This dataset extends the original PAWS dataset by providing aligned Dhivehi translations for each English sentence pair. It is designed to evaluate and train models on paraphrase identification and semantic similarity tasks across English (en) and Dhivehi (dv).

	Dataset Summary

Split Examples Size (bytes)

Train 49,401 38,433,892

Validation 8,000 6,215,265

Test 8,000 6,187,893… See the full description on the dataset page: https://huggingface.co/datasets/alakxender/dhivehi-paws-labeled.