mteb/

FEVER_test_top_250_only_w_correct-v2

FEVERHardNegatives An MTEB dataset Massive Text Embedding Benchmark

FEVER (Fact Extraction and VERification) consists of 185,445 claims generated by altering sentences extracted from Wikipedia and subsequently verified without knowledge of the sentence they were derived from. The hard negative version has been created by pooling the 250 top documents per query from BM25, e5-multilingual-large and e5-mistral-instruct.

Task category t2t

Domains Encyclopaedic, Written… See the full description on the dataset page: https://huggingface.co/datasets/mteb/FEVER_test_top_250_only_w_correct-v2.