Upload app.py

app.py

@@ -796,7 +796,10 @@ def generate_response(prompt: str, max_new_tokens: int = MAX_NEW_TOKENS) -> str:
     """Generate response using vLLM OpenAI-compatible API."""
     # Use /completions endpoint for raw prompt
     url = f"{REMOTE_API_BASE}/completions"
-    headers = {"Content-Type": "application/json"}
+    headers = {
+        "Content-Type": "application/json",
+        "ngrok-skip-browser-warning": "true",  # 绕过 ngrok 免费版的浏览器警告页面
+    }
     payload = {
         "model": MODEL_NAME,
         "prompt": prompt,