# Mapping the BCP-47 codes used in the MADLAD-400 models to the language names.
# Restricted to languages evaluated in Table 14, Section A.9 of:
# [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/pdf/2309.04662)
# Evaluation sets: WMT, Flores-200, NTREX, Gatones
langid_to_language = {
    # Page 16
    "<2en>": "English",
    "<2ru>": "Russian",
    "<2es>": "Spanish",
    "<2fr>": "French",
    "<2de>": "German",
    "<2it>": "Italian",
    "<2pt>": "Portuguese",
    "<2pl>": "Polish",
    "<2nl>": "Dutch",
    "<2vi>": "Vietnamese",
    "<2tr>": "Turkish",
    "<2sv>": "Swedish",
    "<2id>": "Indonesian",
    "<2ro>": "Romanian",
    "<2cs>": "Czech",
    "<2zh>": "Mandarin Chinese",
    "<2hu>": "Hungarian",
    "<2ja>": "Japanese",
    "<2th>": "Thai",
    "<2fi>": "Finnish",
    "<2fa>": "Persian",
    "<2uk>": "Ukrainian",
    "<2da>": "Danish",
    "<2el>": "Greek",
    "<2no>": "Norwegian",
    "<2bg>": "Bulgarian",
    "<2sk>": "Slovak",
    "<2ko>": "Korean",
    "<2ar>": "Arabic",
    "<2lt>": "Lithuanian",
    "<2ca>": "Catalan",
    "<2sl>": "Slovenian",
    "<2he>": "Hebrew",
    "<2et>": "Estonian",
    "<2lv>": "Latvian",
    "<2hi>": "Hindi",
    "<2sq>": "Albanian",
    "<2ms>": "Malay",
    "<2az>": "Azerbaijani",
    "<2sr>": "Serbian",
    "<2ta>": "Tamil",
    "<2hr>": "Croatian",
    "<2kk>": "Kazakh",
    "<2is>": "Icelandic",
    "<2ml>": "Malayalam",
    "<2mr>": "Marathi",
    "<2te>": "Telugu",
    "<2af>": "Afrikaans",
    "<2gl>": "Galician",
    "<2fil>": "Filipino",
    "<2be>": "Belarusian",
    # Page 17
    "<2mk>": "Macedonian",
    "<2eu>": "Basque",
    "<2bn>": "Bengali",
    "<2ka>": "Georgian",
    "<2mn>": "Mongolian",
    "<2bs>": "Bosnian",
    "<2uz>": "Uzbek",
    "<2ur>": "Urdu",
    "<2sw>": "Swahili",
    "<2yue>": "Cantonese",
    "<2ne>": "Nepali",
    "<2kn>": "Kannada",
    "<2gu>": "Gujarati",
    "<2si>": "Sinhala",
    "<2cy>": "Welsh",
    "<2eo>": "Esperanto",
    "<2hy>": "Armenian",
    "<2ky>": "Kyrghyz",
    "<2tg>": "Tajik",
    "<2ga>": "Irish",
    "<2mt>": "Maltese",
    "<2my>": "Myanmar (Burmese)",
    "<2km>": "Khmer",
    "<2tt>": "Tatar",
    "<2so>": "Somali",
    "<2ku>": "Kurdish (Kurmanji)",
    "<2ps>": "Pashto",
    "<2pa>": "Punjabi",
    "<2rw>": "Kinyarwanda",
    "<2lo>": "Lao",
    "<2ha>": "Hausa",
    "<2dv>": "Dhivehi",
    "<2lb>": "Luxembourgish",
    "<2ckb>": "Kurdish (Sorani)",
    "<2mg>": "Malagasy",
    "<2gd>": "Scottish Gaelic",
    "<2am>": "Amharic",
    "<2ug>": "Uyghur",
    "<2ht>": "Haitian Creole",
    "<2hmn>": "Hmong",
    "<2sd>": "Sindhi",
    "<2jv>": "Javanese",
    "<2mi>": "Maori",
    "<2tk>": "Turkmen",
    "<2ceb>": "Cebuano",
    "<2yi>": "Yiddish",
    "<2ba>": "Bashkir",
    "<2fo>": "Faroese",
    "<2or>": "Odia (Oriya)",
    "<2xh>": "Xhosa",
    "<2su>": "Sundanese",
    "<2kl>": "Kalaallisut",
    "<2ny>": "Chichewa",
    "<2sm>": "Samoan",
    "<2sn>": "Shona",
    "<2zu>": "Zulu",
    "<2ig>": "Igbo",
    "<2yo>": "Yoruba",
    "<2pap>": "Papiamento",
    "<2st>": "Sesotho",
    "<2as>": "Assamese",
    "<2oc>": "Occitan",
    "<2cv>": "Chuvash",
    "<2lus>": "Mizo",
    "<2sa>": "Sanskrit",
    "<2bo>": "Tibetan",
    "<2om>": "Oromo",
    "<2ce>": "Chechen",
    # Page 18
    "<2ilo>": "Ilocano",
    "<2lg>": "Luganda",
    "<2ti>": "Tigrinya",
    "<2vec>": "Venetian",
    "<2ts>": "Tsonga",
    "<2ee>": "Ewe",
    "<2av>": "Avar",
    "<2to>": "Tonga (Tonga Islands)",
    "<2tn>": "Tswana",
    "<2nso>": "Sepedi",
    "<2fj>": "Fijian",
    "<2zza>": "Zaza",
    "<2ak>": "Twi",
    "<2dz>": "Dzongkha",
    "<2ln>": "Lingala",
    "<2gn>": "Guarani",
    "<2yua>": "Yucateco",
    "<2war>": "Waray (Philippines)",
    "<2sg>": "Sango",
    "<2ady>": "Adyghe",
    "<2kbp>": "Kabiyè",
    "<2ltg>": "Latgalian",
    "<2iso>": "Isoko",
    "<2kac>": "Kachin",
    "<2bho>": "Bhojpuri",
    "<2ay>": "Aymara",
    "<2qu>": "Quechua",
    "<2pag>": "Pangasinan",
    "<2ve>": "Venda",
    "<2bbc>": "Batak Toba",
    "<2tiv>": "Tiv",
    "<2gom>": "Goan Konkani",
    "<2min>": "Minangkabau",
    "<2bci>": "Baoulé",
    "<2bew>": "Betawi",
    "<2fon>": "Fon",
    # Page 19
    "<2meo>": "Kedah Malay",
    "<2crh>": "Crimean Tatar",
    "<2ace>": "Achinese",
    "<2mai>": "Maithili",
    "<2quc>": "K\u2019iche\u2019",
    "<2shn>": "Shan",
    "<2ss>": "Swati",
    "<2kri>": "Krio",
    "<2kmb>": "Kimbundu",
    "<2wo>": "Wolof",
    "<2ban>": "Balinese",
    "<2bm>": "Bambara",
    # Page 20
    "<2din>": "Dinka",
    "<2mad>": "Madurese",
    "<2mag>": "Magahi",
    "<2kg>": "Kongo",
    "<2dyu>": "Dyula",
    "<2mni>": "Meiteilon (Manipuri)",
    "<2skr>": "Saraiki",
    "<2rn>": "Rundi",
    "<2chr>": "Cherokee",
    "<2hne>": "Chhattisgarhi",
    # Page 21
    "<2awa>": "Awadhi",
    "<2quy>": "Ayacucho Quechua",
    "<2mey>": "Hassaniyya",
    "<2ks>": "Kashimiri",
    "<2ff>": "Fulfulde",
    "<2ber>": "Tamazight (Tfng)",
}