demo fix commit

app.py

@@ -38,7 +38,7 @@ def inference(raw_image):
     image = transform(raw_image).unsqueeze(0).to(device)
     with torch.no_grad():
           caption = model.generate(image, sample=False, num_beams=1, max_length=60, min_length=5) 
-      return caption[0]
+    return caption[0]
 
 
 inputs = [gr.Image(type='pil', interactive=False),]