相关实验视频

Updated: Jan 6, 2026

Author Spotlight: Experimental Approaches for the Synthesis of Low-Valent Metal-Organic Frameworks from Multitopic Phosphine Linkers

Author Spotlight: Experimental Approaches for the Synthesis of Low-Valent Metal-Organic Frameworks from Multitopic Phosphine Linkers

Published on: May 12, 2023

语言模型使无机材料的数据增强合成规划成为可能

Thorben Prein^1,2,3, Elton Pan⁴, Janik Jehkul⁵

¹School of Natural Sciences, Technische Universität München, Garching bei München 85748, Germany.

ACS applied materials & interfaces

|November 26, 2025

概括

大型语言模型 (LMs) 可以从科学文献中预测无机合成条件,而无需微调. 将LM与变压器模型 (SyntMTE) 结合起来,可以实现可扩展和数据高效的合成规划.

科学领域:

材料科学材料科学材料科学
计算化学的计算化学
人工智能的人工智能

背景情况:

无机合成规划传统上使用启发式或有限的机器学习 (ML) 模型,阻碍了广泛的应用.
现有的方法难以应对化学合成数据的广度和复杂性.

研究的目的:

评估现成语言模型 (LMs) 在回忆和预测无机合成条件方面的能力.
调查LM生成数据的潜力,以培养更有效的合成规划模型.
为可扩展和数据效率高的无机合成规划开发混合工作流.

主要方法:

使用预先训练的语言模型 (GPT-4.1,Gemini 2.0 Flash,Llama 4 Maverick) 来进行前体和条件预测.
组合多个LM来提高准确性和降低计算成本.
在LM生成的反应配方和文献数据上训练了一个变压器模型 (SyntMTE).
在持有数据集和涉及固态电解质 (Li7La3Zr2O12) 的案例研究上评估了模型性能.

主要成果:

现成的LM实现了高达53.8%的Top-1前体预测准确度和66.8%的Top-5预测准确度.
LM预测了化和烧结温度,平均绝对误差<126°C,优于专业模型.
整合LM可以提高准确性,并将推断成本降低高达70%.
仅在LM生成数据上训练的模型显示出竞争性表现,仅比文献训练的模型差6%.
在两种数据类型上训练的混合模型提高了高达4%的性能,并复制了固态电解质的实验趋势.

结论:

语言模型具有重要的,未开发的潜力,可以在没有特定任务的微调的情况下进行无机合成规划.
通过LM生成的数据可以增强有限的文献数据集,从而使合成预测模型的训练更加强大和数据效率更高.
结合LM和变压器模型的混合方法为推进无机合成计划提供了可扩展和有效的策略.

关键词:

大型语言模型.关于前体的推建议固态合成固态合成合成条件的预测和预测.合成数据增强合成数据增强

更多相关视频

Author Spotlight: Accelerating Discovery in Microporous Material Chemistry

Author Spotlight: Accelerating Discovery in Microporous Material Chemistry

Published on: October 6, 2023

Synthesis and Characterization of Functionalized Metal-organic Frameworks

Synthesis and Characterization of Functionalized Metal-organic Frameworks

Published on: September 5, 2014

相关概念视频

Molecular Models

Molecular Models

Physical models representing molecular architectures of chemical compounds play essential roles in understanding chemistry. The use of molecular models makes it easier to visualize the structures and shapes of atoms and molecules.

Predicting Molecular Geometry

Predicting Molecular Geometry

VSEPR Theory for Determination of Electron Pair Geometries

Properties of Organometallic Compounds

Properties of Organometallic Compounds

Organometallic compounds are compounds that contain a carbon–metal bond. Carbon belongs to an organyl group like alkyl, aryl, allyl, or benzyl groups. The metal can be from Group I or Group II of the periodic table, a transition metal, or a semimetal.

Synthetic Biology

Synthetic Biology

Synthetic biology is an interdisciplinary science that involves using principles from disciplines such as engineering, molecular biology, cell biology, and systems biology. It involves remodeling existing organisms from nature or constructing completely new synthetic organisms for applications such as protein or enzyme production, bioremediation, value-added macromolecule production, and the addition of desirable traits to crops, to name a few.
Golden rice
Golden rice is a genetically modified...

Synthesis and Decomposition Reactions

Synthesis and Decomposition Reactions

Synthesis and decomposition are two types of redox reactions. Synthesis means to make something, whereas decomposition means to break something. The reactions are accompanied by chemical and energy changes.

Metal-Ligand Bonds

Metal-Ligand Bonds

The hemoglobin in the blood, the chlorophyll in green plants, vitamin B-12, and the catalyst used in the manufacture of polyethylene all contain coordination compounds. Ions of the metals, especially the transition metals, are likely to form complexes.
In these complexes, transition metals form coordinate covalent bonds, a kind of Lewis acid-base interaction in which both of the electrons in the bond are contributed by a donor (Lewis base) to an electron acceptor (Lewis acid). The Lewis acid in...