Training done

added_tokens.json

@@ -1,44 +1,131 @@
 {
-  "</s_benh_vien>": 59529,
-  "</s_chieu>": 59559,
-  "</s_dia_chi>": 59547,
-  "</s_doi_tuong>": 59543,
-  "</s_drug>": 59549,
-  "</s_gioi_tinh>": 59539,
-  "</s_khoa_dieu_tri>": 59527,
-  "</s_lieu_thuoc>": 59551,
-  "</s_ma_doi_tuong>": 59545,
-  "</s_ma_y_te>": 59541,
-  "</s_sang>": 59555,
-  "</s_so_luong>": 59553,
-  "</s_so_vao_vien>": 59531,
-  "</s_so_y_te>": 59525,
-  "</s_ten_benh_nhan>": 59535,
-  "</s_toa_thuoc>": 59533,
-  "</s_trua>": 59557,
-  "</s_tuoi>": 59537,
-  "<ocr-toa-thuoc>": 59560,
-  "<s_benh_vien>": 59528,
-  "<s_chieu>": 59558,
-  "<s_dia_chi>": 59546,
-  "<s_doi_tuong>": 59542,
-  "<s_drug>": 59548,
-  "<s_gioi_tinh>": 59538,
+  "</s_benh_vien>": 60049,
+  "</s_chieu>": 60079,
+  "</s_dia_chi>": 60067,
+  "</s_doi_tuong>": 60063,
+  "</s_drug>": 60069,
+  "</s_gioi_tinh>": 60059,
+  "</s_khoa_dieu_tri>": 60047,
+  "</s_lieu_thuoc>": 60071,
+  "</s_ma_doi_tuong>": 60065,
+  "</s_ma_y_te>": 60061,
+  "</s_sang>": 60075,
+  "</s_so_luong>": 60073,
+  "</s_so_vao_vien>": 60051,
+  "</s_so_y_te>": 60045,
+  "</s_ten_benh_nhan>": 60055,
+  "</s_toa_thuoc>": 60053,
+  "</s_trua>": 60077,
+  "</s_tuoi>": 60057,
+  "<ocr-toa-thuoc>": 60080,
+  "<s_benh_vien>": 60048,
+  "<s_chieu>": 60078,
+  "<s_dia_chi>": 60066,
+  "<s_doi_tuong>": 60062,
+  "<s_drug>": 60068,
+  "<s_gioi_tinh>": 60058,
   "<s_iitcdip>": 57523,
-  "<s_khoa_dieu_tri>": 59526,
-  "<s_lieu_thuoc>": 59550,
-  "<s_ma_doi_tuong>": 59544,
-  "<s_ma_y_te>": 59540,
-  "<s_sang>": 59554,
-  "<s_so_luong>": 59552,
-  "<s_so_vao_vien>": 59530,
-  "<s_so_y_te>": 59524,
+  "<s_khoa_dieu_tri>": 60046,
+  "<s_lieu_thuoc>": 60070,
+  "<s_ma_doi_tuong>": 60064,
+  "<s_ma_y_te>": 60060,
+  "<s_sang>": 60074,
+  "<s_so_luong>": 60072,
+  "<s_so_vao_vien>": 60050,
+  "<s_so_y_te>": 60044,
   "<s_synthdog>": 57524,
-  "<s_ten_benh_nhan>": 59534,
-  "<s_toa_thuoc>": 59532,
-  "<s_trua>": 59556,
-  "<s_tuoi>": 59536,
+  "<s_ten_benh_nhan>": 60054,
+  "<s_toa_thuoc>": 60052,
+  "<s_trua>": 60076,
+  "<s_tuoi>": 60056,
   "<sep/>": 57522,
+  "A!": 59911,
+  "A)": 59952,
+  "A,": 59733,
+  "A;": 59925,
+  "AY": 59679,
+  "E,": 59847,
+  "I!": 59898,
+  "I)": 60025,
+  "I,": 59737,
+  "I3": 59936,
+  "I;": 59905,
+  "IÀ": 59766,
+  "IÁ": 59715,
+  "IÃ": 59819,
+  "IÉ": 59969,
+  "IÊ": 59858,
+  "IÒ": 59835,
+  "IÓ": 59699,
+  "IÔ": 59971,
+  "IÚ": 59974,
+  "IÝ": 60034,
+  "IŨ": 59707,
+  "IƠ": 59769,
+  "IẠ": 59806,
+  "IẢ": 59572,
+  "IẺ": 59876,
+  "IỀ": 59964,
+  "IỌ": 59970,
+  "IỎ": 59797,
+  "IỒ": 59972,
+  "IỖ": 59767,
+  "IỘ": 59973,
+  "IỜ": 59623,
+  "IỞ": 59966,
+  "IỪ": 59975,
+  "IỮ": 59710,
+  "O,": 59738,
+  "O;": 59932,
+  "OE": 59908,
+  "O]": 60043,
+  "OÀ": 59639,
+  "OÁ": 59610,
+  "OÃ": 59823,
+  "OÈ": 59812,
+  "OÉ": 59799,
+  "OẠ": 59650,
+  "OẢ": 59772,
+  "OẸ": 59801,
+  "OẺ": 59989,
+  "OẼ": 60039,
+  "U,": 59742,
+  "U;": 59871,
+  "UY": 59593,
+  "UÀ": 59778,
+  "UÁ": 59853,
+  "UÈ": 59899,
+  "UÉ": 59961,
+  "UÊ": 59698,
+  "UÌ": 59907,
+  "UÍ": 59897,
+  "UÝ": 59647,
+  "UĨ": 59930,
+  "UƠ": 59852,
+  "UẠ": 59909,
+  "UẢ": 59582,
+  "UẺ": 59629,
+  "UẼ": 60026,
+  "UẾ": 59826,
+  "UỀ": 59951,
+  "UỂ": 60038,
+  "UỆ": 59665,
+  "UỈ": 59585,
+  "UỊ": 59948,
+  "UỌ": 60027,
+  "UỚ": 60001,
+  "UỜ": 59991,
+  "UỞ": 59855,
+  "UỲ": 59537,
+  "UỴ": 59538,
+  "UỶ": 59586,
+  "UỸ": 59885,
+  "Y!": 60020,
+  "Y)": 59996,
+  "Y,": 59732,
+  "Y;": 59937,
+  "YD": 59621,
+  "YÔ": 60003,
   "a!": 58437,
   "a)": 58675,
   "a,": 57925,
@@ -1176,6 +1263,135 @@
   "yễn": 58653,
   "yện": 57958,
   "yệt": 57774,
+  "À!": 59856,
+  "À)": 59957,
+  "À,": 59751,
+  "À;": 59943,
+  "ÀM": 59685,
+  "ÀN": 59532,
+  "ÀO": 59551,
+  "ÀU": 59672,
+  "ÀY": 59548,
+  "Á,": 59671,
+  "ÁC": 59529,
+  "ÁI": 59562,
+  "ÁN": 59533,
+  "ÁO": 59564,
+  "ÁP": 59625,
+  "ÁT": 59630,
+  "ÁU": 59763,
+  "ÁY": 59719,
+  "Â": 59791,
+  "Â,": 59790,
+  "ÂM": 59567,
+  "ÂN": 59560,
+  "ÂP": 59887,
+  "ÂU": 59554,
+  "ÂY": 59622,
+  "Ã": 59690,
+  "Ã)": 59955,
+  "ÃI": 59642,
+  "ÃM": 59824,
+  "ÃN": 59583,
+  "ÃO": 59640,
+  "ÃY": 59810,
+  "È": 59600,
+  "È)": 59848,
+  "ÈM": 59859,
+  "ÈN": 59759,
+  "ÈO": 59658,
+  "ÈY": 59906,
+  "ÈÕ": 59999,
+  "ÉC": 59857,
+  "ÉM": 59765,
+  "ÉN": 59700,
+  "ÉO": 59546,
+  "ÉP": 59668,
+  "ÉT": 59687,
+  "Ê": 59618,
+  "Ê,": 59910,
+  "ÊH": 59886,
+  "ÊM": 59649,
+  "ÊN": 59558,
+  "ÊP": 59931,
+  "ÊT": 59869,
+  "ÊU": 59632,
+  "ÊY": 59920,
+  "ÊÔ": 59983,
+  "Ì)": 59954,
+  "Ì,": 60012,
+  "Ì3": 60042,
+  "ÌA": 59680,
+  "ÌI": 60033,
+  "ÌM": 59808,
+  "ÌN": 59638,
+  "ÌU": 59796,
+  "Í'": 59953,
+  "Í)": 59965,
+  "Í,": 59977,
+  "ÍA": 59813,
+  "ÍC": 60032,
+  "ÍM": 59831,
+  "ÍN": 59746,
+  "ÍP": 59870,
+  "ÍT": 59540,
+  "ÍU": 59827,
+  "Ò": 59681,
+  "Ò)": 59995,
+  "Ò,": 60018,
+  "Ò;": 59994,
+  "ÒA": 59664,
+  "ÒE": 59811,
+  "ÒI": 59747,
+  "ÒM": 59878,
+  "ÒN": 59729,
+  "Ó,": 59782,
+  "Ó;": 59888,
+  "ÓA": 59675,
+  "ÓC": 59721,
+  "ÓE": 59798,
+  "ÓI": 59725,
+  "ÓM": 59849,
+  "ÓN": 59851,
+  "ÓP": 59818,
+  "ÓT": 59683,
+  "Ô": 59576,
+  "Ô!": 60021,
+  "Ô,": 60019,
+  "ÔC": 59984,
+  "ÔI": 59557,
+  "ÔM": 59549,
+  "ÔN": 59525,
+  "ÔR": 59935,
+  "ÔT": 59866,
+  "ÔỂ": 60009,
+  "Õ": 59660,
+  "ÕA": 59822,
+  "ÕI": 59758,
+  "ÕM": 59743,
+  "ÕN": 59814,
+  "Ù": 59535,
+  "Ù;": 59986,
+  "ÙA": 59786,
+  "ÙI": 59736,
+  "ÙM": 59776,
+  "ÙN": 59662,
+  "ÙY": 59868,
+  "Ú,": 60015,
+  "ÚA": 59626,
+  "ÚC": 59545,
+  "ÚI": 59682,
+  "ÚM": 59861,
+  "ÚN": 59874,
+  "ÚP": 59802,
+  "ÚT": 59718,
+  "ÚY": 59928,
+  "Ý": 59588,
+  "Ý,": 59938,
+  "ÝN": 60016,
+  "ÝP": 59992,
+  "ÝT": 59787,
+  "ÝU": 60000,
   "à!": 58201,
   "à)": 58710,
   "à,": 57961,
@@ -1442,6 +1658,12 @@
   "ýp": 58975,
   "ýt": 58031,
   "ýu": 59022,
+  "Ă": 59720,
+  "ĂC": 59941,
+  "ĂH": 59921,
+  "ĂK": 59942,
+  "ĂM": 59704,
+  "ĂN": 59606,
   "ă,Ă": 57899,
   "ăc": 58625,
   "ăccoóc": 57900,
@@ -1477,6 +1699,12 @@
   "ăngđa": 59069,
   "ăngđôlin": 59070,
   "ăảm": 59338,
+  "Ĩ": 59615,
+  "ĨA": 59612,
+  "ĨM": 59976,
+  "ĨN": 59919,
+  "ĨU": 59872,
+  "ĨỮ": 60023,
   "ĩ<SUP>o": 58694,
   "ĩa": 57683,
   "ĩa)": 59350,
@@ -1486,6 +1714,13 @@
   "ĩnh": 57750,
   "ĩu": 58242,
   "ĩữ": 59235,
+  "Ũ": 59654,
+  "Ũ,": 59775,
+  "ŨA": 59750,
+  "ŨI": 59773,
+  "ŨM": 59889,
+  "ŨN": 59912,
+  "ŨY": 59884,
   "ũ,": 58002,
   "ũa": 57955,
   "ũa;": 58505,
@@ -1494,6 +1729,13 @@
   "ũn": 58447,
   "ũy": 58289,
   "ũy,": 59068,
+  "Ơ": 59579,
+  "ƠI": 59526,
+  "ƠM": 59661,
+  "ƠN": 59536,
+  "ƠU": 59895,
+  "ƠÌ": 59713,
+  "ƠÍ": 59927,
   "ơ": 57620,
   "ơcanh": 59055,
   "ơcua": 59507,
@@ -1521,6 +1763,12 @@
   "ơu": 58325,
   "ơì": 57866,
   "ơí": 58520,
+  "Ư": 59604,
+  "Ư,": 60013,
+  "ƯA": 59692,
+  "ƯN": 59923,
+  "ƯU": 59547,
+  "ƯỜ": 60008,
   "ư": 57674,
   "ư,": 59160,
   "ưa": 57840,
@@ -1590,6 +1838,19 @@
   "ượp": 58491,
   "ượt": 57842,
   "ượu": 58090,
+  "Ạ": 59561,
+  "Ạ,": 59915,
+  "Ạ;": 60028,
+  "ẠC": 59556,
+  "ẠI": 59573,
+  "ẠM": 59581,
+  "ẠN": 59717,
+  "ẠO": 59568,
+  "ẠP": 59768,
+  "ẠT": 59659,
+  "ẠU": 59845,
+  "ẠY": 59663,
+  "ẠỊ": 60022,
   "ạ,": 58458,
   "ạ;": 59282,
   "ạc": 57584,
@@ -1615,6 +1876,16 @@
   "ạy": 57795,
   "ạị": 59220,
   "ạồm": 58415,
+  "Ả": 59550,
+  "Ả)": 59985,
+  "Ả,": 59947,
+  "Ả;": 59902,
+  "ẢI": 59605,
+  "ẢM": 59566,
+  "ẢN": 59793,
+  "ẢO": 59594,
+  "ẢU": 59846,
+  "ẢY": 59735,
   "ả": 57575,
   "ả)": 58936,
   "ả,": 58650,
@@ -1635,6 +1906,13 @@
   "ảy": 57928,
   "ảy,": 59458,
   "ảẳm": 59344,
+  "ẤC": 59637,
+  "ẤM": 59676,
+  "ẤN": 59635,
+  "ẤP": 59534,
+  "ẤT": 59655,
+  "ẤU": 59694,
+  "ẤY": 59596,
   "ấc": 57743,
   "ấm": 57823,
   "ấm,": 59123,
@@ -1647,6 +1925,10 @@
   "ất;": 58696,
   "ấu": 57844,
   "ấy": 57662,
+  "ẦM": 59757,
+  "ẦN": 59587,
+  "ẦU": 59552,
+  "ẦY": 59712,
   "ầm": 57974,
   "ầm,": 59162,
   "ần": 57631,
@@ -1656,16 +1938,31 @@
   "ầu,": 58314,
   "ầy": 57865,
   "ầy;": 58604,
+  "ẨM": 59684,
+  "ẨN": 59667,
+  "ẨU": 59527,
+  "ẨY": 59805,
   "ẩm": 57832,
   "ẩn": 57803,
   "ẩng": 58417,
   "ẩu": 57530,
   "ẩy": 58063,
+  "ẪM": 59726,
+  "ẪN": 59709,
+  "ẪU": 59764,
+  "ẪY": 59854,
   "ẫm": 57911,
   "ẫn": 57862,
   "ẫng": 58323,
   "ẫu": 57985,
   "ẫy": 58199,
+  "ẬC": 59792,
+  "ẬM": 59777,
+  "ẬN": 59590,
+  "ẬP": 59599,
+  "ẬT": 59524,
+  "ẬU": 59555,
+  "ẬY": 59727,
   "ậc": 58037,
   "ậm": 58007,
   "ậm,": 58269,
@@ -1678,6 +1975,11 @@
   "ậu": 57582,
   "ậu,": 57977,
   "ậy": 57912,
+  "ẮC": 59589,
+  "ẮM": 59711,
+  "ẮN": 59706,
+  "ẮP": 59722,
+  "ẮT": 59597,
   "ắc": 57641,
   "ắm": 57864,
   "ắm)": 58676,
@@ -1691,21 +1993,33 @@
   "ắt": 57663,
   "ắt,": 58512,
   "ắác": 59337,
+  "ẰM": 59734,
+  "ẰN": 59781,
   "ằm,": 59105,
   "ằn": 58020,
   "ằng": 57578,
   "ằng,": 59430,
   "ằng;": 58645,
+  "ẲM": 59904,
+  "ẲN": 59789,
   "ẳm": 58421,
   "ẳn": 58035,
   "ẳnc": 59188,
   "ẳng": 57860,
   "ẳón": 58394,
+  "ẴM": 59723,
+  "ẴN": 59816,
+  "ẴP": 60030,
   "ẵm": 57908,
   "ẵn": 58078,
   "ẵng": 58176,
   "ẵp": 59342,
   "ẵắn": 59340,
+  "ẶC": 59696,
+  "ẶM": 59830,
+  "ẶN": 59701,
+  "ẶP": 59678,
+  "ẶT": 59714,
   "ặc": 57847,
   "ặc,": 58584,
   "ặm": 58114,
@@ -1716,6 +2030,12 @@
   "ặt": 57868,
   "ặt,": 58833,
   "ặạt": 59339,
+  "ẸC": 59945,
+  "ẸM": 59959,
+  "ẸN": 59739,
+  "ẸO": 59836,
+  "ẸP": 59705,
+  "ẸT": 59688,
   "ẹ": 57804,
   "ẹc": 58642,
   "ẹm": 58721,
@@ -1723,6 +2043,11 @@
   "ẹo": 58133,
   "ẹp": 57857,
   "ẹt": 57835,
+  "Ẻ": 59628,
+  "Ẻ,": 60037,
+  "ẺM": 59860,
+  "ẺN": 59880,
+  "ẺO": 59728,
   "ẻ": 57719,
   "ẻ,": 59408,
   "ẻm": 58210,
@@ -1730,12 +2055,26 @@
   "ẻng": 58334,
   "��o": 57913,
   "ẻo,": 58357,
+  "Ẽ": 59807,
+  "Ẽ,": 59990,
+  "ẼM": 59832,
+  "ẼN": 59686,
+  "ẼO": 59760,
+  "ẼÊ": 60031,
   "ẽ": 58066,
   "ẽ,": 58960,
   "ẽm": 58119,
   "ẽn": 57834,
   "ẽo": 57979,
   "ẽê": 59346,
+  "Ế": 59624,
+  "Ế,": 59950,
+  "ẾC": 59867,
+  "ẾM": 59803,
+  "ẾN": 59541,
+  "ẾP": 59619,
+  "ẾT": 59595,
+  "ẾU": 59591,
   "ế": 57708,
   "ế,": 58662,
   "ếc": 58229,
@@ -1753,6 +2092,11 @@
   "ết,": 58524,
   "ếu": 57643,
   "ếu,": 58248,
+  "Ề": 59697,
+  "Ề,": 59916,
+  "ỀM": 59850,
+  "ỀN": 59620,
+  "ỀU": 59616,
   "ề": 57848,
   "ề,": 58459,
   "ềm": 58186,
@@ -1763,6 +2107,11 @@
   "ều": 57687,
   "ều,": 57937,
   "ều;": 58668,
+  "Ể": 59666,
+  "Ể,": 59933,
+  "ỂM": 59574,
+  "ỂN": 59716,
+  "ỂU": 59569,
   "ể": 57799,
   "ể,": 58568,
   "ểm": 57610,
@@ -1771,6 +2120,10 @@
   "ểng": 58889,
   "ểnh": 58282,
   "ểu": 57604,
+  "Ễ,": 60010,
+  "ỄM": 59613,
+  "ỄN": 59653,
+  "ỄU": 59843,
   "ễ": 57749,
   "ễ,": 59152,
   "ễm": 57684,
@@ -1779,6 +2132,14 @@
   "ễnh": 58122,
   "ễnh;": 58944,
   "ễu": 58154,
+  "Ệ": 59656,
+  "Ệ;": 59987,
+  "ỆC": 59783,
+  "ỆM": 59580,
+  "ỆN": 59553,
+  "ỆP": 59530,
+  "ỆT": 59559,
+  "ỆU": 59627,
   "ệ": 57775,
   "ệ;": 58951,
   "ệc": 58024,
@@ -1793,6 +2154,12 @@
   "ệt": 57589,
   "ệt;": 58941,
   "ệu": 57715,
+  "Ỉ": 59584,
+  "Ỉ,": 60011,
+  "ỈA": 59740,
+  "ỈM": 59839,
+  "ỈN": 59893,
+  "ỈU": 59795,
   "ỉ": 57627,
   "ỉ,": 59157,
   "ỉa": 57935,
@@ -1801,6 +2168,14 @@
   "ỉn": 58303,
   "ỉnh": 58156,
   "ỉu": 58041,
+  "Ị": 59631,
+  "Ị,": 60041,
+  "ỊA": 59543,
+  "ỊM": 59982,
+  "ỊN": 59761,
+  "ỊP": 59774,
+  "ỊT": 59708,
+  "ỊU": 59741,
   "ị": 57721,
   "ị,": 59482,
   "ịa": 57556,
@@ -1814,6 +2189,16 @@
   "ịt;": 58321,
   "ịu": 57936,
   "ịệp": 59360,
+  "Ọ": 59648,
+  "Ọ,": 60014,
+  "ỌA": 59651,
+  "ỌC": 59542,
+  "ỌE": 59800,
+  "ỌI": 59815,
+  "ỌM": 59926,
+  "ỌN": 59744,
+  "ỌP": 59863,
+  "ỌT": 59702,
   "ọ,": 59163,
   "ọa": 57767,
   "ọc": 57553,
@@ -1830,6 +2215,15 @@
   "ọp": 58222,
   "ọt": 57854,
   "ọt,": 58945,
+  "Ỏ": 59575,
+  "Ỏ,": 60002,
+  "ỎA": 59862,
+  "ỎC": 59949,
+  "ỎE": 59988,
+  "ỎI": 59531,
+  "ỎL": 59963,
+  "ỎM": 59873,
+  "ỎN": 59879,
   "ỏ": 57614,
   "ỏ,": 59032,
   "ỏa": 58214,
@@ -1841,6 +2235,13 @@
   "ỏn": 58260,
   "ỏng": 57565,
   "ỏng,": 58368,
+  "Ố": 59614,
+  "ỐC": 59617,
+  "ỐI": 59565,
+  "ỐM": 59834,
+  "ỐN": 59598,
+  "ỐP": 59837,
+  "ỐT": 59673,
   "ố": 57686,
   "ốc": 57690,
   "ốc,": 58315,
@@ -1855,6 +2256,12 @@
   "ốp": 58134,
   "ốt": 57817,
   "ốt,": 58347,
+  "Ồ": 59670,
+  "Ồ,": 59962,
+  "ỒI": 59634,
+  "ỒM": 59577,
+  "ỒN": 59794,
+  "ỒÍ": 59960,
   "ồ": 57813,
   "ồ,": 58756,
   "ồi": 57736,
@@ -1865,6 +2272,13 @@
   "ồng": 57705,
   "ồng,": 58445,
   "ồí": 58730,
+  "Ổ,": 59788,
+  "Ổ;": 59929,
+  "ỔC": 59922,
+  "ỔI": 59695,
+  "ỔM": 59875,
+  "ỔN": 59644,
+  "ỔÍ": 59934,
   "ổ": 57771,
   "ổ,": 58033,
   "ổ;": 58542,
@@ -1875,12 +2289,23 @@
   "ổng": 57962,
   "ổng,": 58360,
   "ổí": 58578,
+  "Ỗ": 59677,
+  "Ỗ,": 59748,
+  "ỖI": 59563,
+  "ỖM": 59914,
+  "ỖN": 59840,
   "ỗ": 57824,
   "ỗ,": 57946,
   "ỗi": 57597,
   "ỗm": 58457,
   "ỗn": 58142,
   "ỗng": 57963,
+  "ỘC": 59570,
+  "ỘI": 59608,
+  "ỘM": 59785,
+  "ỘN": 59745,
+  "ỘP": 59864,
+  "ỘT": 59578,
   "ộ": 57541,
   "ộc": 57606,
   "ộc)": 58606,
@@ -1897,6 +2322,16 @@
   "ột": 57618,
   "ột)": 57595,
   "ột,": 58312,
+  "Ớ": 59691,
+  "ỚC": 59646,
+  "ỚI": 59528,
+  "ỚM": 59829,
+  "ỚN": 59784,
+  "ỚP": 59755,
+  "ỚT": 59730,
+  "ỚU": 59896,
+  "ỚÌ": 59754,
+  "ỚỈ": 59944,
   "ớ": 57839,
   "ớc": 57757,
   "ớc,": 57952,
@@ -1912,6 +2347,16 @@
   "ớu": 58327,
   "ớì": 57969,
   "ớỉ": 58638,
+  "Ờ": 59603,
+  "Ờ,": 59669,
+  "ỜI": 59601,
+  "ỜL": 60036,
+  "ỜM": 59841,
+  "ỜN": 59820,
+  "ỜU": 60040,
+  "ỜÌ": 59958,
+  "ỜÍ": 59939,
+  "ỜÕ": 60007,
   "ờ": 57672,
   "ờ,": 57812,
   "ời": 57667,
@@ -1929,6 +2374,12 @@
   "ờì": 58713,
   "ờí": 58620,
   "ờõ": 59141,
+  "Ở,": 59749,
+  "ỞI": 59770,
+  "ỞM": 59918,
+  "ỞN": 59917,
+  "ỞU": 59804,
+  "ỞÌ": 59924,
   "ở": 57918,
   "ở,": 57950,
   "ởi": 57992,
@@ -1938,6 +2389,14 @@
   "ởng": 57747,
   "ởu": 58059,
   "ởì": 58488,
+  "Ỡ": 59674,
+  "ỠI": 59817,
+  "ỠM": 59978,
+  "ỠN": 59883,
+  "ỠU": 60006,
+  "ỠÌ": 59900,
+  "ỠÍ": 59828,
+  "ỠỈ": 59903,
   "ỡ": 57819,
   "ỡi": 58079,
   "ỡm": 58895,
@@ -1947,6 +2406,15 @@
   "ỡì": 58392,
   "ỡí": 58110,
   "ỡỉ": 58419,
+  "Ợ": 59633,
+  "Ợ,": 59894,
+  "ỢC": 59762,
+  "ỢI": 59641,
+  "ỢM": 59838,
+  "ỢN": 59602,
+  "ỢP": 59825,
+  "ỢT": 59693,
+  "ỢU": 59821,
   "ợ": 57724,
   "ợ,": 58313,
   "ợc": 57981,
@@ -1958,6 +2426,16 @@
   "ợp": 58097,
   "ợt": 57841,
   "ợu": 58089,
+  "Ụ": 59645,
+  "Ụ;": 59913,
+  "ỤA": 59703,
+  "ỤC": 59544,
+  "ỤI": 59592,
+  "ỤM": 59809,
+  "ỤN": 59881,
+  "ỤP": 59890,
+  "ỤT": 59689,
+  "ỤY": 59844,
   "ụ;": 58448,
   "ụa": 57855,
   "ụa,": 59146,
@@ -1972,6 +2450,15 @@
   "ụt": 57836,
   "ụt,": 58873,
   "ụy": 58157,
+  "Ủ,": 59940,
+  "Ủ;": 59956,
+  "ỦA": 59753,
+  "ỦI": 59652,
+  "ỦM": 59891,
+  "ỦN": 59892,
+  "ỦO": 59997,
+  "ỦY": 59865,
+  "ỦÓ": 59998,
   "ủ": 57755,
   "ủ,": 58624,
   "ủ;": 58695,
@@ -1984,6 +2471,12 @@
   "ủo": 58997,
   "ủy": 58224,
   "ủó": 58998,
+  "Ứ!": 59980,
+  "ỨA": 59780,
+  "ỨC": 59571,
+  "ỨN": 60004,
+  "ỨT": 59756,
+  "ỨU": 59882,
   "ứ": 57837,
   "ứ!": 58903,
   "ứa": 58017,
@@ -1993,6 +2486,14 @@
   "ứng,": 59124,
   "ứt": 57973,
   "ứu": 58268,
+  "Ừ": 59657,
+  "Ừ!": 59979,
+  "Ừ,": 60005,
+  "Ừ;": 59901,
+  "ỪA": 59752,
+  "ỪM": 59981,
+  "ỪN": 60024,
+  "ỪU": 59842,
   "ừ": 57780,
   "ừ!": 58902,
   "ừ,": 59115,
@@ -2003,12 +2504,18 @@
   "ừng": 57592,
   "ừng,": 57947,
   "ừu": 58146,
+  "ỬA": 59724,
+  "ỬI": 59771,
+  "ỬU": 59833,
   "ử": 57675,
   "ửa": 57909,
   "ửa,": 58446,
   "ửi": 57993,
   "ửng": 58006,
   "ửu": 58127,
+  "Ữ": 59611,
+  "ỮA": 59731,
+  "ỮU": 59607,
   "ữ": 57682,
   "ữa": 57920,
   "ữa,": 58562,
@@ -2016,6 +2523,12 @@
   "ững": 58218,
   "ữu": 57678,
   "ữu,": 59154,
+  "Ự,": 60035,
+  "Ự;": 59946,
+  "ỰA": 59877,
+  "ỰC": 59609,
+  "ỰT": 59643,
+  "ỰU": 59636,
   "ự": 57711,
   "ự,": 59355,
   "ự;": 58649,
@@ -2028,6 +2541,12 @@
   "ựu,": 58506,
   "ỳ": 57546,
   "ỳnh": 57571,
+  "Ỵ": 59539,
+  "ỴA": 59967,
+  "ỴN": 60017,
+  "ỴP": 60029,
+  "ỴT": 59779,
+  "ỴU": 59993,
   "ỵ": 57548,
   "ỵa": 58797,
   "ỵch": 58901,
@@ -2035,6 +2554,7 @@
   "ỵp": 59305,
   "ỵt": 58015,
   "ỵu": 58976,
+  "ỶU": 59968,
   "ỷ": 57630,
   "ỷnh": 58973,
   "ỷu": 58803

preprocessor_config.json

@@ -1,22 +1,4 @@
 {
-  "_valid_processor_keys": [
-    "images",
-    "do_resize",
-    "size",
-    "resample",
-    "do_thumbnail",
-    "do_align_long_axis",
-    "do_pad",
-    "random_padding",
-    "do_rescale",
-    "rescale_factor",
-    "do_normalize",
-    "image_mean",
-    "image_std",
-    "return_tensors",
-    "data_format",
-    "input_data_format"
-  ],
   "do_align_long_axis": false,
   "do_normalize": true,
   "do_pad": true,
@@ -37,8 +19,8 @@
   "processor_class": "DonutProcessor",
   "resample": 2,
   "rescale_factor": 0.00392156862745098,
-  "size": {
-    "height": 1280,
-    "width": 960
-  }
+  "size": [
+    960,
+    1280
+  ]
 }

special_tokens_map.json

@@ -3,27 +3,9 @@
     "<s_iitcdip>",
     "<s_synthdog>"
   ],
-  "bos_token": {
-    "content": "<s>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "cls_token": {
-    "content": "<s>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "eos_token": {
-    "content": "</s>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
+  "bos_token": "<s>",
+  "cls_token": "<s>",
+  "eos_token": "</s>",
   "mask_token": {
     "content": "<mask>",
     "lstrip": true,
@@ -31,25 +13,7 @@
     "rstrip": false,
     "single_word": false
   },
-  "pad_token": {
-    "content": "<pad>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "sep_token": {
-    "content": "</s>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "unk_token": {
-    "content": "<unk>",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  }
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "unk_token": "<unk>"
 }