Abstract: Text-to-image diffusion model alignment is critical for improving the alignment between the generated images and human preferences. While training-based methods are constrained by high ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback