answering is not a problem. llama2-70b works quite well. extracting the right context is the bigger issue 🙂 once the context is correct, answers are automatically correct even with smaller LLMs 🙂

→ View original post on X — @abhi1thakur,