Is it possible to expand the size of semi-structured reports dataset? #210
Description
A lot of RAG based chatbot is based on PDFs like documents, Also how to deal with PDF files is necessary for promoting the performance
of RAG. Thanks in advance for your effort of creating the semi-structured dataset for this senarios. But there is so few of the PDFs and the QA pairs, just 6 docs and 30 QA pairs in total. I see in langchain Simth, the dataset dashboard, most of the experiments are in 100% Faithfulness. Is that possible to expand this benchmark?