README.md

metadata

task_categories:
  - translation
  - text2text-generation
language:
  - hi
  - or
  - bn
  - kn
  - te
  - ta
  - ml
  - gu
  - pa
  - sa
  - mr
  - ne
  - bh
pretty_name: IAST<->Indic Seq2Seq Corpus for Indic languages
size_categories:
  - 1M<n<10M
configs:
  - config_name: sanskrit
    data_files:
      - split: train
        path: files/sanskrit/sanskrit_wikidata_iast.csv
  - config_name: odia
    data_files:
      - split: train
        path: files/odia/odia_wikidata_iast.csv
  - config_name: hindi
    data_files:
      - split: train
        path: files/hindi/hindi_wikidata_iast.csv
  - config_name: bengali
    data_files:
      - split: train
        path: files/bengali/bengali_wikidata_iast.csv
  - config_name: tamil
    data_files:
      - split: train
        path: files/tamil/tamil_wikidata_iast.csv
  - config_name: telugu
    data_files:
      - split: train
        path: files/telugu/telugu_wikidata_iast.csv
  - config_name: kannada
    data_files:
      - split: train
        path: files/kannada/kannada_wikidata_iast.csv
  - config_name: malayalam
    data_files:
      - split: train
        path: files/malayalam/malayalam_wikidata_iast.csv
  - config_name: punjabi
    data_files:
      - split: train
        path: files/punjabi/punjabi_wikidata_iast.csv
  - config_name: gujarati
    data_files:
      - split: train
        path: files/gujarati/gujarati_wikidata_iast.csv
  - config_name: marathi
    data_files:
      - split: train
        path: files/marathi/marathi_wikidata_iast.csv
  - config_name: nepali
    data_files:
      - split: train
        path: files/nepali/nepali_wikidata_iast.csv
  - config_name: bhojpuri
    data_files:
      - split: train
        path: files/bhojpuri/bhojpuri_wikidata_iast.csv
  - config_name: maithili
    data_files:
      - split: train
        path: files/maithili/maithili_wikidata_iast.csv
license: cc-by-sa-3.0

Dataset Details

• Dataset created by transliterating existing datasets to IAST by means of IAST transliteration library
• Languages include Sanskrit, Hindi, Odia, Bengali, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi, Marathi, Bhojpuri, Nepali.
• Pre-existing dataset source(s) - Wikipedia.
• Across all subsets, 'source', 'target', 'source_lang', 'target_lang', 'source' columns are common.
• This is just a hobby dataset, but it should abide by the licenses of the input dataset(s).