Dataset QA from government regulatory manuals PT-BR

Dataset of questions and answers in the domain of government regulatory manuals in Brazilian Portuguese. This dataset was created synthetically to evaluate the ability of embedding models and LLM to retrieve and generate factual and relevant answers from government regulatory manuals written in Brazilian Portuguese. The code used to create this dataset can be found at https://github.com/luizsabiano/manuals_question_generator The code for the entire project, including dataset construction and retrieval and generation experiments, can be found at https://github.com/luizsabiano/rag_eval_man_gov