Skip to content

Dataset · wmt

WMT14 classical multilingual machine translation corpus

wmt/wmt14

A large parallel text corpus in six languages, including Russian, for training translation systems.

What's inside

The set contains between 10M and 100M text examples in English, Russian, German, French, Czech, and Hindi, collected for the WMT 2014 competition. The card includes an important warning: some Common Crawl data contains unaligned sentences and English insertions in foreign-language files.

What you can do with it

  • Training and comparing machine translation models (including for the EN-RU language pair)
  • Developing cleaning and filtering algorithms for noisy parallel corpora
Tasks
Translation
Languages
Czech, German, English, French, Hindi, Russian
Size
10M–100M rows
Downloads this month
9.8K
Likes
39

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("wmt/wmt14")