"does not have a parameter or a buffer named input_scale" in quanto FluxTransformer2DModel

[LianShuaiLong]

@sayakpaul
here is my quanto code:

from optimum.quanto.models import QuantizedDiffusersModel
class QuantizedFluxTransformer2DModel(QuantizedDiffusersModel):
base_class = FluxTransformer2DModel
model = FluxTransformer2DModel.from_pretrained("./FLUX.1-schnell",subfolder="transformer")
qmodel = QuantizedFluxTransformer2DModel.quantize(model,weights=qfloat8)
qmodel.save_pretrained("./FLUX.1-schnell-fp8")

no errors in quanto process
but valueError raise when infer with the quanto model:
ValueError: Linear(in_features=4096, out_features=3072, bias=True) does not have a parameter or a buffer named input_scale
here is my inference code:

pipeline = FluxPipeline.from_pretrained(
"./FLUX.1-schnell",
transformer=None,
torch_dtype=torch.float16
).to("cuda")
transformer = FluxTransformer2DModel.from_pretrained("./FLUX.1-schnell-fp8")
transformer.to("cuda",dtype=torch.float16)
pipeline.transformer = transformer
prompt = "a cat"
image = pipeline(
prompt=prompt,
num_inference_steps=4,
num_images_per_prompt=1,
width=512,
height=512,
generator = torch.Generator.manual_seed(42)
).images[0]
image.save('cat.png')