The aim of the task is to gain a quick and well-founded overview of the many possible applications of Llama-2. To this end, a question-answering system based on scientific publications is to be developed. The scientific papers available on Arxiv.org on the subject of "Llama" will serve as the primary source of information.
For this purpose, I decided to use a Retrieval Augmented Generation (RAG) with Langchain on colab.
Further Research:
- better preprocessing of the text: e.g. the \n- in the QA are disturbing the readibility
- Conversational Retrieval Chain