README.md

metadata

configs:
  - config_name: non-paired_data
    data_files:
      - split: train
        path: non-paired/*_train.jsonl
      - split: test
        path: non-paired/*_test.jsonl
      - split: validation
        path: non-paired/*_valid.jsonl
  - config_name: paired_data
    data_files:
      - split: train
        path: paired/*_train_paired.jsonl
      - split: test
        path: paired/*_test_paired.jsonl
      - split: validation
        path: paired/*_valid_paired.jsonl
    default: true

Test paired sample here

Train paired sample here