|
{ |
|
"</s>": 3, |
|
"<MINED_DATA>": 256099, |
|
"<MMT_BT_DATA>": 256100, |
|
"<SMT_BT_DATA>": 256101, |
|
"<pad>": 0, |
|
"<s>": 2, |
|
"<unk>": 1, |
|
"__afr__": 256001, |
|
"__amh__": 256002, |
|
"__arb__": 256003, |
|
"__ary__": 256004, |
|
"__arz__": 256005, |
|
"__asm__": 256006, |
|
"__azj__": 256007, |
|
"__bel__": 256008, |
|
"__ben__": 256009, |
|
"__bos__": 256010, |
|
"__bul__": 256011, |
|
"__cat__": 256012, |
|
"__ceb__": 256013, |
|
"__ces__": 256014, |
|
"__ckb__": 256015, |
|
"__cmn_Hant__": 256017, |
|
"__cmn__": 256016, |
|
"__cym__": 256018, |
|
"__dan__": 256019, |
|
"__deu__": 256020, |
|
"__ell__": 256021, |
|
"__eng__": 256022, |
|
"__est__": 256023, |
|
"__eus__": 256024, |
|
"__fin__": 256025, |
|
"__fra__": 256026, |
|
"__fuv__": 256027, |
|
"__gaz__": 256028, |
|
"__gle__": 256029, |
|
"__glg__": 256030, |
|
"__guj__": 256031, |
|
"__heb__": 256032, |
|
"__hin__": 256033, |
|
"__hrv__": 256034, |
|
"__hun__": 256035, |
|
"__hye__": 256036, |
|
"__ibo__": 256037, |
|
"__ind__": 256038, |
|
"__isl__": 256039, |
|
"__ita__": 256040, |
|
"__jav__": 256041, |
|
"__jpn__": 256042, |
|
"__kan__": 256043, |
|
"__kat__": 256044, |
|
"__kaz__": 256045, |
|
"__khk__": 256046, |
|
"__khm__": 256047, |
|
"__kir__": 256048, |
|
"__kor__": 256049, |
|
"__lao__": 256050, |
|
"__lit__": 256051, |
|
"__lug__": 256052, |
|
"__luo__": 256053, |
|
"__lvs__": 256054, |
|
"__mai__": 256055, |
|
"__mal__": 256056, |
|
"__mar__": 256057, |
|
"__mkd__": 256058, |
|
"__mlt__": 256059, |
|
"__mni__": 256060, |
|
"__mya__": 256061, |
|
"__nld__": 256062, |
|
"__nno__": 256063, |
|
"__nob__": 256064, |
|
"__npi__": 256065, |
|
"__nya__": 256066, |
|
"__ory__": 256067, |
|
"__pan__": 256068, |
|
"__pbt__": 256069, |
|
"__pes__": 256070, |
|
"__pol__": 256071, |
|
"__por__": 256072, |
|
"__ron__": 256073, |
|
"__rus__": 256074, |
|
"__sat__": 256075, |
|
"__slk__": 256076, |
|
"__slv__": 256077, |
|
"__sna__": 256078, |
|
"__snd__": 256079, |
|
"__som__": 256080, |
|
"__spa__": 256081, |
|
"__srp__": 256082, |
|
"__swe__": 256083, |
|
"__swh__": 256084, |
|
"__tam__": 256085, |
|
"__tel__": 256086, |
|
"__tgk__": 256087, |
|
"__tgl__": 256088, |
|
"__tha__": 256089, |
|
"__tur__": 256090, |
|
"__ukr__": 256091, |
|
"__urd__": 256092, |
|
"__uzn__": 256093, |
|
"__vie__": 256094, |
|
"__yor__": 256095, |
|
"__yue__": 256096, |
|
"__zlm__": 256097, |
|
"__zul__": 256098 |
|
} |
|
|