This release packages the Quora Duplicate Questions slice of the BEIR (Benchmarking IR) benchmark as a flat, table-oriented dataset for retrieval evaluation and tooling. The underlying content comes from Quora’s public Question Pairs resource: pairs of user questions labeled as duplicate or not. BEIR reformulates the problem as duplicate question retrieval: each query is one question text, and systems must rank a large corpus of… See the full description on the dataset page: https://huggingface.co/datasets/orgrctera/beir_quora.
