@misc{indiciaeeee0af7f5d53,
  title = {SRA 2: Variational Autoencoder Self-Representation Alignment for Efficient Diffusion Training},
  author = {Mengmeng Wang and Dengyang Jiang and Liuzhuozheng Li and Yucheng Lin and Guojiang Shen and Xiangjie Kong and Yong Liu and Guang Dai and Jingdong Wang},
  year = {2026},
  url = {https://arxiv.org/abs/2601.17830},
  note = {Source identifier: 2601.17830}
}