config.yaml

corpora:
    ilci:
        splits: ['train', 'test', 'dev']
        langs: [ 'en', 'hi', 'ta', 'te', 'ml', 'ur', 'bn', 'gu', 'mr', 'pa']
        #langs: ['en', 'hi','ta', 'te']
    #wat-ilmpc:
    #     splits: ['dev']
    #     langs: ['en', 'ta']
    #iitb-hi-en:
    #     splits: ['train', 'dev', 'test']
    #     langs: ['en', 'hi']
    #wat-ilmpc:
    #     splits: ['train', 'dev', 'test']
    #     langs: [ 'en', 'hi', 'ta', 'te', 'ml', 'ur', 'bn']
         #langs: ['en', 'hi',  'ta', 'te']
    #national-newscrawl:
    #     splits: ['train']
    #     langs: ['en', 'hi']
    #ufal-en-tam:
    #    splits: ['train', 'test', 'dev']
    #    langs: ['en', 'ta']
    #bible-en-te:
    #    splits: ['train', 'test', 'dev']
    #    langs: ['en', 'te']
    #mann-ki-baat-test:
    #    splits: ['test']
    #    langs: ['en', 'hi',  'ta', 'te']
    #eenadu-en-te:
    #    splits: ['train']
    #    langs: ['en','te']
    odiencorp:
        splits: ['train', 'test', 'dev']
        langs: ['en','or']

dictionary:
    src: '/home/shashanks/ilmulti/ilmulti/sentencepiece/central.dict'
    tgt: '/home/shashanks/ilmulti/ilmulti/sentencepiece/central.dict'