Investigate Why Llama 3.1 is giving incorrect response

We should be able to use atleast llama 3, The responses from llama2 seems outdated.

Start by reproducing the incorrect and outdated responses described for Llama 2 and Llama 3.1, then trace the chatbot's model and retrieval flow. Done means the project can use at least Llama 3 and produces responses that are accurate and current, with the expected behavior documented or tested.

Written by the indexing model from the issue text.