Update app.py

app.py

@@ -623,6 +623,36 @@ def initialize_model(self):
 #         st.error(f"Failed to initialize the system: {str(e)}")
 #         raise
 
+@st.cache_resource(show_spinner=False)
+def initialize_rag_pipeline():
+    """Initialize the RAG pipeline once"""
+    try:
+        data_folder = "ESPN_data"
+        if not os.path.exists(data_folder):
+            os.makedirs(data_folder, exist_ok=True)
+            
+        # Load embeddings first
+        drive_file_id = "1MuV63AE9o6zR9aBvdSDQOUextp71r2NN"
+        with st.spinner("Loading data..."):
+            cache_data = load_from_drive(drive_file_id)
+            if cache_data is None:
+                st.error("Failed to load embeddings from Google Drive")
+                st.stop()
+        
+        # Initialize pipeline
+        rag = RAGPipeline(data_folder)
+        
+        # Store embeddings
+        rag.documents = cache_data['documents']
+        rag.retriever.store_embeddings(cache_data['embeddings'])
+        
+        return rag
+        
+    except Exception as e:
+        logging.error(f"Pipeline initialization error: {str(e)}")
+        st.error(f"Failed to initialize the system: {str(e)}")
+        raise
+
 # def main():
 #     try:
 #         # Environment check
@@ -911,6 +941,10 @@ def initialize_model(self):
 
 def main():
     try:
+        # First, check if model exists
+        model_path = os.path.join("models", "mistral-7b-v0.1.Q4_K_M.gguf")
+        if not os.path.exists(model_path):
+            st.warning("⚠️ First-time setup: The model will be downloaded. This takes a few minutes but only happens once.")
         # Environment check
         if not check_environment():
             return