Deploy expt8 Gradio app

gradio_app.py

@@ -16,10 +16,12 @@ from pydantic import ValidationError
 from app import AskRequest
 try:  # ZeroGPU Spaces require at least one @spaces.GPU function; no-op elsewhere
     import spaces
-    gpu = spaces.GPU(duration=60)
+
+    @spaces.GPU
+    def _gpu_stub():  # never called: the pipeline is CPU-only (FAISS + MiniLM) with a Groq-hosted LLM
+        return None
 except ImportError:
-    def gpu(fn):
-        return fn
+    pass
 from support_rag import BackendError, SupportRAG
 
 logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(name)s %(message)s")
@@ -62,7 +64,6 @@ def metrics(latency_ms: float, intent: str, retries: int) -> str:
         for l, v in cells) + "</div>"
 
 
-@gpu
 def ask(question: str):
     """Returns (answer_html, metrics_html, sources_markdown, sources_accordion_visible)."""
     try: