Dataset · openlanguagedata
FLORES+ multilingual translation benchmark
openlanguagedata/flores_plus
Parallel texts for machine translation evaluation across numerous languages, including low-resource ones.
What's inside
A benchmark for multilingual machine translation featuring texts in numerous languages, including low-resource ones, in tabular and text formats. Originally released by FAIR (Meta) researchers as FLORES and now maintained by OLDI under the CC BY-SA 4.0 license. The size ranges from 100K to 1M examples.
What you can do with it
- Evaluate multilingual translation models
- Test quality on low-resource languages
- Compare MT systems on a uniform benchmark
- Tasks
- Text generation, Translation
- Languages
- Acehnese, acm, acq, Tunisian Arabic, Afrikaans, apc, Akan, Albanian and 4 more
- Size
- 100K–1M rows
- Downloads this month
- 26.7K
- Likes
- 217
- Access
- on request on Hugging Face
Models trained on it
- Helsinki-NLP/opus-mt_tiny_deu-engTranslation 1.5K
- Helsinki-NLP/opus-mt_tiny_spa-engTranslation 678
- Helsinki-NLP/opus-mt_tiny_eng-turTranslation 524
- serkanpeker/Trendyol-LLM-8B-T1-GGUFText generation 334
- Helsinki-NLP/opus-mt_tiny_eng-spaTranslation 323
- serkanpeker/Turkish-Gemma-9b-T1-GGUFText generation 317
- Helsinki-NLP/opus-mt_tiny_eng-deuTranslation 258
- Helsinki-NLP/opus-mt_tiny_zho-engTranslation 246
How to load it
from datasets import load_dataset
ds = load_dataset("openlanguagedata/flores_plus")