将大模型与三维对称性结合，提升物理系统建模精度
Large Language-Geometry Model: When LLM meets Equivariance
- 用提示词引导大模型处理不变特征，专用模块处理空间方向信息
- 在分子动力学、人体运动模拟等任务上超越现有方法
- 适合需要精准空间推理的科学计算场景
准确预测物理系统的三维结构与动态是科学应用的关键。现有基于几何图神经网络（GNN）的方法能有效保证E(3)对称性，但难以利用广泛外部信息；而直接使用大语言模型（LLM）虽可融入外部知识，却缺乏保障对称性的空间推理能力。本文提出EquiLLM框架，将E(3)对称性与LLM能力无缝融合。该框架包含四个核心组件：几何感知提示、等变编码器、大语言模型和等变适配器。其中，由指令提示引导的LLM作为复杂不变特征处理器，三维方向信息则由等变编码器与适配器模块专门处理。实验表明，EquiLLM在分子动力学模拟、人体运动模拟和抗体设计任务中均显著优于先前方法，展现出优异的泛化能力。
原文摘要 · Abstract (English)
Accurately predicting 3D structures and dynamics of physical systems is crucial in scientific applications. Existing approaches that rely on geometric Graph Neural Networks (GNNs) effectively enforce $\mathrm{E}(3)$-equivariance, but they often fall in leveraging extensive broader information. While direct application of Large Language Models (LLMs) can incorporate external knowledge, they lack the capability for spatial reasoning with guaranteed equivariance. In this paper, we propose EquiLLM, a novel framework for representing 3D physical systems that seamlessly integrates E(3)-equivariance with LLM capabilities. Specifically, EquiLLM comprises four key components: geometry-aware prompting, an equivariant encoder, an LLM, and an equivariant adaptor. Essentially, the LLM guided by the instructive prompt serves as a sophisticated invariant feature processor, while 3D directional information is exclusively handled by the equivariant encoder and adaptor modules. Experimental results demonstrate that EquiLLM delivers significant improvements over previous methods across molecular dynamics simulation, human motion simulation, and antibody design, highlighting its promising generalizability.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。