sd1.5 diff model get Errors.
I switched my controlnet to boringcrypto/sd-webui-controlnet, downloaded the "diffusion_pytorch_model.fp16.safetensors" in the diffusion_sd15 folder, prompted a very long list of errors:
.............
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for ControlNet:
Missing key(s) in state_dict: "time_embed.0.weight",...
.............
The following content is too long, if you need I can copy it down.
Then I downloaded "control_mediapipe_face_sd15_v2.safetensors" and worked perfectly.Thanks for your work, it's really great.
Maybe there are some problems when extracting the diff model?
Hi there. The boringcrypto branch does not belong to us and we cannot vouch for the contents. Ours is https://github.com/crucible-ai/sd-webui-controlnet Nonetheless, thank you for the report.
diffusion_pytorch_model.fp16.safetensors is for diffusers and would be used if you happen to use the HuggingFace pipeline.
control_mediapipe_face_sd15_v2.safetensors is for the original controlnet and would be used if you're using a web ui or something based on the original repo.