Bug: set adapters when model is compiled

[chenxiao111222]

diffusers/src/diffusers/loaders/lora_base.py

Line 839 in 7fc53b5

and issubclass(model.__class__, (ModelMixin, PreTrainedModel))

I note that transformer.__class__ is changed from diffusers.models.transformers.transformer_flux.FluxTransformer2DModel to torch._dynamo.eval_frame.OptimizedModule when I use torch.compile, so I can't use set_adapters.

[a-r-r-o-w]

cc @sayakpaul @BenjaminBossan

Maybe we can do an unwrap model on the model to obtain its class here?

[a-r-r-o-w]

Actually, seems like a duplicate of #11408, so I'm marking this as closed