search

wfanyue / DPG-T2I-Personalization

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

Releasing SDXL version? #1

Closed alnaeini closed 3 months ago

alnaeini commented 4 months ago

Any plan on releasing training with SDXL?

wfanyue commented 4 months ago

Hi, Thanks for your interest in our work. We do not conduct experiments with SDXL baseline. But the implementation of the reward model used in our method is simple, you can refer to train_dreambooth_dpg.py and modify the pre-trained model to SDXL. If you have any questions, feel free to contact me.

alnaeini commented 4 months ago

Thanks