hello guys i wanted to use the recursive

I think changing the llm and embed model is onyl required to get away from OpenAI and using something else.

You can take inspiration from these notebook: https://docs.llamaindex.ai/en/stable/examples/vector_stores/SimpleIndexDemoLlama-Local.html

hello thank you for the reply, but it seems from that multidocument agents wont work with llama2 it seems to not be retrieving from the indexes and i wanted multi doc recursive retrival with agents, if you could provide something related to that then i would be really grateful

also i have another question, does openai agent work in conjuction with any other model? in my code im using it to retrieve from the indexes but my final response never seems to retrieve from the indexes, the llm says to the question what is a " reaction catalog?",
ps this was working when i was using gpt 3.5 turbo

Yeah llama2 is not good with agents, infact it is not good many things: Checkout this compatibility report for open source LLM: https://docs.llamaindex.ai/en/stable/module_guides/models/llms.html#open-source-llms

You could try with Zephyr or Starling, but they might not eb as good as OpenAI

@WhiteFang_Jr what about mixtral, can it hold against openAI ?

Havent seen any agent implementation with Mixtral but I have heard it is good in text generation plus following instruction. You could give it a shot!

Alright, otherwise what’s the best working open source LLM, that’s bigger than 7B/13B ?

Alright thank you man!