Improving Molecular Property Prediction in Small Language Models Using Graph-based Tools
使用基于图的工具改进小语言模型中的分子性质预测
机构 * Institute of Informatics and Telecommunications, National Center for Scientific Research “Demokritos(信息与电信研究所,国家科学研究中心“德谟克利特”)
专题命中 规划决策 :agentic(abstract,comments);tool use(abstract);分类 cs.AI、cs.LG
AI总结 研究针对小语言模型在分子性质预测时的结构盲目性问题,提出模块化上下文增强提示框架,通过GNN专家模型和提取子图来辅助预测,在MUTAG和Tox21数据集上实验表明该方法能显著提升预测准确性,验证了基序相关性,但与专门GNN模型仍有差距。
Comments Presented at the 2nd Causal Neuro-symbolic Artificial Intelligence (Causal NeSy): Toward Agentic LLMs with Neuro-Symbolic and Graph Based Reasoning Workshop @ ESWC2026