Institute of Computing Technology, Chinese Academy IR
| DPA-2: a large atomic model as a multi-task learner | |
| Zhang, Duo1,2,3; Liu, Xinzijian1,2; Zhang, Xiangyu4,5; Zhang, Chengqian2,6; Cai, Chun1,2; Bi, Hangrui1,2; Du, Yiming4,5; Qin, Xuejian7,8,9; Peng, Anyang1; Huang, Jiameng2,10; Li, Bowen11; Shan, Yifan7,8,9; Zeng, Jinzhe12,13; Zhang, Yuzhi2; Liu, Siyuan2; Li, Yifan14; Chang, Junhan2,15; Wang, Xinyan2; Zhou, Shuo2,16; Liu, Jianchuan17; Luo, Xiaoshan18,19; Wang, Zhenyu19,20; Jiang, Wanrun1; Wu, Jing21; Yang, Yudi21; Yang, Jiyuan21; Yang, Manyi22; Gong, Fu-Qiang23; Zhang, Linshuang2; Shi, Mengchao2; Dai, Fu-Zhi1; York, Darrin M.12,13; Liu, Shi21,24; Zhu, Tong11,25,26; Zhong, Zhicheng7,8,9; Lv, Jian19; Cheng, Jun27,28; Jia, Weile4; Chen, Mohan1,6; Ke, Guolin2; Weinan, E.29,30; Zhang, Linfeng1,2; Wang, Han6,31 | |
| 2024-12-19 | |
| 发表期刊 | NPJ COMPUTATIONAL MATERIALS
 |
| 卷号 | 10期号:1页码:15 |
| 摘要 | The rapid advancements in artificial intelligence (AI) are catalyzing transformative changes in atomic modeling, simulation, and design. AI-driven potential energy models have demonstrated the capability to conduct large-scale, long-duration simulations with the accuracy of ab initio electronic structure methods. However, the model generation process remains a bottleneck for large-scale applications. We propose a shift towards a model-centric ecosystem, wherein a large atomic model (LAM), pre-trained across multiple disciplines, can be efficiently fine-tuned and distilled for various downstream tasks, thereby establishing a new framework for molecular modeling. In this study, we introduce the DPA-2 architecture as a prototype for LAMs. Pre-trained on a diverse array of chemical and materials systems using a multi-task approach, DPA-2 demonstrates superior generalization capabilities across multiple downstream tasks compared to the traditional single-task pre-training and fine-tuning methodologies. Our approach sets the stage for the development and broad application of LAMs in molecular and materials simulation research. |
| DOI | 10.1038/s41524-024-01493-2 |
| 收录类别 | SCI |
| 语种 | 英语 |
| 资助项目 | National Key R&D Program of China[2021YFA0718900] ; National Key R&D Program of China[2022YFA1403000] ; National Natural Science Foundation of China[12122401] ; National Natural Science Foundation of China[12074007] ; National Natural Science Foundation of China[12135002] ; National Key Research and Development Project of China[2022YFA1004302] ; National Institutes of Health[GM107485] ; National Science Foundation[2209718] ; Natural Science Foundation of Zhejiang Province[2022XHSJJ006] ; National Science Fund for Distinguished Young Scholars[22225302] ; National Science Fund for Distinguished Young Scholars[AI4EC] ; National Science Fund for Distinguished Young Scholars[RD2023100101] ; National Science Fund for Distinguished Young Scholars[RD2022070501] |
| WOS研究方向 | Chemistry ; Materials Science |
| WOS类目 | Chemistry, Physical ; Materials Science, Multidisciplinary |
| WOS记录号 | WOS:001381211200004 |
| 出版者 | NATURE PORTFOLIO |
| 引用统计 | |
| 文献类型 | 期刊论文 |
| 条目标识符 | http://119.78.100.204/handle/2XEOYT63/41077 |
| 专题 | 中国科学院计算技术研究所期刊论文_英文 |
| 通讯作者 | Zhang, Linfeng; Wang, Han |
| 作者单位 | 1.AI Sci Inst, Beijing, Peoples R China 2.DP Technol, Beijing, Peoples R China 3.Peking Univ, Acad Adv Interdisciplinary Studies, Beijing, Peoples R China 4.Chinese Acad Sci, Inst Comp Technol, State Key Lab Processors, Beijing, Peoples R China 5.Univ Chinese Acad Sci, Beijing, Peoples R China 6.Peking Univ, Coll Engn, HEDPS, CAPT, Beijing, Peoples R China 7.Chinese Acad Sci, Ningbo Inst Mat Technol & Engn, Ningbo, Peoples R China 8.Chinese Acad Sci, CAS Key Lab Magnet Mat & Devices, Ningbo, Peoples R China 9.Chinese Acad Sci, Zhejiang Prov Key Lab Magnet Mat & Applicat Techno, Ningbo, Peoples R China 10.Peking Univ, Sch Elect Engn & Comp Sci, Beijing, Peoples R China 11.East China Normal Univ, Shanghai Engn Res Ctr Mol Therapeut & New Drug Dev, Sch Chem & Mol Engn, Shanghai, Peoples R China 12.Rutgers State Univ, Inst Quantitat Biomed, Lab Biomol Simulat Res, Piscataway, NJ USA 13.Rutgers State Univ, Dept Chem & Chem Biol, Piscataway, NJ USA 14.Princeton Univ, Dept Chem, Princeton, NJ USA 15.Peking Univ, Coll Chem & Mol Engn, Beijing, Peoples R China 16.Peking Univ, Yuanpei Coll, Beijing, Peoples R China 17.Xihua Univ, Sch Elect Engn & Elect Informat, Chengdu, Peoples R China 18.Jilin Univ, Coll Phys, State Key Lab Superhard Mat, Changchun, Peoples R China 19.Jilin Univ, Coll Phys, Key Lab Mat Simulat Methods & Software, Minist Educ, Changchun, Peoples R China 20.Jilin Univ, Int Ctr Future Sci, Changchun, Peoples R China 21.Westlake Univ, Sch Sci, Dept Phys, Key Lab Quantum Mat Zhejiang Prov, Hangzhou, Peoples R China 22.Italian Inst Technol, Atomist Simulat, Genoa, Italy 23.Xiamen Univ, Coll Chem & Chem Engn, State Key Lab Phys Chem Solid Surface, iChEM, Xiamen, Peoples R China 24.Westlake Inst Adv Study, Inst Nat Sci, Hangzhou, Peoples R China 25.NYU Shanghai, NYU ECNU Ctr Computat Chem, Shanghai, Peoples R China 26.Inst Adv Algorithms Res, Shanghai, Peoples R China 27.IKKEM, Lab AI Electrochem AI4EC, Xiamen, Peoples R China 28.Xiamen Univ, Inst Artificial Intelligence, Xiamen, Peoples R China 29.Peking Univ, Ctr Machine Learning Res, Beijing, Peoples R China 30.Peking Univ, Sch Math Sci, Beijing, Peoples R China 31.Inst Appl Phys & Computat Math, Lab Computat Phys, Beijing, Peoples R China |
| 推荐引用方式 GB/T 7714 | Zhang, Duo,Liu, Xinzijian,Zhang, Xiangyu,et al. DPA-2: a large atomic model as a multi-task learner[J]. NPJ COMPUTATIONAL MATERIALS,2024,10(1):15. |
| APA | Zhang, Duo.,Liu, Xinzijian.,Zhang, Xiangyu.,Zhang, Chengqian.,Cai, Chun.,...&Wang, Han.(2024).DPA-2: a large atomic model as a multi-task learner.NPJ COMPUTATIONAL MATERIALS,10(1),15. |
| MLA | Zhang, Duo,et al."DPA-2: a large atomic model as a multi-task learner".NPJ COMPUTATIONAL MATERIALS 10.1(2024):15. |
| 条目包含的文件 | 条目无相关文件。 | |||||
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论