| { | |
| "add_prefix_space": true, | |
| "added_tokens_decoder": { | |
| "0": { | |
| "content": "<pad>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "1": { | |
| "content": "<unk>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "2": { | |
| "content": "<s>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "3": { | |
| "content": "</s>", | |
| "lstrip": false, | |
| "normalized": false, | |
| "rstrip": false, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256001": { | |
| "content": "__ace__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256002": { | |
| "content": "__ace_Latn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256003": { | |
| "content": "__acm__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256004": { | |
| "content": "__acq__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256005": { | |
| "content": "__aeb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256006": { | |
| "content": "__afr__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256007": { | |
| "content": "__ajp__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256008": { | |
| "content": "__aka__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256009": { | |
| "content": "__amh__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256010": { | |
| "content": "__apc__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256011": { | |
| "content": "__arb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256012": { | |
| "content": "__ars__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256013": { | |
| "content": "__ary__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256014": { | |
| "content": "__arz__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256015": { | |
| "content": "__asm__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256016": { | |
| "content": "__ast__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256017": { | |
| "content": "__awa__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256018": { | |
| "content": "__ayr__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256019": { | |
| "content": "__azb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256020": { | |
| "content": "__azj__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256021": { | |
| "content": "__bak__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256022": { | |
| "content": "__bam__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256023": { | |
| "content": "__ban__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256024": { | |
| "content": "__bel__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256025": { | |
| "content": "__bem__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256026": { | |
| "content": "__ben__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256027": { | |
| "content": "__bho__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256028": { | |
| "content": "__bjn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256029": { | |
| "content": "__bjn_Latn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256030": { | |
| "content": "__bod__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256031": { | |
| "content": "__bos__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256032": { | |
| "content": "__bug__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256033": { | |
| "content": "__bul__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256034": { | |
| "content": "__cat__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256035": { | |
| "content": "__ceb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256036": { | |
| "content": "__ces__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256037": { | |
| "content": "__cjk__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256038": { | |
| "content": "__ckb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256039": { | |
| "content": "__crh__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256040": { | |
| "content": "__cym__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256041": { | |
| "content": "__dan__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256042": { | |
| "content": "__deu__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256043": { | |
| "content": "__dik__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256044": { | |
| "content": "__dyu__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256045": { | |
| "content": "__dzo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256046": { | |
| "content": "__ell__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256047": { | |
| "content": "__eng__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256048": { | |
| "content": "__epo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256049": { | |
| "content": "__est__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256050": { | |
| "content": "__eus__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256051": { | |
| "content": "__ewe__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256052": { | |
| "content": "__fao__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256053": { | |
| "content": "__pes__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256054": { | |
| "content": "__fij__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256055": { | |
| "content": "__fin__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256056": { | |
| "content": "__fon__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256057": { | |
| "content": "__fra__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256058": { | |
| "content": "__fur__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256059": { | |
| "content": "__fuv__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256060": { | |
| "content": "__gla__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256061": { | |
| "content": "__gle__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256062": { | |
| "content": "__glg__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256063": { | |
| "content": "__grn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256064": { | |
| "content": "__guj__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256065": { | |
| "content": "__hat__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256066": { | |
| "content": "__hau__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256067": { | |
| "content": "__heb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256068": { | |
| "content": "__hin__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256069": { | |
| "content": "__hne__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256070": { | |
| "content": "__hrv__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256071": { | |
| "content": "__hun__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256072": { | |
| "content": "__hye__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256073": { | |
| "content": "__ibo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256074": { | |
| "content": "__ilo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256075": { | |
| "content": "__ind__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256076": { | |
| "content": "__isl__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256077": { | |
| "content": "__ita__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256078": { | |
| "content": "__jav__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256079": { | |
| "content": "__jpn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256080": { | |
| "content": "__kab__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256081": { | |
| "content": "__kac__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256082": { | |
| "content": "__kam__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256083": { | |
| "content": "__kan__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256084": { | |
| "content": "__kas__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256085": { | |
| "content": "__kas_Deva__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256086": { | |
| "content": "__kat__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256087": { | |
| "content": "__knc__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256088": { | |
| "content": "__knc_Latn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256089": { | |
| "content": "__kaz__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256090": { | |
| "content": "__kbp__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256091": { | |
| "content": "__kea__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256092": { | |
| "content": "__khm__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256093": { | |
| "content": "__kik__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256094": { | |
| "content": "__kin__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256095": { | |
| "content": "__kir__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256096": { | |
| "content": "__kmb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256097": { | |
| "content": "__kon__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256098": { | |
| "content": "__kor__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256099": { | |
| "content": "__kmr__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256100": { | |
| "content": "__lao__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256101": { | |
| "content": "__lvs__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256102": { | |
| "content": "__lij__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256103": { | |
| "content": "__lim__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256104": { | |
| "content": "__lin__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256105": { | |
| "content": "__lit__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256106": { | |
| "content": "__lmo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256107": { | |
| "content": "__ltg__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256108": { | |
| "content": "__ltz__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256109": { | |
| "content": "__lua__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256110": { | |
| "content": "__lug__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256111": { | |
| "content": "__luo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256112": { | |
| "content": "__lus__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256113": { | |
| "content": "__mag__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256114": { | |
| "content": "__mai__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256115": { | |
| "content": "__mal__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256116": { | |
| "content": "__mar__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256117": { | |
| "content": "__min__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256118": { | |
| "content": "__mkd__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256119": { | |
| "content": "__plt__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256120": { | |
| "content": "__mlt__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256121": { | |
| "content": "__mni__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256122": { | |
| "content": "__khk__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256123": { | |
| "content": "__mos__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256124": { | |
| "content": "__mri__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256125": { | |
| "content": "__zsm__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256126": { | |
| "content": "__mya__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256127": { | |
| "content": "__nld__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256128": { | |
| "content": "__nno__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256129": { | |
| "content": "__nob__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256130": { | |
| "content": "__npi__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256131": { | |
| "content": "__nso__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256132": { | |
| "content": "__nus__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256133": { | |
| "content": "__nya__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256134": { | |
| "content": "__oci__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256135": { | |
| "content": "__gaz__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256136": { | |
| "content": "__ory__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256137": { | |
| "content": "__pag__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256138": { | |
| "content": "__pan__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256139": { | |
| "content": "__pap__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256140": { | |
| "content": "__pol__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256141": { | |
| "content": "__por__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256142": { | |
| "content": "__prs__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256143": { | |
| "content": "__pbt__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256144": { | |
| "content": "__quy__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256145": { | |
| "content": "__ron__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256146": { | |
| "content": "__run__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256147": { | |
| "content": "__rus__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256148": { | |
| "content": "__sag__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256149": { | |
| "content": "__san__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256150": { | |
| "content": "__sat__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256151": { | |
| "content": "__scn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256152": { | |
| "content": "__shn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256153": { | |
| "content": "__sin__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256154": { | |
| "content": "__slk__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256155": { | |
| "content": "__slv__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256156": { | |
| "content": "__smo__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256157": { | |
| "content": "__sna__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256158": { | |
| "content": "__snd__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256159": { | |
| "content": "__som__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256160": { | |
| "content": "__sot__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256161": { | |
| "content": "__spa__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256162": { | |
| "content": "__als__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256163": { | |
| "content": "__srd__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256164": { | |
| "content": "__srp__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256165": { | |
| "content": "__ssw__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256166": { | |
| "content": "__sun__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256167": { | |
| "content": "__swe__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256168": { | |
| "content": "__swh__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256169": { | |
| "content": "__szl__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256170": { | |
| "content": "__tam__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256171": { | |
| "content": "__tat__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256172": { | |
| "content": "__tel__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256173": { | |
| "content": "__tgk__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256174": { | |
| "content": "__tgl__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256175": { | |
| "content": "__tha__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256176": { | |
| "content": "__tir__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256177": { | |
| "content": "__taq__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256178": { | |
| "content": "__taq_Tfng__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256179": { | |
| "content": "__tpi__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256180": { | |
| "content": "__tsn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256181": { | |
| "content": "__tso__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256182": { | |
| "content": "__tuk__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256183": { | |
| "content": "__tum__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256184": { | |
| "content": "__tur__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256185": { | |
| "content": "__twi__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256186": { | |
| "content": "__tzm__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256187": { | |
| "content": "__uig__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256188": { | |
| "content": "__ukr__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256189": { | |
| "content": "__umb__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256190": { | |
| "content": "__urd__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256191": { | |
| "content": "__uzn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256192": { | |
| "content": "__vec__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256193": { | |
| "content": "__vie__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256194": { | |
| "content": "__war__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256195": { | |
| "content": "__wol__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256196": { | |
| "content": "__xho__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256197": { | |
| "content": "__ydd__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256198": { | |
| "content": "__yor__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256199": { | |
| "content": "__yue__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256200": { | |
| "content": "__cmn__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256201": { | |
| "content": "__cmn_Hant__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| }, | |
| "256202": { | |
| "content": "__zul__", | |
| "lstrip": true, | |
| "normalized": false, | |
| "rstrip": true, | |
| "single_word": false, | |
| "special": true | |
| } | |
| }, | |
| "additional_special_tokens": [ | |
| "<pad>", | |
| "<unk>", | |
| "<s>", | |
| "</s>", | |
| "__ace__", | |
| "__ace_Latn__", | |
| "__acm__", | |
| "__acq__", | |
| "__aeb__", | |
| "__afr__", | |
| "__ajp__", | |
| "__aka__", | |
| "__amh__", | |
| "__apc__", | |
| "__arb__", | |
| "__ars__", | |
| "__ary__", | |
| "__arz__", | |
| "__asm__", | |
| "__ast__", | |
| "__awa__", | |
| "__ayr__", | |
| "__azb__", | |
| "__azj__", | |
| "__bak__", | |
| "__bam__", | |
| "__ban__", | |
| "__bel__", | |
| "__bem__", | |
| "__ben__", | |
| "__bho__", | |
| "__bjn__", | |
| "__bjn_Latn__", | |
| "__bod__", | |
| "__bos__", | |
| "__bug__", | |
| "__bul__", | |
| "__cat__", | |
| "__ceb__", | |
| "__ces__", | |
| "__cjk__", | |
| "__ckb__", | |
| "__crh__", | |
| "__cym__", | |
| "__dan__", | |
| "__deu__", | |
| "__dik__", | |
| "__dyu__", | |
| "__dzo__", | |
| "__ell__", | |
| "__eng__", | |
| "__epo__", | |
| "__est__", | |
| "__eus__", | |
| "__ewe__", | |
| "__fao__", | |
| "__pes__", | |
| "__fij__", | |
| "__fin__", | |
| "__fon__", | |
| "__fra__", | |
| "__fur__", | |
| "__fuv__", | |
| "__gla__", | |
| "__gle__", | |
| "__glg__", | |
| "__grn__", | |
| "__guj__", | |
| "__hat__", | |
| "__hau__", | |
| "__heb__", | |
| "__hin__", | |
| "__hne__", | |
| "__hrv__", | |
| "__hun__", | |
| "__hye__", | |
| "__ibo__", | |
| "__ilo__", | |
| "__ind__", | |
| "__isl__", | |
| "__ita__", | |
| "__jav__", | |
| "__jpn__", | |
| "__kab__", | |
| "__kac__", | |
| "__kam__", | |
| "__kan__", | |
| "__kas__", | |
| "__kas_Deva__", | |
| "__kat__", | |
| "__knc__", | |
| "__knc_Latn__", | |
| "__kaz__", | |
| "__kbp__", | |
| "__kea__", | |
| "__khm__", | |
| "__kik__", | |
| "__kin__", | |
| "__kir__", | |
| "__kmb__", | |
| "__kon__", | |
| "__kor__", | |
| "__kmr__", | |
| "__lao__", | |
| "__lvs__", | |
| "__lij__", | |
| "__lim__", | |
| "__lin__", | |
| "__lit__", | |
| "__lmo__", | |
| "__ltg__", | |
| "__ltz__", | |
| "__lua__", | |
| "__lug__", | |
| "__luo__", | |
| "__lus__", | |
| "__mag__", | |
| "__mai__", | |
| "__mal__", | |
| "__mar__", | |
| "__min__", | |
| "__mkd__", | |
| "__plt__", | |
| "__mlt__", | |
| "__mni__", | |
| "__khk__", | |
| "__mos__", | |
| "__mri__", | |
| "__zsm__", | |
| "__mya__", | |
| "__nld__", | |
| "__nno__", | |
| "__nob__", | |
| "__npi__", | |
| "__nso__", | |
| "__nus__", | |
| "__nya__", | |
| "__oci__", | |
| "__gaz__", | |
| "__ory__", | |
| "__pag__", | |
| "__pan__", | |
| "__pap__", | |
| "__pol__", | |
| "__por__", | |
| "__prs__", | |
| "__pbt__", | |
| "__quy__", | |
| "__ron__", | |
| "__run__", | |
| "__rus__", | |
| "__sag__", | |
| "__san__", | |
| "__sat__", | |
| "__scn__", | |
| "__shn__", | |
| "__sin__", | |
| "__slk__", | |
| "__slv__", | |
| "__smo__", | |
| "__sna__", | |
| "__snd__", | |
| "__som__", | |
| "__sot__", | |
| "__spa__", | |
| "__als__", | |
| "__srd__", | |
| "__srp__", | |
| "__ssw__", | |
| "__sun__", | |
| "__swe__", | |
| "__swh__", | |
| "__szl__", | |
| "__tam__", | |
| "__tat__", | |
| "__tel__", | |
| "__tgk__", | |
| "__tgl__", | |
| "__tha__", | |
| "__tir__", | |
| "__taq__", | |
| "__taq_Tfng__", | |
| "__tpi__", | |
| "__tsn__", | |
| "__tso__", | |
| "__tuk__", | |
| "__tum__", | |
| "__tur__", | |
| "__twi__", | |
| "__tzm__", | |
| "__uig__", | |
| "__ukr__", | |
| "__umb__", | |
| "__urd__", | |
| "__uzn__", | |
| "__vec__", | |
| "__vie__", | |
| "__war__", | |
| "__wol__", | |
| "__xho__", | |
| "__ydd__", | |
| "__yor__", | |
| "__yue__", | |
| "__cmn__", | |
| "__cmn_Hant__", | |
| "__zul__" | |
| ], | |
| "bos_token": "<s>", | |
| "clean_up_tokenization_spaces": true, | |
| "cls_token": "<s>", | |
| "eos_token": "</s>", | |
| "max_length": 128, | |
| "model_max_length": 1000000000000000019884624838656, | |
| "pad_to_multiple_of": 2, | |
| "pad_token": "<pad>", | |
| "pad_token_type_id": 0, | |
| "padding_side": "right", | |
| "processor_class": "SeamlessM4TProcessor", | |
| "sep_token": "</s>", | |
| "sp_model_kwargs": {}, | |
| "src_lang": "__eng__", | |
| "stride": 0, | |
| "tgt_lang": "__fra__", | |
| "tokenizer_class": "SeamlessM4TTokenizer", | |
| "truncation_side": "right", | |
| "truncation_strategy": "longest_first", | |
| "unk_token": "<unk>" | |
| } | |