Upload tokenizer
a7517e3
verified
|
{ |
|
"<mask>": 50000, |
|
"ace_Arab": 50001, |
|
"ace_Latn": 50002, |
|
"acm_Arab": 50003, |
|
"acq_Arab": 50004, |
|
"aeb_Arab": 50005, |
|
"afr_Latn": 50006, |
|
"ajp_Arab": 50007, |
|
"aka_Latn": 50008, |
|
"als_Latn": 50009, |
|
"amh_Ethi": 50010, |
|
"apc_Arab": 50011, |
|
"arb_Arab": 50012, |
|
"arg_Latn": 50013, |
|
"arn_Latn": 50014, |
|
"ars_Arab": 50015, |
|
"ary_Arab": 50016, |
|
"arz_Arab": 50017, |
|
"asm_Beng": 50018, |
|
"ast_Latn": 50019, |
|
"awa_Deva": 50020, |
|
"ayr_Latn": 50021, |
|
"azb_Arab": 50022, |
|
"azj_Latn": 50023, |
|
"bak_Cyrl": 50024, |
|
"bam_Latn": 50025, |
|
"ban_Latn": 50026, |
|
"bel_Cyrl": 50027, |
|
"bem_Latn": 50028, |
|
"ben_Beng": 50029, |
|
"bho_Deva": 50030, |
|
"bjn_Arab": 50031, |
|
"bjn_Latn": 50032, |
|
"bod_Tibt": 50033, |
|
"bos_Latn": 50034, |
|
"bug_Latn": 50035, |
|
"bul_Cyrl": 50036, |
|
"cat_Latn": 50037, |
|
"ceb_Latn": 50038, |
|
"ces_Latn": 50039, |
|
"cjk_Latn": 50040, |
|
"ckb_Arab": 50041, |
|
"crh_Latn": 50042, |
|
"cym_Latn": 50043, |
|
"dan_Latn": 50044, |
|
"deu_Latn": 50045, |
|
"dik_Latn": 50046, |
|
"dyu_Latn": 50047, |
|
"dzo_Tibt": 50048, |
|
"ell_Grek": 50049, |
|
"eng_Latn": 50050, |
|
"epo_Latn": 50051, |
|
"est_Latn": 50052, |
|
"eus_Latn": 50053, |
|
"ewe_Latn": 50054, |
|
"fao_Latn": 50055, |
|
"fij_Latn": 50056, |
|
"fin_Latn": 50057, |
|
"fon_Latn": 50058, |
|
"fra_Latn": 50059, |
|
"fur_Latn": 50060, |
|
"fuv_Latn": 50061, |
|
"gaz_Latn": 50062, |
|
"gla_Latn": 50063, |
|
"gle_Latn": 50064, |
|
"glg_Latn": 50065, |
|
"grn_Latn": 50066, |
|
"guj_Gujr": 50067, |
|
"hat_Latn": 50068, |
|
"hau_Latn": 50069, |
|
"heb_Hebr": 50070, |
|
"hin_Deva": 50071, |
|
"hne_Deva": 50072, |
|
"hrv_Latn": 50073, |
|
"hun_Latn": 50074, |
|
"hye_Armn": 50075, |
|
"ibo_Latn": 50076, |
|
"ilo_Latn": 50077, |
|
"ind_Latn": 50078, |
|
"isl_Latn": 50079, |
|
"ita_Latn": 50080, |
|
"jav_Latn": 50081, |
|
"jpn_Jpan": 50082, |
|
"kab_Latn": 50083, |
|
"kac_Latn": 50084, |
|
"kam_Latn": 50085, |
|
"kan_Knda": 50086, |
|
"kas_Arab": 50087, |
|
"kas_Deva": 50088, |
|
"kat_Geor": 50089, |
|
"kaz_Cyrl": 50090, |
|
"kbp_Latn": 50091, |
|
"kea_Latn": 50092, |
|
"khk_Cyrl": 50093, |
|
"khm_Khmr": 50094, |
|
"kik_Latn": 50095, |
|
"kin_Latn": 50096, |
|
"kir_Cyrl": 50097, |
|
"kmb_Latn": 50098, |
|
"kmr_Latn": 50099, |
|
"knc_Arab": 50100, |
|
"knc_Latn": 50101, |
|
"kon_Latn": 50102, |
|
"kor_Hang": 50103, |
|
"lao_Laoo": 50104, |
|
"lij_Latn": 50105, |
|
"lim_Latn": 50106, |
|
"lin_Latn": 50107, |
|
"lit_Latn": 50108, |
|
"lmo_Latn": 50109, |
|
"ltg_Latn": 50110, |
|
"ltz_Latn": 50111, |
|
"lua_Latn": 50112, |
|
"lug_Latn": 50113, |
|
"luo_Latn": 50114, |
|
"lus_Latn": 50115, |
|
"lvs_Latn": 50116, |
|
"mag_Deva": 50117, |
|
"mai_Deva": 50118, |
|
"mal_Mlym": 50119, |
|
"mar_Deva": 50120, |
|
"min_Latn": 50121, |
|
"mkd_Cyrl": 50122, |
|
"mlt_Latn": 50123, |
|
"mni_Beng": 50124, |
|
"mos_Latn": 50125, |
|
"mri_Latn": 50126, |
|
"mya_Mymr": 50127, |
|
"nld_Latn": 50128, |
|
"nno_Latn": 50129, |
|
"nob_Latn": 50130, |
|
"npi_Deva": 50131, |
|
"nso_Latn": 50132, |
|
"nus_Latn": 50133, |
|
"nya_Latn": 50134, |
|
"oci_Latn": 50135, |
|
"ory_Orya": 50136, |
|
"pag_Latn": 50137, |
|
"pan_Guru": 50138, |
|
"pap_Latn": 50139, |
|
"pbt_Arab": 50140, |
|
"pes_Arab": 50141, |
|
"plt_Latn": 50142, |
|
"pol_Latn": 50143, |
|
"por_Latn": 50144, |
|
"prs_Arab": 50145, |
|
"quy_Latn": 50146, |
|
"ron_Latn": 50147, |
|
"run_Latn": 50148, |
|
"rus_Cyrl": 50149, |
|
"sag_Latn": 50150, |
|
"san_Deva": 50151, |
|
"sat_Beng": 50152, |
|
"scn_Latn": 50153, |
|
"shn_Mymr": 50154, |
|
"sin_Sinh": 50155, |
|
"slk_Latn": 50156, |
|
"slv_Latn": 50157, |
|
"smo_Latn": 50158, |
|
"sna_Latn": 50159, |
|
"snd_Arab": 50160, |
|
"som_Latn": 50161, |
|
"sot_Latn": 50162, |
|
"spa_Latn": 50163, |
|
"srd_Latn": 50164, |
|
"srp_Cyrl": 50165, |
|
"ssw_Latn": 50166, |
|
"sun_Latn": 50167, |
|
"swe_Latn": 50168, |
|
"swh_Latn": 50169, |
|
"szl_Latn": 50170, |
|
"tam_Taml": 50171, |
|
"taq_Latn": 50172, |
|
"taq_Tfng": 50173, |
|
"tat_Cyrl": 50174, |
|
"tel_Telu": 50175, |
|
"tgk_Cyrl": 50176, |
|
"tgl_Latn": 50177, |
|
"tha_Thai": 50178, |
|
"tir_Ethi": 50179, |
|
"tpi_Latn": 50180, |
|
"tsn_Latn": 50181, |
|
"tso_Latn": 50182, |
|
"tuk_Latn": 50183, |
|
"tum_Latn": 50184, |
|
"tur_Latn": 50185, |
|
"twi_Latn": 50186, |
|
"tzm_Tfng": 50187, |
|
"uig_Arab": 50188, |
|
"ukr_Cyrl": 50189, |
|
"umb_Latn": 50190, |
|
"urd_Arab": 50191, |
|
"uzn_Latn": 50192, |
|
"vec_Latn": 50193, |
|
"vie_Latn": 50194, |
|
"war_Latn": 50195, |
|
"wol_Latn": 50196, |
|
"xho_Latn": 50197, |
|
"ydd_Hebr": 50198, |
|
"yor_Latn": 50199, |
|
"yue_Hant": 50200, |
|
"zho_Hans": 50201, |
|
"zho_Hant": 50202, |
|
"zsm_Latn": 50203, |
|
"zul_Latn": 50204 |
|
} |
|
|