Is there an example of using ReActAgent without using OpenAI LLM? I set one up through HuggingFaceLLM but the response time is much longer and unable to get right results in contrast to OpenAI.

I have a basic question because I am not familiar with Python, which Python version should I use for Llama-index?

Is it possible to use local llm's such as llama.cpp for DocumentSummaryIndex？ I keep getting llama_tokenize_with_model: too many tokens
Error

I have a question: instead of getting response from LLM using query_engine, how do I get the retrieved context from the step before sending to LLM for generating response