Dataset · mteb
STS14 semantic textual similarity benchmark
mteb/sts14-sts
The classic SemEval STS 2014 English dataset within MTEB for evaluating text embedding quality.
What's inside
The dataset contains English text pairs from blogs, web pages, and conversational speech for semantic similarity evaluation. The size ranges from 1,000 to 10,000 entries. The dataset is adapted for the MTEB library for standardized model evaluation.
What you can do with it
- Evaluating text embedding model quality on sentence similarity tasks
- Comparing encoders within a unified MTEB benchmark
- Tasks
- Sentence similarity
- Languages
- English
- Size
- 1K–10K rows
- Downloads this month
- 19.9K
- Likes
- 2
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("mteb/sts14-sts")