Abstract: We introduce SODA, a self-supervised diffusion model, designed for representation learning. The model incorpo-rates an image encoder, which distills a source view into a compact ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results