Skip to content

Dataset · PhilipMay

Multilingual STS Benchmark (STSb Multi MT)

PhilipMay/stsb_multi_mt

English STS Benchmark and its translations into five languages for training sentence embeddings.

What's inside

The dataset is based on the STS Benchmark, a collection of English datasets for Semantic Textual Similarity tasks from SemEval 2012–2017. It includes texts from image captions, news headlines, and user forums. Translations into German, Spanish, French, Italian, and Dutch were made via deepl.com; the size is between 10k and 100k examples.

What you can do with it

  • Training multilingual sentence transformers
  • Evaluating sentence embedding quality on STS
  • Cross-lingual validation of text similarity models
Tasks
Text classification
Languages
German, English, Spanish, French, Italian, Dutch, Polish, Portuguese and 2 more
Size
10K–100K rows
Downloads this month
2K
Likes
69

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("PhilipMay/stsb_multi_mt")