Custom pipeline : lpw_stable_diffusion is incompatible with new version of diffusers

[adhikjoshi]

Describe the bug

DiffusionPipeline.from_pretrained(
                model_id, custom_pipeline="lpw_stable_diffusion",
                torch_dtype=torch.float16, revision="fp16",
                safety_checker=None, vae=vae, use_auth_token=hf_token
            ).to("cuda")

Gives this error

module transformers has no attribute CLIPImageProcessor

Reproduction

No response

Logs

/opt/conda/envs/ldm/lib/python3.8/site-packages/huggingface_hub/file_download.py:592: FutureWarning: `cached_download` is the legacy way to download files from the HF hub, please consider upgrading to `hf_hub_download`
  warnings.warn(
The config attributes {'requires_safety_checker': True} were passed to StableDiffusionLongPromptWeightingPipeline, but are not expected and will be ignored. Please verify your model_index.json configuration file.
Keyword arguments {'requires_safety_checker': True} not recognized.
module transformers has no attribute CLIPImageProcessor

System Info

• transformers==4.19.2
• python=3.8.5
• pip=20.3
• diffusers==0.8.1

[adhikjoshi]

After some trial and errors, it turns out that StableDiffusionLongPromptWeightingPipeline is not working with latest transformers and diffusers.

@SkyTNT Would you update package to support latest versions of both?
@patrickvonplaten @irgolic If you could weigh in,

[SkyTNT]

I will take a look.

[adhikjoshi]

This has different set of errors now

#1607