How to use this VAE on LCM

[alfredplpl]

Thanks to open your work. Btw, I ran the code:

from diffusers import DiffusionPipeline,StableDiffusionPipeline
import torch
from consistencydecoder import ConsistencyDecoder, save_image, load_image

pipe = DiffusionPipeline.from_pretrained("SimianLuo/LCM_Dreamshaper_v7", custom_pipeline="latent_consistency_txt2img", custom_revision="main", revision="fb9c5d")

decoder_consistency = ConsistencyDecoder(device="cuda:0") # Model size: 2.49 GB

pipe.to(torch_device="cuda", torch_dtype=torch.float32)

prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"

# Can be set to 1~50 steps. LCM support fast inference even <= 4 steps. Recommend: 1~8 steps.
num_inference_steps = 4

latent = pipe(prompt=prompt, num_inference_steps=num_inference_steps, guidance_scale=8.0, output_type="latent").images[0]

latent=latent.unsqueeze(0).to("cuda:0")

sample_consistency = decoder_consistency(latent)
save_image(sample_consistency, "con.png")

I got the image:

What is wrong?