Update app.py

app.py

@@ -75,7 +75,7 @@ else:
 device = torch.device('cuda')
 
 base_model = "black-forest-labs/FLUX.1-dev"
-pipe = FluxPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16, token=huggingface_token).to(device)
+pipe = FluxPipeline.from_single_file("https://huggingface.co/marduk191/Flux.1_collection/resolve/main/flux.1_dev_fp8_fp16t5-marduk191.safetensors", torch_dtype=torch.bfloat16, token=huggingface_token).to(device)
 
 # Load and fuse LoRA BEFORE quantizing
 print('Loading and fusing lora, please wait...')