#import streamlit as st
#from transformers import pipeline

#pipe = pipeline('sentiment-analysis')
#text = st.text_area('enter some text!')

#if text:
#   out = pipe(text)
#   st.json(out)
#   

  
      # !pip install diffusers transformers
from diffusers import DiffusionPipeline

model_id = "CompVis/ldm-text2im-large-256"

# load model and scheduler
ldm = DiffusionPipeline.from_pretrained(model_id)

# run pipeline in inference (sample random noise and denoise)
prompt = "A painting of a squirrel eating a burger"
images = ldm([prompt], num_inference_steps=50, eta=0.3, guidance_scale=6)["sample"]

# save images
for idx, image in enumerate(images):
    image.save(f"squirrel-{idx}.png")