Is there a way to only trigger output parsers when valid data is available. If no usefull data is exists the chatbot should just write "no data is available". I have an output parser that displays data in a dataframe.

I got a simple RAG chatbot wich runs as it should in the console.
But when I run it inside a streamlit chat im getting complete different answers.

Why is this happening?

Hi ther, i ran into a problem.

im trying to make a chatbot with llama-index.

I try to use a LLM form HuggingFace (see in the picture)

But when i run the chat_engine.chat(...) im getting following error:
NotImplementedError: Messages passed in must be of odd length. (picture)

Im kind of stuck. Thanks for your help.