AReA: Adaptive Receptive-Field Attention Speeds Up High-Resolution Diffusion

A paper from the University of Toronto proposes AReA, a training-free adaptive receptive-field attention mechanism that tackles the quadratic token growth and heavy full-attention cost of diffusion Transformers, enabling faster and more stable high-resolution image generation.

2026-10-14 ~ 2026-10-15 · 2 related posts

• AReA: training-free adaptive receptive-field attention fixes high-res DiT generation — uoft-cs · 2026-10-14
• AReA paper brings adaptive receptive-field attention to high-res diffusion transformers — Apprehensive_Sky892 · 2026-10-15