Beijun D.

No worries! I was thinking of switching it to something other than streamlit anyway. Thank you for your help!

Thank you for trying! I appreciate all of the help!

Dustin N. It's about 30 traces for 5-7 minutes. I've tried doing it with about 90 traces which took about 18 minutes

Hi Dustin N. I'm not running anything in a Jupyter notebook. I've broken up the evaluate_rag.ipynb code into multiple async functions which I then call in the streamlit app.

    async def _executor(self):
        nest_asyncio.apply()
        session = px.launch_app()
        # print(f"🚀 Open the Phoenix UI if you haven't already: {session.url}")
        await self._generate_traces()
            

        await self._relevanceEvaluator()
        await self._qa_hallucination_evaluator() 

All the async functions being broken down contain pretty much the same code in the notebook with the exception of the model. Sometimes I use OpenAIModel and sometimes LiteLLM.

Roger Y. Yes, whenever the app is run.