gao. 2023. Chameleon: Plug-and-play compositional reasoning with large language models. arXiv preprint arXiv:2304.09842.
  23. Xing Han Lù, Zdeněk Kasner, and Siva Reddy. 2024. Weblinx: Real-world website navigation with multi-turn dialogue.
  24. Shen-yun Miao, Chao-Chun Liang, and Keh-Yih Su. 2020. A diverse corpus for evaluating and developing English math word problem solvers. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 975–984, Online. Association for Computational Linguistics. find
  25. Swaroop Mishra, Matthew Finlayson, Pan Lu, Leonard Tang, Sean Welleck, Chitta Baral, Tanmay Rajpurohit, Oyvind Tafjord, Ashish Sabharwal, Peter Clark, and Ashwin Kalyan. 2022. LILA: A Unified Benchmark for Mathematical Reasoning. In Proceedings of the 2022 Conference on Empirical Methods in Natural…
  26. OpenAI. 2023a. Gpt-4 technical report. abs/2303.08774.
  27. ArXiv, Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023a. Llama: Open and efficient foundation language models.…
  28. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. 2023b. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288.
  29. Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. MuSiQue: Multi-hop questions via single-hop question composition. Transactions of the Association for Computational Linguistics, 10:539–554. find
  30. Yizhong Wang, Hamish Ivison, Pradeep Dasigi, Jack Hessel, Tushar Khot, Khyathi Raghavi Chandu, David Wadden, Kelsey MacMillan, Noah A Smith, Iz Beltagy, et al. 2023a. How far can camels go? exploring the state of instruction tuning on open resources. arXiv preprint arXiv:2306.04751.
  31. OpenAI. 2023b. Gpt-4v(ision) system card. find
  32. 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, Online. Associat…
  33. Shuofei Qiao, Ningyu Zhang, Runnan Fang, Yujie Luo, Wangchunshu Zhou, Yuchen Eleanor Jiang, Chengfei Lv, and Huajun Chen. 2024. Autoact: Automatic agent learning from scratch via self-planning. arXiv preprint arXiv:2401.05268. — [arXiv:2401.05268](https://arxiv.org/abs/2401.05268) find
  34. Dustin Schwenk, Apoorv Khandelwal, Christopher Clark, Kenneth Marino, and Roozbeh Mottaghi. 2022. A-okvqa: A benchmark for visual question answering using world knowledge. In European Conference on Computer Vision, pages 146–162. Springer.
  35. Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. 2023. Hugginggpt: Solving ai tasks with chatgpt and its friends in huggingface. arXiv preprint arXiv:2303.17580.
  36. Noah Shinn, Beck Labash, and Ashwin Gopinath. 2023. Reflexion: an autonomous agent with dynamic memory and self-reflection. arXiv preprint arXiv:2303.11366.
  37. Mohit Shridhar, Xingdi Yuan, Marc-Alexandre Cote, Yonatan Bisk, Adam Trischler, and Matthew Hausknecht. 2020. Alfworld: Aligning text and embodied environments for interactive learning. In International Conference on Learning Representations.
  38. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A. Smith, Daniel Khashabi, and Hannaneh Hajishirzi. 2023b. Self-instruct: Aligning language models with self-generated instructions. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long… find
  39. Yizhong Wang, Swaroop Mishra, Pegah Alipoormolabashi, Yeganeh Kordi, Amirreza Mirzaei, Atharva Naik, Arjun Ashok, Arut Selvan Dhanasekaran, Anjana Arunkumar, David Stap, Eshaan Pathak, Giannis Karamanolakis, Haizhi Lai, Ishan Purohit, Ishani Mondal, Jacob Anderson, Kirby Kuznia, Krima Doshi, Kuntal… find
  40. Johannes Welbl, Nelson F. Liu, and Matt Gardner. 2017. Crowdsourcing Multiple Choice Science Questions. In Proceedings of the 3rd Workshop on Noisy User-generated Text, pages 94–106, Copenhagen, Denmark. Association for Computational Linguistics.
  41. Peter West, Chandra Bhagavatula, Jack Hessel, Jena Hwang, Liwei Jiang, Ronan Le Bras, Ximing Lu, Sean Welleck, and Yejin Choi. 2022. Symbolic knowledge distillation: from general language models to commonsense models. In Proceedings of the 2022 Conference of the North American Chapter of the Associa…
  42. Jian Xie, Kai Zhang, Jiangjie Chen, Tinghui Zhu, Renze Lou, Yuandong Tian, Yanghua Xiao, and Yu Su. 2024. Travelplanner: A benchmark for real-world planning with language agents. arXiv preprint arXiv:2402.01622.
  43. Binfeng Xu, Zhiyuan Peng, Bowen Lei, Subhabrata Mukherjee, Yuchen Liu, and Dongkuan Xu. 2023. Rewoo: Decoupling reasoning from observations for efficient augmented language models. arXiv preprint arXiv:2305.18323.
  44. John Yang, Akshara Prabhakar, Karthik Narasimhan, and Shunyu Yao. 2023. Intercode: Standardizing and benchmarking interactive coding with execution feedback.
  45. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. HotpotQA: A dataset for diverse, explainable multi-hop question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2369…
  46. Shunyu Yao, Howard Chen, John Yang, and Karthik Narasimhan. 2022a. Webshop: Towards scalable real-world web interaction with grounded language agents. Advances in Neural Information Processing Systems, 35:20744–20757.
  47. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2022b. React: Synergizing reasoning and acting in language models. In The Eleventh International Conference on Learning Representations.
  48. Rowan Zellers, Yonatan Bisk, Ali Farhadi, and Yejin Choi. 2019. From recognition to cognition: Visual commonsense reasoning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6720–6731. find
  49. 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.
  50. 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.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yin-2023-agent-lumos-unified-modular-training/