Xie Saining’s Team Releases Second-Generation Representation Autoencoder RAEv2

刚刚，谢赛宁团队放出第二代表征自编码器

Xie Saining’s team, Adobe Research, and the Australian National University released RAEv2, which reaches gFID 1.06 after 80 epochs on ImageNet-256 and reduces EPFID@2 from 177 epochs to 35 epochs while keeping compute at 189 GFLOPs.

Why it matters: HKR-K and HKR-R pass with concrete benchmark and training-efficiency claims. HKR-H is weak because the angle is a normal research release, so it lands at the featured threshold rather than a must-write item.