fix attempt 2

app.py

@@ -32,7 +32,7 @@ generator = torch.manual_seed(0)
 def infer(prompt, negative_prompt, image):
     # implement your inference function here
     
-    cond_input = conditioning_image_transforms(image)
+    cond_input = conditioning_image_transforms(T.PILToTensor(image))
     
     output = pipe(
         prompt,
@@ -42,7 +42,7 @@ def infer(prompt, negative_prompt, image):
         num_inference_steps=20
             )
 
-    return torch.tensor(output[0])
+    return output[0]
 
     # you need to pass inputs and outputs according to inference function
 title = "Categorical Conditioning Controlnet for One-Shot Image Stylization."