LLM 如何变革材料科学与化学的 14 个例证：基于一场大语言模型黑客松的反思
材料科学
2023-11-23 v4 机器学习
化学物理
摘要
GPT-4 等大语言模型（LLMs）引起了众多科学家的兴趣。近期研究表明，这些模型在化学与材料科学中可能有用。为探索这些可能性，我们组织了一场黑客松。本文记录了作为该黑客松一部分所构建的项目。参与者将 LLMs 用于多种应用，包括预测分子与材料的属性、设计工具的新界面、从非结构化数据中提取知识，以及开发新的教育应用。主题的多样性，以及可用不到两天生成可运行原型这一事实，凸显出 LLMs 将深刻影响我们领域的未来。丰富的想法与项目集也表明，LLMs 的应用不限于材料科学与化学，而可为广泛科学学科带来潜在益处。
引用
@article{arxiv.2306.06283,
title = {14 Examples of How LLMs Can Transform Materials Science and Chemistry: A Reflection on a Large Language Model Hackathon},
author = {Kevin Maik Jablonka and Qianxiang Ai and Alexander Al-Feghali and Shruti Badhwar and Joshua D. Bocarsly and Andres M Bran and Stefan Bringuier and L. Catherine Brinson and Kamal Choudhary and Defne Circi and Sam Cox and Wibe A. de Jong and Matthew L. Evans and Nicolas Gastellu and Jerome Genzling and María Victoria Gil and Ankur K. Gupta and Zhi Hong and Alishba Imran and Sabine Kruschwitz and Anne Labarre and Jakub Lála and Tao Liu and Steven Ma and Sauradeep Majumdar and Garrett W. Merz and Nicolas Moitessier and Elias Moubarak and Beatriz Mouriño and Brenden Pelkie and Michael Pieler and Mayk Caldas Ramos and Bojana Ranković and Samuel G. Rodriques and Jacob N. Sanders and Philippe Schwaller and Marcus Schwarting and Jiale Shi and Berend Smit and Ben E. Smith and Joren Van Herck and Christoph Völker and Logan Ward and Sean Warren and Benjamin Weiser and Sylvester Zhang and Xiaoqi Zhang and Ghezal Ahmad Zia and Aristana Scourtas and KJ Schmidt and Ian Foster and Andrew D. White and Ben Blaiszik},
journal= {arXiv preprint arXiv:2306.06283},
year = {2023}
}