update

hf_demo.py

@@ -11,8 +11,9 @@ from PIL import Image
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float16
+print(f"Using {device} device, dtype={dtype}")
 pipe = DiffusionPipeline.from_pretrained("rhfeiyang/art-free-diffusion-v1",
-                                            dtype=dtype).to(device)
+                                         torch_dtype=dtype).to(device)
 
 from inference import get_lora_network, inference, get_validation_dataloader
 lora_map = {
@@ -42,7 +43,7 @@ def demo_inference_gen(adapter_choice:str, prompt:str, samples:int=1,seed:int=0,
         adapter_path = f"data/Art_adapters/{adapter_path}/adapter_alpha1.0_rank1_all_up_1000steps.pt"
 
     prompts = [prompt]*samples
-    infer_loader = get_validation_dataloader(prompts)
+    infer_loader = get_validation_dataloader(prompts,num_workers=0)
     network = get_lora_network(pipe.unet, adapter_path)["network"]
     pred_images = inference(network, pipe.tokenizer, pipe.text_encoder, pipe.vae, pipe.unet, pipe.scheduler, infer_loader,
                             height=512, width=512, scales=[1.0],
@@ -52,7 +53,7 @@ def demo_inference_gen(adapter_choice:str, prompt:str, samples:int=1,seed:int=0,
     return pred_images
 @spaces.GPU
 def demo_inference_stylization(adapter_path:str, prompts:list, image:list, start_noise=800,seed:int=0):
-    infer_loader = get_validation_dataloader(prompts, image)
+    infer_loader = get_validation_dataloader(prompts, image,num_workers=0)
     network = get_lora_network(pipe.unet, adapter_path,"all_up")["network"]
     pred_images = inference(network, pipe.tokenizer, pipe.text_encoder, pipe.vae, pipe.unet, pipe.scheduler, infer_loader,
                             height=512, width=512, scales=[0.,1.],