Update app.py

app.py

@@ -61,7 +61,8 @@ def generate_image_embeddings(prompt_image,
                               device='cuda:0'):
     prompt_image = prompt_image.convert('RGB')
     inputs = vision_processor(images=prompt_image,
-                              return_tensors="pt")
+                              return_tensors="pt",
+                              padding=True)
     inputs = inputs.to(device)
     image_outputs = vision_encoder(**inputs)
     img_feats = image_outputs.image_embeds.view(1, -1)