使用基于图的工具改进小语言模型中的分子性质预测
英文标题:Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools
小型语言模型在基于SMILES的零样本分子性质预测中存在结构盲区。该研究提出上下文增强提示框架,在推理时调用图神经网络工具,提供预测提示与解释子图。在MUTAG和Tox21上的实验显示,丰富提示可带来超过25%的相对准确率提升,Tox21最高提升74%,并通过边丢弃干预验证了基序的功能相关性。该工作展示了将图结构信息注入语言模型推理的有效性,但性能仍不及专用GNN模型。
640 B
612 B
380 B
9.1 KB
32.7 KB
22.8 KB
24.2 KB
34.7 KB
中文
[提交于 2026年7月14日]
使用基于图的工具改进小语言模型中的分子性质预测
查看题为《使用基于图的工具改进小语言模型中的分子性质预测》的论文PDF,作者为 Konstantinos Bougiatiotis 与另一位合作者
摘要:小语言模型(SLMs)在基于SMILES字符串的零样本分子性质预测中展现出潜力,但由于序列表示未能充分提供关键的图拓扑线索,它们常常存在结构盲区。我们提出了一种模块化的上下文增强提示框架,可在推理时启用代理式工具使用:训练好的图神经网络专家模型提供带有置信度的预测提示,另一个GNN则提取特定于实例的解释性子图(例如,子图SMILES和相应的解释性段落)。我们在MUTAG和Tox21数据集上评估了三种常用SLM,测试了从仅使用SMILES到使用所有可用工具的五种提示配置。在两个数据集上,使用图衍生上下文丰富提示可带来显著的准确率提升,通常相对提高超过25%,在Tox21上最高提升74%。我们进一步通过基于必要性的边丢弃干预验证了所提取基序的功能相关性。尽管取得了这些提升,与专用GNN模型相比仍存在持续差距,这凸显了文本条件推理在分子结构中的价值与局限性。
提交历史
来自: Konstantinos Bougiatiotis [ 查看邮箱 ]
[v1] 2026年7月14日周二 13:10:30 UTC (1,567 KB)
English / 原文
[Submitted on 14 Jul 2026]
Title:Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools
View a PDF of the paper titled Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools, by Konstantinos Bougiatiotis and 1 other authors
Abstract:Small language models (SLMs) have shown promise for zero-shot molecular property prediction from SMILES strings, yet they often suffer from structural blindness because sequence representations under-specify key graph-topological cues. We propose a modular Context-Augmented Prompting framework that enables agentic tool use at inference time: a trained GNN expert model provides a predictive hint with confidence, and a GNN extracts an instance-specific explanatory subgraph (e.g., a subgraph SMILES and an accompanying explanatory paragraph). We evaluate three commonly used SLMs on MUTAG and Tox21 under five prompting configurations ranging from SMILES-only to using all available tools at hand. Across two datasets, enriching prompts with graph-derived context yields substantial accuracy gains, often exceeding 25% relative improvement and up to 74% on Tox21. We further validate the functional relevance of the extracted motifs via a necessity-based edge-drop intervention. Despite the observed gains, a persistent gap remains to specialized GNN models, highlighting both the value and limits of text-conditioned reasoning for molecular structure.
Submission history
From: Konstantinos Bougiatiotis [
]
[v1]
Tue, 14 Jul 2026 13:10:30 UTC (1,567 KB)