Update app.py

app.py

@@ -36,8 +36,11 @@ def generate_response(message, history):
     inputs["token_type_ids"] = torch.ones_like(inputs["input_ids"])
     
     with torch.no_grad():
-        out = model.generate(**inputs, max_new_tokens=256, do_sample=False)
-    return tokenizer.decode(out[0], skip_special_tokens=False)
+        text = model.generate(**inputs, max_new_tokens=256, do_sample=False)
+    start = text.find('<|object_ref_end|>') + len('<|object_ref_end|>')
+    end = text.find('<|box_end|>')
+    result = text[start:end]
+    return tokenizer.decode(result, skip_special_tokens=False)
 chatbot = gr.ChatInterface(
     fn=generate_response,
     title="Simple ZeroGPU Chatbot",