Skip to content

Dataset · mteb

Biomedical semantic textual similarity benchmark

mteb/biosses-sts

English medical text pairs for evaluating embedding quality in MTEB.

What's inside

The BIOSSES dataset within the Massive Text Embedding Benchmark (MTEB), designed to evaluate sentence-similarity in the medical domain. It contains under 1,000 English examples and is ready for direct model evaluation via the mteb library.

What you can do with it

  • Evaluate text embedding quality in the biomedical domain using the MTEB framework
  • Compare fine-tuned medical retrievers and encoders on similarity search tasks
Tasks
Sentence similarity
Languages
English
Size
up to 1K rows
Downloads this month
21.5K
Likes
2

How to load it

from datasets import load_dataset

ds = load_dataset("mteb/biosses-sts")