update top k

Author: vbarda

backend/retrieval_graph/graph.py

@@ -204,7 +204,7 @@ async def respond(
     configuration = AgentConfiguration.from_runnable_config(config)
     model = load_chat_model(configuration.response_model)
     # TODO: add a re-ranker here
-    top_k = 10
+    top_k = 20
     context = format_docs(state.documents[:top_k])
     prompt = configuration.response_system_prompt.format(context=context)
     messages = [{"role": "system", "content": prompt}] + state.messages