search

tencent-ailab / IP-Adapter

The image prompt adapter is designed to enable a pretrained text-to-image diffusion model to generate images with image prompt.
Apache License 2.0
4.46k stars 289 forks source link

train #356

Open masaisai111 opened 1 month ago

masaisai111 commented 1 month ago

May I ask why the author does not want to use pipe to load each module, such as unet, when reasoning during training

xiaohu2015 commented 1 month ago

you can also use pipe to load each module, here I followed diffusers repo

masaisai111 commented 1 month ago

image Here I think, pipe.unet=unet(load alone), is this the correct understanding

masaisai111 commented 1 month ago

here，base_model_path==args.pretrained_model_name_or_path