Thanks for your implementation. I have a question about the tuning stage in PTI code (lora_diffusion/cli_lora_pti.py). I noticed that the prior-preservation class images are not used in this stage, lacking of related args. The only loss is the masked reconstruction constraint. Is the prior-preservation not implemented in phase 2? Then, could you provide a suitable PTI config for person(face) in PTI code? Thanks~
Thanks for your implementation. I have a question about the tuning stage in PTI code (lora_diffusion/cli_lora_pti.py). I noticed that the prior-preservation class images are not used in this stage, lacking of related args. The only loss is the masked reconstruction constraint. Is the prior-preservation not implemented in phase 2? Then, could you provide a suitable PTI config for person(face) in PTI code? Thanks~