清华与阿里联合论文提出ViT³：线性复杂度视觉Transformer，CVPR 2026 Oral
Tsinghua and Alibaba Joint Paper Introduces ViT³: A Vision Transformer with Linear Complexity — CVPR 2026 Oral
视觉Transformer终于突破了计算瓶颈，做边缘计算或高分辨率图像处理的开发者可以直接关注，这可能是部署到手机等设备的关键技术。
清华大学与阿里巴巴联合发表论文，提出ViT³（Vision Test-Time Training）架构，这是一种纯Transformer模型，在视觉任务中实现了线性计算复杂度。
该架构使得在边缘设备上进行高分辨率图像理解成为可能，解决了传统Transformer在视觉应用中计算量过大的问题。该论文已被CVPR 2026接收为Oral论文，标志着视觉Transformer在效率上取得重要突破。
Tsinghua and Alibaba Joint Paper Introduces ViT³: A Vision Transformer with Linear Complexity — CVPR 2026 Oral
A joint paper from Tsinghua University and Alibaba presented at CVPR 2026 introduces ViT³ (Vision Test-Time Training), a pure transformer architecture that achieves linear computational complexity for visual tasks, enabling practical high-resolution image understanding on edge devices.
This report is categorized under AITOP's "Papers" desk, focusing on China's AI industry and domestic foundation model ecosystem. Key entities and technical topics under continuous tracking include 视觉Transformer, 线性复杂度, 边缘计算, 高分辨率图像.
Primary source verification: published directly by pandaily with first-party authority.