一个开源的大型编码器-解码器基础模型家族,用于化学
Eduardo Soares1, Emilio Vital Brazil2, Victor Shirasuna2
1IBM Research, Rio de Janeiro, Brazil. eduardo.soares@ibm.com.
Communications chemistry
|July 2, 2025
概括
我们开发了用于分子建模的新基础模型,在数百万个化学序列上进行了训练. 这些模型擅长预测分子特性和反应结果,为化学分析提供了强大的工具,使用最小的数据进行化学分析.
科学领域:
- 计算化学和化学信息学
- 机器学习在药物发现中的应用.
- 开发用于分子建模的人工智能.
背景情况:
- 最初在自然语言处理方面取得成功的基础模型现在正在适应分子建模.
- 化学语言模型的大规模预训练对于在各种任务中进行有效的表示学习至关重要.
- 现有的分子性质估计和反应预测方法可以通过先进的AI技术来改进.
研究的目的:
- 引入一个新的编码器-解码器化学基础模型家族.
- 将这些模型预训练在一个广泛的分子序列数据集上.
- 评估他们对财产估计和反应结果预测任务的表现.
主要方法:
- 开发用于分子序列的编码器-解码器基础模型.
- 在PubChem的9100万个分子序列的精选数据集上进行预训练.
- 在基准数据集上对模型性能进行评估,用于财产估计和反应预测.
- 分析学习的分子表示及其属性.
主要成果:
- 提出的化学基础模型与现有的最先进的方法相匹配或超越.
- 模型变体在属性估计和反应结果预测方面表现强.
- 学习的分子表示表现显示出化学相关的特征,并支持少量射击学习.
- 基于解码器的重建目标是学习嵌入空间结构的关键.
结论:
- 开发的基础模型为分子分析和推理提供了通用解决方案.
- 这些模型需要最小的监督,使它们高度适应.
- 这些发现突出了大规模预训练的潜力,以提高分子智能.
相关概念视频
02:00Molecular Models
40.7K
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.
40.7K
02:58Chemistry of the Cell
44.1K
The cell is chemically composed of water, organic molecules and inorganic ions.
Water
The polarity of the water molecule and its resulting hydrogen bonding makes water a unique substance with special properties that are intimately tied to the processes of life. Life originally evolved in an aqueous environment, and most of an organism’s cellular chemistry and metabolism occur inside the aqueous contents of the cell’s cytoplasm. Special properties of water are its high heat capacity...
Water
The polarity of the water molecule and its resulting hydrogen bonding makes water a unique substance with special properties that are intimately tied to the processes of life. Life originally evolved in an aqueous environment, and most of an organism’s cellular chemistry and metabolism occur inside the aqueous contents of the cell’s cytoplasm. Special properties of water are its high heat capacity...
44.1K
02:37Experimental Determination of Chemical Formula
40.1K
The elemental makeup of a compound defines its chemical identity, and chemical formulas are the most concise way of representing this elemental makeup. When a compound’s formula is unknown, measuring the mass of its constituent elements is often the first step in determining the formula experimentally.
40.1K
02:40Chemical Bonds
18.4K
Atoms participate in a chemical bond formation to acquire a completed valence-shell electron configuration similar to that of the noble gas nearest to it in atomic number. Ionic, covalent, and metallic bonds are some of the important types of chemical bonds. Bond energy and bond length determine the strength of a chemical bond.
Types of Chemical Bonds
An ionic bond is formed due to electrostatic attraction between cations and anions. Often, the ions are formed by the transfer of electrons...
18.4K
01:26Chemical Synapses
3.3K
Chemical synapses are specialized sites between two neurons or between a neuron and a non-neuronal cell like a muscle, glandular or sensory cell.
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
Because chemical synapses depend on the release of neurotransmitter molecules from synaptic vesicles to pass on their signal, there is an approximately one millisecond delay between when the axon potential reaches the presynaptic terminal and when the neurotransmitter leads to opening of postsynaptic ion channels. Additionally, this signaling is...
3.3K
02:17What is Organic Chemistry?
80.8K
Organic chemistry is the study of compounds of carbon called organic compounds. Organic compounds either originate from living organisms or are synthesized by chemists. A defining trait of these compounds is the presence of carbon as the principal element, which is bonded to other carbon atoms and other elements such as hydrogen, oxygen, nitrogen, and sulfur. The existence of a wide array of organic molecules is a consequence of carbon atoms’ ability to form up to four strong bonds to...
80.8K