Browse Datasets

FLORES-200

Professional Translations for 200 Languages

translation-evaluation embeddings alignment 3,001 sentences per language CC-BY-SA 4.0
Access Dataset Read Paper

Description

A many-to-many evaluation dataset covering 200 languages with professional translations. Gold standard for evaluating embeddings and alignment quality.

Quick Start

Python 
from datasets import load_dataset

dataset = load_dataset("facebook/flores", "all")

Project Ideas

Embedding Evaluation

Evaluate cosine similarity between English and Dinka sentence embeddings

Intermediate

Alignment Fine-tuning

Fine-tune a multilingual model to align Kanuri and English

Advanced

Ethical Considerations

  • High quality but limited domain (Wikipedia-based)