כתבה
arXiv cs.AI ·
HiRAE: Hierarchical Representation Autoencoding with Residual Budgets
HiRAE: פיתוח פרקטיקה היררכית לאוטוקודינג של ייצוגים ויזואליים, עם תקציבים של תיקוני תצלומים.
תקציר מקורי באנגליתarXiv:2609.37775v1 Announce Type: cross Abstract: Pretrained visual representations support image generation, but may not fully preserve the fine-grained details needed for faithful reconstruction. Meanwhile, intermediate encoder layers contain complementary visual details, but learning to fuse them for reconstruction can produce a latent distribution that is difficult to model. Existing fusion methods require empirical tuning of layer selection or staged optimization of fusion and decoding, increasing configuration effort or training complexity. We introduce HiRAE (Hierarchical Representation Autoencoder), which learns a hierarchical fusion framework over the full encoder hierarchy to improve reconstruction fidelity while maintaining compatibility with generative modeling. HiRAE groups en
קרא במקור המקורי
arxiv.org
פתח כתבה מקורית