推理能力能否显著改善化学的大型语言模型的知识?─基于与LLM的对话
Dong-Xu Cui1, Shi-Yu Long2, Yi-Xuan Tang3
1College of Chemical Engineering, Lanzhou University of Arts and Science, Lanzhou 730000, China.
Journal of chemical information and modeling
|August 25, 2025
概括
推理增强的大型语言模型 (LLM) 显示化学任务性能有所改善,但与结构准确性和空间识别有所困难. 对于专门的化学应用,需要特定领域的培训.
科学领域:
- 人工智能
- 计算化学
- 化学信息学
背景情况:
- 大型语言模型 (LLM) 越来越多地应用于科学领域.
- 提高LLM的推理能力对化学认知的影响尚不清楚.
- 快速的工程策略需要对化学任务进行系统的评估.
研究的目的:
- 在九个关键化学任务上系统评估五个推理增强的LLM.
- 将其性能与传统的LLM和计算工具进行比较.
- 研究推理和提示工程对化学认知的影响.
主要方法:
- 在九个化学任务中对五个推理增强的LLM (Deepseek-R1-0528,OpenAI-o4 mini,Gemini-2.5-pro,doubao-seed-1.6-thinking,qwen-max-latest) 进行系统评估.
- 与传统的LLM和已建立的计算化学工具进行比较.
- 基于即时复杂性和模型推理能力的性能变化分析.
主要成果:
- 支持推理的LLM在基础化学任务中表现出显著的性能提升.
- 过于复杂的提示通常不会改善,有时甚至会阻碍表现.
- 所有评估的LLM在生成准确的CIF文件方面都存在局限性,特别是在债券拓方面.
- 在立体化学识别和识别罕见对称组方面仍然存在挑战,这表明空间识别不足.
结论:
- 推理框架增强了LLM的逻辑连贯性,但没有克服固有的空间识别限制.
- 目前的LLM需要特定领域的培训,以有效地将一般推理与专业化应用联系起来.
- 对优化训练模式的进一步研究对于在化学领域推进人工智能至关重要.
更多相关视频
05:47Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
575
07:35A Knowledge Graph Approach to Elucidate the Role of Organellar Pathways in Disease via Biomedical Reports
Published on: October 13, 2023
1.7K
相关概念视频
Language and Cognition
440
Language serves as a bridge between ideas and communication, influencing how individuals perceive and interact with the world. Psychologists have long debated whether language shapes thought or vice versa. This discussion gained grip with Edward Sapir and Benjamin Lee Whorf in the 1940s, who proposed that language determines thought, a concept known as linguistic determinism. They suggested that the vocabulary and structure of a language influence how its speakers think and perceive reality.
440
Inductive Effects on Chemical Shift: Overview
1.3K
The protons in unsubstituted alkanes are strongly shielded with chemical shifts below 1.8 ppm. Methine, methylene, and methyl protons appear at approximately 1.7, 1.2 and 0.7 ppm, while the proton signal from methane appears at 0.23 ppm. An electronegative substituent, such as chlorine, withdraws the electron density from the protons, increasing their chemical shift. Progressive substitution of the hydrogens in methane by chlorine shifts the proton signals increasingly downfield, to 3.05 ppm in...
1.3K
The Small x Assumption
46.5K
If a reaction has a small equilibrium constant, the equilibrium position favors the reactants. In such reactions, a negligible change in concentration may occur if the initial concentrations of reactants are high and the Kc value is small. In such circumstances, the equilibrium concentration is approximately equal to its initial concentration. This estimation can be used to simplify the equilibrium calculations by assuming that some equilibrium concentrations are equal to the initial...
46.5K
Inductive Reasoning
62.7K
Inductive reasoning is a form of logical thinking that uses related observations to arrive at a general conclusion. It is uncertain and operates in degrees to which the conclusions are credible. As such, inductive arguments can be weak or strong, rather than valid or invalid, and conclusions can be used to formulate testable, falsifiable hypotheses.
Inductive reasoning is common in descriptive science. A life scientist makes observations and records them. This data can be qualitative or...
Inductive reasoning is common in descriptive science. A life scientist makes observations and records them. This data can be qualitative or...
62.7K
Predicting Molecular Geometry
35.9K
VSEPR Theory for Determination of Electron Pair Geometries
35.9K
Reaction Quotient
49.1K
The status of a reversible reaction is conveniently assessed by evaluating its reaction quotient (Q). For a reversible reaction described by m A + n B ⇌ x C + y D, the reaction quotient is derived directly from the stoichiometry of the balanced equation as
49.1K
