Using the public deployment

Seems that providing an input image with transparency (for outpainting, for example) throws an error:
The size of tensor a (768) must match the size of tensor b (96) at non-singleton dimension 3
the paranthesed numbers being dependent on the dimensions of the provided image.

https://github.com/fofr/cog-sdxl-multi-controlnet-lora, which this package is based on, does not throw this error