Jul 2026
Scaling Properties of Text Conditioning in Visual Generation
Surprisingly, it is found that the converged diffusion loss scales with the amount of structured language in the prompt, and two complementary measures are adapted: a white-box likelihood metric (GPG) and a black-box attribute metric (ED).
Zilong Chen, Chao-Rui Deng, Kunchang Li et al.
· arXiv.org · 2 citations