goldfish-models commited on
Commit
d1b305f
1 Parent(s): 558cb16

Upload pus_arab_5mb tokenizer.

Browse files
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"[XXXXX231]": 44276, "[XXXXX424]": 44469, "[XXXXX262]": 44307, "[XXXXX582]": 44627, "[XXXXX309]": 44354, "[XXXXX370]": 44415, "[XXXXX619]": 44664, "[XXXXX74]": 44119, "[XXXXX700]": 44745, "[XXXXX98]": 44143, "[XXXXX560]": 44605, "[XXXXX206]": 44251, "[XXXXX51]": 44096, "[XXXXX191]": 44236, "[XXXXX607]": 44652, "[XXXXX990]": 45035, "[XXXXX644]": 44689, "[XXXXX851]": 44896, "[XXXXX153]": 44198, "[XXXXX716]": 44761, "[XXXXX590]": 44635, "[XXXXX922]": 44967, "[XXXXX487]": 44532, "[XXXXX115]": 44160, "[XXXXX170]": 44215, "[XXXXX1001]": 45046, "[XXXXX618]": 44663, "[XXXXX339]": 44384, "[XXXXX987]": 45032, "[XXXXX301]": 44346, "[XXXXX641]": 44686, "[XXXXX1009]": 45054, "[XXXXX821]": 44866, "[XXXXX741]": 44786, "[XXXXX378]": 44423, "[XXXXX751]": 44796, "[XXXXX407]": 44452, "[XXXXX938]": 44983, "[XXXXX764]": 44809, "[XXXXX518]": 44563, "[XXXXX728]": 44773, "[XXXXX773]": 44818, "[XXXXX713]": 44758, "[XXXXX431]": 44476, "[XXXXX961]": 45006, "[XXXXX553]": 44598, "[XXXXX65]": 44110, "[XXXXX845]": 44890, "[XXXXX557]": 44602, "[XXXXX823]": 44868, "[XXXXX666]": 44711, "[XXXXX945]": 44990, "[XXXXX480]": 44525, "[XXXXX392]": 44437, "[XXXXX924]": 44969, "[XXXXX67]": 44112, "[XXXXX939]": 44984, "[XXXXX729]": 44774, "[XXXXX798]": 44843, "[XXXXX485]": 44530, "[XXXXX257]": 44302, "[XXXXX193]": 44238, "[XXXXX867]": 44912, "[XXXXX421]": 44466, "[XXXXX490]": 44535, "[XXXXX882]": 44927, "[XXXXX382]": 44427, "[XXXXX928]": 44973, "[XXXXX905]": 44950, "[XXXXX903]": 44948, "[XXXXX135]": 44180, "[XXXXX853]": 44898, "[XXXXX537]": 44582, "[XXXXX265]": 44310, "[XXXXX855]": 44900, "[XXXXX88]": 44133, "[XXXXX852]": 44897, "[XXXXX200]": 44245, "[XXXXX606]": 44651, "[XXXXX282]": 44327, "[XXXXX996]": 45041, "[XXXXX137]": 44182, "[XXXXX107]": 44152, "[XXXXX616]": 44661, "[XXXXX565]": 44610, "[XXXXX332]": 44377, "[XXXXX430]": 44475, "[XXXXX422]": 44467, "[XXXXX375]": 44420, "[XXXXX575]": 44620, "[XXXXX82]": 44127, "[XXXXX712]": 44757, "[XXXXX172]": 44217, "[XXXXX676]": 44721, "[XXXXX784]": 44829, "[XXXXX735]": 44780, "[XXXXX441]": 44486, "[XXXXX69]": 44114, "[XXXXX930]": 44975, "[XXXXX433]": 44478, "[XXXXX781]": 44826, "[XXXXX815]": 44860, "[XXXXX160]": 44205, "[XXXXX80]": 44125, "[XXXXX310]": 44355, "[XXXXX598]": 44643, "[XXXXX87]": 44132, "[XXXXX209]": 44254, "[XXXXX496]": 44541, "[XXXXX691]": 44736, "[XXXXX907]": 44952, "[XXXXX805]": 44850, "[XXXXX702]": 44747, "[XXXXX913]": 44958, "[XXXXX119]": 44164, "[XXXXX650]": 44695, "[XXXXX701]": 44746, "[XXXXX673]": 44718, "[XXXXX195]": 44240, "[XXXXX112]": 44157, "[XXXXX230]": 44275, "[XXXXX241]": 44286, "[XXXXX605]": 44650, "[XXXXX765]": 44810, "[XXXXX752]": 44797, "[XXXXX973]": 45018, "[XXXXX792]": 44837, "[XXXXX114]": 44159, "[XXXXX175]": 44220, "[XXXXX957]": 45002, "[XXXXX90]": 44135, "[XXXXX854]": 44899, "[XXXXX303]": 44348, "[XXXXX647]": 44692, "[XXXXX954]": 44999, "[XXXXX204]": 44249, "[XXXXX674]": 44719, "[XXXXX232]": 44277, "[XXXXX327]": 44372, "[XXXXX706]": 44751, "[XXXXX581]": 44626, "[XXXXX692]": 44737, "[XXXXX149]": 44194, "[XXXXX668]": 44713, "[XXXXX278]": 44323, "[XXXXX941]": 44986, "[XXXXX493]": 44538, "[XXXXX264]": 44309, "[XXXXX569]": 44614, "[XXXXX144]": 44189, "[XXXXX977]": 45022, "[XXXXX699]": 44744, "[XXXXX294]": 44339, "[XXXXX835]": 44880, "[XXXXX47]": 44092, "[XXXXX442]": 44487, "[XXXXX19]": 44064, "[XXXXX693]": 44738, "[XXXXX101]": 44146, "[XXXXX929]": 44974, "[XXXXX686]": 44731, "[XXXXX281]": 44326, "[XXXXX923]": 44968, "[XXXXX795]": 44840, "[XXXXX62]": 44107, "[XXXXX612]": 44657, "[XXXXX829]": 44874, "[XXXXX33]": 44078, "[XXXXX342]": 44387, "[XXXXX533]": 44578, "[XXXXX468]": 44513, "[XXXXX1]": 44046, "[XXXXX742]": 44787, "[XXXXX203]": 44248, "[XXXXX150]": 44195, "[XXXXX840]": 44885, "[XXXXX329]": 44374, "[XXXXX360]": 44405, "[XXXXX744]": 44789, "[XXXXX423]": 44468, "[XXXXX224]": 44269, "[XXXXX128]": 44173, "[XXXXX997]": 45042, "[XXXXX544]": 44589, "[XXXXX807]": 44852, "[XXXXX357]": 44402, "[XXXXX508]": 44553, "[XXXXX252]": 44297, "[XXXXX816]": 44861, "[XXXXX99]": 44144, "[XXXXX563]": 44608, "[XXXXX516]": 44561, "[XXXXX753]": 44798, "[XXXXX955]": 45000, "[XXXXX8]": 44053, "[XXXXX89]": 44134, "[XXXXX658]": 44703, "[XXXXX964]": 45009, "[XXXXX205]": 44250, "[XXXXX879]": 44924, "[XXXXX401]": 44446, "[XXXXX249]": 44294, "[XXXXX355]": 44400, "[XXXXX169]": 44214, "[XXXXX545]": 44590, "[XXXXX809]": 44854, "[XXXXX465]": 44510, "[XXXXX269]": 44314, "[XXXXX196]": 44241, "[XXXXX474]": 44519, "[XXXXX466]": 44511, "[XXXXX749]": 44794, "[XXXXX720]": 44765, "[XXXXX219]": 44264, "[XXXXX636]": 44681, "[XXXXX250]": 44295, "[XXXXX672]": 44717, "[XXXXX935]": 44980, "[XXXXX556]": 44601, "[XXXXX399]": 44444, "[XXXXX912]": 44957, "[XXXXX437]": 44482, "[XXXXX968]": 45013, "[XXXXX859]": 44904, "[XXXXX12]": 44057, "[XXXXX632]": 44677, "[XXXXX440]": 44485, "[XXXXX969]": 45014, "[MASK]": 44044, "[XXXXX256]": 44301, "[XXXXX323]": 44368, "[XXXXX201]": 44246, "[XXXXX118]": 44163, "[XXXXX220]": 44265, "[XXXXX477]": 44522, "[XXXXX777]": 44822, "[XXXXX434]": 44479, "[XXXXX727]": 44772, "[XXXXX223]": 44268, "[XXXXX386]": 44431, "[XXXXX455]": 44500, "[XXXXX624]": 44669, "[XXXXX54]": 44099, "[XXXXX737]": 44782, "[XXXXX189]": 44234, "[XXXXX836]": 44881, "[XXXXX707]": 44752, "[XXXXX743]": 44788, "[XXXXX886]": 44931, "[XXXXX272]": 44317, "[XXXXX59]": 44104, "[XXXXX649]": 44694, "[XXXXX270]": 44315, "[XXXXX458]": 44503, "[XXXXX236]": 44281, "[XXXXX369]": 44414, "[XXXXX345]": 44390, "[XXXXX226]": 44271, "[XXXXX103]": 44148, "[XXXXX199]": 44244, "[XXXXX341]": 44386, "[XXXXX225]": 44270, "[XXXXX528]": 44573, "[XXXXX229]": 44274, "[XXXXX43]": 44088, "[XXXXX298]": 44343, "[XXXXX46]": 44091, "[XXXXX354]": 44399, "[XXXXX168]": 44213, "[XXXXX906]": 44951, "[XXXXX418]": 44463, "[XXXXX902]": 44947, "[XXXXX222]": 44267, "[XXXXX682]": 44727, "[XXXXX526]": 44571, "[XXXXX292]": 44337, "[XXXXX687]": 44732, "[XXXXX580]": 44625, "[XXXXX965]": 45010, "[XXXXX803]": 44848, "[XXXXX917]": 44962, "[XXXXX919]": 44964, "[XXXXX100]": 44145, "[XXXXX260]": 44305, "[XXXXX228]": 44273, "[XXXXX83]": 44128, "[XXXXX503]": 44548, "[XXXXX898]": 44943, "[XXXXX239]": 44284, "[XXXXX559]": 44604, "[XXXXX670]": 44715, "[XXXXX208]": 44253, "[XXXXX291]": 44336, "[XXXXX500]": 44545, "[XXXXX478]": 44523, "[XXXXX715]": 44760, "[XXXXX736]": 44781, "[XXXXX931]": 44976, "[XXXXX944]": 44989, "[XXXXX456]": 44501, "[XXXXX495]": 44540, "[XXXXX86]": 44131, "[XXXXX634]": 44679, "[XXXXX268]": 44313, "[XXXXX147]": 44192, "[XXXXX388]": 44433, "[XXXXX340]": 44385, "[XXXXX878]": 44923, "[XXXXX91]": 44136, "[XXXXX27]": 44072, "[XXXXX703]": 44748, "[XXXXX568]": 44613, "[XXXXX980]": 45025, "[XXXXX176]": 44221, "[XXXXX280]": 44325, "[XXXXX828]": 44873, "[XXXXX126]": 44171, "[XXXXX768]": 44813, "[XXXXX880]": 44925, "[XXXXX830]": 44875, "[XXXXX276]": 44321, "[XXXXX96]": 44141, "[XXXXX888]": 44933, "[XXXXX609]": 44654, "[XXXXX975]": 45020, "[XXXXX152]": 44197, "[XXXXX14]": 44059, "[XXXXX30]": 44075, "[XXXXX9]": 44054, "[XXXXX599]": 44644, "[XXXXX146]": 44191, "[XXXXX164]": 44209, "[XXXXX258]": 44303, "[XXXXX552]": 44597, "[XXXXX655]": 44700, "[XXXXX138]": 44183, "[XXXXX947]": 44992, "[XXXXX212]": 44257, "[XXXXX683]": 44728, "[XXXXX187]": 44232, "[XXXXX71]": 44116, "[XXXXX617]": 44662, "[XXXXX489]": 44534, "[XXXXX850]": 44895, "[XXXXX642]": 44687, "[XXXXX266]": 44311, "[XXXXX227]": 44272, "[XXXXX6]": 44051, "[XXXXX267]": 44312, "[XXXXX698]": 44743, "[XXXXX648]": 44693, "[XXXXX587]": 44632, "[XXXXX413]": 44458, "[XXXXX368]": 44413, "[XXXXX322]": 44367, "[XXXXX451]": 44496, "[XXXXX894]": 44939, "[XXXXX445]": 44490, "[XXXXX660]": 44705, "[XXXXX548]": 44593, "[XXXXX383]": 44428, "[XXXXX600]": 44645, "[XXXXX293]": 44338, "[XXXXX583]": 44628, "[XXXXX381]": 44426, "[XXXXX25]": 44070, "[XXXXX438]": 44483, "[XXXXX839]": 44884, "[XXXXX145]": 44190, "[XXXXX535]": 44580, "[XXXXX884]": 44929, "[XXXXX419]": 44464, "[XXXXX589]": 44634, "[XXXXX566]": 44611, "[XXXXX277]": 44322, "[XXXXX925]": 44970, "[XXXXX690]": 44735, "[XXXXX901]": 44946, "[XXXXX491]": 44536, "[XXXXX790]": 44835, "[XXXXX731]": 44776, "[XXXXX4]": 44049, "[XXXXX412]": 44457, "[XXXXX376]": 44421, "[XXXXX679]": 44724, "[XXXXX472]": 44517, "[XXXXX656]": 44701, "[XXXXX989]": 45034, "[XXXXX398]": 44443, "[XXXXX359]": 44404, "[XXXXX757]": 44802, "[XXXXX347]": 44392, "[XXXXX746]": 44791, "[XXXXX255]": 44300, "[XXXXX767]": 44812, "[XXXXX317]": 44362, "[XXXXX482]": 44527, "[XXXXX812]": 44857, "[XXXXX813]": 44858, "[XXXXX711]": 44756, "[XXXXX337]": 44382, "[XXXXX643]": 44688, "[XXXXX391]": 44436, "[XXXXX914]": 44959, "[XXXXX77]": 44122, "[XXXXX550]": 44595, "[XXXXX504]": 44549, "[XXXXX596]": 44641, "[XXXXX974]": 45019, "[XXXXX688]": 44733, "[XXXXX611]": 44656, "[XXXXX689]": 44734, "[XXXXX502]": 44547, "[XXXXX129]": 44174, "[XXXXX22]": 44067, "[XXXXX825]": 44870, "[XXXXX653]": 44698, "[XXXXX39]": 44084, "[XXXXX210]": 44255, "[XXXXX299]": 44344, "[XXXXX494]": 44539, "[XXXXX321]": 44366, "[XXXXX896]": 44941, "[XXXXX75]": 44120, "[XXXXX319]": 44364, "[XXXXX625]": 44670, "[XXXXX998]": 45043, "[XXXXX7]": 44052, "[XXXXX320]": 44365, "[XXXXX28]": 44073, "[XXXXX551]": 44596, "[XXXXX637]": 44682, "[XXXXX116]": 44161, "[XXXXX871]": 44916, "[XXXXX397]": 44442, "[XXXXX523]": 44568, "[XXXXX154]": 44199, "[XXXXX393]": 44438, "[XXXXX389]": 44434, "[XXXXX483]": 44528, "[XXXXX403]": 44448, "[XXXXX732]": 44777, "[XXXXX608]": 44653, "[XXXXX796]": 44841, "[XXXXX1008]": 45053, "[XXXXX286]": 44331, "[XXXXX704]": 44749, "[XXXXX849]": 44894, "[XXXXX585]": 44630, "[XXXXX133]": 44178, "[XXXXX988]": 45033, "[XXXXX501]": 44546, "[XXXXX847]": 44892, "[XXXXX435]": 44480, "[XXXXX971]": 45016, "[XXXXX336]": 44381, "[XXXXX136]": 44181, "[XXXXX802]": 44847, "[XXXXX289]": 44334, "[XXXXX179]": 44224, "[XXXXX24]": 44069, "[XXXXX84]": 44129, "[XXXXX733]": 44778, "[XXXXX20]": 44065, "[XXXXX654]": 44699, "[XXXXX94]": 44139, "[XXXXX48]": 44093, "[XXXXX417]": 44462, "[XXXXX214]": 44259, "[XXXXX863]": 44908, "[XXXXX287]": 44332, "[XXXXX772]": 44817, "[XXXXX554]": 44599, "[XXXXX1007]": 45052, "[XXXXX571]": 44616, "[XXXXX578]": 44623, "[XXXXX248]": 44293, "[XXXXX32]": 44077, "[XXXXX638]": 44683, "[XXXXX785]": 44830, "[XXXXX1005]": 45050, "[XXXXX774]": 44819, "[XXXXX15]": 44060, "[XXXXX34]": 44079, "[XXXXX148]": 44193, "[XXXXX915]": 44960, "[XXXXX782]": 44827, "[XXXXX273]": 44318, "[XXXXX363]": 44408, "[XXXXX761]": 44806, "[XXXXX620]": 44665, "[XXXXX58]": 44103, "[XXXXX870]": 44915, "[XXXXX531]": 44576, "[XXXXX467]": 44512, "[XXXXX921]": 44966, "[XXXXX284]": 44329, "[XXXXX185]": 44230, "[XXXXX471]": 44516, "[XXXXX771]": 44816, "[XXXXX794]": 44839, "[XXXXX450]": 44495, "[XXXXX862]": 44907, "[XXXXX353]": 44398, "[XXXXX577]": 44622, "[XXXXX626]": 44671, "[XXXXX865]": 44910, "[XXXXX117]": 44162, "[XXXXX134]": 44179, "[XXXXX61]": 44106, "[XXXXX395]": 44440, "[XXXXX125]": 44170, "[XXXXX444]": 44489, "[XXXXX1010]": 45055, "[XXXXX371]": 44416, "[XXXXX35]": 44080, "[XXXXX522]": 44567, "[XXXXX469]": 44514, "[XXXXX242]": 44287, "[XXXXX295]": 44340, "[XXXXX667]": 44712, "[XXXXX529]": 44574, "[XXXXX780]": 44825, "[XXXXX541]": 44586, "[XXXXX890]": 44935, "[XXXXX860]": 44905, "[XXXXX283]": 44328, "[XXXXX665]": 44710, "[XXXXX946]": 44991, "[XXXXX70]": 44115, "[XXXXX681]": 44726, "[XXXXX365]": 44410, "[XXXXX2]": 44047, "[XXXXX937]": 44982, "[XXXXX875]": 44920, "[XXXXX396]": 44441, "[XXXXX892]": 44937, "[XXXXX769]": 44814, "[XXXXX539]": 44584, "[XXXXX511]": 44556, "[XXXXX506]": 44551, "[XXXXX81]": 44126, "[XXXXX163]": 44208, "[XXXXX486]": 44531, "[XXXXX629]": 44674, "[XXXXX933]": 44978, "[XXXXX760]": 44805, "[XXXXX110]": 44155, "[XXXXX547]": 44592, "[XXXXX13]": 44058, "[XXXXX139]": 44184, "[XXXXX739]": 44784, "[XXXXX221]": 44266, "[XXXXX967]": 45012, "[XXXXX174]": 44219, "[XXXXX808]": 44853, "[XXXXX756]": 44801, "[XXXXX215]": 44260, "[XXXXX595]": 44640, "[XXXXX162]": 44207, "[XXXXX473]": 44518, "[XXXXX604]": 44649, "[XXXXX631]": 44676, "[XXXXX385]": 44430, "[XXXXX958]": 45003, "[XXXXX92]": 44137, "[XXXXX324]": 44369, "[XXXXX646]": 44691, "[XXXXX748]": 44793, "[XXXXX981]": 45026, "[XXXXX719]": 44764, "[XXXXX874]": 44919, "[XXXXX379]": 44424, "[XXXXX290]": 44335, "[XXXXX380]": 44425, "[XXXXX198]": 44243, "[XXXXX873]": 44918, "[XXXXX156]": 44201, "[XXXXX520]": 44565, "[XXXXX234]": 44279, "[XXXXX18]": 44063, "[XXXXX844]": 44889, "[XXXXX671]": 44716, "[XXXXX725]": 44770, "[XXXXX254]": 44299, "[XXXXX411]": 44456, "[XXXXX936]": 44981, "[XXXXX443]": 44488, "[XXXXX916]": 44961, "[XXXXX766]": 44811, "[XXXXX507]": 44552, "[XXXXX10]": 44055, "[XXXXX567]": 44612, "[XXXXX576]": 44621, "[XXXXX130]": 44175, "[XXXXX564]": 44609, "[XXXXX525]": 44570, "[XXXXX356]": 44401, "[XXXXX182]": 44227, "[XXXXX615]": 44660, "[XXXXX415]": 44460, "[XXXXX97]": 44142, "[XXXXX697]": 44742, "[XXXXX864]": 44909, "[XXXXX261]": 44306, "[XXXXX106]": 44151, "[XXXXX333]": 44378, "[XXXXX102]": 44147, "[XXXXX373]": 44418, "[XXXXX963]": 45008, "[XXXXX72]": 44117, "[XXXXX109]": 44154, "[XXXXX920]": 44965, "[XXXXX348]": 44393, "[XXXXX1000]": 45045, "[XXXXX452]": 44497, "[XXXXX705]": 44750, "[XXXXX763]": 44808, "[XXXXX940]": 44985, "[XXXXX999]": 45044, "[XXXXX696]": 44741, "[XXXXX111]": 44156, "[XXXXX775]": 44820, "[XXXXX639]": 44684, "[XXXXX31]": 44076, "[XXXXX887]": 44932, "[XXXXX420]": 44465, "[XXXXX314]": 44359, "[XXXXX509]": 44554, "[XXXXX26]": 44071, "[XXXXX243]": 44288, "[XXXXX216]": 44261, "[XXXXX645]": 44690, "[XXXXX108]": 44153, "[XXXXX811]": 44856, "[XXXXX315]": 44360, "[XXXXX180]": 44225, "[XXXXX38]": 44083, "[XXXXX759]": 44804, "[XXXXX897]": 44942, "[XXXXX334]": 44379, "[XXXXX453]": 44498, "[XXXXX367]": 44412, "[XXXXX56]": 44101, "[XXXXX678]": 44723, "[XXXXX297]": 44342, "[XXXXX470]": 44515, "<pad>": 44043, "[XXXXX238]": 44283, "[XXXXX165]": 44210, "[XXXXX738]": 44783, "[XXXXX218]": 44263, "[XXXXX192]": 44237, "[XXXXX562]": 44607, "[XXXXX29]": 44074, "[XXXXX695]": 44740, "[XXXXX449]": 44494, "[XXXXX505]": 44550, "[XXXXX833]": 44878, "[XXXXX352]": 44397, "[XXXXX621]": 44666, "[XXXXX237]": 44282, "[XXXXX995]": 45040, "[XXXXX669]": 44714, "[XXXXX602]": 44647, "[XXXXX994]": 45039, "[XXXXX970]": 45015, "[XXXXX622]": 44667, "[XXXXX909]": 44954, "[XXXXX78]": 44123, "[XXXXX447]": 44492, "[XXXXX484]": 44529, "[XXXXX657]": 44702, "[XXXXX985]": 45030, "[XXXXX142]": 44187, "[XXXXX36]": 44081, "[XXXXX416]": 44461, "[XXXXX157]": 44202, "[XXXXX900]": 44945, "[XXXXX308]": 44353, "[XXXXX50]": 44095, "[XXXXX461]": 44506, "[XXXXX570]": 44615, "[XXXXX167]": 44212, "[XXXXX498]": 44543, "[XXXXX42]": 44087, "[XXXXX213]": 44258, "[XXXXX235]": 44280, "[XXXXX492]": 44537, "[XXXXX726]": 44771, "[XXXXX374]": 44419, "[XXXXX723]": 44768, "[XXXXX73]": 44118, "[XXXXX247]": 44292, "[XXXXX358]": 44403, "[XXXXX868]": 44913, "[XXXXX956]": 45001, "[XXXXX63]": 44108, "[XXXXX710]": 44755, "[XXXXX948]": 44993, "[XXXXX1004]": 45049, "[XXXXX542]": 44587, "[XXXXX515]": 44560, "[XXXXX694]": 44739, "[XXXXX105]": 44150, "[XXXXX55]": 44100, "[XXXXX662]": 44707, "[XXXXX800]": 44845, "[XXXXX972]": 45017, "[XXXXX848]": 44893, "[XXXXX194]": 44239, "[XXXXX233]": 44278, "[XXXXX993]": 45038, "[XXXXX140]": 44185, "[XXXXX122]": 44167, "[XXXXX842]": 44887, "[XXXXX817]": 44862, "[XXXXX143]": 44188, "[XXXXX820]": 44865, "[XXXXX364]": 44409, "[XXXXX846]": 44891, "[XXXXX245]": 44290, "[XXXXX76]": 44121, "[XXXXX630]": 44675, "[XXXXX724]": 44769, "[XXXXX384]": 44429, "[XXXXX387]": 44432, "[XXXXX561]": 44606, "[XXXXX16]": 44061, "[XXXXX991]": 45036, "[XXXXX512]": 44557, "[XXXXX346]": 44391, "[XXXXX312]": 44357, "[XXXXX984]": 45029, "[XXXXX21]": 44066, "[XXXXX514]": 44559, "[XXXXX714]": 44759, "[XXXXX79]": 44124, "[XXXXX178]": 44223, "[XXXXX53]": 44098, "[XXXXX627]": 44672, "[XXXXX414]": 44459, "[XXXXX448]": 44493, "[XXXXX601]": 44646, "[XXXXX819]": 44864, "[XXXXX338]": 44383, "[XXXXX861]": 44906, "[XXXXX124]": 44169, "[XXXXX211]": 44256, "[XXXXX463]": 44508, "[XXXXX708]": 44753, "[XXXXX302]": 44347, "[XXXXX927]": 44972, "[XXXXX652]": 44697, "[XXXXX588]": 44633, "[XXXXX408]": 44453, "[XXXXX932]": 44977, "[XXXXX747]": 44792, "[XXXXX151]": 44196, "[XXXXX734]": 44779, "[XXXXX793]": 44838, "[XXXXX307]": 44352, "[XXXXX436]": 44481, "[XXXXX786]": 44831, "[XXXXX462]": 44507, "[XXXXX591]": 44636, "[XXXXX57]": 44102, "[XXXXX776]": 44821, "[XXXXX330]": 44375, "[XXXXX799]": 44844, "[XXXXX950]": 44995, "[XXXXX161]": 44206, "[XXXXX866]": 44911, "[XXXXX628]": 44673, "[XXXXX877]": 44922, "[XXXXX722]": 44767, "[XXXXX832]": 44877, "[XXXXX131]": 44176, "[XXXXX952]": 44997, "[XXXXX546]": 44591, "[XXXXX663]": 44708, "[XXXXX5]": 44050, "[XXXXX942]": 44987, "[XXXXX603]": 44648, "[XXXXX750]": 44795, "[XXXXX394]": 44439, "[XXXXX400]": 44445, "[XXXXX801]": 44846, "[XXXXX831]": 44876, "[XXXXX190]": 44235, "[XXXXX881]": 44926, "[XXXXX717]": 44762, "[XXXXX331]": 44376, "[XXXXX304]": 44349, "[XXXXX275]": 44320, "[XXXXX409]": 44454, "[XXXXX460]": 44505, "[XXXXX513]": 44558, "[XXXXX824]": 44869, "[XXXXX895]": 44940, "[XXXXX573]": 44618, "[XXXXX543]": 44588, "[XXXXX534]": 44579, "[XXXXX572]": 44617, "[XXXXX349]": 44394, "[XXXXX962]": 45007, "[XXXXX549]": 44594, "[XXXXX464]": 44509, "[XXXXX926]": 44971, "[XXXXX325]": 44370, "[XXXXX613]": 44658, "[XXXXX432]": 44477, "[XXXXX597]": 44642, "[XXXXX127]": 44172, "[XXXXX263]": 44308, "[XXXXX37]": 44082, "[XXXXX454]": 44499, "[XXXXX377]": 44422, "[XXXXX540]": 44585, "[XXXXX538]": 44583, "[XXXXX758]": 44803, "[XXXXX640]": 44685, "[XXXXX593]": 44638, "[XXXXX1002]": 45047, "[XXXXX288]": 44333, "[XXXXX827]": 44872, "[XXXXX791]": 44836, "[XXXXX271]": 44316, "[XXXXX978]": 45023, "[XXXXX721]": 44766, "[XXXXX530]": 44575, "[XXXXX806]": 44851, "[XXXXX837]": 44882, "[XXXXX306]": 44351, "[XXXXX755]": 44800, "[XXXXX177]": 44222, "[XXXXX904]": 44949, "[XXXXX814]": 44859, "[XXXXX951]": 44996, "[XXXXX343]": 44388, "[XXXXX328]": 44373, "[XXXXX651]": 44696, "[XXXXX40]": 44085, "[XXXXX240]": 44285, "[XXXXX788]": 44833, "[XXXXX300]": 44345, "[XXXXX66]": 44111, "[XXXXX633]": 44678, "[XXXXX586]": 44631, "[XXXXX911]": 44956, "[XXXXX390]": 44435, "[XXXXX510]": 44555, "[XXXXX93]": 44138, "[XXXXX120]": 44165, "[XXXXX730]": 44775, "[XXXXX181]": 44226, "[XXXXX910]": 44955, "[XXXXX3]": 44048, "[XXXXX677]": 44722, "[XXXXX166]": 44211, "[XXXXX949]": 44994, "[XXXXX899]": 44944, "[XXXXX251]": 44296, "[XXXXX789]": 44834, "[XXXXX132]": 44177, "[XXXXX574]": 44619, "[XXXXX778]": 44823, "[XXXXX623]": 44668, "[XXXXX253]": 44298, "[SEP]": 44042, "[XXXXX976]": 45021, "[XXXXX0]": 44045, "[XXXXX296]": 44341, "[XXXXX614]": 44659, "[XXXXX318]": 44363, "[XXXXX313]": 44358, "[XXXXX983]": 45028, "[XXXXX876]": 44921, "[XXXXX499]": 44544, "[XXXXX64]": 44109, "[XXXXX184]": 44229, "[XXXXX197]": 44242, "[XXXXX410]": 44455, "[XXXXX49]": 44094, "[XXXXX745]": 44790, "[XXXXX869]": 44914, "[XXXXX592]": 44637, "[XXXXX558]": 44603, "[XXXXX959]": 45004, "[XXXXX426]": 44471, "[XXXXX23]": 44068, "[XXXXX366]": 44411, "[XXXXX680]": 44725, "[XXXXX481]": 44526, "[XXXXX217]": 44262, "[XXXXX405]": 44450, "[XXXXX259]": 44304, "[XXXXX425]": 44470, "[XXXXX274]": 44319, "[XXXXX584]": 44629, "[XXXXX843]": 44888, "[XXXXX476]": 44521, "[XXXXX979]": 45024, "[XXXXX521]": 44566, "[XXXXX856]": 44901, "[XXXXX960]": 45005, "[XXXXX113]": 44158, "[XXXXX52]": 44097, "[XXXXX95]": 44140, "[XXXXX883]": 44928, "[XXXXX311]": 44356, "[XXXXX918]": 44963, "[CLS]": 44041, "[XXXXX754]": 44799, "[XXXXX404]": 44449, "[XXXXX457]": 44502, "[XXXXX579]": 44624, "[XXXXX822]": 44867, "[XXXXX402]": 44447, "[XXXXX770]": 44815, "[XXXXX428]": 44473, "[XXXXX527]": 44572, "[XXXXX536]": 44581, "[XXXXX305]": 44350, "[XXXXX891]": 44936, "[XXXXX344]": 44389, "[XXXXX779]": 44824, "[XXXXX635]": 44680, "[XXXXX104]": 44149, "[XXXXX202]": 44247, "[XXXXX121]": 44166, "[XXXXX872]": 44917, "[XXXXX246]": 44291, "[XXXXX351]": 44396, "[XXXXX155]": 44200, "[XXXXX285]": 44330, "[XXXXX885]": 44930, "[XXXXX532]": 44577, "[XXXXX1003]": 45048, "[XXXXX659]": 44704, "[XXXXX60]": 44105, "[XXXXX207]": 44252, "[XXXXX244]": 44289, "[XXXXX787]": 44832, "[XXXXX610]": 44655, "[XXXXX889]": 44934, "[XXXXX123]": 44168, "[XXXXX797]": 44842, "[XXXXX173]": 44218, "[XXXXX186]": 44231, "[XXXXX372]": 44417, "[XXXXX524]": 44569, "[XXXXX519]": 44564, "[XXXXX908]": 44953, "[XXXXX992]": 45037, "[XXXXX893]": 44938, "[XXXXX85]": 44130, "[XXXXX362]": 44407, "[XXXXX11]": 44056, "[XXXXX497]": 44542, "[XXXXX943]": 44988, "[XXXXX517]": 44562, "[XXXXX826]": 44871, "[XXXXX664]": 44709, "[XXXXX335]": 44380, "[XXXXX141]": 44186, "[XXXXX857]": 44902, "[XXXXX488]": 44533, "[XXXXX439]": 44484, "[XXXXX406]": 44451, "[XXXXX459]": 44504, "[XXXXX427]": 44472, "[XXXXX68]": 44113, "[XXXXX841]": 44886, "[XXXXX316]": 44361, "[XXXXX475]": 44520, "[XXXXX158]": 44203, "[XXXXX1006]": 45051, "[XXXXX810]": 44855, "[XXXXX361]": 44406, "[XXXXX45]": 44090, "[XXXXX685]": 44730, "[XXXXX555]": 44600, "[XXXXX804]": 44849, "[XXXXX44]": 44089, "[XXXXX986]": 45031, "[XXXXX17]": 44062, "[XXXXX762]": 44807, "[XXXXX858]": 44903, "[XXXXX188]": 44233, "[XXXXX183]": 44228, "[XXXXX934]": 44979, "[XXXXX709]": 44754, "[XXXXX740]": 44785, "[XXXXX159]": 44204, "[XXXXX350]": 44395, "[XXXXX171]": 44216, "[XXXXX966]": 45011, "[XXXXX594]": 44639, "[XXXXX684]": 44729, "[XXXXX429]": 44474, "[XXXXX446]": 44491, "[XXXXX838]": 44883, "[XXXXX326]": 44371, "[XXXXX279]": 44324, "[XXXXX479]": 44524, "[XXXXX675]": 44720, "[XXXXX818]": 44863, "[XXXXX953]": 44998, "[XXXXX718]": 44763, "[XXXXX661]": 44706, "[XXXXX834]": 44879, "[XXXXX783]": 44828, "[XXXXX982]": 45027, "[XXXXX41]": 44086}
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]"]}
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9579070d1fae994403b4351cf15f52572c51f2b6671aa2dc8796ca1e08e2cd5d
3
+ size 1073551
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/5mb/pus_arab_5mb", "model_input_names": ["input_ids", "attention_mask"], "special_tokens_map_file": "models/5mb/pus_arab_5mb/special_tokens_map.json", "tokenizer_class": "AlbertTokenizer"}