MinaGabriel/sentence-relevance-extractor

Sentence Relevance Extractor (SRE) Sentence Relevance Extractor (SRE) is a large-scale dataset for binary evidence selection in multi-document, multi-hop question answering. The goal: Given a question and a sentence from the context, predict whether this sentence is relevant evidence ("Yes") or irrelevant ("No"). This dataset is suitable for training: Sentence-level RAG rerankers Binary relevance classifiers Optimization-based truth discovery systems Multi-hop QA evidence… See the full description on the dataset page: https://huggingface.co/datasets/MinaGabriel/sentence-relevance-extractor.