README.md

metadata

dataset_info:
  features:
    - name: english
      dtype: string
    - name: kurdish
      dtype: string
  splits:
    - name: train
      num_bytes: 49594900
      num_examples: 148844
  download_size: 25408908
  dataset_size: 49594900
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
task_categories:
  - translation
language:
  - ku
  - en
pretty_name: Kurdish - English Sentences
size_categories:
  - 100K<n<1M

Summary

Extracted subset from Helsinki-NLP/opus-100 and reshaped it into 2 columns. Note: noticed some low quality pairs. It would be a good project to classify and select high quality pairs.