Skip to content

Dataset · Helsinki-NLP

Parallel News Commentary Corpus

Helsinki-NLP/news_commentary

Texts from the OPUS News-Commentary collection in six languages for training and evaluating machine translation models.

What's inside

Inside are parallel texts for machine translation from the OPUS News-Commentary set. The dataset contains 1 to 10 million records in English, German, French, Spanish, Czech, and Arabic.

What you can do with it

  • Training and fine-tuning machine translation models (seq2seq, LLM)
  • Evaluating translation quality between European languages and Arabic
Tasks
Translation
Languages
Arabic, Czech, German, English, Spanish, French, Italian, Japanese and 4 more
Size
1M–10M rows
Downloads this month
5.8K
Likes
39

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("Helsinki-NLP/news_commentary")