test input

app.py

@@ -72,7 +72,7 @@ def deepseek(image, text_input):
     conversation = [
         {
             "role": "<|User|>",
-            "content": f"<image>\n<|ref|>{text_input}<|/ref|>.",
+            "content": f"<image><|ref|>{text_input}<|/ref|>.",
             "images": ["./images/visual_grounding_1.jpeg"],
         },
         {"role": "<|Assistant|>", "content": ""},