PuSH - Publication Server of Helmholtz Zentrum München

Export: TXT Text RIS Endnote (RIS) BIB BibTeX
Van Herck, J.* ; Gil, M.V.* ; Jablonka, K.M.* ; Abrudan, A.* ; Anker, A.S.* ; Asgari, M.* ; Blaiszik, B.* ; Buffo, A.* ; Choudhury, L.* ; Corminboeuf, C.* ; Daglar, H.* ; Elahi, A.M.* ; Foster, I.T.* ; García, S.A.* ; Garvin, M.* ; Godin, G.* ; Good, L.L.* ; Gu, J.* ; Xiao Hu, N.* ; Jin, X.* ; Junkers, T.* ; Keskin, S.* ; Knowles, T.P.J.* ; Laplaza, R.* ; Lessona, M.* ; Majumdar, S.K.* ; Mashhadimoslem, H.* ; McIntosh, R.D.* ; Moosavi, S.M.* ; Mouriño, B.* ; Nerli, F.* ; Pevida, C.* ; Poudineh, N.* ; Rajabi-Kochi, M.* ; Saar, K.L.* ; Hooriabad Saboor, F.* ; Sagharichiha, M.* ; Schmidt, K.J.* ; Shi, J.* ; Simone, E.* ; Svatunek, D.* ; Taddei, M.* ; Tetko, I.V. ; Tolnai, D.* ; Vahdatifar, S.* ; Whitmer, J.* ; Wieland, D.C.F.* ; Willumeit-Römer, R.* ; Züttel, A.* ; Smit, B.*

Assessment of fine-tuned large language models for real-world chemistry and material science applications.

Chem. Sci. 16, 670-684 (2025)
Publ. Version/Full Text DOI PMC
Free journal
Creative Commons Lizenzvertrag
Open Access Green as soon as Postprint is submitted to ZB.
The current generation of large language models (LLMs) has limited chemical knowledge. Recently, it has been shown that these LLMs can learn and predict chemical properties through fine-tuning. Using natural language to train machine learning models opens doors to a wider chemical audience, as field-specific featurization techniques can be omitted. In this work, we explore the potential and limitations of this approach. We studied the performance of fine-tuning three open-source LLMs (GPT-J-6B, Llama-3.1-8B, and Mistral-7B) for a range of different chemical questions. We benchmark their performances against "traditional" machine learning models and find that, in most cases, the fine-tuning approach is superior for a simple classification problem. Depending on the size of the dataset and the type of questions, we also successfully address more sophisticated problems. The most important conclusions of this work are that, for all datasets considered, their conversion into an LLM fine-tuning training set is straightforward and that fine-tuning with even relatively small datasets leads to predictive models. These results suggest that the systematic use of LLMs to guide experiments and simulations will be a powerful technique in any research study, significantly reducing unnecessary experiments or computations.
Impact Factor
Scopus SNIP
Web of Science
Times Cited
Altmetric
7.400
0.000
1
Tags
Annotations
Special Publikation
Hide on homepage

Edit extra information
Edit own tags
Private
Edit own annotation
Private
Hide on publication lists
on hompage
Mark as special
publikation
Publication type Article: Journal article
Document type Scientific Article
Language english
Publication Year 2025
Prepublished in Year 2024
HGF-reported in Year 2024
ISSN (print) / ISBN 2041-6520
e-ISSN 2041-6539
Journal Chemical Science
Quellenangaben Volume: 16, Issue: 2, Pages: 670-684 Article Number: , Supplement: ,
Publisher Royal Society of Chemistry (RSC)
Publishing Place Thomas Graham House, Science Park, Milton Rd, Cambridge Cb4 0wf, Cambs, England
Reviewing status Peer reviewed
Institute(s) Institute of Structural Biology (STB)
POF-Topic(s) 30203 - Molecular Targets and Therapies
Research field(s) Enabling and Novel Technologies
PSP Element(s) G-503000-001
Grants UK Research and Innovation (UKRI) under the UK government's Horizon Europe
European Research Council (ERC)
Data Sciences Institute at the University of Toronto
Novo Nordisk Foundation
USorb-DAC Project through Grantham Foundation for the Protection of the Environment
Carl Zeiss Foundation
European Regional Development Fund (ERDF)
Galician Government
Spanish Ministry of Science and Innovation
Spanish National Research Council (CSIC)
European Union NextGenerationEU/PRTR
Spanish Agencia Estatal de Investigacion (AEI) - MICIU/AEI
European Research Council under the European Union's Horizon 2020 research and innovation program through the ERC grant DiProPhys
National Institutes of Health Oxford-Cambridge Scholars Program
Cambridge Trust's Cambridge International Scholarship

European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme
European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant
NCCR MARVEL, a National Centre of Competence in Research - Swiss National Science Foundation
Italian MUR
St. John's College Research Fellowship programme
Rhodes Trust
Schmidt Science Fellowship
Frances and Augustus Newman Foundation
European Research Council under the European Union's Seventh Framework Programme (FP7/2007-2013)
Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases at the National Institutes of Health
Swiss Science Foundation
DOI 10.1039/d4sc04401k
WOS ID 001373013200001
Scopus ID 85212108442
PubMed ID 39664810
Erfassungsdatum 2024-12-13
[➜Report correction]