'StableDiffusionInpaintPipeline' has no attribute 'from_ckpt'

[Pirog17000]

Describe the bug

Title is self-explanatory. There's no way to load inpaint model from safetensors using from_ckpt

I'm using the same approach as written in 'Load Safetensors' docs page:

pipeline = StableDiffusionPipeline.from_ckpt(
    "https://huggingface.co/WarriorMama777/OrangeMixs/blob/main/Models/AbyssOrangeMix/AbyssOrangeMix.safetensors"
)

but for Inpaint pipeline it doesn't work:
pipe = StableDiffusionInpaintPipeline.from_ckpt("downloaded_inpaint_model.safetensors")
throws an error 'StableDiffusionInpaintPipeline' has no attribute 'from_ckpt'

Maybe I'm just doing something wrong or understood it wrong and inpaint models should be loaded from safetensor other way?

Reproduction

Logs

No response

System Info

• diffusers version: 0.17.0.dev0
• Platform: Windows-10-10.0.19045-SP0
• Python version: 3.10.9
• PyTorch version (GPU?): 1.13.1+cu117 (True)
• Huggingface_hub version: 0.13.2
• Transformers version: 4.28.1
• Accelerate version: 0.17.0.dev0
• xFormers version: 0.0.17+57edbb7.d20230228

[Pirog17000]

pipe = StableDiffusionInpaintPipeline.from_ckpt("downloaded_inpaint_model.safetensors", use_safetensors=True)
loads the pipeline, but later on throws another error:

Error(s) in loading state_dict for UNet2DConditionModel:
        size mismatch for conv_in.weight: copying a param with shape
torch.Size([320, 9, 3, 3]) from checkpoint, the shape in current model is
torch.Size([320, 4, 3, 3]).

[Pirog17000]

converted file with --num_in_channels 9 into diffusers, expected to see single 'safetensors', got a folder with bunch of safetensors. Anyway, seems it's working, but not as expected from the beginning :)

Any way to convert 'to diffusers into single .safetensors file and read it easily?'

[patrickvonplaten]

Will try to make sure that the StableDiffusionInpaintPipeline also works with native SD v1-5 checkpoint.

[patrickvonplaten]

See: #3533

[sayakpaul]

Since #3533 is merged, I am closing this issue. Feel free reopen in case problems persist.