This dataset is subset of neural-bridge/rag-dataset-1200 dataset from which only the datapoints that has context in it are sampled.
Further the content that has max token length of 768 are retained.
The score is predicted using LLM.
The dataset has completeness_score which inddicates how much information does the content has to answer the query and relevancy_score which indicates how relevant is the content to the query.
The label and actual_label are same just used for checking correctness of… See the full description on the dataset page:
https://huggingface.co/datasets/Ellio98/Context-Question-Relevancy-and-Completeness.
