goldfish-models commited on
Commit
34d61b5
1 Parent(s): 303ad5d

Upload haw_latn_100mb tokenizer.

Browse files
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"[XXXXX950]": 50954, "[XXXXX588]": 50592, "[XXXXX847]": 50851, "[XXXXX101]": 50105, "[XXXXX339]": 50343, "[XXXXX543]": 50547, "[XXXXX1137]": 51141, "[XXXXX932]": 50936, "[XXXXX18]": 50022, "[XXXXX623]": 50627, "[XXXXX784]": 50788, "[XXXXX747]": 50751, "[XXXXX157]": 50161, "[XXXXX748]": 50752, "[XXXXX946]": 50950, "[XXXXX705]": 50709, "[XXXXX526]": 50530, "[XXXXX1184]": 51188, "[XXXXX882]": 50886, "[XXXXX40]": 50044, "[XXXXX10]": 50014, "[XXXXX344]": 50348, "[XXXXX714]": 50718, "[XXXXX522]": 50526, "[XXXXX362]": 50366, "[XXXXX533]": 50537, "[XXXXX232]": 50236, "[XXXXX532]": 50536, "[XXXXX579]": 50583, "[XXXXX169]": 50173, "[XXXXX280]": 50284, "[XXXXX987]": 50991, "[XXXXX240]": 50244, "[XXXXX783]": 50787, "[XXXXX151]": 50155, "[XXXXX463]": 50467, "[XXXXX425]": 50429, "[XXXXX24]": 50028, "[XXXXX778]": 50782, "[XXXXX495]": 50499, "[XXXXX552]": 50556, "[XXXXX1163]": 51167, "[XXXXX848]": 50852, "[XXXXX123]": 50127, "[XXXXX673]": 50677, "[XXXXX363]": 50367, "[XXXXX229]": 50233, "[XXXXX1055]": 51059, "[XXXXX158]": 50162, "[XXXXX652]": 50656, "[XXXXX710]": 50714, "[XXXXX412]": 50416, "[XXXXX590]": 50594, "[XXXXX980]": 50984, "[XXXXX749]": 50753, "[XXXXX786]": 50790, "[XXXXX41]": 50045, "[XXXXX792]": 50796, "[XXXXX702]": 50706, "[XXXXX610]": 50614, "[XXXXX1060]": 51064, "[XXXXX261]": 50265, "[XXXXX800]": 50804, "[XXXXX594]": 50598, "[XXXXX650]": 50654, "[XXXXX162]": 50166, "[XXXXX413]": 50417, "[XXXXX424]": 50428, "[XXXXX267]": 50271, "[XXXXX508]": 50512, "[XXXXX444]": 50448, "[XXXXX796]": 50800, "[XXXXX656]": 50660, "[XXXXX1128]": 51132, "[XXXXX118]": 50122, "[XXXXX1165]": 51169, "[XXXXX87]": 50091, "[XXXXX1172]": 51176, "[XXXXX1009]": 51013, "[XXXXX247]": 50251, "[XXXXX320]": 50324, "[XXXXX134]": 50138, "[XXXXX477]": 50481, "[XXXXX972]": 50976, "[XXXXX780]": 50784, "[XXXXX82]": 50086, "[XXXXX836]": 50840, "[XXXXX390]": 50394, "[XXXXX1175]": 51179, "[XXXXX573]": 50577, "[XXXXX994]": 50998, "[XXXXX434]": 50438, "[XXXXX273]": 50277, "[XXXXX350]": 50354, "[XXXXX401]": 50405, "[XXXXX883]": 50887, "[XXXXX843]": 50847, "[XXXXX414]": 50418, "[XXXXX430]": 50434, "[XXXXX685]": 50689, "[XXXXX728]": 50732, "[XXXXX620]": 50624, "[XXXXX961]": 50965, "[XXXXX335]": 50339, "[XXXXX1123]": 51127, "[XXXXX669]": 50673, "[XXXXX184]": 50188, "[XXXXX73]": 50077, "[XXXXX612]": 50616, "[XXXXX327]": 50331, "[XXXXX902]": 50906, "[XXXXX183]": 50187, "[XXXXX812]": 50816, "[XXXXX734]": 50738, "[XXXXX299]": 50303, "[XXXXX723]": 50727, "[XXXXX777]": 50781, "[XXXXX254]": 50258, "[XXXXX255]": 50259, "[XXXXX281]": 50285, "[XXXXX289]": 50293, "[XXXXX569]": 50573, "[XXXXX465]": 50469, "[XXXXX907]": 50911, "[XXXXX172]": 50176, "[XXXXX429]": 50433, "[XXXXX858]": 50862, "[XXXXX876]": 50880, "[XXXXX758]": 50762, "[XXXXX824]": 50828, "[XXXXX353]": 50357, "[XXXXX613]": 50617, "[XXXXX757]": 50761, "[XXXXX188]": 50192, "[XXXXX227]": 50231, "[XXXXX1068]": 51072, "[XXXXX805]": 50809, "[XXXXX855]": 50859, "[XXXXX649]": 50653, "[XXXXX214]": 50218, "[XXXXX564]": 50568, "[XXXXX245]": 50249, "[XXXXX545]": 50549, "[XXXXX314]": 50318, "[XXXXX139]": 50143, "[XXXXX308]": 50312, "[XXXXX400]": 50404, "[XXXXX1074]": 51078, "[XXXXX1170]": 51174, "[XXXXX1140]": 51144, "[XXXXX1065]": 51069, "[XXXXX369]": 50373, "[XXXXX854]": 50858, "[XXXXX12]": 50016, "[XXXXX252]": 50256, "[XXXXX1134]": 51138, "[XXXXX708]": 50712, "[XXXXX504]": 50508, "[XXXXX937]": 50941, "[XXXXX835]": 50839, "[XXXXX512]": 50516, "[XXXXX30]": 50034, "[XXXXX1102]": 51106, "[XXXXX1098]": 51102, "[XXXXX376]": 50380, "[XXXXX977]": 50981, "[XXXXX115]": 50119, "[XXXXX302]": 50306, "[XXXXX492]": 50496, "[XXXXX128]": 50132, "[XXXXX1159]": 51163, "[XXXXX1045]": 51049, "[XXXXX161]": 50165, "[XXXXX966]": 50970, "[XXXXX286]": 50290, "[XXXXX880]": 50884, "[XXXXX368]": 50372, "[XXXXX1192]": 51196, "[XXXXX535]": 50539, "[XXXXX947]": 50951, "[XXXXX374]": 50378, "[XXXXX431]": 50435, "[XXXXX554]": 50558, "[XXXXX838]": 50842, "[XXXXX951]": 50955, "[XXXXX639]": 50643, "[XXXXX850]": 50854, "[XXXXX1119]": 51123, "[XXXXX938]": 50942, "[XXXXX547]": 50551, "[XXXXX962]": 50966, "[XXXXX908]": 50912, "[XXXXX346]": 50350, "[XXXXX59]": 50063, "[XXXXX312]": 50316, "[XXXXX1131]": 51135, "[XXXXX839]": 50843, "[XXXXX638]": 50642, "[XXXXX436]": 50440, "[XXXXX373]": 50377, "[XXXXX861]": 50865, "[XXXXX1190]": 51194, "[XXXXX1024]": 51028, "[XXXXX967]": 50971, "[XXXXX61]": 50065, "[XXXXX844]": 50848, "[XXXXX960]": 50964, "[XXXXX1178]": 51182, "[XXXXX616]": 50620, "[XXXXX349]": 50353, "[XXXXX832]": 50836, "[XXXXX1110]": 51114, "[XXXXX1046]": 51050, "[XXXXX455]": 50459, "[XXXXX96]": 50100, "[XXXXX1099]": 51103, "[XXXXX108]": 50112, "[XXXXX1191]": 51195, "[XXXXX979]": 50983, "[XXXXX975]": 50979, "[XXXXX904]": 50908, "[XXXXX963]": 50967, "[XXXXX513]": 50517, "[XXXXX1003]": 51007, "[XXXXX601]": 50605, "[XXXXX890]": 50894, "[XXXXX234]": 50238, "[XXXXX119]": 50123, "[XXXXX1081]": 51085, "[XXXXX717]": 50721, "[XXXXX50]": 50054, "[XXXXX662]": 50666, "[XXXXX1194]": 51198, "[XXXXX65]": 50069, "[XXXXX857]": 50861, "[XXXXX442]": 50446, "[XXXXX821]": 50825, "[XXXXX1000]": 51004, "[XXXXX58]": 50062, "[XXXXX1002]": 51006, "[XXXXX263]": 50267, "[XXXXX377]": 50381, "[XXXXX803]": 50807, "[XXXXX279]": 50283, "[XXXXX497]": 50501, "[XXXXX1005]": 51009, "[XXXXX375]": 50379, "[XXXXX340]": 50344, "[XXXXX531]": 50535, "[XXXXX655]": 50659, "[XXXXX1136]": 51140, "[XXXXX631]": 50635, "[XXXXX864]": 50868, "[XXXXX689]": 50693, "[XXXXX974]": 50978, "[XXXXX548]": 50552, "[XXXXX223]": 50227, "[XXXXX48]": 50052, "[XXXXX457]": 50461, "[XXXXX1080]": 51084, "[XXXXX85]": 50089, "[XXXXX287]": 50291, "[XXXXX661]": 50665, "[XXXXX420]": 50424, "[XXXXX621]": 50625, "[XXXXX743]": 50747, "[XXXXX215]": 50219, "[XXXXX725]": 50729, "[XXXXX153]": 50157, "[XXXXX707]": 50711, "[XXXXX706]": 50710, "[XXXXX982]": 50986, "[XXXXX729]": 50733, "[XXXXX511]": 50515, "[XXXXX1079]": 51083, "[XXXXX586]": 50590, "[XXXXX929]": 50933, "[XXXXX367]": 50371, "[XXXXX764]": 50768, "[XXXXX295]": 50299, "[XXXXX68]": 50072, "[XXXXX910]": 50914, "[XXXXX467]": 50471, "[XXXXX144]": 50148, "[XXXXX488]": 50492, "[XXXXX901]": 50905, "[XXXXX665]": 50669, "[XXXXX196]": 50200, "[XXXXX331]": 50335, "[XXXXX149]": 50153, "[XXXXX1039]": 51043, "[XXXXX642]": 50646, "[XXXXX683]": 50687, "[XXXXX637]": 50641, "[XXXXX544]": 50548, "[XXXXX1101]": 51105, "[XXXXX489]": 50493, "[XXXXX92]": 50096, "[XXXXX587]": 50591, "[XXXXX599]": 50603, "[XXXXX831]": 50835, "[XXXXX433]": 50437, "[XXXXX439]": 50443, "[XXXXX629]": 50633, "[XXXXX292]": 50296, "[XXXXX13]": 50017, "[XXXXX310]": 50314, "[XXXXX1058]": 51062, "[XXXXX416]": 50420, "[XXXXX271]": 50275, "[XXXXX57]": 50061, "[XXXXX294]": 50298, "[XXXXX250]": 50254, "[XXXXX19]": 50023, "[XXXXX126]": 50130, "[XXXXX259]": 50263, "[XXXXX811]": 50815, "[XXXXX888]": 50892, "[XXXXX992]": 50996, "[XXXXX268]": 50272, "[XXXXX117]": 50121, "[XXXXX958]": 50962, "[XXXXX606]": 50610, "[XXXXX1106]": 51110, "[XXXXX905]": 50909, "[XXXXX1036]": 51040, "[XXXXX581]": 50585, "[XXXXX1114]": 51118, "[XXXXX342]": 50346, "[XXXXX33]": 50037, "[XXXXX63]": 50067, "[XXXXX97]": 50101, "[XXXXX724]": 50728, "[XXXXX1069]": 51073, "[XXXXX140]": 50144, "[XXXXX732]": 50736, "[XXXXX341]": 50345, "[XXXXX1097]": 51101, "[XXXXX1129]": 51133, "[XXXXX94]": 50098, "[XXXXX957]": 50961, "[XXXXX1040]": 51044, "[XXXXX928]": 50932, "[XXXXX618]": 50622, "[XXXXX1120]": 51124, "[XXXXX452]": 50456, "[XXXXX605]": 50609, "[XXXXX1078]": 51082, "[XXXXX456]": 50460, "[XXXXX454]": 50458, "[XXXXX8]": 50012, "[XXXXX840]": 50844, "[XXXXX154]": 50158, "[XXXXX1016]": 51020, "[XXXXX345]": 50349, "[XXXXX667]": 50671, "[XXXXX608]": 50612, "[XXXXX711]": 50715, "[XXXXX1073]": 51077, "[XXXXX102]": 50106, "[XXXXX654]": 50658, "[XXXXX790]": 50794, "[XXXXX670]": 50674, "[XXXXX524]": 50528, "[XXXXX124]": 50128, "[XXXXX71]": 50075, "[XXXXX389]": 50393, "[XXXXX739]": 50743, "[XXXXX110]": 50114, "[XXXXX114]": 50118, "[XXXXX438]": 50442, "[XXXXX626]": 50630, "[XXXXX718]": 50722, "[XXXXX160]": 50164, "[XXXXX274]": 50278, "[XXXXX1067]": 51071, "[XXXXX239]": 50243, "[XXXXX893]": 50897, "[XXXXX789]": 50793, "[XXXXX360]": 50364, "[XXXXX603]": 50607, "[XXXXX856]": 50860, "[XXXXX671]": 50675, "[XXXXX863]": 50867, "[XXXXX179]": 50183, "[XXXXX825]": 50829, "[XXXXX695]": 50699, "[XXXXX1183]": 51187, "[XXXXX484]": 50488, "[XXXXX450]": 50454, "[XXXXX782]": 50786, "[XXXXX593]": 50597, "[XXXXX691]": 50695, "[XXXXX133]": 50137, "[XXXXX56]": 50060, "[XXXXX27]": 50031, "[XXXXX1017]": 51021, "[XXXXX818]": 50822, "[XXXXX461]": 50465, "[XXXXX909]": 50913, "[XXXXX779]": 50783, "[XXXXX451]": 50455, "[XXXXX897]": 50901, "[XXXXX870]": 50874, "[XXXXX913]": 50917, "[XXXXX658]": 50662, "[XXXXX187]": 50191, "[XXXXX1156]": 51160, "[XXXXX1139]": 51143, "[XXXXX490]": 50494, "[XXXXX426]": 50430, "[XXXXX185]": 50189, "[XXXXX814]": 50818, "[XXXXX319]": 50323, "[XXXXX64]": 50068, "[XXXXX394]": 50398, "[XXXXX871]": 50875, "[XXXXX822]": 50826, "[XXXXX218]": 50222, "[XXXXX567]": 50571, "[XXXXX701]": 50705, "[XXXXX704]": 50708, "[XXXXX981]": 50985, "[XXXXX684]": 50688, "[XXXXX521]": 50525, "[XXXXX1085]": 51089, "[XXXXX156]": 50160, "[XXXXX72]": 50076, "[XXXXX391]": 50395, "[XXXXX257]": 50261, "[XXXXX217]": 50221, "[XXXXX248]": 50252, "[XXXXX459]": 50463, "[XXXXX793]": 50797, "[XXXXX830]": 50834, "[XXXXX675]": 50679, "[XXXXX520]": 50524, "[XXXXX1193]": 51197, "[XXXXX555]": 50559, "[XXXXX502]": 50506, "[XXXXX1108]": 51112, "[XXXXX195]": 50199, "[XXXXX633]": 50637, "[XXXXX311]": 50315, "[XXXXX121]": 50125, "[XXXXX206]": 50210, "[XXXXX659]": 50663, "[XXXXX1034]": 51038, "[XXXXX506]": 50510, "[XXXXX192]": 50196, "[XXXXX186]": 50190, "[XXXXX6]": 50010, "[XXXXX956]": 50960, "[XXXXX589]": 50593, "[XXXXX404]": 50408, "[XXXXX428]": 50432, "[XXXXX98]": 50102, "[XXXXX296]": 50300, "[XXXXX470]": 50474, "[XXXXX1032]": 51036, "[XXXXX1187]": 51191, "[XXXXX178]": 50182, "[XXXXX625]": 50629, "[XXXXX351]": 50355, "[XXXXX316]": 50320, "[XXXXX660]": 50664, "[XXXXX1083]": 51087, "[XXXXX150]": 50154, "[XXXXX944]": 50948, "[XXXXX336]": 50340, "[XXXXX357]": 50361, "[XXXXX116]": 50120, "[XXXXX482]": 50486, "[XXXXX347]": 50351, "[XXXXX572]": 50576, "[XXXXX1164]": 51168, "[XXXXX802]": 50806, "[XXXXX406]": 50410, "[XXXXX827]": 50831, "[XXXXX1059]": 51063, "[XXXXX1092]": 51096, "[XXXXX644]": 50648, "[XXXXX343]": 50347, "[XXXXX1169]": 51173, "[XXXXX580]": 50584, "[XXXXX1023]": 51027, "[XXXXX318]": 50322, "[XXXXX1185]": 51189, "[XXXXX9]": 50013, "[XXXXX614]": 50618, "[XXXXX719]": 50723, "[XXXXX740]": 50744, "[XXXXX91]": 50095, "[XXXXX67]": 50071, "[XXXXX95]": 50099, "[XXXXX105]": 50109, "[XXXXX209]": 50213, "[XXXXX230]": 50234, "[XXXXX359]": 50363, "[XXXXX1150]": 51154, "[XXXXX794]": 50798, "[XXXXX199]": 50203, "[XXXXX171]": 50175, "[XXXXX1126]": 51130, "[XXXXX462]": 50466, "[XXXXX326]": 50330, "[XXXXX219]": 50223, "[XXXXX29]": 50033, "[XXXXX991]": 50995, "[XXXXX775]": 50779, "[XXXXX930]": 50934, "[XXXXX496]": 50500, "[XXXXX371]": 50375, "[XXXXX769]": 50773, "[XXXXX788]": 50792, "[XXXXX1042]": 51046, "[XXXXX275]": 50279, "[XXXXX1070]": 51074, "[XXXXX851]": 50855, "[XXXXX408]": 50412, "[XXXXX878]": 50882, "[XXXXX1047]": 51051, "[XXXXX976]": 50980, "[XXXXX1095]": 51099, "[XXXXX15]": 50019, "[XXXXX407]": 50411, "[XXXXX486]": 50490, "[XXXXX5]": 50009, "[XXXXX1033]": 51037, "[XXXXX538]": 50542, "[XXXXX602]": 50606, "[XXXXX709]": 50713, "[XXXXX1117]": 51121, "[XXXXX297]": 50301, "[XXXXX1182]": 51186, "[XXXXX34]": 50038, "[XXXXX290]": 50294, "[XXXXX113]": 50117, "[XXXXX752]": 50756, "[XXXXX787]": 50791, "[XXXXX568]": 50572, "[XXXXX1168]": 51172, "[XXXXX321]": 50325, "[XXXXX678]": 50682, "[XXXXX31]": 50035, "[XXXXX328]": 50332, "[XXXXX1176]": 51180, "[XXXXX47]": 50051, "[XXXXX378]": 50382, "[XXXXX869]": 50873, "[XXXXX284]": 50288, "[XXXXX515]": 50519, "[XXXXX942]": 50946, "[XXXXX208]": 50212, "[XXXXX834]": 50838, "[XXXXX1148]": 51152, "[XXXXX23]": 50027, "[XXXXX1008]": 51012, "[XXXXX246]": 50250, "[XXXXX1022]": 51026, "[XXXXX494]": 50498, "[XXXXX558]": 50562, "[XXXXX537]": 50541, "[XXXXX136]": 50140, "[XXXXX730]": 50734, "[XXXXX35]": 50039, "[XXXXX534]": 50538, "[XXXXX596]": 50600, "[XXXXX20]": 50024, "[XXXXX476]": 50480, "[XXXXX577]": 50581, "[XXXXX676]": 50680, "[XXXXX468]": 50472, "[XXXXX615]": 50619, "[XXXXX697]": 50701, "[XXXXX77]": 50081, "[XXXXX235]": 50239, "[XXXXX686]": 50690, "[XXXXX51]": 50055, "[XXXXX152]": 50156, "[XXXXX722]": 50726, "[XXXXX1049]": 51053, "[XXXXX1173]": 51177, "[XXXXX11]": 50015, "[XXXXX453]": 50457, "[XXXXX563]": 50567, "[XXXXX383]": 50387, "[XXXXX978]": 50982, "[XXXXX682]": 50686, "[XXXXX828]": 50832, "[XXXXX380]": 50384, "[XXXXX52]": 50056, "[XXXXX249]": 50253, "[XXXXX1007]": 51011, "[XXXXX193]": 50197, "[XXXXX1063]": 51067, "[XXXXX1084]": 51088, "[XXXXX1107]": 51111, "[XXXXX264]": 50268, "[XXXXX756]": 50760, "[XXXXX1030]": 51034, "[XXXXX300]": 50304, "[XXXXX1062]": 51066, "[XXXXX819]": 50823, "[XXXXX405]": 50409, "[XXXXX76]": 50080, "[XXXXX466]": 50470, "[XXXXX514]": 50518, "[XXXXX361]": 50365, "[XXXXX1112]": 51116, "[XXXXX411]": 50415, "[XXXXX630]": 50634, "[XXXXX640]": 50644, "[XXXXX884]": 50888, "[XXXXX791]": 50795, "[CLS]": 50000, "[XXXXX598]": 50602, "[XXXXX491]": 50495, "[XXXXX813]": 50817, "[XXXXX356]": 50360, "[XXXXX103]": 50107, "[XXXXX464]": 50468, "[XXXXX807]": 50811, "[XXXXX997]": 51001, "[XXXXX1144]": 51148, "[XXXXX923]": 50927, "[XXXXX927]": 50931, "[XXXXX175]": 50179, "[XXXXX210]": 50214, "[XXXXX475]": 50479, "[XXXXX895]": 50899, "[XXXXX926]": 50930, "[XXXXX611]": 50615, "[XXXXX323]": 50327, "[XXXXX1155]": 51159, "[XXXXX1053]": 51057, "[XXXXX446]": 50450, "[XXXXX720]": 50724, "[XXXXX1115]": 51119, "[XXXXX607]": 50611, "[XXXXX773]": 50777, "[XXXXX816]": 50820, "[XXXXX627]": 50631, "[XXXXX645]": 50649, "[XXXXX953]": 50957, "[XXXXX881]": 50885, "[XXXXX885]": 50889, "[XXXXX721]": 50725, "[XXXXX1043]": 51047, "[XXXXX381]": 50385, "[XXXXX768]": 50772, "[XXXXX1157]": 51161, "[XXXXX1027]": 51031, "[XXXXX651]": 50655, "[XXXXX303]": 50307, "[XXXXX221]": 50225, "[XXXXX392]": 50396, "[XXXXX437]": 50441, "[XXXXX781]": 50785, "[XXXXX168]": 50172, "[XXXXX180]": 50184, "[XXXXX440]": 50444, "[XXXXX690]": 50694, "[XXXXX334]": 50338, "[XXXXX510]": 50514, "[XXXXX42]": 50046, "[XXXXX663]": 50667, "[XXXXX755]": 50759, "[XXXXX1082]": 51086, "[XXXXX14]": 50018, "[XXXXX1186]": 51190, "[XXXXX875]": 50879, "[XXXXX382]": 50386, "[XXXXX37]": 50041, "[XXXXX111]": 50115, "[XXXXX922]": 50926, "[XXXXX713]": 50717, "[XXXXX1113]": 51117, "[XXXXX1001]": 51005, "[XXXXX164]": 50168, "[XXXXX1025]": 51029, "[XXXXX527]": 50531, "[XXXXX421]": 50425, "[XXXXX480]": 50484, "[XXXXX1093]": 51097, "[XXXXX530]": 50534, "[XXXXX1132]": 51136, "[XXXXX313]": 50317, "[XXXXX417]": 50421, "[XXXXX145]": 50149, "[XXXXX372]": 50376, "[XXXXX53]": 50057, "[XXXXX283]": 50287, "[XXXXX49]": 50053, "[XXXXX820]": 50824, "[XXXXX1031]": 51035, "[XXXXX106]": 50110, "[XXXXX841]": 50845, "[XXXXX66]": 50070, "[XXXXX1161]": 51165, "[XXXXX266]": 50270, "[XXXXX410]": 50414, "[XXXXX201]": 50205, "[XXXXX648]": 50652, "[XXXXX505]": 50509, "[XXXXX703]": 50707, "[XXXXX422]": 50426, "[XXXXX301]": 50305, "[XXXXX993]": 50997, "[XXXXX726]": 50730, "[XXXXX242]": 50246, "[XXXXX903]": 50907, "[XXXXX518]": 50522, "[XXXXX574]": 50578, "[XXXXX1077]": 51081, "[XXXXX798]": 50802, "[XXXXX1127]": 51131, "[XXXXX332]": 50336, "[XXXXX1012]": 51016, "[XXXXX989]": 50993, "[XXXXX386]": 50390, "[XXXXX74]": 50078, "[XXXXX1171]": 51175, "[XXXXX1035]": 51039, "[XXXXX0]": 50004, "[XXXXX921]": 50925, "[XXXXX277]": 50281, "[XXXXX22]": 50026, "[XXXXX26]": 50030, "[XXXXX479]": 50483, "[XXXXX760]": 50764, "[XXXXX866]": 50870, "[XXXXX940]": 50944, "[XXXXX251]": 50255, "[XXXXX1145]": 51149, "[XXXXX194]": 50198, "[XXXXX322]": 50326, "[XXXXX736]": 50740, "[XXXXX306]": 50310, "[XXXXX1111]": 51115, "[XXXXX891]": 50895, "[XXXXX485]": 50489, "[XXXXX93]": 50097, "[XXXXX143]": 50147, "[XXXXX565]": 50569, "[XXXXX1141]": 51145, "[XXXXX509]": 50513, "[XXXXX1167]": 51171, "[XXXXX473]": 50477, "[XXXXX653]": 50657, "[XXXXX1181]": 51185, "[XXXXX1038]": 51042, "[XXXXX899]": 50903, "[XXXXX965]": 50969, "[XXXXX1096]": 51100, "[XXXXX1090]": 51094, "[XXXXX549]": 50553, "[XXXXX260]": 50264, "[XXXXX738]": 50742, "[XXXXX7]": 50011, "[XXXXX1075]": 51079, "[XXXXX879]": 50883, "[XXXXX1125]": 51129, "[XXXXX1143]": 51147, "[XXXXX1122]": 51126, "[XXXXX759]": 50763, "[XXXXX402]": 50406, "[XXXXX680]": 50684, "[XXXXX801]": 50805, "[XXXXX132]": 50136, "[XXXXX396]": 50400, "[XXXXX309]": 50313, "[XXXXX911]": 50915, "[XXXXX887]": 50891, "[XXXXX1146]": 51150, "[XXXXX469]": 50473, "[XXXXX333]": 50337, "[XXXXX763]": 50767, "[XXXXX959]": 50963, "[XXXXX478]": 50482, "[XXXXX868]": 50872, "[XXXXX917]": 50921, "[XXXXX692]": 50696, "[XXXXX556]": 50560, "[XXXXX69]": 50073, "[XXXXX448]": 50452, "[XXXXX687]": 50691, "[XXXXX507]": 50511, "[XXXXX55]": 50059, "[XXXXX551]": 50555, "[XXXXX1071]": 51075, "[XXXXX189]": 50193, "[XXXXX754]": 50758, "[XXXXX617]": 50621, "[XXXXX935]": 50939, "[XXXXX753]": 50757, "[XXXXX867]": 50871, "[XXXXX1037]": 51041, "[XXXXX833]": 50837, "[XXXXX1195]": 51199, "[XXXXX1052]": 51056, "[XXXXX873]": 50877, "[XXXXX244]": 50248, "[XXXXX1]": 50005, "[XXXXX403]": 50407, "[XXXXX419]": 50423, "[XXXXX258]": 50262, "[XXXXX348]": 50352, "[XXXXX307]": 50311, "[XXXXX387]": 50391, "[XXXXX418]": 50422, "[XXXXX936]": 50940, "[XXXXX366]": 50370, "[XXXXX278]": 50282, "[XXXXX969]": 50973, "[XXXXX715]": 50719, "[XXXXX127]": 50131, "[XXXXX700]": 50704, "[XXXXX228]": 50232, "[XXXXX1004]": 51008, "[XXXXX222]": 50226, "[XXXXX135]": 50139, "[XXXXX865]": 50869, "[XXXXX666]": 50670, "[XXXXX141]": 50145, "[XXXXX1086]": 51090, "[XXXXX1154]": 51158, "[XXXXX191]": 50195, "[XXXXX1160]": 51164, "[XXXXX83]": 50087, "[XXXXX900]": 50904, "[XXXXX604]": 50608, "[XXXXX120]": 50124, "[XXXXX745]": 50749, "[XXXXX1105]": 51109, "[XXXXX766]": 50770, "[XXXXX968]": 50972, "[XXXXX646]": 50650, "[XXXXX472]": 50476, "[XXXXX1138]": 51142, "[XXXXX216]": 50220, "[XXXXX1166]": 51170, "[XXXXX1054]": 51058, "[XXXXX60]": 50064, "[XXXXX845]": 50849, "[XXXXX925]": 50929, "[XXXXX591]": 50595, "[XXXXX674]": 50678, "[XXXXX560]": 50564, "[XXXXX842]": 50846, "[XXXXX562]": 50566, "[XXXXX1188]": 51192, "[XXXXX317]": 50321, "[XXXXX298]": 50302, "[XXXXX379]": 50383, "[XXXXX197]": 50201, "[XXXXX1061]": 51065, "[XXXXX398]": 50402, "[XXXXX231]": 50235, "[XXXXX325]": 50329, "[XXXXX795]": 50799, "[XXXXX934]": 50938, "[XXXXX1006]": 51010, "[XXXXX859]": 50863, "[XXXXX1057]": 51061, "[XXXXX78]": 50082, "[XXXXX355]": 50359, "[XXXXX924]": 50928, "[XXXXX762]": 50766, "[XXXXX742]": 50746, "[XXXXX483]": 50487, "[XXXXX849]": 50853, "[XXXXX256]": 50260, "[XXXXX575]": 50579, "[XXXXX224]": 50228, "[XXXXX877]": 50881, "[XXXXX1026]": 51030, "[XXXXX557]": 50561, "[XXXXX487]": 50491, "[XXXXX262]": 50266, "[XXXXX198]": 50202, "[XXXXX75]": 50079, "[XXXXX474]": 50478, "[XXXXX1121]": 51125, "[XXXXX399]": 50403, "[XXXXX39]": 50043, "[XXXXX996]": 51000, "[XXXXX112]": 50116, "[XXXXX1010]": 51014, "[XXXXX393]": 50397, "[XXXXX806]": 50810, "[XXXXX619]": 50623, "[XXXXX293]": 50297, "[XXXXX166]": 50170, "[XXXXX1015]": 51019, "[XXXXX999]": 51003, "[XXXXX735]": 50739, "[XXXXX941]": 50945, "[XXXXX733]": 50737, "[XXXXX829]": 50833, "[XXXXX173]": 50177, "[XXXXX1180]": 51184, "[XXXXX716]": 50720, "[XXXXX635]": 50639, "[XXXXX945]": 50949, "[XXXXX395]": 50399, "[XXXXX550]": 50554, "[XXXXX449]": 50453, "[XXXXX846]": 50850, "[XXXXX458]": 50462, "[XXXXX622]": 50626, "[XXXXX540]": 50544, "[XXXXX817]": 50821, "[XXXXX17]": 50021, "[XXXXX220]": 50224, "[XXXXX582]": 50586, "[XXXXX973]": 50977, "[XXXXX737]": 50741, "[XXXXX1013]": 51017, "[XXXXX624]": 50628, "[XXXXX3]": 50007, "[XXXXX1158]": 51162, "[XXXXX952]": 50956, "[XXXXX1149]": 51153, "[XXXXX36]": 50040, "[XXXXX750]": 50754, "[XXXXX664]": 50668, "[XXXXX525]": 50529, "[XXXXX329]": 50333, "[XXXXX912]": 50916, "[XXXXX1041]": 51045, "[XXXXX746]": 50750, "[XXXXX529]": 50533, "[XXXXX176]": 50180, "[XXXXX539]": 50543, "[XXXXX641]": 50645, "[XXXXX751]": 50755, "[XXXXX131]": 50135, "[XXXXX147]": 50151, "[XXXXX609]": 50613, "[XXXXX4]": 50008, "[XXXXX1020]": 51024, "[XXXXX542]": 50546, "[XXXXX305]": 50309, "[XXXXX165]": 50169, "[XXXXX241]": 50245, "[XXXXX771]": 50775, "[XXXXX32]": 50036, "[XXXXX985]": 50989, "[XXXXX1089]": 51093, "[XXXXX174]": 50178, "[MASK]": 50003, "[XXXXX583]": 50587, "[XXXXX983]": 50987, "[XXXXX894]": 50898, "[XXXXX939]": 50943, "[XXXXX358]": 50362, "[XXXXX315]": 50319, "[XXXXX519]": 50523, "[XXXXX481]": 50485, "[XXXXX1048]": 51052, "[XXXXX447]": 50451, "[XXXXX1179]": 51183, "[XXXXX104]": 50108, "[XXXXX1152]": 51156, "[XXXXX571]": 50575, "[XXXXX1100]": 51104, "[XXXXX874]": 50878, "[XXXXX694]": 50698, "[XXXXX636]": 50640, "[XXXXX681]": 50685, "[XXXXX986]": 50990, "[XXXXX54]": 50058, "[XXXXX471]": 50475, "[XXXXX148]": 50152, "[XXXXX43]": 50047, "[XXXXX38]": 50042, "[XXXXX352]": 50356, "[XXXXX1029]": 51033, "[XXXXX546]": 50550, "[XXXXX523]": 50527, "[XXXXX1072]": 51076, "[XXXXX501]": 50505, "[XXXXX89]": 50093, "[XXXXX338]": 50342, "[XXXXX125]": 50129, "<pad>": 50002, "[XXXXX896]": 50900, "[XXXXX385]": 50389, "[XXXXX553]": 50557, "[XXXXX226]": 50230, "[XXXXX237]": 50241, "[XXXXX815]": 50819, "[XXXXX79]": 50083, "[XXXXX632]": 50636, "[XXXXX122]": 50126, "[XXXXX1091]": 51095, "[XXXXX1050]": 51054, "[XXXXX304]": 50308, "[XXXXX657]": 50661, "[XXXXX964]": 50968, "[XXXXX203]": 50207, "[XXXXX44]": 50048, "[XXXXX931]": 50935, "[XXXXX28]": 50032, "[XXXXX212]": 50216, "[XXXXX1142]": 51146, "[XXXXX46]": 50050, "[XXXXX744]": 50748, "[XXXXX712]": 50716, "[XXXXX137]": 50141, "[XXXXX955]": 50959, "[XXXXX86]": 50090, "[XXXXX541]": 50545, "[XXXXX330]": 50334, "[XXXXX799]": 50803, "[XXXXX600]": 50604, "[XXXXX772]": 50776, "[XXXXX364]": 50368, "[XXXXX804]": 50808, "[XXXXX898]": 50902, "[XXXXX995]": 50999, "[XXXXX265]": 50269, "[XXXXX918]": 50922, "[XXXXX886]": 50890, "[XXXXX100]": 50104, "[XXXXX354]": 50358, "[XXXXX595]": 50599, "[XXXXX1177]": 51181, "[XXXXX517]": 50521, "[XXXXX177]": 50181, "[XXXXX460]": 50464, "[XXXXX767]": 50771, "[XXXXX1087]": 51091, "[XXXXX253]": 50257, "[XXXXX282]": 50286, "[XXXXX776]": 50780, "[XXXXX889]": 50893, "[XXXXX808]": 50812, "[XXXXX810]": 50814, "[XXXXX181]": 50185, "[XXXXX1109]": 51113, "[XXXXX727]": 50731, "[XXXXX765]": 50769, "[XXXXX570]": 50574, "[XXXXX1028]": 51032, "[XXXXX761]": 50765, "[XXXXX643]": 50647, "[XXXXX860]": 50864, "[XXXXX276]": 50280, "[XXXXX443]": 50447, "[XXXXX155]": 50159, "[XXXXX852]": 50856, "[XXXXX797]": 50801, "[XXXXX693]": 50697, "[XXXXX919]": 50923, "[XXXXX1021]": 51025, "[XXXXX1135]": 51139, "[XXXXX243]": 50247, "[XXXXX163]": 50167, "[XXXXX688]": 50692, "[XXXXX1124]": 51128, "[XXXXX285]": 50289, "[XXXXX1103]": 51107, "[XXXXX528]": 50532, "[XXXXX384]": 50388, "[XXXXX146]": 50150, "[XXXXX270]": 50274, "[XXXXX107]": 50111, "[XXXXX435]": 50439, "[XXXXX862]": 50866, "[XXXXX190]": 50194, "[XXXXX933]": 50937, "[XXXXX1066]": 51070, "[XXXXX81]": 50085, "[XXXXX211]": 50215, "[XXXXX597]": 50601, "[XXXXX1133]": 51137, "[XXXXX441]": 50445, "[XXXXX21]": 50025, "[XXXXX445]": 50449, "[XXXXX1076]": 51080, "[XXXXX1174]": 51178, "[XXXXX423]": 50427, "[XXXXX272]": 50276, "[XXXXX592]": 50596, "[XXXXX914]": 50918, "[XXXXX1094]": 51098, "[XXXXX785]": 50789, "[XXXXX370]": 50374, "[XXXXX236]": 50240, "[XXXXX388]": 50392, "[XXXXX1064]": 51068, "[XXXXX920]": 50924, "[XXXXX823]": 50827, "[XXXXX90]": 50094, "[XXXXX500]": 50504, "[XXXXX536]": 50540, "[XXXXX498]": 50502, "[XXXXX584]": 50588, "[XXXXX1147]": 51151, "[XXXXX88]": 50092, "[XXXXX916]": 50920, "[XXXXX1130]": 51134, "[XXXXX288]": 50292, "[XXXXX998]": 51002, "[XXXXX696]": 50700, "[XXXXX1051]": 51055, "[XXXXX397]": 50401, "[XXXXX225]": 50229, "[XXXXX668]": 50672, "[XXXXX1153]": 51157, "[XXXXX672]": 50676, "[XXXXX970]": 50974, "[XXXXX1044]": 51048, "[XXXXX1104]": 51108, "[XXXXX291]": 50295, "[XXXXX770]": 50774, "[XXXXX138]": 50142, "[XXXXX1018]": 51022, "[XXXXX167]": 50171, "[XXXXX1056]": 51060, "[XXXXX954]": 50958, "[XXXXX365]": 50369, "[XXXXX943]": 50947, "[XXXXX204]": 50208, "[XXXXX324]": 50328, "[XXXXX130]": 50134, "[XXXXX80]": 50084, "[XXXXX634]": 50638, "[XXXXX409]": 50413, "[XXXXX238]": 50242, "[XXXXX129]": 50133, "[XXXXX2]": 50006, "[XXXXX1011]": 51015, "[XXXXX1014]": 51018, "[XXXXX25]": 50029, "[XXXXX337]": 50341, "[XXXXX1162]": 51166, "[XXXXX1019]": 51023, "[XXXXX516]": 50520, "[XXXXX45]": 50049, "[XXXXX948]": 50952, "[XXXXX1088]": 51092, "[XXXXX984]": 50988, "[XXXXX503]": 50507, "[XXXXX915]": 50919, "[XXXXX1151]": 51155, "[XXXXX269]": 50273, "[XXXXX415]": 50419, "[XXXXX561]": 50565, "[XXXXX647]": 50651, "[XXXXX853]": 50857, "[XXXXX677]": 50681, "[XXXXX62]": 50066, "[XXXXX1118]": 51122, "[XXXXX837]": 50841, "[XXXXX499]": 50503, "[XXXXX493]": 50497, "[XXXXX576]": 50580, "[XXXXX566]": 50570, "[XXXXX233]": 50237, "[XXXXX559]": 50563, "[XXXXX578]": 50582, "[XXXXX1189]": 51193, "[XXXXX699]": 50703, "[XXXXX432]": 50436, "[XXXXX990]": 50994, "[SEP]": 50001, "[XXXXX585]": 50589, "[XXXXX906]": 50910, "[XXXXX205]": 50209, "[XXXXX70]": 50074, "[XXXXX698]": 50702, "[XXXXX731]": 50735, "[XXXXX872]": 50876, "[XXXXX1116]": 51120, "[XXXXX202]": 50206, "[XXXXX200]": 50204, "[XXXXX16]": 50020, "[XXXXX949]": 50953, "[XXXXX892]": 50896, "[XXXXX213]": 50217, "[XXXXX99]": 50103, "[XXXXX109]": 50113, "[XXXXX679]": 50683, "[XXXXX628]": 50632, "[XXXXX741]": 50745, "[XXXXX159]": 50163, "[XXXXX84]": 50088, "[XXXXX142]": 50146, "[XXXXX774]": 50778, "[XXXXX971]": 50975, "[XXXXX427]": 50431, "[XXXXX170]": 50174, "[XXXXX182]": 50186, "[XXXXX826]": 50830, "[XXXXX207]": 50211, "[XXXXX988]": 50992, "[XXXXX809]": 50813}
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "<unk>", "sep_token": "[SEP]", "pad_token": "<pad>", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false}, "additional_special_tokens": ["[XXXXX0]", "[XXXXX1]", "[XXXXX2]", "[XXXXX3]", "[XXXXX4]", "[XXXXX5]", "[XXXXX6]", "[XXXXX7]", "[XXXXX8]", "[XXXXX9]", "[XXXXX10]", "[XXXXX11]", "[XXXXX12]", "[XXXXX13]", "[XXXXX14]", "[XXXXX15]", "[XXXXX16]", "[XXXXX17]", "[XXXXX18]", "[XXXXX19]", "[XXXXX20]", "[XXXXX21]", "[XXXXX22]", "[XXXXX23]", "[XXXXX24]", "[XXXXX25]", "[XXXXX26]", "[XXXXX27]", "[XXXXX28]", "[XXXXX29]", "[XXXXX30]", "[XXXXX31]", "[XXXXX32]", "[XXXXX33]", "[XXXXX34]", "[XXXXX35]", "[XXXXX36]", "[XXXXX37]", "[XXXXX38]", "[XXXXX39]", "[XXXXX40]", "[XXXXX41]", "[XXXXX42]", "[XXXXX43]", "[XXXXX44]", "[XXXXX45]", "[XXXXX46]", "[XXXXX47]", "[XXXXX48]", "[XXXXX49]", "[XXXXX50]", "[XXXXX51]", "[XXXXX52]", "[XXXXX53]", "[XXXXX54]", "[XXXXX55]", "[XXXXX56]", "[XXXXX57]", "[XXXXX58]", "[XXXXX59]", "[XXXXX60]", "[XXXXX61]", "[XXXXX62]", "[XXXXX63]", "[XXXXX64]", "[XXXXX65]", "[XXXXX66]", "[XXXXX67]", "[XXXXX68]", "[XXXXX69]", "[XXXXX70]", "[XXXXX71]", "[XXXXX72]", "[XXXXX73]", "[XXXXX74]", "[XXXXX75]", "[XXXXX76]", "[XXXXX77]", "[XXXXX78]", "[XXXXX79]", "[XXXXX80]", "[XXXXX81]", "[XXXXX82]", "[XXXXX83]", "[XXXXX84]", "[XXXXX85]", "[XXXXX86]", "[XXXXX87]", "[XXXXX88]", "[XXXXX89]", "[XXXXX90]", "[XXXXX91]", "[XXXXX92]", "[XXXXX93]", "[XXXXX94]", "[XXXXX95]", "[XXXXX96]", "[XXXXX97]", "[XXXXX98]", "[XXXXX99]", "[XXXXX100]", "[XXXXX101]", "[XXXXX102]", "[XXXXX103]", "[XXXXX104]", "[XXXXX105]", "[XXXXX106]", "[XXXXX107]", "[XXXXX108]", "[XXXXX109]", "[XXXXX110]", "[XXXXX111]", "[XXXXX112]", "[XXXXX113]", "[XXXXX114]", "[XXXXX115]", "[XXXXX116]", "[XXXXX117]", "[XXXXX118]", "[XXXXX119]", "[XXXXX120]", "[XXXXX121]", "[XXXXX122]", "[XXXXX123]", "[XXXXX124]", "[XXXXX125]", "[XXXXX126]", "[XXXXX127]", "[XXXXX128]", "[XXXXX129]", "[XXXXX130]", "[XXXXX131]", "[XXXXX132]", "[XXXXX133]", "[XXXXX134]", "[XXXXX135]", "[XXXXX136]", "[XXXXX137]", "[XXXXX138]", "[XXXXX139]", "[XXXXX140]", "[XXXXX141]", "[XXXXX142]", "[XXXXX143]", "[XXXXX144]", "[XXXXX145]", "[XXXXX146]", "[XXXXX147]", "[XXXXX148]", "[XXXXX149]", "[XXXXX150]", "[XXXXX151]", "[XXXXX152]", "[XXXXX153]", "[XXXXX154]", "[XXXXX155]", "[XXXXX156]", "[XXXXX157]", "[XXXXX158]", "[XXXXX159]", "[XXXXX160]", "[XXXXX161]", "[XXXXX162]", "[XXXXX163]", "[XXXXX164]", "[XXXXX165]", "[XXXXX166]", "[XXXXX167]", "[XXXXX168]", "[XXXXX169]", "[XXXXX170]", "[XXXXX171]", "[XXXXX172]", "[XXXXX173]", "[XXXXX174]", "[XXXXX175]", "[XXXXX176]", "[XXXXX177]", "[XXXXX178]", "[XXXXX179]", "[XXXXX180]", "[XXXXX181]", "[XXXXX182]", "[XXXXX183]", "[XXXXX184]", "[XXXXX185]", "[XXXXX186]", "[XXXXX187]", "[XXXXX188]", "[XXXXX189]", "[XXXXX190]", "[XXXXX191]", "[XXXXX192]", "[XXXXX193]", "[XXXXX194]", "[XXXXX195]", "[XXXXX196]", "[XXXXX197]", "[XXXXX198]", "[XXXXX199]", "[XXXXX200]", "[XXXXX201]", "[XXXXX202]", "[XXXXX203]", "[XXXXX204]", "[XXXXX205]", "[XXXXX206]", "[XXXXX207]", "[XXXXX208]", "[XXXXX209]", "[XXXXX210]", "[XXXXX211]", "[XXXXX212]", "[XXXXX213]", "[XXXXX214]", "[XXXXX215]", "[XXXXX216]", "[XXXXX217]", "[XXXXX218]", "[XXXXX219]", "[XXXXX220]", "[XXXXX221]", "[XXXXX222]", "[XXXXX223]", "[XXXXX224]", "[XXXXX225]", "[XXXXX226]", "[XXXXX227]", "[XXXXX228]", "[XXXXX229]", "[XXXXX230]", "[XXXXX231]", "[XXXXX232]", "[XXXXX233]", "[XXXXX234]", "[XXXXX235]", "[XXXXX236]", "[XXXXX237]", "[XXXXX238]", "[XXXXX239]", "[XXXXX240]", "[XXXXX241]", "[XXXXX242]", "[XXXXX243]", "[XXXXX244]", "[XXXXX245]", "[XXXXX246]", "[XXXXX247]", "[XXXXX248]", "[XXXXX249]", "[XXXXX250]", "[XXXXX251]", "[XXXXX252]", "[XXXXX253]", "[XXXXX254]", "[XXXXX255]", "[XXXXX256]", "[XXXXX257]", "[XXXXX258]", "[XXXXX259]", "[XXXXX260]", "[XXXXX261]", "[XXXXX262]", "[XXXXX263]", "[XXXXX264]", "[XXXXX265]", "[XXXXX266]", "[XXXXX267]", "[XXXXX268]", "[XXXXX269]", "[XXXXX270]", "[XXXXX271]", "[XXXXX272]", "[XXXXX273]", "[XXXXX274]", "[XXXXX275]", "[XXXXX276]", "[XXXXX277]", "[XXXXX278]", "[XXXXX279]", "[XXXXX280]", "[XXXXX281]", "[XXXXX282]", "[XXXXX283]", "[XXXXX284]", "[XXXXX285]", "[XXXXX286]", "[XXXXX287]", "[XXXXX288]", "[XXXXX289]", "[XXXXX290]", "[XXXXX291]", "[XXXXX292]", "[XXXXX293]", "[XXXXX294]", "[XXXXX295]", "[XXXXX296]", "[XXXXX297]", "[XXXXX298]", "[XXXXX299]", "[XXXXX300]", "[XXXXX301]", "[XXXXX302]", "[XXXXX303]", "[XXXXX304]", "[XXXXX305]", "[XXXXX306]", "[XXXXX307]", "[XXXXX308]", "[XXXXX309]", "[XXXXX310]", "[XXXXX311]", "[XXXXX312]", "[XXXXX313]", "[XXXXX314]", "[XXXXX315]", "[XXXXX316]", "[XXXXX317]", "[XXXXX318]", "[XXXXX319]", "[XXXXX320]", "[XXXXX321]", "[XXXXX322]", "[XXXXX323]", "[XXXXX324]", "[XXXXX325]", "[XXXXX326]", "[XXXXX327]", "[XXXXX328]", "[XXXXX329]", "[XXXXX330]", "[XXXXX331]", "[XXXXX332]", "[XXXXX333]", "[XXXXX334]", "[XXXXX335]", "[XXXXX336]", "[XXXXX337]", "[XXXXX338]", "[XXXXX339]", "[XXXXX340]", "[XXXXX341]", "[XXXXX342]", "[XXXXX343]", "[XXXXX344]", "[XXXXX345]", "[XXXXX346]", "[XXXXX347]", "[XXXXX348]", "[XXXXX349]", "[XXXXX350]", "[XXXXX351]", "[XXXXX352]", "[XXXXX353]", "[XXXXX354]", "[XXXXX355]", "[XXXXX356]", "[XXXXX357]", "[XXXXX358]", "[XXXXX359]", "[XXXXX360]", "[XXXXX361]", "[XXXXX362]", "[XXXXX363]", "[XXXXX364]", "[XXXXX365]", "[XXXXX366]", "[XXXXX367]", "[XXXXX368]", "[XXXXX369]", "[XXXXX370]", "[XXXXX371]", "[XXXXX372]", "[XXXXX373]", "[XXXXX374]", "[XXXXX375]", "[XXXXX376]", "[XXXXX377]", "[XXXXX378]", "[XXXXX379]", "[XXXXX380]", "[XXXXX381]", "[XXXXX382]", "[XXXXX383]", "[XXXXX384]", "[XXXXX385]", "[XXXXX386]", "[XXXXX387]", "[XXXXX388]", "[XXXXX389]", "[XXXXX390]", "[XXXXX391]", "[XXXXX392]", "[XXXXX393]", "[XXXXX394]", "[XXXXX395]", "[XXXXX396]", "[XXXXX397]", "[XXXXX398]", "[XXXXX399]", "[XXXXX400]", "[XXXXX401]", "[XXXXX402]", "[XXXXX403]", "[XXXXX404]", "[XXXXX405]", "[XXXXX406]", "[XXXXX407]", "[XXXXX408]", "[XXXXX409]", "[XXXXX410]", "[XXXXX411]", "[XXXXX412]", "[XXXXX413]", "[XXXXX414]", "[XXXXX415]", "[XXXXX416]", "[XXXXX417]", "[XXXXX418]", "[XXXXX419]", "[XXXXX420]", "[XXXXX421]", "[XXXXX422]", "[XXXXX423]", "[XXXXX424]", "[XXXXX425]", "[XXXXX426]", "[XXXXX427]", "[XXXXX428]", "[XXXXX429]", "[XXXXX430]", "[XXXXX431]", "[XXXXX432]", "[XXXXX433]", "[XXXXX434]", "[XXXXX435]", "[XXXXX436]", "[XXXXX437]", "[XXXXX438]", "[XXXXX439]", "[XXXXX440]", "[XXXXX441]", "[XXXXX442]", "[XXXXX443]", "[XXXXX444]", "[XXXXX445]", "[XXXXX446]", "[XXXXX447]", "[XXXXX448]", "[XXXXX449]", "[XXXXX450]", "[XXXXX451]", "[XXXXX452]", "[XXXXX453]", "[XXXXX454]", "[XXXXX455]", "[XXXXX456]", "[XXXXX457]", "[XXXXX458]", "[XXXXX459]", "[XXXXX460]", "[XXXXX461]", "[XXXXX462]", "[XXXXX463]", "[XXXXX464]", "[XXXXX465]", "[XXXXX466]", "[XXXXX467]", "[XXXXX468]", "[XXXXX469]", "[XXXXX470]", "[XXXXX471]", "[XXXXX472]", "[XXXXX473]", "[XXXXX474]", "[XXXXX475]", "[XXXXX476]", "[XXXXX477]", "[XXXXX478]", "[XXXXX479]", "[XXXXX480]", "[XXXXX481]", "[XXXXX482]", "[XXXXX483]", "[XXXXX484]", "[XXXXX485]", "[XXXXX486]", "[XXXXX487]", "[XXXXX488]", "[XXXXX489]", "[XXXXX490]", "[XXXXX491]", "[XXXXX492]", "[XXXXX493]", "[XXXXX494]", "[XXXXX495]", "[XXXXX496]", "[XXXXX497]", "[XXXXX498]", "[XXXXX499]", "[XXXXX500]", "[XXXXX501]", "[XXXXX502]", "[XXXXX503]", "[XXXXX504]", "[XXXXX505]", "[XXXXX506]", "[XXXXX507]", "[XXXXX508]", "[XXXXX509]", "[XXXXX510]", "[XXXXX511]", "[XXXXX512]", "[XXXXX513]", "[XXXXX514]", "[XXXXX515]", "[XXXXX516]", "[XXXXX517]", "[XXXXX518]", "[XXXXX519]", "[XXXXX520]", "[XXXXX521]", "[XXXXX522]", "[XXXXX523]", "[XXXXX524]", "[XXXXX525]", "[XXXXX526]", "[XXXXX527]", "[XXXXX528]", "[XXXXX529]", "[XXXXX530]", "[XXXXX531]", "[XXXXX532]", "[XXXXX533]", "[XXXXX534]", "[XXXXX535]", "[XXXXX536]", "[XXXXX537]", "[XXXXX538]", "[XXXXX539]", "[XXXXX540]", "[XXXXX541]", "[XXXXX542]", "[XXXXX543]", "[XXXXX544]", "[XXXXX545]", "[XXXXX546]", "[XXXXX547]", "[XXXXX548]", "[XXXXX549]", "[XXXXX550]", "[XXXXX551]", "[XXXXX552]", "[XXXXX553]", "[XXXXX554]", "[XXXXX555]", "[XXXXX556]", "[XXXXX557]", "[XXXXX558]", "[XXXXX559]", "[XXXXX560]", "[XXXXX561]", "[XXXXX562]", "[XXXXX563]", "[XXXXX564]", "[XXXXX565]", "[XXXXX566]", "[XXXXX567]", "[XXXXX568]", "[XXXXX569]", "[XXXXX570]", "[XXXXX571]", "[XXXXX572]", "[XXXXX573]", "[XXXXX574]", "[XXXXX575]", "[XXXXX576]", "[XXXXX577]", "[XXXXX578]", "[XXXXX579]", "[XXXXX580]", "[XXXXX581]", "[XXXXX582]", "[XXXXX583]", "[XXXXX584]", "[XXXXX585]", "[XXXXX586]", "[XXXXX587]", "[XXXXX588]", "[XXXXX589]", "[XXXXX590]", "[XXXXX591]", "[XXXXX592]", "[XXXXX593]", "[XXXXX594]", "[XXXXX595]", "[XXXXX596]", "[XXXXX597]", "[XXXXX598]", "[XXXXX599]", "[XXXXX600]", "[XXXXX601]", "[XXXXX602]", "[XXXXX603]", "[XXXXX604]", "[XXXXX605]", "[XXXXX606]", "[XXXXX607]", "[XXXXX608]", "[XXXXX609]", "[XXXXX610]", "[XXXXX611]", "[XXXXX612]", "[XXXXX613]", "[XXXXX614]", "[XXXXX615]", "[XXXXX616]", "[XXXXX617]", "[XXXXX618]", "[XXXXX619]", "[XXXXX620]", "[XXXXX621]", "[XXXXX622]", "[XXXXX623]", "[XXXXX624]", "[XXXXX625]", "[XXXXX626]", "[XXXXX627]", "[XXXXX628]", "[XXXXX629]", "[XXXXX630]", "[XXXXX631]", "[XXXXX632]", "[XXXXX633]", "[XXXXX634]", "[XXXXX635]", "[XXXXX636]", "[XXXXX637]", "[XXXXX638]", "[XXXXX639]", "[XXXXX640]", "[XXXXX641]", "[XXXXX642]", "[XXXXX643]", "[XXXXX644]", "[XXXXX645]", "[XXXXX646]", "[XXXXX647]", "[XXXXX648]", "[XXXXX649]", "[XXXXX650]", "[XXXXX651]", "[XXXXX652]", "[XXXXX653]", "[XXXXX654]", "[XXXXX655]", "[XXXXX656]", "[XXXXX657]", "[XXXXX658]", "[XXXXX659]", "[XXXXX660]", "[XXXXX661]", "[XXXXX662]", "[XXXXX663]", "[XXXXX664]", "[XXXXX665]", "[XXXXX666]", "[XXXXX667]", "[XXXXX668]", "[XXXXX669]", "[XXXXX670]", "[XXXXX671]", "[XXXXX672]", "[XXXXX673]", "[XXXXX674]", "[XXXXX675]", "[XXXXX676]", "[XXXXX677]", "[XXXXX678]", "[XXXXX679]", "[XXXXX680]", "[XXXXX681]", "[XXXXX682]", "[XXXXX683]", "[XXXXX684]", "[XXXXX685]", "[XXXXX686]", "[XXXXX687]", "[XXXXX688]", "[XXXXX689]", "[XXXXX690]", "[XXXXX691]", "[XXXXX692]", "[XXXXX693]", "[XXXXX694]", "[XXXXX695]", "[XXXXX696]", "[XXXXX697]", "[XXXXX698]", "[XXXXX699]", "[XXXXX700]", "[XXXXX701]", "[XXXXX702]", "[XXXXX703]", "[XXXXX704]", "[XXXXX705]", "[XXXXX706]", "[XXXXX707]", "[XXXXX708]", "[XXXXX709]", "[XXXXX710]", "[XXXXX711]", "[XXXXX712]", "[XXXXX713]", "[XXXXX714]", "[XXXXX715]", "[XXXXX716]", "[XXXXX717]", "[XXXXX718]", "[XXXXX719]", "[XXXXX720]", "[XXXXX721]", "[XXXXX722]", "[XXXXX723]", "[XXXXX724]", "[XXXXX725]", "[XXXXX726]", "[XXXXX727]", "[XXXXX728]", "[XXXXX729]", "[XXXXX730]", "[XXXXX731]", "[XXXXX732]", "[XXXXX733]", "[XXXXX734]", "[XXXXX735]", "[XXXXX736]", "[XXXXX737]", "[XXXXX738]", "[XXXXX739]", "[XXXXX740]", "[XXXXX741]", "[XXXXX742]", "[XXXXX743]", "[XXXXX744]", "[XXXXX745]", "[XXXXX746]", "[XXXXX747]", "[XXXXX748]", "[XXXXX749]", "[XXXXX750]", "[XXXXX751]", "[XXXXX752]", "[XXXXX753]", "[XXXXX754]", "[XXXXX755]", "[XXXXX756]", "[XXXXX757]", "[XXXXX758]", "[XXXXX759]", "[XXXXX760]", "[XXXXX761]", "[XXXXX762]", "[XXXXX763]", "[XXXXX764]", "[XXXXX765]", "[XXXXX766]", "[XXXXX767]", "[XXXXX768]", "[XXXXX769]", "[XXXXX770]", "[XXXXX771]", "[XXXXX772]", "[XXXXX773]", "[XXXXX774]", "[XXXXX775]", "[XXXXX776]", "[XXXXX777]", "[XXXXX778]", "[XXXXX779]", "[XXXXX780]", "[XXXXX781]", "[XXXXX782]", "[XXXXX783]", "[XXXXX784]", "[XXXXX785]", "[XXXXX786]", "[XXXXX787]", "[XXXXX788]", "[XXXXX789]", "[XXXXX790]", "[XXXXX791]", "[XXXXX792]", "[XXXXX793]", "[XXXXX794]", "[XXXXX795]", "[XXXXX796]", "[XXXXX797]", "[XXXXX798]", "[XXXXX799]", "[XXXXX800]", "[XXXXX801]", "[XXXXX802]", "[XXXXX803]", "[XXXXX804]", "[XXXXX805]", "[XXXXX806]", "[XXXXX807]", "[XXXXX808]", "[XXXXX809]", "[XXXXX810]", "[XXXXX811]", "[XXXXX812]", "[XXXXX813]", "[XXXXX814]", "[XXXXX815]", "[XXXXX816]", "[XXXXX817]", "[XXXXX818]", "[XXXXX819]", "[XXXXX820]", "[XXXXX821]", "[XXXXX822]", "[XXXXX823]", "[XXXXX824]", "[XXXXX825]", "[XXXXX826]", "[XXXXX827]", "[XXXXX828]", "[XXXXX829]", "[XXXXX830]", "[XXXXX831]", "[XXXXX832]", "[XXXXX833]", "[XXXXX834]", "[XXXXX835]", "[XXXXX836]", "[XXXXX837]", "[XXXXX838]", "[XXXXX839]", "[XXXXX840]", "[XXXXX841]", "[XXXXX842]", "[XXXXX843]", "[XXXXX844]", "[XXXXX845]", "[XXXXX846]", "[XXXXX847]", "[XXXXX848]", "[XXXXX849]", "[XXXXX850]", "[XXXXX851]", "[XXXXX852]", "[XXXXX853]", "[XXXXX854]", "[XXXXX855]", "[XXXXX856]", "[XXXXX857]", "[XXXXX858]", "[XXXXX859]", "[XXXXX860]", "[XXXXX861]", "[XXXXX862]", "[XXXXX863]", "[XXXXX864]", "[XXXXX865]", "[XXXXX866]", "[XXXXX867]", "[XXXXX868]", "[XXXXX869]", "[XXXXX870]", "[XXXXX871]", "[XXXXX872]", "[XXXXX873]", "[XXXXX874]", "[XXXXX875]", "[XXXXX876]", "[XXXXX877]", "[XXXXX878]", "[XXXXX879]", "[XXXXX880]", "[XXXXX881]", "[XXXXX882]", "[XXXXX883]", "[XXXXX884]", "[XXXXX885]", "[XXXXX886]", "[XXXXX887]", "[XXXXX888]", "[XXXXX889]", "[XXXXX890]", "[XXXXX891]", "[XXXXX892]", "[XXXXX893]", "[XXXXX894]", "[XXXXX895]", "[XXXXX896]", "[XXXXX897]", "[XXXXX898]", "[XXXXX899]", "[XXXXX900]", "[XXXXX901]", "[XXXXX902]", "[XXXXX903]", "[XXXXX904]", "[XXXXX905]", "[XXXXX906]", "[XXXXX907]", "[XXXXX908]", "[XXXXX909]", "[XXXXX910]", "[XXXXX911]", "[XXXXX912]", "[XXXXX913]", "[XXXXX914]", "[XXXXX915]", "[XXXXX916]", "[XXXXX917]", "[XXXXX918]", "[XXXXX919]", "[XXXXX920]", "[XXXXX921]", "[XXXXX922]", "[XXXXX923]", "[XXXXX924]", "[XXXXX925]", "[XXXXX926]", "[XXXXX927]", "[XXXXX928]", "[XXXXX929]", "[XXXXX930]", "[XXXXX931]", "[XXXXX932]", "[XXXXX933]", "[XXXXX934]", "[XXXXX935]", "[XXXXX936]", "[XXXXX937]", "[XXXXX938]", "[XXXXX939]", "[XXXXX940]", "[XXXXX941]", "[XXXXX942]", "[XXXXX943]", "[XXXXX944]", "[XXXXX945]", "[XXXXX946]", "[XXXXX947]", "[XXXXX948]", "[XXXXX949]", "[XXXXX950]", "[XXXXX951]", "[XXXXX952]", "[XXXXX953]", "[XXXXX954]", "[XXXXX955]", "[XXXXX956]", "[XXXXX957]", "[XXXXX958]", "[XXXXX959]", "[XXXXX960]", "[XXXXX961]", "[XXXXX962]", "[XXXXX963]", "[XXXXX964]", "[XXXXX965]", "[XXXXX966]", "[XXXXX967]", "[XXXXX968]", "[XXXXX969]", "[XXXXX970]", "[XXXXX971]", "[XXXXX972]", "[XXXXX973]", "[XXXXX974]", "[XXXXX975]", "[XXXXX976]", "[XXXXX977]", "[XXXXX978]", "[XXXXX979]", "[XXXXX980]", "[XXXXX981]", "[XXXXX982]", "[XXXXX983]", "[XXXXX984]", "[XXXXX985]", "[XXXXX986]", "[XXXXX987]", "[XXXXX988]", "[XXXXX989]", "[XXXXX990]", "[XXXXX991]", "[XXXXX992]", "[XXXXX993]", "[XXXXX994]", "[XXXXX995]", "[XXXXX996]", "[XXXXX997]", "[XXXXX998]", "[XXXXX999]", "[XXXXX1000]", "[XXXXX1001]", "[XXXXX1002]", "[XXXXX1003]", "[XXXXX1004]", "[XXXXX1005]", "[XXXXX1006]", "[XXXXX1007]", "[XXXXX1008]", "[XXXXX1009]", "[XXXXX1010]", "[XXXXX1011]", "[XXXXX1012]", "[XXXXX1013]", "[XXXXX1014]", "[XXXXX1015]", "[XXXXX1016]", "[XXXXX1017]", "[XXXXX1018]", "[XXXXX1019]", "[XXXXX1020]", "[XXXXX1021]", "[XXXXX1022]", "[XXXXX1023]", "[XXXXX1024]", "[XXXXX1025]", "[XXXXX1026]", "[XXXXX1027]", "[XXXXX1028]", "[XXXXX1029]", "[XXXXX1030]", "[XXXXX1031]", "[XXXXX1032]", "[XXXXX1033]", "[XXXXX1034]", "[XXXXX1035]", "[XXXXX1036]", "[XXXXX1037]", "[XXXXX1038]", "[XXXXX1039]", "[XXXXX1040]", "[XXXXX1041]", "[XXXXX1042]", "[XXXXX1043]", "[XXXXX1044]", "[XXXXX1045]", "[XXXXX1046]", "[XXXXX1047]", "[XXXXX1048]", "[XXXXX1049]", "[XXXXX1050]", "[XXXXX1051]", "[XXXXX1052]", "[XXXXX1053]", "[XXXXX1054]", "[XXXXX1055]", "[XXXXX1056]", "[XXXXX1057]", "[XXXXX1058]", "[XXXXX1059]", "[XXXXX1060]", "[XXXXX1061]", "[XXXXX1062]", "[XXXXX1063]", "[XXXXX1064]", "[XXXXX1065]", "[XXXXX1066]", "[XXXXX1067]", "[XXXXX1068]", "[XXXXX1069]", "[XXXXX1070]", "[XXXXX1071]", "[XXXXX1072]", "[XXXXX1073]", "[XXXXX1074]", "[XXXXX1075]", "[XXXXX1076]", "[XXXXX1077]", "[XXXXX1078]", "[XXXXX1079]", "[XXXXX1080]", "[XXXXX1081]", "[XXXXX1082]", "[XXXXX1083]", "[XXXXX1084]", "[XXXXX1085]", "[XXXXX1086]", "[XXXXX1087]", "[XXXXX1088]", "[XXXXX1089]", "[XXXXX1090]", "[XXXXX1091]", "[XXXXX1092]", "[XXXXX1093]", "[XXXXX1094]", "[XXXXX1095]", "[XXXXX1096]", "[XXXXX1097]", "[XXXXX1098]", "[XXXXX1099]", "[XXXXX1100]", "[XXXXX1101]", "[XXXXX1102]", "[XXXXX1103]", "[XXXXX1104]", "[XXXXX1105]", "[XXXXX1106]", "[XXXXX1107]", "[XXXXX1108]", "[XXXXX1109]", "[XXXXX1110]", "[XXXXX1111]", "[XXXXX1112]", "[XXXXX1113]", "[XXXXX1114]", "[XXXXX1115]", "[XXXXX1116]", "[XXXXX1117]", "[XXXXX1118]", "[XXXXX1119]", "[XXXXX1120]", "[XXXXX1121]", "[XXXXX1122]", "[XXXXX1123]", "[XXXXX1124]", "[XXXXX1125]", "[XXXXX1126]", "[XXXXX1127]", "[XXXXX1128]", "[XXXXX1129]", "[XXXXX1130]", "[XXXXX1131]", "[XXXXX1132]", "[XXXXX1133]", "[XXXXX1134]", "[XXXXX1135]", "[XXXXX1136]", "[XXXXX1137]", "[XXXXX1138]", "[XXXXX1139]", "[XXXXX1140]", "[XXXXX1141]", "[XXXXX1142]", "[XXXXX1143]", "[XXXXX1144]", "[XXXXX1145]", "[XXXXX1146]", "[XXXXX1147]", "[XXXXX1148]", "[XXXXX1149]", "[XXXXX1150]", "[XXXXX1151]", "[XXXXX1152]", "[XXXXX1153]", "[XXXXX1154]", "[XXXXX1155]", "[XXXXX1156]", "[XXXXX1157]", "[XXXXX1158]", "[XXXXX1159]", "[XXXXX1160]", "[XXXXX1161]", "[XXXXX1162]", "[XXXXX1163]", "[XXXXX1164]", "[XXXXX1165]", "[XXXXX1166]", "[XXXXX1167]", "[XXXXX1168]", "[XXXXX1169]", "[XXXXX1170]", "[XXXXX1171]", "[XXXXX1172]", "[XXXXX1173]", "[XXXXX1174]", "[XXXXX1175]", "[XXXXX1176]", "[XXXXX1177]", "[XXXXX1178]", "[XXXXX1179]", "[XXXXX1180]", "[XXXXX1181]", "[XXXXX1182]", "[XXXXX1183]", "[XXXXX1184]", "[XXXXX1185]", "[XXXXX1186]", "[XXXXX1187]", "[XXXXX1188]", "[XXXXX1189]", "[XXXXX1190]", "[XXXXX1191]", "[XXXXX1192]", "[XXXXX1193]", "[XXXXX1194]", "[XXXXX1195]"]}
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74f725f6c6f6c7accb15d26886e307367059da54078525e6911a93a34036b3d0
3
+ size 1074984
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"do_lower_case": false, "remove_space": true, "keep_accents": true, "bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "<unk>", "sep_token": "[SEP]", "pad_token": "<pad>", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": false, "__type": "AddedToken"}, "sp_model_kwargs": {}, "name_or_path": "models/100mb/haw_latn_100mb", "model_input_names": ["input_ids", "attention_mask"], "special_tokens_map_file": "models/100mb/haw_latn_100mb/special_tokens_map.json", "tokenizer_class": "AlbertTokenizer"}