Update app.py

app.py

@@ -5,20 +5,40 @@ import modin.pandas as pd
 from PIL import Image
 from diffusers import StableDiffusion3Pipeline #DiffusionPipeline #, StableDiffusion3Pipeline
 from huggingface_hub import hf_hub_download
+from diffusers import BitsAndBytesConfig, SD3Transformer2DModel
 
 device = 'cuda' if torch.cuda.is_available() else 'cpu'
 torch.cuda.max_memory_allocated(device=device)
 torch.cuda.empty_cache() 
 
-pipe = StableDiffusion3Pipeline.from_pretrained("stabilityai/stable-diffusion-3.5-large-turbo", torch_dtype=torch.bfloat16)
-pipe = pipe.to(device)
-pipe.enable_model_cpu_offload()
+model_id = "stabilityai/stable-diffusion-3.5-large-turbo"
 
+nf4_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype=torch.bfloat16
+)
+model_nf4 = SD3Transformer2DModel.from_pretrained(
+    model_id,
+    subfolder="transformer",
+    quantization_config=nf4_config,
+    torch_dtype=torch.bfloat16
+)
+
+t5_nf4 = T5EncoderModel.from_pretrained("diffusers/t5-nf4", torch_dtype=torch.bfloat16)
+
+pipeline = StableDiffusion3Pipeline.from_pretrained(
+    model_id, 
+    transformer=model_nf4,
+    text_encoder_3=t5_nf4,
+    torch_dtype=torch.bfloat16
+)
+pipeline.enable_model_cpu_offload()
 
 def genie (Prompt, height, width, seed):
     generator = np.random.seed(0) if seed == 0 else torch.manual_seed(seed)
        
-    image = pipe(Prompt, num_inference_steps=4, height=height, width=width, guidance_scale=0.0,).images[0]
+    image = pipeline(Prompt, num_inference_steps=4, height=height, width=width, guidance_scale=0.0,).images[0]
 
     return image