NEAT:面向 3D 分子生成的邻域引导高效自回归 Set Transformer
NEAT: Neighborhood-Guided, Efficient, Autoregressive Set Transformer for 3D Molecular Generation
AI 导读
研究者提出 NEAT(Neighborhood-Guided, Efficient, Autoregressive Set Transformer),一种用于 3D 分子生成的自回归 Transformer 模型,采用邻域引导训练策略,将分子图视为原子集合,学习与顺序无关的分布,保证原子级置换不变性。
正文 · 原文
Abstract:Transformer-based autoregressive models offer an efficient alternative to diffusion- and flow-based approaches for generating 3D molecules. One challenge remains: standard transformer architectures require a sequential ordering of tokens, which is not inherently defined for the atoms in a molecule. Prior works have addressed this by using canonical atom orderings. However, these approaches are not permutation invariant w.r.t. atoms and bias next-token prediction towards ordering conventions. We overcome this limitation by introducing a novel neighborhood-guided training strategy. Our model, NEAT (Neighborhood-Guided, Efficient, Autoregressive Set Transformer) treats molecular graphs as sets of atoms and learns an order-agnostic distribution over admissible tokens at the graph boundary, thereby ensuring atom-level permutation invariance. NEAT achieves state-of-the-art generation quality on the QM9 and GEOM-Drugs datasets, enables seamless completion of molecular prefixes, and offers a significant speed advantage over existing baselines.
| Comments: | NeurIPS 2026 |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2512.05844 [cs.LG] |
| (or arXiv:2512.05844v4 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2512.05844 arXiv-issued DOI via DataCite |
Submission history
From: Daniel Rose [view email]
[v1]
Fri, 5 Dec 2025 16:18:07 UTC (1,529 KB)
[v2]
Thu, 29 Jan 2026 13:23:10 UTC (9,054 KB)
[v3]
Tue, 5 May 2026 18:38:58 UTC (11,570 KB)
[v4]
Fri, 13 Nov 2026 16:53:21 UTC (11,795 KB)
来源:arXiv cs.AI · arxiv.org