研究人员开发了空间推测解码 (SSD)，这是一个旨在加速自回归图像生成的新框架。该方法通过利用图像固有的二维空间局部性，解决了将图像视为一维序列所带来的计算瓶颈。SSD 同时预测相邻的水平和向下标记，从而在保持 DPG-Bench 和 GenEval 等基准测试的高保真度的同时，将推理速度提高了 13.3 倍。 AI
影响 该方法可以实现实时、高分辨率的自回归图像生成，显著提高 AI 驱动的视觉内容创作的效率。
排序理由 该集群包含一篇详细介绍图像生成新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2606.20543v1 Announce Type: new Abstract: Autoregressive models excel in visual generation by treating images as 1D sequences of discrete tokens, mirroring language modeling. However, this flattening discards the intrinsic 2D spatial locality of visual signals, creating sev…
Autoregressive models excel in visual generation by treating images as 1D sequences of discrete tokens, mirroring language modeling. However, this flattening discards the intrinsic 2D spatial locality of visual signals, creating severe computational bottlenecks during inference. …