Elamur Research⌕
☰
home / papers / 2024 / MLR-Copilot: Autonomous Machine Learning Research based on Large Language Models Agents

MLR-Copilot: Autonomous Machine Learning Research based on Large Language Models Agents

Paper · 2024 · Ruochen Li, Teerth Patel, Qingyun Wang et al.

Summary

MLR-Copilot is an LLM-agent framework for automating the full machine learning research cycle. Earlier auto-experimentation systems began with a predefined task and a ready code template. This system starts from a published paper instead. It runs in three stages. First, IdeaAgent takes selected sections of the input paper (via the Semantic Scholar API) and extracts the task, research gaps and keywords. It then gathers recent related work and proposes a methodology with an experiment plan. IdeaAgent was trained with supervised fine-tuning and then fine-grained RL, using reward models for novelty, feasibility and effectiveness built from OpenReview review data. Second, ExperimentAgent adapts prototype code from the source paper and can optionally fetch models and datasets from HuggingFace. Third, it runs the experiments, debugs them over several iterations and can accept human feedback. The authors evaluate the system manually and automatically on five ML research tasks and report novel, feasible ideas with executable results. The supplied excerpt omits the detailed scores. The work is an early end-to-end research copilot, released in August 2024 alongside AI Scientist.

Key points

  • Pipeline: idea generation (IdeaAgent) → experiment implementation → implementation execution (both handled by ExperimentAgent), with execution, debugging and optional human feedback looping back into the implementation.
  • IdeaAgent training data: 4,271 ICLR/NeurIPS papers from 2023–2024, collected via OpenReview. Supervised fine-tuning used a derived set of 1,000 papers with extracted ideas and plans. RL used rewards for novelty, feasibility and effectiveness, with no steering at inference time.
  • A research idea is represented as RI = {P, R, h, e}: source-paper information, retrieved recent work, generated methodology and experiment plan.
  • Evaluation covers five ML research tasks, including a sentiment-analysis case study on the ELLIPSE dataset. Figure 1 shows an example execution result of 0.787 on a semantic textual similarity task.
  • Code, data and models are at github.com/du-nlp-lab/MLR-Copilot, and a GUI demo is on HuggingFace Spaces (du-lab/MLR-Copilot). The authors note a limitation: the stages are mostly sequential, with little ability to go back from execution to ideation.

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 Illinois Urbana-ChampaignUrbana · United States228 works · 74,591 citations🎓The University of Texas at DallasRichardson · United States66 works · 5,836 citations

 Fields, topics and concepts

📚 Shelves 3The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksInformation Systems · 1,888 works

Cite

Cite as Ruochen Li, Teerth Patel, Qingyun Wang and Xinya Du. 2024. MLR-Copilot: Autonomous Machine Learning Research based on Large Language Models Agents. arXiv:2408.14033.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2408.14033 · arXiv 2408.14033

@misc{li2024mlr,
  title = {{MLR-Copilot: Autonomous Machine Learning Research based on Large Language Models Agents}},
  author = {Ruochen Li, Teerth Patel, Qingyun Wang and Xinya Du},
  year = {2024},
  doi = {10.48550/arxiv.2408.14033},
  eprint = {2408.14033},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2408.14033}
}

References

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

  1. Jinheon Baek, Sujay Kumar Jauhar, Silviu Cucerzan, and Sung Ju Hwang. 2024. Researchagent: Iterative research idea generation over scientific literature with large language models. arXiv preprint arXiv:2404.07738.
  2. Lutz Bornmann, Ruediger Mutz, and Hans-Dieter Daniel. 2010. The growth of scientific knowledge: a bibliometric perspective on the expansion and acceleration of scientific output. Journal of the American Society for Information Science and Technology, 61(12):2155–2160. find
  3. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
  4. Shyam Choudhury. 2021. Emerging technologies in scientific research: Opportunities and challenges. Journal of Scientific Research and Development, 12(3):45– 56. find
  5. Arman Cohan and Nazli Goharian. 2018. Scientific document summarization via citation contextualization and scientific discourse. International Journal on Digital Libraries, 19(2–3):287–303. find
  6. Arghavan Moradi Dakhel, Vahid Majdinasab, Amin Nikanjam, Foutse Khomh, Michel C Desmarais, and Zhen Ming Jack Jiang. 2023. Github copilot ai pair programmer: Asset or liability? Journal of Systems and Software, 203:111734. find
  7. John Doe and Jane Smith. 2023. Ellipse: A dataset for feedback prediction in machine learning. Journal of Machine Learning Research, 24:1–10. find
  8. A. Franklin, M. Benner, N. Rambis, P. Baffour, R. Holbrook, S. Crossley, and ulrichboser. 2022. Feedback prize - english language learning. find
  9. GitHub, Inc. Github copilot. https://github.com/features/copilot. Accessed: 2024-08-05. ↗
  10. Juraj Gottweis, Wei-Hung Weng, Alexander Daryin, Tao Tu, Anil Palepu, Petar Sirkovic, Artiom Myaskovsky, Felix Weissenberger, Keran Rong, Ryutaro Tanno, et al. 2025. Towards an ai co-scientist. arXiv preprint arXiv:2502.18864.
  11. Glen M. Hocky and Andrew D. White. 2022. Natural language processing models that automate programming will transform chemistry research and teaching. Digital Discovery, 1:79–83. find
  12. Qian Huang, Jian Vora, Percy Liang, and Jure Leskovec. 2023. Mlagentbench: Evaluating language agents on machine learning experimentation. arXiv preprint arXiv:2310.03302.
  13. Tal Ifargan, Lukas Hafner, Maor Kern, Ori Alcalay, and Roy Kishony. 2025. Autonomous llm-driven research—from data to human-verifiable research papers. NEJM AI, 2(1):AIoa2400555.
  14. Peter Jansen, Oyvind Tafjord, Marissa Radensky, Pao Siangliulue, Tom Hope, Bhavana Dalvi Mishra, Bodhisattwa Prasad Majumder, Daniel S Weld, and Peter Clark. 2025. Codescientist: End-to-end semi-automated scientific discovery with code-based experimentation. arXiv preprint arXiv:2503.22708.
  15. Quentin Lhoest, Albert Villanova del Moral, Yacine Jernite, Abhishek Thakur, Patrick von Platen, Suraj Patil, Julien Chaumond, Mariama Drame, Julien Plu, Lewis Tunstall, Joe Davison, Mario Šaško, Gunjan Chhablani, Bhavitvya Malik, Simon Brandeis, Teven Le Scao, Victor Sanh, Canwen Xu, Nicolas Patry,… — Datasets: A Community Library for Natural Language Processing
  16. Ruochen Li, Liqiang Jing, Chi Han, Jiawei Zhou, and Xinya Du. 2024a. Learning to generate research idea with dynamic control. find
  17. Ruochen Li, Liqiang Jing, Chi Han, Jiawei Zhou, and Xinya Du. 2024b. Learning to generate research idea with dynamic control. arXiv preprint arXiv:2412.14626. — [arXiv:2412.14626](https://arxiv.org/abs/2412.14626) find
  18. Chris Lu, Cong Lu, Robert Tjarko Lange, Jakob Foerster, Jeff Clune, and David Ha. 2024. The AI Scientist: Towards fully automated open-ended scientific discovery. arXiv preprint arXiv:2408.06292.
  19. Andrew L Maas, Raymond E Daly, Peter T Pham, Dan Huang, Andrew Y Ng, and Christopher Potts. 2011. Learning word vectors for sentiment analysis. Proceedings of the 49th annual meeting of the association for computational linguistics: Human language technologies, pages 142–150.
  20. OpenAI. 2023. GPT-4 technical report. arXiv preprint arXiv:2303.08774.
  21. Nedjma Ousidhoum, Shamsuddeen Hassan Muhammad, Mohamed Abdalla, Idris Abdulmumin, Ibrahim Said Ahmad, Sanchit Ahuja, Alham Fikri Aji, Vladimir Araujo, Abinew Ali Ayele, Pavan Baswani, Meriem Beloucif, Chris Biemann, Sofia Bourhim, Christine De Kock, Genet Shanko Dekebo, Oumaima Hourrane, Gopichand K… find
  22. Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th annual acm symposium on user interface software and technology, pages 1–22. find
  23. Kenny Powell. 2015. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches. Sage Publications. find
  24. Biqing Qi, Kaiyan Zhang, Haoxiang Li, Kai Tian, Sihang Zeng, Zhang-Ren Chen, and Bowen Zhou. 2023a. Large language models are zero shot hypothesis proposers.
  25. Biqing Qi, Kaiyan Zhang, Haoxiang Li, Kai Tian, Sihang Zeng, Zhang-Ren Chen, and Bowen Zhou. 2023b. Large language models are zero shot hypothesis proposers.
  26. Samuel Schmidgall, Yusheng Su, Ze Wang, Ximeng Sun, Jialian Wu, Xiaodong Yu, Jiang Liu, Zicheng Liu, and Emad Barsoum. 2025. Agent laboratory: Using llm agents as research assistants. arXiv preprint arXiv:2501.04227.
  27. ScienceDirect. 2023. Automl: A systematic review on automated machine learning. ScienceDirect. find
  28. John Smith, Jane Doe, and Wei Zhang. 2023. Mlagentbench2023: A framework for automating research idea generation and implementation using llm agents. Journal of Computational Research, 45(3):123–145. find
  29. Jiabin Tang, Lianghao Xia, Zhonghang Li, and Chao Huang. 2025. Ai-researcher: Autonomous scientific innovation. arXiv preprint arXiv:2505.18705.
  30. Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton-Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller,… — Llama 2: Open Foundation and Fine-Tuned Chat Models
  31. Vijay Viswanathan, Chenyang Zhao, Amanda Bertsch, Tongshuang Wu, and Graham Neubig. 2023. Prompt2model: Generating deployable models from natural language instructions. CoRR, abs/2308.12261. find
  32. Qingyun Wang, Doug Downey, Heng Ji, and Tom Hope. 2024. Scimon: Scientific inspiration machines optimized for novelty. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL2024).
  33. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lho… — Transformers: State-of-the-Art Natural Language Processing
  34. Yutaro Yamada, Robert Tjarko Lange, Cong Lu, Shengran Hu, Chris Lu, Jakob Foerster, Jeff Clune, and David Ha. 2025. The ai scientist-v2: Workshop-level automated scientific discovery via agentic tree search. arXiv preprint arXiv:2504.08066.
  35. Zonglin Yang, Xinya Du, Junxian Li, Jie Zheng, Soujanya Poria, and Erik Cambria. 2023. Large language models for automated open-domain scientific hypotheses discovery. arXiv preprint arXiv:2309.02726.
  36. Haofei Yu, Zhaochen Hong, Zirui Cheng, Kunlun Zhu, Keyang Xuan, Jinwei Yao, Tao Feng, and Jiaxuan You. 2024. Researchtown: Simulator of human research community. arXiv preprint arXiv:2412.17767.
  37. Shujian Zhang, Chengyue Gong, Lemeng Wu, Xingchao Liu, and Mi Zhou. 2023. Automl-gpt: Automatic machine learning with gpt. ArXiv, abs/2305.02499.
  38. Yue Zhang and Zhiyang Teng. 2023. Natural Language Processing: A Machine Learning Perspective. MIT Press. find
  39. Ruiqi Zhong, Peter Zhang, Steve Li, Jinwoo Ahn, Dan Klein, and Jacob Steinhardt. 2023. Goal driven discovery of distributional differences via language descriptions. Advances in Neural Information Processing Systems, 36:40204–40237.
  40. Shuyan Zhou, Frank F Xu, Hao Zhu, Xuhui Zhou, Robert Lo, Abishek Sridhar, Xianyi Cheng, Tianyue Ou, Yonatan Bisk, Daniel Fried, et al. 2023. Webarena: A realistic web environment for building autonomous agents. arXiv preprint arXiv:2307.13854.
  41. Minjun Zhu, Yixuan Weng, Linyi Yang, and Yue Zhang. 2025. Deepreview: Improving llm-based paper review with human-like deep thinking process. arXiv preprint arXiv:2503.08569.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/li-2024-mlr-copilot-autonomous-machine/