I think I asked this before, but is

I think I asked this before, but is there a way to designate one model to handle function calling for an agent, and another handle the generation
for example:

Plain Text

        query_engine = get_query_engine(
        user_id=user_id, vector_store=vector_store, model=model, filters=filters
    )
    query_engine_tool = QueryEngineTool.from_defaults(query_engine=query_engine)

    model_instance = LLM_INSTANCES[model]
    return ReActAgent.from_tools(
        tools=[query_engine_tool],
        llm=model_instance,
        chat_history=history,
    )

I would love to pass gpt4 in to handle all the function calling and decision making, but allow another model to do the final generation in someway

Find answers from the community

I think I asked this before, but is