How to reproduce results of huggingface api using diffuser

Hi, I train a diffuser model and push to the hub. And I find that I cannot reproduce the result of the hub api using codes like this and it seems the results in the API are better.

from diffusers import StableDiffusionPipeline, EulerDiscreteScheduler
import torch
model_id = "my_repo"
scheduler = EulerDiscreteScheduler.from_pretrained(model_id, subfolder="scheduler")
pipe = StableDiffusionPipeline.from_pretrained(model_id, scheduler=scheduler, torch_dtype=torch.float16)
pipe ="cuda")

prompt = "A photo of the sks face of the nurse"
image = pipe(prompt).images[0]"generated_image.png")

Thanks a lot!