This dataset is created to evaluate large language models and GraphRAG systems operating on the Turkish Penal Code (TCK) texts. It is used to test the systems' performance in accurately retrieving legal context and generating correct answers while strictly adhering to the provided legal text.
soru: The legal question text to be provided as input to the system.
cevap: The verified legal response that the system is expected to… See the full description on the dataset page:
https://huggingface.co/datasets/hzfokyn/turk-ceza-kanunu-evaluation-dataset.
