search

wfanyue / DPG-T2I-Personalization

[ECCV 2024] Powerful and Flexible: Personalized Text-to-Image Generation via Reinforcement Learning
36 stars 1 forks source link

Personalization Alignment #2

Open alnaeini opened 4 months ago

wfanyue commented 4 months ago

Hi, could you provide more explanation about "Personalization Alignment"?

alnaeini commented 4 months ago

Sure.

As mentioned in the paper, The current reward model using DINO embedding to increase image fidelity(ImageAlignment). My questions was when DINO reward implementation going to be released?

wfanyue commented 3 months ago

Thanks for your interest! I will release as soon as possible on recent 1~2 weekends. I will sync with you on this issue when I update the code.