Semantic Textual Similarity Benchmark (STSbenchmark) dataset. This version removes duplicate sentence pairs from the validation and test splits when the same order-insensitive pair appears more than once in an evaluation split or appears in another split.

Task category STS (text-to-text)

Domains Blog, News, Written

Reference Machine translated multilingual STS benchmark dataset.

Source datasets:… See the full description on the dataset page: https://huggingface.co/datasets/mteb/STSBenchmarkv2.