Add pruna integration for loading model through diffusers.from_pretrained / pipeline

[davidberenstein1957]

Is your feature request related to a problem? Please describe.
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...].

Pruna is an open-source AI model optimisation framework.
As discussed with @SunMarc about doing this for transformers but we could do something similar for diffusers as well. Something like *Pipeline.from_pretrained interface as an alternative to the PrunaModel interface.

Currently, the code looks as follows.

Currently, the code looks as follows.

from pruna import PrunaModel

loaded_model = PrunaModel.from_hub(
    "PrunaAI/FLUX.1-dev-smashed"
)

We could go for something like.

import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained("PrunaAI/FLUX.1-dev-smashed")

Describe the solution you'd like.
It would be a nice integration.

Describe alternatives you've considered.
We could also add the library as an explicit tab selector within the Hub, similar to llama-cpp/unsloth and other frameworks.

Additional context.
NA

[yiyixuxu]

cc @sayakpaul here
I think you just need to register with hub to add a tab selector, no?

[sayakpaul]

Yeah that is what my feeling is.