Newest papers

Efficient Edge-aware Attention Network for Graph Generation

Vincent Jung, Alba Carballo-Castro, Yiming Qin, Lonneke van der Plas, Pascal Frossard B · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28618 · cited by 0 works (OpenAlex) · open-access PDF

French

Les modèles de génération de graphes utilisent généralement des architectures de transformateur spécifiques aux graphes pour mettre à jour conjointement les caractéristiques des nœuds et des bords. Cependant, ces architectures ne bénéficient pas directement des récentes implémentations d'attention consciente du matériel, limitant leur évolutivité. L'adaptation est cependant non triviale: les caractéristiques de bord et de niveau de graphes doivent être injectées dans des scores d'attention parallèles, les masques de graphes doivent être respectés et les mises à jour dynamiques de bord restent une source majeure d'utilisation de la mémoire carré. Nous proposons une couche d'attention consciente du bord pour la génération de graphes qui intègre des informations structurelles

Machine translation (Meta M2M100); the English original follows.

Abstract

Graph generative models commonly use graph-specific Transformer architectures to jointly update node and edge features. However, these architectures do not directly benefit from recent hardware-aware attention implementations, limiting their scalability. The adaptation is, however, non-trivial: edge and graph-level features must be injected into pairwise attention scores, graph masks must be respected, and dynamic edge updates remain a major source of quadratic memory usage. We propose an edge-aware attention layer for graph generation that incorporates structural information through head-wise score modulation, while updating edge representations with a lightweight residual mechanism that retains attention-score dependent edge updates. This layer can be effectively implemented using PyTorch's FlexAttention. We achieve similar generation quality to our baseline while reducing peak GPU memory by around 65% and substantially improving training and sampling throughput. These results suggest that the proposed architecture is a practical step toward scalable graph foundation models.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Scalability, Graph, Theoretical computer science, Limiting, Residual

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28618, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange