Upload tokenizer

added_tokens.json

@@ -20,13 +20,10 @@
   "Hai_bê": 64139,
   "Hi_oo": 64051,
   "Hiệ": 64179,
-  "Huy_h": 64285,
   "Hòa": 64225,
-  "Hội_gười": 64304,
   "I_o": 64097,
   "Iijima_Isao": 64046,
   "Ishiba": 64007,
-  "Jea_e": 64284,
   "Kaiha_a": 64054,
   "Ke": 64055,
   "Koji": 64053,
@@ -39,8 +36,6 @@
   "Nam_Tow": 64264,
   "Nam_chố": 64159,
   "Nam_hực": 64128,
-  "Nam_vẫ": 64282,
-  "Nam_ă": 64311,
   "Nam_ại": 64031,
   "Nam_ỉ": 64243,
   "Nam_ừ": 64016,
@@ -54,7 +49,6 @@
   "Phạm_Mi_h": 64021,
   "Phạm_Qua": 64032,
   "Phầ_lớ": 64249,
-  "Phố_Việ": 64281,
   "S_isuk": 64271,
   "Shige_u": 64005,
   "TPHCM.": 64183,
@@ -84,9 +78,7 @@
   "Xuâ_Hòa": 64152,
   "Yoshi_o": 64052,
   "Yoshiko": 64008,
-  "a_g": 64296,
   "a_h": 64278,
-  "a_hà_h": 64309,
   "a_o": 64056,
   "aka_Mao": 64049,
   "am_Tow": 64266,
@@ -102,7 +94,6 @@
   "bắ": 64012,
   "bắ_đầu": 64274,
   "bằ": 64251,
-  "bổ_g": 64316,
   "chiế": 64226,
   "chiế_lược": 64067,
   "chuyê_cơ": 64001,
@@ -112,18 +103,15 @@
   "chuyể_đổi": 64089,
   "chí_h": 64084,
   "chấ": 64077,
-  "chức_ă": 64283,
   "co_gười": 64254,
   "co_phố": 64267,
   "co_si": 64250,
-  "co_số": 64291,
   "co_xa": 64155,
   "cà_g": 64261,
   "cá_hâ": 64117,
   "cù_g": 64009,
   "cơ_sở_hạ": 64131,
   "cườ": 64102,
-  "cải_hiệ": 64313,
   "cậy_chí": 64081,
   "diễ_a": 64199,
   "diệ": 64069,
@@ -132,25 +120,20 @@
   "dâ": 64258,
   "dự_g": 64164,
   "dự_á": 64095,
-  "eu_o": 64315,
   "ghiệp": 64134,
   "ghiệp_hố": 64236,
-  "ghĩ": 64299,
   "ghĩa": 64060,
   "ghị": 64198,
   "giai_đọa": 64196,
   "goại": 64111,
-  "guyệ": 64306,
   "guồ": 64132,
   "guồ_lực": 64200,
   "gày": 64017,
-  "gôi_hà": 64292,
   "gười": 64154,
   "gầ": 64201,
   "gắ": 64268,
   "h_Vũ": 64030,
   "h_chí": 64270,
-  "h_cư": 64287,
   "h_cảm": 64212,
   "h_hổ": 64181,
   "h_hức": 64014,
@@ -158,7 +141,6 @@
   "h_vực": 64080,
   "h_đạo": 64115,
   "hau": 64276,
-  "hiêu": 64301,
   "hiế": 64227,
   "hiế_hực": 64191,
   "hiều": 64094,
@@ -166,7 +148,6 @@
   "hiều_ý": 64203,
   "hiệ": 64129,
   "hiệm": 64170,
-  "hà_cửa": 64294,
   "hà_g": 64110,
   "hà_h": 64189,
   "hà_lã": 64118,
@@ -177,7 +158,6 @@
   "hâ_lực": 64133,
   "hì_h": 64150,
   "hòa": 64070,
-  "hòa_òa": 64302,
   "hóa": 64135,
   "hăm_TPHCM": 64222,
   "hăm_chí": 64013,
@@ -188,7 +168,6 @@
   "hươ": 64137,
   "hướ": 64210,
   "hấ": 64138,
-  "hấy": 64290,
   "hậ": 64202,
   "hằm": 64101,
   "hế": 64061,
@@ -200,7 +179,6 @@
   "hỏ": 64235,
   "hỗ_ợ": 64127,
   "hợp_ác": 64082,
-  "hứ": 64297,
   "hữ": 64108,
   "hữu_ghị": 64259,
   "i_Masafumi": 64045,
@@ -214,9 +192,6 @@
   "iể": 64093,
   "iể_hà": 64165,
   "iể_hực": 64076,
-  "kha_g": 64295,
-  "khô_g": 64289,
-  "khă": 64314,
   "khả_ă": 64143,
   "khẳ": 64119,
   "khỏa": 64180,
@@ -240,7 +215,6 @@
   "mặ_hà": 64144,
   "mừ": 64157,
   "o_Naoki": 64058,
-  "o_lớ": 64300,
   "phươ": 64121,
   "phấ_đấu": 64140,
   "phầ": 64177,
@@ -249,7 +223,6 @@
   "quốc_phò": 64123,
   "quốc_ế": 64125,
   "si_h": 64173,
-  "suấ": 64312,
   "sâ": 64024,
   "sả_xuấ": 64194,
   "u_g": 64105,
@@ -261,15 +234,11 @@
   "vậ_lực": 64234,
   "vữ": 64186,
   "xa_h": 64090,
-  "xuố": 64288,
-  "xúc_độ": 64303,
   "xươ": 64228,
   "Ô_g": 64213,
   "ác_o": 64088,
-  "âm_sự": 64286,
   "âm_ă": 64106,
   "âm_đặc": 64175,
-  "âm_ới": 64307,
   "ê_hế": 64263,
   "ê_hế_giới": 64074,
   "ê_ấ": 64078,
@@ -277,7 +246,6 @@
   "í_h": 64246,
   "í_hức": 64206,
   "òa": 64068,
-  "ô_g": 64310,
   "ă_mó": 64277,
   "Đô_g": 64103,
   "Đả_g": 64241,
@@ -286,7 +254,6 @@
   "đà_phá": 64075,
   "đòa": 64010,
   "đú": 64223,
-  "đơ_sơ": 64293,
   "đạ": 64188,
   "đại_hóa": 64136,
   "đấ_ước": 64168,
@@ -295,7 +262,6 @@
   "đề_liê": 64209,
   "đị": 64120,
   "đối_ác": 64087,
-  "đứ": 64308,
   "ơ_Chí_h": 64242,
   "ưu_iê": 64109,
   "ươ": 64217,
@@ -307,10 +273,8 @@
   "ườ": 64142,
   "ưở": 64027,
   "ưở_g": 64025,
-  "ấ_hiều": 64298,
   "ấ_đô": 64253,
   "ẩm_hực": 64275,
-  "ập_hợp": 64305,
   "ập_u": 64126,
   "ụ_cộ": 64083,
   "Ủy": 64171,

tokenizer_config.json

@@ -360,14 +360,6 @@
       "single_word": false,
       "special": false
     },
-    "99": {
-      "content": "1",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "100": {
       "content": "số",
       "lstrip": false,
@@ -400,14 +392,6 @@
       "single_word": false,
       "special": false
     },
-    "118": {
-      "content": "hay",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "120": {
       "content": "Hà_Nội",
       "lstrip": false,
@@ -432,14 +416,6 @@
       "single_word": false,
       "special": false
     },
-    "133": {
-      "content": "giá",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "135": {
       "content": "...",
       "lstrip": false,
@@ -472,14 +448,6 @@
       "single_word": false,
       "special": false
     },
-    "162": {
-      "content": "Sau",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "163": {
       "content": "4",
       "lstrip": false,
@@ -504,14 +472,6 @@
       "single_word": false,
       "special": false
     },
-    "173": {
-      "content": "5",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "181": {
       "content": "cấp",
       "lstrip": false,
@@ -528,38 +488,6 @@
       "single_word": false,
       "special": false
     },
-    "193": {
-      "content": "em",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "207": {
-      "content": "mọi",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "218": {
-      "content": "Tôi",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "219": {
-      "content": "lúc",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "222": {
       "content": "học",
       "lstrip": false,
@@ -640,14 +568,6 @@
       "single_word": false,
       "special": false
     },
-    "267": {
-      "content": "xã_hội",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "268": {
       "content": "Phó",
       "lstrip": false,
@@ -744,14 +664,6 @@
       "single_word": false,
       "special": false
     },
-    "359": {
-      "content": "khó",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "378": {
       "content": "độ",
       "lstrip": false,
@@ -776,22 +688,6 @@
       "single_word": false,
       "special": false
     },
-    "426": {
-      "content": "9",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "434": {
-      "content": "Năm",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "447": {
       "content": "hỏi",
       "lstrip": false,
@@ -800,14 +696,6 @@
       "single_word": false,
       "special": false
     },
-    "450": {
-      "content": "Pháp",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "471": {
       "content": "sớm",
       "lstrip": false,
@@ -832,14 +720,6 @@
       "single_word": false,
       "special": false
     },
-    "499": {
-      "content": "bước",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "506": {
       "content": "hộ",
       "lstrip": false,
@@ -960,14 +840,6 @@
       "single_word": false,
       "special": false
     },
-    "654": {
-      "content": "hôm",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "669": {
       "content": "Đó",
       "lstrip": false,
@@ -984,14 +856,6 @@
       "single_word": false,
       "special": false
     },
-    "680": {
-      "content": "Bà",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "681": {
       "content": "giới",
       "lstrip": false,
@@ -1000,14 +864,6 @@
       "single_word": false,
       "special": false
     },
-    "696": {
-      "content": "100",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "711": {
       "content": "bức",
       "lstrip": false,
@@ -1016,14 +872,6 @@
       "single_word": false,
       "special": false
     },
-    "733": {
-      "content": "đời",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "740": {
       "content": "lập",
       "lstrip": false,
@@ -1080,14 +928,6 @@
       "single_word": false,
       "special": false
     },
-    "825": {
-      "content": "chủ_yếu",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "835": {
       "content": "phố",
       "lstrip": false,
@@ -1152,14 +992,6 @@
       "single_word": false,
       "special": false
     },
-    "956": {
-      "content": "đại_học",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "988": {
       "content": "chở",
       "lstrip": false,
@@ -1192,22 +1024,6 @@
       "single_word": false,
       "special": false
     },
-    "1093": {
-      "content": "hội",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "1106": {
-      "content": "Mỗi",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1111": {
       "content": "i",
       "lstrip": false,
@@ -1248,14 +1064,6 @@
       "single_word": false,
       "special": false
     },
-    "1287": {
-      "content": "cầm",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1292": {
       "content": "châu_Á",
       "lstrip": false,
@@ -1272,14 +1080,6 @@
       "single_word": false,
       "special": false
     },
-    "1338": {
-      "content": "bây_giờ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1351": {
       "content": "Cả",
       "lstrip": false,
@@ -1328,14 +1128,6 @@
       "single_word": false,
       "special": false
     },
-    "1464": {
-      "content": "bao",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1517": {
       "content": "a",
       "lstrip": false,
@@ -1368,14 +1160,6 @@
       "single_word": false,
       "special": false
     },
-    "1606": {
-      "content": "lúa",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1615": {
       "content": "Chiều",
       "lstrip": false,
@@ -1384,14 +1168,6 @@
       "single_word": false,
       "special": false
     },
-    "1663": {
-      "content": "70",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1664": {
       "content": "h",
       "lstrip": false,
@@ -1408,14 +1184,6 @@
       "single_word": false,
       "special": false
     },
-    "1750": {
-      "content": "lao",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "1775": {
       "content": "đám",
       "lstrip": false,
@@ -1440,14 +1208,6 @@
       "single_word": false,
       "special": false
     },
-    "1912": {
-      "content": "độc_lập",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "2089": {
       "content": "pháp_lý",
       "lstrip": false,
@@ -1472,14 +1232,6 @@
       "single_word": false,
       "special": false
     },
-    "2153": {
-      "content": "sở",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "2185": {
       "content": "phó",
       "lstrip": false,
@@ -1504,14 +1256,6 @@
       "single_word": false,
       "special": false
     },
-    "2209": {
-      "content": "cơ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "2217": {
       "content": "đậm",
       "lstrip": false,
@@ -1568,14 +1312,6 @@
       "single_word": false,
       "special": false
     },
-    "2469": {
-      "content": "hưu",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "2557": {
       "content": "chức_vụ",
       "lstrip": false,
@@ -1624,14 +1360,6 @@
       "single_word": false,
       "special": false
     },
-    "2913": {
-      "content": "lò",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "2991": {
       "content": "ô",
       "lstrip": false,
@@ -1688,14 +1416,6 @@
       "single_word": false,
       "special": false
     },
-    "3810": {
-      "content": "kỹ_sư",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "3988": {
       "content": "ổ",
       "lstrip": false,
@@ -1728,22 +1448,6 @@
       "single_word": false,
       "special": false
     },
-    "4216": {
-      "content": "1992",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "4278": {
-      "content": "1994",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "4318": {
       "content": "MTTQ",
       "lstrip": false,
@@ -1856,14 +1560,6 @@
       "single_word": false,
       "special": false
     },
-    "5854": {
-      "content": "Nội_Bài",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "6007": {
       "content": "hư",
       "lstrip": false,
@@ -1992,14 +1688,6 @@
       "single_word": false,
       "special": false
     },
-    "8915": {
-      "content": "ời",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "8942": {
       "content": "lê",
       "lstrip": false,
@@ -2032,14 +1720,6 @@
       "single_word": false,
       "special": false
     },
-    "9412": {
-      "content": "ài",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "9456": {
       "content": "ề",
       "lstrip": false,
@@ -2168,14 +1848,6 @@
       "single_word": false,
       "special": false
     },
-    "12416": {
-      "content": "ợ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "13291": {
       "content": "ộc",
       "lstrip": false,
@@ -2216,14 +1888,6 @@
       "single_word": false,
       "special": false
     },
-    "15002": {
-      "content": "Cò",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "15145": {
       "content": "úc",
       "lstrip": false,
@@ -2264,14 +1928,6 @@
       "single_word": false,
       "special": false
     },
-    "16788": {
-      "content": "iểu",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "17341": {
       "content": "iếp",
       "lstrip": false,
@@ -2296,14 +1952,6 @@
       "single_word": false,
       "special": false
     },
-    "20017": {
-      "content": "đỡ_đầu",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "20463": {
       "content": "kiều_hối",
       "lstrip": false,
@@ -2352,14 +2000,6 @@
       "single_word": false,
       "special": false
     },
-    "24096": {
-      "content": "há",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "30251": {
       "content": "ă",
       "lstrip": false,
@@ -2400,14 +2040,6 @@
       "single_word": false,
       "special": false
     },
-    "35477": {
-      "content": "hy",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "37463": {
       "content": "ổ_chức",
       "lstrip": false,
@@ -2456,14 +2088,6 @@
       "single_word": false,
       "special": false
     },
-    "46921": {
-      "content": "a_đời",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
     "48978": {
       "content": "â",
       "lstrip": false,
@@ -4743,294 +4367,6 @@
       "rstrip": false,
       "single_word": false,
       "special": false
-    },
-    "64281": {
-      "content": "Phố_Việ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64282": {
-      "content": "Nam_vẫ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64283": {
-      "content": "chức_ă",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64284": {
-      "content": "Jea_e",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64285": {
-      "content": "Huy_h",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64286": {
-      "content": "âm_sự",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64287": {
-      "content": "h_cư",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64288": {
-      "content": "xuố",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64289": {
-      "content": "khô_g",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64290": {
-      "content": "hấy",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64291": {
-      "content": "co_số",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64292": {
-      "content": "gôi_hà",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64293": {
-      "content": "đơ_sơ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64294": {
-      "content": "hà_cửa",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64295": {
-      "content": "kha_g",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64296": {
-      "content": "a_g",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64297": {
-      "content": "hứ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64298": {
-      "content": "ấ_hiều",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64299": {
-      "content": "ghĩ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64300": {
-      "content": "o_lớ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64301": {
-      "content": "hiêu",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64302": {
-      "content": "hòa_òa",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64303": {
-      "content": "xúc_độ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64304": {
-      "content": "Hội_gười",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64305": {
-      "content": "ập_hợp",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64306": {
-      "content": "guyệ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64307": {
-      "content": "âm_ới",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64308": {
-      "content": "đứ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64309": {
-      "content": "a_hà_h",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64310": {
-      "content": "ô_g",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64311": {
-      "content": "Nam_ă",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64312": {
-      "content": "suấ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64313": {
-      "content": "cải_hiệ",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64314": {
-      "content": "khă",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64315": {
-      "content": "eu_o",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "64316": {
-      "content": "bổ_g",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
     }
   },
   "bos_token": "<s>",