metadata
task_categories:
- translation
- text2text-generation
language:
- hi
- or
- bn
- kn
- te
- ta
- ml
- gu
- pa
pretty_name: IAST<->Indic Seq2Seq Corpus for Indic languages
size_categories:
- 100K<n<1M
configs:
- config_name: wikipedia
data_files:
- split: train
path:
- files/odia/odia_wikidata_iast.csv
- files/hindi/hindi_wikidata_iast.csv
- files/bengali/bengali_wikidata_iast.csv
- files/tamil/tamil_wikidata_iast.csv
- files/telugu/telugu_wikidata_iast.csv
- files/kannada/kannada_wikidata_iast.csv
- files/malayalam/malayalam_wikidata_iast.csv
- files/punjabi/punjabi_wikidata_iast.csv
- files/gujarati/gujarati_wikidata_iast.csv
Dataset Details
- Dataset created by transliterating existing datasets to IAST by means of IAST transliteration library
- Languages include Hindi, Odia, Bengali, Tamil, Telugu, Kannada, Malayalam, Gujarati, Punjabi.
- 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).